The effect of introducing an electronic medical record system on data quality and factors associated with data quality across 187 HIV clinics in Kenya: An interrupted time series analysis from 2011-2018

The effect of introducing an electronic medical record system on data quality and factors associated with data quality across 187 HIV clinics in Kenya: An interrupted time series analysis from 2011-2018 | 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 Research Article The effect of introducing an electronic medical record system on data quality and factors associated with data quality across 187 HIV clinics in Kenya: An interrupted time series analysis from 2011-2018 Beryne Odeny, Orvalho Augusto, Bradley H. Wagenaar, James P. Hughes, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5672455/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: The objective of this evaluation was to estimate the effect of electronic medical record system (EMR) implementation on the quality of data uploaded to the District Health Information System Version 2 platform (DHIS2). Methods: This was an interrupted time series analysis of DHIS2 data quality. Data were extracted from 187 Kenyan health facilities from January 2011 to June 2018 (i.e., spanning 30 quarters). The primary exposure was presence of EMR, and the primary data quality outcomes were quarterly composite discrepancy scores and composite completeness scores. The composite discrepancy score depicted the extent of deviation of observed values from plausible values based on internal consistency checks. Higher discrepancy scores reflected worse data quality. The composite completeness score (CCS score) was a percentage measure of the extent of documentation of pre-selected variables. A 2017 cross-sectional facility survey was used to assess factors associated with data quality. We conducted an interrupted time series analysis to determine changes in the trend of data quality scores before and after EMR implementation. We conducted multivariable linear regression analyses to determine factors associated with data quality. Results: There was no statistically significant level change or effect in composite discrepancy scores comparing pre-EMR period and the post-EMR period. In the cross-sectional analysis, on average health centers had higher composite discrepancy scores compared to dispensaries thus worse data quality (0.066; 95% CI: 0.002-0.130, p=0.045), high volume facilities (>500 patients) had higher discrepancy scores than low volume facilities (0.090; 95% CI: 0.043-0.138, p<0.001), and operating the KenyaEMR system was associated with less discrepancy scores and thus better data quality (0.058; 95% CI: -0.107- -0.008, p=0.024] than the IQCare system. Regarding CCS, there was a significant drop in composite completeness scores (CCS) after transitioning to EMR. The average CCS in the first quarter post-EMR was lower than the average CCS in the quarter preceding EMR implementation (6.96; 95% CI: -9.15 – -4.77, p<0.001). After six quarters post-EMR implementation, CCS declined steadily with an average quarterly change in CCS that was 1.20 percentage points lower than the average quarterly trend pre-EMR (95% CI: -1.70 – -0.69, <0.001). In cross-sectional analysis, health centers (8.16; 95% CI: 3.94 – 12.37, p<0.001) and hospitals (10.39; 95% CI: 5.96 – 14.80, p<0.001), high facility volume (4.54; 95% CI: 1.06 – 8.02, p=0.010) and high HIV burden county (3.95; 95% CI: 0.19 – 7.70, p= 0.039) were associated with higher CCS compared to dispensaries, low facility volume, and low HIV burden, respectively. Conclusions: EMR implementation did not demonstrate evidence for significant positive impact on DHIS2 data quality, as indicated by the lack of improvement in composite discrepancy scores and a drop in composite completeness scores post-EMR implementation. Our findings suggest that EMRs are not sufficient to ensure high-quality data. Facility characteristics (like higher level facility, high volume, and being in a high HIV burden county), and KenyaEMR use appear to be associated with discrepancy and completeness of data. Further research to explore the mechanistic link between EMRs, data quality, and context will be necessary to optimize the use of EMRs to improve data quality in routine health information system data in LMICs. Figures Figure 1 Figure 2 Figure 3 Introduction Robust routine health information systems (RHIS) are essential for optimal health system evaluation, quality improvement, governance, and health management [ 1 – 3 ]. The World Health Organization's (WHO) global digital health strategy aims to “improve health for everyone, everywhere by accelerating the adoption of appropriate digital health” – this includes EMRs [ 4 , 5 ]. Electronic Medical Records (EMRs) are considered essential building blocks for strong RHIS [ 6 , 7 ]. EMRs streamline data capture, data retrieval, and reporting which facilitate data use for clinical decision making and enhanced patient care. Furthermore, EMRs have also been found to reduce data recording errors [ 3 , 8 , 9 ] as well as time/ monetary costs of data management [ 10 – 13 ]. EMRs have a range of purposes crucial to health system strengthening including data reporting, aggregation and management, supporting clinical decision making, and interlinking health departments [ 4 , 5 , 14 – 17 ]. This paper aims to add to the base of evidence on the utility of EMRs as a tool for strengthening data management – specifically, data quality in aggregate data reporting [ 18 , 19 ]. EMR utilization in healthcare in low- and middle-income countries (LMICs) has dramatically increased over the past two decades. The United States (U.S.) President’s Emergency Plan for AIDS Relief (PEPFAR) has funded scale up and implementation of EMRs specifically for HIV care in health facilities throughout Kenya – one of the high-burden HIV countries in sub-Saharan Africa (SSA). Kenya was among the first high HIV burden countries to publish national strategies for EMR integration in the health system [ 4 , 5 ]. In Kenya, the District Health Information System (DHIS2) – a national electronic health information system – has been used for over a decade to house aggregated facility-level data. Version 2 of the program is an open-source platform ( www.DHIS2.org ) [ 20 ]. Health facilities collate patient-level data to prepare summaries that are uploaded to the DHIS2 routinely. The data in DHIS2 are used in planning health service delivery, including planning of health personnel, supply chain management, among other functions, and as such high standards of data quality need to be maintained [ 21 – 23 ]. Before the introduction of EMRs in Kenya, facility-level HIV data were captured on paper charts and registries. These paper records were used to generate HIV-specific facility summaries for the DHIS2. Phased introduction of EMRs in 2012 led to progressive migration from paper records to a hybrid implementation of paper and electronic records in most HIV facilities. According to programmatic records by mid-2018, about 23% of target facilities have fully transitioned to EMR and 75% are using both paper and EMRs. Collation of electronic data was found to simplify the aggregation of facility-level HIV data, thus use of EMRs to generate HIV-specific facility reports for the DHIS2 expanded as EMRs became widespread. While the DHIS2 hosts data from other sectors beyond HIV service delivery, EMRs are primarily tailored for HIV data management, thus EMR reports uploaded to DHIS2 are HIV-specific. While EMRs are associated with better data quality [ 24 , 25 ], they are not immune to error [ 18 ], thus Routine Data Quality Assessments (RDQAs) are essential to uphold EMR data quality [ 26 ]. Improved data quality in EMRs would potentially lead to improved quality of aggregate data uploaded to the DHIS2. To our knowledge, there is little high-quality evidence on the longitudinal effects of EMRs on data quality in aggregated data systems such as DHIS2 in LMICs. Using an interrupted time series design, we aim to evaluate whether introduction of EMRs improved data quality of the aggregate HIV data in the DHIS2. Specifically, we assessed data quality as captured by a “composite discrepancy scores” and “composite completeness scores.” Composite discrepancy scores (a plausibility check) encapsulate the degree of discrepancy or deviation from the expected values across groups of indicators and related data checks [ 27 ]. The composite completeness score (CCS score) is a percentage measure of the extent of documentation for pre-selected variables. The objectives of this evaluation are two-fold: 1) to evaluate the evidence for effects of EMR implementation on composite discrepancy and completeness scores for HIV data across 187 health facilities in Kenya from January 2011 to June 2018; and 2) to assess facility and EMR implementation correlates of data quality (i.e., reflected by the composite discrepancy and completeness scores) using data from a cross-sectional facility survey conducted in 2017. Methods Setting This was a quasi-experimental study utilizing time series data from the national DHIS2 RHIS in Kenya. Data was extracted quarterly from January 2011 to June 2018 for 187 facilities implementing EMRs – specifically, KenyaEMR and IQCare EMR software. Data sources Data sources were the DHIS2 database and I-TECH and Palladium program records on EMR deployment and implementation process. I-TECH and Palladium were the two predominant EMR technical assistance providers in Kenya during the timeframe of this analysis. Program records were used to capture the date of KenyaEMR or IQCare deployment at each health facility. The primary exposure is presence of EMR (a binary pre or post variable). The primary outcomes were composite discrepancy and completeness scores, as defined below. Data checks and HIV-related indicators [ 28 ] We used HIV-related indicators – encompassing general adult and pediatric HIV care, antenatal care (ANC), Labor & Delivery care (L&D), and Prevention of mother-to-child Transmission of HIV (PMTCT) – that were uploaded monthly to KHIS. The data consisted of aggregate health service utilization statistics by department. Appropriate data checks were determined a priori, and these checks primarily summarized relationships between indicators to ensure the data were complete, consistent, and plausible. For example, one data check compared the total number of patients in HIV care in a specific quarter versus the total number on ART in that quarter. The difference between the two indicator values was expected to be zero or greater (the logic being that those enrolled would always be more or equal to those receiving ART). A series of data checks were used to construct composite scores for each unique facility. Table 2 below summarizes the data checks and indicators explored in this analysis. As outlined in Table 2 , four ANC and PMTCT data checks (#1 – #4) were computed as differences between related indicator values. Similarly, four general HIV care data checks (#5 – #8) were computed as differences between general HIV care indicator values. The difference for all data checks were expected to be greater than or equal to zero (except data check # 8 which was expected to only be 0, Table 2 ). Computing the composite discrepancy score The composite discrepancy score was based on assessment of plausibility across indicators and eight predefined data checks. Individual data checks were derived by comparing groups of indicators. For example, one data check was derived by comparing the total number of patients currently on ART (an indicator) minus sum of patients on ART across all age groups (another indicator) and this was ideally equal to zero; the further away the data check value was from the expected value (zero in this case), the higher the absolute difference between the observed and expected values, thus a higher individual discrepancy score which meant worse data quality (Appendix 1). These individual scores were computed for each data check and were based on Z -score deviations that depicted the extent of discrepancy of observed values i.e., how far observed values were from expected values. The composite discrepancy score was a continuous variable computed as an average of all the Z- scores for individual data logic checks, for each unique facility quarter. For additional context, a score that is 0.01 lower, is a comparatively small change but represents a small improvement in data quality performance, and vice versa. Additional details on the development of this score are provided in the Appendix 2 and in a paper detailing the composite score development [ 28 ]. Computing the composite completeness score The completeness score aims to determine whether aggregate data on the two HIV indicators that constitute a particular data check are present or missing (Table 2 ). If data on the indicators are missing, it means that individual patient data was not aggregated into the summaries that are uploaded to the KHIS. For example, data check 1 compares two indicators: 1) the number of women tested for HIV in antenatal care (ANC) and 2) the number of women testing HIV positive in ANC. When both values are missing, the data check will miss a value for that specific facility. This is a measure of completeness that looks at two related variables that should both be present. The completeness score was based on the proportion of data checks which were complete for each observation i.e., each unique facility quarter. A binary score of 1 or 0 was assigned to each data check based on the presence or missingness of the data (1 = present, 0 = missing). The completeness score was a continuous score computed as the proportion (percent) of data checks with complete data for each observation (i.e., unique facility-quarters). The minimum possible score was 0% (0 complete out of 8 checks) and the maximum possible was 100% (8 complete out of 8 checks) [ 28 ]. Cross-sectional survey A cross-sectional facility survey done in October to December of 2017 was used to collect data on non-time varying facility and EMR implementation factors which are potentially correlated with data quality such as facility level, facility volume, and EMR type. Survey data on EMR implementation characteristics was only available for 115 health facilities. Ethical considerations This evaluation was approved by the AMREF Ethical Scientific Review Committee (ESRC), Kenya, United States (US) Center for Disease Control and Prevention (CDC), and the University of Washington (UW) Human Subjects Review committee. Statistical Analysis Descriptive statistics An analysis plan was prepared prospectively prior to data analysis. Descriptive statistics were used to analyze the distribution of health facility characteristics. We reported the number of facilities included in the analysis by facility level (national hospital, county hospital or sub-county hospital, health center, dispensary), facility volume ( ≥ 500 versus < 500 HIV patients in care), region/county (46 counties classified into low and high-HIV burden counties), and EMR type (KenyaEMR, IQCare or other). Median and interquartile ratios were used for continuous data, whereas proportions were used for binary and categorical data. Baseline characteristics were summarized and presented in Table 1 . Proposed impact model [ 29 ] The outcomes, composite discrepancy and completeness scores, were hypothesized to be responsive to the intervention of EMR implementation on a gradual basis. Based on the literature, we hypothesized a 6-quarter (i.e., 18 months) time lag after EMR introduction to account for the “wash-in period” after EMR introduction. It was expected that EMR implementation would improve over time until optimal functioning was achieved [ 15 , 24 ]. Therefore, a gradual change in the composite discrepancy score was anticipated over the 6-quarter “wash-in period”, followed by a plateauing of the intervention effect after 6 quarters as demonstrated in Fig. 1 (impact model). The final impact model which informed the statistical model and analysis, excluded the “wash-in period”, thus only two segments were modeled: pre-EMR, and post-EMR (starting after 6 quarters post-intervention). Inferential analyses using Interrupted Time Series (ITS) Multilevel models were used to control for clustering at the health facility level. Changes in the data quality scores pre- and post- EMR implementation were estimated using generalized linear models (GLM) with Gaussian distribution and identity link and bootstrapped standard errors. Random effects for facilities (including random intercepts) were used to account for intra-cluster correlation. Data was assumed to be missing at random. We included data points beginning at 14 quarters pre-EMR implementation. The primary model included dummy variables for quarter (i.e., season) to account for seasonality and random intercept to account for facilities having different baseline DQ scores. Table 1 Definition of covariates in the final ITS model Intercept : baseline mean DQ score at time = 0 (i.e., Quarter 1, 2011) for an average facility. Time : the mean change in DQ score per quarter before EMR implementation for an average facility. Season indicator : the difference in mean DQ score comparing each Quarter X to Quarter 1 (reference) for an average facility. EMR : mean change in DQ score after the 6-quarter “wash-in” period compared to DQ score immediately before EMR introduction for an average facility. This is the level change. Time_after_EMR : Slope in the post-EMR period, after the wash-in period of six months. This is the mean change in score each quarter post-EMR implementation compared to quarterly trend before EMR implementation for an average facility. This is the slope change. Cross-sectional survey analysis: Linear regression models were used to assess the relationship between EMR and facility characteristics and the composite discrepancy and completeness scores (as a measure of data quality). We used the composite discrepancy and completeness scores from quarter 4 of June 2017, i.e., October to November 2017. We accounted for clustering of measures at the facility level. By this time, all facilities had an EMR and the survey was conducted post-EMR. A priori, based on programmatic relevance, we selected EMR type, facility volume, and being in a high-burden county to be included in the multivariable model. We accounted for clustering at the facility level and assessed for interaction between the facility level and facility volume. We conducted a sensitivity multivariable analysis in which we used the full model. Health facilities missing data on facility and EMR implementation characteristics were excluded from the analysis, i.e., complete case analyses. All statistical analyses for this evaluation were done using R version 3.6.2 (2019-12-12). Results Based on the 2017 cross-sectional survey, most facilities (73%) in the sample were health centers and sub-county hospitals (Table 2 ). Fifty six percent (56%) of facilities were high volume, attending to more than 500 patients in their HIV clinics. Thirty six percent (36%) of facilities were in high-HIV burden counties. Most facilities in this sample implemented KenyaEMR software (56%). With regard to EMR implementation characteristics, 55% facilities used retrospectively entered data after completing paper forms. Most facilities (55%) reported at least one routine data quality assessment exercise that year. Only 17% of facilities reported equipment failure in the preceding week, and most facilities had at least one or two people trained to use EMR. --- INSERT Table 2 --- Table 2 Health Facility (n = 187) and EMR implementation characteristics (n = 115) Health facility characteristics (n = 187) n (%) Facility level Dispensary 45 (24) Health Center 68 (36.5) County/ Sub-County Hospital 68 (36.5) Missing 6 (3) Facility volume Low volume (< 500) 105 (42) High volume ( ≥ 500) 79 (56) Missing 3 (2) High HIV burden county No 120 (64) Yes 67 (36) EMR type International Quality Care (IQCare) 82 (44) KenyaEMR 105 (56) EMR implementation characteristics (data from N = 115 facilities) Implementation mode Hybrid † 23 (20) Point of care entry 26 (23) Retrospective data entry 64 (55) Uncategorized 2 (2) Annual number of RDQAs ⁑ 0 27 (23) 1 63 (55) ≥ 2 9 (8) Missing 16 (14) Equipment failure (past week) No 95 (83) Yes 20 (17) † Hybrid mode – facility practices both point of care entry and retrospective data entry ⁑ RDQAs- Routine Data Quality Assessments within the year preceding the survey