Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities

Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities Matthew Johnson, Winfred Dotse-Gborgbortsi, Anthony Ofosu, Chigozie Edson Utazi, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6790840/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Background Suboptimal resource allocation limits uptake of essential health services in sub-Saharan Africa. Reliable population denominators and intervention coverage estimates may inform allocation such that health facility resources better match demand. An approach utilising routinely collected health data has previously been used for region-/district-level estimation. This ecological study aimed to explore its use at lower administrative levels, taking Ghana as an example. Methods Data collected by Ghanaian health facilities from 2017–2020 were obtained from the District Health Information Management System and used to estimate childhood vaccination coverage and diarrhoea service utilisation in 2020 for the Eastern region, districts therein, and selected sub-districts and facilities. The estimation approach assumed shared utilisation patterns for benchmark/target interventions (providing denominators and numerators, respectively), and was calibrated and evaluated using household survey data. Results Region-level coverage estimates of 90.3% and 92.4% were obtained for second (Penta2) and third doses of the pentavalent vaccine, and 83.8% for the first dose of the measles-rubella vaccine. The variability and proportion of implausible (> 100%) coverage estimates increased through lower levels; the Penta2 district-level median was 93.6% (IQR:87.3%-96.2%) and facility-level median was 96.5% (IQR:92.0%-104.5%). At all levels, diarrhoea service utilisation estimates were lower than the survey-derived estimate; the district-level median was 7.7% (IQR:5.9%-9.6%) and facility-level median was 2.3% (IQR:0.0%-5.5%). Conclusions Whilst this approach is useful for uncovering local variation in service uptake, integrating routinely collected health data with alternative approaches that explicitly recognise inter-facility competition and spatial/aspatial determinants of health-seeking may produce more accurate estimates, better informing facility-level resource allocation. Health sciences/Health care/Public health Health sciences/Diseases/Infectious diseases Health sciences/Health care/Disease prevention Figures Figure 1 Figure 2 Figure 3 Figure 4 Plain English Summary Inadequate supply of medicines and other resources limits uptake of essential health services in sub-Saharan Africa. Improved population denominators and intervention coverage estimates may better inform the resourcing of health facilities. We evaluated an estimation approach that uses data collected by facilities as a product of routine service delivery but has not yet been implemented at facility-level. Our estimates generally appeared plausible at the highest administrative levels. Although we show that it is possible to apply this estimation approach at sub-district and facility-levels, the variability and proportion of implausible estimates increased at facility-level especially. To inform facility-level resourcing, these data should be integrated with alternative estimation approaches that can account for additional determinants of health-seeking and service utilisation. Background Sustainable Development Goal 3.8 commits countries to Universal Health Coverage (UHC), a multifaceted concept encompassing the quality, affordability and accessibility of health services [ 1 ]. The latter has led many countries of sub-Saharan Africa (SSA) to take steps to bring services closer to communities. Recent policies from Kenya [ 2 ] and Zambia [ 3 ] aimed to increase the proportion of the population living within 5km of a health facility (HF), whilst Rwandan policy aimed to reduce health-seekers’ walking time to the nearest HF to 45 minutes [ 4 ]. In Ghana, the Community-Based Health Planning and Services (CHPS) initiative, introduced in 2005 [ 5 ], has densified the HF network, in rural areas especially, by establishing a new service tier at the base of the primary care hierarchy. By 2021, 90.7% of residential structures were within 5km of the nearest HF [ 6 ]. Though this has alleviated geographic barriers to access, the readiness of CHPS ‘compounds’ to provide primary care services has often fallen short of higher-tier HFs, with inadequate and/or unreliable supply of essential medicines singled out as a specific problem [ 7 ]. Realising the coverage gains from investments in physical infrastructure thus also depends on HFs being adequately resourced to meet the demand placed upon them. In Ghana, despite significant progress since introduction of the Expanded Programme on Immunisation, complete coverage of recommended vaccines amongst children aged 12–23 months has not yet been achieved [ 8 ]. Recent systematic reviews have highlighted the potential for vaccine stockout at HFs, amongst other factors, to impede further progress in SSA [ 9 , 10 ]. The data collected by HFs as a product of routine service delivery may have a pivotal role in this regard. Using these routinely collected health data (RCHD) to generate reliable coverage estimates at the level of individual HFs could help decision-makers to identify and target communities not yet reached by services, but also inform need-based procurement and allocation of vaccines across the HF network [ 10 , 11 ]. Moreover, the range of indicators captured by RCHD might permit development of cross-cutting methods to help decision-makers address multiple population health challenges. Whilst diarrhoeal diseases persist amongst the leading causes of child mortality in SSA [ 12 , 13 ], a substantial proportion of these deaths could be prevented by universal coverage of oral rehydration solution (ORS), a low-cost and highly effective treatment for dehydration [ 14 ]. Yet despite its benefits, and adoption into many national health policies [ 15 ], utilisation remains low: treatment was estimated from recent survey data to have occurred in just 49.1% of acute diarrhoeal episodes amongst children aged under 5 years (u5s) in 30 countries of SSA, rising to 60.8% in Ghana [ 16 ]. Like vaccines, non-availability of ORS packets in HFs has been cited as a supply-side barrier to utilisation [ 17 ] that might be alleviated by using improved demand forecasts to guide resource allocation [ 18 ]. Implicitly, intervention coverage estimation requires a denominator (the number eligible for/targeted by the service) and numerator (the number of users) [ 19 ]. Traditional demographic data sources, including censuses and vital registration systems, are often incomplete, unreliable and outdated in low- and middle-income countries (LMICs), prompting a recent trend towards deriving the denominator (in addition to the numerator) from RCHD [ 20 ]. Though historically under-utilised in SSA, the view that RCHD could contribute to health system management and population health improvement has gained traction, precipitating renewed efforts to improve data quality (DQ) [ 21 , 22 ]. Meanwhile, more than 70 LMICs, predominately in SSA and Asia, have adopted District Health Information Software (DHIS2) for collection, reporting and analysis of RCHD [ 23 ]. This platform was implemented in Ghana in 2012 as the District Health Information Management System (DHIMS2). In 2017 Maina and colleagues proposed a data-driven approach to denominator and intervention coverage estimation using RCHD extracted from DHIS2 [ 24 ]. Briefly, this approach is predicated on the assumption that the population targeted by an intervention can be inferred from the actual number of service events for a related ‘benchmark’ intervention considered to have near-universal (> 90%) coverage, and thus near-complete RCHD. Following this logic, the number of benchmark intervention service events can be extracted from DHIS2 and calibrated using contemporaneous household survey data to obtain the ‘target’ intervention denominator. The number of target intervention service events is also extracted as the numerator for the coverage calculation. Initially applied in first-level administrative units of Kenya, this approach has since been used to measure childhood vaccination coverage in first- and/or second-level units (in Ghana, regions and districts, respectively) elsewhere in SSA [ 25 – 27 ]. On the whole, implementation studies suggest this is a promising approach to denominator and coverage estimation at these administrative levels and that, in this application, RCHD may outperform traditional demographic data sources. To our knowledge, however, there are few published examples of implementation at a lower administrative level [ 28 ], and none in individual HFs. We aimed to evaluate Maina and colleagues’ approach as a means of estimating HF-level coverage of essential child health interventions in a subnational area of Ghana. The first dose of the pentavalent vaccine (Penta1) is typically received by Ghanaian children 6 weeks after birth [ 29 ] and was used as the benchmark intervention. We estimated denominators and coverage/utilisation of four target interventions (three preventive and one curative) selected to cover the first year of life: the second (Penta2) and third (Penta3) doses of the pentavalent vaccine, first dose of the measles-rubella vaccine (MRV1), and outpatient diarrhoea service utilisation amongst children aged under 1 year (u1s). In Ghana, these vaccinations are typically received 10 weeks, 14 weeks and 9 months after birth, respectively [ 29 ]. Coverage/utilisation was estimated at four administrative levels, culminating with individual HFs. Methods Study setting As at the 2021 census, Ghana was subdivided into 16 regions and 261 districts [ 30 ]. Public sector healthcare is provided by Ghana Health Service (GHS) which, to facilitate service delivery and accessibility, further subdivides districts into sub-districts, then CHPS zones. The latter, the smallest unit of service delivery, are centred on specific communities and encapsulate populations of up to 5,000 persons [ 31 ]. Following this administrative structure, regional hospitals deliver secondary care services whilst district-centric networks of hospitals, health centres, clinics and CHPS compounds deliver primary care services. Within HFs, patient-level data collected during routine service events are aggregated by month, manually transcribed to paper-based forms and registers, then entered to DHIMS2. Ghanaian policy requires all public, private and faith-based HFs to report RCHD to DHIMS2 [ 32 ]. This ecological study was conducted in Ghana’s Eastern region (Fig. 1 ), which was subdivided into 33 districts and had a total population of 2,925,653 persons at the 2021 census [ 30 ]. Coverage/utilisation was estimated for: (i) the Eastern region as a whole; (ii) all individual districts therein; (iii) sub-districts of three ‘study’ districts (Atiwa West, Denkyembour and Kwaebibirem), and (iv) individual HFs within these study districts. Sub-districts and HFs were anonymised in all results. Shapefiles depicting the national boundaries of countries of West and Central Africa and national and subnational (region and district) boundary locations of Ghana were sourced from the Humanitarian Data Exchange under a Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Sub-district boundary locations and health facility point locations were provided by the Ghana Health Service Data and other resources Standard Stata scripts for district-level implementation of Maina and colleagues’ approach are hosted by the Countdown to 2030 Collaboration [ 33 ] and were adapted for this study. The 2017/18 Ghana Multiple Indicator Cluster Survey (GMICS) [ 34 ] was contemporaneous with the study period (January 2017 to December 2020) and used to calibrate and evaluate our implementation. All analyses used Stata SE Version 16.0 (Stata-Corp, College Station, TX, USA) and R v4.4.1 with RStudio v2024.04.2 + 764 (R Core Team, Vienna, Austria). The results are reported in accordance with the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) statement [ 35 ]. RCHD describing the health indicators used for this study (Table 1 ) were extracted from DHIMS2 and provided by GHS. The study team did not have direct access to the DHIMS2 database. We estimated coverage/utilisation for the 2020 calendar year, taking individual HFs as the lowest-level units of analysis. The DHIMS2 extract was aggregated by month, health indicator and HF, but also identified the sub-district and district within which each was situated. HFs were retained for analysis where their location was known and DHIMS2 provided evidence that they were operational and delivering relevant services throughout 2017–2020. Having divided the DHIMS2 extract into eight discrete intervals of six months’ duration (January to June and July to December of each calendar year), HFs were retained where one or more service events were recorded in all intervals for each of: Penta1, all-cause outpatient attendances (all ages combined), and outpatient attendances for diarrhoeal disease (all ages combined). After excluding HFs failing to meet these criteria ( Supplementary Figure S1 ), the HF-level data were reaggregated by sub-district, district and for the Eastern region as a whole to enable coverage/utilisation estimation at each administrative level. Sub-districts with no HFs meeting the selection criteria were excluded from all analyses, and some sub-districts were grouped (without crossing district boundaries) to ensure that units at this administrative level contained at least three HFs. All districts contained at least three HFs meeting the selection criteria. Table 1 Health indicators extracted from DHIMS2 and used to implement Maina and colleagues’ method Health indicator Purpose Pentavalent vaccination, 1st dose Benchmark intervention; data quality assessment Pentavalent vaccination, 2nd dose Target intervention; data quality assessment Pentavalent vaccination, 3rd dose Target intervention; data quality assessment Measles-rubella vaccination, 1st dose Target intervention; data quality assessment Diarrhoeal disease outpatient attendances: less than 28 days of age Target intervention; data quality assessment Diarrhoeal disease outpatient attendances: 1–11 months of age Target intervention; data quality assessment Diarrhoeal disease outpatient attendances: 1–4 years of age Data quality assessment Diarrhoeal disease outpatient attendances: total for all age groups combined Data quality assessment All-cause outpatient attendances: less than 28 days of age Data quality assessment All-cause outpatient attendances: 1–11 months of age Data quality assessment All-cause outpatient attendances: 1–4 years of age Data quality assessment All-cause outpatient attendances: total for all age groups combined Data quality assessment All-cause inpatient admissions: less than 28 days of age Data quality assessment All-cause inpatient admissions: 1–11 months of age Data quality assessment All-cause inpatient admissions: 1–4 years of age Data quality assessment All-cause inpatient admissions: total for all age groups combined Data quality assessment Antenatal care, 1st visit Data quality assessment Antenatal care, at least 4 visits Data quality assessment Bacillus Calmette-Guerin vaccination Data quality assessment Caesarean sections Data quality assessment Deliveries in health facilities Data quality assessment Family planning, new visits/acceptors Data quality assessment Family planning, revisits Data quality assessment Intermittent preventive treatment for malaria in infants, 2nd dose Data quality assessment Postnatal care within 48 hours after delivery Data quality assessment Stillbirths (fresh) Data quality assessment Stillbirths (macerated) Data quality assessment Total number of deaths in health facilities amongst children under 5 years of age Data quality assessment Total number of maternal deaths in health facilities Data quality assessment Data quality assessment and adjustment The quality of DHIMS2 data in 2020 was assessed using standard DQ checks from the Countdown scripts ( Supplementary Table S3 ). All were performed at region-/district-levels and are described, with results, in the supplementary file . An outline of the adjustments applied based on these checks follows. A reporting completeness check and adjustment preceded all others. This invoked equation [ 1 ], where \(\:n\) represents the number of service events for a health indicator, \(\:c\) the observed reporting rate, and \(\:k\) the adjustment factor. If incomplete reporting is assumed to indicate service non-provision then \(\:k=0\) , but if it is assumed that the service was provided at a lower volume than HFs with complete reporting, \(\:k\) is set to a value between 0 and 1. As a high level of reporting completeness was observed ( Supplementary Table S5 ), the impact of this adjustment was considered negligible and a default \(\:k\) -value of 0.25 assumed for all services. \(\:{n}_{adjusted}={n}_{reported}+{n}_{reported}\text{*}\left(\frac{1}{c}-1\right)\text{*}k\) [1] Missing values were imputed using the median of non-missing values for the same administrative unit/health indicator/year combination. After checking internal consistency over time, outliers from 2020 (reported values exceeding five median absolute deviations above/below the 2017–2019 median) were adjusted by imputing the 2017–2019 median for the same administrative unit/health indicator combination. Denominator estimation To calibrate the denominator, the DQ-adjusted number of Penta1 service events during 2020 was first uplifted for non-utilisation using the GMICS estimate of 4.8% in the Eastern region. The result was assumed to represent the total number of u1s eligible for Penta2 and Penta3, and was used as the denominator for these target interventions without further adjustment. For MRV1, the result was further adjusted downwards to account for post-neonatal mortality. No adjustment for neonatal mortality was made as children were assumed to have received Penta1 after the neonatal period, at 6 weeks after birth. In the absence of region-specific mortality rates, the national GMICS estimate of 14 post-neonatal deaths per 1,000 live births was used. The same adjustment was applied in respect of diarrhoea service utilisation amongst u1s. Coverage/utilisation estimation To estimate coverage/utilisation of the target intervention, DQ-adjusted numerators from 2020 were divided by the appropriate denominator. As the denominator for diarrhoea service utilisation relates to the number of u1s eligible for the service but DHIMS2 only reports the number of service events taking place amongst this population, an additional step was required to obtain a numerator expressed in terms of service users. The number of diarrhoeal service events amongst children aged less than 28 days or 1–11 months (Table 1 ) was summed, then divided by an episode rate of 1.82 episodes per child-year. This rate had been estimated by the 2016 Global Burden of Disease study for u5s in Ghana [ 36 ]. The result was then divided by the denominator to generate utilisation estimates. Evaluation All DHIMS2-derived coverage/utilisation estimates were compared to survey-derived estimates for the Eastern region, which were recalculated (with 95% confidence intervals) from GMICS microdata using the ‘survey’ R package [ 37 ]. Differences between the surv5ey-derived estimate for the Eastern region and DHIMS2-derived estimates for HFs, sub-districts, districts and the region were used to compute ‘mean absolute deviation’ for each administrative level. Though the computation was analogous to mean absolute error, this comparison was taken as demonstrating the variability of DHIMS2-derived estimates through successive administrative levels as opposed to estimation ‘error’. To provide an assessment of their consistency across indicators (within administrative levels), deviation values were further standardised as a percentage of the survey-reported estimate and used to calculate Spearman’s rank correlation coefficients. DHIMS2-derived estimates for HFs within the study districts were examined for evidence of spatial autocorrelation using the ‘spdep’ R package [ 38 ]. The Global Moran’s I statistic [ 39 ], a summary measure of autocorrelation, was separately calculated for each indicator with the neighbourhood structure of HFs defined using the Euclidean distances separating each from its eight nearest neighbours. Distances were converted to a row-standardised spatial weights matrix assigning greater importance to neighbouring estimates. Moran’s scatterplots were produced to visualise the extent to which the global statistic accurately reflected local spatial patterns [ 40 ]. Each scatterplot was divided into quadrants by splitting x- and y-axes on the means of coverage/utilisation and spatial lag (the weighted average of neighbouring observations), respectively. The Global Moran’s I statistic and 95% confidence interval were also plotted. Estimates exerting disproportionate influence on the global statistic were identified using regression diagnostics including Cook’s distance and covariance ratios [ 41 ], and highlighted on the scatterplots and HF location maps. Ethics Ethical approval was obtained from the GHS Ethics Review Committee (ID:001/03/23) and Ethics Committee of the Faculty of Environmental and Life Sciences, University of Southampton (ID:78697). Results Overall, of 1,084 Eastern region HFs whose location was known, 403 (37.2%) were retained for analysis (Table 2 ). Of these, CHPS compounds were most prevalent (65.0%), followed by health centres (27.5%) and hospitals (5.7%). The 49 HFs retained from the three study districts comprised CHPS compounds (73.4%), health centres (20.4%) and hospitals (6.1%) only. Most of the 681 Eastern region HFs excluded by the selection algorithm ( Supplementary Figure S1 ) were non-operational through all, or part, of 2017–2020 and, together, accounted for a minor proportion of recorded service events. First, 468 were public