Incompleteness and misclassification of maternal deaths in Zimbabwe: data from two reproductive age mortality surveys, 2007-08 and 2018-19

Incompleteness and misclassification of maternal deaths in Zimbabwe: data from two reproductive age mortality surveys, 2007-08 and 2018-19 | 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 Incompleteness and misclassification of maternal deaths in Zimbabwe: data from two reproductive age mortality surveys, 2007-08 and 2018-19 Reuben Musarandega, Lennarth Nystrom, Grant Murewanhema, Chipo Gwanzura, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3891799/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Introduction: Reducing maternal mortality is a high-priority global health goal, especially in sub-Saharan Africa, where the maternal mortality ratios (MMRs) of most of the countries is higher than the average global MMR. We implemented two cross-sectional reproductive age mortality surveys, in 2007-08 and 2018-19, to assess changes in the MMR and causes of death in Zimbabwe after a raft of interventions implemented to reduce maternal mortality. This paper analysed the missingness and misclassification of deaths in the surveys. Methods: We compared percentages of missed deaths in each survey using the Chi-square test. The risk ratios of missing deaths in different data sources in each survey were calculated using log-linear regression models. Proportions of misclassified deaths were compared using Fisher’s exact test and sensitivity and specificity of incompleteness and misclassification of deaths compared using the six-box method and the Chi-square test. Results: The proportion of missed pregnancy-related deaths decreased from 27% in 2007-08 to 18% in 2018-19 (p=0.044) and the proportion of missed maternal deaths decreased from 30% in 2007-08 to 19% in 2018-19. Misclassification of maternal deaths in health records was 17% in 2007-08 and 8.5% in 2018-19 (p=0.160). The proportion of pregnancy-related deaths identified through health records increased from 11% in 2007-08 to 60% in 2018-19 (p<0.001). Sensitivity of incompleteness and misclassification of deaths was 95% in 2007-08 and 77% in 2018-19, and specificity was 29% and 83% respectively. Conclusion: Incompleteness and misclassification of maternal deaths are still a challenge in Zimbabwe. Maternal death studies must triangulate data sources to improve the completeness of data and efforts to reduce misclassification of deaths should continue to improve maternal mortality estimates. Maternal deaths Pregnancy-related deaths Maternal mortality Missingness Incompleteness Misclassification Figures Figure 1 1. INTRODUCTION Reducing maternal mortality is a high-priority global health goal [ 1 ], particularly in sub-Saharan Africa, where the average maternal mortality ratio (MMR) was estimated at 536 maternal deaths per 100 000 live births in 2020 compared to the global average MMR of 223 maternal deaths per 100 000 live births [ 2 ]. Zimbabwe (357, uncertainty interval (UI): 255 to 456) was among the East and Southern Africa countries with an MMR above the global average MMR (223; UI: 202 to 255) [ 2 ]. The incompleteness of data also referred to as “missingness” or under-reporting [ 2 , 3 ], is one of the problems impeding progress on the goal to reduce maternal mortality [ 4 ], It affects all sources of data – surveys, surveillance and civil registration and vital statistics (CRVS) data [ 5 ]. Incompleteness of maternal deaths occurs when deaths are not recorded and reported in the data [ 6 ]. Deaths that occur outside health institutions are affected the most as these are often not documented. Deaths occurring in institutions are documented but sometimes not reported in surveillance systems [ 6 ]. The incompleteness of maternal deaths causes under-estimation of MMRs [ 7 ]. Researchers use a wide range of statistical techniques to adjust MMRs for missingness such as the Brass Growth Balance Method for censuses and household surveys [ 8 , 9 ], and the Bayesian Maternal Mortality Misclassification (BMI) model for global estimates [ 2 , 4 ]. However, these techniques may end up over-estimating the MMRs [ 10 ]. As a result, countries may dispute the MMR estimates generated by global models [ 11 ]. Misclassification occurs when a maternal death is classified as non-maternal or a non-maternal death classified as maternal [ 4 , 12 ]. Assigning incorrect causes leads to misclassification of deaths, which leads to under or over-estimating the MMR [ 12 ]. Deaths can also be misclassified by assigning them to the wrong cause-group when coders are not competent enough to correctly code the deaths [ 3 ]. Inaccurate MMR estimates will lead to inappropriate planning and misallocating resources to maternal mortality interventions. Ultimately, national, and global goals are hindered. Achievement of Sustainable Development Goal (SDG) 3.1 for maternal mortality is threatened by this challenge. We performed two reproductive age mortality surveys (RAMOS) in 2007-08 and 2018-19 to assess changes in the MMR and causes of maternal deaths in Zimbabwe following a raft of interventions implemented to reduce maternal mortality. A RAMOS enumerates the deaths of all women of reproductive age and identifies maternal deaths out of these, to calculate the maternal mortality ratio. In this paper, we estimated the level of incompleteness and misclassification of pregnancy-related and maternal deaths in the two surveys. We recommend measures to improve the completeness and classification of maternal deaths in Zimbabwe and similar countries. 2. MATERIAL AND METHODS 2.1 Study design We conducted the two cross-sectional RAMOS in 11 districts of Zimbabwe, applying multi-stage cluster sampling where the population was stratified into provinces (n=10). One district was selected from each province using simple random sampling. An additional district was chosen in Harare – the province with the largest population. Papers published previously described the study methods in detail [13-15]. 2.2 Study setting Three of the 11 study districts (Figure 1) were partially urban (Bindura, Kwekwe and Mutare); three were urban (Harare Southeastern, Harare Western and Nkulumane) and five were rural (Chivi, Matobo, Mutoko, Tsholotsho and Zvimba). [Figure 1] 2.3 Data collection process and data sources We used trained research nurse midwives to collect data in the two surveys, guided by standard operating procedures from the study protocol. They collected live births from maternity registers in hospitals and health centres. They recorded reproductive age deaths from patient registers and charts at maternity units, operating theatres, high dependency and intensive care units, gynaecological, medical, and surgical wards, mortuaries, police posts, and casualty departments in hospitals and maternity records at health centres. Medical staff completed the medical charts while attending to patients in the hospitals and health centres. We collected additional deaths from the government’s Registrar General’s (RG’s) offices. The RG’s office registered deaths on death notification forms in civil registration and vital statistics (CRVS) records through reports submitted by health facilities, the police, or members of the public. Police submitted reports of deaths that they attended to at home and took to a hospital for a post-mortem. At the health facilities, the medical officer who attended to the death that occurred at the health institution or brought by the police for post-mortem documented the underlying and antecedent or contributory causes of the death on the death record according to International Classification of Disease Version 10 (ICD-10) [16]. The public reported home deaths that the police could not attend to, where village heads or headmen wrote a death confirmation letter. The relatives registered the death at the RG’s offices through the community head’s letter and their verbal report. The RG’s offices also conducted community registration outreaches periodically where they registered births and deaths posthumously in the community [17, 18]. The RG’s office created a death record for all deaths reported by the three means, on which they documented the date, place and causes of death available on the source of information (medical certificate, police report or family members’ report). They issued a death certificate with a reference number (year of issue/sequential number of deaths recorded that year) and filed the certificates in box files labelled by year and stored in secure records’ rooms. We also collected deaths that occurred in the community in both surveys. In the 2007-08 survey, we collected these through village death registers, and verbal autopsy (VA) forms adapted from the WHO Verbal Autopsy instrument [19]. Village health workers (VHW) and village heads recorded suspected pregnancy-related deaths in the registers. The trained research midwives visited the woman's family and conducted the Verbal Autopsy with the relatives who attended to the deceased (husband, mother, sister or other). In the 2018-19 survey we collected community deaths and other institutional deaths missed in the districts from the maternal and perinatal death surveillance and response (MPDSR) system [20, 21]. Nurses and doctors completed the MPDSR maternal death notification forms in health institutions and in the community. In the community, VHWs notified the local health institution of suspected pregnancy-related deaths. Community-health nurses from the health institutions visited the home, and, when the family cooperated, they investigated the death. When confirmed to be pregnancy-related, they recorded the death on the notification forms. Death notification forms were completed in quadruplicates - one copy was kept at the reporting institution and other copies sent to the Ministry of Health and Childcare’s (MoHCC) district, provincial and head office. At the head office, monitoring and evaluation officers entered the data into the MPDSR database [22]. A data collection instrument adapted from the WHO 2007 systematic review was used to abstract data from all source records in the two surveys [23]. We collected the data for each survey in two rounds. The 2007-08 survey was conducted from May 1, 2007, to June 15, 2008, and repeated from May 1 to July 31, 2020. The 2018-19 survey was done from July 1 to July 31, 2020, and from May 3 to July 20, 2021. Some deduplication of data was done in Stata software [24]. 2.6 Data recollection, verification, and cleaning In 2020, trained research nurse-midwives reviewed the questionnaires and VA forms for data collected in 2007-08. They verified all pregnancy-related deaths and cross-checked their entries in the study database. They recollected the 2007-08 pregnancy-related deaths from CRVS and health records. For some women, death outcomes changed as additional information became available through the recollected data. Two different teams of research midwives collected and recollected the 2018-19 data. They cross-checked deaths collected from the districts with those in the MPDSR database. New deaths and additional information were identified in the recollection and comparison of data from the field with the MPDSR database. 2.7 Study variables The following information was collected for each woman: location (province, district, and classification of the place of residence as rural/urban), age (completed years), pregnancy status (pregnant or not), and causes of death (as stated on medical and death certificates). For pregnancy-related deaths, data were also collected on pregnancy-related and delivery complications (eclampsia, cardiomyopathy, sepsis, embolism, transfusions, heart attack, respiratory distress, shock, and anaesthesia complications), place of death (home or institutional), type of death (maternal or non-maternal), classification of the cause of death as direct or indirect and source of data (health records, CRVS, VA or MPDSR system). 2.8 Qualitative data During the 2018-19 survey, the research midwives collected qualitative data on the availability, accessibility, storage, and security of records in health facilities and the RG’s offices, through observation and interviews with relevant staff (records staff, nurses, and RG’s officers). They used a structured observation and interview guide. The qualitative assessment themes were adapted from WHO and South African guidelines for medical record reviews [25] , [26]. 2.9 Reviewing the causes and classification of deaths. Six obstetricians trained to use the ICD-MM manual reviewed and coded the 2007-08 and 2018-19 deaths. WHO developed the ICD-MM manual as a simplified, user-friendly tool to code the causes of maternal deaths using ICD-10 rules [16, 27]. The obstetrician reviews generated additional variables: verified causes of death coded into ICD-MM groups, reassigned type of death (maternal or non-maternal), and reclassified deaths as direct or indirect. The documented the reasons for changing the cause of death where it changed (deaths identified during the additional data collection, new information available on the death, different level of the reviewer from the one who classified the death in the data source, e.g. death originally classified by nurses or general medical officer and now reviewed by obstetricians). 2.10 Data analysis We calculated the percentages of incompleteness of pregnancy-related and maternal deaths (total initially collected / total re-collected deaths) in the two surveys. Using the Chi-square test, we compared the percentage incompleteness of deaths between 2007-08 and 2018-19 (Table 1). We also calculated the percentage incompleteness of deaths for different causes of incompleteness and compared them using the Chi-square-test (Table 2). We calculated risk ratios (RR) for missed deaths in different data sources, with 95% confidence intervals, using log-linear regression models unadjusted and adjusted for deaths identified in more than one data source (Table 3). We calculated the percentages of misclassified deaths (died or alive) and causes of death for each data source and compared them using Fisher’s exact test (Table 4). We also calculated the sensitivity [(true maternal deaths / (true maternal deaths + misclassified true maternal deaths)] and specificity [(true non-maternal deaths / (true non-maternal deaths + misclassified non-maternal deaths)] of incompleteness of deaths using the six-box method [3, 16]. P<0.05 was assessed as statistically significant. All statistical analyses were performed using Stata version 17 [24]. Findings of the qualitative assessment were synthesised manually using thematic analysis. 