COMPOSITE DISCREPANCY SCORES Table 3 provides a descriptive summary of the discrepancy score which had a mean (SD) of 0.07 (0.22), and a median (IQR) of 0.02 (0-0.06). Table 3 Summary of composite discrepancy scores Total N = 187 facilities over 30 quarters Score approach Mean (SD) Median (IQR) Minimum Maximum Discrepancy score 0.07 (0.22) 0.02 (0–0.06) 0 6.16 SD – Standard Deviation; IQR – interquartile Range The interrupted time series graph, centered on time of EMR implementation, shows an initial drop in discrepancy scores between 14 and 13 quarters pre-EMR (Fig. 2 ) likely due to a third of facilities contributing to data in these quarters; however, scores have remained stable over the duration of the evaluation signaling minimal changes to the trend of data quality before and after EMR implementation. --- INSERT FIGURE 2 --- Table 4 summarizes the parameters of the time series model. The average baseline composite discrepancy score at EMR implementation (centered time = 0) was 0.0753 [ 95% confidence interval (CI): 0.0482–0.1025]. In the pre-EMR period, the average quarterly discrepancy score decreased by 0.0012 points (95% CI: -0.0045 – +0.0020, p-value = 0.458), i.e., there was no significant trend. Regarding level change, comparing the values post-EMR, and pre-EMR period; the average composite discrepancy score in the first quarter post-EMR was 0.0118 points higher [ 95% CI: (-0.0248–0.0484), p-value = 0.528] than the average score in the quarter preceding EMR implementation for the average facility. This level change was not statistically significant. Comparing the trends post- and pre-EMR implementation; after six quarters post-EMR, the average change in the composite discrepancy score was 0.0004 points higher (95% CI: -0.0038–0.0047, p-value = 0.844 than the average quarterly trend pre-implementation for the average facility. This change was not significant (Table 4 ). Average discrepancy scores were compared across the four seasons in the pre-EMR period. Season did not predict level of discrepancy score when comparing other seasons with the first calendar quarter (Table 4 ). --- INSERT Table 4 --- Table 4 Interrupted time series analysis of crude time trends in discrepancy scores across 187 health facilities with EMRs, 2011–2018, Kenya β*(95% CI) p-value Discrepancy score Average quarterly change in pre-EMR period (slope) -0.0012 (-0.0045 – +0.0020) 0.458 Level change following EMR implementation 0.0118 (-0.0248 – +0.0484) 0.528 Average quarterly change in 6 quarters post-EMR (slope) 0.0004 (-0.0038–0.0047) 0.844 Baseline score at EMR implementation † 0.0753 (0.0482–0.1025) -- Average quarterly change in reference to season 1 (January to March) Season 2 - April to June -0.0173 (-0.0353 – -0.0007) 0.059 Season 3 - July to September -0.0097 (-0.0287 – +0.0092) 0.315 Season 4 - October to December -0.0104 (-0.0292 – +0.0084) 0.278 Computation of Intracluster correlation coefficient (ICC) Between-cluster variation 0.015 Within- cluster variation 0.034 ICC 0.306 β*represents unadjusted quarterly change in mean DQ score using generalized linear mixed models with random intercept for clinics. Time was centered at date of EMR deployment. Time was modeled with a wash-in period of 6 quarters after EMR introduction † Baseline quarter in the time series at EMR implementation With regard to the post-EMR cross-sectional survey of health facility and EMR implementation factors associated with discrepancy scores, significant findings in unadjusted and adjusted analysis have been described below and in Table 5 . In unadjusted analysis, facility type, facility volume, being in a high HIV-burden county, and EMR type were associated with data quality. After adjustment, all these factors remained significant except for the association with high HIV-burden county. Health centers, on average, had a composite discrepancy score that was 0.066 (95% CI: 0.002–0.130) points higher than that of dispensaries, thus a higher degree of discrepancy indicating poorer data quality (p-value = 0.045). High volume facilities (> 500 patients on ART,) on average, had a score that was 0.090 (95% CI: 0.043–0.138) points higher than that of low volume facilities, thus a higher degree of discrepancy indicating poorer data quality (p-value < 0.001). Facilities with KenyaEMR, on average, had a score that was 0.058 (95% CI: -0.107- -0.008) points lower than that of facilities with IQCare, thus a lower degree of discrepancy indicating better data quality (p-value = 0.024). Implementation mode (whether point of care, retrospective data entry, or hybrid), annual number of RDQAs, and having experienced equipment failure in the preceding week had no significant association with the discrepancy score. --- INSERT Table 5 --- Table 5 Facility and EMR implementation correlates of composite discrepancy scores post-EMR – cross-sectional analysis of survey done in 2017, quarter 4 (October – December) N = 187 Characteristic N (%) Median (IQR) β*-unadjusted (95% CI) p-value β ** - adjusted (95% CI) p-value Facility level*** Dispensary 45 (24) Ref Ref Health Center 68 (36.5) 0.066 (0.002–0.130) 0.045 0.093 (0.026–0.159) 0.007 County/Sub-county hosp 68 (36.5) 0.042 (-0.022-0.106) 0.197 0.045(-0.25-0.114) 0.212 Missing 6 (3) Facility volume < 500 105 (42) Ref Ref ≥ 500 79 (56) 0.090 (0.043–0.138) < 0.001 0.084 (0.033–0.136) 0.001 Missing 3 (2) High HIV burden county No 120 Ref Ref Yes 67 0.065 (0.015–0.116) 0.011 0.042 (-0.011-0.094) 0.122 EMR type IQCare 82 (44) Ref Ref KenyaEMR 105 (56) -0.058 (-0.107- -0.008) 0.024 -0.082 (-0.132- -0.031) 0.002 EMR implementation characteristics (data from N = 115 facilities) Implementation mode Hybrid † 23 (20) Ref Ref Point of care entry 26 (23) -0.024 (-0.137-0.090) 0.684 -0.012 (-0.121-0.097) 0.827 Retrospective data entry 64 (55) 0.016 (-0.079-0.111) 0.739 -0.012 (-0.109-0.085) 0.808 Uncategorized 2 (2) -0.037 (-0.326-0.252) 0.801 -0.013 (0.033–0.184) 0.945 Annual number of RDQAs ⁑ 0 27 (23) Ref Ref 1 63 (55) -0.033 (-0.131-0.065) 0.511 -0.039 (-0.135-0.057) 0.426 ≥ 2 9 (8) 0.009 (-0.154-0.171) 0.916 -0.008 (-0.172-0.156) 0.922 Missing 16 (14) Equipment failure (past week) No 95 (83) Ref Ref Yes 20 (17) -0.020 (-0.118-0.078) 0.684 -0.034 (-0.128-0.060) 0.474 β*represents the crude relationship between DQ score and key facility and EMR implementation factors β** represents the adjusted relationship with DQ score adjusting for being in a high vs low HIV burden county, facility volume, and EMR type † Hybrid mode – facility practices both point of care entry and retrospective data entry in parallel ⁑ RDQAs- Routine Data Quality Assessments ANALYSIS OF COMPOSITE COMPLETENESS SCORES Table 6 Summary statistics of Composite Completeness Scores (CCS) Total N = 187 facilities over 30 quarters Score approach Mean (SD) Median (IQR) Minimum Maximum CCS score (%) 57.8 (28.4) 75.0 (37.5–75.0) 12.50 100 SD – Standard Deviation; IQR – interquartile Range Table 7 Interrupted time series analysis of crude time trends in Composite Completeness scores (CCS scores) across 187 health facilities with EMRs, 2011–2018, Kenya β*(95% CI) p-value CCS score (%) Average quarterly change in pre-EMR period (slope) 5.05 (4.86–5.23) < 0.001 Level change following EMR implementation -6.96 (-9.15 – -4.77) < 0.001 Average quarterly change in 6 quarters post-EMR period (slope) -1.20 (-1.70 – -0.69) < 0.001 Baseline score at EMR implementation 79.92 (77.93–81.91) -- Average quarterly change in reference to season 1 (January to March) Season 2 - April to June -2.21 (-3.47 – -0.95) 0.001 Season 3 - July to September 0.33 (-0.97 – +1.64) 0.615 Season 4 - October to December -0.43 (-1.74 – +0.88) 0.520 Computation of Intracluster correlation coefficient (ICC) Between-cluster variation 63.86 Within- cluster variation 352.13 ICC 0.15 β*represents unadjusted quarterly change in mean DQ score using generalized linear mixed models with random intercept for clinics. Time was centered at date of EMR deployment The average baseline CCS score for the average facility at EMR implementation was 79.92% (95% CI: 77.93–81.91). Pre-EMR, the average quarterly CCS score trend was significantly increasing by 5.05 percentage points per quarter (95% CI:4.86–5.23, p < 0.001). See Table 7 . There was a significant level change after transitioning to EMR. The average CCS score in the first quarter post-EMR was 6.96 percentage points lower than the average CCS score in the quarter preceding EMR implementation for the average facility (95% CI: (-9.15 – -4.77, p < 0.001). After 6 quarters post-EMR implementation, CCS scores declined and the average quarterly change in CCS was 1.20 percentage points lower than the average quarterly trend pre-EMR (95% CI: -1.70 – -0.69, < 0.001). CCS scores were compared across the four seasons in the pre-EMR period. The average quarterly CCS score in the April - June season was 2.21 percentage points lower than that in the January – March season; this was significant (95% CI: -3.47 – -0.95, p = 0.001). There were no significant differences in average CCS scores when comparing the January – March season with the other seasons (i.e., July – September and October – December). Table 8 Facility and EMR implementation characteristics associated with Composite Completeness scores (CCS) in the October-December 2017 cross-sectional survey N = 187 Characteristic N (%) Median (IQR) β*-unadjusted (95% CI) p-value β ** - adjusted (95% CI) p-value Facility level*** Dispensary 45 (24) Ref Ref Health Center 68 (36.5) 6.49 (2.34–10.65) 0.002 8.16 (3.94–12.37) < 0.001 County/Sub-county hosp 68 (36.5) 9.43 (5.28–13.59) < 0.001 10.39 (5.96–14.80) < 0.001 Missing 6 (3) Facility volume*** < 500 105 (42) Ref Ref ≥ 500 79 (56) 7.61 (4.31–10.91) < 0.001 4.54 (1.06–8.02) 0.010 Missing 3 (2) High HIV burden county No 120 Ref Ref Yes 67 2.68 (-0.87–6.22) 0.139 3.95 (0.19–7.70) 0.039 EMR type IQCare 82 (44) Ref Ref KenyaEMR 105 (56) 2.16 (-1.32–5.65) 0.224 2.67 (-0.61–5.95) 0.111 EMR implementation characteristics (data from N = 115 facilities) Implementation mode Hybrid † 23 (20) Ref Ref Point of care entry 26 (23) 4.54 (-2.73–11.81) 0.224 3.75 (-3.37–10.87) 0.302 Retrospective data entry 64 (55) 3.44 (-2.73–9.61) 0.227 5.18 (-1.99–12.35) 0.157 Uncategorized 2 (2) -0.27 (-18.99–18.45) 0.977 -1.19 (-26.55–24.16) 0.926 Annual number of RDQAs ⁑ 0 27 (23) Ref Ref 1 63 (55) 6.61 (0.52–12.71) 0.033 6.05 (-0.96–13.06) 0.093 ≥ 2 9 (8) 11.57 (1.38–21.77) 0.026 9.77 (-1.95–21.48) 0.102 Missing 16 (14) Equipment failure (past week) No 95 (83) Ref Ref Yes 20 (17) -0.10 (-6.34–6.14) 0.975 -0.15 (-6.76–6.47) 0.965 ⁑ RDQAs- Routine Data Quality Assessments; CI – Confidence interval; IQR – Interquartile range β*represents the crude relationship between DQ score and key facility and EMR implementation factors β**For facility characteristics, the multivariable model included facility level, being in a high vs low HIV burden county, facility volume, and EMR type. For EMR characteristics, the multivariable model included facility level, being in a high vs low HIV burden county, facility volume, EMR type, mode of data entry, annual number of RDQAs and equipment failure. ***No interaction found between facility level and facility volume. † Hybrid mode – facility practices both point of care entry and retrospective data entry in parallel All facility level characteristics, except EMR type, were significantly associated with the composite completeness scores. In multivariable analysis, health centers and hospitals had higher completeness compared to dispensaries i.e., 8.16 percentage point higher (95% CI: 3.94–12.37, p < 0.001) and 10.39 percentage point higher (95% CI: 5.96–14.80, p < 0.001) completeness, respectively. Higher facility volume was associated with a 4.54 percentage point higher completeness compared to low volume facility (95% CI: 1.06–8.02, p = 0.010). Being in a high HIV burden county was associated with a 3.95 percentage point higher completeness (95% CI: 0.19–7.70, p = 0.039) compared to being in a low HIV burden county. Among EMR implementation characteristics, the only characteristic that was associated with higher CCS in bivariable analysis was having one or more annual RDQAs. Discussion In this evaluation we evaluated the impact of EMR implementation on data quality in the DHIS-2 system, as captured by composite discrepancy scores and CCS, across 187 facilities in Kenya from 2011 to 2018. Overall, changes in the discrepancy score comparing different periods pre- and post-EMR were very small (most estimates were smaller than 0.01) and thus within 1 Z- score i.e., 1 standard deviation (SD), and CCS seemed to significantly decline after EMR implementation. We found no evidence that EMR implementation was associated with favorable changes in DHIS2 in the discrepancy scores and CCS. This could be attributed to the low (favorable) baseline discrepancy scores and prior local efforts to improve completeness in paper-based systems before EMRs were introduced. A cross-sectional facility survey done in 2017 unveiled health facility and EMR implementation factors that were associated with discrepancy and completeness of data. While the differences in Z-score deviations were mostly less than 0.1, the differences were significant enough to reveal associated factors. High HIV burden counties were associated with worse data quality (high discrepancy scores), potentially due to high workload and collateral increase in documentation burden. Operating KenyaEMR software was associated with lower discrepancy scores – thus better data quality – compared to using IQCare. The underlying reasons could be usability profiles of the two EMR systems. KenyaEMR was more closely aligned to the paper-based MOH compared to IQCare. With regard to CCS, higher level facilities (i.e., health centers and hospitals), high volume clinics and high HIV burden counties were associated with higher completeness scores. This could have been due to access to more resources, staff (like data managers) and frequency of supervision visits in busier health facilities [ 30 , 31 ]. We found that mode of implementation (i.e., point-of-care, retrospective or hybrid data entry) was not associated with discrepancy or completeness. A small proportion of facilities (23%) maintained fidelity to the intended mode of EMR use, i.e., point-of-care entry. There is existing research describing data quality in the context of electronic data systems, but less research examining the evidence for causal effects of EMR implementation on data quality. Previous studies in sub-Saharan Africa have shown a mixed picture in terms of levels of data accuracy for HIV-related indicators. Several studies have shown high data quality of EMR records [ 9 , 30 , 32 – 37 ], while others have demonstrated overall poor EMR data quality [ 38 ]. Notably, there were no studies that compared data quality in EMRs and paper records longitudinally. A pre-post study in Iran found that vocal-electronic documentation was associated with higher quality data than paper documentation [ 39 ]. Two cross-sectional studies compared electronic and paper records concurrently. These studies found that data quality was comparable across the electronic and paper records [ 36 , 40 ]. A one-month randomized controlled study in Ethiopia demonstrated better data quality and efficiency with electronic compared to paper-based survey data [ 41 ]. Our lack of positive findings suggest that presence of EMRs is not insufficient to achieve strong aggregated data quality used in reporting; optimization of EMR usage and data exchange is a necessary requirement for enhancing data quality [ 42 ]. Some barriers to improved data quality during EMR implementation include sub-optimal EMR use which may be associated with insignificant changes in data quality post-EMR implementation [ 30 ]. Recent studies evaluating the actual use of EMR implementation have demonstrated concerningly low EMR usage and data exchange [ 30 , 43 ]. Ngugi et al, and colleagues measured seven EMR usage indicators and found gaps in active use of EMRs and data completeness in Kenya [ 43 ]. Factors affecting EMR usage need further exploration and remedial interventions [ 26 , 36 , 38 , 44 , 45 ]. An important consideration is variability in the approach health workers use to generate EMR aggregate summaries that are reported in DHIS2. It is likely that some health workers may manually transcribe EMRs summaries for retrospective upload to DHIS2, while others may not use the EMR and thus the EMR and DHIS2 would run parallel to each other. Dual/ parallel documentation in EMR and paper records is huge barrier to EMR usage and good data quality. In this evaluation, we found that 75% of facilities had a hybrid or retrospective data entry system – they performed both point-of-care and retrospective data entry, which would contribute to inefficiencies and errors in data management. A study in Malawi found that retrospective data entry by clerks was associated with more errors than point of care entry by clinicians [ 46 ]. Assisting facilities to channel efforts toward point-of-care EMR usage would be a possible solution for improving data quality [ 8 , 46 , 47 ]. A study in Kenya established the benefit of RDQAs in improving EMR data quality [ 25 ]. While the sub-analysis of factors in the present evaluation showed that there was no association between RDQAs and data quality, it is acknowledged that the cross-sectional survey may have had inherent bias and insufficient power to detect all significant associations. This evaluation points to a constellation of factors beyond the presence of EMR that may impact DHIS2 data quality. Other factors that may have been at play include facility level, facility volume, HIV burden, EMR type, and frequency of RDQAs as demonstrated by the cross-sectional survey. An important consideration is variability in the approach health workers used to generate EMR aggregate summaries that are uploaded to DHIS2. It is likely that some health workers manually transcribed EMRs summaries for retrospective upload to DHIS2, while others did not use the EMR summaries, and thus the EMR summaries and DHIS2 reports would not be synchronized. In resource limited settings, it can take a considerable duration of time to successfully implement and sustain EMR software – this should be kept in mind when evaluating EMRs [ 48 ]. A parallel focus on EMR usage, sustainability, and contextual factors is necessary for optimizing RHIS in LMICs [ 49 ]. EMR implementation is feasible; however, the challenges of streamlining use of EMR, and lack of interoperability with DHIS2, are considerable. Other potential implementation challenges include health provider and health system constraints, and EMR set up and maintenance costs [ 7 , 49 – 51 ]. More ongoing work is needed to understand systemic deficits that hinder health provider acceptance, uptake, and meaningful use of EMRs [ 52 ]. It can be hypothesized that poor uptake and use of EMR would limit realization of the benefits of EMRs for DHIS2 data reporting. This work should be coupled with continuous quality improvement efforts and design of contextually appropriate implementation strategies to strengthen the uptake of EMRs. Strategies such as RDQAs – as demonstrated in this study –, training health providers, infrastructure support, modifying electronic tools, designing usable and interoperable systems, and sustainable technical support are critical for optimal implementation of EMRs [ 6 , 53 – 55 ]. Strengths: This evaluation serves as a blueprint for future longitudinal evaluations of mature EMRs. To our knowledge this is the first nationwide longitudinal analyses of data quality that spans pre-EMR and post-EMR periods that assesses the causal relationship between EMR implementation and aggregate data quality. Prior studies in low- and middle-income countries (LMICs) have considerable limitations including small sample sizes, had shorter follow-up periods or cross-sectional designs, use of less robust designs, or lack long-term comparative evaluations of EMRs and paper systems. In light of this, the present evaluation is a much-needed addition to the literature. Another strength includes the use of the time series models which eliminate confounding by time invariant factors. The stepped/ differential introduction of EMRs across multiple facilities by default creates a multiple baseline design which is useful in controlling for possible time-varying confounders [ 56 ]. Limitations: While we present a strong quasi-experimental design, it is not without limitations. First, while we aim to obtain