sector/faith-based HFs that had not reported data to DHIMS2 during the first and second intervals and thus considered non-operational at the start of this period. Second, 143 public sector/faith-based HFs considered operational at the start of the period but failing to report data to DHIMS2 during one or more intervals were excluded. Of these, 29 had not reported data to DHIMS2 during any of the final three intervals, suggesting that they may have ceased to operate mid-period. The remaining 114 HFs were likely operational throughout, but excluded owing to short-term periods of incomplete reporting. Finally, all 70 private sector HFs were excluded as none had reported data to DHIMS2. After exclusions, one sub-district from the three study districts no longer contained any HFs and was also excluded from further analysis. The 20 remaining sub-districts were grouped where necessary (without crossing district boundaries) to produce 12 administrative units at this level, each containing at least three HFs. Table 2 The number and distribution of valid health facility locations received from Ghana Health Service and retained for analysis, by type and ownership Health facility group Type of health facility Number (row %) Total (column %) Number (%) excluded a Public sector Faith-based Private sector Unknown Eastern region Health facilities with valid locations Hospital 23 (52.3) 5 (11.4) 15 (34.1) 1 (2.3) 44 (4.1) - Clinic 10 (18.5) 4 (7.4) 33 (61.1) 7 (13.0) 54 (5.0) - Health centre 123 (80.9) 10 (6.6) 6 (3.9) 13 (8.6) 152 (14.0) - CHPS compound 795 (97.7) 2 (0.2) 0 (0.0) 17 (2.1) 814 (75.1) Maternity clinic 3 (15.0) 0 (0.0) 16 (80.0) 1 (5.0) 20 (1.8) - All types 954 (88.0) 21 (1.9) 70 (6.5) 39 (3.6) 1,084 (100.0) - Health facilities retained by the selection algorithm Hospital 18 (78.3) 4 (17.4) 0 (0.0) 1 (4.3) 23 (5.7) 21 (47.7) Clinic 2 (28.6) 4 (57.1) 0 (0.0) 1 (14.3) 7 (1.7) 47 (87.0) Health centre 94 (84.7) 8 (7.2) 0 (0.0) 9 (8.1) 111 (27.5) 41 (27.0) CHPS compound 255 (97.3) 0 (0.0) 0 (0.0) 7 (2.7) 262 (59.1) 552 (67.8) Maternity clinic 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 20 (100.0) All types 369 (91.6) 16 (4.0) 0 (0.0) 18 (4.5) 403 (100.0) 681 (62.8) Study districts Health facilities retained by the selection algorithm Hospital 2 (66.7) 1 (33.3) 0 (0.0) 0 (0.0) 3 (6.1) 0 (0.0) Clinic 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 1 (100.0) Health centre 8 (80.0) 0 (0.0) 0 (0.0) 2 (20.0) 10 (20.4) 2 (16.7) CHPS compound 34 (94.4) 0 (0.0) 0 (0.0) 2 (5.6) 36 (73.4) 25 (41.0) Maternity clinic 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) - All types 44 (89.8) 1 (2.0) 0 (0.0) 4 (8.2) 49 (100.0) 28 (36.4) a The number (%) of excluded health facilities was derived by comparison to the 1,084 valid health facility locations received from Ghana Health Service (see also Supplementary Figure S1 ) DHIMS2-derived Penta2 coverage estimates for the region (90.3%) and most districts (median:93.6%, IQR:87.3%-96.2%) closely approximated the survey-derived estimate (89.8%, 95%CI:82.9%-96.7%) (Table 3 ). Penta3 estimates for the region (92.4%) and most districts (median:96.2%, IQR:87.8%-100.0%) exceeded the survey-derived estimate and upper confidence limit (80.1%, 95%CI:68.7%-91.5%) (Fig. 2 ), suggesting that the DHIMS2-derived denominator was too low, yielding a coverage over-estimate. MRV1 followed a similar pattern to Penta2, notwithstanding greater difference between DHIMS2-derived (83.8%) and survey-derived estimates (81.3%, 95%CI:72.6%-90.0%) for the region, and wider distribution of district-level DHIMS2-derived estimates (median:87.8%, IQR:80.2%-94.2%) (Fig. 2 ). Estimates of diarrhoea service utilisation for the region (7.8%) and most districts (median:7.7%, IQR:5.9%-9.6%) fell below the survey-derived estimate and lower confidence limit (15.4%, 95%CI:8.7%-22.1%) (Fig. 2 ), suggesting that the DHIMS2-derived denominator was too high, yielding an under-estimate. The distribution of DHIMS2-derived estimates appeared to widen amongst sub-districts and HFs (Fig. 3 ), with several vaccination coverage estimates exceeding 100% at these lower levels. Estimates of Penta2, Penta3 and MRV1 coverage exceeded 100% in 34.7%, 42.9% and 28.6% of HFs (Table 3 ), respectively. In addition, DHIMS2-derived estimates tended to deviate from survey-derived equivalents by an increasing margin through successive administrative levels (Table 3 ). Table 3 Comparisons of survey-derived and DHIMS2-derived coverage/utilisation estimates, by health indicator and administrative level Administrative level Estimate b, c Penta2 Penta3 MRV1 Diarrhoea service utilisation Eastern region Survey-derived estimate (95% CI) 89.8 (82.9–96.7) 80.1 (68.7–91.5) 81.3 (72.6–90.0) 15.4 (8.7–22.1) DHIMS2-derived estimate 90.3 92.4 83.8 7.8 Deviation 0.4 12.3 2.5 7.6 Districts (all) (n = 33) Median (IQR) DHIMS2-derived estimate 93.6 (87.3, 96.2) 96.2 (87.8, 100.0) 87.8 (80.2, 94.2) 7.7 (5.9, 9.6) Mean absolute deviation 5.6 14.1 11.7 7.4 Number (%) units with coverage estimate > 100% 2 (6.1) 9 (27.2) 4 (12.1) - Districts (study) a (n = 3) Median (IQR) DHIMS2-derived estimate 96.2 (94.9, 96.3) 98.9 (97.2, 99.9) 84.5 (83.9, 86.8) 4.5 (4.1, 5.2) Mean absolute deviation 5.6 18.4 4.4 10.6 Number (%) units with coverage estimate > 100% 0 (0.0) 1 (0.3) 0 (0.0) - Sub-districts a (n = 12) Median (IQR) DHIMS2-derived estimate 95.6 (94.7, 97.3) 98.7 (97.1, 102.0) 88.1 (83.1, 98.2) 4.3 (3.0, 6.2) Mean absolute deviation 6.9 20.7 11.6 10.4 Number (%) units with coverage estimate > 100% 2 (16.7) 4 (33.3) 2 (16.7) - Health facilities a (n = 49) Median (IQR) DHIMS2-derived estimate 96.5 (92.0, 104.5) 97.5 (89.3, 106.3) 93.1 (80.6, 103.0) 2.3 (0.0, 5.5) Mean absolute deviation 11.0 21.5 17.0 12.1 Number (%) units with coverage estimate > 100% 17 (34.7) 21 (42.9) 14 (28.6) - a Rows show estimates for administrative units located within the three study districts only (Atiwa West, Denkyembour and Kwaebibirem districts) b Survey-derived estimates for the Eastern region were recalculated (with 95% confidence intervals) using microdata from the 2017/18 Ghana Multiple Indicator Cluster Survey c Deviation is the difference between the survey-derived estimate for the Eastern region and DHIMS2-derived estimates for each administrative unit, and was used to compute mean absolute deviation for each administrative level Data points represent DHIMS2-derived estimates for districts of the Eastern region The solid red vertical line represents the survey-derived coverage/utilisation estimate and the red shaded area its associated 95% confidence interval The solid blue vertical line represents the DHIMS2-derived coverage/utilisation estimate for the Eastern region as a whole The solid grey vertical line (shown on the vaccination coverage plots only) is placed at 100% coverage Data points represent DHIMS2-derived estimates for the three study districts of the Eastern region (grey points) and all sub-districts (dark blue points) and individual health facilities (light blue points) therein The solid red vertical line represents the survey-derived coverage/utilisation estimate and the red shaded area its associated 95% confidence interval The solid blue vertical line represents the DHIMS2-derived coverage/utilisation estimate for the Eastern region as a whole The solid grey vertical line (shown on the vaccination coverage plots only) is placed at 100% coverage Penta2, Penta3 and MRV1 deviation values were moderately/strongly correlated at all administrative levels (Table 4 ), suggesting that their direction and magnitude was largely consistent across these indicators. The same pattern is discernible from district-level dot plots (Fig. 2 ), which indicate, for example, over-estimation for all vaccinations in Okere district, but under-estimation in Nsawam-Adoagyiri district. Deviation values for diarrhoea service utilisation were not correlated with any other indicator. Table 4 Correlation of deviation values for all districts of the Eastern region and all sub-districts and health facilities of the three study districts, by health indicator Administrative level Health indicator Penta2 Penta3 MRV1 Diarrhoea service utilisation Districts (all) (n = 33) Penta2 - 0.677 *** 0.578 *** 0.094 Penta3 0.677 *** - 0.834 *** 0.172 MRV1 0.578 *** 0.834 *** - 0.284 Diarrhoea service utilisation 0.094 0.172 0.284 - Sub-districts a (n = 12) Penta2 - 0.601 ** 0.874 *** 0.028 Penta3 0.601 ** - 0.308 -0.042 MRV1 0.874 *** 0.308 - 0.308 Diarrhoea service utilisation 0.028 -0.042 0.308 - Health facilities a (n = 49) Penta2 - 0.617 *** 0.559 *** -0.015 Penta3 0.617 *** - 0.467 *** 0.066 MRV1 0.559 *** 0.467 *** - 0.094 Diarrhoea service utilisation -0.015 0.066 0.094 - Deviation values were standardised by conversion to a percentage of the survey-derived estimate for the Eastern region. Spearman’s rank correlation coefficient was then calculated across health indicators and within administrative levels : *** correlation significant at < 0.01 level; ** correlation significant at < 0.05 level; * correlation significant at < 0.1 level a Rows show estimates for administrative units located within the three study districts only (Atiwa West, Denkyembour and Kwaebibirem districts) The Global Moran’s I statistic suggested a slight tendency towards negative spatial autocorrelation for each indicator, but failed to reach statistical significance. Still, the presence of influential points in the upper-left and lower-right quadrants of each scatterplot (Fig. 4 ) indicates that the global statistic obscures low–high and high–low clustering; that is, instances of HFs falling below or exceeding the mean coverage/utilisation and spatial lag but surrounded by neighbours with dissimilar estimates. The location maps confirmed that these points pertained to HFs with vaccination coverage estimates substantially exceeding 100% but whose closest neighbours had lower estimates, or vice versa (Fig. 4 ). Shapefiles depicting district boundary locations were sourced from the Humanitarian Data Exchange under a Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Health facility point locations were provided by the Ghana Health Service Points on the maps indicate the locations of health facilities across the study districts and are shaded by their DHIMS2-derived coverage/utilisation estimate. Points on the scatterplots represent individual health facilities and are plotted according to their coverage/utilisation estimate and spatial lag value. Any points found to exert a disproportionate influence on the Global Moran’s I statistic are highlighted on both the map and scatterplot Discussion This study aimed to estimate subnational coverage/utilisation of child health interventions in Ghana’s Eastern region and, to our knowledge, is the first to implement Maina and colleagues’ approach at the level of individual HFs. Vaccination coverage estimates appeared plausible at region- and district-levels but frequently exceeded 100% and deviated from survey-derived equivalents by an increasing margin through lower levels. This may reflect local variation in vaccination uptake, but patterns of over-/under-estimation across proximate subregions of Uganda [ 26 ] and districts of Sierra Leone [ 27 ] were largely attributed by previous implementation studies to cross-boundary health-seeking causing actual service utilisation levels to depart from the denominator. The implicit assumption that health-seekers will not cross administrative boundaries is less likely to hold in smaller units, particularly HF catchment areas. Bypassing, or travel beyond the nearest HF, to access maternal/child health services has been observed across Ghana and SSA; besides spatial determinants, including the number, location and convenience of HFs [ 42 – 44 ], some health-seekers may favour higher-tier HFs [ 45 , 46 ] or eschew those perceived as falling short of an expected standard [ 45 , 47 ]. In this way some HFs may possess aspatial qualities or characteristics that make them more or less attractive to health-seekers, motivating bypassing. Maina and colleagues’ approach arguably accounts for competition between proximate HFs indirectly in that DHIMS2-derived numerators and denominators implicitly capture the health-seeking choices made by service users. Still, the presence of low–high/high–low clustering amongst HF-level coverage estimates suggests that the factors underlying these choices may not always be shared between benchmark/target interventions, with potential to induce numerator/denominator mismatch. The finding that deviation values for Penta2, Penta3 and MRV1 were moderately/strongly correlated across HFs may point to a difference between Penta1 and subsequent vaccinations, in particular. One scenario that might give rise to such a trend in urban areas is that women returning to formal employment after maternity leave, for which Ghana’s standard entitlement is three months [ 48 ], may need to share caring duties between family members or take children to HFs nearer the home or workplace. Whilst Maina and colleagues’ approach appears relatively robust to differences of this nature at the highest levels of aggregation, our findings suggest that its estimates become more unstable at the level of individual HFs especially. Diarrhoea service utilisation was found to be under-estimated at all administrative levels. As preventive interventions, including childhood vaccinations, are targeted to all members of the eligible population, a relatively large number of service events can be expected to take place within even the lowest-level administrative units. Curative interventions, however, are only utilised by persons experiencing and receiving treatment for a specific health need: according to the 2017/18 Ghana Multiple Indicator Cluster Survey (GMICS), diarrhoea prevalence was 15.4% (95%CI:8.7%-22.1%) amongst u1s in the Eastern region, with treatment-seeking to public/private HFs occurring in just 46.5% (95%CI:19.9%-73.1%) of these episodes. As such, while data released by Ghana Health Service (GHS) consistently placed diarrhoea amongst the top five reasons for all-ages outpatient attendance from 2002–2016 [ 49 ], the number of u1s presenting to HFs, and thus providing the numerator for the utilisation calculation, was considerably lower. Indeed, we found that low numerator values yielded utilisation estimates close to 0% in several HFs. The estimation process was further complicated by the numerator and denominator describing different quantities: whilst the Penta1-derived denominator represented the estimated number of u1s eligible for the diarrhoea service, the numerator extracted from DHIMS2 quantified service events amongst this population. To obtain a numerator expressed in the same unit as the denominator we divided the number of service events by an episode rate that, being estimated for u5s nationally as of 2016, was not necessarily appropriate for the time period or target population. Overall, the assumptions associated with establishing both numerator and denominator appear to have produced estimates that cannot reliably guide the resourcing of HFs for this curative intervention, suggesting that it may be better served by an approach that does not presuppose a relationship between benchmark/target interventions. Still, a Penta1-derived denominator may have relevance to preventive interventions designed to reduce infectious disease incidence in this population. A randomised controlled trial conducted in Ghana’s Volta region suggested that health education delivered to caregivers at regular home visits led to increased frequency of handwashing and other behaviours that reduce children’s infectious disease risk [ 50 ]. Denominator estimates generated using this approach could thus contribute to the resourcing of HFs as a base for health promotion and education programmes delivered to local communities. Refining the approach for implementation amongst individual health facilities Previous implementation studies typically used vaccinations delivered during neonatal or early post-neonatal periods as a benchmark for subsequent child health interventions. Maina and colleagues recommended consulting contemporaneous household survey data to verify that the prospective benchmark intervention has minimum coverage of 90%, but our analysis suggests that this criterion alone is insufficient for HF-level implementation. As an example, Ghanaian children typically receive the Bacillus Calmette-Guerin (BCG) vaccination at birth [ 29 ] and, though its GMICS coverage estimate was 91.8% in the Eastern region, service events were most commonly reported to DHIMS2 by HFs that had also reported childbirths. As such, while BCG may provide a viable benchmark at higher levels of aggregation [ 26 ], the same may not be true for HFs that are not mandated to deliver childbirth services, such as CHPS compounds [ 51 ]. Still fewer benchmark interventions may be available in CHPS zones that lack a built compound, but where a limited package of services is being delivered in a mobile capacity. We therefore recommend that HF-level estimation be preceded by an additional review of DHIS2 to ensure that the benchmark intervention was delivered by HFs at all primary care tiers through the time period of interest. Consideration should also be given to the calibration adjustments associated with translating the benchmark to a denominator. Having used Penta1 as the benchmark, our DHIMS2-derived denominator was calibrated to adjust for non-utilisation. In the absence of sub-regional equivalents, we applied the GMICS rate of Penta1 non-utilisation for the Eastern region to all administrative units as a constant adjustment factor. This disregards sub-regional variation in vaccination uptake, however, which may be significant at the most granular geographies [ 52 ]. Equally important are limitations and DQ issues associated with the survey, which may also exhibit a geographic pattern [ 53 ]. As an example, vaccination uptake may be incorrectly recorded if non-availability of a child’s vaccination card necessitates caregiver self-report, which may be subject to recall or other biases [ 54 ]. Some implementation studies have explored using first antenatal care visits as a benchmark for child health interventions [ 27 , 28 ]. This was ruled out for the present study, however, as establishing this benchmark/target link entails a longer sequence of survey-informed calibration adjustments (including stillbirth, twinning and mortality rates), each of which may obscure distinct patterns of geographic variation and create more opportunity for inaccuracy to enter the process. We also recommend, therefore, that users working at HF-level select a benchmark intervention that minimises the number and impact of calibration adjustments. We also evaluated DHIMS2-derived estimates by comparison with survey-derived equivalents for the Eastern region. Although their agreement appeared to diminish through lower administrative levels, this comparison also disregarded legitimate sources of sub-regional variation. Diarrhoeal risk, for example, is highly influenced by the prevailing meteorological and environmental conditions, such that survey-derived prevalence and/or treatment-seeking estimates (which utilise a two-week reference period) may differ markedly from the annual average in some locations [ 55 ]. Moreover, diarrhoeal risk has been shown to increase in Ghana’s peri-urban districts, which are often characterised by inadequate sanitation and close living conditions [ 56 ]. Even within a single region, our DHIMS2-derived estimates of diarrhoea service utilisation varied widely at district- (range:2.8%-17.1%) and HF-levels (range:0.0%-24.5%). Modern geostatistical methods incorporating spatially varying environmental and socioeconomic covariates have been used to further disaggregate survey-derived infectious disease incidence [ 57 ] and vaccination coverage [ 52 ] estimates from SSA, and may provide more appropriate comparators for sub-regional DHIMS2-derived estimates. The additional methodological complexity may be offset if offering a clearer picture of the performance of Maina and colleagues’ approach at these levels. Notably, as HFs are rarely enclosed within unambiguous administrative boundaries in SSA, disaggregation to this level would necessitate a preliminary step to delineate catchment areas [ 58 ], although using survey-derived comparators for sub-districts, or even districts, would still represent a valuable advance. Still, as survey organisations work to sometimes lengthy release cycles, the absence of an iteration coinciding with the estimation year may also limit the extent to which survey-derived metrics should be considered a ‘gold standard’ for calibration and evaluation. We indirectly accounted for health system reconfiguration over time by way of our HF selection algorithm but were unable to account for other contextual changes with implications for the agreement of DHIMS2-derived and survey-derived estimates, such as population movement or the inter-annual epidemiology of diarrhoea. Scaling estimation to national-level health facility networks Several HF-level estimation approaches have been explored in SSA, but have largely been implemented in a single or small number of HFs, thereby facilitating access to data that are not always accessible to decision-makers in this setting [ 59 ]. This is key, for deployment throughout the national HF network depends upon ease of implementation by way of data structures that are consistent and common to all. A clear advantage of Maina and colleagues’ approach, therefore, is its emphasis upon data generated via pre-existing and nationally consistent administrative processes and made readily available to decision-makers at all levels of health system management via DHIS2. In keeping with the trend towards health sector decentralisation in LMICs [ 60 ], this approach could, in principle, be administered by District Health Management Teams (DHMTs) across the country to inform resource allocation and evidence-based decision-making at the lowest levels of service delivery. Having used RCHD extracted