3. RESULTS We identified 237 pregnancy-related and 208 maternal deaths in the first round of the 2007-08 survey, and 325 and 296 after the second round of data collection. In the 2018-19 survey, we identified 112 pregnancy-related and 104 maternal deaths in the first round and 137 pregnancy-related and 130 maternal deaths after the second round of data collection. The proportion of missed pregnancy-related deaths (27% vs 18%; p=0.044) and maternal deaths (30% vs 19%; p=0.037) declined from 2007-08 to 2018-19 (Table 1). [Table 1] Missing deaths in the 1 st round of data collection was the main cause of incompleteness of data in both surveys (25% in 2007-08 vs 18% in 2018-19; p=0.136) (Table 2). The data were incomplete because of deaths missed in data collection, deaths misclassified as maternal/non-maternal and incomplete data cleaning in 2007-08. In the 2018-19 survey incompleteness was because of deaths missed in data collection and misclassified deaths. [Table 2] The data source that identified the highest number of pregnancy-related deaths in 2007-08 was VAs (78%) and health records in 2018-19 (60%). The number and proportion (36 [11%] vs 82 [60%]; p<0.001) of deaths identified through health records increased between 2007-08 and 2018-19, while the number and proportion of deaths identified through CRVS declined by 45% (121 [78%] vs 67 [42%]; p=0.020) (Table 3). In 2007-08, VAs were seven times (RR=7.1 [5.1 – 9.7]) and CRVS three times (RR=3.4 [2.4 – 4.7]) more likely to identify a maternal death than health records, while in 2018-19 CRVS (RR=0.8 [0.7 – 0.9]) and the MPDSR (RR=0.7 [0.6 – 0.9]) were less likely to identify a death than health records. After adjusting for deaths identified in more than one source, in 2007-08, CRVS were two times (RR=2.4 [1.4 – 4.1]) and VAs nine times (RR=9.2 [5.7 – 15]) more likely to identify unique deaths than health facility records. In 2018-19, CRVS (RR=0.4 [0.2 – 1.0]) and the MPDSR (RR=1.4 [0.9 – 2.4]) were equally likely to identify unique deaths as health facility records. For deaths identified in two sources, in 2007-08, health facility records and VAs were equally likely (RR=0.5 [0.2 – 1.5]) while VAs and CRVS were six times more likely (RR=6.3 [3.4 – 12]) to identify the same deaths than health facility records and CRVS. In 2018-19, health records and the MPDSR were twice more likely (RR=2.4 [1.3 – 4.4]) and CRVS and the MPDSR less likely (R=0.3 [0.1 – 0.9]) to identify the same death as health facility records and CRVS. [Table 3] In 2007-08, misclassification of maternal and non-maternal deaths was higher in health facility records (17%) than in CRVS (1.7%; p=0.002) and VA records (5.6%; p=0.016) (Table 4). In 2018-19, misclassification was not different between health facility records and CRVS (p=0.691) or the MPDSR (p=0.489). Misclassification of causes of death was not different between health facility records and CRVS or VA/MPDSR in both 2007-08 and 2018-19 survey (p>0.05). However, misclassification of causes of death appeared to increase between 2007-08 and 2018-19 in each data source (p<0.001). [Table 4] In 2007-08, the sensitivity of the study was 95%, and the specificity was 29%, thus the study had a high probability of correctly identifying true maternal deaths with a high chance (71%) of misclassifying non-maternal deaths as maternal. In 2018-19, the sensitivity was 77%, and the specificity was 83% (Table 5). [Table 5] The assessment of documentation and record keeping showed numerous gaps in Zimbabwe’s maternal mortality data systems, which included lost records, absence of record storerooms in health facilities, assigning of non-medical causes of death in CRVS records and collecting few community deaths in the MPDSR (Table 6). [Table 6] 4. DISCUSSION Identifying all maternal deaths and correctly identifying and classifying their causes is critical to accurately measuring maternal mortality. Our study presents essential findings on incompleteness, misclassification, quality of documentation and record keeping for pregnancy-related and maternal deaths in Zimbabwe. All the data sources in the two surveys provided no more than 60% of the deaths, except VAs which identified 78% of total deaths in 2007-08 because they collected community deaths better. This signifies that missingness of deaths remains a challenge for all data sources in Zimbabwe, as in other developing countries [ 5 , 7 , 28 ]. Corroborating this finding, WHO has reported high missingness of deaths in developing countries in its estimates and used adjustment factors of up to 150% to correct the under-reporting [ 6 , 28 ]. Deaths occurring in the community, and in private and rural health institutions which lack supervision are prone to underreporting [ 29 , 30 ]. Even in studies where community health workers collected the data, 30–90% underreporting has been observed [ 31 ]. Failure to know that a woman was pregnant before her death and reluctance to report the deaths causes the incompleteness of pregnancy-related community deaths [ 5 ]. Selective reporting by health authorities contributes to incompleteness of institutional maternal deaths [ 32 ]. In an MPDSR evaluation conducted in Ethiopia, study participants described maternal deaths as “political,” reporting that authorities suppressed maternal death reports to evade public rebuke because the avoidable nature of maternal deaths provokes public anger. They said authorities pressured health workers to underreport maternal deaths and paint a picture of success in policies intended to reduce maternal deaths [ 32 ]. Above all, most data sources are unable to identify community deaths, as communities avoid reporting home deaths [ 22 , 30 ]. We found an increase in the number and proportion of deaths identified through health records between 2007-08 and 2018-19 possibly due to increase in institutional deliveries. The 2015-16 Zimbabwe Demographic and Health Survey (ZDHS) reported an increase in institutional deliveries from 65% (57% rural and 85% urban) in 2010/11 to 72% (68% rural and 81% urban) in 2015/16 [ 33 ], and the MICS reported 86% (82% rural and 94% urban) in 2019 [ 34 ]. Improved documentation of deaths through MPDSR could also be responsible [ 16 ]. Increase in institutional deliveries could also improve the completeness of pregnancy-related and maternal deaths from 2007-08 to 2018-19. As institutional deliveries increased, more deaths would occur in institutions where they are better documented and reported than community deaths [ 35 ]. Causes of institutional deaths would be documented better and more accurately since medical personnel attend to the deaths. Furthermore, by 2018-19, doctors and nurses had been trained to document and audit maternal deaths through the MPDSR [ 20 , 36 ]. Analysis of the proportion of deaths identified by two or more data sources evaluated the transmission of information across the data systems. High proportions of deaths recorded by two or more data sources were expected, because, by policy, health institutions are required to transfer death records to the RG’s offices. Major hospitals also have RG’s offices within them which collects hospital death reports. In assessing the extent to which the data sources identified the same deaths, low proportions of deaths were identified by any two data sources. In 2018-19, MPDSR and health records identified only 3% of the deaths. Health and CRVS records identified less than 4% of deaths in 2007-08 and less than 10% in 2018-19. CRVS should be recording more deaths because the RG’s offices are conducting mobile registration in the communities countrywide [ 18 ]. Higher misclassification of maternal and non-maternal deaths was observed in 2007-08. During this period, the MPDSR did not exist [ 20 ], and health workers were possibly unaware of the definition of maternal death. Even in 2015, an MPDSR evaluation observed that 50% of nurses who used the system in Zimbabwe failed to correctly define a maternal death [ 30 ]. High sensitivity of the data sources to correctly identify maternal deaths was observed in 2007-08 and 2018-19, while specificity was low in 2007-08 where data collectors classified late maternal deaths (deaths occurring 42 days after the termination of pregnancy) as maternal deaths. In 2018-19, data collectors strictly applied the definition of maternal death in the data collection, making the sensitivity and specificity high. The sensitivity declined between the 2007-08 and 2018-19 surveys due to misclassification of the causes of death, signifying that the capacity to assign accurate causes of death is still low [ 6 , 12 , 27 , 37 ]. Misclassification of the causes of death was higher in 2018-19 than in 2007-08 because the 2018-19 patient notes were available for obstetricians to use to review and reclassify the deaths. The storage of health records was a challenge. CRVS records were better stored than health facility and MPDSR records. Some health facilities lacked storage rooms. Others did not enforce record-keeping standards, and others lacked human resources for record-filling. The completeness of data in the records was a challenge. Supporting this finding, the 2015 MPDRS evaluation reported an average completeness of notification forms of 76%, and only 53% forms had accompanying documents [ 30 ]. The completeness of records and medical histories helps assessors to assign the correct causes of death [ 4 ]. Strengths and limitations of the study The study recollected, triangulated, and linked the data across sources. Trained obstetricians reviewed and reclassified the deaths using the ICD-MM manual. This enabled the analysis of incompleteness and misclassification of the deaths. However, the study included fewer community deaths in the 2018-19 survey [ 15 ], and there were no patient notes for 2007-08 deaths, for the obstetricians to use in the review. Despite these limitations, the three data sources (health records, CRVS records and MPDSR records/VAs) and the recollection process generated sufficient data for estimating incompleteness and misclassification of deaths. The findings support recommendations to strengthen data systems for maternal mortality in low-resource countries. Conclusions Incompleteness and misclassification of maternal deaths are still significant problems in Zimbabwe. Health facility records, CRVS and the MPDSR still inaccurately assign causes to pregnancy-related and maternal deaths. The transfer of death records from health institutions to RG’s offices and MoHCC’s head office remains a challenge in the data systems. Maternal mortality studies using health, CRVS and MPDSR records should triangulate data sources to increase the completeness of data. Record keeping and transmission should be strengthened in these data systems, supported by strengthened policy frameworks and resource allocations to collect community deaths. Medical staff require ongoing capacity strengthening to assign accurate causes of death. Collecting community deaths and assigning correct causes should be an important component of future maternal mortality studies. Abbreviations BMI - Bayesian Maternal Mortality Misclassification, CRVS – Civil Registration and Vital Statistics, ICD – International Classification of Diseases, ICD-MM – International Classification of Diseases for Maternal Mortality, MMR – Maternal Mortality Ratio, MPDSR – Maternal and Perinatal Death Surveillance and Response, MoHCC – Ministry of Health and Child Care, RG – Registrar General, RR – Risk Ratio, VA – Verbal Autopsy, VHW – Village Health Worker, RAMOS – Reproductive Age Mortality Study, SDG – Sustainable Development Goal, UI – Uncertainty Interval, Declarations Ethics approval and consent to participate. The University of Pretoria and the Medical Research Council of Zimbabwe institutional review boards (IRBs) approved the study. The MoHCC and RG’s department granted permissions for the study. All institutions approved the collection of data with personally identifying information (PII), including name, date of birth, place of birth and date of death to link and de-duplicate individual women across the data sources. The IRBs approved the collection of clinical notes to be used by obstetricians to code the causes of death. In the 2007-08 survey, verbal autopsy informants gave written informed consent to be interviewed. Informed consent was waived in the 2018-19 survey where all data were collected from secondary records. Consent for publication Not applicable Availability of data and materials Data will be made available on a reasonable request. Competing interests The authors have no competing interests to declare. Funding The UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), and WHO-Geneva funded in part the recollection and validation of the 2007-08 and 2018-19 data, through a sub-grant from the Improving Maternal Health Measurement (IMHM) Project at the Women & Health Initiative of the Harvard T.H. Chan School of Public Health funded by the Bill & Melinda Gates Foundation [Grant Number OPP1169546]. The production of the manuscript was not funded. Authors’ contributions RM 1 conceptualized, conducted the analysis, and drafted the paper. RM 2 guided the analysis. LN, GM, CG, SN, RM 2 and SPM reviewed several versions of the paper. All authors approved the final manuscript. Acknowledgments The authors acknowledge the contributions of all 2007-2008 and 2018-2019 Zimbabwe Maternal and Perinatal Mortality Study (ZMPMS) group members (see online supplementary file), who contributed to the protocol development, data collection, data processing, and those who reviewed and classified the deaths. The support received from various MoHCC and RG’s department staff is acknowledged and appreciated. References WHO. Sustainable Development Goal 3: Health. https://www.who.int/topics/sustainable-development-goals/targets/en/. Accessed 14 Nov 2020. WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Trends in maternal mortality 2000 to 2020: Estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva: World Health Organization; 2023. Licence: CC BY-NC-SA 3.0 IGO. WHO. Maternal mortality: Guidance to improve national reporting. Geneva: World Health Organization; 2022. WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Trends in maternal mortality 2000 to 2017: Estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Geneva: World Health Organization; 2019. Licence: CC BY-NC-SA 3.0 IGO. Graham WJ, Ahmed S, Stanton C, Abou-Zahr C, Campbell OM: Measuring maternal mortality: an overview of opportunities and options for developing countries. BMC Med 2008;6: 12. Peterson E, Chou D, Moller AB, Gemmill A, Say L, Alkema L: Estimating misclassification errors in the reporting of maternal mortality in national civil registration vital statistics systems: A Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity for multiple countries and years with missing data. Stat Med 2022. Adair T, Lopez AD: How reliable are self-reported estimates of birth registration completeness? Comparison with vital statistics systems. PLoS One 2021;16(6): e0252140. WHO. WHO guidance for measuring maternal mortality from a census. Geneva: World Health Organization. 