evidence for causal effects, we recognize that this robust study design is still susceptible to confounding by time varying factors and that some of the assumptions for the time series analysis may not have been fulfilled. Examples of time varying factors include recurrent staff turnovers and program-level changes to data documentation and tools. Second, we were not able to obtain granular monthly data, but were restricted to quarterly data until 2018 which limits our understanding of the month-to-month variations in data quality to date. Third, the cross-sectional survey design, for understanding facility and EMR implementation correlates of data quality, is vulnerable to selection bias and residual confounding inherent to the design. We were also not able to understand the level of interoperability between EMR and DHIS2 and how providers navigated these technical issues to generate DHIS2 summaries. Given power calculations were not used to determine the sample size for the survey, the cross-sectional analysis may have been underpowered to detect all significant associations between facility or EMR implementation factors and data quality. Finally, this evaluation focused on HIV-specific aspects of DHIS2, thus findings cannot be generalized to other health service departments, and further research would be needed to assess effects beyond HIV service delivery. Conclusion With regard to implications for research, practice, and policy, we found no evidence to suggest that EMRs in HIV facilities in Kenya improved aggregated data quality as captured by the discrepancy and completeness scores. This does not detract from the multiple mechanisms through which EMRs could have effects on quality of care and health outcomes. We found that higher facility level, higher facility volume, and KenyaEMR type were associated with lower discrepancy; and facility level, higher patient volume, high HIV burden were associated with higher completeness. These findings suggest that the presence of EMRs is insufficient to achieve high DHIS2 data quality. Further understanding how these factors interact with EMRs and data management processes would be necessary to optimize the value of EMRs and DHIS2 data quality. In the interim, implementation research is needed to understand factors associated with sustained EMR use and data quality, and further characterize EMR implementation gaps. Further research is needed to track the impact of strategies to augment EMR-DHIS2 interoperability for better data quality beyond the studied time frame. Declarations Author Contribution B.M.O. conception, design, analysis, prepared figures and tables, and wrote the manuscript. O.A. design, analysis and interpretation of data. B.H.W. interpretation of the data and revision of manuscript. J.P.H. design, analysis, and interpretation of data. S.G. interpretation of data. A.N. conception, acquisition, and analysis. N.P. conception, design, interpretation, and revision of manuscript. All authors reviewed the manuscript and approved its submission for publication. Acknowledgement We would like to acknowledge the Ministry of Health-Kenya, I-TECH, and Palladium for facilitating access to the data for analysis. Data Availability The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request. References Aqil A, Lippeveld T, Hozumi D. PRISM framework: a paradigm shift for designing, strengthening and evaluating routine health information systems. Health Policy Plan. 2009;24:217–28. 10.1093/heapol/czp010 . Wagenaar BH, Sherr K, Fernandes Q, Wagenaar AC. Using routine health information systems for well-designed health evaluations in low-and middle-income countries. Health Policy Plan. 2016;31:129–35. Leon N, Balakrishna Y, Hohlfeld A, Odendaal WA, Schmidt B-M, Zweigenthal V, et al. Routine Health Information System (RHIS) improvements for strengthened health system management. Cochrane Database Syst Rev. 2020;8:CD012012. 10.1002/14651858.CD012012.pub2 . WHO. Global diffusion of eHealth: making universal health coverage achievable: report of the third global survey on eHealth. World Health Organization; 2017. WHO. WHO guideline: recommendations on digital interventions for health system strengthening: web supplement 2: summary of findings and GRADE tables. World Health Organization; 2019. WHO. Electronic health records: manual for developing countries. Manila: WHO Regional Office for the Western Pacific; 2006. Jacob N, Rice B, Kalk E, Heekes A, Morgan J, Brinkmann S, et al. Consolidating strategic information to monitor progress against the UNAIDS 90-90-90 targets: evaluating the operational feasibility of an electronic HIV testing register in Cape Town, South Africa. BMC Health Serv Res. 2020;20:720. 10.1186/s12913-020-05517-7 . Ley B, Rijal KR, Marfurt J, Adhikari NR, Banjara MR, Shrestha UT, et al. Analysis of erroneous data entries in paper based and electronic data collection. BMC Res Notes. 2019;12:537. 10.1186/s13104-019-4574-8 . Castelnuovo B, Kiragga A, Afayo V, Ncube M, Orama R, Magero S, et al. Implementation of provider-based electronic medical records and improvement of the quality of data in a large HIV program in Sub-Saharan Africa. PLoS ONE. 2012;7:e51631. 10.1371/journal.pone.0051631 . McLean E, Dube A, Saul J, Branson K, Luhanga M, Mwiba O, et al. Implementing electronic data capture at a well-established health and demographic surveillance site in rural northern Malawi. Glob Health Action. 2017;10:1367162. 10.1080/16549716.2017.1367162 . Mukasa O, Mushi HP, Maire N, Ross A, de Savigny D. Do surveys with paper and electronic devices differ in quality and cost? Experience from the Rufiji Health and demographic surveillance system in Tanzania. Glob Health Action. 2017;10:1387984. 10.1080/16549716.2017.1387984 . Flaxman AD, Stewart A, Joseph JC, Alam N, Alam SS, Chowdhury H, et al. Collecting verbal autopsies: improving and streamlining data collection processes using electronic tablets. Popul Health Metr. 2018;16:3. 10.1186/s12963-018-0161-9 . Singh Y, Jackson D, Bhardwaj S, Titus N, Goga A. National surveillance using mobile systems for health monitoring: complexity, functionality and feasibility. BMC Infect Dis. 2019;19:786. 10.1186/s12879-019-4338-z . WHO. eHealth and innovation in women’s and children’s health: a baseline review: based on the findings of the 2013 survey of CoIA countries by the WHO Global Observatory for eHealth. March 2014. 2014. Oluoch TO. Electronic medical records and clinical Decision Support Systems in HIV care in resource-limited settings. 9789462596313; 2015. Oluoch T, Kwaro D, Ssempijja V, Katana A, Langat P, Okeyo N et al. Better adherence to pre-antiretroviral therapy guidelines after implementing an electronic medical record system in rural Kenyan HIV clinics: a multicenter pre-post study. doi:D - NLM: HHSPA796290 OTO - NOTNLM. Oluoch T, Santas X, Fau - Kwaro D, Kwaro D, Fau - Were M, Were M, Fau - Biondich P, Biondich P, Fau - Bailey C, Bailey C, Fau - Abu-Hanna A et al. The effect of electronic medical record-based clinical decision support on HIV care in resource-constrained settings: a systematic review. Chan KS, Fowles JB, Weiner JP. Review: Electronic Health Records and the Reliability and Validity of Quality Measures: A Review of the Literature. Med Care Res Rev. 2010;67:503–27. 10.1177/1077558709359007 . Bian J, Lyu T, Loiacono A, Viramontes TM, Lipori G, Guo Y, et al. Assessing the practice of data quality evaluation in a national clinical data research network through a systematic scoping review in the era of real-world data. J Am Med Inf Assoc JAMIA. 2020;27:1999–2010. 10.1093/jamia/ocaa245 . District Health Information Software 2 [Internet]. Oslo: Health Information Systems Programme. 2012. [Accessed April 9, March 2022]. Available: http://www.hisp.uio.no/ Braa J, Sahay S. The DHIS2 open source software platform: evolution over time and space. LF Celi Glob Health Inf. 2017;451. Braa J, Heywood A, Sahay S. Improving quality and use of data through data-use workshops: Zanzibar, United Republic of Tanzania. Bull World Health Organ. 2012;90:379–84. Dehnavieh R, Haghdoost A, Khosravi A, Hoseinabadi F, Rahimi H, Poursheikhali A, et al. The District Health Information System (DHIS2): A literature review and meta-synthesis of its strengths and operational challenges based on the experiences of 11 countries. Health Inf Manag. 2018;1833358318777713. 10.1177/1833358318777713 . Amoroso CL, Akimana B, Wise B, Fraser HS. Using electronic medical records for HIV care in rural Rwanda. 2010. pp. 337–341. Muthee V, Bochner AF, Osterman A, Liku N, Akhwale W, Kwach J, et al. The impact of routine data quality assessments on electronic medical record data quality in Kenya. PLoS ONE. 2018;13:e0195362. 10.1371/journal.pone.0195362 . Tsai CH, Eghdam A, Davoody N, Wright G, Flowerday S, Koch S. Effects of Electronic Health Record Implementation and Barriers to Adoption and Use: A Scoping Review and Qualitative Analysis of the Content. Life Basel Switz. 2020;10. 10.3390/life10120327 . Lee K, Weiskopf N, Pathak J. A framework for data quality assessment in clinical research datasets. American Medical Informatics Association; 2017. p. 1080. Odeny B, Hughes JP, Gloyd S, Njoroge A, Wagenaar BH, Puttkammer N. Development of novel composite data quality scores to evaluate facility-level data quality in the District Health Information System (DHIS2) in Kenya: a nationwide retrospective cohort study. doi:Under review. Xiao H, Augusto O, Wagenaar BH. Reflection on modern methods: a common error in the segmented regression parameterization of interrupted time-series analyses. Int J Epidemiol. 2021;50:1011–5. Garrib A, Stoops N, McKenzie A, Dlamini L, Govender T, Rohde J, et al. An evaluation of the District Health Information System in rural South Africa. South Afr Med J Suid-Afr Tydskr Vir Geneeskd. 2008;98:549–52. Noureldin M, Mosallam R, Hassan SZ. Quality of documentation of electronic medical information systems at primary health care units in Alexandria, Egypt. East Mediterr Health J Rev Sante Mediterr Orient Al-Majallah Al-Sihhiyah Li-Sharq Al-Mutawassit. 2014;20: 105–11. Nicol E, Dudley L, Bradshaw D. Assessing the quality of routine data for the prevention of mother-to-child transmission of HIV: An analytical observational study in two health districts with high HIV prevalence in South Africa. Int J Med Inf. 2016;95:60–70. 10.1016/j.ijmedinf.2016.09.006 . O’Hagan R, Marx MA, Finnegan KE, Naphini P, Ng’ambi K, Laija K, et al. National assessment of data quality and associated systems-level factors in Malawi. Glob Health Sci Pract. 2017;5:367–81. Nisingizwe MP, Iyer HS, Gashayija M, Hirschhorn LR, Amoroso C, Wilson R, et al. Toward utilization of data for program management and evaluation: quality assessment of five years of health management information system data in Rwanda. Glob Health Action. 2014;7:25829. Nagbe T, Yealue K, Yeabah T, Rude JM, Fallah M, Skrip L, et al. Integrated disease surveillance and response implementation in Liberia, findings from a data quality audit, 2017. Pan Afr Med J. 2019;33:10. 10.11604/pamj.supp.2019.33.2.17608 . Abiy R, Gashu K, Asemaw T, Mitiku M, Fekadie B, Abebaw Z, et al. A Comparison of Electronic Medical Record Data to Paper Records in Antiretroviral Therapy Clinic in Ethiopia: What is affecting the Quality of the Data? Online J Public Health Inf. 2018;10:e212. 10.5210/ojphi.v10i2.8309 . Ahmed R, Robinson R, Elsony A, Thomson R, Squire SB, Malmborg R, et al. A comparison of smartphone and paper data-collection tools in the Burden of Obstructive Lung Disease (BOLD) study in Gezira state, Sudan. PLoS ONE. 2018;13:e0193917. 10.1371/journal.pone.0193917 . Endriyas M, Alano A, Mekonnen E, Ayele S, Kelaye T, Shiferaw M, et al. Understanding performance data: health management information system data accuracy in Southern Nations Nationalities and People’s Region, Ethiopia. BMC Health Serv Res. 2019;19:175. 10.1186/s12913-019-3991-7 . Mohammadi Firouzeh M, Jafarjalal E, Emamzadeh Ghasemi HS, Bahrani N, Sardashti S. Evaluation of vocal-electronic nursing documentation: A comparison study in Iran. Inf Health Soc Care. 2017;42:250–60. 10.1080/17538157.2016.1178119 . Walther B, Hossin S, Townend J, Abernethy N, Parker D, Jeffries D. Comparison of electronic data capture (EDC) with the standard data capture method for clinical trial data. PLoS ONE. 2011;6:e25348. 10.1371/journal.pone.0025348 . Zeleke AA, Worku AG, Demissie A, Otto-Sobotka F, Wilken M, Lipprandt M et al. Evaluation of Electronic and Paper-Pen Data Capturing Tools for Data Quality in a Public Health Survey in a Health and Demographic Surveillance Site, Ethiopia: Randomized Controlled Crossover Health Care Information Technology Evaluation. JMIR MHealth UHealth. 2019;7: e10995. 10.2196/10995 Nicol E, Hanmer LA. Routine Health Information Systems in South Africa–Opportunities for Improvement. Stud Health Technol Inf. 2015;216:993. Ngugi P, Babic A, Were MC. A multivariate statistical evaluation of actual use of electronic health record systems implementations in Kenya. PLoS ONE. 2021;16:e0256799. 10.1371/journal.pone.0256799 . Gloyd S, Wagenaar BH, Woelk GB, Kalibala S. Opportunities and challenges in conducting secondary analysis of HIV programmes using data from routine health information systems and personal health information. J Int AIDS Soc. 2016;19:20847. 10.7448/IAS.19.5.20847 . Bisrat A, Minda D, Assamnew B, Abebe B, Abegaz T. Implementation challenges and perception of care providers on Electronic Medical Records at St. Paul’s and Ayder Hospitals, Ethiopia. BMC Med Inf Decis Mak. 2021;21:306. 10.1186/s12911-021-01670-z . Douglas G, McKay M, Bwanali M, Gadabu O, Mumba S, Gondwe A. A comparison of data captured at point of care versus data captured retrospectively. AMIA Annu Symp Proc AMIA Symp. 2008; 934. Tweya H, Feldacker C, Ben-Smith A, Harries AD, Komatsu R, Jahn A, et al. Simplifying ART cohort monitoring: can pharmacy stocks provide accurate estimates of patients retained on antiretroviral therapy in Malawi? BMC Health Serv Res. 2012;12:210. 10.1186/1472-6963-12-210 . Tweya H, Feldacker C, Gadabu OJ, Ng’ambi W, Mumba SL, Phiri D, et al. Developing a point-of-care electronic medical record system for TB/HIV co-infected patients: experiences from Lighthouse Trust, Lilongwe, Malawi. BMC Res Notes. 2016;9:146. 10.1186/s13104-016-1943-4 . Myburgh H, Murphy JP, van Huyssteen M, Foster N, Grobbelaar CJ, Struthers HE, et al. Implementation of an electronic monitoring and evaluation system for the antiretroviral treatment programme in the Cape Winelands district, South Africa: a qualitative evaluation. PLoS ONE. 2015;10:e0127223. 10.1371/journal.pone.0127223 . Aminpour F, Sadoughi F, Ahamdi M. Utilization of open source electronic health record around the world: A systematic review. J Res Med Sci. 2014;19:57–64. Hillestad R, Bigelow J, Bower A, Girosi F, Meili R, Scoville R, et al. Can electronic medical record systems transform health care? Potential health benefits, savings, and costs. Health Aff (Millwood). 2005;24:1103–17. Bhavaraju SR. From subconscious to conscious to artificial intelligence: A focus on electronic health records. Neurol India. 2018;66:1270–5. 10.4103/0028-3886.241377 . Akanbi MO, Ocheke AN, Agaba PA, Daniyam CA, Agaba EI, Okeke EN et al. Use of Electronic Health Records in sub-Saharan Africa: Progress and challenges. 2012. Muthee V, Bochner AF, Kang’a S, Owiso G, Akhwale W, Wanyee S et al. Site readiness assessment preceding the implementation of a HIV care and treatment electronic medical record system in Kenya. King JD, Buolamwini J, Cromwell EA, Panfel A, Teferi T, Zerihun M, et al. A novel electronic data collection system for large-scale surveys of neglected tropical diseases. PLoS ONE. 2013;8:e74570. 10.1371/journal.pone.0074570 . Bernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2017;46:348–55. Additional Declarations No competing interests reported. Supplementary Files Appendix.docx Cite Share Download PDF Status: Posted 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-5672455","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":395905389,"identity":"2c747f77-8081-4d78-8b1d-68c58f402b87","order_by":0,"name":"Beryne Odeny","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAxElEQVRIiWNgGAWjYBACAwhlAyIYDzAwMBOtJY0BpJokLYdJ0GLO3vtMuqDifOKG8+cPHGCosE5sIKTFsue4mfSMM7cTN9xIBtpyJp2wFoMbaWzSvG23czfcADqMse0wEVruPwNq+Xcud8P5w0At/4jRcoMNqKXhQO6GA0CHMTYQocWyJ43ZmudYcv3MG8kGBxKOpRsT1GLOfozxNk+NnTHf+YMPH3yosZYlqAUIWCTgzAQilIMA8wciFY6CUTAKRsFIBQCXzEJWu1EYcwAAAABJRU5ErkJggg==","orcid":"","institution":"Washington University in St. Louis","correspondingAuthor":true,"prefix":"","firstName":"Beryne","middleName":"","lastName":"Odeny","suffix":""},{"id":395905390,"identity":"db293add-7d9c-4cce-aabb-29215cfb8da5","order_by":1,"name":"Orvalho Augusto","email":"","orcid":"","institution":"University of Washington","correspondingAuthor":false,"prefix":"","firstName":"Orvalho","middleName":"","lastName":"Augusto","suffix":""},{"id":395905391,"identity":"6c3aea14-94b3-4060-9317-0a291661fa52","order_by":2,"name":"Bradley H. Wagenaar","email":"","orcid":"","institution":"University of Washington","correspondingAuthor":false,"prefix":"","firstName":"Bradley","middleName":"H.","lastName":"Wagenaar","suffix":""},{"id":395905392,"identity":"d830b68a-5ba0-4fbf-9ddc-bd26ca32461c","order_by":3,"name":"James P. Hughes","email":"","orcid":"","institution":"University of Washington","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"P.","lastName":"Hughes","suffix":""},{"id":395905393,"identity":"97240f61-1fee-439c-8e5c-5cad758364d2","order_by":4,"name":"Anne Njoroge","email":"","orcid":"","institution":"International Training and Education Center for Health (I-TECH)","correspondingAuthor":false,"prefix":"","firstName":"Anne","middleName":"","lastName":"Njoroge","suffix":""},{"id":395905394,"identity":"9f12d2c9-8c67-4794-868b-91e7c50c7eb5","order_by":5,"name":"Steve Gloyd","email":"","orcid":"","institution":"University of Washington","correspondingAuthor":false,"prefix":"","firstName":"Steve","middleName":"","lastName":"Gloyd","suffix":""},{"id":395905395,"identity":"528d29b9-3ab0-4b3d-bd13-bd3c18b5d4b4","order_by":6,"name":"Nancy Puttkammer","email":"","orcid":"","institution":"University of Washington","correspondingAuthor":false,"prefix":"","firstName":"Nancy","middleName":"","lastName":"Puttkammer","suffix":""}],"badges":[],"createdAt":"2024-12-19 00:08:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5672455/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5672455/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":72909340,"identity":"9daa84e5-911c-4e82-bfb7-7589c47baf72","added_by":"auto","created_at":"2025-01-03 14:30:57","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":80065,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5672455/v1/565a1850b02ed05c22f81765.png"},{"id":72910652,"identity":"8c28bc78-2f80-45e1-bc4b-f62994226ec1","added_by":"auto","created_at":"2025-01-03 14:38:57","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":80962,"visible":true,"origin":"","legend":"\u003cp\u003eTime series plot of the average composite discrepancy scores across all 187 facilities\u003c/p\u003e\n\u003cp\u003eFigure 2 footnote: The data centered on the date of EMR implementation and adjusted for seasonality and autocorrelation: Blue dots in figure 1 represent the average discrepancy score for all the facilities by quarter; gray dots represent individual facility discrepancy scores; the solid line represents the time series plot.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5672455/v1/b33ca6fd2ffe5af44c6753a1.png"},{"id":72910653,"identity":"cae1725c-ca17-4437-825d-489f47833b80","added_by":"auto","created_at":"2025-01-03 14:38:57","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":91594,"visible":true,"origin":"","legend":"\u003cp\u003eInterrupted time series plot of CCS over time in 187 facilities (includes the wash-in period of 6 months after EMR implementation)\u003c/p\u003e\n\u003cp\u003eFigure 3 footnote: The data centered on the date of EMR implementation and adjusted for seasonality and autocorrelation: Blue dots in figure 1 represent the average discrepancy score for all the facilities by quarter; the solid line represents the time series plot; the dashed lines represent the prediction intervals.