from DHIMS2 and experienced no computational difficulties with HF-level implementation across three districts of Ghana, we have no reason to foresee any technical impediment to replication by DHMTs. This arrangement could further enhance the scalability of the approach by affording decision-makers the opportunity to tailor its implementation to the local context. As an example, whilst its denominator estimates are derived using actual service utilisation levels from the recent past, interventions to increase treatment-seeking for childhood diarrhoea are being trialled in many settings [ 18 ]. Indeed, a study undertaken in three regions of Ghana found that efforts to improve the supply of ORS and promote its use had increased utilisation over a three-year period [ 61 ]. To guard against under-estimation, and therefore under-resourcing of local HFs, in this scenario, the anticipated impact of relevant interventions could be factored into the approach as an additional adjustment. Implications of data quality and completeness Our assessment indicated that DHIMS2 DQ was generally of a high standard in 2020 ( supplementary file ). This finding was, in part, an artefact of our selection criteria being linked to reporting completeness, but confirms that much progress has been achieved in Ghana. As completeness approached 100% amongst HFs retained for analysis, we adjusted for incomplete reporting by applying a common \(\:k\) -value across all services and administrative units. Yet whilst incomplete reporting may ostensibly point to suboptimal DQ, the same trend may equally arise in lower-level administrative units where intermittent stockouts or staffing issues have disrupted service delivery, or inter-facility competition has distorted health-seeking flows [ 20 ]. Indeed, a DHIS2 DQ assessment from Kenya showed that reporting completeness may vary between HFs, across services and over time, and that a short-term decline during 2017 was linked to a national health worker strike [ 62 ]. To better account for the impact of wider organisational events on local service delivery, there is value in prospective collection of information to contextualise outlying DQ metrics (such as the duration of supply chain or staffing issues), and DHMTs liaising directly with health workers to co-determine more appropriate adjustments for affected HFs. Efforts to involve health workers in decision-making processes may have the further benefit of helping them to recognise the value of high-quality data to broader health system objectives, fostering greater commitment to future DQ improvement initiatives [ 63 , 64 ]. Irrespective of the cause, incomplete reporting may preclude replication of Maina and colleagues’ approach in some HFs. A three-year ‘look-back’ period is used to assess internal consistency over time and adjust outliers. Though less problematic in large-scale administrative units aggregating several HFs, this duration may be prohibitive in smaller units where health system reconfiguration has created discontinuities of service, and therefore data. Our data suggested that 497 public sector/faith-based HFs of the Eastern region were inactive at start and/or end of the study period. At the same time, the exclusion of all private sector HFs indicates that its adherence to mandatory reporting processes was poor. This is not unique to Ghana, however, with a similar contrast between public and private sectors observed elsewhere in SSA [ 65 , 66 ]. Although many LMICs rely upon the private sector to extend vaccination coverage to underserved communities [ 67 ], its contribution cannot be fully appreciated from incomplete or unreliable data, hampering decision-makers’ ability to track progress and close persistent gaps. Inter-facility competition will influence levels of demand throughout the HF network, underlining the importance of complete data for optimal resource allocation. Sustainable policy levers are needed to maintain compliance with national reporting standards: Ghana’s Health Facilities Regulatory Agency licenses HFs to operate and has a role via its accreditation scheme, but there is also merit in DHMTs engaging with private providers directly to build relationships, promote information sharing and demonstrate the mutual benefits of high-quality data capturing the universe of local HFs [ 68 ]. As Maina and colleagues’ approach depends upon the relationship between benchmark/target interventions, the accuracy and internal consistency of related indicators are crucial, and should be a focus of DQ improvement efforts. We found that the volume of Penta3 service events frequently exceeded that of Penta1, resulting in numerator/denominator mismatch and coverage over-estimation, often beyond 100%. Again, deviations from the expectation may arise if the factors underlying HF choice differ across benchmark/target interventions, but selective attachment of coverage targets to certain services may also induce over-reporting or disproportionate focus upon Penta3 in some Ghanaian HFs [ 69 ]. Persistent weaknesses affecting the manual, paper-based data collection and reporting procedures followed by health workers in SSA may also contribute: studies from Uganda and Nigeria, for example, found inconsistencies between physical records and DHIS2 [ 70 , 71 ]. Importantly, DQ improvement is a continual process and may reach a point where, to maximise HF retention, it becomes possible to detach the standard checks from the approach in favour of a generalised assessment programme supported by regular training and monitoring visits to HFs by dedicated staff. Unsupervised machine learning methods have been used with data extracted from DHIS2 as a means of identifying under-performing HFs [ 62 ], and could be leveraged to facilitate DQ assessment across the network. Strengths and limitations As the first to evaluate Maina and colleagues’ approach at HF-level, our study considers issues and limitations not yet raised by previous implementation studies. Though we offer recommendations to address the issues we have observed, our subnational study area may not capture all that might manifest if scaled to national-level. Having used RCHD that are similar in format and content to other DHIS2-adopter countries, we believe that many of our observations concerning DQ and the performance of this estimation approach are likely generalisable both within and beyond Ghana. As previously noted, however, our evaluation cannot provide a conclusive picture of performance at sub-regional levels and would benefit from comparison with survey-derived estimates generated using geostatistical modelling methods. We also recognise potential limitations associated with using RCHD extracted from DHIMS2. Though standard data collection procedures are used in Ghanaian HFs, the accuracy and generalisability of these data may be affected if their interpretation or understanding varies amongst health workers, or if alternative definitional approaches are followed in other countries. As our data extract did not provide the age of children attending vaccination service events, we assumed that all took place according to Ghana’s standard schedule. Yet this may not always be the case, raising the possibility of misclassification bias; vaccination of ineligible children could generate excess reporting in some HFs and may have contributed to our findings around Penta1/Penta3 service volumes. Conclusion Reliable population denominators and intervention coverage estimates could help decision-makers to ensure that HFs are better resourced to meet the demand placed upon them. Maina and colleagues’ estimation approach is predicated upon an assumed relationship between benchmark/target interventions and has shown promise at region- and district-levels. Yet having implemented this approach at lower administrative levels, we found that the variability of vaccination coverage estimates and proportion of implausible estimates (> 100%) increased through district-, sub-district- and HF-levels. The finding that deviations from survey-derived equivalents were correlated at HF-level suggests shared underlying drivers, likely related to inconsistent patterns of service utilisation across benchmark/target interventions. Estimation of diarrhoea service utilisation was further challenged by the additional assumptions required to establish both the numerator and denominator. The organisation and configuration of the local health system may also impose certain restrictions upon HF-level implementation: services that are optional at lower tiers of the primary care hierarchy are unable to provide a viable benchmark, while newly-established or intermittently-reporting HFs may have insufficient longitudinal data. Indeed, a number of HFs were excluded from our study following initial DQ and reporting completeness checks, including all from the private sector. Overall, whilst this approach has potential beneath district-level, we found that estimates became more unstable through lower levels, with implications for their use to guide the resourcing of individual HFs. Incorporating RCHD within alternate approaches that explicitly recognise inter-facility competition and the varied spatial/aspatial determinants of health-seeking and service utilisation, but do not depend upon a benchmark/target relationship, might retain the advantages of Maina and colleagues’ approach whilst addressing some of its shortcomings in respect of HF-level estimation. Abbreviations BCG Bacillus Calmette-Guerin vaccination CHPS Community-Based Health Planning and Services (initiative) DHIMS2 District Health Information Management System DHIS2 District Health Information Software DQ data quality GHS Ghana Health Service GMICS 2017/18 Ghana Multiple Indicator Cluster Survey HF health facility LMIC low- and middle-income country MRV1 first dose of the measles-rubella vaccine ORS oral rehydration solution Penta1 first dose of the pentavalent vaccine Penta2 second dose of the pentavalent vaccine Penta3 third dose of the pentavalent vaccine RCHD routinely collected health data SSA sub-Saharan Africa u1s children aged under 1 year u5s children aged under 5 years UHC Universal Health Coverage Declarations Data availability statement The data describing individual health facilities and routinely collected health data analysed for this study are not publicly available due to confidentiality and data licensing restrictions from the Ghana Health Service. They can be obtained from the Ghana Health Service (https://ghs.gov.gh/) on reasonable request. The 2017/18 Ghana Multiple Indicator Cluster Survey microdata are available from UNICEF (https://mics.unicef.org/). For mapping purposes, shapefiles depicting the national boundaries of countries of West and Central Africa and national and subnational (region and district) boundary locations of Ghana were sourced from the Humanitarian Data Exchange (https://data.humdata.org/). The Humanitarian Data Exchange web platform stipulates that these data are made publicly available under the Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Sub-district boundary locations and health facility point locations were provided directly by the Ghana Health Service and are not publicly available. Standard Stata scripts for implementation of the method utilised for this study are openly available from the Countdown to 2030 Collaboration at https://www.countdown2030.org/tools-for-analysis/health-facility-data-and-analysis. The adapted scripts are available from the corresponding author on reasonable request. Acknowledgements The authors thank Mr Thomas Ankomah of the Ghana Health Service for providing technical advice and assisting with extraction of routinely collected health data from the District Health Information Management System (DHIMS2). The authors also thank Dr Winfred Ofosu, Ghana Health Service Regional Director of Health Services for the Eastern region, for granting permission to access and use routinely collected health data from this region. Author contributions MJ and JW conceptualised and designed the study. MJ analysed the data and wrote the first draft of the manuscript. WD-G and AO facilitated access to and interpretation of data from the Ghana Health Service. All authors provided feedback and edited drafts of the manuscript, and read and approved the final manuscript. Competing interests statement AO, at the time of the study, was the Deputy Director General of Ghana Health Service, which generates and owns the data describing individual health facilities and routinely collected health data analysed for this study and is responsible for the delivery of public health services in Ghana. WDG previously worked with Ghana Health Service as a public health information officer until 2017. Funding statement MJ received funding from the UK Economic and Social Research Council (ESRC) South Coast DTP, grant ID: ES/P000673/1. NM is a recipient of an NIHR Research Professorship award (Ref: RP-2017-08-ST2-008). The funders had no role in the study. For the purposes of open access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. References United Nations. SDG Indicators: Global indicator framework for the Sustainable Development Goals and targets of the 2030 Agenda for Sustainable Development . 2021 [cited 2021 11th August 2021]; Available from: https://unstats.un.org/sdgs/indicators/indicators-list/ . Ministry of Health. Kenya Health Sector Strategic Plan . 2018 [cited 2021 4th June 2021]; Available from: https://www.health.go.ke/wp-content/uploads/2020/11/Kenya-Health-Sector-Strategic-Plan-2018-231.pdf . Ministry of Health. Zambia National Health Strategic Plan 2017–2021 . 2017 [cited 2022 20th June 2022]; Available from: https://www.moh.gov.zm/?page_id=1306 . Ministry of Health, Fourth Health Sector Strategic Plan: July 2018 - June 2024 . 2018, Ministry of Health: Kigali. Ghana Health Service, Community-Based Health Planning and Services (CHPS): The Operational Policy . 2005, Ghana Health Service: Accra. Ghana Statistical Service, Ghana 2021 Population and Housing Census: Volume 2 Proximity of Residential Structures to Essential Service Facilities . 2021, Ghana Statistical Service: Accra. Ayanore, M., et al., Sub-national variations in general service readiness of primary health care facilities in Ghana: Health policy and equity implications towards the attainment of Universal Health Coverage . PLoS One, 2022. 17(6): p. e0269546. Budu, E., et al., Trend and determinants of complete vaccination coverage among children aged 12–23 months in Ghana: Analysis of data from the 1998 to 2014 Ghana Demographic and Health Surveys . PLoS One, 2020. 15(10): p. e0239754. Bangura, J.B., et al., Barriers to childhood immunization in sub-Saharan Africa: A systematic review . BMC Public Health, 2020. 20(1): p. 1108. Perieres, L., et al., Reasons given for non-vaccination and under-vaccination of children and adolescents in sub-Saharan Africa: A systematic review . Hum Vaccin Immunother, 2022. 18(5): p. 2076524. Favin, M., et al., Why children are not vaccinated: a review of the grey literature . Int Health, 2012. 4(4): p. 229–38. Paulson, K.R., et al., Global, regional, and national progress towards Sustainable Development Goal 3.2 for neonatal and child health: all-cause and cause-specific mortality findings from the Global Burden of Disease Study 2019 . The Lancet, 2021. 398(10303): p. 870–905. Perin, J., et al., Global, regional, and national causes of under-5 mortality in 2000-19: an updated systematic analysis with implications for the Sustainable Development Goals . Lancet Child Adolesc Health, 2022. 6(2): p. 106–115. Bhutta, Z.A., et al., Interventions to address deaths from childhood pneumonia and diarrhoea equitably: what works and at what cost? The Lancet, 2013. 381(9875): p. 1417–1429. Fischer Walker, C.L., et al., Zinc and low osmolarity oral rehydration salts for diarrhoea: a renewed call to action . Bull World Health Organ, 2009. 87(10): p. 780–6. Ahinkorah, B.O., et al., Prevalence and predictors of oral rehydration therapy, zinc, and other treatments for diarrhoea among children under-five in sub-Saharan Africa . Plos One, 2022. 17(10). Ezezika, O., et al., Barriers and facilitators to implementation of oral rehydration therapy in low- and middle-income countries: A systematic review . PLoS One, 2021. 16(4): p. e0249638. Schroder, K., et al., Increasing coverage of pediatric diarrhea treatment in high-burden countries . J Glob Health, 2019. 9(1): p. 0010503. Tanahashi T, Health service coverage and its evaluation . Bulletin of the World Health Organisation, 1978. 56(2): p. 295–303. Mwinnyaa, G., et al., Estimating population-based coverage of reproductive, maternal, newborn, and child health (RMNCH) interventions from health management information systems: a comprehensive review . BMC Health Serv Res, 2021. 21(Suppl 2): p. 1083. Lee, J., et al., Interventions to improve district-level routine health data in low-income and middle-income countries: a systematic review . BMJ Glob Health, 2021. 6(6). Ndabarora, E., J.A. Chipps, and L. Uys, Systematic review of health data quality management and best practices at community and district levels in LMIC . Information Development, 2013. 30(2): p. 103–120. DHIS2. DHIS2 . 2022 [cited 2022 5th June 2022]; Available from: https://dhis2.org/. Maina, I., et al., Using health-facility data to assess subnational coverage of maternal and child health indicators, Kenya . Bull World Health Organ, 2017. 95(10): p. 683–694. Agiraembabazi, G., et al., Can routine health facility data be used to monitor subnational coverage of maternal, newborn and child health services in Uganda? BMC Health Serv Res, 2021. 21(Suppl 1): p. 512. Simmons, E.M., et al., Assessing coverage of essential maternal and child health interventions using health-facility data in Uganda . Popul Health Metr, 2020. 18(1): p. 26. Maïga, A., et al., Measuring coverage of maternal and child health services using routine health facility data: a Sierra Leone case study . BMC Health Serv Res, 2021. 21(Suppl 1): p. 547. Malembaka, E.B., et al., The use of health facility data to assess the effects of armed conflicts on maternal and child health: experience from the Kivu, DR Congo . BMC Health Serv Res, 2021. 21(Suppl 1): p. 195. Walana, W., et al., Childhood immunization in Ghana: Tracing the history and projecting the future . Public Health Challenges, 2024. 3(e176). Ghana Statistical Service, Ghana 2021 Population and Housing Census General Report: Volume 3A Population of Regions and Districts . 2021, Ghana Statistical Service: Accra. Ministry of Health, National Community-Based Health Planning and Services (CHPS) Policy . 2016, Ministry of Health: Accra. Ministry of Health, Health Information System Strategic Plan 2022–2025 . 2022, Ministry of Health: Accra. Countdown to 2030 Collaboration. Health Facility Data & Analysis Center . 2024 [cited 2024 7th October 2024]; Available from: https://www.countdown2030.org/tools-for-analysis/health-facility-data-and-analysis . Ghana Statistical Service, Multiple Indicator Cluster Survey (MICS 2017/18): Survey Findings Report . 2018, Ghana Statistical Service: Accra. Benchimol, E.I., et al., The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement . PLoS Med, 2015. 12(10): p. e1001885. Troeger, C., et al., Estimates of the global, regional, and national morbidity, mortality, and aetiologies of diarrhoea in 195 countries: a systematic analysis for the Global Burden of Disease Study 2016 . The Lancet Infectious Diseases, 2018. 18(11): p. 1211–1228. Lumley T, Analysis of complex survey samples . Journal of Statistical Software, 2004. 9(8): p. 1–19. Pebesma E and Bivand R, Spatial Data Science: With Applications in R . 1st edition ed. 2023. Anselin L, Local Indicators of Spatial Association—LISA . Geographical Analysis, 1995. 27(2): p. 93–115. Anselin L, The Moran scatterplot as an ESDA tool to assess local instability in spatial association , in Spatial Analytical Perspectives on GIS , Fischer MM, Editor. 1996, Routledge: London. p. 111–125. Cook RD and Weisberg S, Residuals and Influence in Regression . 1982, New York: Chapman and Hall. Amoro, V.A., G.A. Abiiro, and K.A. Alatinga, Bypassing primary healthcare facilities for maternal healthcare in North West Ghana: socio-economic correlates and financial implications . BMC Health Serv Res, 2021. 21(1): p. 545. Dotse-Gborgbortsi, W., et al., The influence of distance and quality on utilisation of birthing services at health facilities in Eastern Region, Ghana . BMJ Glob Health, 2019. 4(Suppl 5): p. e002020. Escamilla, V., et al., The Role of Distance and Quality on Facility Selection for Maternal and Child Health Services in Urban Kenya . J Urban Health, 2018. 95(1): p. 1–12. Ocholla, I.A., et al., Geographical accessibility in assessing bypassing behaviour for inpatient neonatal care, Bungoma County-Kenya . BMC Pregnancy Childbirth, 2020. 20(1): p. 287. Yao, J. and V. Agadjanian, Bypassing health facilities in rural Mozambique: spatial, institutional, and individual determinants . BMC Health Serv Res, 2018. 18(1): p. 1006. Kahabuka C, et al., Unfulfilled expectations to services offered at primary health care facilities: Experiences of caretakers of underfive children in rural Tanzania . BMC Health Services Research, 2012. 12(158). Manukure, J.M., F. Annor, and A.A. Gyasi-Gyamerah, ‘Leaving the child behind was not easy’: lived experiences of mothers returning to work after maternity leave in Ghana . Community, Work & Family, 2024: p. 1–24. Ghana Health Service, The Health Sector in Ghana: Facts and Figs. 2017 . 2017. Ma, Y., et al., Evaluating the impact of community health volunteer home visits on child diarrhea and fever in the Volta Region, Ghana: A cluster-randomized controlled trial . PLoS Med, 2019. 16(6): p. e1002830. Health Facilities Regulatory Agency, Specific Requirements for CHPS Compound, v1.3 . 2017, Health Facilities Regulatory Agency: Accra. Mosser, J.F., et al., Mapping diphtheria-pertussis-tetanus vaccine coverage in Africa, 2000–2016: a spatial and temporal modelling study . The Lancet, 2019. 