2013. DHS: Guide to DHS Statistics: Maternal Mortality and Pregnancy-related Mortality. Maryland: Demographic and Health Survey. 2021. Danel I, Graham W, Stupp P, Castillo P: "Applying the Sisterhood Method for Estimating Maternal Mortality to a Health Facility-based Sample: A Comparison with Results from a Household- based Sample.". Stud Fam Plann 1997;28(3): 266. The World Bank. Civil registration and vital statistics: key to better data on maternal mortality. Geneva: https://blogs.worldbank.org/health/civil-registration-and-vital-statistics-key-better-data-maternal-mortality. Accessed 17 Jan 2024. Kao S, Chen LM, Shi L, Weinrich MC: Underreporting and misclassification of maternal mortality in Taiwan. Acta Obstet Gynecol Scand 1997;76(7): 629-636. Musarandega R, Machekano R, Pattinson R, Munjanja SP, Zimbabwe M, Perinatal Mortality Study g: Protocol for analysing the epidemiology of maternal mortality in Zimbabwe: A civil registration and vital statistics trend study. PLoS One 2021;16(6): e0252106. Musarandega R, Ngwenya S, Murewanhema G, Machekano R, Magwali T, Nyström L, et al.: Changes in causes of pregnancy-related and maternal mortality in Zimbabwe 2007-08 to 2018-19: findings from two reproductive age mortality surveys. BMC Public Health 2021. Musarandega R, Cresswell J, Magwali T, Makosa D, Machekano R, Ngwenya S, et al.: Maternal mortality decline in Zimbabwe, 2007/2008 to 2018/2019: findings from mortality surveys using civil registration, vital statistics and health system data. BMJ Glob Health 2022;7(8): e009465. WHO. The WHO Application of ICD-10 to deaths during pregnancy, childbirth and the puerperium: ICD-MM. World Health Organization. Geneva. 2011. All Africa. Zimbabwe: Countrywide mobile registration blitz begins. The Herald, 2021. https://allafrica.com/stories/202109140535.html. Accessed 16 May 2022. Kubatana. BVR, IDs' mobile registration: Citizens speak. Kubatana.net; 2017. https://kubatana.net/2017/09/19/bvr-ids-mobile-registration-citizens-speak/. Accessed 16 May 2022. WHO. The 2007 WHO Verbal Autopsy Instrument. https://www.who.int/standards/classifications/other-classifications/verbal-autopsy-standards-ascertaining-and-attributing-causes-of-death-tool. Accessed 9 Dec 2021. MOHCC: Maternal and Perinatal Death Surveillance and Response Report. Ministry of Health and Child Care Zimbabwe, Family and Child Health Department; 2018. MOHCC. Guidelines for maternal and perinatal death audits in Zimbabwe. Harare: Zimbabwe Ministry of Health and Child Care; 2013. Maphosa M, Juru TP, Masuka N, Mungati M, Gombe N, Nsubuga P, et al.: Evaluation of the Maternal Death Surveillance and response system in Hwange District, Zimbabwe, 2017. BMC Pregnancy Childbirth 2019;19(1): 103. Khan KS, Wojdyla D, Say L, Gülmezoglu AM, Van Look PFA: WHO analysis of causes of maternal death: a systematic review. The Lancet 2006;367(9516): 1066-1074. STATA. Statistical software for data science. Texas: STATA Coporation LLC; 2022. https://www.stata.com/. Accessed 21 Jan 2022. WHO. Medical Records Manual: A guide for Developing Countries. World Health Organization. 2006. HPCSA. Guideline for the good practice in the health care profession: Guidelines on the keeping of patient records. Booklet 9. Pretoria: Health Professions Council of South Africa; 2016. Say L, Chou D: Better understanding of maternal deaths—the new WHO cause classification system. BJOG 2011;118(s2): 15-17. Wilmoth JR, Mizoguchi N, Oestergaard MZ, Say L, Mathers CD, Zureick-Brown S, et al.: A New Method for Deriving Global Estimates of Maternal Mortality. Stat Politics Policy 2012;3(2). Rutgers S: Two years maternal mortality in Matebeleland north Province, Zimbabwe. Cent Afr J Med 2001;47(2): 39-43. Faith Mutsigiri-Murewanhema PTM, Tsitsi Juru, Notion Tafara Gombe, Donewell Bangure, More Mungati, Mufuta Tshimanga: Evaluation of the maternal mortality surveillance system in Mutare district, Zimbabwe, 2014-2015: a cross sectional study. Pan Afr Med J 2017;27(204). Silva R, Amouzou A, Munos M, Marsh A, Hazel E, Victora C, et al.: Can Community Health Workers Report Accurately on Births and Deaths? Results of Field Assessments in Ethiopia, Malawi and Mali. PLoS One 2016;11(1): e0144662. Melberg A, Mirkuzie AH, Sisay TA, Sisay MM, Moland KM: 'Maternal deaths should simply be 0': politicization of maternal death reporting and review processes in Ethiopia. Health Policy Plan 2019;34(7): 492-498. ZimStat and ICF International. Zimbabwe Demographic and Health Survey 2015: final report. Zimbabwe national statistics agency (ZIMSTAT) and ICF International Rockville, Maryland, USA; 2016. ZimStat and UNICEF. Multiple indicator cluster survey 2017: final report. https://www.unicef.org/zimbabwe/reports/zimbabwe-2019-mics-survey-findings-report; 2019. Accessed 20 Sep 2020. Ekirapa-Kiracho E, Waiswa P, Rahman MH, Makumbi F, Kiwanuka N, Okui O, et al.: Increasing access to institutional deliveries using demand and supply side incentives: early results from a quasi-experimental study. BMC Int Health Hum Rights 2011;11(1): S11. MCHIP. Assessment of maternal and perinatal death surveillance and response implementation in Zimbabwe. Technical report. Harare: Maternal and Child Health Integrated Program; 2017. Ameh C, Adegoke A, Pattinson R, van den Broek N: Using the new ICD-MM classification system for attribution of cause of maternal death—a pilot study. BJOG 2014;121(s4): 32-40. Tables Table 1: Missingness of pregnancy-related and maternal deaths in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; The P-value for the Chi-square -test of difference in the proportion of missed deaths between the two surveys. Number of deaths 2007-08 2018-19 P-value Before verifying After verifying Missingness Before verifying After verifying Missingness Pregnancy-related 237 325 27% 112 137 18% 0.044 Maternal 208 296 30% 104 130 19% 0.037 Table 2: Number and proportion of missed pregnancy-related deaths by the cause of missingness in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; the P-value for the Chi-square test of difference in proportion of missed deaths by the cause of missingness between the two surveys. Causes of missingness Number (%) of deaths missed by cause of missingness. P-value 2007-08 (N=325) 2018-19 (N=137) Deaths missed in the 1 st round of data collection 80 (25%) 25 (18%) 0.136 Deaths incorrectly classified as maternal/non-maternal deaths 8 (3.4%) 2 (1.8%) 0.493 Deaths not entered in the database 18 (5.5%) 0 (0%) 0.005 Twin deliveries duplicated in the database 4 (1.2%) 0 (0%) 0.192 Deaths duplicated in the 1 st database 11 (4.7%) 0 (0%) 0.029 Table 3: Number and percentage of deaths identified and log-linear regression risk ratio (RR), with 95% confidence intervals, of completeness of pregnancy-related deaths in different data sources in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; the P-value for Fisher’s exact test of difference in the percentage of deaths identified by each data source in the two surveys. 2007-08 (n=325) 2018-19 (n=137) P-value, comparing percent of deaths identified by a data source in the surveys Data source where deaths were identified N (%) RR (95% CI) Data source where deaths were identified N (%) RR (95% CI) Health facility records 36 (11) Reference Health facility records 82 (60) Reference <0.001 CRVS records 121 (37) 3.4 ()2.4 – 4.7) CRVS records 67 (49) 0.8 (0.7 – 0.9) 0.020 VAs records 254 (78) 7.1 (5.1 – 9.7) MPDSR database 58 (42) 0.7 (0.6 – 0.9) <0.001 Health facility records only 19 (5.9) Reference Health facility records only 27 (20) Reference <0.001 CRVS records only 46 (14) 2.4 (1.4 – 4.1) CRVS records only 12 (8.8) 0.4 (0.2 – 1.0) <0.001 VAs only 174 (53) 9.2 (5.7 – 15) MPDSR database only 39 (28) 1.4 (0.9 – 2.4) <0.001 Health facility and CRVS records 11 (3.4) Reference Health facility and CRVS records 13 (9.5) Reference 0.007 Health facility and VA records 6 (1.9) 0.5 (0.2 – 1.5) Health records and MPDSR 31 (23) 2.4 (1.3 – 4.4) <0.001 CRVS and VA records 69 (21) 6.3 (3.4 – 12) CRVS and MPDSR 4 (2.9) 0.3 (0.1 – 0.9) <0.001 N/A – Not applicable in the first three rows because the data sources in these rows count all deaths identified by each source including those that were identified in more than one source. Also, not applicable in the second three rows because the deaths in each row were not exclusive to those data sources. Table 4: Number and percentage of misclassified deaths by the type of misclassification and data source in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; Fisher’s exact p-value comparing misclassification between surveys. Type of misclassification Data source Survey year, number, and percentage of misclassified deaths P-value comparing 2007-08 vs 2018-19 2007-08 2018-19 Number and percentage P-value comparing misclassification in data sources 3 Number and percentage P-value comparing misclassification in data sources 3 Misclassification of survival outcome (died/alive) Health facility records 2/41 (4.9%) Reference N/A 4 Reference N/A CRVS (RG’s) records 2/121 (1.7%) 0.266 N/A 4 N/A N/A VA or MPDSR 2 5/252 (2.0%) 0.255 N/A 4 N/A N/A Misclassification of maternal and non-maternal deaths Health facility records 7/41 (17%) Reference 7/82 (8.5%) Reference 0.160 CRVS (RG’s) records 2/121 (1.7%) 0.002 4/67 (6.0%) 0.755 0.189 VA or MPDSR 2 14/252 (5.6%) 0.016 9/58 (16%) 0.281 0.009 Misclassification of type of maternal death (Direct/Indirect) Health facility records 2/41 (4.9%) Reference 4/82 (4.9%) Reference 1.000 CRVS (RG’s) records 8/121 (6.6%) 0.690 2/67 (3.0%) 0.691 0.499 VA or MPDSR 2 12/252 (4.8%) 0.974 5/58 (8.6%) 0.489 0.332 Misclassification of causes of death 1 Health facility records 7/41 (17%) Reference 43/82 (52%) Reference <0.001 CRVS (RG’s) records 17/121 (14%) 0.619 29/67 (43%) 0.323 <0.001 VA or MPDSR 2 37/252 (15%) 0.643 35/58 (60%) 0.391 <0.001 1 Reviewers assigned a different cause of death after identifying a different underlying cause. 2 The data sources were Verbal Autopsy (VA) 2007-08 and MPDSR for 2018-19 3 For each type of misclassification, the Fisher’s exact or Chi-square p-value compares the percentage of the reference category (health records) and each of the other data sources (CRVS and VA or MPDSR). 4 There was no misclassification of survival outcomes (died/alive) because the study collected data for deceased women only Table 6: Qualitative findings on documentation and record keeping in the Zimbabwe Maternal and Perinatal Mortality Surveys; results from the 2020 data collection process. Data Source Key observations Health records In large hospitals with dedicated records rooms and staff recent patient charts/notes/registers were available and accessible. Health facilities without record rooms had poorly archived patient records; records were dumped on the floor in some health facilities. Some health facilities with records rooms did not follow standard record filing and removal practices. In some health facilities, the individual staff kept maternal death records, and when these staffs were off duty, the records couldn’t be accessed. Nurses in one district hospital without a medical officer reported that they signed death certificates but were not allowed to assign medical causes of death on the certificates. Hence, they omitted the causes of death or assigned non-medical causes such as “short illness” or “long illness.” CRVS records All RG’s offices visited had secure records rooms managed by dedicated records staff. Death records were sequentially filed in box files by registration date, and the boxes were labelled and sequentially filed by year. For some deaths that occurred in health institutions, medical certificates were attached to the death records, but this practice was not consistent across districts. Deaths reported by family members and deaths registered during mobile registration campaigns had vague causes of death, such as “natural causes”, “natural death”, “swollen leg”, “running stomach”, “headache”, etc. MPDSR database Death notification forms completed by health facilities were available at the MoHCC’s provincial offices and head office. In provincial offices, the forms were filed in box files but kept in individuals’ offices. When these individuals were not present, the records could not be accessed. Forms received at the head office were captured into an electronic database. After electronic capture, the forms were not properly filed because of a shortage of storage space and staff. Though the system is meant to capture community deaths, staff in all districts reported that (i) they faced challenges when investigating community deaths, especially among members of the apostolic religious sects, (ii) most women in these sects deliver at home, (ii) many community deaths occur in these communities, (iv) their faith healers, religious leaders and family members do not cooperate with health workers who come to enquire about community deaths. Additional Declarations No competing interests reported. Supplementary Files Supplementaryfile1.