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5672455/v1/b49f562b8066c4177c027c5b.png"},{"id":72910844,"identity":"abf21919-b977-4caa-a0ba-b31d3a767be0","added_by":"auto","created_at":"2025-01-03 14:46:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1200359,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5672455/v1/7877bfcc-3fc5-4e5e-b135-eb81bddbe63d.pdf"},{"id":72909342,"identity":"792eb350-791f-4966-9db0-df6a7dacc75c","added_by":"auto","created_at":"2025-01-03 14:30:57","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":17782,"visible":true,"origin":"","legend":"","description":"","filename":"Appendix.docx","url":"https://assets-eu.researchsquare.com/files/rs-5672455/v1/418fa47e6e92674e0e0e4bb3.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"The effect of introducing an electronic medical record system on data quality and factors associated with data quality across 187 HIV clinics in Kenya: An interrupted time series analysis from 2011-2018","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRobust routine health information systems (RHIS) are essential for optimal health system evaluation, quality improvement, governance, and health management [\u003cspan class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e3\u003c/span\u003e]. The World Health Organization\u0026apos;s (WHO) global digital health strategy aims to \u0026ldquo;improve health for everyone, everywhere by accelerating the adoption of appropriate digital health\u0026rdquo; \u0026ndash; this includes EMRs [\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e5\u003c/span\u003e]. Electronic Medical Records (EMRs) are considered essential building blocks for strong RHIS [\u003cspan class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e7\u003c/span\u003e]. EMRs streamline data capture, data retrieval, and reporting which facilitate data use for clinical decision making and enhanced patient care. Furthermore, EMRs have also been found to reduce data recording errors [\u003cspan class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e9\u003c/span\u003e] as well as time/ monetary costs of data management [\u003cspan class=\"CitationRef\"\u003e10\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e13\u003c/span\u003e]. EMRs have a range of purposes crucial to health system strengthening including data reporting, aggregation and management, supporting clinical decision making, and interlinking health departments [\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e]. This paper aims to add to the base of evidence on the utility of EMRs as a tool for strengthening data management \u0026ndash; specifically, data quality in aggregate data reporting [\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e\n\u003cp\u003eEMR utilization in healthcare in low- and middle-income countries (LMICs) has dramatically increased over the past two decades. The United States (U.S.) President\u0026rsquo;s Emergency Plan for AIDS Relief (PEPFAR) has funded scale up and implementation of EMRs specifically for HIV care in health facilities throughout Kenya \u0026ndash; one of the high-burden HIV countries in sub-Saharan Africa (SSA). Kenya was among the first high HIV burden countries to publish national strategies for EMR integration in the health system [\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e5\u003c/span\u003e]. In Kenya, the District Health Information System (DHIS2) \u0026ndash; a national electronic health information system \u0026ndash; has been used for over a decade to house aggregated facility-level data. Version 2 of the program is an open-source platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.DHIS2.org\u003c/span\u003e\u003c/span\u003e) [\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e]. Health facilities collate patient-level data to prepare summaries that are uploaded to the DHIS2 routinely. The data in DHIS2 are used in planning health service delivery, including planning of health personnel, supply chain management, among other functions, and as such high standards of data quality need to be maintained [\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e\n\u003cp\u003eBefore the introduction of EMRs in Kenya, facility-level HIV data were captured on paper charts and registries. These paper records were used to generate HIV-specific facility summaries for the DHIS2. Phased introduction of EMRs in 2012 led to progressive migration from paper records to a hybrid implementation of paper and electronic records in most HIV facilities. According to programmatic records by mid-2018, about 23% of target facilities have fully transitioned to EMR and 75% are using both paper and EMRs. Collation of electronic data was found to simplify the aggregation of facility-level HIV data, thus use of EMRs to generate HIV-specific facility reports for the DHIS2 expanded as EMRs became widespread. While the DHIS2 hosts data from other sectors beyond HIV service delivery, EMRs are primarily tailored for HIV data management, thus EMR reports uploaded to DHIS2 are HIV-specific. While EMRs are associated with better data quality [\u003cspan class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e], they are not immune to error [\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e], thus Routine Data Quality Assessments (RDQAs) are essential to uphold EMR data quality [\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e]. Improved data quality in EMRs would potentially lead to improved quality of aggregate data uploaded to the DHIS2.\u003c/p\u003e\n\u003cp\u003eTo our knowledge, there is little high-quality evidence on the longitudinal effects of EMRs on data quality in aggregated data systems such as DHIS2 in LMICs. Using an interrupted time series design, we aim to evaluate whether introduction of EMRs improved data quality of the aggregate HIV data in the DHIS2. Specifically, we assessed data quality as captured by a \u0026ldquo;composite discrepancy scores\u0026rdquo; and \u0026ldquo;composite completeness scores.\u0026rdquo; Composite discrepancy scores (a plausibility check) encapsulate the degree of discrepancy or deviation from the expected values across groups of indicators and related data checks [\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e]. The composite completeness score (CCS score) is a percentage measure of the extent of documentation for pre-selected variables. The objectives of this evaluation are two-fold: 1) to evaluate the evidence for effects of EMR implementation on composite discrepancy and completeness scores for HIV data across 187 health facilities in Kenya from January 2011 to June 2018; and 2) to assess facility and EMR implementation correlates of data quality (i.e., reflected by the composite discrepancy and completeness scores) using data from a cross-sectional facility survey conducted in 2017.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eSetting\u003c/p\u003e\n\u003cp\u003eThis was a quasi-experimental study utilizing time series data from the national DHIS2 RHIS in Kenya. Data was extracted quarterly from January 2011 to June 2018 for 187 facilities implementing EMRs \u0026ndash; specifically, KenyaEMR and IQCare EMR software.\u003c/p\u003e\n\u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eData sources\u003c/p\u003e\n\u003c/div\u003e\n\u003cp\u003eData sources were the DHIS2 database and I-TECH and Palladium program records on EMR deployment and implementation process. I-TECH and Palladium were the two predominant EMR technical assistance providers in Kenya during the timeframe of this analysis. Program records were used to capture the date of KenyaEMR or IQCare deployment at each health facility. The primary exposure is presence of EMR (a binary pre or post variable). The primary outcomes were composite discrepancy and completeness scores, as defined below.\u003c/p\u003e\n\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003cp\u003eData checks and HIV-related indicators [\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/p\u003e\n \u003cp\u003eWe used HIV-related indicators \u0026ndash; encompassing general adult and pediatric HIV care, antenatal care (ANC), Labor \u0026amp; Delivery care (L\u0026amp;D), and Prevention of mother-to-child Transmission of HIV (PMTCT) \u0026ndash; that were uploaded monthly to KHIS. The data consisted of aggregate health service utilization statistics by department. Appropriate data checks were determined a priori, and these checks primarily summarized relationships between indicators to ensure the data were complete, consistent, and plausible. For example, one data check compared the total number of patients in HIV care in a specific quarter versus the total number on ART in that quarter. The difference between the two indicator values was expected to be zero or greater (the logic being that those enrolled would always be more or equal to those receiving ART). A series of data checks were used to construct composite scores for each unique facility. Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e below summarizes the data checks and indicators explored in this analysis.\u003c/p\u003e\n \u003cp\u003eAs outlined in Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e, four ANC and PMTCT data checks (#1 \u0026ndash; #4) were computed as differences between related indicator values. Similarly, four general HIV care data checks (#5 \u0026ndash; #8) were computed as differences between general HIV care indicator values. The difference for all data checks were expected to be greater than or equal to zero (except data check # 8 which was expected to only be 0, Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eComputing the composite discrepancy score\u003c/h3\u003e\n\u003cp\u003eThe composite discrepancy score was based on assessment of plausibility across indicators and eight predefined data checks. Individual data checks were derived by comparing groups of indicators. For example, one data check was derived by comparing the total number of patients currently on ART (an indicator) minus sum of patients on ART across all age groups (another indicator) and this was ideally equal to zero; the further away the data check value was from the expected value (zero in this case), the higher the absolute difference between the observed and expected values, thus a higher individual discrepancy score which meant worse data quality (Appendix 1). These individual scores were computed for each data check and were based on \u003cem\u003eZ\u003c/em\u003e-score deviations that depicted the extent of discrepancy of observed values i.e., how far observed values were from expected values. The composite discrepancy score was a continuous variable computed as an average of all the \u003cem\u003eZ-\u003c/em\u003escores for individual data logic checks, for each unique facility quarter. For additional context, a score that is 0.01 lower, is a comparatively small change but represents a small improvement in data quality performance, and vice versa. Additional details on the development of this score are provided in the Appendix 2 and in a paper detailing the composite score development [\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eComputing the composite completeness score\u003c/h3\u003e\n\u003cp\u003eThe completeness score aims to determine whether aggregate data on the two HIV indicators that constitute a particular data check are present or missing (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). If data on the indicators are missing, it means that individual patient data was not aggregated into the summaries that are uploaded to the KHIS. For example, data check 1 compares two indicators: 1) the number of women tested for HIV in antenatal care (ANC) and 2) the number of women testing HIV positive in ANC. When both values are missing, the data check will miss a value for that specific facility. This is a measure of completeness that looks at two related variables that should both be present.\u003c/p\u003e\n\u003cp\u003eThe completeness score was based on the proportion of data checks which were complete for each observation i.e., each unique facility quarter. A binary score of 1 or 0 was assigned to each data check based on the presence or missingness of the data (1\u0026thinsp;=\u0026thinsp;present, 0\u0026thinsp;=\u0026thinsp;missing). The completeness score was a continuous score computed as the proportion (percent) of data checks with complete data for each observation (i.e., unique facility-quarters). The minimum possible score was 0% (0 complete out of 8 checks) and the maximum possible was 100% (8 complete out of 8 checks) [\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eCross-sectional survey\u003c/h3\u003e\n\u003cp\u003eA cross-sectional facility survey done in October to December of 2017 was used to collect data on non-time varying facility and EMR implementation factors which are potentially correlated with data quality such as facility level, facility volume, and EMR type. Survey data on EMR implementation characteristics was only available for 115 health facilities.\u003c/p\u003e\n\u003ch3\u003eEthical considerations\u003c/h3\u003e\n\u003cp\u003eThis evaluation was approved by the AMREF Ethical Scientific Review Committee (ESRC), Kenya, United States (US) Center for Disease Control and Prevention (CDC), and the University of Washington (UW) Human Subjects Review committee.\u003c/p\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003cp\u003eStatistical Analysis\u003c/p\u003e\n \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\n \u003cp\u003eDescriptive statistics\u003c/p\u003e\n \u003cp\u003eAn analysis plan was prepared prospectively prior to data analysis. Descriptive statistics were used to analyze the distribution of health facility characteristics. We reported the number of facilities included in the analysis by facility level (national hospital, county hospital or sub-county hospital, health center, dispensary), facility volume (\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;500 versus \u0026lt;\u0026thinsp;500 HIV patients in care), region/county (46 counties classified into low and high-HIV burden counties), and EMR type (KenyaEMR, IQCare or other). Median and interquartile ratios were used for continuous data, whereas proportions were used for binary and categorical data. Baseline characteristics were summarized and presented in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eProposed impact model\u003c/strong\u003e [\u003cspan class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/p\u003e\n \u003cp\u003eThe outcomes, composite discrepancy and completeness scores, were hypothesized to be responsive to the intervention of EMR implementation on a gradual basis. Based on the literature, we hypothesized a 6-quarter (i.e., 18 months) time lag after EMR introduction to account for the \u0026ldquo;wash-in period\u0026rdquo; after EMR introduction. It was expected that EMR implementation would improve over time until optimal functioning was achieved [\u003cspan class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e24\u003c/span\u003e]. Therefore, a gradual change in the composite discrepancy score was anticipated over the 6-quarter \u0026ldquo;wash-in period\u0026rdquo;, followed by a plateauing of the intervention effect after 6 quarters as demonstrated in Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e (impact model). The final impact model which informed the statistical model and analysis, excluded the \u0026ldquo;wash-in period\u0026rdquo;, thus only two segments were modeled: pre-EMR, and post-EMR (starting after 6 quarters post-intervention).\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003ch3\u003eInferential analyses using Interrupted Time Series (ITS)\u003c/h3\u003e\n\u003cp\u003eMultilevel models were used to control for clustering at the health facility level. Changes in the data quality scores pre- and post- EMR implementation were estimated using generalized linear models (GLM) with Gaussian distribution and identity link and bootstrapped standard errors. Random effects for facilities (including random intercepts) were used to account for intra-cluster correlation. Data was assumed to be missing at random. We included data points beginning at 14 quarters pre-EMR implementation.\u003c/p\u003e\n\u003cp\u003eThe primary model included dummy variables for quarter (i.e., season) to account for seasonality and random intercept to account for facilities having different baseline DQ scores.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n \u003cdiv align=\"left\" class=\"colspec\"\u003e\u003cimg src=\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAtAAAACICAYAAADOFdwvAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAADsMAAA7DAcdvqGQAADfISURBVHhe7d3/jxvVvT/+pz+/Q/A6PyEuQjuuVATVIjqbVnSDFKRkTKiiXpXWTqmqSFQELyi6XHKz4A1CV4QQb0FXiqrsLrqRqqtb1ilUV3oLm3gjpRI26CZZkC1A3B9iC3FRf7Ljhv4B5/MD85p75njGnll7vyXPhzRK9sx4fOZ8mzMz54wTSikFIiIiIiKK5P8zA4iIiIiIKBw70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70LQttdttTExMoF6vm6uIiIiIthQ70LQtzc7OYmJiwgwmIiIi2nK3VQe6UqkgkUhsm7ua3W7X6ygmEglMT0+j2Wyam22K5eVlJBKJSMvCwgK63S4ymQwmJibQbrcxPz/vHUO73UYmk0EikcD8/Lz3HeZ+zH2K+fl5HDlyBDdu3PDChllYWEA6nUYikUA6nUapVDI3oS1SqVQwPT2NRCKBiYkJX5kYp4mJCczOzprBt4xbtYxL27EdbKc2eTupVCpe2dsueUXxbWVdm52d3bLv3jBqhwLgLeVy2QvP5/NeuOM4vs8Ui0UFQHU6HV/4VrFtW9m2rVqtllJKqWw2q5LJpLnZprFtWwFQS0tL5irV6XS8tC0Wi2plZUW1Wi1lWZZyHEfVajVVq9UUAJXP51Wr1VKFQqEvD4ap1Woqm80q5eZxrVYzN+lTKBRUMpn0tl1aWlIAVKPRMDeNpVgsKsuyvPJk27YqFovmZjRAuVz2lalGoxFaxnSFQsFXx4OWQqHgbd9qtSLtd6catYxL3Qxb9HpqbhtE30ZPc/Ozsti27WundY7jKNu2zeAtsd422TzeoEXaDkmjKG3bdiB1uNFoqE6nEyk9xk0/rwctkrYrKytemGVZ5m6Ucs9l5jaNRkMlk8m+/SaTSZXP5/v6DNKX2Gm2sq5JX+FWsvNKgEtOIPpJVGmVw7btvkJvkpP5VjRkxWJRJZNJr6FWWsM6DsViMXZhlfQw4yUkbaWxkr/lxFir1VQymfTSvVAo9OXPIJ1Ox5dvkjeDKrykmXlyHjVf8/m8r8PSarVUMpmMdTw7leM4Y7lQkJOtmWZR9m/btu/EBffCTGnlbmVlxfjU9hLlOKMYVxmXixLzM7Zt9+WRfgET1I7KxbZc7OrM79EvvuN2+IO+e5igY4xilDZZL5/yt+S9dM52agc6m832nUvk/LuZHMfp+05pl/V6JhebYWksnV+zfEm+6Oc3OU7zHBTUgQ5KJxoun88HtiM7wY4dwnH06FFYloVWq+ULf++992DbNi5evIhUKuVbZ/ryyy/NoE3z9ttvI5fLYXJy0gv7+OOPfdtstqmpKRQKBfR6vcBH4alUCtls1vv7ypUrSCaTOHjwIODGP5fLeel+4cIFPP7449725rANcwjH+fPnsba2ht27d3uPevbu3et9Psg777wDy7K8OAAYeYhOu93G4uIizp07h5mZGQDA5OQkcrkcfvCDH5ibU4hLly6h1+vh6aef9oVXq1Xf30FarRZee+01pFIpLz8fffRRwC2HjuNgz549xqduTeMq47t27QIAr0yLAwcO9JXrXbt2wXEcIKCdLJVKsCwLALBv3z7fOgR8TyqVwq9+9SsgYt53u1288MILsG17aBs+TqO2ya+99poZBLjtaj6f9/6emZmBUqovH7Yr8xwLYNsMa5mcnMQbb7zhC9u1axcOHDiAZDKJd955x7cObj7LeWxQ+UqlUjh69Ciy2SzW1tZ8x3zixAl8d530f9bW1nx/x7GwsBB7eEMmk0EmkzGDQ8OHkfNwHO122zfkKZ1OY3l52dwM9XodiYDhs81mE4uLi/jZz37mC98xzB71TiJ3CYVc6Zt3LeTOKrTHjfpVqr5sBomPfkcp7G7deq3nDrRy4yHDFoY9Di8UCr4rR9u2+x7VdzqddR+TmUZBgtJs1MdUYXf8gqysrHh345LJZN9d0VarpbLZrLe+UCiofD7fd2dE7n5JOcxms747Ya1WyyvvcB+5y908SWP5nHLjlUwmVTKZjHQcunHdOc1ms335EHSXbxi5CxT2GbkzpX9XrVZTlmWppPsUodVqeds5jqM6nY63DQKeZKmQNDfblkHGlY7jKuOO40RuExzH8R6H62VV2qlBQ0iCvse8u6e0fNX3o4fpi1lfBom7vdqANtk8VqEPFZA4mmF6WyDfHRSmW1pa8sqyZVmB+RKk0+moojZUzWzD9CcRskg9MsPEoLjIkwjHcfrq5DCy7TBy7pO463V2ZWXFq5fmvoLKqNLuNkt+6cNElJt/cg7QF3M/gwTFZ5igehYWHlTXTHHjLPXDtm21srLilV0ElNGg49PbVlnCht1sV/FybJuRsVnlclm13PG4YYVDttUbVrNibBapgFKxy+WysixLZbPZWCfoQaQRWQ9pSIZ1dPQOc8d9rC7pL4/WBuXJIJJfZqXTtdxxr3LSazQaynE7Fuv5TiH71YeTBDGHeZgNl6SBnCw67hAVGJ3BpaUllUwmvThLuZTvlv1I+eh0OspxHK+xKRQKvsZrZWXFyxfbtmOXAznBjMqyLK8hbblj4pPr6NBns9mBDauUPXmELh1fCS+4Y/FbrZZ3IikWi942cqLVSadGtpE6EaetGEc6jrOMJ7VH3VKGwvJCyp+klSgWi6pYLAammZA010nZNNuToP3IMZsXo1HFzSe1AW2ymW66oHORhDnufBKldTaz2WxfmJ73RbdzLdtIPYsSb71MyDk0aYxxNts1CTPzTUWMi+2OM5fjCtp/EPM7G+5Fj0nOfbJevxHkuBeGUh51UsfNfLNtuy9Ngsqt5N96BMVnmLB0CwsPirMu6NgHkfOfWc6C6no+nw9sw6Uc7FThqblDyElOroLCSAOpk4yO0tjCuLoctAwj8VVuYV9Pp2IYaUTWS79TsF3pJz3JyzgNwCCyb8uyAsuHrNdPZOYYOMft5OoNTNG926MzP6fcci3C9mOWNTl56WkQ1pgO4oy54ycnpvV2RqSOD4KQDpfkoXyvxEW/exuUltJ5EnIyDioLYcaRjuMq4xJ/cwnKj5o278DsYMkJM+zCTNJX2rNWq+XFOyh/8vl8336kM2l2tqOKm09qA9rkQfkkaaTHUcL0NJIw/WIkaLukcZdcLnyGPUEMElQXgtoQJ6QDHSUukr5BZW8Q+U5zMennPumsK7cOSNsbdJyStpJvUueCykLQ5xHzfGkeh7mElR8RlC+Dws26JscwaBkk6d59rtW+e4GALHKTQs9zy7L62vBRL5S3gx07BlrkcjksLi7iF7/4BXK5nLna89lnn8G2bTM4MvdiI9IyzLVr1zA9PQ0AuHjxIt544w38+te/Xve4MhlfpC9zc3OoVqt94VHHRr322mtIJpOoVqvb9nVZn332GSzLQiqVwokTJ1Cr1XDmzJmxxDeXy6FcLuPGjRvYu3dv3z5PnjyJfD6PqakpNJtNLCws4MKFCzh27Bjgjg2rVqv4l3/5F984u8uXL/eVw3379vWls7zCT/aTzWYHjtcDgE8//RQAfGOOr169ih/+8IfaVn4y9k5fqtUq5ubm+sLjjI/7n//5HwDA97//fczMzKDVaqHVauGVV14xNx2o2Wyi1+thamrKXOWRcXX333+/L1zq06lTp7y0++abbwAA//Zv/+Zt99VXX/nypFQqodVq4fTp0+h2u6hUKvjtb38L27ZDx61uVDqOq4z/93//NwCg0+lAKYVWqwXHcQLL1Mcff+y1T3v27PHKVaFQwLlz5wB3vOdjjz3m+xy0McNPPPEEEokELMvC5cuXUavVAtvna9eu9ZXPzz//HMlk0jcWOYy8LlNf4M6dMMPN8Ze6cbfJ63XPPfeYQd6Yct3XX38NuGW/1+v55pqkUinYtu1ts1nixGXPnj2BZS8KOc82Gg1zVZ8XX3zRG7+8vLyMZ555xtykj9TZ3bt34+2338ba2ppv/sG4FItFb5H5BnrYI488Yn5kJGZde+SRR3zfBwCO4/SFhen1elhbW8PevXt9y7PPPgsAuHnzJuCew1qtFu677z7f58+fPw8A2L9/vy98RzF71DuNPEYYxnZn9evkCizu3YpRBV11hV01rpd+Fb5ekj5xHxVvFsdxQvN0XDrumHC9jJl386Rs6ekUNkY06Y4fNck46