393(10183): p. 1843–1855. Seidler, V., et al., Subnational variations in the quality of household survey data in sub-Saharan Africa . Nat Commun, 2025. 16(1): p. 3771. Cutts, F.T., H.S. Izurieta, and D.A. Rhoda, Measuring coverage in MNCH: design, implementation, and interpretation challenges associated with tracking vaccination coverage using household surveys . PLoS Med, 2013. 10(5): p. e1001404. Fischer Walker, C.L., O. Fontaine, and R.E. Black, Measuring coverage in MNCH: current indicators for measuring coverage of diarrhea treatment interventions and opportunities for improvement . PLoS Med, 2013. 10(5): p. e1001385. Asare, E.O., J.L. Warren, and V.E. Pitzer, Spatiotemporal patterns of diarrhea incidence in Ghana and the impact of meteorological and socio-demographic factors . Front Epidemiol, 2022. 2. Reiner, R.C., Jr., L.B.D.T.B. Collaborators, and S.I. Hay, The overlapping burden of the three leading causes of disability and death in sub-Saharan African children . Nat Commun, 2022. 13(1): p. 7457. Macharia, P.M., et al., Defining service catchment areas in low-resource settings . BMJ Glob Health, 2021. 6(7). Johnson M, et al., A scoping review of the methods used to estimate health facility catchment populations for child health indicators in sub-Saharan Africa . Population Health Metrics, 2025. 23(11). Cobos Munoz, D., et al., Decentralization of health systems in low and middle income countries: a systematic review . Int J Public Health, 2017. 62(2): p. 219–229. El-Khoury M, Banke K, and Sloane P, Improved Childhood Diarrhea Treatment Practices in Ghana: A Pre-Post Evaluation of a Comprehensive Private-Sector Program . Global Health: Science and Practice, 2016. 4(2): p. 264–275. Gesicho, M.B., M.C. Were, and A. Babic, Evaluating performance of health care facilities at meeting HIV-indicator reporting requirements in Kenya: an application of K-means clustering algorithm . BMC Med Inform Decis Mak, 2021. 21(1): p. 6. Gimbel, S., et al., Improving data quality across 3 sub-Saharan African countries using the Consolidated Framework for Implementation Research (CFIR): results from the African Health Initiative . BMC Health Serv Res, 2017. 17(Suppl 3): p. 828. Mutale W, et al., Improving health information systems for decision making across five sub-Saharan African countries: Implementation strategies from the African Health Initiative . BMC Health Services Research, 2013. 13(59). Muhoza, P., et al., A data quality assessment of the first four years of malaria reporting in the Senegal DHIS2, 2014–2017 . BMC Health Serv Res, 2022. 22(1): p. 18. Githinji, S., et al., Completeness of malaria indicator data reporting via the District Health Information Software 2 in Kenya, 2011–2015 . Malar J, 2017. 16(1): p. 344. Sharma, G., et al., Private sector engagement for immunisation programmes: a pragmatic scoping review of 25 years of evidence on good practice in low-income and middle-income countries . BMJ Glob Health, 2024. 8(Suppl 5). World Health Organization, Engaging the private health service delivery sector through governance in mixed health systems: strategy report of the WHO Advisory Group on the Governance of the Private Sector for Universal Health Coverage . 2020, World Health Organization: Geneva. Ziema, S.A. and L. Asem, Assessment of immunization data quality of routine reports in Ho municipality of Volta region, Ghana . BMC Health Serv Res, 2020. 20(1): p. 1013. Ward, K., et al., Enhancing Workforce Capacity to Improve Vaccination Data Quality, Uganda . Emerg Infect Dis, 2017. 23(13): p. S85-93. Bhattacharya, A.A., et al., Quality of routine facility data for monitoring priority maternal and newborn indicators in DHIS2: A case study from Gombe State, Nigeria . PLoS One, 2019. 14(1): p. e0211265. Additional Declarations Yes there is potential Competing Interest. AO, at the time of the study, was the Deputy Director General of Ghana Health Service, which generates and owns the data describing individual health facilities and routinely collected health data analysed for this study and is responsible for the delivery of public health services in Ghana. WDG previously worked with Ghana Health Service as a public health information officer until 2017. Supplementary Files JohnsonUsingroutinelycollectedhealthdatatoestimateS.docx Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities: Supplementary file Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6790840","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":469196125,"identity":"3bd8b103-4d5e-4108-8a59-35f775c90576","order_by":0,"name":"Matthew Johnson","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIie3RMQrCMBTG8Q8CugRcA4I5gfBKwUk8S6TQToWOgoN1sYvg2s0reIRXOrj0AA4OiuDc0UFE6dSpzeiQ/xQCv/DCA1yuv4wNgz6T1s2glywZCfuAsCYQjJqXqTWZbvnGNV2jY3YoHgkWGio0nWTGbIqcnvGpKoWfI/BSFXIvKSWJ+KSCwVhCGKgo7SdvEpHOG7KxJKDS4NKQ38FmsGJPoff7i+9LOns7+TTd5FIF9es91zrb3h9ytdajYUidBKpqv0k2ixxl3ZO7XC6XC19GmktHB+QkkAAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-5597-6615","institution":"University of Southampton","correspondingAuthor":true,"prefix":"","firstName":"Matthew","middleName":"","lastName":"Johnson","suffix":""},{"id":469196126,"identity":"3df9f51e-5f59-4a6c-ba9b-c866c6e8763d","order_by":1,"name":"Winfred Dotse-Gborgbortsi","email":"","orcid":"","institution":"University of Southampton","correspondingAuthor":false,"prefix":"","firstName":"Winfred","middleName":"","lastName":"Dotse-Gborgbortsi","suffix":""},{"id":469196127,"identity":"b4ace7e5-aa1a-46cb-beec-aa0aa740b32d","order_by":2,"name":"Anthony Ofosu","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Anthony","middleName":"","lastName":"Ofosu","suffix":""},{"id":469196128,"identity":"f1df32a2-ffbb-445d-ba57-cb9af363a9f2","order_by":3,"name":"Chigozie Edson Utazi","email":"","orcid":"","institution":"University of Southampton","correspondingAuthor":false,"prefix":"","firstName":"Chigozie","middleName":"Edson","lastName":"Utazi","suffix":""},{"id":469196129,"identity":"3081adfc-4a13-4fe8-8c6b-f21eb49329ae","order_by":4,"name":"Nuala McGrath","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Nuala","middleName":"","lastName":"McGrath","suffix":""},{"id":469196130,"identity":"be6e1d4b-3147-4126-b152-acb3d7460dbc","order_by":5,"name":"James Wright","email":"","orcid":"https://orcid.org/0000-0002-8842-2181","institution":"","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"","lastName":"Wright","suffix":""}],"badges":[],"createdAt":"2025-05-31 12:05:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6790840/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6790840/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86625585,"identity":"d5770aeb-1b74-43a5-931e-94586cb2b471","added_by":"auto","created_at":"2025-07-14 05:11:33","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":168306,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eMap indicating the locations of the study districts and the health facilities retained for analysis within the Eastern Region, Ghana\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/97768c3c45d78c0e1c8693fe.jpg"},{"id":86625584,"identity":"66960ab6-dfa0-444b-a648-e2cb6c473497","added_by":"auto","created_at":"2025-07-14 05:11:33","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":71150,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eLevel of agreement between survey-derived and DHIMS2-derived coverage/utilisation estimates for the Eastern region and all districts therein, by health indicator\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/e26227fa6d5fbcc1e12702d3.jpg"},{"id":86626534,"identity":"36e32bc6-adc3-443c-979f-96070a593580","added_by":"auto","created_at":"2025-07-14 05:29:35","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":62286,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eLevel of agreement between survey-derived and DHIMS2-derived coverage/utilisation estimates for the three study districts and all sub-districts and health facilities therein, by health indicator\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/d85653e9b8f198ff91798677.jpg"},{"id":86625589,"identity":"c48c1f7f-e722-4ba7-bce2-cfb1db9c9bdd","added_by":"auto","created_at":"2025-07-14 05:11:33","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":120115,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eMaps and Moran’s scatterplots describing the spatial autocorrelation associated with DHIMS2-derived coverage/utilisation estimates across health facilities of the three study districts, by health indicator\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/36b3dbcdd319d45bccc388dc.jpg"},{"id":86627469,"identity":"5b18b0d8-ca94-43d9-9f14-e34d3ddb3fe4","added_by":"auto","created_at":"2025-07-14 05:37:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3661863,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/dcb9fb34-6500-4586-8e77-d307d75662d8.pdf"},{"id":86625587,"identity":"9da544c0-7c92-4bd7-afed-3106bfe9f35a","added_by":"auto","created_at":"2025-07-14 05:11:33","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":293636,"visible":true,"origin":"","legend":"Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities: Supplementary file","description":"","filename":"JohnsonUsingroutinelycollectedhealthdatatoestimateS.docx","url":"https://assets-eu.researchsquare.com/files/rs-6790840/v1/e34fb0d3a962b580276866d8.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nAO, at the time of the study, was the Deputy Director General of Ghana Health Service, which generates and owns the data describing individual health facilities and routinely collected health data analysed for this study and is responsible for the delivery of public health services in Ghana. WDG previously worked with Ghana Health Service as a public health information officer until 2017.","formattedTitle":"Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities","fulltext":[{"header":"Plain English Summary","content":"\u003cp\u003eInadequate supply of medicines and other resources limits uptake of essential health services in sub-Saharan Africa. Improved population denominators and intervention coverage estimates may better inform the resourcing of health facilities. We evaluated an estimation approach that uses data collected by facilities as a product of routine service delivery but has not yet been implemented at facility-level. Our estimates generally appeared plausible at the highest administrative levels. Although we show that it is possible to apply this estimation approach at sub-district and facility-levels, the variability and proportion of implausible estimates increased at facility-level especially. To inform facility-level resourcing, these data should be integrated with alternative estimation approaches that can account for additional determinants of health-seeking and service utilisation.\u003c/p\u003e"},{"header":"Background","content":"\u003cp\u003eSustainable Development Goal 3.8 commits countries to Universal Health Coverage (UHC), a multifaceted concept encompassing the quality, affordability and accessibility of health services [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. The latter has led many countries of sub-Saharan Africa (SSA) to take steps to bring services closer to communities. Recent policies from Kenya [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] and Zambia [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] aimed to increase the proportion of the population living within 5km of a health facility (HF), whilst Rwandan policy aimed to reduce health-seekers\u0026rsquo; walking time to the nearest HF to 45 minutes [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. In Ghana, the Community-Based Health Planning and Services (CHPS) initiative, introduced in 2005 [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], has densified the HF network, in rural areas especially, by establishing a new service tier at the base of the primary care hierarchy. By 2021, 90.7% of residential structures were within 5km of the nearest HF [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Though this has alleviated geographic barriers to access, the readiness of CHPS \u0026lsquo;compounds\u0026rsquo; to provide primary care services has often fallen short of higher-tier HFs, with inadequate and/or unreliable supply of essential medicines singled out as a specific problem [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Realising the coverage gains from investments in physical infrastructure thus also depends on HFs being adequately resourced to meet the demand placed upon them.\u003c/p\u003e \u003cp\u003eIn Ghana, despite significant progress since introduction of the Expanded Programme on Immunisation, complete coverage of recommended vaccines amongst children aged 12\u0026ndash;23 months has not yet been achieved [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Recent systematic reviews have highlighted the potential for vaccine stockout at HFs, amongst other factors, to impede further progress in SSA [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The data collected by HFs as a product of routine service delivery may have a pivotal role in this regard. Using these routinely collected health data (RCHD) to generate reliable coverage estimates at the level of individual HFs could help decision-makers to identify and target communities not yet reached by services, but also inform need-based procurement and allocation of vaccines across the HF network [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Moreover, the range of indicators captured by RCHD might permit development of cross-cutting methods to help decision-makers address \u003cem\u003emultiple\u003c/em\u003e population health challenges. Whilst diarrhoeal diseases persist amongst the leading causes of child mortality in SSA [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], a substantial proportion of these deaths could be prevented by universal coverage of oral rehydration solution (ORS), a low-cost and highly effective treatment for dehydration [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Yet despite its benefits, and adoption into many national health policies [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], utilisation remains low: treatment was estimated from recent survey data to have occurred in just 49.1% of acute diarrhoeal episodes amongst children aged under 5 years (u5s) in 30 countries of SSA, rising to 60.8% in Ghana [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Like vaccines, non-availability of ORS packets in HFs has been cited as a supply-side barrier to utilisation [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] that might be alleviated by using improved demand forecasts to guide resource allocation [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eImplicitly, intervention coverage estimation requires a denominator (the number eligible for/targeted by the service) and numerator (the number of users) [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Traditional demographic data sources, including censuses and vital registration systems, are often incomplete, unreliable and outdated in low- and middle-income countries (LMICs), prompting a recent trend towards deriving the denominator (in addition to the numerator) from RCHD [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Though historically under-utilised in SSA, the view that RCHD could contribute to health system management and population health improvement has gained traction, precipitating renewed efforts to improve data quality (DQ) [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Meanwhile, more than 70 LMICs, predominately in SSA and Asia, have adopted District Health Information Software (DHIS2) for collection, reporting and analysis of RCHD [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. This platform was implemented in Ghana in 2012 as the District Health Information Management System (DHIMS2). In 2017 Maina and colleagues proposed a data-driven approach to denominator and intervention coverage estimation using RCHD extracted from DHIS2 [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Briefly, this approach is predicated on the assumption that the population targeted by an intervention can be inferred from the actual number of service events for a related \u0026lsquo;benchmark\u0026rsquo; intervention considered to have near-universal (\u0026gt;\u0026thinsp;90%) coverage, and thus near-complete RCHD. Following this logic, the number of benchmark intervention service events can be extracted from DHIS2 and calibrated using contemporaneous household survey data to obtain the \u0026lsquo;target\u0026rsquo; intervention denominator. The number of target intervention service events is also extracted as the numerator for the coverage calculation. Initially applied in first-level administrative units of Kenya, this approach has since been used to measure childhood vaccination coverage in first- and/or second-level units (in Ghana, regions and districts, respectively) elsewhere in SSA [\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. On the whole, implementation studies suggest this is a promising approach to denominator and coverage estimation at these administrative levels and that, in this application, RCHD may outperform traditional demographic data sources. To our knowledge, however, there are few published examples of implementation at a lower administrative level [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], and none in individual HFs.\u003c/p\u003e \u003cp\u003eWe aimed to evaluate Maina and colleagues\u0026rsquo; approach as a means of estimating HF-level coverage of essential child health interventions in a subnational area of Ghana. The first dose of the pentavalent vaccine (Penta1) is typically received by Ghanaian children 6 weeks after birth [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] and was used as the benchmark intervention. We estimated denominators and coverage/utilisation of four target interventions (three preventive and one curative) selected to cover the first year of life: the second (Penta2) and third (Penta3) doses of the pentavalent vaccine, first dose of the measles-rubella vaccine (MRV1), and outpatient diarrhoea service utilisation amongst children aged under 1 year (u1s). In Ghana, these vaccinations are typically received 10 weeks, 14 weeks and 9 months after birth, respectively [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Coverage/utilisation was estimated at four administrative levels, culminating with individual HFs.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy setting\u003c/h2\u003e \u003cp\u003eAs at the 2021 census, Ghana was subdivided into 16 regions and 261 districts [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Public sector healthcare is provided by Ghana Health Service (GHS) which, to facilitate service delivery and accessibility, further subdivides districts into sub-districts, then CHPS zones. The latter, the smallest unit of service delivery, are centred on specific communities and encapsulate populations of up to 5,000 persons [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Following this administrative structure, regional hospitals deliver secondary care services whilst district-centric networks of hospitals, health centres, clinics and CHPS compounds deliver primary care services. Within HFs, patient-level data collected during routine service events are aggregated by month, manually transcribed to paper-based forms and registers, then entered to DHIMS2. Ghanaian policy requires all public, private and faith-based HFs to report RCHD to DHIMS2 [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThis ecological study was conducted in Ghana\u0026rsquo;s Eastern region (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), which was subdivided into 33 districts and had a total population of 2,925,653 persons at the 2021 census [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Coverage/utilisation was estimated for: (i) the Eastern region as a whole; (ii) all individual districts therein; (iii) sub-districts of three \u0026lsquo;study\u0026rsquo; districts (Atiwa West, Denkyembour and Kwaebibirem), and (iv) individual HFs within these study districts. Sub-districts and HFs were anonymised in all results.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eShapefiles depicting the national boundaries of countries of West and Central Africa and national and subnational (region and district) boundary locations of Ghana were sourced from the Humanitarian Data Exchange under a Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Sub-district boundary locations and health facility point locations were provided by the Ghana Health Service\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData and other resources\u003c/h3\u003e\n\u003cp\u003eStandard Stata scripts for district-level implementation of Maina and colleagues\u0026rsquo; approach are hosted by the Countdown to 2030 Collaboration [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e] and were adapted for this study. The 2017/18 Ghana Multiple Indicator Cluster Survey (GMICS) [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] was contemporaneous with the study period (January 2017 to December 2020) and used to calibrate and evaluate our implementation. All analyses used Stata SE Version 16.0 (Stata-Corp, College Station, TX, USA) and R v4.4.1 with RStudio v2024.04.2\u0026thinsp;+\u0026thinsp;764 (R Core Team, Vienna, Austria). The results are reported in accordance with the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) statement [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRCHD describing the health indicators used for this study (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) were extracted from DHIMS2 and provided by GHS. The study team did not have direct access to the DHIMS2 database. We estimated coverage/utilisation for the 2020 calendar year, taking individual HFs as the lowest-level units of analysis. The DHIMS2 extract was aggregated by month, health indicator and HF, but also identified the sub-district and district within which each was situated. HFs were retained for analysis where their location was known and DHIMS2 provided evidence that they were operational and delivering relevant services throughout 2017\u0026ndash;2020. Having divided the DHIMS2 extract into eight discrete intervals of six months\u0026rsquo; duration (January to June and July to December of each calendar year), HFs were retained where one or more service events were recorded in all intervals for each of: Penta1, all-cause outpatient attendances (all ages combined), and outpatient attendances for diarrhoeal disease (all ages combined). After excluding HFs failing to meet these criteria (\u003cb\u003eSupplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e), the HF-level data were reaggregated by sub-district, district and for the Eastern region as a whole to enable coverage/utilisation estimation at each administrative level. Sub-districts with no HFs meeting the selection criteria were excluded from all analyses, and some sub-districts were grouped (without crossing district boundaries) to ensure that units at this administrative level contained at least three HFs. All districts contained at least three HFs meeting the selection criteria.