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 20 Feb, 2024 Editor assigned by journal 20 Feb, 2024 Submission checks completed at journal 19 Feb, 2024 First submitted to journal 23 Jan, 2024 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-3891799","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":273605238,"identity":"0c08df5a-0d2d-4b8b-8641-463092e235c1","order_by":0,"name":"Reuben Musarandega","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtklEQVRIiWNgGAWjYBADORBx4AEpWozBWhJI0ZLYACKJ0mI+7Yzxh5977NLnhx1+CLTFTk63gYAWmds5ZpI9z5JzN95OMwBqSTY2O0BAi4R0jhkDzwHm3I2zE0BaDiRuI0KL8cc/B+rTDWenfyBai4E0z4HDCfJABrFa0sqkZQ4cN9wgnVNwIMGAKL8kb/745kC1vPzs9M0fPlTYyRHUAgcGYJUGxCoHAfkGUlSPglEwCkbBiAIA0tpDdJqLJmgAAAAASUVORK5CYII=","orcid":"","institution":"University of Pretoria","correspondingAuthor":true,"prefix":"","firstName":"Reuben","middleName":"","lastName":"Musarandega","suffix":""},{"id":273605239,"identity":"29355380-3a66-4bda-8f4d-318522c020e9","order_by":1,"name":"Lennarth Nystrom","email":"","orcid":"","institution":"Umea University","correspondingAuthor":false,"prefix":"","firstName":"Lennarth","middleName":"","lastName":"Nystrom","suffix":""},{"id":273605240,"identity":"2b8f3f43-a4e4-4c8e-93c1-18d3b9ad760e","order_by":2,"name":"Grant Murewanhema","email":"","orcid":"","institution":"University of Zimbabwe","correspondingAuthor":false,"prefix":"","firstName":"Grant","middleName":"","lastName":"Murewanhema","suffix":""},{"id":273605241,"identity":"b35afaf3-5ed6-459a-ac08-7ebcc1dca253","order_by":3,"name":"Chipo Gwanzura","email":"","orcid":"","institution":"University of Zimbabwe","correspondingAuthor":false,"prefix":"","firstName":"Chipo","middleName":"","lastName":"Gwanzura","suffix":""},{"id":273605242,"identity":"3fbaefad-8926-4b2c-b51b-97d4073ffbd4","order_by":4,"name":"Solwayo Ngwenya","email":"","orcid":"","institution":"National University of Science and Technology; and Mpilo Central Hospital","correspondingAuthor":false,"prefix":"","firstName":"Solwayo","middleName":"","lastName":"Ngwenya","suffix":""},{"id":273605243,"identity":"36f53532-0ac8-485f-b330-804db563240b","order_by":5,"name":"Robert Pattinson","email":"","orcid":"","institution":"University of Pretoria","correspondingAuthor":false,"prefix":"","firstName":"Robert","middleName":"","lastName":"Pattinson","suffix":""},{"id":273605244,"identity":"639c461d-0ecd-40c8-9955-c8bd32389b9e","order_by":6,"name":"Rhoderick Machekano","email":"","orcid":"","institution":"Stellenbosch University","correspondingAuthor":false,"prefix":"","firstName":"Rhoderick","middleName":"","lastName":"Machekano","suffix":""},{"id":273605245,"identity":"cafda09d-b3d8-4b22-8d85-529f00075840","order_by":7,"name":"Stephen Peter Munjanja","email":"","orcid":"","institution":"University of Zimbabwe","correspondingAuthor":false,"prefix":"","firstName":"Stephen","middleName":"Peter","lastName":"Munjanja","suffix":""}],"badges":[],"createdAt":"2024-01-23 18:17:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3891799/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3891799/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":51396712,"identity":"1be0f4f6-3bd6-40cf-854b-2cf2817ca368","added_by":"auto","created_at":"2024-02-20 20:21:22","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":304312,"visible":true,"origin":"","legend":"\u003cp\u003eMap of study districts for Zimbabwe maternal and perinatal mortality study 2007-08 and 2018-19.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-3891799/v1/55ae847991bc517d99bed89d.jpeg"},{"id":51396752,"identity":"6919ff7c-c3fe-4f5d-aab1-4067f2c864e2","added_by":"auto","created_at":"2024-02-20 20:22:03","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":931001,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3891799/v1/398a56eb-5cba-40cd-a0ca-de2c8e6dd9aa.pdf"},{"id":51396709,"identity":"1e4b9896-9680-45ea-a5b9-df5eff9d7bdc","added_by":"auto","created_at":"2024-02-20 20:21:17","extension":"docx","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":20949,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-3891799/v1/92a3c991e13462ee3d33048b.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Incompleteness and misclassification of maternal deaths in Zimbabwe: data from two reproductive age mortality surveys, 2007-08 and 2018-19","fulltext":[{"header":"1. INTRODUCTION","content":"\u003cp\u003eReducing maternal mortality is a high-priority global health goal [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], particularly in sub-Saharan Africa, where the average maternal mortality ratio (MMR) was estimated at 536 maternal deaths per 100 000 live births in 2020 compared to the global average MMR of 223 maternal deaths per 100 000 live births [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Zimbabwe (357, uncertainty interval (UI): 255 to 456) was among the East and Southern Africa countries with an MMR above the global average MMR (223; UI: 202 to 255) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe incompleteness of data also referred to as \u0026ldquo;missingness\u0026rdquo; or under-reporting [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e], is one of the problems impeding progress on the goal to reduce maternal mortality [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], It affects all sources of data \u0026ndash; surveys, surveillance and civil registration and vital statistics (CRVS) data [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Incompleteness of maternal deaths occurs when deaths are not recorded and reported in the data [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Deaths that occur outside health institutions are affected the most as these are often not documented. Deaths occurring in institutions are documented but sometimes not reported in surveillance systems [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. The incompleteness of maternal deaths causes under-estimation of MMRs [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Researchers use a wide range of statistical techniques to adjust MMRs for missingness such as the Brass Growth Balance Method for censuses and household surveys [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], and the Bayesian Maternal Mortality Misclassification (BMI) model for global estimates [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. However, these techniques may end up over-estimating the MMRs [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. As a result, countries may dispute the MMR estimates generated by global models [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMisclassification occurs when a maternal death is classified as non-maternal or a non-maternal death classified as maternal [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Assigning incorrect causes leads to misclassification of deaths, which leads to under or over-estimating the MMR [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Deaths can also be misclassified by assigning them to the wrong cause-group when coders are not competent enough to correctly code the deaths [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Inaccurate MMR estimates will lead to inappropriate planning and misallocating resources to maternal mortality interventions. Ultimately, national, and global goals are hindered. Achievement of Sustainable Development Goal (SDG) 3.1 for maternal mortality is threatened by this challenge.\u003c/p\u003e \u003cp\u003eWe performed two reproductive age mortality surveys (RAMOS) in 2007-08 and 2018-19 to assess changes in the MMR and causes of maternal deaths in Zimbabwe following a raft of interventions implemented to reduce maternal mortality. A RAMOS enumerates the deaths of all women of reproductive age and identifies maternal deaths out of these, to calculate the maternal mortality ratio. In this paper, we estimated the level of incompleteness and misclassification of pregnancy-related and maternal deaths in the two surveys. We recommend measures to improve the completeness and classification of maternal deaths in Zimbabwe and similar countries.\u003c/p\u003e"},{"header":"2. MATERIAL AND METHODS","content":"\u003cp\u003e\u003cstrong\u003e2.1 Study design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe conducted the two cross-sectional RAMOS in 11 districts of Zimbabwe, applying multi-stage cluster sampling where the population was stratified into provinces (n=10). One district was selected from each province using simple random sampling. An additional district was chosen in Harare \u0026ndash; the province with the largest population. Papers published previously described the study methods in detail\u0026nbsp;[13-15]. \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Study setting\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThree of the 11 study\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003edistricts (Figure 1) were partially urban (Bindura, Kwekwe and Mutare); three were urban (Harare Southeastern, Harare Western and Nkulumane) and five were rural (Chivi, Matobo, Mutoko, Tsholotsho and Zvimba).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Figure 1]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Data collection process and data sources\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe\u0026nbsp;used trained research nurse midwives to collect data in the two surveys, guided by standard operating procedures from the study protocol. They collected live births from maternity registers in hospitals and health centres. They recorded reproductive age deaths from patient registers and charts at maternity units, operating theatres, high dependency and intensive care units, gynaecological, medical, and surgical wards, mortuaries, police posts, and casualty departments in hospitals and maternity records at health centres. Medical staff completed the medical charts while attending to patients in the hospitals and health centres.\u003c/p\u003e\n\u003cp\u003eWe collected additional deaths from\u0026nbsp;the government\u0026rsquo;s Registrar General\u0026rsquo;s (RG\u0026rsquo;s) offices. The RG\u0026rsquo;s office registered deaths on death notification forms in civil registration and vital statistics (CRVS) records through reports submitted by health facilities, the police, or members of the public. Police submitted reports of deaths that they attended to at home and took to a hospital for a post-mortem. At the health facilities, the medical officer who attended to the death that occurred at the health institution or brought by the police for post-mortem documented the underlying and antecedent or contributory causes of the death on the death record according to International Classification of Disease Version 10 (ICD-10)\u0026nbsp;[16].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe public reported home deaths that the police could not attend to, where village heads or headmen wrote a death confirmation letter. The relatives registered the death at the RG\u0026rsquo;s offices through the community head\u0026rsquo;s letter and their verbal report. The RG\u0026rsquo;s offices also conducted community registration outreaches periodically where they registered births and deaths posthumously in the community\u0026nbsp;[17, 18].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe RG\u0026rsquo;s office created a death record for all deaths reported by the three means, on which they documented the date, place and causes of death available on the source of information (medical certificate, police report or family members\u0026rsquo; report). They issued a death certificate with a reference number (year of issue/sequential number of deaths recorded that year) and filed the certificates in box files labelled by year and stored in secure records\u0026rsquo; rooms. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe also collected deaths that occurred\u0026nbsp;in the\u0026nbsp;community in both surveys. In the 2007-08 survey, we collected these through village death registers, and\u0026nbsp;verbal autopsy (VA) forms adapted from the WHO Verbal Autopsy instrument\u0026nbsp;[19].\u0026nbsp;Village health workers (VHW) and village heads recorded suspected pregnancy-related deaths in the registers. The trained research midwives visited the woman\u0026apos;s family and conducted the Verbal Autopsy with the relatives who attended to the deceased (husband, mother, sister or other).\u003c/p\u003e\n\u003cp\u003eIn the 2018-19 survey we collected community deaths and other institutional deaths missed in the districts from the maternal and perinatal death surveillance and response (MPDSR) system\u0026nbsp;[20, 21].\u0026nbsp;Nurses and doctors completed the MPDSR maternal death notification forms in health institutions and in the community. In the community, VHWs notified the local health institution of suspected pregnancy-related deaths. Community-health nurses from the health institutions visited the home, and, when the family cooperated, they investigated the death. When confirmed to be pregnancy-related, they recorded the death on the notification forms. Death notification forms were completed in quadruplicates - one copy was kept at the reporting institution and other copies sent to the Ministry of Health and Childcare\u0026rsquo;s (MoHCC) district, provincial and head office. At the head office, monitoring and evaluation officers entered the data into the MPDSR database\u0026nbsp;[22]. \u0026nbsp;A data collection instrument adapted from the WHO 2007 systematic review was used to abstract data from all source records in the two surveys\u0026nbsp;[23].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe collected the data for each survey in two rounds. The 2007-08 survey was conducted from May 1, 2007, to June 15, 2008, and repeated from May 1 to July 31, 2020. The 2018-19 survey was done from July 1 to July 31, 2020, and from May 3 to July 20, 2021.\u0026nbsp;Some deduplication of data was done in Stata software\u0026nbsp;[24].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.6 Data recollection, verification, and cleaning\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn 2020, trained research nurse-midwives reviewed the questionnaires and VA forms for data collected in 2007-08. They verified all pregnancy-related deaths and cross-checked their entries in the study database. They recollected the 2007-08 pregnancy-related deaths from CRVS and health records. For some women, death outcomes changed as additional information became available through the recollected data. Two different teams of research midwives collected and recollected the 2018-19 data. They cross-checked deaths collected from the districts with those in the MPDSR database. New deaths and additional information were identified in the recollection and comparison of data from the field with the MPDSR database.