UKh4LszI3cgzTIqTwh05t0f8/F/VM4Y7pwWA+60B915GyboTr9J0tok4Xp6FgqFvnjpQ02UlhdwhzE5jhOYZ8OMIx3HVcaDxqOHcdzH3Ep7OlJzH7WrkCEIwnHvLup/h32v3J0cd1sYFrdBxh0PDLiDGFQPBoXp+zHD5O+gJUrcO+6bJiTf5LO6oHRwAu5AR41L0P6iCPrOIPq5r+OO05U6LW1BUB0y01bKuVn/VMjnzeOMI2h/w4SlY1B4WF3T6cceBQKGagUJKtsq4CnfTrTj70Cvrq7iwIEDZrBPt9vF2trawLtYw5h3MwYtg4TdLTPpL67PZDLodrvmJhuq2+3izJkzWFpaGind4qrX68jlckhE+NGNarW64XFLpVJ45pln0Ov1vLtRf//73wEAtVoNSilcu3YN586d88VFrr71sFKphF6vhwceeMAL09etrKxgcXHR95RA7tyYdz5XV1d9b0Rpt9vo9Xq+tylcvXoVAPCjH/3IC9ssQXfaw0xPT4felf3www+RTCYH5vNf/vIX7w6Ortlsendvxerqqi9ecnfkJz/5iRf2t7/9DY7jQCmFGzdu4OLFizh69Ki3fjNFKeNBd7/b7bZvm7W1Ne8O6zDVatV7w8nDDz+M69ev44UXXsDJkycB9w4xQtqwarXqu9N87NgxrK2toVKp+LaD+xYfBOynWq0G3t3eKFHbZLjt4sLCAnK5XGiZ3QrSFunLxYsXzc182u02vve97+HmzZu4ePEibty4MfSuYxTrictGSaVSyOVyuHDhAvL5fKy73gcPHkQ+n8fi4mLsJz5bKehNN2F1bVSrq6tmUB95KqWfw5rNJlqt1s59+4ZrR3eg5eT38MMPm6t85DVMZsclziMSs0EYtAwihUk/KXa7XVSrVe+RZbvdxq9//Wv84Q9/QKfTAQA899xz3vab4amnnsKePXs2veOwd+9ePP/881BK4Y033sDp06cDH6PKSc/M03fffdd7xdZ6BP18eFBn2NTtdgMfUYtut4u33noLMBoS/eIol8vh3LlzQ1+HJEMMnn/+eS9MhkzoDeSHH3646a8CE9Vqta9efvDBBwCAu+++2wsbdIEEt4Ee9rq61dXVwA7XtWvX+jrxa2trvnjJRcb3v/99bat+pVJp00+iccp4sVj0tUH68AdpJ+U1gHp4Op32hdXrdViW5X1+165daLVamJ6e9sqtXByZ5cp83SC0dJVOt07C9Holdf3BBx/0wjZalDZZwjKZDB588EGUSiWcOHHCW7dVpL5/8cUX5qqh3nvvPfR6Pd9xSFs3TFB9GyUuo1heXh54MSPnsPWcy2Qo48mTJ4fexAq6iN9o169f9/0dNkwpqK6NKp/PY21tre875VeNJfzy5ct97ZV08nf08I2d3oGWMTRBY8R0QVdAAHDHHXcA2ol9dna2787NuMl4Qilc8rPXyWTSq+CXLl2CZVmYmZlBKpXCsWPHcOHCBd9+NtLy8jKuXr2KP/7xj+Yqz0b9lLfS3o/65JNPAlo+6aSRlryVn+BdW1vDqVOnAPdknE6nMT097cVX4lSv1707/HJ3TMbbzs7Oeo1lpVLB4uKi787M/fffj2Qy6ZUbOV79avree+8F3Hzudrt46qmnvCclEtd2u42rV69644K73a53x1XIRZ68W7Ner2N2dhb5fN5Xnj/66KO+O7VSlod17sdNOkGffvqpl47Ly8s4ffo08vm81ympVCro9Xp9jauo1+totVpDT1y9Xg9wt5fy1nWfOunvKZY6p184yx3+Xbt2eXny2GOP4erVq176LS8v46233tr0xj5KGY9CLhL0i6t2u41sNtt3gfH73//eV/4eeOABJJNJHD9+HHDzNugOF9zPAsCdd97phU1OTsKyLLz77rvalt+5fPlyaKfj888/R9e927vRFy5R2mS4NxV+8YtfDB0PK+Xmq6++MlcBAL799ltAG48/LEzv1JphqVQK+XweL7/8si/+s7OzfR0bk5w3Zbt6vY5PPvkEMN7z3O12+zpq8tlKpeK1L1Hj0u12vTI5qna7jd/97ne+sJs3b/rajKmpKSilAp8QDmtbUqkUXnrppUjzN+666y6v3Wg2m0NvDuiC3is9zD/+4z+i1Wohl8uhXq+jVCrh0KFDvvorBtU1oZSKdVF4/PhxJJNJHDp0CKVSCfV6HQsLC3j22Wd980WuX7/eN0rg8uXL3v/r9Xrgu6N3BHNMx04hY4ZkMcfX6JwB4/Dk1S6WZcUeK7oeSfeNAjKeC+7bCczXEuljmGQMURzmPqKS8b3D0mJlg37KW1d0X5cVRMZmylhgBPxc8NLSknc8EicZmytjjW3t57lb7ntJ9bGAYW93kddcBX2vkDx23Fcq1dxfarS0N3vk3df7hJUFpb3TWcpp0Di1oHFvMn5Yvj8qZ8Sxu/JavqL27loz3pL2Hfd1aub3mePMzfU6Gd+Y1365MGjcXdCY6Ib7ujq9/ne0X86T+hr3jQFqDOkYpYwrt55IOpvj4Gvau66DFn1st96mhtU7/bP6sZntsV7eJP7mWNKw75HtpczETXsMOR+YorTJUh6lTXPc9wybpDzJfsxjNtc3Go3AMOXGS8KkDQoKk/Iq64LKSBCpe3paS71JuuPY9bKnj1fVP6u3L8Piou/PnIswiP65oEXKorR5CEh7oW8jcTDzwGzzbff1o9KPkPKua7hv+ZDvjltu10Ov+5LWQecChNS1UbWMd+Xr51ZZr+ePkDxIGm+N2mni9cp2KGkgt5oUpmGNe3EMHej1CppsqdM7xTIxQRpI6SBK5SnE/Clv3crKysB4mB2yMOVyWSW191mbaWs28DQ66YwMop90paNZcN9nTd+JWsY77rvBVchrzrYj6ZBudTyjtsl6utZqNa9d2er409YI6kBvV1tZ16SjPKx+7VQ7eghHFHX31To//elPzVWbLmicahj9kdm3334b+Fhm3Obn59Hr9UJ/krbt/sS1jH8b9095i1KphA8//NB7ZZap2+2i1WpFGif50Ucf4cCBA96wgcuXL+PIkSOAezwb9Yqi25n5uiRTvV5HtVrFQw89hIT2urfTp09Hqhu3gzhlPJVKeXVuZmYGjuP0vTZsu9kuYyDjtMlw5ynMzMzg4MGD2LNnz7ZP56DX/MmyndQDXsUadG7YLm7evNk3/Gm72sq69oc//AEIGD57yzB71LeSjvuYOGz4xmYrBLxCK4hcMcqd0Ww2O/Bu7DjIHZYoi9wVK2zAT3mbd56Xlpb6rl7l9UJRfmhBj5PcMZc7n/L6pkajsa7XlFE/SeM4dzvCHoffzuKUcWh3eKQem3VmO5HH5dshz6O2yeaPPsgxxCnndGuQoRo7Ie+3qq7pw6E2uu+ylW7ZDrSMWdqssUhROI4T+b2HK9q41+w6f8EtDr3AD1ukAz2oc9pax095d9x3dprfZ3YGZNzXMGac5LGrkE6K4zgbnr63C+nARc1zfdygmc+3s6hlXLkX2PoYxO16Ypf6uJ3GPcZpk/V5D9vpGGjzdNzfBdjuw/62uq7p5+9bWULFnfpJRERERHQbu+XHQBMRERERjRM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMWx6B7peryOXyyGTyZirtkyz2UQikUClUjFXRVapVJBOp5FIJDA9PY1ms2luEsv8/Dymp6e9v+v1OmZnZ/vSbXp6GvPz874wItr+2u02JiYmRmp34jLble2gVCohk8lgYWHBXLVpwtpX+j/dbhfLy8tIp9Oo1+vm6k2zHcvwVut2u5idnUUikcDExMSG9gkWFhYwMTGBRCKBXC6HbrdrbhJZUBsY1B4EbbcdbHoH+tChQ7hw4YIZvKPV63UcO3YMV65cQaPRQK/Xw5/+9Cdzs5EcOnQIi4uLZjDtMJlMBolEYuAiDYcelsvlzF35dLtd3/bSmAYt09PTW9pZgdsIm/Ha6jjR5pOOa7VaNVdtKravw50/fx4vv/wyWq2WuYq22FNPPYWpqSkopZDP57G4uIh2u21uNrKFhQV8+umnuHHjBlZWVrC6uoorV66Ym63bdmkPIlNbAIByHMcM3rGy2awqFotm8NhFTTfbts2gsdro/W+2zT6eRqOhAASWmWKx6AtfWlpSABQA1Wq1fNvqisWiAqAsy1KdTscLz2azSq/mjUZD2batAKhCoeCFb4VOp6Ns21a2bfviHNVm59uoCoWCqtVqZvCOsJFx73Q6ofVhM9m2Hal9vZ2Vy2UFYMPKwkbayDI8buVyOXJ9kPPJZkgmkxuehrVaLVJ70Ol0try+bvod6FvRdrqj3m63sba2ZgaPzUbvf7NtxfFMTU2ZQZ4TJ074/n7ggQfgOA4A4M033/St07399tuwbRvpdBqpVMoLf/jhh33bTU1N4eLFiwCA06dP+9ZttlQq5Vvi2Ip8G9V2aifi2si4x837jbJd4rGd3XnnnWbQjrGRZXjc3n//fTMo1Gbdra3X6+j1embwlrl06ZIZtOnYgb6FdLtdZLNZM3hsNnr/m227HU+9Xke9Xu/rRD/22GNwHCf0sVypVMKBAwcidwCibrddbbd8i2J2dnbHPvreyXEnwg4rw6VSicOJhmg2m5idnTWDN91IHeig8YtBYfV63QvTNZtNb+KdOcaz2Wwil8t5n5udnfUGq+uDzOfn55HQJgBWKhVMT08j4Q6mjzKu0hy0rh9D2LKwsOCN4wSAubk5JBIJbxKKPqlQ4q/Tj88c9N9utzE/P4+JiQnfZ0zmxJdMJuPdlZO46HHWJ8jo66IK2r/5HUFLlIk5CwsLvrJQKpWwsLDgKzuSP/r4WZnMok+iSCQSvoku+uSXSqWCTCaDiYmJwOMJio98Dkaal0qlvrwbxe9//3szyHPy5EnAHYdoOnnyJI4fP24Gh5JOuGVZvvCFhYW+chpGz5dBS9zJRmZeST3QJ5CMK99kIowsQe2VxD+sPYoS3/n5ee+EuHfvXiTcOme2O0L2J8egr9ePo16ve21dlDpmtivrjbu+H2nD9Uldg+Iv9O/Sy0jU+o4hbSginAtKpZIXz+XlZd+6qILiFtS2Srkzjdp+BpVrSYdB5yC97On5ZaZRt9v10jidTuPbb7/1rYcbBz2d9XO1Xla63a53vJlMBt1uF/V63fvuQccpzDKMCMcSVIZFWBkZVsb1dsZ8aYCciybc9iWTyXjrzfySfUiaVSoVHD58GND6FIPa0EQigbm5Oe//+vaD6uGw4zNlMhns3bsX0NJQr59yrBMTE311SU/jdDrtqwdhbaBJ367dbmPfvn3o9XqoVqteGsvx63GDli5Rylds5piOODqdjspms8qyLF94Pp/vCysUCt44RxmrubS0pJQ2rmplZUUppVSr1VLZbNYb81mr1VQymVT5fN77v4wHbrVayrZttbS0pMrlsu97ZPyo7DeIvr9hY27CmJ9ttVoKgHd8ZjxarZZKJpN96+VvGbdqZo8cs9DTQTiO4/ucpE8ymfTCJNwcLxuFuf9xyOfzyrZtL79lPK+kqZQPPY1XVlYUtLF4UuY6nY7qdDrKsiyvDBaLRS+tCoWCajQaKplMqkajEXg8hULBy6tOp6Py+byCOwZZ9mNZlmo0Giqfz6tsNuv7fBSSv+Ziji+r1WrecUs+6nlWLpe973ccp29MmKSlco9lZWVFWZblHX+YVqu17voQhxlnPa+kDZCxbnqbMq58k3XmePClpSUvfQa1R1HjK/kg+RvW7hSLRV9dkLZB4qcfR1j7GcZsV9Ybd30/5XJZrayseO3LsPgro753Op111fdhbeiwc4HEWfa3tLSkLMvqqz9RFAoFX9yUe4ySzhtpULnW00M/fr3s2bbtxVuOQ/Ku485RyOfzqtPpePVAP1bZl/m3zE8wy0qn0/HG7DqOM3IZjnosZhlWQ8qIGW+9jBcKBV/5lfZUZLNZXzm1bdurT7JPy7JUuVz24gFA5fN5pWKMARZ6G6+HDaqHg44vjMRLT0Opl9JWyn7le8vlsq98yPpGo+HLO/1YzeMP2w5G30e2yxrnY8nPjTByDZfM1xPVbFDkZCbMA5cwSRz5fNCitAQ2T3q2OznKXMzvMpkZFpf5WWkgpBCZ2+TzeV96SEMljYkK6SAEHYsZFvQ5OT69ccrn80MbqyBB+x+FxE1PK2WkV1D+mJU5m8360tRsVORvabSEeTwdd0JT0CLfD7exHoV5PMpoeEVN60BLY6V/znEc7zPOgA60eRxmepukc7jRBsVZTwszP8eZb4VCoS8Oel0c1h5FiW/QNma5lmMw66V+IaAC6ryERckvM92C4hUl7rIf6XyoiPEPqu/S2ZP4m+mih0kchrWhg84FnU5HJZPJvniutwMdlD5mGm4kBJTrYeegKGlcLBZ9F1Iq4EJGOthB20j6mmVOwsy0NuMTxtxf1GMx82hQGVEhZVzKqp6ueudT+kNBi3y3/h1Cv+AKOp5BzLIWpR6qkOMbxExTpV14hm1jWZYvHvrNK337oLwbFhaUjpIWev5YljX0fLdeIw3hAICDBw/Csiy88847vnA97NKlS/jpT3/qWz/IJ598gmKxCPVdB9+36Hbt2uX7e21tDbVare8zMmkqKv1RQNgy6JGDvE4G2qMeXbvdRjKZ9P5OpVK4du0ajh496ttuXGZmZuA4jjcMoNvtYnV1Ffv37zc3jc18dBK0DHp08vHHHwMAJicnzVWxlEolnDt3DpVKBblcznusZRo2CebLL78EvmuN+hZ9bPJGjCN+/vnnzSCfXC4Hy7Jw5swZdLtd77HgzMyMuWkfOQbHcfDuu+/2pbc8Ol9whyYtLi5ibm4uNO/0R+2DlkGPH8dplHx7+umnUa1WvfRst9v4h3/4B2991PZoVHIM99xzjy9c2s6//vWvvvDtQE/PKPEPqu9mWYxiWBs66Fzw5Zdfotfr9cXTbKfHZcEd5hFUF0ZtP4VZroedg6K4fPly3+fMNFtbW8N9993nC5Pzytdff+0L324GlRGdnrZSB/W+Ry6Xw40bNwAAn3/+ORzH6dunUmpgOy11JKiMxBWlHurMshPH0aNHcf36dW9M8qFDh3zrW62W75w7NTWFGzduDJxIP4qnn34ayWTSm3AvQ4TW08ZEMXIHGgCy2aw3rqVUKuHxxx/HM888g8XFRXS7XXz44Yc4ePCg+bGBLl++bAZFIg30KMyCH7SYE71M8/PzOHDgAHbt2oXV1VVz9abPCD558iRarRZKpRIuXboUa9LZIBcvXuxLG3MxG6SNIOO43n//ffzmN79BsVg0N4llHA1ZXDMzMwMbWQA4deoUer0e3nvvPSwvL+PIkSPmJgP98Y9/RKvV6hvrPDMz46XZiRMnkM/nUSwWQ/NuZmamL5+DlmHHM27rybfJyUnYtu2N3Tt//nxfe7Xe9mg9zHGmwy76tpvNiv+wNnQc54JxOHHihPcmHdNGtp/DzkHjcvPmTd/f4zivbJb1lpFBb4C4evWqNwY8Kqkjd999t7lq3TajHsq49rm5OTz66KP4z//8T3OTWG8UGVUqlcJLL73kTbh/5513Yp8j4xhLB/qXv/wler0eKpUKPvvsM8zMzODJJ58E3IKm3ymIYnJyEtVqFbOzs763DpgTDU2WZWFubg7Ly8teAZYB/ZtpeXkZp0+fxpUrV3D06NG+q5+77rrL68ToFW3Y8Y1Cvwv91ltvbdjd7vUKeruEuP/++82gPvv27UM2m8W5c+f6Oj9xSAN26NAhbyIV3I6ZOTlio9Tr9cAJR9DuQv/ud7/D6upq7DKTSqXw//7f/8Pi4mLod+xEo+bbiy++iMXFxcBfEF1vexSXHMN//Md/+MK/+eYbIGI92EpR4i937gZd6EQ5zmFt6KBzwR133AGM0HkyPfjgg2bQlht2DorirrvuGtoZtCwLFy5c8G0jdWQ7potuUBkJI2VndnbWV4YrlQpKpRLuvfde9Ho938RBuE8BgtoW8cEHH8CyrHXlkylKPRyX5557Dt1uFxcvXkQul+vrpCeTSSwuLvraYJnAuFHkLnShUECpVBp7O60bSwd6amoKlmXh1VdfxQ9+8APAPek4joPDhw/jJz/5ibetXlDNMLmSPX78uJfwlmV5j7Jk5q15ZSXOnj0LAHj22Wexe/duJBIJ7N69G48++qi56dhIY/HVV195YXIc33zzDbrdrq+jUqlU8PLLLwMAFhcXvXim02nfO3vNdDL/DQt77LHHvDCzkMpd6FQq5XuE0nZ/JjPKo8JB+1+PJ598EslkEtls1ktLs6GROxqffvop4J58ZXjQkSNHUCqV0Ov1vNcUNZtN745hu91Gs9nsu0sizOOZnJxEoVBAr9fDE0884ZW9vXv3Yv/+/b601sVJQzm+oDiVSiUcOnTIewz6xRdf+MoWADzzzDNotVo4deqUL7zb7eL69eu+OEqa6WFyt/nw4cOYn58fePGykbruWyxkEZIuej2XMNluXPkm9u/fj2QyiZ///Of45S9/6Vs3rD2KEt9HHnnE22Z5eRntdtvbXraVY7hw4YJ3wmm323jrrbdQLBaRSqW8/enHI/8PKk8m8/PrjXtQekaJv9T3F154weuAyHC4M2fOYHl5OVJ9H9aGDjoXTE1NwXEcnDlzxvfk9Pr1696FEgBMT08jnU4PrR/Safjiiy8Ad1/S/kT5/CiC8gERzkFm2YNWBuTf559/Hr1eD88995yX5/KmoCNHjqBer+Ps2bNotVp45ZVXvHr85ptvwnEc70aGWeYA9LVTZnkcxNxflGMJKsODyghC0nZqagrZbBa9Xs97G8XExASOHTuG/fv3I5fLwbZtrK2t4aGHHvLairffftt3zr169apX/uv1OhYXF734yFO7mzdvotlsDr3RIecHKWdR6iFCjm8Q6YDLvwDwt7/9Db1eD91uF+12Gx988IG3rlKp4I033gDcNJa0OHDgAB5//HFgSN4NC3Mcxytz+pBauQt94cIFvPTSS1443DglhgzBjcUcFL1ehUIhcMKBObMTxmB9GRyuhynjF9OS7oz3Tqfj296yrL6B9ivuWwbC1gcJGqAehQxY15eiO0FL4uBobwqRmdLK/U79+PTvlsH9sqiAdAsLk0H6+nfpbNvum1wgE04sy/JNEAgybP/r0Wg0+tJL0lJIWZK0qtVq3gSFjjaTP+nOzJfJHIVCwZdP5jGGHU+hUPBm/tq27U0+lP2Ys32jpqGZt0GLTMoxw2W/MvNbJn8ElUOZMKIv5oQLPS5F9xcQJc03ehJhUJzl+/WwWq3Wl2a1Wm1s+abLu2+HCBLWHkWNr0xyS7qT12pGu6enddGdvAXjbRsqoM6b+zHzWGfGa71x17exjDZfDYm/ctNS9iH5Z7lttZTpYfVdDWlDZR96PPT1He0tLXJcMhFZJhxJ+TL3G0Qmmlru2xWKxaJyHMcrf5JX4yb5YJbrQecgeSODLEH1To65XC779lMul/vOIRIm8dDfbGGWJxXSrul/xynD5mfDjsUswyKsjAwq451Ox9fOOI7ja4M67tvJ5PPZbNa3Hm7bpNcB85wsZdN8UYJJP05ZxKB6OOj4gpjpDjft5S0bEldplx13sq7SJhrKd0mdCMq7qGFKm6yqf5fouBMpzcmDUvajHHMUY+tA72SSQRvRwG0nHXf2eZhyubxt0kCvKDvJdkrDuOTko4zGieh2JRdJo9qoDjTtPGxXN15Ze71rkGEXJlGNZQjHrULGN92qXnnlFeTzeTPY89FHH236pK9bza2ShrlcbqQJTES3gvfee8+bz0NE21+328Wrr76K3/zmN+YqwF1vvsFtvW7bDnSlUsHExARKpRK+/fZb2La9Ya9W2UrLy8ve2KNSqYR//ud/NjfxJloFrdsKMpYryni47WK7peF6yVhBottZu93G8vIyfvzjH49lYhcivm6Sbm3m+G0aH/nFzN27dyOVSgW+TKBSqeC9994b+ha1yMxb0reLmvbrNtlstm8Mza1CxgLbtj22McsbyRzvxEddG0/Gt5k/iEBEo5F2zBzjSrcnnts2jsxD2Mz+XEJ9l6lERERERBTBbTuEg4iIiIhoPdiBJiIiIiKKgR1oIiIiIqIY2IEmIiIiIoqBHWgiIiIiohjYgSYiIiIiioEdaCIiIiKiGNiBJiIiIiKKgR1oIiIiIqIY2IEmIiIiIoqBHWgiIiIiohjYgSYiIiIiioEdaCIiIiKiGNiBvo00m03Mzs6awSOp1+uYnZ1FJpMxV3m63S6Wl5eRTqdRr9fN1ZsqSnzDVCoVpNNpJBIJTE9PI5fLYXp62tzsltRsNpHJZJBIJJBOp1EqlQLDZ2dnMTExgXa7be4i1Pz8/LZNx263i9nZWSQSCUxMTGB+ft7cZN3q9Tpyudy6yuIwuVwO3W7XDCYiojFhB/o2USqV8Nvf/hZHjx4FACQSiaHLwsKCuZs+hw4dwuLiohnsc/78ebz88stotVrmqtiazSbm5+fX3RmPEt8g9Xodx44dw5UrV9BoNNDr9bC2tmZudkvqdrvYt28fisUiOp0O0uk03nrrrcDw1dVV8+M72lNPPYWpqSkopZDP57G4uBjr4mCQQ4cO4cKFC2bwWPzsZz/D9773PTSbTXMVERGNg6KxKpfLqlgsmsFbqtFoKMuyVKfTUUop1Wq1lG3bqtVqedsA8MW7WCyqfD7v/T0IAOU4jhnsUywWFQBVq9XMVbHYtq2SyeRI+7Jte2h8Tdlsdtvl62ZZWloKTK+w8O3Ctm0zKJZGo6HG3USacYpSd9ZrZWVF2bbt1XsiIhof3oEes/fff98M2nJzc3M4deoUUqkUAOCvf/0rXnzxRUxOTpqbek6cOIEbN26YwVvu2rVreOmll8zgWCQd4tioO4U7wX/913+ZQcCA8O2g3W6P/ISgWq2aQSMZR5zikCFG58+fN1cREdGI2IEeo1KptK7hARupXq/j6tWryOVyXtjMzIzv7zAyzpVoJ+l2u8hms2bwltqqOP3qV7/CmTNnOB6aiGjMxtKBXlhY8CZXpdNpVCoVb13QWNV6vd433jZoMT8nZGLPxMQEEokEMpmMN9avVCohk8lgYWEB8/PzSCQSqFQqoeFw4zM9PY2EO1FodnbWO+HoE+AqlQoymUzgJKlKpYLDhw8D7h1fPf7NZhO5XM47Ltl/2L4//fRT36Q7iW86nUaz2US73fZN3DLjonvnnXewZ88eMzgWfaKYpHfQdy4vL3t5oqdhmLB0GbdSqeSVz+XlZXM1MKAMLywsIJFIAFq+yvHPz89jYmLC24dexiT/EgFjySuViq+86ev1tC6VSkMnrelpGDTJbVDZFmHxkXparVZRrVa9fAoLD5ssatZXfYJbUDpiQNkw64x8dmJiwsuzTCbj3emV/II2CXRiYsIbSx8mkUhgbm7O+7/sR08DuN8lYTCOp9lsIp1OY3p6OjROQraV9NGZ6ZeJ0N6JmZkZ9Ho9XLp0SdsjERGNzBzTEVehUFArKytKKaU6nY7K5/MKgG98rSgWi4HhcWWzWW/8q4zntSxL1Wo1b3ys4zjeuuPHjweGLy0teZ+R/cnfMlaxWCx6ny0UCqrRaKhkMqkajYYvTvJZcyxxq9VS2WzWO27Zfz6fD933P/3TP3nhkmadTkdZlqUsy1LFYlF1Oh3VarW8fYWR7Ycx462zLEtls1mltHGh+ncC8KXhyspK3zbmGOhB6TKMua9BVlZWfHFbWlpSlmX5xp1GKcNm+mSzWQVAwa1Cetmzbdv7vkKh4NtXuVxWhULBG5e6tLSkAHjfn81m1dLSkrdP2X8QyX/ZXvalf35Q2VYR4qOUUo7jBI7TNcP18qznjW3bKp/Pq06n45UfKU9mOqohZUP/Dtmm0+kox3GUZVnePhzH6Us7vd4Wi8XAY9JJOdNJfurHJ+VFGcdTLpe98qdC4gRAWZbl5Vm5XO5L/7jtnexLSPoTEdH4hJ+dI+h0Ot7JwlyCOmMI6VjHISeYoKVWq3mdjkKh4PtcWHjQyUU6gHISkxNpuVz2bWeS79CPXU64QYsasO+gTmLQCd3sxJjM+IQZtF0ymRzYoULARCg5bumYmcczLF0GMfcVptPp9MVduRcEEt+oZdj8WwV0iILyX8IkrrZt930PtPRzHMf3ebNs6vL5vG99p9PxdaCilO1h8VEB+T0o3MyblZUVX8dWufGWDrQKSMdhZcP8Dj1MmPtUbh6aHd9BzH3qYVG+25y8FxYnMw2hlaH1tne6oHwiIqLRjDSE48svvwS+OyP0LSdOnPDen9put5FOpwEAlmXhX//1X73HnoOWoCEcn3/+ORzH6fs+pRRmZma87Xbt2uX7nDDD19bWcN999/nC9u/fDwD4+uuvfeF33nmn7+8oPvnkExSLxb64fnee/D/r2fdmuXHjBvbv349SqYTp6elIk6sef/xxQCsjpqjpMoovv/wSvV4P99xzjy9cyqJsgwFleNzW1tZQq9X6vuvixYsAgB/+8IeYm5tDOp3GwsICXnvtNXMXnna7jWQy6f2dSqVw7do171WFUcr2sPiM6uuvv/bFEQDOnTs3cHz9RpUN27axd+9eTE9Po1Qq4dy5c+YmY7WeyaqmUds7IiLaGCN1oEVQRxfuidKyLADA9evXAQCtVguvvvpq38kgaNFPELqrV6/2jeMcxc2bN31/j+PEp7t8+bIZtKOUSiX86Ec/wmeffYZ///d/h+M45iZ9vv32WwDA3Xffba7ybKd0CSvDG+Hjjz82gzyvv/46VlZWkEwmMTc3N/RdvsPeDhKlbA+Kzzisra0FjpkfZCPKxsWLF1EsFtHr9XD48GFMT0+PtR3ZKONu74iIaHQjdaClc3To0CHfxJV6vR46WWtU9957L3q9nm8iDdzJXoM6GmEsy8KFCxd8Jyg52T/44IPaluszOTmJarWK2dlZXyfCnCi0USzL6utExdFsNnH48GGcPXsWr7/+OqampsxNAr3//vuwLCv0VXmbkS533HEHMKSDuNll2LIszM3NYXl52StzMkkMAGZnZ5HL5XDt2jXUajVgwOvU7rrrLrRarb6JgZKGUcr2sPiMSu6MZrNZXz7Pz8+Hdqo3qmy88sorOHHiBK5fv46lpSWsra2FPiEJM442IY5xtXdh9ZCIiNZnpA705OQkCoUCer0ennjiCW/oxd69e71HxeOWy+Vg2zbW1tbw0EMPed/59ttvY2pqyrvzaQoLP3v2LFqtFl555RVvlv+bb74Jx3Fw8OBBIOAuXhi5Y37z5k00m02USiUcP34cyWQSi4uLsCzLi6+8dSBs3xKux/urr74C3A6OkDiHOXDgAD755BMz2EdOxEFx+fvf/w4A+N///V/AfZOBPE2QDqdlWbh+/bp3F7dUKqFUKuHPf/6ztx/zeIalyyDmvsJMTU3BcRycOXPGGzJQKpVw/fp1r4MWpQxLJ07SX+gdTmjx0dNRwuTfs2fPAgCeffZZ7N69G4lEArt378ajjz4KuN8lHeL7778fExMToY/nX375ZQDA4uKit690Oo2HH34YiFi2h8VHPmeWs7BwM2+efPJJWJaFtbU1L5/T6TQ++eQTr1NnpuOwsmF+hx4m+3jssce8v+VtG4uLi95F0Y9//GNAu8gKIvmtd+JlqNUXX3wBuOVJ7pbLG3H09NCZcTKPW/+/HM962ztdtVr18lPeoKJfLBIR0TqYg6LXo1Ao+N5AoE+IsyzLmziIMUwiVO5kKX22ezabVY1Gw5tQA3dmu0zECQsX5XLZm0yVTCZ9byWQCULy2WET12RGvj6pp9Fo+PYvbyQI27ceDneykExA0sP0vxGSlbJdGHO/su+gbSSOkt/mmyv0tyPobykJOh41IF0GMeM6bHKU/lYNiXM2m1X5fN5XFsPKsBl3uBO8zHQz86NYLPZ9VsqdTKxDQHksFouqXC576wdNDlNu/uppGKdsi7D4mMcEN73Dws3jlXyWt2pIuJ7PZjqKsLIR9B3mPmq1mvdGG8uyfG/eKLpv8QiaXKrT92fGTSY5Wpbl/fKo4ziqXC774mJOnjTjpO87KF2lbMdt78zvhDahUT5jTlomIqJ4Euq7k8WGSafTWF1dxeTkJDKZDKrVKlZWVkZ+HEvRZTIZHDlyhGlOdJuZnZ1FMpnE66+/bq4iIqIRbGoHmrZGs9nEz3/+c/z5z3+OPIaZiHY2qfdXrlwJnDxKRETrN9IY6CjS6TQ7z1tsamoKp06dwr59+2JNPCKinalSqXgXzew8ExGN34Z0oNvttjfZ5eTJk+Zq2gK5XA5/+ctf8Kc//clcRUS3mLNnz2J1dZVPnIiINsiGD+EgIiIiIrqVbMgdaCIiIiKiWxU70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUAzvQREREREQxsANNRERERBQDO9BERERERDGwA01EREREFAM70EREREREMbADTUREREQUw/8PvLeXOwIAhx8AAAAASUVORK5CYII=\"\u003e\u003c/div\u003e\u0026nbsp;\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDefinition of covariates in the final ITS model\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"1\"\u003e\u003c/colgroup\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntercept\u003c/strong\u003e: baseline mean DQ score at time\u0026thinsp;=\u0026thinsp;0 (i.e., Quarter 1, 2011) for an average facility.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTime\u003c/strong\u003e: the mean change in DQ score per quarter before EMR implementation for an average facility.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eSeason indicator\u003c/strong\u003e: the difference in mean DQ score comparing each Quarter X to Quarter 1 (reference) for an average facility.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eEMR\u003c/strong\u003e: mean change in DQ score after the 6-quarter \u0026ldquo;wash-in\u0026rdquo; period compared to DQ score immediately before EMR introduction for an average facility. This is the level change.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTime_after_EMR\u003c/strong\u003e: Slope in the post-EMR period, after the wash-in period of six months. This is the mean change in score each quarter post-EMR implementation compared to quarterly trend before EMR implementation for an average facility. This is the slope change.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eCross-sectional survey analysis:\u003c/p\u003e\n\u003cp\u003eLinear regression models were used to assess the relationship between EMR and facility characteristics and the composite discrepancy and completeness scores (as a measure of data quality). We used the composite discrepancy and completeness scores from quarter 4 of June 2017, i.e., October to November 2017. We accounted for clustering of measures at the facility level. By this time, all facilities had an EMR and the survey was conducted post-EMR. A priori, based on programmatic relevance, we selected EMR type, facility volume, and being in a high-burden county to be included in the multivariable model. We accounted for clustering at the facility level and assessed for interaction between the facility level and facility volume. We conducted a sensitivity multivariable analysis in which we used the full model. Health facilities missing data on facility and EMR implementation characteristics were excluded from the analysis, i.e., complete case analyses. All statistical analyses for this evaluation were done using R version 3.6.2 (2019-12-12).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eBased on the 2017 cross-sectional survey, most facilities (73%) in the sample were health centers and sub-county hospitals (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). Fifty six percent (56%) of facilities were high volume, attending to more than 500 patients in their HIV clinics. Thirty six percent (36%) of facilities were in high-HIV burden counties. Most facilities in this sample implemented KenyaEMR software (56%). With regard to EMR implementation characteristics, 55% facilities used retrospectively entered data after completing paper forms. Most facilities (55%) reported at least one routine data quality assessment exercise that year. Only 17% of facilities reported equipment failure in the preceding week, and most facilities had at least one or two people trained to use EMR.\u003c/p\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003cp\u003e--- INSERT Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e ---\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eHealth Facility (n\u0026thinsp;=\u0026thinsp;187) and EMR implementation characteristics (n\u0026thinsp;=\u0026thinsp;115)\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHealth facility characteristics (n\u0026thinsp;=\u0026thinsp;187)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003en (%)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility level\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDispensary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45 (24)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHealth Center\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCounty/ Sub-County Hospital\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (3)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility volume\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLow volume (\u0026lt;\u0026thinsp;500)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (42)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHigh volume (\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;500)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHigh HIV burden county\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e120 (64)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e67 (36)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEMR type\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eInternational Quality Care (IQCare)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e82 (44)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKenyaEMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eEMR implementation characteristics (data from N\u0026thinsp;=\u0026thinsp;115 facilities)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eImplementation mode\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHybrid\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (20)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePoint of care entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRetrospective data entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e64 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUncategorized\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAnnual number of RDQAs\u003csup\u003e⁑\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e63 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 (8)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (14)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEquipment failure (past week)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95 (83)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (17)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003eHybrid mode \u0026ndash; facility practices both point of care entry and retrospective data entry\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e⁑\u003c/sup\u003e RDQAs- Routine Data Quality Assessments within the year preceding the survey\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003cp\u003eCOMPOSITE DISCREPANCY SCORES\u003c/p\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e provides a descriptive summary of the discrepancy score which had a mean (SD) of 0.07 (0.22), and a median (IQR) of 0.02 (0-0.06).