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eHealth indicators extracted from DHIMS2 and used to implement Maina and colleagues\u0026rsquo; method\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHealth indicator\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePurpose\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePentavalent vaccination, 1st dose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBenchmark intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePentavalent vaccination, 2nd dose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePentavalent vaccination, 3rd dose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMeasles-rubella vaccination, 1st dose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiarrhoeal disease outpatient attendances: less than 28 days of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiarrhoeal disease outpatient attendances: 1\u0026ndash;11 months of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget intervention; data quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiarrhoeal disease outpatient attendances: 1\u0026ndash;4 years of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiarrhoeal disease outpatient attendances: total for all age groups combined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause outpatient attendances: less than 28 days of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause outpatient attendances: 1\u0026ndash;11 months of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause outpatient attendances: 1\u0026ndash;4 years of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause outpatient attendances: total for all age groups combined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause inpatient admissions: less than 28 days of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause inpatient admissions: 1\u0026ndash;11 months of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause inpatient admissions: 1\u0026ndash;4 years of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll-cause inpatient admissions: total for all age groups combined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAntenatal care, 1st visit\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAntenatal care, at least 4 visits\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBacillus Calmette-Guerin vaccination\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCaesarean sections\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDeliveries in health facilities\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFamily planning, new visits/acceptors\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFamily planning, revisits\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntermittent preventive treatment for malaria in infants, 2nd dose\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePostnatal care within 48 hours after delivery\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStillbirths (fresh)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStillbirths (macerated)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal number of deaths in health facilities amongst children under 5 years of age\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal number of maternal deaths in health facilities\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eData quality assessment\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\n\u003ch3\u003eData quality assessment and adjustment\u003c/h3\u003e\n\u003cp\u003eThe quality of DHIMS2 data in 2020 was assessed using standard DQ checks from the Countdown scripts (\u003cb\u003eSupplementary Table S3\u003c/b\u003e). All were performed at region-/district-levels and are described, with results, in the \u003cb\u003esupplementary file\u003c/b\u003e. An outline of the adjustments applied based on these checks follows.\u003c/p\u003e \u003cp\u003eA reporting completeness check and adjustment preceded all others. This invoked equation [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], where \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:n\\)\u003c/span\u003e\u003c/span\u003e represents the number of service events for a health indicator, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:c\\)\u003c/span\u003e\u003c/span\u003e the observed reporting rate, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e the adjustment factor. If incomplete reporting is assumed to indicate service non-provision then \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k=0\\)\u003c/span\u003e\u003c/span\u003e, but if it is assumed that the service was provided at a lower volume than HFs with complete reporting, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e is set to a value between 0 and 1. As a high level of reporting completeness was observed (\u003cb\u003eSupplementary Table S5\u003c/b\u003e), the impact of this adjustment was considered negligible and a default \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e-value of 0.25 assumed for all services.\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:{n}_{adjusted}={n}_{reported}+{n}_{reported}\\text{*}\\left(\\frac{1}{c}-1\\right)\\text{*}k\\)\u003c/span\u003e \u003c/span\u003e[1]\u003c/p\u003e \u003cp\u003eMissing values were imputed using the median of non-missing values for the same administrative unit/health indicator/year combination. After checking internal consistency over time, outliers from 2020 (reported values exceeding five median absolute deviations above/below the 2017\u0026ndash;2019 median) were adjusted by imputing the 2017\u0026ndash;2019 median for the same administrative unit/health indicator combination.\u003c/p\u003e\n\u003ch3\u003eDenominator estimation\u003c/h3\u003e\n\u003cp\u003eTo calibrate the denominator, the DQ-adjusted number of Penta1 service events during 2020 was first uplifted for non-utilisation using the GMICS estimate of 4.8% in the Eastern region. The result was assumed to represent the total number of u1s eligible for Penta2 and Penta3, and was used as the denominator for these target interventions without further adjustment. For MRV1, the result was further adjusted downwards to account for post-neonatal mortality. No adjustment for neonatal mortality was made as children were assumed to have received Penta1 after the neonatal period, at 6 weeks after birth. In the absence of region-specific mortality rates, the national GMICS estimate of 14 post-neonatal deaths per 1,000 live births was used. The same adjustment was applied in respect of diarrhoea service utilisation amongst u1s.\u003c/p\u003e\n\u003ch3\u003eCoverage/utilisation estimation\u003c/h3\u003e\n\u003cp\u003eTo estimate coverage/utilisation of the target intervention, DQ-adjusted numerators from 2020 were divided by the appropriate denominator. As the denominator for diarrhoea service utilisation relates to the number of u1s eligible for the service but DHIMS2 only reports the number of service events taking place amongst this population, an additional step was required to obtain a numerator expressed in terms of service users. The number of diarrhoeal service events amongst children aged less than 28 days or 1\u0026ndash;11 months (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) was summed, then divided by an episode rate of 1.82 episodes per child-year. This rate had been estimated by the 2016 Global Burden of Disease study for u5s in Ghana [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. The result was then divided by the denominator to generate utilisation estimates.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation\u003c/h2\u003e \u003cp\u003eAll DHIMS2-derived coverage/utilisation estimates were compared to survey-derived estimates for the Eastern region, which were recalculated (with 95% confidence intervals) from GMICS microdata using the \u0026lsquo;survey\u0026rsquo; R package [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Differences between the surv5ey-derived estimate for the Eastern region and DHIMS2-derived estimates for HFs, sub-districts, districts and the region were used to compute \u0026lsquo;mean absolute deviation\u0026rsquo; for each administrative level. Though the computation was analogous to mean absolute error, this comparison was taken as demonstrating the variability of DHIMS2-derived estimates through successive administrative levels as opposed to estimation \u0026lsquo;error\u0026rsquo;. To provide an assessment of their consistency across indicators (within administrative levels), deviation values were further standardised as a percentage of the survey-reported estimate and used to calculate Spearman\u0026rsquo;s rank correlation coefficients.\u003c/p\u003e \u003cp\u003eDHIMS2-derived estimates for HFs within the study districts were examined for evidence of spatial autocorrelation using the \u0026lsquo;spdep\u0026rsquo; R package [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. The Global Moran\u0026rsquo;s I statistic [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e], a summary measure of autocorrelation, was separately calculated for each indicator with the neighbourhood structure of HFs defined using the Euclidean distances separating each from its eight nearest neighbours. Distances were converted to a row-standardised spatial weights matrix assigning greater importance to neighbouring estimates. Moran\u0026rsquo;s scatterplots were produced to visualise the extent to which the global statistic accurately reflected local spatial patterns [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Each scatterplot was divided into quadrants by splitting x- and y-axes on the means of coverage/utilisation and spatial lag (the weighted average of neighbouring observations), respectively. The Global Moran\u0026rsquo;s I statistic and 95% confidence interval were also plotted. Estimates exerting disproportionate influence on the global statistic were identified using regression diagnostics including Cook\u0026rsquo;s distance and covariance ratios [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], and highlighted on the scatterplots and HF location maps.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eEthics\u003c/h3\u003e\n\u003cp\u003e Ethical approval was obtained from the GHS Ethics Review Committee (ID:001/03/23) and Ethics Committee of the Faculty of Environmental and Life Sciences, University of Southampton (ID:78697).\u003c/p\u003e \u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eOverall, of 1,084 Eastern region HFs whose location was known, 403 (37.2%) were retained for analysis (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Of these, CHPS compounds were most prevalent (65.0%), followed by health centres (27.5%) and hospitals (5.7%). The 49 HFs retained from the three study districts comprised CHPS compounds (73.4%), health centres (20.4%) and hospitals (6.1%) only.\u003c/p\u003e \u003cp\u003eMost of the 681 Eastern region HFs excluded by the selection algorithm (\u003cb\u003eSupplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e) were non-operational through all, or part, of 2017\u0026ndash;2020 and, together, accounted for a minor proportion of recorded service events. First, 468 were public sector/faith-based HFs that had not reported data to DHIMS2 during the first and second intervals and thus considered non-operational at the start of this period. Second, 143 public sector/faith-based HFs considered operational at the start of the period but failing to report data to DHIMS2 during one or more intervals were excluded. Of these, 29 had not reported data to DHIMS2 during any of the final three intervals, suggesting that they may have ceased to operate mid-period. The remaining 114 HFs were likely operational throughout, but excluded owing to short-term periods of incomplete reporting. Finally, all 70 private sector HFs were excluded as none had reported data to DHIMS2. After exclusions, one sub-district from the three study districts no longer contained any HFs and was also excluded from further analysis. The 20 remaining sub-districts were grouped where necessary (without crossing district boundaries) to produce 12 administrative units at this level, each containing at least three HFs.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eThe number and distribution of valid health facility locations received from Ghana Health Service and retained for analysis, by type and ownership\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003eHealth facility group\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eType of health facility\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c7\" namest=\"c4\"\u003e \u003cp\u003eNumber (row %)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003cp\u003e(column %)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eNumber (%) excluded \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePublic sector\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eFaith-based\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePrivate sector\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eUnknown\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"11\" rowspan=\"12\"\u003e \u003cp\u003e\u003cb\u003eEastern region\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003e\u003cb\u003eHealth facilities with valid locations\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHospital\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23 (52.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5 (11.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e15 (34.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1 (2.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e44 (4.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eClinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10 (18.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4 (7.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e33 (61.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e7 (13.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e54 (5.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHealth centre\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e123 (80.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10 (6.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6 (3.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e13 (8.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e152 (14.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eCHPS compound\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e795 (97.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2 (0.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e17 (2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e814 (75.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eMaternity clinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3 (15.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16 (80.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1 (5.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e20 (1.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eAll types\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e954 (88.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e21 (1.9)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e70 (6.5)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003e39 (3.6)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cb\u003e1,084 (100.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u003cb\u003e-\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003e\u003cb\u003eHealth facilities retained by the selection algorithm\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHospital\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e18 (78.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4 (17.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1 (4.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e23 (5.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e21 (47.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eClinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (28.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4 (57.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1 (14.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7 (1.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e47 (87.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHealth centre\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e94 (84.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8 (7.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e9 (8.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e111 (27.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e41 (27.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eCHPS compound\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e255 (97.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e7 (2.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e262 (59.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e552 (67.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eMaternity clinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e20 (100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eAll types\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e369 (91.6)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e16 (4.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e0 (0.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003e18 (4.5)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cb\u003e403 (100.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u003cb\u003e681 (62.8)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003e\u003cb\u003eStudy districts\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003e\u003cb\u003eHealth facilities retained by the selection algorithm\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHospital\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (66.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1 (33.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3 (6.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eClinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e1 (100.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eHealth centre\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8 (80.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e2 (20.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e10 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eCHPS compound\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e34 (94.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e2 (5.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e36 (73.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e25 (41.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eMaternity clinic\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eAll types\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e44 (89.8)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e1 (2.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e0 (0.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003e4 (8.2)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cb\u003e49 (100.0)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u003cb\u003e28 (36.4)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"9\" nameend=\"c9\" namest=\"c1\"\u003e \u003cp\u003e \u003csup\u003e \u003cb\u003ea\u003c/b\u003e \u003c/sup\u003e \u003cb\u003eThe number (%) of excluded health facilities was derived by comparison to the 1,084 valid health facility locations received from Ghana Health Service (see also Supplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDHIMS2-derived Penta2 coverage estimates for the region (90.3%) and most districts (median:93.6%, IQR:87.3%-96.2%) closely approximated the survey-derived estimate (89.8%, 95%CI:82.9%-96.7%) (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Penta3 estimates for the region (92.4%) and most districts (median:96.2%, IQR:87.8%-100.0%) exceeded the survey-derived estimate and upper confidence limit (80.1%, 95%CI:68.7%-91.5%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), suggesting that the DHIMS2-derived denominator was too low, yielding a coverage over-estimate. MRV1 followed a similar pattern to Penta2, notwithstanding greater difference between DHIMS2-derived (83.8%) and survey-derived estimates (81.3%, 95%CI:72.6%-90.0%) for the region, and wider distribution of district-level DHIMS2-derived estimates (median:87.8%, IQR:80.2%-94.2%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Estimates of diarrhoea service utilisation for the region (7.8%) and most districts (median:7.7%, IQR:5.9%-9.6%) fell below the survey-derived estimate and lower confidence limit (15.4%, 95%CI:8.7%-22.1%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), suggesting that the DHIMS2-derived denominator was too high, yielding an under-estimate. The distribution of DHIMS2-derived estimates appeared to widen amongst sub-districts and HFs (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), with several vaccination coverage estimates exceeding 100% at these lower levels. Estimates of Penta2, Penta3 and MRV1 coverage exceeded 100% in 34.7%, 42.9% and 28.6% of HFs (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), respectively. In addition, DHIMS2-derived estimates tended to deviate from survey-derived equivalents by an increasing margin through successive administrative levels (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparisons of survey-derived and DHIMS2-derived coverage/utilisation estimates, by health indicator and administrative level\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdministrative level\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEstimate \u003csup\u003eb, c\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePenta2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePenta3\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMRV1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDiarrhoea service utilisation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eEastern region\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSurvey-derived estimate (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e89.8 (82.9\u0026ndash;96.