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.7 Study variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe following information was collected for each woman: location (province, district, and classification of the place of residence as rural/urban), age (completed years), pregnancy status (pregnant or not), and causes of death (as stated on medical and death certificates). For pregnancy-related deaths, data were also collected on pregnancy-related and delivery complications (eclampsia, cardiomyopathy, sepsis, embolism, transfusions, heart attack, respiratory distress, shock, and anaesthesia complications), place of death (home or institutional), type of death (maternal or non-maternal), classification of the cause of death as direct or indirect and source of data (health records, CRVS, VA or MPDSR system).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.8 Qualitative data\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDuring the 2018-19 survey, the research midwives collected qualitative data on the availability, accessibility, storage, and security of records in health facilities and the RG\u0026rsquo;s offices, through observation and interviews with relevant staff (records staff, nurses, and RG\u0026rsquo;s officers). They used a structured observation and interview guide. The\u0026nbsp;qualitative assessment\u0026nbsp;themes were adapted from WHO and South African\u0026nbsp;guidelines for medical record reviews\u0026nbsp;[25]\u003csup\u003e,\u003c/sup\u003e[26].\u0026nbsp;\u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.9 Reviewing the causes and classification of deaths.\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSix obstetricians trained to use the ICD-MM manual reviewed and coded the 2007-08 and 2018-19 deaths. WHO developed the ICD-MM manual as a simplified, user-friendly tool to code the causes of maternal deaths using ICD-10 rules\u0026nbsp;[16, 27]. The obstetrician reviews generated additional variables: verified causes of death coded into ICD-MM groups, reassigned type of death (maternal or non-maternal), and reclassified deaths as direct or indirect. The documented the reasons for changing the cause of death where it changed (deaths identified during the additional data collection, new information available on the death, different level of the reviewer from the one who classified the death in the data source, e.g. death originally classified by nurses or general medical officer and now reviewed by obstetricians).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.10 Data analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe calculated the percentages of incompleteness of pregnancy-related and maternal deaths (total initially collected / total re-collected deaths) in the two surveys. Using the Chi-square test, we compared the percentage incompleteness of deaths between 2007-08 and 2018-19\u0026nbsp;(Table 1).\u0026nbsp;We also calculated the percentage incompleteness of deaths for different causes of incompleteness and compared them using the Chi-square-test (Table 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe calculated risk ratios (RR) for missed deaths in different data sources, with 95% confidence intervals, using log-linear regression models unadjusted and adjusted for deaths identified in more than one data source (Table 3).\u0026nbsp;We calculated the percentages of misclassified deaths (died or alive) and causes of death for each data source\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eand compared them using Fisher\u0026rsquo;s exact test (Table 4). We also calculated the sensitivity [(true maternal deaths / (true maternal deaths + misclassified true maternal deaths)] and specificity [(true non-maternal deaths / (true non-maternal deaths + misclassified non-maternal deaths)] of incompleteness of deaths using the six-box method [3, 16]. P\u0026lt;0.05 was assessed as statistically significant. All statistical analyses were performed using Stata version 17 [24]. Findings of the qualitative assessment were synthesised manually using thematic analysis. \u0026nbsp;\u003c/p\u003e"},{"header":"3. RESULTS","content":"\u003cp\u003eWe identified 237 pregnancy-related and 208 maternal deaths in the first round of the 2007-08 survey, and 325 and 296 after the second round of data collection. In the 2018-19 survey, we identified 112 pregnancy-related and 104 maternal deaths in the first round and 137 pregnancy-related and 130 maternal deaths after the second round of data collection. The proportion of missed pregnancy-related deaths (27% vs 18%; p=0.044) and maternal deaths (30% vs 19%; p=0.037) declined from 2007-08 to 2018-19 (Table 1).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 1]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMissing deaths in the 1\u003csup\u003est\u003c/sup\u003e round of data collection was the main cause of incompleteness of data in both surveys (25% in 2007-08 vs 18% in 2018-19; p=0.136) (Table 2). The data were incomplete because of deaths missed in data collection, deaths misclassified as maternal/non-maternal and incomplete data cleaning in 2007-08. In the 2018-19 survey incompleteness was because of deaths missed in data collection and misclassified deaths.\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 2]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data source that identified the highest number of pregnancy-related deaths in 2007-08 was VAs (78%) and health records in 2018-19 (60%). The number and proportion (36 [11%] vs 82 [60%]; p\u0026lt;0.001) of deaths identified through health records increased between 2007-08 and 2018-19, while the number and proportion of deaths identified through CRVS declined by 45% (121 [78%] vs 67 [42%]; p=0.020) (Table 3). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn 2007-08, VAs were seven times (RR=7.1 [5.1 – 9.7]) and CRVS three times (RR=3.4 [2.4 – 4.7]) more likely to identify a maternal death than health records, while in 2018-19 CRVS (RR=0.8 [0.7 – 0.9])\u0026nbsp;and the MPDSR (RR=0.7 [0.6 – 0.9])\u0026nbsp;were less likely to identify a death than health records.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAfter adjusting for deaths identified in more than one source, in 2007-08, CRVS were two times (RR=2.4 [1.4 – 4.1])\u0026nbsp;and\u0026nbsp;VAs nine times (RR=9.2 [5.7 – 15]) more likely to identify unique deaths than health facility records. In 2018-19,\u0026nbsp;CRVS\u0026nbsp;(RR=0.4 [0.2 – 1.0]) and the MPDSR (RR=1.4 [0.9 – 2.4]) were equally likely to identify unique deaths\u0026nbsp;as health facility records.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor deaths identified in two sources, in 2007-08, health facility records and VAs were equally likely (RR=0.5 [0.2 – 1.5]) while VAs and CRVS were six times more likely (RR=6.3\u0026nbsp;[3.4 – 12])\u0026nbsp;to identify the same deaths than health facility records and CRVS. In 2018-19, health records and the MPDSR were twice more likely (RR=2.4 [1.3 – 4.4]) and CRVS and the MPDSR less likely (R=0.3 [0.1 – 0.9]) to identify the same death as health facility records and CRVS.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 3]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn 2007-08, misclassification of maternal and non-maternal deaths was higher in health facility records (17%) than in CRVS (1.7%; p=0.002) and VA records (5.6%; p=0.016) (Table 4). In 2018-19, misclassification was not different between health facility records and CRVS (p=0.691) or the MPDSR (p=0.489). Misclassification of causes of death was not different between health facility records and CRVS or VA/MPDSR in both 2007-08 and 2018-19 survey (p\u0026gt;0.05). However, misclassification of causes of death appeared to increase between 2007-08 and 2018-19 in each data source (p\u0026lt;0.001). \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 4]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn 2007-08, the sensitivity of the study was 95%, and the specificity was 29%, thus the study had a high probability of correctly identifying true maternal deaths with a high chance (71%) of misclassifying non-maternal deaths as maternal. In 2018-19, the sensitivity was 77%, and the specificity was 83% (Table 5). \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 5]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe assessment of documentation and record keeping showed numerous gaps in Zimbabwe’s maternal mortality data systems, which included lost records, absence of record storerooms in health facilities, assigning of non-medical causes of death in CRVS records and collecting few community deaths in the MPDSR (Table 6). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e[Table 6]\u003c/strong\u003e\u003c/p\u003e"},{"header":"4. DISCUSSION","content":"\u003cp\u003eIdentifying all maternal deaths and correctly identifying and classifying their causes is critical to accurately measuring maternal mortality. Our study presents essential findings on incompleteness, misclassification, quality of documentation and record keeping for pregnancy-related and maternal deaths in Zimbabwe.\u003c/p\u003e \u003cp\u003eAll the data sources in the two surveys provided no more than 60% of the deaths, except VAs which identified 78% of total deaths in 2007-08 because they collected community deaths better. This signifies that missingness of deaths remains a challenge for all data sources in Zimbabwe, as in other developing countries [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Corroborating this finding, WHO has reported high missingness of deaths in developing countries in its estimates and used adjustment factors of up to 150% to correct the under-reporting [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Deaths occurring in the community, and in private and rural health institutions which lack supervision are prone to underreporting [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Even in studies where community health workers collected the data, 30–90% underreporting has been observed [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Failure to know that a woman was pregnant before her death and reluctance to report the deaths causes the incompleteness of pregnancy-related community deaths [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eSelective reporting by health authorities contributes to incompleteness of institutional maternal deaths [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. In an MPDSR evaluation conducted in Ethiopia, study participants described maternal deaths as “political,” reporting that authorities suppressed maternal death reports to evade public rebuke because the avoidable nature of maternal deaths provokes public anger. They said authorities pressured health workers to underreport maternal deaths and paint a picture of success in policies intended to reduce maternal deaths [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Above all, most data sources are unable to identify community deaths, as communities avoid reporting home deaths [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe found an increase in the number and proportion of deaths identified through health records between 2007-08 and 2018-19 possibly due to increase in institutional deliveries. The 2015-16 Zimbabwe Demographic and Health Survey (ZDHS) reported an increase in institutional deliveries from 65% (57% rural and 85% urban) in 2010/11 to 72% (68% rural and 81% urban) in 2015/16 [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], and the MICS reported 86% (82% rural and 94% urban) in 2019 [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Improved documentation of deaths through MPDSR could also be responsible [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIncrease in institutional deliveries could also improve the completeness of pregnancy-related and maternal deaths from 2007-08 to 2018-19. As institutional deliveries increased, more deaths would occur in institutions where they are better documented and reported than community deaths [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Causes of institutional deaths would be documented better and more accurately since medical personnel attend to the deaths. Furthermore, by 2018-19, doctors and nurses had been trained to document and audit maternal deaths through the MPDSR [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAnalysis of the proportion of deaths identified by two or more data sources evaluated the transmission of information across the data systems. High proportions of deaths recorded by two or more data sources were expected, because, by policy, health institutions are required to transfer death records to the RG’s offices. Major hospitals also have RG’s offices within them which collects hospital death reports. In assessing the extent to which the data sources identified the same deaths, low proportions of deaths were identified by any two data sources. In 2018-19, MPDSR and health records identified only 3% of the deaths. Health and CRVS records identified less than 4% of deaths in 2007-08 and less than 10% in 2018-19. CRVS should be recording more deaths because the RG’s offices are conducting mobile registration in the communities countrywide [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHigher misclassification of maternal and non-maternal deaths was observed in 2007-08. During this period, the MPDSR did not exist [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], and health workers were possibly unaware of the definition of maternal death. Even in 2015, an MPDSR evaluation observed that 50% of nurses who used the system in Zimbabwe failed to correctly define a maternal death [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHigh sensitivity of the data sources to correctly identify maternal deaths was observed in 2007-08 and 2018-19, while specificity was low in 2007-08 where data collectors classified late maternal deaths (deaths occurring 42 days after the termination of pregnancy) as maternal deaths. In 2018-19, data collectors strictly applied the definition of maternal death in the data collection, making the sensitivity and specificity high. The sensitivity declined between the 2007-08 and 2018-19 surveys due to misclassification of the causes of death, signifying that the capacity to assign accurate causes of death is still low [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Misclassification of the causes of death was higher in 2018-19 than in 2007-08 because the 2018-19 patient notes were available for obstetricians to use to review and reclassify the deaths.