\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eSummary of composite discrepancy scores\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eTotal N\u0026thinsp;=\u0026thinsp;187 facilities over 30 quarters\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eScore approach\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMedian (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMinimum\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMaximum\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiscrepancy score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.07 (0.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.02 (0\u0026ndash;0.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"5\"\u003e\n \u003cp\u003eSD \u0026ndash; Standard Deviation; IQR \u0026ndash; interquartile Range\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n \u003cp\u003eThe interrupted time series graph, centered on time of EMR implementation, shows an initial drop in discrepancy scores between 14 and 13 quarters pre-EMR (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e) likely due to a third of facilities contributing to data in these quarters; however, scores have remained stable over the duration of the evaluation signaling minimal changes to the trend of data quality before and after EMR implementation.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n \u003cp\u003e--- INSERT FIGURE \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e ---\u003c/p\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e summarizes the parameters of the time series model. The average baseline composite discrepancy score at EMR implementation (centered time\u0026thinsp;=\u0026thinsp;0) was 0.0753 [ 95% confidence interval (CI): 0.0482\u0026ndash;0.1025]. In the pre-EMR period, the average quarterly discrepancy score decreased by 0.0012 points (95% CI: -0.0045 \u0026ndash; +0.0020, p-value\u0026thinsp;=\u0026thinsp;0.458), i.e., there was no significant trend.\u003c/p\u003e\n \u003cp\u003eRegarding level change, comparing the values post-EMR, and pre-EMR period; the average composite discrepancy score in the first quarter post-EMR was 0.0118 points higher [ 95% CI: (-0.0248\u0026ndash;0.0484), p-value\u0026thinsp;=\u0026thinsp;0.528] than the average score in the quarter preceding EMR implementation for the average facility. This level change was not statistically significant. Comparing the trends post- and pre-EMR implementation; after six quarters post-EMR, the average change in the composite discrepancy score was 0.0004 points higher (95% CI: -0.0038\u0026ndash;0.0047, p-value\u0026thinsp;=\u0026thinsp;0.844 than the average quarterly trend pre-implementation for the average facility. This change was not significant (Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eAverage discrepancy scores were compared across the four seasons in the pre-EMR period. Season did not predict level of discrepancy score when comparing other seasons with the first calendar quarter (Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003cp\u003e--- INSERT Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e ---\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable id=\"Tab4\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eInterrupted time series analysis of crude time trends in discrepancy scores across 187 health facilities with EMRs, 2011\u0026ndash;2018, Kenya\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u0026beta;*(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cem\u003eDiscrepancy score\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in pre-EMR period (slope)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.0012 (-0.0045 \u0026ndash; +0.0020)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.458\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eLevel change following EMR implementation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0118 (-0.0248 \u0026ndash; +0.0484)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.528\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in 6 quarters post-EMR (slope)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0004 (-0.0038\u0026ndash;0.0047)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eBaseline score at EMR implementation\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0753 (0.0482\u0026ndash;0.1025)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e--\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in reference to season 1 (January to March)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 2 - April to June\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.0173 (-0.0353 \u0026ndash; -0.0007)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.059\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 3 - July to September\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.0097 (-0.0287 \u0026ndash; +0.0092)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.315\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 4 - October to December\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.0104 (-0.0292 \u0026ndash; +0.0084)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.278\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003e\u003cstrong\u003eComputation of Intracluster correlation coefficient (ICC)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween-cluster variation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWithin- cluster variation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e0.034\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eICC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e0.306\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003e\u0026beta;*represents unadjusted quarterly change in mean DQ score using generalized linear mixed models with random intercept for clinics. Time was centered at date of EMR deployment. Time was modeled with a wash-in period of 6 quarters after EMR introduction\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003eBaseline quarter in the time series at EMR implementation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eWith regard to the post-EMR cross-sectional survey of health facility and EMR implementation factors associated with discrepancy scores, significant findings in unadjusted and adjusted analysis have been described below and in Table \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e. In unadjusted analysis, facility type, facility volume, being in a high HIV-burden county, and EMR type were associated with data quality. After adjustment, all these factors remained significant except for the association with high HIV-burden county. Health centers, on average, had a composite discrepancy score that was 0.066 (95% CI: 0.002\u0026ndash;0.130) points higher than that of dispensaries, thus a higher degree of discrepancy indicating poorer data quality (p-value\u0026thinsp;=\u0026thinsp;0.045). High volume facilities \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e(\u0026gt;\u003c/span\u003e\u0026thinsp;500 patients on ART,) on average, had a score that was 0.090 (95% CI: 0.043\u0026ndash;0.138) points higher than that of low volume facilities, thus a higher degree of discrepancy indicating poorer data quality (p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Facilities with KenyaEMR, on average, had a score that was 0.058 (95% CI: -0.107- -0.008) points lower than that of facilities with IQCare, thus a lower degree of discrepancy indicating better data quality (p-value\u0026thinsp;=\u0026thinsp;0.024). Implementation mode (whether point of care, retrospective data entry, or hybrid), annual number of RDQAs, and having experienced equipment failure in the preceding week had no significant association with the discrepancy score.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n \u003cp\u003e--- INSERT Table \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e ---\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab5\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eFacility and EMR implementation correlates of composite discrepancy scores post-EMR \u0026ndash; cross-sectional analysis of survey done in 2017, quarter 4 (October \u0026ndash; December)\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"7\"\u003e\n \u003cp\u003eN\u0026thinsp;=\u0026thinsp;187\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCharacteristic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eN (%)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eMedian (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;*-unadjusted\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e** - \u003cstrong\u003eadjusted (95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility level***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDispensary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45 (24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHealth Center\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e0.066 (0.002\u0026ndash;0.130)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.045\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.093 (0.026\u0026ndash;0.159)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.007\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCounty/Sub-county hosp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e0.042 (-0.022-0.106)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.197\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.045(-0.25-0.114)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.212\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility volume\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e0.090 (0.043\u0026ndash;0.138)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.084 (0.033\u0026ndash;0.136)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHigh HIV burden county\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e0.065 (0.015\u0026ndash;0.116)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.011\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.042 (-0.011-0.094)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.122\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEMR type\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIQCare\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e82 (44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKenyaEMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e-0.058 (-0.107- -0.008)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.024\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.082 (-0.132- -0.031)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.002\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"7\"\u003e\n \u003cp\u003e\u003cstrong\u003eEMR implementation characteristics (data from N\u0026thinsp;=\u0026thinsp;115 facilities)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eImplementation mode\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHybrid\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e23 (20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePoint of care entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e26 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.024 (-0.137-0.090)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.684\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.012 (-0.121-0.097)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.827\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRetrospective data entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e64 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.016 (-0.079-0.111)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.739\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.012 (-0.109-0.085)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.808\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUncategorized\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e2 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.037 (-0.326-0.252)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.801\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.013 (0.033\u0026ndash;0.184)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.945\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAnnual number of RDQAs\u003csup\u003e⁑\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e27 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e63 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.033 (-0.131-0.065)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.511\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.039 (-0.135-0.057)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.426\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e9 (8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.009 (-0.154-0.171)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.916\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.008 (-0.172-0.156)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.922\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e16 (14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEquipment failure (past week)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e95 (83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e20 (17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.020 (-0.118-0.078)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.684\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.034 (-0.128-0.060)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.474\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"7\"\u003e\n \u003cp\u003e\u0026beta;*represents the crude relationship between DQ score and key facility and EMR implementation factors\u003c/p\u003e\n \u003cp\u003e\u0026beta;** represents the adjusted relationship with DQ score adjusting for being in a high vs low HIV burden county, facility volume, and EMR type\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003eHybrid mode \u0026ndash; facility practices both point of care entry and retrospective data entry in parallel\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e⁑\u003c/sup\u003e RDQAs- Routine Data Quality Assessments\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\n \u003cp\u003eANALYSIS OF COMPOSITE COMPLETENESS SCORES\u0026nbsp;\u003c/p\u003e\u003cbr\u003e\n \u003ctable id=\"Tab6\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eSummary statistics of Composite Completeness Scores (CCS)\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eTotal N\u0026thinsp;=\u0026thinsp;187 facilities over 30 quarters\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eScore approach\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMedian (IQR)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMinimum\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMaximum\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCCS score (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e57.8 (28.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e75.0 (37.5\u0026ndash;75.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"5\"\u003e\n \u003cp\u003eSD \u0026ndash; Standard Deviation; IQR \u0026ndash; interquartile Range\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u0026nbsp;\u003ctable id=\"Tab7\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eInterrupted time series analysis of crude time trends in Composite Completeness scores (CCS scores) across 187 health facilities with EMRs, 2011\u0026ndash;2018, Kenya\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u0026beta;*(95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cem\u003eCCS score (%)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in pre-EMR period (slope)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.05 (4.86\u0026ndash;5.23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eLevel change following EMR implementation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-6.96 (-9.15 \u0026ndash; -4.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in 6 quarters post-EMR period (slope)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-1.20 (-1.70 \u0026ndash; -0.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eBaseline score at EMR implementation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79.92 (77.93\u0026ndash;81.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e--\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAverage quarterly change in reference to season 1 (January to March)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 2 - April to June\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-2.21 (-3.47 \u0026ndash; -0.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 3 - July to September\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.33 (-0.97 \u0026ndash; +1.64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.615\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eSeason 4 - October to December\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.43 (-1.74 \u0026ndash; +0.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.520\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003e\u003cstrong\u003eComputation of Intracluster correlation coefficient (ICC)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween-cluster variation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e63.86\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWithin- cluster variation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e352.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eICC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003e0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003e\u0026beta;*represents unadjusted quarterly change in mean DQ score using generalized linear mixed models with random intercept for clinics. Time was centered at date of EMR deployment\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eThe average baseline CCS score for the average facility at EMR implementation was 79.92% (95% CI: 77.93\u0026ndash;81.91). Pre-EMR, the average quarterly CCS score trend was significantly increasing by 5.05 percentage points per quarter (95% CI:4.86\u0026ndash;5.23, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). See Table \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003eThere was a significant level change after transitioning to EMR. The average CCS score in the first quarter post-EMR was 6.96 percentage points lower than the average CCS score in the quarter preceding EMR implementation for the average facility (95% CI: (-9.15 \u0026ndash; -4.77, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). After 6 quarters post-EMR implementation, CCS scores declined and the average quarterly change in CCS was 1.20 percentage points lower than the average quarterly trend pre-EMR (95% CI: -1.70 \u0026ndash; -0.69, \u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n \u003cp\u003eCCS scores were compared across the four seasons in the pre-EMR period. The average quarterly CCS score in the April - June season was 2.21 percentage points lower than that in the January \u0026ndash; March season; this was significant (95% CI: -3.47 \u0026ndash; -0.95, p\u0026thinsp;=\u0026thinsp;0.001). There were no significant differences in average CCS scores when comparing the January \u0026ndash; March season with the other seasons (i.e., July \u0026ndash; September and October \u0026ndash; December).\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab8\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eFacility and EMR implementation characteristics associated with Composite Completeness scores (CCS) in the October-December 2017 cross-sectional survey\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003eN\u0026thinsp;=\u0026thinsp;187\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eCharacteristic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eN (%)\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eMedian (IQR)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;*-unadjusted\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e** - \u003cstrong\u003eadjusted (95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility level***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDispensary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45 (24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHealth Center\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.49 (2.34\u0026ndash;10.