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e80.1 (68.7\u0026ndash;91.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e81.3 (72.6\u0026ndash;90.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e15.4 (8.7\u0026ndash;22.1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDHIMS2-derived estimate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e90.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e92.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e83.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDeviation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eDistricts (all)\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;33)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian (IQR) DHIMS2-derived estimate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e93.6 (87.3, 96.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e96.2 (87.8, 100.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e87.8 (80.2, 94.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.7 (5.9, 9.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean absolute deviation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e14.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e7.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber (%) units with coverage estimate\u0026thinsp;\u0026gt;\u0026thinsp;100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (6.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9 (27.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4 (12.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eDistricts (study)\u003c/b\u003e \u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;3)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian (IQR) DHIMS2-derived estimate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.2 (94.9, 96.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.9 (97.2, 99.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e84.5 (83.9, 86.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e4.5 (4.1, 5.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean absolute deviation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e18.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e10.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber (%) units with coverage estimate\u0026thinsp;\u0026gt;\u0026thinsp;100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (0.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eSub-districts\u003c/b\u003e \u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;12)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian (IQR) DHIMS2-derived estimate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95.6 (94.7, 97.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.7 (97.1, 102.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e88.1 (83.1, 98.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e4.3 (3.0, 6.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean absolute deviation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e11.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e10.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber (%) units with coverage estimate\u0026thinsp;\u0026gt;\u0026thinsp;100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4 (33.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003eHealth facilities\u003c/b\u003e \u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;49)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian (IQR) DHIMS2-derived estimate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.5 (92.0, 104.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e97.5 (89.3, 106.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e93.1 (80.6, 103.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.3 (0.0, 5.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMean absolute deviation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e21.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e17.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e12.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber (%) units with coverage estimate\u0026thinsp;\u0026gt;\u0026thinsp;100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17 (34.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e21 (42.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e14 (28.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003e\u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e \u003cb\u003eRows show estimates for administrative units located within the three study districts only (Atiwa West, Denkyembour and Kwaebibirem districts)\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cb\u003eb\u003c/b\u003e\u003c/sup\u003e \u003cb\u003eSurvey-derived estimates for the Eastern region were recalculated (with 95% confidence intervals) using microdata from the 2017/18 Ghana Multiple Indicator Cluster Survey\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cb\u003ec\u003c/b\u003e\u003c/sup\u003e \u003cb\u003eDeviation is the difference between the survey-derived estimate for the Eastern region and DHIMS2-derived estimates for each administrative unit, and was used to compute mean absolute deviation for each administrative level\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eData points represent DHIMS2-derived estimates for districts of the Eastern region\u003c/h2\u003e \u003cp\u003e \u003cb\u003eThe solid red vertical line represents the survey-derived coverage/utilisation estimate and the red shaded area its associated 95% confidence interval\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe solid blue vertical line represents the DHIMS2-derived coverage/utilisation estimate for the Eastern region as a whole\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe solid grey vertical line (shown on the vaccination coverage plots only) is placed at 100% coverage\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eData points represent DHIMS2-derived estimates for the three study districts of the Eastern region (grey points) and all sub-districts (dark blue points) and individual health facilities (light blue points) therein\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe solid red vertical line represents the survey-derived coverage/utilisation estimate and the red shaded area its associated 95% confidence interval\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe solid blue vertical line represents the DHIMS2-derived coverage/utilisation estimate for the Eastern region as a whole\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe solid grey vertical line (shown on the vaccination coverage plots only) is placed at 100% coverage\u003c/b\u003e \u003c/p\u003e \u003cp\u003ePenta2, Penta3 and MRV1 deviation values were moderately/strongly correlated at all administrative levels (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), suggesting that their direction and magnitude was largely consistent across these indicators. The same pattern is discernible from district-level dot plots (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), which indicate, for example, over-estimation for all vaccinations in Okere district, but under-estimation in Nsawam-Adoagyiri district. Deviation values for diarrhoea service utilisation were not correlated with any other indicator.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCorrelation of deviation values for all districts of the Eastern region and all sub-districts and health facilities of the three study districts, by health indicator\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAdministrative level\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHealth indicator\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePenta2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePenta3\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMRV1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eDiarrhoea service utilisation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eDistricts (all)\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;33)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta2\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.677 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.578 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta3\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.677 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.834 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.172\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eMRV1\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.578 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.834 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.284\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDiarrhoea service utilisation\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.172\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.284\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eSub-districts\u003c/b\u003e \u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;12)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta2\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.601 **\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.874 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.028\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta3\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.601 **\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-0.042\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eMRV1\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.874 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.308\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDiarrhoea service utilisation\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.028\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.308\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e\u003cb\u003eHealth facilities\u003c/b\u003e \u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003e(n\u0026thinsp;=\u0026thinsp;49)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta2\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.617 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.559 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-0.015\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePenta3\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.617 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.467 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.066\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eMRV1\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.559 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.467 ***\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDiarrhoea service utilisation\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.066\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eDeviation values were standardised by conversion to a percentage of the survey-derived estimate for the Eastern region. Spearman\u0026rsquo;s rank correlation coefficient was then calculated across health indicators and within administrative levels\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e\u003cb\u003e*** correlation significant at \u0026lt;\u0026thinsp;0.01 level; ** correlation significant at \u0026lt;\u0026thinsp;0.05 level; * correlation significant at \u0026lt;\u0026thinsp;0.1 level\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003csup\u003e\u003cb\u003ea\u003c/b\u003e\u003c/sup\u003e \u003cb\u003eRows show estimates for administrative units located within the three study districts only (Atiwa West, Denkyembour and Kwaebibirem districts)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe Global Moran\u0026rsquo;s I statistic suggested a slight tendency towards negative spatial autocorrelation for each indicator, but failed to reach statistical significance. Still, the presence of influential points in the upper-left and lower-right quadrants of each scatterplot (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) indicates that the global statistic obscures low\u0026ndash;high and high\u0026ndash;low clustering; that is, instances of HFs falling below or exceeding the mean coverage/utilisation and spatial lag but surrounded by neighbours with dissimilar estimates. The location maps confirmed that these points pertained to HFs with vaccination coverage estimates substantially exceeding 100% but whose closest neighbours had lower estimates, or vice versa (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eShapefiles depicting district boundary locations were sourced from the Humanitarian Data Exchange under a Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Health facility point locations were provided by the Ghana Health Service\u003c/b\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003ePoints on the maps indicate the locations of health facilities across the study districts and are shaded by their DHIMS2-derived coverage/utilisation estimate. Points on the scatterplots represent individual health facilities and are plotted according to their coverage/utilisation estimate and spatial lag value. Any points found to exert a disproportionate influence on the Global Moran\u0026rsquo;s I statistic are highlighted on both the map and scatterplot\u003c/b\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study aimed to estimate subnational coverage/utilisation of child health interventions in Ghana\u0026rsquo;s Eastern region and, to our knowledge, is the first to implement Maina and colleagues\u0026rsquo; approach at the level of individual HFs. Vaccination coverage estimates appeared plausible at region- and district-levels but frequently exceeded 100% and deviated from survey-derived equivalents by an increasing margin through lower levels. This may reflect local variation in vaccination uptake, but patterns of over-/under-estimation across proximate subregions of Uganda [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] and districts of Sierra Leone [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] were largely attributed by previous implementation studies to cross-boundary health-seeking causing actual service utilisation levels to depart from the denominator. The implicit assumption that health-seekers will not cross administrative boundaries is less likely to hold in smaller units, particularly HF catchment areas. Bypassing, or travel beyond the nearest HF, to access maternal/child health services has been observed across Ghana and SSA; besides spatial determinants, including the number, location and convenience of HFs [\u003cspan additionalcitationids=\"CR43\" citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e], some health-seekers may favour higher-tier HFs [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e] or eschew those perceived as falling short of an expected standard [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. In this way some HFs may possess aspatial qualities or characteristics that make them more or less attractive to health-seekers, motivating bypassing. Maina and colleagues\u0026rsquo; approach arguably accounts for competition between proximate HFs indirectly in that DHIMS2-derived numerators and denominators implicitly capture the health-seeking choices made by service users. Still, the presence of low\u0026ndash;high/high\u0026ndash;low clustering amongst HF-level coverage estimates suggests that the factors underlying these choices may not always be shared between benchmark/target interventions, with potential to induce numerator/denominator mismatch. The finding that deviation values for Penta2, Penta3 and MRV1 were moderately/strongly correlated across HFs may point to a difference between Penta1 and subsequent vaccinations, in particular. One scenario that might give rise to such a trend in urban areas is that women returning to formal employment after maternity leave, for which Ghana\u0026rsquo;s standard entitlement is three months [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e], may need to share caring duties between family members or take children to HFs nearer the home or workplace. Whilst Maina and colleagues\u0026rsquo; approach appears relatively robust to differences of this nature at the highest levels of aggregation, our findings suggest that its estimates become more unstable at the level of individual HFs especially.\u003c/p\u003e \u003cp\u003eDiarrhoea service utilisation was found to be under-estimated at all administrative levels. As preventive interventions, including childhood vaccinations, are targeted to all members of the eligible population, a relatively large number of service events can be expected to take place within even the lowest-level administrative units. Curative interventions, however, are only utilised by persons experiencing and receiving treatment for a specific health need: according to the 2017/18 Ghana Multiple Indicator Cluster Survey (GMICS), diarrhoea prevalence was 15.4% (95%CI:8.7%-22.1%) amongst u1s in the Eastern region, with treatment-seeking to public/private HFs occurring in just 46.5% (95%CI:19.9%-73.1%) of these episodes. As such, while data released by Ghana Health Service (GHS) consistently placed diarrhoea amongst the top five reasons for all-ages outpatient attendance from 2002\u0026ndash;2016 [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e], the number of u1s presenting to HFs, and thus providing the numerator for the utilisation calculation, was considerably lower. Indeed, we found that low numerator values yielded utilisation estimates close to 0% in several HFs. The estimation process was further complicated by the numerator and denominator describing different quantities: whilst the Penta1-derived denominator represented the estimated number of u1s eligible for the diarrhoea service, the numerator extracted from DHIMS2 quantified service events amongst this population. To obtain a numerator expressed in the same unit as the denominator we divided the number of service events by an episode rate that, being estimated for u5s nationally as of 2016, was not necessarily appropriate for the time period or target population. Overall, the assumptions associated with establishing both numerator and denominator appear to have produced estimates that cannot reliably guide the resourcing of HFs for this curative intervention, suggesting that it may be better served by an approach that does not presuppose a relationship between benchmark/target interventions. Still, a Penta1-derived denominator may have relevance to preventive interventions designed to reduce infectious disease incidence in this population. A randomised controlled trial conducted in Ghana\u0026rsquo;s Volta region suggested that health education delivered to caregivers at regular home visits led to increased frequency of handwashing and other behaviours that reduce children\u0026rsquo;s infectious disease risk [\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. Denominator estimates generated using this approach could thus contribute to the resourcing of HFs as a base for health promotion and education programmes delivered to local communities.\u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eRefining the approach for implementation amongst individual health facilities\u003c/h2\u003e \u003cp\u003ePrevious implementation studies typically used vaccinations delivered during neonatal or early post-neonatal periods as a benchmark for subsequent child health interventions. Maina and colleagues recommended consulting contemporaneous household survey data to verify that the prospective benchmark intervention has minimum coverage of 90%, but our analysis suggests that this criterion alone is insufficient for HF-level implementation. As an example, Ghanaian children typically receive the Bacillus Calmette-Guerin (BCG) vaccination at birth [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] and, though its GMICS coverage estimate was 91.8% in the Eastern region, service events were most commonly reported to DHIMS2 by HFs that had also reported childbirths. As such, while BCG may provide a viable benchmark at higher levels of aggregation [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e], the same may not be true for HFs that are not mandated to deliver childbirth services, such as CHPS compounds [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. Still fewer benchmark interventions may be available in CHPS zones that lack a built compound, but where a limited package of services is being delivered in a mobile capacity. We therefore recommend that HF-level estimation be preceded by an additional review of DHIS2 to ensure that the benchmark intervention was delivered by HFs at all primary care tiers through the time period of interest.