\u003c/p\u003e \u003cp\u003eThe storage of health records was a challenge. CRVS records were better stored than health facility and MPDSR records. Some health facilities lacked storage rooms. Others did not enforce record-keeping standards, and others lacked human resources for record-filling. The completeness of data in the records was a challenge. Supporting this finding, the 2015 MPDRS evaluation reported an average completeness of notification forms of 76%, and only 53% forms had accompanying documents [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. The completeness of records and medical histories helps assessors to assign the correct causes of death [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003cb\u003eStrengths and limitations of the study\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe study recollected, triangulated, and linked the data across sources. Trained obstetricians reviewed and reclassified the deaths using the ICD-MM manual. This enabled the analysis of incompleteness and misclassification of the deaths. However, the study included fewer community deaths in the 2018-19 survey [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], and there were no patient notes for 2007-08 deaths, for the obstetricians to use in the review. Despite these limitations, the three data sources (health records, CRVS records and MPDSR records/VAs) and the recollection process generated sufficient data for estimating incompleteness and misclassification of deaths. The findings support recommendations to strengthen data systems for maternal mortality in low-resource countries.\u003c/p\u003e "},{"header":"Conclusions","content":"\u003cp\u003eIncompleteness and misclassification of maternal deaths are still significant problems in Zimbabwe. Health facility records, CRVS and the MPDSR still inaccurately assign causes to pregnancy-related and maternal deaths. The transfer of death records from health institutions to RG’s offices and MoHCC’s head office remains a challenge in the data systems. Maternal mortality studies using health, CRVS and MPDSR records should triangulate data sources to increase the completeness of data. Record keeping and transmission should be strengthened in these data systems, supported by strengthened policy frameworks and resource allocations to collect community deaths. Medical staff require ongoing capacity strengthening to assign accurate causes of death. Collecting community deaths and assigning correct causes should be an important component of future maternal mortality studies.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eBMI - Bayesian Maternal Mortality Misclassification, CRVS \u0026ndash; Civil Registration and Vital Statistics, ICD \u0026ndash; International Classification of Diseases, ICD-MM \u0026ndash; International Classification of Diseases for Maternal Mortality, MMR \u0026ndash; Maternal Mortality Ratio, MPDSR \u0026ndash; Maternal and Perinatal Death Surveillance and Response, MoHCC \u0026ndash; Ministry of Health and Child Care, RG \u0026ndash; Registrar General, RR \u0026ndash; Risk Ratio, VA \u0026ndash; Verbal Autopsy, VHW \u0026ndash; Village Health Worker, RAMOS \u0026ndash; Reproductive Age Mortality Study, SDG \u0026ndash; Sustainable Development Goal, UI \u0026ndash; Uncertainty Interval, \u0026nbsp;\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe University of Pretoria and the Medical Research Council of Zimbabwe institutional review boards (IRBs) approved the study. The MoHCC and RG\u0026rsquo;s department granted permissions for the study. All institutions approved the collection of data with personally identifying information (PII), including name, date of birth, place of birth and date of death to link and de-duplicate individual women across the data sources. The IRBs approved the collection of clinical notes to be used by obstetricians to code the causes of death. In the 2007-08 survey, verbal autopsy informants gave written informed consent to be interviewed. Informed consent was waived in the 2018-19 survey where all data were collected from secondary records.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData will be made available on a reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no competing interests to declare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), and WHO-Geneva\u0026nbsp;funded in part the recollection and validation of the 2007-08 and 2018-19 data, through a sub-grant from the Improving Maternal Health Measurement (IMHM) Project at the Women \u0026amp; Health Initiative of the Harvard T.H. Chan School of Public Health funded by the Bill \u0026amp; Melinda Gates Foundation [Grant Number OPP1169546].\u0026nbsp;The production of the manuscript was not funded.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRM\u003csup\u003e1\u003c/sup\u003e conceptualized, conducted the analysis, and drafted the paper. RM\u003csup\u003e2\u003c/sup\u003e guided the analysis. LN, GM, CG, SN, RM\u003csup\u003e2\u003c/sup\u003e and SPM reviewed several versions of the paper. All authors approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge the contributions of all 2007-2008 and 2018-2019 Zimbabwe Maternal and Perinatal Mortality Study (ZMPMS) group members (see online supplementary file), who contributed to the protocol development, data collection, data processing, and those who reviewed and classified the deaths. The support received from various MoHCC and RG\u0026rsquo;s department staff is acknowledged and appreciated.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWHO. Sustainable Development Goal 3: Health. https://www.who.int/topics/sustainable-development-goals/targets/en/. Accessed 14 Nov 2020.\u003c/li\u003e\n\u003cli\u003eWHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Trends in maternal mortality 2000 to 2020: Estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. Geneva: World Health Organization; 2023. Licence: CC BY-NC-SA 3.0 IGO.\u003c/li\u003e\n\u003cli\u003eWHO. Maternal mortality: Guidance to improve national reporting. Geneva: World Health Organization; 2022.\u003c/li\u003e\n\u003cli\u003eWHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Trends in maternal mortality 2000 to 2017: Estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Geneva: World Health Organization; 2019. Licence: CC BY-NC-SA 3.0 IGO.\u003c/li\u003e\n\u003cli\u003eGraham WJ, Ahmed S, Stanton C, Abou-Zahr C, Campbell OM: Measuring maternal mortality: an overview of opportunities and options for developing countries. \u003cem\u003eBMC Med\u003c/em\u003e 2008;6: 12.\u003c/li\u003e\n\u003cli\u003ePeterson E, Chou D, Moller AB, Gemmill A, Say L, Alkema L: Estimating misclassification errors in the reporting of maternal mortality in national civil registration vital statistics systems: A Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity for multiple countries and years with missing data. \u003cem\u003eStat Med\u003c/em\u003e 2022.\u003c/li\u003e\n\u003cli\u003eAdair T, Lopez AD: How reliable are self-reported estimates of birth registration completeness? Comparison with vital statistics systems. \u003cem\u003ePLoS One\u003c/em\u003e 2021;16(6): e0252140.\u003c/li\u003e\n\u003cli\u003eWHO. WHO guidance for measuring maternal mortality from a census. Geneva: World Health Organization. 2013.\u003c/li\u003e\n\u003cli\u003eDHS: Guide to DHS Statistics: Maternal Mortality and Pregnancy-related Mortality. Maryland: \u003cem\u003eDemographic and Health Survey.\u003c/em\u003e 2021.\u003c/li\u003e\n\u003cli\u003eDanel I, Graham W, Stupp P, Castillo P: \u0026quot;Applying the Sisterhood Method for Estimating Maternal Mortality to a Health Facility-based Sample: A Comparison with Results from a Household- based Sample.\u0026quot;. \u003cem\u003eStud Fam Plann\u003c/em\u003e 1997;28(3): 266.\u003c/li\u003e\n\u003cli\u003eThe World Bank. Civil registration and vital statistics: key to better data on maternal mortality. Geneva: https://blogs.worldbank.org/health/civil-registration-and-vital-statistics-key-better-data-maternal-mortality. Accessed 17 Jan 2024.\u003c/li\u003e\n\u003cli\u003eKao S, Chen LM, Shi L, Weinrich MC: Underreporting and misclassification of maternal mortality in Taiwan. \u003cem\u003eActa Obstet Gynecol Scand\u003c/em\u003e 1997;76(7): 629-636.\u003c/li\u003e\n\u003cli\u003eMusarandega R, Machekano R, Pattinson R, Munjanja SP, Zimbabwe M, Perinatal Mortality Study g: Protocol for analysing the epidemiology of maternal mortality in Zimbabwe: A civil registration and vital statistics trend study. \u003cem\u003ePLoS One\u003c/em\u003e 2021;16(6): e0252106.\u003c/li\u003e\n\u003cli\u003eMusarandega R, Ngwenya S, Murewanhema G, Machekano R, Magwali T, Nystr\u0026ouml;m L, et al.: Changes in causes of pregnancy-related and maternal mortality in Zimbabwe 2007-08 to 2018-19: findings from two reproductive age mortality surveys. \u003cem\u003eBMC Public Health\u003c/em\u003e 2021.\u003c/li\u003e\n\u003cli\u003eMusarandega R, Cresswell J, Magwali T, Makosa D, Machekano R, Ngwenya S, et al.: Maternal mortality decline in Zimbabwe, 2007/2008 to 2018/2019: findings from mortality surveys using civil registration, vital statistics and health system data. \u003cem\u003eBMJ Glob Health\u003c/em\u003e 2022;7(8): e009465.\u003c/li\u003e\n\u003cli\u003eWHO. The WHO Application of ICD-10 to deaths during pregnancy, childbirth and the puerperium: ICD-MM. World Health Organization. Geneva. 2011.\u003c/li\u003e\n\u003cli\u003eAll Africa. Zimbabwe: Countrywide mobile registration blitz begins. The Herald, 2021. https://allafrica.com/stories/202109140535.html. Accessed 16 May 2022.\u003c/li\u003e\n\u003cli\u003eKubatana. BVR, IDs\u0026apos; mobile registration: Citizens speak. Kubatana.net; 2017. https://kubatana.net/2017/09/19/bvr-ids-mobile-registration-citizens-speak/. Accessed 16 May 2022.\u003c/li\u003e\n\u003cli\u003eWHO. The 2007 WHO Verbal Autopsy Instrument. https://www.who.int/standards/classifications/other-classifications/verbal-autopsy-standards-ascertaining-and-attributing-causes-of-death-tool. Accessed 9 Dec 2021.\u003c/li\u003e\n\u003cli\u003eMOHCC: Maternal and Perinatal Death Surveillance and Response Report. Ministry of Health and Child Care Zimbabwe, Family and Child Health Department; 2018.\u003c/li\u003e\n\u003cli\u003eMOHCC. Guidelines for maternal and perinatal death audits in Zimbabwe. Harare: Zimbabwe Ministry of Health and Child Care; 2013.\u003c/li\u003e\n\u003cli\u003eMaphosa M, Juru TP, Masuka N, Mungati M, Gombe N, Nsubuga P, et al.: Evaluation of the Maternal Death Surveillance and response system in Hwange District, Zimbabwe, 2017. \u003cem\u003eBMC Pregnancy Childbirth\u003c/em\u003e 2019;19(1): 103.\u003c/li\u003e\n\u003cli\u003eKhan KS, Wojdyla D, Say L, G\u0026uuml;lmezoglu AM, Van Look PFA: WHO analysis of causes of maternal death: a systematic review. \u003cem\u003eThe Lancet\u003c/em\u003e 2006;367(9516): 1066-1074.\u003c/li\u003e\n\u003cli\u003eSTATA. Statistical software for data science. Texas: STATA Coporation LLC; 2022. https://www.stata.com/. Accessed 21 Jan 2022.\u003c/li\u003e\n\u003cli\u003eWHO. Medical Records Manual: A guide for Developing Countries. World Health Organization. 2006.\u003c/li\u003e\n\u003cli\u003eHPCSA. Guideline for the good practice in the health care profession: Guidelines on the keeping of patient records. Booklet 9. Pretoria: Health Professions Council of South Africa; 2016.\u003c/li\u003e\n\u003cli\u003eSay L, Chou D: Better understanding of maternal deaths\u0026mdash;the new WHO cause classification system. \u003cem\u003eBJOG\u003c/em\u003e 2011;118(s2): 15-17.\u003c/li\u003e\n\u003cli\u003eWilmoth JR, Mizoguchi N, Oestergaard MZ, Say L, Mathers CD, Zureick-Brown S, et al.: A New Method for Deriving Global Estimates of Maternal Mortality. \u003cem\u003eStat Politics Policy\u003c/em\u003e 2012;3(2).\u003c/li\u003e\n\u003cli\u003eRutgers S: Two years maternal mortality in Matebeleland north Province, Zimbabwe. \u003cem\u003eCent Afr J Med\u003c/em\u003e 2001;47(2): 39-43.\u003c/li\u003e\n\u003cli\u003eFaith Mutsigiri-Murewanhema PTM, Tsitsi Juru, Notion Tafara Gombe, Donewell Bangure, More Mungati, Mufuta Tshimanga: Evaluation of the maternal mortality surveillance system in Mutare district, Zimbabwe, 2014-2015: a cross sectional study. \u003cem\u003ePan Afr Med J\u003c/em\u003e 2017;27(204).\u003c/li\u003e\n\u003cli\u003eSilva R, Amouzou A, Munos M, Marsh A, Hazel E, Victora C, et al.: Can Community Health Workers Report Accurately on Births and Deaths? Results of Field Assessments in Ethiopia, Malawi and Mali. \u003cem\u003ePLoS One\u003c/em\u003e 2016;11(1): e0144662.\u003c/li\u003e\n\u003cli\u003eMelberg A, Mirkuzie AH, Sisay TA, Sisay MM, Moland KM: \u0026apos;Maternal deaths should simply be 0\u0026apos;: politicization of maternal death reporting and review processes in Ethiopia. \u003cem\u003eHealth Policy Plan\u003c/em\u003e 2019;34(7): 492-498.\u003c/li\u003e\n\u003cli\u003eZimStat and ICF International. Zimbabwe Demographic and Health Survey 2015: final report. Zimbabwe national statistics agency (ZIMSTAT) and ICF International Rockville, Maryland, USA; 2016.\u003c/li\u003e\n\u003cli\u003eZimStat and UNICEF. Multiple indicator cluster survey 2017: final report. https://www.unicef.org/zimbabwe/reports/zimbabwe-2019-mics-survey-findings-report; 2019. Accessed 20 Sep 2020.\u003c/li\u003e\n\u003cli\u003eEkirapa-Kiracho E, Waiswa P, Rahman MH, Makumbi F, Kiwanuka N, Okui O, et al.: Increasing access to institutional deliveries using demand and supply side incentives: early results from a quasi-experimental study. \u003cem\u003eBMC Int Health Hum Rights\u003c/em\u003e 2011;11(1): S11.\u003c/li\u003e\n\u003cli\u003eMCHIP. Assessment of maternal and perinatal death surveillance and response implementation in Zimbabwe. Technical report. Harare: Maternal and Child Health Integrated Program; 2017.\u003c/li\u003e\n\u003cli\u003eAmeh C, Adegoke A, Pattinson R, van den Broek N: Using the new ICD-MM classification system for attribution of cause of maternal death\u0026mdash;a pilot study. \u003cem\u003eBJOG\u003c/em\u003e 2014;121(s4): 32-40.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1: Missingness of pregnancy-related and maternal deaths in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; The P-value for the Chi-square -test of difference in the proportion of missed deaths between the two surveys.