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.002\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.16 (3.94\u0026ndash;12.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCounty/Sub-county hosp\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68 (36.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9.43 (5.28\u0026ndash;13.59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10.39 (5.96\u0026ndash;14.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 (3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFacility volume***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.61 (4.31\u0026ndash;10.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.54 (1.06\u0026ndash;8.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHigh HIV burden county\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.68 (-0.87\u0026ndash;6.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.95 (0.19\u0026ndash;7.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.039\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEMR type\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIQCare\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e82 (44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKenyaEMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.16 (-1.32\u0026ndash;5.65)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.67 (-0.61\u0026ndash;5.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.111\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eEMR implementation characteristics (data from N\u0026thinsp;=\u0026thinsp;115 facilities)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eImplementation mode\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHybrid\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePoint of care entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.54 (-2.73\u0026ndash;11.81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.75 (-3.37\u0026ndash;10.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.302\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRetrospective data entry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e64 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.44 (-2.73\u0026ndash;9.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.227\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.18 (-1.99\u0026ndash;12.35)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.157\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUncategorized\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 (2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.27 (-18.99\u0026ndash;18.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.977\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-1.19 (-26.55\u0026ndash;24.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.926\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAnnual number of RDQAs\u003csup\u003e⁑\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e63 (55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.61 (0.52\u0026ndash;12.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.033\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.05 (-0.96\u0026ndash;13.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.093\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u0026ge;\u003c/span\u003e\u0026thinsp;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 (8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.57 (1.38\u0026ndash;21.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.026\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9.77 (-1.95\u0026ndash;21.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMissing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16 (14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEquipment failure (past week)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95 (83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRef\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.10 (-6.34\u0026ndash;6.14)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.15 (-6.76\u0026ndash;6.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.965\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003e\u003csup\u003e⁑\u003c/sup\u003e RDQAs- Routine Data Quality Assessments; CI \u0026ndash; Confidence interval; IQR \u0026ndash; Interquartile range\u003c/p\u003e\n \u003cp\u003e\u0026beta;*represents the crude relationship between DQ score and key facility and EMR implementation factors\u003c/p\u003e\n \u003cp\u003e\u0026beta;**For facility characteristics, the multivariable model included facility level, being in a high vs low HIV burden county, facility volume, and EMR type. For EMR characteristics, the multivariable model included facility level, being in a high vs low HIV burden county, facility volume, EMR type, mode of data entry, annual number of RDQAs and equipment failure.\u003c/p\u003e\n \u003cp\u003e***No interaction found between facility level and facility volume.\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003eHybrid mode \u0026ndash; facility practices both point of care entry and retrospective data entry in parallel\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n \u003cp\u003eAll facility level characteristics, except EMR type, were significantly associated with the composite completeness scores. In multivariable analysis, health centers and hospitals had higher completeness compared to dispensaries i.e., 8.16 percentage point higher (95% CI: 3.94\u0026ndash;12.37, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and 10.39 percentage point higher (95% CI: 5.96\u0026ndash;14.80, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) completeness, respectively. Higher facility volume was associated with a 4.54 percentage point higher completeness compared to low volume facility (95% CI: 1.06\u0026ndash;8.02, p\u0026thinsp;=\u0026thinsp;0.010). Being in a high HIV burden county was associated with a 3.95 percentage point higher completeness (95% CI: 0.19\u0026ndash;7.70, p\u0026thinsp;=\u0026thinsp;0.039) compared to being in a low HIV burden county. Among EMR implementation characteristics, the only characteristic that was associated with higher CCS in bivariable analysis was having one or more annual RDQAs.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this evaluation we evaluated the impact of EMR implementation on data quality in the DHIS-2 system, as captured by composite discrepancy scores and CCS, across 187 facilities in Kenya from 2011 to 2018. Overall, changes in the discrepancy score comparing different periods pre- and post-EMR were very small (most estimates were smaller than 0.01) and thus within 1 \u003cem\u003eZ-\u003c/em\u003escore i.e., 1 standard deviation (SD), and CCS seemed to significantly decline after EMR implementation. We found no evidence that EMR implementation was associated with favorable changes in DHIS2 in the discrepancy scores and CCS. This could be attributed to the low (favorable) baseline discrepancy scores and prior local efforts to improve completeness in paper-based systems before EMRs were introduced.\u003c/p\u003e \u003cp\u003eA cross-sectional facility survey done in 2017 unveiled health facility and EMR implementation factors that were associated with discrepancy and completeness of data. While the differences in Z-score deviations were mostly less than 0.1, the differences were significant enough to reveal associated factors. High HIV burden counties were associated with worse data quality (high discrepancy scores), potentially due to high workload and collateral increase in documentation burden. Operating KenyaEMR software was associated with lower discrepancy scores \u0026ndash; thus better data quality \u0026ndash; compared to using IQCare. The underlying reasons could be usability profiles of the two EMR systems. KenyaEMR was more closely aligned to the paper-based MOH compared to IQCare. With regard to CCS, higher level facilities (i.e., health centers and hospitals), high volume clinics and high HIV burden counties were associated with higher completeness scores. This could have been due to access to more resources, staff (like data managers) and frequency of supervision visits in busier health facilities [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. We found that mode of implementation (i.e., point-of-care, retrospective or hybrid data entry) was not associated with discrepancy or completeness. A small proportion of facilities (23%) maintained fidelity to the intended mode of EMR use, i.e., point-of-care entry.\u003c/p\u003e \u003cp\u003eThere is existing research describing data quality in the context of electronic data systems, but less research examining the evidence for causal effects of EMR implementation on data quality. Previous studies in sub-Saharan Africa have shown a mixed picture in terms of levels of data accuracy for HIV-related indicators. Several studies have shown high data quality of EMR records [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan additionalcitationids=\"CR33 CR34 CR35 CR36\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], while others have demonstrated overall poor EMR data quality [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Notably, there were no studies that compared data quality in EMRs and paper records longitudinally. A pre-post study in Iran found that vocal-electronic documentation was associated with higher quality data than paper documentation [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Two cross-sectional studies compared electronic and paper records concurrently. These studies found that data quality was comparable across the electronic and paper records [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. A one-month randomized controlled study in Ethiopia demonstrated better data quality and efficiency with electronic compared to paper-based survey data [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur lack of positive findings suggest that presence of EMRs is not insufficient to achieve strong aggregated data quality used in reporting; optimization of EMR usage and data exchange is a necessary requirement for enhancing data quality [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Some barriers to improved data quality during EMR implementation include sub-optimal EMR use which may be associated with insignificant changes in data quality post-EMR implementation [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Recent studies evaluating the actual use of EMR implementation have demonstrated concerningly low EMR usage and data exchange [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Ngugi et al, and colleagues measured seven EMR usage indicators and found gaps in active use of EMRs and data completeness in Kenya [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Factors affecting EMR usage need further exploration and remedial interventions [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. An important consideration is variability in the approach health workers use to generate EMR aggregate summaries that are reported in DHIS2. It is likely that some health workers may manually transcribe EMRs summaries for retrospective upload to DHIS2, while others may not use the EMR and thus the EMR and DHIS2 would run parallel to each other.\u003c/p\u003e \u003cp\u003eDual/ parallel documentation in EMR and paper records is huge barrier to EMR usage and good data quality. In this evaluation, we found that 75% of facilities had a hybrid or retrospective data entry system \u0026ndash; they performed both point-of-care and retrospective data entry, which would contribute to inefficiencies and errors in data management. A study in Malawi found that retrospective data entry by clerks was associated with more errors than point of care entry by clinicians [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Assisting facilities to channel efforts toward point-of-care EMR usage would be a possible solution for improving data quality [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. A study in Kenya established the benefit of RDQAs in improving EMR data quality [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. While the sub-analysis of factors in the present evaluation showed that there was no association between RDQAs and data quality, it is acknowledged that the cross-sectional survey may have had inherent bias and insufficient power to detect all significant associations.\u003c/p\u003e \u003cp\u003eThis evaluation points to a constellation of factors beyond the presence of EMR that may impact DHIS2 data quality. Other factors that may have been at play include facility level, facility volume, HIV burden, EMR type, and frequency of RDQAs as demonstrated by the cross-sectional survey. An important consideration is variability in the approach health workers used to generate EMR aggregate summaries that are uploaded to DHIS2. It is likely that some health workers manually transcribed EMRs summaries for retrospective upload to DHIS2, while others did not use the EMR summaries, and thus the EMR summaries and DHIS2 reports would not be synchronized. In resource limited settings, it can take a considerable duration of time to successfully implement and sustain EMR software \u0026ndash; this should be kept in mind when evaluating EMRs [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. A parallel focus on EMR usage, sustainability, and contextual factors is necessary for optimizing RHIS in LMICs [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eEMR implementation is feasible; however, the challenges of streamlining use of EMR, and lack of interoperability with DHIS2, are considerable. Other potential implementation challenges include health provider and health system constraints, and EMR set up and maintenance costs [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan additionalcitationids=\"CR50\" citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. More ongoing work is needed to understand systemic deficits that hinder health provider acceptance, uptake, and meaningful use of EMRs [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. It can be hypothesized that poor uptake and use of EMR would limit realization of the benefits of EMRs for DHIS2 data reporting. This work should be coupled with continuous quality improvement efforts and design of contextually appropriate implementation strategies to strengthen the uptake of EMRs. Strategies such as RDQAs \u0026ndash; as demonstrated in this study \u0026ndash;, training health providers, infrastructure support, modifying electronic tools, designing usable and interoperable systems, and sustainable technical support are critical for optimal implementation of EMRs [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan additionalcitationids=\"CR54\" citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eStrengths:\u003c/p\u003e \u003cp\u003eThis evaluation serves as a blueprint for future longitudinal evaluations of mature EMRs. To our knowledge this is the first nationwide longitudinal analyses of data quality that spans pre-EMR and post-EMR periods that assesses the causal relationship between EMR implementation and aggregate data quality. Prior studies in low- and middle-income countries (LMICs) have considerable limitations including small sample sizes, had shorter follow-up periods or cross-sectional designs, use of less robust designs, or lack long-term comparative evaluations of EMRs and paper systems. In light of this, the present evaluation is a much-needed addition to the literature. Another strength includes the use of the time series models which eliminate confounding by time invariant factors. The stepped/ differential introduction of EMRs across multiple facilities by default creates a multiple baseline design which is useful in controlling for possible time-varying confounders [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eLimitations:\u003c/p\u003e \u003cp\u003eWhile we present a strong quasi-experimental design, it is not without limitations. First, while we aim to obtain evidence for causal effects, we recognize that this robust study design is still susceptible to confounding by time varying factors and that some of the assumptions for the time series analysis may not have been fulfilled. Examples of time varying factors include recurrent staff turnovers and program-level changes to data documentation and tools. Second, we were not able to obtain granular monthly data, but were restricted to quarterly data until 2018 which limits our understanding of the month-to-month variations in data quality to date. Third, the cross-sectional survey design, for understanding facility and EMR implementation correlates of data quality, is vulnerable to selection bias and residual confounding inherent to the design. We were also not able to understand the level of interoperability between EMR and DHIS2 and how providers navigated these technical issues to generate DHIS2 summaries. Given power calculations were not used to determine the sample size for the survey, the cross-sectional analysis may have been underpowered to detect all significant associations between facility or EMR implementation factors and data quality. Finally, this evaluation focused on HIV-specific aspects of DHIS2, thus findings cannot be generalized to other health service departments, and further research would be needed to assess effects beyond HIV service delivery.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWith regard to implications for research, practice, and policy, we found no evidence to suggest that EMRs in HIV facilities in Kenya improved aggregated data quality as captured by the discrepancy and completeness scores. This does not detract from the multiple mechanisms through which EMRs could have effects on quality of care and health outcomes. We found that higher facility level, higher facility volume, and KenyaEMR type were associated with lower discrepancy; and facility level, higher patient volume, high HIV burden were associated with higher completeness. These findings suggest that the presence of EMRs is insufficient to achieve high DHIS2 data quality. Further understanding how these factors interact with EMRs and data management processes would be necessary to optimize the value of EMRs and DHIS2 data quality. In the interim, implementation research is needed to understand factors associated with sustained EMR use and data quality, and further characterize EMR implementation gaps. Further research is needed to track the impact of strategies to augment EMR-DHIS2 interoperability for better data quality beyond the studied time frame.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAuthor Contribution\u003c/p\u003e\n\u003cp\u003eB.M.O. conception, design, analysis, prepared figures and tables, and wrote the manuscript. O.A. design, analysis and interpretation of data. B.H.W. interpretation of the data and revision of manuscript. J.P.H. design, analysis, and interpretation of data. S.G. interpretation of data. A.N. conception, acquisition, and analysis. N.P. conception, design, interpretation, and revision of manuscript. All authors reviewed the manuscript and approved its submission for publication.\u003c/p\u003e\n\u003cp\u003eAcknowledgement\u003c/p\u003e\n\u003cp\u003eWe would like to acknowledge the Ministry of Health-Kenya, I-TECH, and Palladium for facilitating access to the data for analysis.\u003c/p\u003e\n\u003cp\u003eData Availability\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003e\u003cspan\u003eAqil A, Lippeveld T, Hozumi D. PRISM framework: a paradigm shift for designing, strengthening and evaluating routine health information systems. Health Policy Plan. 2009;24:217\u0026ndash;28. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/heapol/czp010\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWagenaar BH, Sherr K, Fernandes Q, Wagenaar AC. Using routine health information systems for well-designed health evaluations in low-and middle-income countries. Health Policy Plan. 2016;31:129\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLeon N, Balakrishna Y, Hohlfeld A, Odendaal WA, Schmidt B-M, Zweigenthal V, et al. Routine Health Information System (RHIS) improvements for strengthened health system management. Cochrane Database Syst Rev. 2020;8:CD012012. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/14651858.CD012012.pub2\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWHO. Global diffusion of eHealth: making universal health coverage achievable: report of the third global survey on eHealth. World Health Organization; 2017.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWHO. WHO guideline: recommendations on digital interventions for health system strengthening: web supplement 2: summary of findings and GRADE tables. World Health Organization; 2019.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWHO. Electronic health records: manual for developing countries. Manila: WHO Regional Office for the Western Pacific; 2006.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eJacob N, Rice B, Kalk E, Heekes A, Morgan J, Brinkmann S, et al. Consolidating strategic information to monitor progress against the UNAIDS 90-90-90 targets: evaluating the operational feasibility of an electronic HIV testing register in Cape Town, South Africa. BMC Health Serv Res. 2020;20:720. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12913-020-05517-7\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLey B, Rijal KR, Marfurt J, Adhikari NR, Banjara MR, Shrestha UT, et al. Analysis of erroneous data entries in paper based and electronic data collection. BMC Res Notes. 2019;12:537. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13104-019-4574-8\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eCastelnuovo B, Kiragga A, Afayo V, Ncube M, Orama R, Magero S, et al. Implementation of provider-based electronic medical records and improvement of the quality of data in a large HIV program in Sub-Saharan Africa. PLoS ONE. 2012;7:e51631. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0051631\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMcLean E, Dube A, Saul J, Branson K, Luhanga M, Mwiba O, et al. Implementing electronic data capture at a well-established health and demographic surveillance site in rural northern Malawi. Glob Health Action. 2017;10:1367162. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/16549716.2017.1367162\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMukasa O, Mushi HP, Maire N, Ross A, de Savigny D. Do surveys with paper and electronic devices differ in quality and cost? Experience from the Rufiji Health and demographic surveillance system in Tanzania. Glob Health Action. 2017;10:1387984. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/16549716.2017.1387984\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eFlaxman AD, Stewart A, Joseph JC, Alam N, Alam SS, Chowdhury H, et al. Collecting verbal autopsies: improving and streamlining data collection processes using electronic tablets. Popul Health Metr. 2018;16:3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12963-018-0161-9\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eSingh Y, Jackson D, Bhardwaj S, Titus N, Goga A. National surveillance using mobile systems for health monitoring: complexity, functionality and feasibility. BMC Infect Dis. 2019;19:786. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12879-019-4338-z\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWHO. eHealth and innovation in women\u0026rsquo;s and children\u0026rsquo;s health: a baseline review: based on the findings of the 2013 survey of CoIA countries by the WHO Global Observatory for eHealth. March 2014. 2014.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eOluoch TO. Electronic medical records and clinical Decision Support Systems in HIV care in resource-limited settings. 9789462596313; 2015.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eOluoch T, Kwaro D, Ssempijja V, Katana A, Langat P, Okeyo N et al. Better adherence to pre-antiretroviral therapy guidelines after implementing an electronic medical record system in rural Kenyan HIV clinics: a multicenter pre-post study. doi:D - NLM: HHSPA796290 OTO - NOTNLM.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eOluoch T, Santas X, Fau - Kwaro D, Kwaro D, Fau - Were M, Were M, Fau - Biondich P, Biondich P, Fau - Bailey C, Bailey C, Fau - Abu-Hanna A et al. The effect of electronic medical record-based clinical decision support on HIV care in resource-constrained settings: a systematic review.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eChan KS, Fowles JB, Weiner JP. Review: Electronic Health Records and the Reliability and Validity of Quality Measures: A Review of the Literature. Med Care Res Rev. 2010;67:503\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1077558709359007\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBian J, Lyu T, Loiacono A, Viramontes TM, Lipori G, Guo Y, et al. Assessing the practice of data quality evaluation in a national clinical data research network through a systematic scoping review in the era of real-world data. J Am Med Inf Assoc JAMIA. 2020;27:1999\u0026ndash;2010. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jamia/ocaa245\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDistrict Health Information Software 2 [Internet]. Oslo: Health Information Systems Programme. 2012. [Accessed April 9, March 2022]. Available: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.hisp.uio.no/\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBraa J, Sahay S. The DHIS2 open source software platform: evolution over time and space. LF Celi Glob Health Inf. 2017;451.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBraa J, Heywood A, Sahay S. Improving quality and use of data through data-use workshops: Zanzibar, United Republic of Tanzania. Bull World Health Organ. 2012;90:379\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDehnavieh R, Haghdoost A, Khosravi A, Hoseinabadi F, Rahimi H, Poursheikhali A, et al. The District Health Information System (DHIS2): A literature review and meta-synthesis of its strengths and operational challenges based on the experiences of 11 countries. Health Inf Manag. 2018;1833358318777713. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1833358318777713\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAmoroso CL, Akimana B, Wise B, Fraser HS. Using electronic medical records for HIV care in rural Rwanda. 2010. pp. 337\u0026ndash;341.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMuthee V, Bochner AF, Osterman A, Liku N, Akhwale W, Kwach J, et al. The impact of routine data quality assessments on electronic medical record data quality in Kenya. PLoS ONE. 2018;13:e0195362. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0195362\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eTsai CH, Eghdam A, Davoody N, Wright G, Flowerday S, Koch S. Effects of Electronic Health Record Implementation and Barriers to Adoption and Use: A Scoping Review and Qualitative Analysis of the Content. Life Basel Switz. 2020;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/life10120327\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eLee K, Weiskopf N, Pathak J. A framework for data quality assessment in clinical research datasets. American Medical Informatics Association; 2017. p. 1080.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eOdeny B, Hughes JP, Gloyd S, Njoroge A, Wagenaar BH, Puttkammer N. Development of novel composite data quality scores to evaluate facility-level data quality in the District Health Information System (DHIS2) in Kenya: a nationwide retrospective cohort study. doi:Under review.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eXiao H, Augusto O, Wagenaar BH. Reflection on modern methods: a common error in the segmented regression parameterization of interrupted time-series analyses. Int J Epidemiol. 2021;50:1011\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eGarrib A, Stoops N, McKenzie A, Dlamini L, Govender T, Rohde J, et al. An evaluation of the District Health Information System in rural South Africa. South Afr Med J Suid-Afr Tydskr Vir Geneeskd. 2008;98:549\u0026ndash;52.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNoureldin M, Mosallam R, Hassan SZ. Quality of documentation of electronic medical information systems at primary health care units in Alexandria, Egypt. East Mediterr Health J Rev Sante Mediterr Orient Al-Majallah Al-Sihhiyah Li-Sharq Al-Mutawassit. 2014;20: 105\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNicol E, Dudley L, Bradshaw D. Assessing the quality of routine data for the prevention of mother-to-child transmission of HIV: An analytical observational study in two health districts with high HIV prevalence in South Africa. Int J Med Inf. 2016;95:60\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ijmedinf.2016.09.006\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eO\u0026rsquo;Hagan R, Marx MA, Finnegan KE, Naphini P, Ng\u0026rsquo;ambi K, Laija K, et al. National assessment of data quality and associated systems-level factors in Malawi. Glob Health Sci Pract. 2017;5:367\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNisingizwe MP, Iyer HS, Gashayija M, Hirschhorn LR, Amoroso C, Wilson R, et al. Toward utilization of data for program management and evaluation: quality assessment of five years of health management information system data in Rwanda. Glob Health Action. 2014;7:25829.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNagbe T, Yealue K, Yeabah T, Rude JM, Fallah M, Skrip L, et al. Integrated disease surveillance and response implementation in Liberia, findings from a data quality audit, 2017. Pan Afr Med J. 2019;33:10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.11604/pamj.supp.2019.33.2.17608\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAbiy R, Gashu K, Asemaw T, Mitiku M, Fekadie B, Abebaw Z, et al. A Comparison of Electronic Medical Record Data to Paper Records in Antiretroviral Therapy Clinic in Ethiopia: What is affecting the Quality of the Data? Online J Public Health Inf. 2018;10:e212. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5210/ojphi.v10i2.8309\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAhmed R, Robinson R, Elsony A, Thomson R, Squire SB, Malmborg R, et al. A comparison of smartphone and paper data-collection tools in the Burden of Obstructive Lung Disease (BOLD) study in Gezira state, Sudan. PLoS ONE. 2018;13:e0193917. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0193917\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eEndriyas M, Alano A, Mekonnen E, Ayele S, Kelaye T, Shiferaw M, et al. Understanding performance data: health management information system data accuracy in Southern Nations Nationalities and People\u0026rsquo;s Region, Ethiopia. BMC Health Serv Res. 2019;19:175. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12913-019-3991-7\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMohammadi Firouzeh M, Jafarjalal E, Emamzadeh Ghasemi HS, Bahrani N, Sardashti S. Evaluation of vocal-electronic nursing documentation: A comparison study in Iran. Inf Health Soc Care. 2017;42:250\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/17538157.2016.1178119\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eWalther B, Hossin S, Townend J, Abernethy N, Parker D, Jeffries D. Comparison of electronic data capture (EDC) with the standard data capture method for clinical trial data. PLoS ONE. 2011;6:e25348. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0025348\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eZeleke AA, Worku AG, Demissie A, Otto-Sobotka F, Wilken M, Lipprandt M et al. Evaluation of Electronic and Paper-Pen Data Capturing Tools for Data Quality in a Public Health Survey in a Health and Demographic Surveillance Site, Ethiopia: Randomized Controlled Crossover Health Care Information Technology Evaluation. JMIR MHealth UHealth. 2019;7: e10995. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2196/10995\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNicol E, Hanmer LA. Routine Health Information Systems in South Africa\u0026ndash;Opportunities for Improvement. Stud Health Technol Inf. 2015;216:993.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eNgugi P, Babic A, Were MC. A multivariate statistical evaluation of actual use of electronic health record systems implementations in Kenya. PLoS ONE. 2021;16:e0256799. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0256799\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eGloyd S, Wagenaar BH, Woelk GB, Kalibala S. Opportunities and challenges in conducting secondary analysis of HIV programmes using data from routine health information systems and personal health information. J Int AIDS Soc. 2016;19:20847. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7448/IAS.19.5.20847\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBisrat A, Minda D, Assamnew B, Abebe B, Abegaz T. Implementation challenges and perception of care providers on Electronic Medical Records at St. Paul\u0026rsquo;s and Ayder Hospitals, Ethiopia. BMC Med Inf Decis Mak. 2021;21:306. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12911-021-01670-z\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eDouglas G, McKay M, Bwanali M, Gadabu O, Mumba S, Gondwe A. A comparison of data captured at point of care versus data captured retrospectively. AMIA Annu Symp Proc AMIA Symp. 2008; 934.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eTweya H, Feldacker C, Ben-Smith A, Harries AD, Komatsu R, Jahn A, et al. Simplifying ART cohort monitoring: can pharmacy stocks provide accurate estimates of patients retained on antiretroviral therapy in Malawi? BMC Health Serv Res. 2012;12:210. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1472-6963-12-210\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eTweya H, Feldacker C, Gadabu OJ, Ng\u0026rsquo;ambi W, Mumba SL, Phiri D, et al. Developing a point-of-care electronic medical record system for TB/HIV co-infected patients: experiences from Lighthouse Trust, Lilongwe, Malawi. BMC Res Notes. 2016;9:146. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13104-016-1943-4\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMyburgh H, Murphy JP, van Huyssteen M, Foster N, Grobbelaar CJ, Struthers HE, et al. Implementation of an electronic monitoring and evaluation system for the antiretroviral treatment programme in the Cape Winelands district, South Africa: a qualitative evaluation. PLoS ONE. 2015;10:e0127223. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0127223\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAminpour F, Sadoughi F, Ahamdi M. Utilization of open source electronic health record around the world: A systematic review. J Res Med Sci. 2014;19:57\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eHillestad R, Bigelow J, Bower A, Girosi F, Meili R, Scoville R, et al. Can electronic medical record systems transform health care? Potential health benefits, savings, and costs. Health Aff (Millwood). 2005;24:1103\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBhavaraju SR. From subconscious to conscious to artificial intelligence: A focus on electronic health records. Neurol India. 2018;66:1270\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4103/0028-3886.241377\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eAkanbi MO, Ocheke AN, Agaba PA, Daniyam CA, Agaba EI, Okeke EN et al. Use of Electronic Health Records in sub-Saharan Africa: Progress and challenges. 2012.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eMuthee V, Bochner AF, Kang\u0026rsquo;a S, Owiso G, Akhwale W, Wanyee S et al. Site readiness assessment preceding the implementation of a HIV care and treatment electronic medical record system in Kenya.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eKing JD, Buolamwini J, Cromwell EA, Panfel A, Teferi T, Zerihun M, et al. A novel electronic data collection system for large-scale surveys of neglected tropical diseases. PLoS ONE. 2013;8:e74570. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0074570\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003eBernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2017;46:348\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-5672455/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5672455/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e The objective of this evaluation was to estimate the effect of electronic medical record system (EMR) implementation on the quality of data uploaded to the District Health Information System Version 2 platform (DHIS2).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e This was an interrupted time series analysis of DHIS2 data quality. Data were extracted from 187 Kenyan health facilities from January 2011 to June 2018 (i.e., spanning 30 quarters). The primary exposure was presence of EMR, and the primary data quality outcomes were quarterly composite discrepancy scores and composite completeness scores. The composite discrepancy score depicted the extent of deviation of observed values from plausible values based on internal consistency checks. Higher discrepancy scores reflected worse data quality. The composite completeness score (CCS score) was a percentage measure of the extent of documentation of pre-selected variables. A 2017 cross-sectional facility survey was used to assess factors associated with data quality. We conducted an interrupted time series analysis to determine changes in the trend of data quality scores before and after EMR implementation. We conducted multivariable linear regression analyses to determine factors associated with data quality.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eThere was no statistically significant level change or effect in composite discrepancy scores comparing pre-EMR period and the post-EMR period. In the cross-sectional analysis, on average health centers had higher composite discrepancy scores compared to dispensaries thus worse data quality (0.066; 95% CI: 0.002-0.130, p=0.045), high volume facilities (\u0026gt;500 patients) had higher discrepancy scores than low volume facilities (0.090; 95% CI: 0.043-0.138, p\u0026lt;0.001), and operating the KenyaEMR system was associated with less discrepancy scores and thus better data quality (0.058; 95% CI: -0.107- -0.008, p=0.024] than the IQCare system. Regarding CCS, there was a significant drop in composite completeness scores (CCS) after transitioning to EMR. The average CCS in the first quarter post-EMR was lower than the average CCS in the quarter preceding EMR implementation (6.96; 95% CI: -9.15 – -4.77, p\u0026lt;0.001). After six quarters post-EMR implementation, CCS declined steadily with an average quarterly change in CCS that was 1.20 percentage points lower than the average quarterly trend pre-EMR (95% CI: -1.70 – -0.69, \u0026lt;0.001). In cross-sectional analysis, health centers (8.16; 95% CI: 3.94 – 12.37, p\u0026lt;0.001) and hospitals (10.39; 95% CI: 5.96 – 14.80, p\u0026lt;0.001), high facility volume (4.54; 95% CI: 1.06 – 8.02, p=0.010) and high HIV burden county (3.95; 95% CI: 0.19 – 7.70, p= 0.039) were associated with higher CCS compared to dispensaries, low facility volume, and low HIV burden, respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eEMR implementation did not demonstrate evidence for significant positive impact on DHIS2 data quality, as indicated by the lack of improvement in composite discrepancy scores and a drop in composite completeness scores post-EMR implementation. Our findings suggest that EMRs are not sufficient to ensure high-quality data. Facility characteristics (like higher level facility, high volume, and being in a high HIV burden county), and KenyaEMR use appear to be associated with discrepancy and completeness of data. Further research to explore the mechanistic link between EMRs, data quality, and context will be necessary to optimize the use of EMRs to improve data quality in routine health information system data in LMICs.\u003c/p\u003e","manuscriptTitle":"The effect of introducing an electronic medical record system on data quality and factors associated with data quality across 187 HIV clinics in Kenya: An interrupted time series analysis from 2011-2018","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-01-03 14:30:52","doi":"10.21203/rs.3.rs-5672455/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"3594e9c2-6160-4f51-9fbc-790a64bf01f4","owner":[],"postedDate":"January 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-01-13T10:38:12+00:00","versionOfRecord":[],"versionCreatedAt":"2025-01-03 14:30:52","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5672455","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5672455","identity":"rs-5672455","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}