\u003c/p\u003e \u003cp\u003eConsideration should also be given to the calibration adjustments associated with translating the benchmark to a denominator. Having used Penta1 as the benchmark, our DHIMS2-derived denominator was calibrated to adjust for non-utilisation. In the absence of sub-regional equivalents, we applied the GMICS rate of Penta1 non-utilisation for the Eastern region to \u003cem\u003eall\u003c/em\u003e administrative units as a constant adjustment factor. This disregards sub-regional variation in vaccination uptake, however, which may be significant at the most granular geographies [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. Equally important are limitations and DQ issues associated with the survey, which may also exhibit a geographic pattern [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e]. As an example, vaccination uptake may be incorrectly recorded if non-availability of a child\u0026rsquo;s vaccination card necessitates caregiver self-report, which may be subject to recall or other biases [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. Some implementation studies have explored using first antenatal care visits as a benchmark for child health interventions [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. This was ruled out for the present study, however, as establishing this benchmark/target link entails a longer sequence of survey-informed calibration adjustments (including stillbirth, twinning and mortality rates), each of which may obscure distinct patterns of geographic variation and create more opportunity for inaccuracy to enter the process. We also recommend, therefore, that users working at HF-level select a benchmark intervention that minimises the number and impact of calibration adjustments.\u003c/p\u003e \u003cp\u003eWe also evaluated DHIMS2-derived estimates by comparison with survey-derived equivalents for the Eastern region. Although their agreement appeared to diminish through lower administrative levels, this comparison also disregarded legitimate sources of sub-regional variation. Diarrhoeal risk, for example, is highly influenced by the prevailing meteorological and environmental conditions, such that survey-derived prevalence and/or treatment-seeking estimates (which utilise a two-week reference period) may differ markedly from the annual average in some locations [\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. Moreover, diarrhoeal risk has been shown to increase in Ghana\u0026rsquo;s peri-urban districts, which are often characterised by inadequate sanitation and close living conditions [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e]. Even within a single region, our DHIMS2-derived estimates of diarrhoea service utilisation varied widely at district- (range:2.8%-17.1%) and HF-levels (range:0.0%-24.5%). Modern geostatistical methods incorporating spatially varying environmental and socioeconomic covariates have been used to further disaggregate survey-derived infectious disease incidence [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e] and vaccination coverage [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e] estimates from SSA, and may provide more appropriate comparators for sub-regional DHIMS2-derived estimates. The additional methodological complexity may be offset if offering a clearer picture of the performance of Maina and colleagues\u0026rsquo; approach at these levels. Notably, as HFs are rarely enclosed within unambiguous administrative boundaries in SSA, disaggregation to this level would necessitate a preliminary step to delineate catchment areas [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e], although using survey-derived comparators for sub-districts, or even districts, would still represent a valuable advance. Still, as survey organisations work to sometimes lengthy release cycles, the absence of an iteration coinciding with the estimation year may also limit the extent to which survey-derived metrics should be considered a \u0026lsquo;gold standard\u0026rsquo; for calibration and evaluation. We indirectly accounted for health system reconfiguration over time by way of our HF selection algorithm but were unable to account for other contextual changes with implications for the agreement of DHIMS2-derived and survey-derived estimates, such as population movement or the inter-annual epidemiology of diarrhoea.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eScaling estimation to national-level health facility networks\u003c/h2\u003e \u003cp\u003eSeveral HF-level estimation approaches have been explored in SSA, but have largely been implemented in a single or small number of HFs, thereby facilitating access to data that are not always accessible to decision-makers in this setting [\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. This is key, for deployment throughout the national HF network depends upon ease of implementation by way of data structures that are consistent and common to all. A clear advantage of Maina and colleagues\u0026rsquo; approach, therefore, is its emphasis upon data generated via pre-existing and nationally consistent administrative processes and made readily available to decision-makers at all levels of health system management via DHIS2. In keeping with the trend towards health sector decentralisation in LMICs [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e], this approach could, in principle, be administered by District Health Management Teams (DHMTs) across the country to inform resource allocation and evidence-based decision-making at the lowest levels of service delivery. Having used RCHD extracted from DHIMS2 and experienced no computational difficulties with HF-level implementation across three districts of Ghana, we have no reason to foresee any technical impediment to replication by DHMTs. This arrangement could further enhance the scalability of the approach by affording decision-makers the opportunity to tailor its implementation to the local context. As an example, whilst its denominator estimates are derived using actual service utilisation levels from the recent past, interventions to increase treatment-seeking for childhood diarrhoea are being trialled in many settings [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Indeed, a study undertaken in three regions of Ghana found that efforts to improve the supply of ORS and promote its use had increased utilisation over a three-year period [\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e]. To guard against under-estimation, and therefore under-resourcing of local HFs, in this scenario, the anticipated impact of relevant interventions could be factored into the approach as an additional adjustment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eImplications of data quality and completeness\u003c/h2\u003e \u003cp\u003eOur assessment indicated that DHIMS2 DQ was generally of a high standard in 2020 (\u003cb\u003esupplementary file\u003c/b\u003e). This finding was, in part, an artefact of our selection criteria being linked to reporting completeness, but confirms that much progress has been achieved in Ghana. As completeness approached 100% amongst HFs retained for analysis, we adjusted for incomplete reporting by applying a common \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e-value across all services and administrative units. Yet whilst incomplete reporting may ostensibly point to suboptimal DQ, the same trend may equally arise in lower-level administrative units where intermittent stockouts or staffing issues have disrupted service delivery, or inter-facility competition has distorted health-seeking flows [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Indeed, a DHIS2 DQ assessment from Kenya showed that reporting completeness may vary between HFs, across services and over time, and that a short-term decline during 2017 was linked to a national health worker strike [\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. To better account for the impact of wider organisational events on local service delivery, there is value in prospective collection of information to contextualise outlying DQ metrics (such as the duration of supply chain or staffing issues), and DHMTs liaising directly with health workers to co-determine more appropriate adjustments for affected HFs. Efforts to involve health workers in decision-making processes may have the further benefit of helping them to recognise the value of high-quality data to broader health system objectives, fostering greater commitment to future DQ improvement initiatives [\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIrrespective of the cause, incomplete reporting may preclude replication of Maina and colleagues\u0026rsquo; approach in some HFs. A three-year \u0026lsquo;look-back\u0026rsquo; period is used to assess internal consistency over time and adjust outliers. Though less problematic in large-scale administrative units aggregating several HFs, this duration may be prohibitive in smaller units where health system reconfiguration has created discontinuities of service, and therefore data. Our data suggested that 497 public sector/faith-based HFs of the Eastern region were inactive at start and/or end of the study period. At the same time, the exclusion of \u003cem\u003eall\u003c/em\u003e private sector HFs indicates that its adherence to mandatory reporting processes was poor. This is not unique to Ghana, however, with a similar contrast between public and private sectors observed elsewhere in SSA [\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e, \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e]. Although many LMICs rely upon the private sector to extend vaccination coverage to underserved communities [\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e], its contribution cannot be fully appreciated from incomplete or unreliable data, hampering decision-makers\u0026rsquo; ability to track progress and close persistent gaps. Inter-facility competition will influence levels of demand throughout the HF network, underlining the importance of complete data for optimal resource allocation. Sustainable policy levers are needed to maintain compliance with national reporting standards: Ghana\u0026rsquo;s Health Facilities Regulatory Agency licenses HFs to operate and has a role via its accreditation scheme, but there is also merit in DHMTs engaging with private providers directly to build relationships, promote information sharing and demonstrate the mutual benefits of high-quality data capturing the universe of local HFs [\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAs Maina and colleagues\u0026rsquo; approach depends upon the relationship between benchmark/target interventions, the accuracy and internal consistency of related indicators are crucial, and should be a focus of DQ improvement efforts. We found that the volume of Penta3 service events frequently exceeded that of Penta1, resulting in numerator/denominator mismatch and coverage over-estimation, often beyond 100%. Again, deviations from the expectation may arise if the factors underlying HF choice differ across benchmark/target interventions, but selective attachment of coverage targets to certain services may also induce over-reporting or disproportionate focus upon Penta3 in some Ghanaian HFs [\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e]. Persistent weaknesses affecting the manual, paper-based data collection and reporting procedures followed by health workers in SSA may also contribute: studies from Uganda and Nigeria, for example, found inconsistencies between physical records and DHIS2 [\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e, \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e]. Importantly, DQ improvement is a continual process and may reach a point where, to maximise HF retention, it becomes possible to detach the standard checks from the approach in favour of a generalised assessment programme supported by regular training and monitoring visits to HFs by dedicated staff. Unsupervised machine learning methods have been used with data extracted from DHIS2 as a means of identifying under-performing HFs [\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e], and could be leveraged to facilitate DQ assessment across the network.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eStrengths and limitations\u003c/h2\u003e \u003cp\u003eAs the first to evaluate Maina and colleagues\u0026rsquo; approach at HF-level, our study considers issues and limitations not yet raised by previous implementation studies. Though we offer recommendations to address the issues we have observed, our subnational study area may not capture all that might manifest if scaled to national-level. Having used RCHD that are similar in format and content to other DHIS2-adopter countries, we believe that many of our observations concerning DQ and the performance of this estimation approach are likely generalisable both within and beyond Ghana. As previously noted, however, our evaluation cannot provide a conclusive picture of performance at sub-regional levels and would benefit from comparison with survey-derived estimates generated using geostatistical modelling methods. We also recognise potential limitations associated with using RCHD extracted from DHIMS2. Though standard data collection procedures are used in Ghanaian HFs, the accuracy and generalisability of these data may be affected if their interpretation or understanding varies amongst health workers, or if alternative definitional approaches are followed in other countries. As our data extract did not provide the age of children attending vaccination service events, we assumed that all took place according to Ghana\u0026rsquo;s standard schedule. Yet this may not always be the case, raising the possibility of misclassification bias; vaccination of ineligible children could generate excess reporting in some HFs and may have contributed to our findings around Penta1/Penta3 service volumes.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eReliable population denominators and intervention coverage estimates could help decision-makers to ensure that HFs are better resourced to meet the demand placed upon them. Maina and colleagues\u0026rsquo; estimation approach is predicated upon an assumed relationship between benchmark/target interventions and has shown promise at region- and district-levels. Yet having implemented this approach at lower administrative levels, we found that the variability of vaccination coverage estimates and proportion of implausible estimates (\u0026gt;\u0026thinsp;100%) increased through district-, sub-district- and HF-levels. The finding that deviations from survey-derived equivalents were correlated at HF-level suggests shared underlying drivers, likely related to inconsistent patterns of service utilisation across benchmark/target interventions. Estimation of diarrhoea service utilisation was further challenged by the additional assumptions required to establish both the numerator and denominator. The organisation and configuration of the local health system may also impose certain restrictions upon HF-level implementation: services that are optional at lower tiers of the primary care hierarchy are unable to provide a viable benchmark, while newly-established or intermittently-reporting HFs may have insufficient longitudinal data. Indeed, a number of HFs were excluded from our study following initial DQ and reporting completeness checks, including all from the private sector. Overall, whilst this approach has potential beneath district-level, we found that estimates became more unstable through lower levels, with implications for their use to guide the resourcing of individual HFs. Incorporating RCHD within alternate approaches that explicitly recognise inter-facility competition and the varied spatial/aspatial determinants of health-seeking and service utilisation, but do not depend upon a benchmark/target relationship, might retain the advantages of Maina and colleagues\u0026rsquo; approach whilst addressing some of its shortcomings in respect of HF-level estimation.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eBCG\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eBacillus Calmette-Guerin vaccination\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCHPS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCommunity-Based Health Planning and Services (initiative)\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDHIMS2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDistrict Health Information Management System\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDHIS2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eDistrict Health Information Software\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDQ\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edata quality\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGHS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGhana Health Service\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGMICS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e2017/18 Ghana Multiple Indicator Cluster Survey\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHF\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ehealth facility\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eLMIC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003elow- and middle-income country\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMRV1\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003efirst dose of the measles-rubella vaccine\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eORS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eoral rehydration solution\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePenta1\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003efirst dose of the pentavalent vaccine\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePenta2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003esecond dose of the pentavalent vaccine\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePenta3\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ethird dose of the pentavalent vaccine\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRCHD\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eroutinely collected health data\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSSA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003esub-Saharan Africa\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eu1s\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003echildren aged under 1 year\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eu5s\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003echildren aged under 5 years\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eUHC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eUniversal Health Coverage\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data describing individual health facilities and routinely collected health data analysed for this study are not publicly available due to confidentiality and data licensing restrictions from the Ghana Health Service. They can be obtained from the Ghana Health Service (https://ghs.gov.gh/) on reasonable request.\u0026nbsp;The\u0026nbsp;2017/18 Ghana Multiple Indicator Cluster Survey microdata are available from\u0026nbsp;UNICEF (https://mics.unicef.org/).\u0026nbsp;For mapping purposes, shapefiles depicting the national boundaries of countries of West and Central Africa and\u0026nbsp;national and subnational (region and district) boundary locations of Ghana were sourced from the Humanitarian Data Exchange (https://data.humdata.org/). The Humanitarian Data Exchange web platform stipulates that these data are made publicly available under the Creative Commons Attribution for Intergovernmental Organisations (CC BY-IGO) licence. Sub-district boundary locations and health facility point locations were provided directly by the Ghana Health Service and are not publicly available.\u003c/p\u003e\n\u003cp\u003eStandard Stata scripts for implementation of the method utilised for this study are\u0026nbsp;openly available from\u0026nbsp;the Countdown to 2030 Collaboration at https://www.countdown2030.org/tools-for-analysis/health-facility-data-and-analysis. The adapted scripts are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank Mr Thomas Ankomah of the Ghana Health Service for providing technical advice and assisting with extraction of routinely collected health data from the\u0026nbsp;District Health Information Management System (DHIMS2). The authors also thank Dr Winfred Ofosu, Ghana Health Service\u0026nbsp;Regional Director of Health Services for the Eastern region, for granting permission to access and use routinely collected health data from this region.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMJ and JW conceptualised and designed the study. MJ analysed the data and wrote the first draft of the manuscript. WD-G and AO facilitated access to and interpretation of data from the Ghana Health Service. All authors provided feedback and edited drafts of the manuscript, and read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAO, at the time of the study, was the Deputy Director General of Ghana Health Service, which generates and owns the data describing individual health facilities and routinely collected health data analysed for this study and is responsible for the delivery of public health services in Ghana. WDG previously worked with Ghana Health Service as a public health information officer until 2017.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMJ received funding from the UK Economic and Social Research Council (ESRC) South Coast DTP, grant ID: ES/P000673/1. NM is a recipient of an NIHR Research Professorship award (Ref: RP-2017-08-ST2-008). The funders had no role in the study.\u003c/p\u003e\n\u003cp\u003eFor the purposes of open access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eUnited Nations. \u003cem\u003eSDG Indicators: Global indicator framework for the Sustainable Development Goals and targets of the 2030 Agenda for Sustainable Development\u003c/em\u003e. 2021 [cited 2021 11th August 2021]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://unstats.un.org/sdgs/indicators/indicators-list/\u003c/span\u003e\u003cspan address=\"https://unstats.un.org/sdgs/indicators/indicators-list/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health. \u003cem\u003eKenya Health Sector Strategic Plan\u003c/em\u003e. 2018 [cited 2021 4th June 2021]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.health.go.ke/wp-content/uploads/2020/11/Kenya-Health-Sector-Strategic-Plan-2018-231.pdf\u003c/span\u003e\u003cspan address=\"https://www.health.go.ke/wp-content/uploads/2020/11/Kenya-Health-Sector-Strategic-Plan-2018-231.