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"870\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.959723820483314%\" rowspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of deaths\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"35.44303797468354%\" colspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2007-08\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"35.67318757192175%\" colspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2018-19\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.924050632911392%\" rowspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"15.6957928802589%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBefore verifying\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.181229773462782%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAfter verifying\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.961165048543688%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMissingness\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.181229773462782%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBefore verifying\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.6957928802589%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAfter verifying\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.284789644012946%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMissingness\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.959723820483314%\" valign=\"top\"\u003e\n \u003cp\u003ePregnancy-related\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.162255466052935%\"\u003e\n \u003cp\u003e237\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.507479861910241%\"\u003e\n \u003cp\u003e325\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.773302646720369%\"\u003e\n \u003cp\u003e27%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.507479861910241%\"\u003e\n \u003cp\u003e112\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.162255466052935%\"\u003e\n \u003cp\u003e137\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.003452243958574%\"\u003e\n \u003cp\u003e18%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.924050632911392%\"\u003e\n \u003cp\u003e0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.959723820483314%\" valign=\"top\"\u003e\n \u003cp\u003eMaternal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.162255466052935%\"\u003e\n \u003cp\u003e208\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.507479861910241%\"\u003e\n \u003cp\u003e296\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.773302646720369%\"\u003e\n \u003cp\u003e30%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.507479861910241%\"\u003e\n \u003cp\u003e104\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.162255466052935%\"\u003e\n \u003cp\u003e130\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.003452243958574%\"\u003e\n \u003cp\u003e19%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.924050632911392%\"\u003e\n \u003cp\u003e0.037\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2: Number and proportion of missed pregnancy-related deaths by the cause of missingness in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; the P-value for the Chi-square test of difference in proportion of missed deaths by the cause of missingness between the two surveys.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"688\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCauses of missingness\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"43.604651162790695%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber (%) of deaths missed by cause of missingness.\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\" rowspan=\"2\" valign=\"top\"\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 width=\"50%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2007-08 (N=325)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"50%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2018-19 (N=137)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" valign=\"top\"\u003e\n \u003cp\u003eDeaths missed in the 1\u003csup\u003est\u003c/sup\u003e round of data collection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e80 (25%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e25 (18%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\"\u003e\n \u003cp\u003e0.136\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" valign=\"top\"\u003e\n \u003cp\u003eDeaths incorrectly classified as maternal/non-maternal deaths\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e8 (3.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e2 (1.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\"\u003e\n \u003cp\u003e0.493\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" valign=\"top\"\u003e\n \u003cp\u003eDeaths not entered in the database\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e18 (5.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" valign=\"top\"\u003e\n \u003cp\u003eTwin deliveries duplicated in the database\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e4 (1.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\"\u003e\n \u003cp\u003e0.192\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"36.627906976744185%\" valign=\"top\"\u003e\n \u003cp\u003eDeaths duplicated in the 1\u003csup\u003est\u003c/sup\u003e database\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e11 (4.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.802325581395348%\"\u003e\n \u003cp\u003e0 (0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.767441860465116%\"\u003e\n \u003cp\u003e0.029\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3: Number and percentage of deaths identified and log-linear regression risk ratio (RR), with 95% confidence intervals, of completeness of pregnancy-related deaths in different data sources in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19; the P-value for Fisher\u0026rsquo;s exact test of difference in the percentage of deaths identified by each data source in the two surveys.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"942\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"40.12738853503185%\" colspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003e2007-08 (n=325)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"41.40127388535032%\" colspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2018-19 (n=137)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\" rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value, comparing percent of deaths identified by a data source in the surveys\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.005208333333332%\"\u003e\n \u003cp\u003e\u003cstrong\u003eData source where deaths were identified\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.369791666666666%\"\u003e\n \u003cp\u003e\u003cstrong\u003eN (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.84375%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRR (95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.567708333333332%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eData source where deaths were identified\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.588541666666666%\"\u003e\n \u003cp\u003e\u003cstrong\u003eN (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.625%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRR (95% CI)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e36 (11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e82 (60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS records\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e121 (37)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e3.4 ()2.4 \u0026ndash; 4.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e67 (49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e0.8 (0.7 \u0026ndash; 0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e0.020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eVAs records\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e254 (78)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e7.1 (5.1 \u0026ndash; 9.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eMPDSR database\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e58 (42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e0.7 (0.6 \u0026ndash; 0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e19 (5.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003eReference\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\"\u003e\n \u003cp\u003eHealth facility records only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e27 (20)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003eReference\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\"\u003e\n \u003cp\u003eCRVS records only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e46 (14)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e2.4 (1.4 \u0026ndash; 4.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\"\u003e\n \u003cp\u003eCRVS records only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e12 (8.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e0.4 (0.2 \u0026ndash; 1.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\"\u003e\n \u003cp\u003eVAs only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e174 (53)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e9.2 (5.7 \u0026ndash; 15)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\"\u003e\n \u003cp\u003eMPDSR database only\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e39 (28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e1.4 (0.9 \u0026ndash; 2.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility and CRVS records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e11 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility and CRVS records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e13 (9.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility and VA records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e6 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e0.5 (0.2 \u0026ndash; 1.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eHealth records and MPDSR\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e31 (23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e2.4 (1.3 \u0026ndash; 4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.940552016985137%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS and VA records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.084925690021231%\"\u003e\n \u003cp\u003e69 (21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.101910828025478%\"\u003e\n \u003cp\u003e6.3 (3.4 \u0026ndash; 12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.21443736730361%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS and MPDSR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.447983014861995%\"\u003e\n \u003cp\u003e4 (2.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.738853503184714%\"\u003e\n \u003cp\u003e0.3 (0.1 \u0026ndash; 0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.471337579617835%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"7\" valign=\"top\"\u003e\n \u003cp\u003eN/A \u0026ndash; Not applicable in the first three rows because the data sources in these rows count all deaths identified by each source including those that were identified in more than one source. Also, not applicable in the second three rows because the deaths in each row were not exclusive to those data sources. \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4: Number and percentage of misclassified deaths by the type of misclassification and data source in the Zimbabwe Maternal and Perinatal Mortality Surveys, 2007-08 and 2018-19;\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eFisher\u0026rsquo;s exact p-value comparing misclassification between surveys.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"723\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.012448132780083%\" rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eType of misclassification\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.203319502074688%\" rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eData source\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"58.78284923928077%\" colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eSurvey year, number, and percentage of misclassified deaths\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.001383125864454%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value comparing 2007-08 vs 2018-19\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"40.077071290944126%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2007-08\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"41.811175337186896%\" colspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e2018-19\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.11175337186898%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"18.497109826589597%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber and percentage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.57996146435453%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value comparing misclassification in data sources\u003csup\u003e3\u003c/sup\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.845857418111752%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber and percentage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.965317919075144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value comparing misclassification in data sources\u003csup\u003e3\u003c/sup\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.11175337186898%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.012448132780083%\" rowspan=\"3\"\u003e\n \u003cp\u003eMisclassification of survival outcome (died/alive)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.203319502074688%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.278008298755188%\"\u003e\n \u003cp\u003e2/41\u003c/p\u003e\n \u003cp\u003e(4.9%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.491009681881051%\"\u003e\n \u003cp\u003eReference\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.246196403872752%\"\u003e\n \u003cp\u003eN/A\u003csup\u003e4\u003c/sup\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.767634854771785%\"\u003e\n \u003cp\u003eReference\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.001383125864454%\"\u003e\n \u003cp\u003eN/A\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS (RG\u0026rsquo;s) records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e2/121\u003c/p\u003e\n \u003cp\u003e(1.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.266\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003eN/A\u003csup\u003e4\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eVA or MPDSR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e5/252\u003c/p\u003e\n \u003cp\u003e(2.