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health. \u003cem\u003eZambia National Health Strategic Plan 2017\u0026ndash;2021\u003c/em\u003e. 2017 [cited 2022 20th June 2022]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.moh.gov.zm/?page_id=1306\u003c/span\u003e\u003cspan address=\"https://www.moh.gov.zm/?page_id=1306\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health, \u003cem\u003eFourth Health Sector Strategic Plan: July 2018 - June 2024\u003c/em\u003e. 2018, Ministry of Health: Kigali.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhana Health Service, \u003cem\u003eCommunity-Based Health Planning and Services (CHPS): The Operational Policy\u003c/em\u003e. 2005, Ghana Health Service: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhana Statistical Service, \u003cem\u003eGhana 2021 Population and Housing Census: Volume 2 Proximity of Residential Structures to Essential Service Facilities\u003c/em\u003e. 2021, Ghana Statistical Service: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAyanore, M., et al., \u003cem\u003eSub-national variations in general service readiness of primary health care facilities in Ghana: Health policy and equity implications towards the attainment of Universal Health Coverage\u003c/em\u003e. PLoS One, 2022. 17(6): p. e0269546.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBudu, E., et al., \u003cem\u003eTrend and determinants of complete vaccination coverage among children aged 12\u0026ndash;23 months in Ghana: Analysis of data from the 1998 to 2014 Ghana Demographic and Health Surveys\u003c/em\u003e. PLoS One, 2020. 15(10): p. e0239754.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBangura, J.B., et al., \u003cem\u003eBarriers to childhood immunization in sub-Saharan Africa: A systematic review\u003c/em\u003e. BMC Public Health, 2020. 20(1): p. 1108.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerieres, L., et al., \u003cem\u003eReasons given for non-vaccination and under-vaccination of children and adolescents in sub-Saharan Africa: A systematic review\u003c/em\u003e. Hum Vaccin Immunother, 2022. 18(5): p. 2076524.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFavin, M., et al., \u003cem\u003eWhy children are not vaccinated: a review of the grey literature\u003c/em\u003e. Int Health, 2012. 4(4): p. 229\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaulson, K.R., et al., \u003cem\u003eGlobal, regional, and national progress towards Sustainable Development Goal 3.2 for neonatal and child health: all-cause and cause-specific mortality findings from the Global Burden of Disease Study 2019\u003c/em\u003e. The Lancet, 2021. 398(10303): p. 870\u0026ndash;905.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerin, J., et al., \u003cem\u003eGlobal, regional, and national causes of under-5 mortality in 2000-19: an updated systematic analysis with implications for the Sustainable Development Goals\u003c/em\u003e. Lancet Child Adolesc Health, 2022. 6(2): p. 106\u0026ndash;115.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhutta, Z.A., et al., \u003cem\u003eInterventions to address deaths from childhood pneumonia and diarrhoea equitably: what works and at what cost?\u003c/em\u003e The Lancet, 2013. 381(9875): p. 1417\u0026ndash;1429.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFischer Walker, C.L., et al., \u003cem\u003eZinc and low osmolarity oral rehydration salts for diarrhoea: a renewed call to action\u003c/em\u003e. Bull World Health Organ, 2009. 87(10): p. 780\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhinkorah, B.O., et al., \u003cem\u003ePrevalence and predictors of oral rehydration therapy, zinc, and other treatments for diarrhoea among children under-five in sub-Saharan Africa\u003c/em\u003e. Plos One, 2022. 17(10).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEzezika, O., et al., \u003cem\u003eBarriers and facilitators to implementation of oral rehydration therapy in low- and middle-income countries: A systematic review\u003c/em\u003e. PLoS One, 2021. 16(4): p. e0249638.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchroder, K., et al., \u003cem\u003eIncreasing coverage of pediatric diarrhea treatment in high-burden countries\u003c/em\u003e. J Glob Health, 2019. 9(1): p. 0010503.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTanahashi T, \u003cem\u003eHealth service coverage and its evaluation\u003c/em\u003e. Bulletin of the World Health Organisation, 1978. 56(2): p. 295\u0026ndash;303.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMwinnyaa, G., et al., \u003cem\u003eEstimating population-based coverage of reproductive, maternal, newborn, and child health (RMNCH) interventions from health management information systems: a comprehensive review\u003c/em\u003e. BMC Health Serv Res, 2021. 21(Suppl 2): p. 1083.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee, J., et al., \u003cem\u003eInterventions to improve district-level routine health data in low-income and middle-income countries: a systematic review\u003c/em\u003e. BMJ Glob Health, 2021. 6(6).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNdabarora, E., J.A. Chipps, and L. Uys, \u003cem\u003eSystematic review of health data quality management and best practices at community and district levels in LMIC\u003c/em\u003e. Information Development, 2013. 30(2): p. 103\u0026ndash;120.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDHIS2. \u003cem\u003eDHIS2\u003c/em\u003e. 2022 [cited 2022 5th June 2022]; Available from: https://dhis2.org/.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaina, I., et al., \u003cem\u003eUsing health-facility data to assess subnational coverage of maternal and child health indicators, Kenya\u003c/em\u003e. Bull World Health Organ, 2017. 95(10): p. 683\u0026ndash;694.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAgiraembabazi, G., et al., \u003cem\u003eCan routine health facility data be used to monitor subnational coverage of maternal, newborn and child health services in Uganda?\u003c/em\u003e BMC Health Serv Res, 2021. 21(Suppl 1): p. 512.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSimmons, E.M., et al., \u003cem\u003eAssessing coverage of essential maternal and child health interventions using health-facility data in Uganda\u003c/em\u003e. Popul Health Metr, 2020. 18(1): p. 26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMa\u0026iuml;ga, A., et al., \u003cem\u003eMeasuring coverage of maternal and child health services using routine health facility data: a Sierra Leone case study\u003c/em\u003e. BMC Health Serv Res, 2021. 21(Suppl 1): p. 547.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalembaka, E.B., et al., \u003cem\u003eThe use of health facility data to assess the effects of armed conflicts on maternal and child health: experience from the Kivu, DR Congo\u003c/em\u003e. BMC Health Serv Res, 2021. 21(Suppl 1): p. 195.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalana, W., et al., \u003cem\u003eChildhood immunization in Ghana: Tracing the history and projecting the future\u003c/em\u003e. Public Health Challenges, 2024. 3(e176).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhana Statistical Service, \u003cem\u003eGhana 2021 Population and Housing Census General Report: Volume 3A Population of Regions and Districts\u003c/em\u003e. 2021, Ghana Statistical Service: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health, \u003cem\u003eNational Community-Based Health Planning and Services (CHPS) Policy\u003c/em\u003e. 2016, Ministry of Health: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health, \u003cem\u003eHealth Information System Strategic Plan 2022\u0026ndash;2025\u003c/em\u003e. 2022, Ministry of Health: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCountdown to 2030 Collaboration. \u003cem\u003eHealth Facility Data \u0026amp; Analysis Center\u003c/em\u003e. 2024 [cited 2024 7th October 2024]; Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.countdown2030.org/tools-for-analysis/health-facility-data-and-analysis\u003c/span\u003e\u003cspan address=\"https://www.countdown2030.org/tools-for-analysis/health-facility-data-and-analysis\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhana Statistical Service, \u003cem\u003eMultiple Indicator Cluster Survey (MICS 2017/18): Survey Findings Report\u003c/em\u003e. 2018, Ghana Statistical Service: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenchimol, E.I., et al., \u003cem\u003eThe REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement\u003c/em\u003e. PLoS Med, 2015. 12(10): p. e1001885.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTroeger, C., et al., \u003cem\u003eEstimates of the global, regional, and national morbidity, mortality, and aetiologies of diarrhoea in 195 countries: a systematic analysis for the Global Burden of Disease Study 2016\u003c/em\u003e. The Lancet Infectious Diseases, 2018. 18(11): p. 1211\u0026ndash;1228.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLumley T, \u003cem\u003eAnalysis of complex survey samples\u003c/em\u003e. Journal of Statistical Software, 2004. 9(8): p. 1\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePebesma E and Bivand R, \u003cem\u003eSpatial Data Science: With Applications in R\u003c/em\u003e. 1st edition ed. 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAnselin L, \u003cem\u003eLocal Indicators of Spatial Association\u0026mdash;LISA\u003c/em\u003e. Geographical Analysis, 1995. 27(2): p. 93\u0026ndash;115.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAnselin L, \u003cem\u003eThe Moran scatterplot as an ESDA tool to assess local instability in spatial association\u003c/em\u003e, in \u003cem\u003eSpatial Analytical Perspectives on GIS\u003c/em\u003e, Fischer MM, Editor. 1996, Routledge: London. p. 111\u0026ndash;125.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCook RD and Weisberg S, \u003cem\u003eResiduals and Influence in Regression\u003c/em\u003e. 1982, New York: Chapman and Hall.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmoro, V.A., G.A. Abiiro, and K.A. Alatinga, \u003cem\u003eBypassing primary healthcare facilities for maternal healthcare in North West Ghana: socio-economic correlates and financial implications\u003c/em\u003e. BMC Health Serv Res, 2021. 21(1): p. 545.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDotse-Gborgbortsi, W., et al., \u003cem\u003eThe influence of distance and quality on utilisation of birthing services at health facilities in Eastern Region, Ghana\u003c/em\u003e. BMJ Glob Health, 2019. 4(Suppl 5): p. e002020.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEscamilla, V., et al., \u003cem\u003eThe Role of Distance and Quality on Facility Selection for Maternal and Child Health Services in Urban Kenya\u003c/em\u003e. J Urban Health, 2018. 95(1): p. 1\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOcholla, I.A., et al., \u003cem\u003eGeographical accessibility in assessing bypassing behaviour for inpatient neonatal care, Bungoma County-Kenya\u003c/em\u003e. BMC Pregnancy Childbirth, 2020. 20(1): p. 287.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYao, J. and V. Agadjanian, \u003cem\u003eBypassing health facilities in rural Mozambique: spatial, institutional, and individual determinants\u003c/em\u003e. BMC Health Serv Res, 2018. 18(1): p. 1006.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKahabuka C, et al., \u003cem\u003eUnfulfilled expectations to services offered at primary health care facilities: Experiences of caretakers of underfive children in rural Tanzania\u003c/em\u003e. BMC Health Services Research, 2012. 12(158).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eManukure, J.M., F. Annor, and A.A. Gyasi-Gyamerah, \u003cem\u003e\u0026lsquo;Leaving the child behind was not easy\u0026rsquo;: lived experiences of mothers returning to work after maternity leave in Ghana\u003c/em\u003e. Community, Work \u0026amp; Family, 2024: p. 1\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhana Health Service, \u003cem\u003eThe Health Sector in Ghana: Facts and Figs. 2017\u003c/em\u003e. 2017.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMa, Y., et al., \u003cem\u003eEvaluating the impact of community health volunteer home visits on child diarrhea and fever in the Volta Region, Ghana: A cluster-randomized controlled trial\u003c/em\u003e. PLoS Med, 2019. 16(6): p. e1002830.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHealth Facilities Regulatory Agency, \u003cem\u003eSpecific Requirements for CHPS Compound, v1.3\u003c/em\u003e. 2017, Health Facilities Regulatory Agency: Accra.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMosser, J.F., et al., \u003cem\u003eMapping diphtheria-pertussis-tetanus vaccine coverage in Africa, 2000\u0026ndash;2016: a spatial and temporal modelling study\u003c/em\u003e. The Lancet, 2019. 393(10183): p. 1843\u0026ndash;1855.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSeidler, V., et al., \u003cem\u003eSubnational variations in the quality of household survey data in sub-Saharan Africa\u003c/em\u003e. Nat Commun, 2025. 16(1): p. 3771.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCutts, F.T., H.S. Izurieta, and D.A. Rhoda, \u003cem\u003eMeasuring coverage in MNCH: design, implementation, and interpretation challenges associated with tracking vaccination coverage using household surveys\u003c/em\u003e. PLoS Med, 2013. 10(5): p. e1001404.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFischer Walker, C.L., O. Fontaine, and R.E. Black, \u003cem\u003eMeasuring coverage in MNCH: current indicators for measuring coverage of diarrhea treatment interventions and opportunities for improvement\u003c/em\u003e. PLoS Med, 2013. 10(5): p. e1001385.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAsare, E.O., J.L. Warren, and V.E. Pitzer, \u003cem\u003eSpatiotemporal patterns of diarrhea incidence in Ghana and the impact of meteorological and socio-demographic factors\u003c/em\u003e. Front Epidemiol, 2022. 2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReiner, R.C., Jr., L.B.D.T.B. Collaborators, and S.I. Hay, \u003cem\u003eThe overlapping burden of the three leading causes of disability and death in sub-Saharan African children\u003c/em\u003e. Nat Commun, 2022. 13(1): p. 7457.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacharia, P.M., et al., \u003cem\u003eDefining service catchment areas in low-resource settings\u003c/em\u003e. BMJ Glob Health, 2021. 6(7).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnson M, et al., \u003cem\u003eA scoping review of the methods used to estimate health facility catchment populations for child health indicators in sub-Saharan Africa\u003c/em\u003e. Population Health Metrics, 2025. 23(11).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCobos Munoz, D., et al., \u003cem\u003eDecentralization of health systems in low and middle income countries: a systematic review\u003c/em\u003e. Int J Public Health, 2017. 62(2): p. 219\u0026ndash;229.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEl-Khoury M, Banke K, and Sloane P, \u003cem\u003eImproved Childhood Diarrhea Treatment Practices in Ghana: A Pre-Post Evaluation of a Comprehensive Private-Sector Program\u003c/em\u003e. Global Health: Science and Practice, 2016. 4(2): p. 264\u0026ndash;275.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGesicho, M.B., M.C. Were, and A. Babic, \u003cem\u003eEvaluating performance of health care facilities at meeting HIV-indicator reporting requirements in Kenya: an application of K-means clustering algorithm\u003c/em\u003e. BMC Med Inform Decis Mak, 2021. 21(1): p. 6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGimbel, S., et al., \u003cem\u003eImproving data quality across 3 sub-Saharan African countries using the Consolidated Framework for Implementation Research (CFIR): results from the African Health Initiative\u003c/em\u003e. BMC Health Serv Res, 2017. 17(Suppl 3): p. 828.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMutale W, et al., \u003cem\u003eImproving health information systems for decision making across five sub-Saharan African countries: Implementation strategies from the African Health Initiative\u003c/em\u003e. BMC Health Services Research, 2013. 13(59).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuhoza, P., et al., \u003cem\u003eA data quality assessment of the first four years of malaria reporting in the Senegal DHIS2, 2014\u0026ndash;2017\u003c/em\u003e. BMC Health Serv Res, 2022. 22(1): p. 18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGithinji, S., et al., \u003cem\u003eCompleteness of malaria indicator data reporting via the District Health Information Software 2 in Kenya, 2011\u0026ndash;2015\u003c/em\u003e. Malar J, 2017. 16(1): p. 344.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSharma, G., et al., \u003cem\u003ePrivate sector engagement for immunisation programmes: a pragmatic scoping review of 25 years of evidence on good practice in low-income and middle-income countries\u003c/em\u003e. BMJ Glob Health, 2024. 8(Suppl 5).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization, \u003cem\u003eEngaging the private health service delivery sector through governance in mixed health systems: strategy report of the WHO Advisory Group on the Governance of the Private Sector for Universal Health Coverage\u003c/em\u003e. 2020, World Health Organization: Geneva.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZiema, S.A. and L. Asem, \u003cem\u003eAssessment of immunization data quality of routine reports in Ho municipality of Volta region, Ghana\u003c/em\u003e. BMC Health Serv Res, 2020. 20(1): p. 1013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWard, K., et al., \u003cem\u003eEnhancing Workforce Capacity to Improve Vaccination Data Quality, Uganda\u003c/em\u003e. Emerg Infect Dis, 2017. 23(13): p. S85-93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhattacharya, A.A., et al., \u003cem\u003eQuality of routine facility data for monitoring priority maternal and newborn indicators in DHIS2: A case study from Gombe State, Nigeria\u003c/em\u003e. PLoS One, 2019. 14(1): p. e0211265.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6790840/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6790840/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eSuboptimal resource allocation limits uptake of essential health services in sub-Saharan Africa. Reliable population denominators and intervention coverage estimates may inform allocation such that health facility resources better match demand. An approach utilising routinely collected health data has previously been used for region-/district-level estimation. This ecological study aimed to explore its use at lower administrative levels, taking Ghana as an example.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eData collected by Ghanaian health facilities from 2017\u0026ndash;2020 were obtained from the District Health Information Management System and used to estimate childhood vaccination coverage and diarrhoea service utilisation in 2020 for the Eastern region, districts therein, and selected sub-districts and facilities. The estimation approach assumed shared utilisation patterns for benchmark/target interventions (providing denominators and numerators, respectively), and was calibrated and evaluated using household survey data.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eRegion-level coverage estimates of 90.3% and 92.4% were obtained for second (Penta2) and third doses of the pentavalent vaccine, and 83.8% for the first dose of the measles-rubella vaccine. The variability and proportion of implausible (\u0026gt;\u0026thinsp;100%) coverage estimates increased through lower levels; the Penta2 district-level median was 93.6% (IQR:87.3%-96.2%) and facility-level median was 96.5% (IQR:92.0%-104.5%). At all levels, diarrhoea service utilisation estimates were lower than the survey-derived estimate; the district-level median was 7.7% (IQR:5.9%-9.6%) and facility-level median was 2.3% (IQR:0.0%-5.5%).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eWhilst this approach is useful for uncovering local variation in service uptake, integrating routinely collected health data with alternative approaches that explicitly recognise inter-facility competition and spatial/aspatial determinants of health-seeking may produce more accurate estimates, better informing facility-level resource allocation.\u003c/p\u003e","manuscriptTitle":"Using routinely collected health data to estimate child health service coverage in Ghanaian health facilities","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-14 05:11:28","doi":"10.21203/rs.3.rs-6790840/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-medicine","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsmed","sideBox":"Learn more about [Communications Medicine](http://www.nature.com/commsmed)","snPcode":"43856","submissionUrl":"https://mts-commsmed.nature.com/cgi-bin/main.plex","title":"Communications Medicine","twitterHandle":"@commsmedicine","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"37e5dfc4-0f40-4cf9-a5dd-715122ea2387","owner":[],"postedDate":"July 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":49814950,"name":"Health sciences/Health care/Public health"},{"id":49814951,"name":"Health sciences/Diseases/Infectious diseases"},{"id":49814952,"name":"Health sciences/Health care/Disease prevention"}],"tags":[],"updatedAt":"2026-01-27T12:01:14+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-14 05:11:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6790840","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6790840","identity":"rs-6790840","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}