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003eN/A\u003csup\u003e4\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.012448132780083%\" rowspan=\"3\"\u003e\n \u003cp\u003eMisclassification of maternal and non-maternal deaths\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.203319502074688%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.278008298755188%\"\u003e\n \u003cp\u003e7/41\u003c/p\u003e\n \u003cp\u003e(17%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.491009681881051%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.246196403872752%\"\u003e\n \u003cp\u003e7/82\u003c/p\u003e\n \u003cp\u003e(8.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.767634854771785%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.001383125864454%\"\u003e\n \u003cp\u003e0.160\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS (RG\u0026rsquo;s) records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e2/121\u003c/p\u003e\n \u003cp\u003e(1.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e4/67\u003c/p\u003e\n \u003cp\u003e(6.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.755\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eVA or MPDSR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e14/252\u003c/p\u003e\n \u003cp\u003e(5.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e9/58\u003c/p\u003e\n \u003cp\u003e(16%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.281\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.012448132780083%\" rowspan=\"3\"\u003e\n \u003cp\u003eMisclassification of type of maternal death (Direct/Indirect)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.203319502074688%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.278008298755188%\"\u003e\n \u003cp\u003e2/41\u003c/p\u003e\n \u003cp\u003e(4.9%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.491009681881051%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.246196403872752%\"\u003e\n \u003cp\u003e4/82\u003c/p\u003e\n \u003cp\u003e(4.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.767634854771785%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.001383125864454%\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS (RG\u0026rsquo;s) records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e8/121\u003c/p\u003e\n \u003cp\u003e(6.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.690\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e2/67\u003c/p\u003e\n \u003cp\u003e(3.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.691\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e0.499\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eVA or MPDSR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e12/252\u003c/p\u003e\n \u003cp\u003e(4.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.974\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e5/58\u003c/p\u003e\n \u003cp\u003e(8.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.489\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e0.332\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.012448132780083%\" rowspan=\"3\"\u003e\n \u003cp\u003eMisclassification of causes of death\u003csup\u003e1\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.203319502074688%\" valign=\"top\"\u003e\n \u003cp\u003eHealth facility records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.278008298755188%\"\u003e\n \u003cp\u003e7/41\u003c/p\u003e\n \u003cp\u003e(17%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.491009681881051%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"14.246196403872752%\"\u003e\n \u003cp\u003e43/82\u003c/p\u003e\n \u003cp\u003e(52%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.767634854771785%\"\u003e\n \u003cp\u003eReference\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.001383125864454%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS (RG\u0026rsquo;s) records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e17/121\u003c/p\u003e\n \u003cp\u003e(14%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.619\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e29/67\u003c/p\u003e\n \u003cp\u003e(43%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.323\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.5%\" valign=\"top\"\u003e\n \u003cp\u003eVA or MPDSR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16%\"\u003e\n \u003cp\u003e37/252\u003c/p\u003e\n \u003cp\u003e(15%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.666666666666668%\"\u003e\n \u003cp\u003e0.643\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.166666666666668%\"\u003e\n \u003cp\u003e35/58\u003c/p\u003e\n \u003cp\u003e(60%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19%\"\u003e\n \u003cp\u003e0.391\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.666666666666666%\"\u003e\n \u003cp\u003e\u0026lt;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"7\" valign=\"top\"\u003e\n \u003cp\u003e\u003csup\u003e1\u0026nbsp;\u003c/sup\u003eReviewers assigned a different cause of death after identifying a different underlying cause.\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e2\u0026nbsp;\u003c/sup\u003eThe data sources were Verbal Autopsy (VA) 2007-08 and MPDSR for 2018-19 \u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e3\u003c/sup\u003e For each type of misclassification, the Fisher\u0026rsquo;s exact or Chi-square p-value compares the percentage of the reference category (health records) and each of the other data sources (CRVS and VA or MPDSR).\u003c/p\u003e\n \u003cp\u003e\u003csup\u003e4\u003c/sup\u003e There was no misclassification of survival outcomes (died/alive) because the study collected data for deceased women only\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"https://myfiles.space/user_files/132203_cef980177e9a226b/132203_custom_files/img1708460294.png\" style=\"width: 668px; height: 611.078px;\" width=\"668\" height=\"611.078\"\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 6: Qualitative findings on documentation and record keeping in the Zimbabwe Maternal and Perinatal Mortality Surveys; results from the 2020 data collection process. \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"684\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.54385964912281%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eData Source\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"82.45614035087719%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eKey observations\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.54385964912281%\" valign=\"top\"\u003e\n \u003cp\u003eHealth records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"82.45614035087719%\" valign=\"top\"\u003e\n \u003cul\u003e\n \u003cli\u003eIn large hospitals with dedicated records rooms and staff recent patient charts/notes/registers were available and accessible.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eHealth facilities without record rooms had poorly archived patient records; records were dumped on the floor in some health facilities.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eSome health facilities with records rooms did not follow standard record filing and removal practices.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eIn some health facilities, the individual staff kept maternal death records, and when these staffs were off duty, the records couldn\u0026rsquo;t be accessed. \u0026nbsp;\u003c/li\u003e\n \u003cli\u003eNurses in one district hospital without a medical officer reported that they signed death certificates but were not allowed to assign medical causes of death on the certificates. Hence, they omitted the causes of death or assigned non-medical causes such as \u0026ldquo;short illness\u0026rdquo; or \u0026ldquo;long illness.\u0026rdquo;\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.54385964912281%\" valign=\"top\"\u003e\n \u003cp\u003eCRVS records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"82.45614035087719%\" valign=\"top\"\u003e\n \u003cul\u003e\n \u003cli\u003eAll RG\u0026rsquo;s offices visited had secure records rooms managed by dedicated records staff. Death records were sequentially filed in box files by registration date, and the boxes were labelled and sequentially filed by year.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eFor some deaths that occurred in health institutions, medical certificates were attached to the death records, but this practice was not consistent across districts.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eDeaths reported by family members and deaths registered during mobile registration campaigns had vague causes of death, such as \u0026ldquo;natural causes\u0026rdquo;, \u0026ldquo;natural death\u0026rdquo;, \u0026ldquo;swollen leg\u0026rdquo;, \u0026ldquo;running stomach\u0026rdquo;, \u0026ldquo;headache\u0026rdquo;, etc.\u0026nbsp;\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"17.54385964912281%\" valign=\"top\"\u003e\n \u003cp\u003eMPDSR database\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"82.45614035087719%\" valign=\"top\"\u003e\n \u003cul\u003e\n \u003cli\u003eDeath notification forms completed by health facilities were available at the MoHCC\u0026rsquo;s provincial offices and head office. In provincial offices, the forms were filed in box files but kept in individuals\u0026rsquo; offices. When these individuals were not present, the records could not be accessed.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eForms received at the head office were captured into an electronic database. After electronic capture, the forms were not properly filed because of a shortage of storage space and staff.\u003c/li\u003e\n \u003cli\u003eThough the system is meant to capture community deaths, staff in all districts reported that (i) they faced challenges when investigating community deaths, especially among members of the apostolic religious sects, (ii) most women in these sects deliver at home, (ii) many community deaths occur in these communities, (iv) their faith healers, religious leaders and family members do not cooperate with health workers who come to enquire about community deaths. \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-epidemiology-and-global-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Journal of Epidemiology and Global Health](https://www.springer.com/journal/44197)","snPcode":"44197","submissionUrl":"https://submission.nature.com/new-submission/44197/3","title":"Journal of Epidemiology and Global Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Maternal deaths, Pregnancy-related deaths, Maternal mortality, Missingness, Incompleteness, Misclassification","lastPublishedDoi":"10.21203/rs.3.rs-3891799/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3891799/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eIntroduction: \u003c/strong\u003eReducing maternal mortality is a high-priority global health goal, especially in sub-Saharan Africa, where the maternal mortality ratios (MMRs) of most of the countries is higher than the average global MMR. We implemented two cross-sectional reproductive age mortality surveys, in 2007-08 and 2018-19, to assess changes in the MMR and causes of death in Zimbabwe after a raft of interventions implemented to reduce maternal mortality. This paper analysed the missingness and misclassification of deaths in the surveys.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe compared percentages\u003cstrong\u003e \u003c/strong\u003eof missed deaths\u003cstrong\u003e \u003c/strong\u003ein each survey using the Chi-square test. The risk ratios of missing deaths in different data sources in each survey were calculated using log-linear regression models. Proportions of misclassified deaths were compared using Fisher’s exact test and sensitivity and specificity of incompleteness and misclassification of deaths compared using the six-box method and the Chi-square test.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eThe proportion of missed pregnancy-related deaths decreased from 27% in 2007-08 to 18% in 2018-19 (p=0.044) and the proportion of missed maternal deaths decreased from 30% in 2007-08 to 19% in 2018-19. Misclassification of maternal deaths in health records was 17% in 2007-08 and 8.5% in 2018-19 (p=0.160). The proportion of pregnancy-related deaths identified through health records increased from 11% in 2007-08 to 60% in 2018-19 (p\u0026lt;0.001). Sensitivity of incompleteness and misclassification of deaths was 95% in 2007-08 and 77% in 2018-19, and specificity was 29% and 83% respectively.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eIncompleteness and misclassification of maternal deaths are still a challenge in Zimbabwe. Maternal death studies must triangulate data sources to improve the completeness of data and efforts to reduce misclassification of deaths should continue to improve maternal mortality estimates.\u003c/p\u003e","manuscriptTitle":"Incompleteness and misclassification of maternal deaths in Zimbabwe: data from two reproductive age mortality surveys, 2007-08 and 2018-19","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-02-20 20:20:42","doi":"10.21203/rs.3.rs-3891799/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-02-21T01:43:13+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-02-20T08:27:04+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-02-19T05:23:10+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Epidemiology and Global Health","date":"2024-01-23T18:10:24+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-epidemiology-and-global-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Journal of Epidemiology and Global Health](https://www.springer.com/journal/44197)","snPcode":"44197","submissionUrl":"https://submission.nature.com/new-submission/44197/3","title":"Journal of Epidemiology and Global Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Open","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"050c5a31-8fde-4ef7-aab8-45e7f8dcd506","owner":[],"postedDate":"February 20th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2024-10-15T18:23:12+00:00","versionOfRecord":[],"versionCreatedAt":"2024-02-20 20:20:42","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3891799","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3891799","identity":"rs-3891799","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}