Skip to main content

Incompleteness and misclassification of maternal death recording: a systematic review and meta-analysis



To quantify the extent of incompleteness and misclassification of maternal and pregnancy related deaths, and to identify general and context-specific factors associated with incompleteness and/or misclassification of maternal death data.


We conducted a systematic review of incompleteness and/or misclassification of maternal and pregnancy-related deaths. We conducted a narrative synthesis to identify methods used to capture and classify maternal deaths, as well as general and context specific factors affecting the completeness and misclassification of maternal death recording. We conducted a meta-analysis of proportions to obtain estimates of incompleteness and misclassification of maternal death recording, overall and disaggregated by income and surveillance system types.


Of 2872 title-abstracts identified, 29 were eligible for inclusions in the qualitative synthesis, and 20 in the meta-analysis. Included studies relied principally on record linkage and review for identifying deaths, and on review of medical records and verbal autopsies to correctly classify cause of death. Deaths to women towards the extremes of the reproductive age range, those not classified by a medical examiner or a coroner, and those from minority ethnic groups in their setting were more likely misclassified or unrecorded. In the meta-analysis, we found maternal death recording to be incomplete by 34% (95% CI: 28–48), with 60% sensitivity (95% CI: 31–81.). Overall, we found maternal mortality was under-estimated by 39% (95% CI: 30–48) due to incompleteness and/or misclassification. Reporting of deaths away from the intrapartum, due to indirect causes or occurring at home were less complete than their counterparts. There was substantial between and within group variability across most results.


Maternal deaths were under-estimated in almost all contexts, but the extent varied across settings. Countries should aim towards establishing Civil Registration and Vital Statistics systems where they are not instituted. Efforts to improve the completeness and accuracy of maternal cause of death recording, such as Confidential Enquiries into Maternal Deaths, are needed even where CRVS is considered to be well-functioning.

Peer Review reports


Maternal mortality is an important measure of women’s health during the reproductive ages. The United Nations Maternal Mortality Estimation Interagency Group (UN MMEIG) is one of the entities that produce global maternal mortality estimates [20, 46]. Their most recent estimates found that, in 2020, 223 women died due to maternal causes per 100,000 live births, with the majority of these deaths being from preventable causes [46]. In recognition of its importance as a global indicator of women’s health and development, maternal mortality is the focus of Sustainable Development target 3.1: to reduce the global maternal mortality ratio to less than 70 deaths per 100,000 live births by 2030, with no country having a maternal mortality ratio (MMR) higher than 140.

Investing and prioritising interventions to reduce maternal mortality requires accurate and timely data on the levels and trends of maternal mortality. Interventions can furthermore be tailored to sub-national level, if the data is of adequate quality. The need for accurate measurement of maternal mortality has been emphasised by the WHO in the Strategies toward Ending Preventable Maternal Mortality” (EPMM) report published in 2015. The report highlighted a need to improve measurement systems and data quality and to ensure that all maternal deaths are counted [40]. The 2021 EPMM Strategy published on April 2021 further stressed WHO’s commitment to this recommendation [43].

Civil Registration and Vital Statistics systems (CRVS) are the preferred source of data for producing comparable maternal mortality statistics. Many low- and middle-income countries (LMICs) do not have complete and accurate CRVS systems, and instead rely on other data sources, such as, facility-based systems, district management information systems, and Health and Demographic surveillance systems. These systems often do not have adequate coverage, and rarely provide nationally representative estimates, as a result, LMICS often count on ad hoc population-based surveys, though these tend to be sporadic and untimely.

Even in countries with a well-functioning CRVS, mortality data are often incomplete. Estimation errors are common across different adult mortality indicators [28], but there can be additional challenges in the measurement of maternal mortality. These challenges include the additional information required to classify a death as maternal, such as accurate assignment of cause of death, pregnancy status and/or timing of the death relative to pregnancy.

Errors in maternal mortality estimation can be conceptually and operationally categorised into two types, 1) incompleteness (also known as missingness or under-reporting), defined as whether the death is registered into a designated data collection system (often a CRVS) or an alternative routine data reporting system. 2) misclassification refers to whether the cause of death is accurately documented, which affects whether the death is considered maternal or non-maternal, and is expressed as sensitivity and specificity of the surveillance system classification of cause of death [45]. Reducing incompleteness requires interventions aimed at establishing and/or improving death and cause of death registration systems, while reducing misclassification requires interventions aimed at improving the accuracy of cause of death classification, such as the use of the International Classification of Disease (ICD) for certifying and coding the cause of death [42].

Improving the performance of registration systems requires continuous monitoring of the extent of incompleteness and/or misclassification of maternal and pregnancy related deaths. Furthermore, understanding which subset(s) of maternal deaths are more likely to be incomplete or misclassified can provide insights for researchers and stakeholders in implementing targeted interventions or studies aimed at identifying and/or classifying maternal deaths.

Currently, the MMEIG estimates sensitivity and specificity of maternal death recording using a Bayesian hierarchical modelling approach. Due to the requirement of national representativeness for inclusion of studies, they do not estimate under-reporting and misclassification in non-CRVS surveillance systems, for example in facility based or in health and demographic surveillance systems, hence, their estimates may not generalise to non-CRVS systems. Such systems may still provide valuable information on maternal mortality, especially in countries with limited data. Furthermore, many countries are working towards improving the coverage of their surveillance systems, and in this review, we evaluate and compare the quality of maternal death recording and classification between several surveillance systems, in an attempt to provide insight into their quality for maternal mortality measurement.

The primary aim of this systematic review and meta-analysis is to quantify the extent of incompleteness and misclassification of maternal deaths, and describe the methods used to identify missing and misclassified maternal deaths. Additionally, we describe the socio-demographic and clinical characteristics of women whose deaths are more likely to be unreported or misclassified, and the key contextual factors associated with incompleteness and/or misclassification.

Materials and methods

This study is a narrative review and a meta-analysis reporting on a sub-set of data from a larger systematic review of studies that report on the level of maternal and or pregnancy related deaths, and the completeness and/or misclassification of these deaths. The review was conducted as part of the UN MMEIG estimation process for the 2022 round of global MMR estimates [46].

The main outcomes for the meta-analysis were i) incompleteness, and ii) misclassification of maternal or pregnancy related deaths within a routine surveillance system (specificity and sensitivity). We used the ICD-11 definitions of maternal and pregnancy related deaths, and of direct and indirect obstetric deaths [44].

In the narrative review, we aimed to synthesise information on the methods used by the identified studies to obtain the true number of maternal deaths, and the context specific challenges they identified for measuring maternal mortality.

Definition of key terms and measures

Sensitivity refers to the proportion of correctly classified maternal deaths out of all true maternal deaths. Specificity refers to the proportion of correctly classified non-maternal deaths out of all true non-maternal deaths. False negative referred to the maternal deaths that are incorrectly recorded as non-maternal. incompleteness refers to the proportion of the true maternal deaths identified in the population that were not previously recorded in the routine surveillance system. The proportion under-estimated was defined as the proportion of deaths not reported from the total maternal deaths, whether from incompleteness, false negatives, or both.

The above metrics likely differ according to the system being compared to, and we therefore stratified our analysis based on the comparison used. These categories were CRVS; national Health Management Information System (if the system described was instituted nationally, but was facility based); facility-based. (if it only includes deaths from one or several facilities but not all facilities in the country); Health and Demographic Surveillance system (if it was instituted in a given area (not national) and dedicated for ongoing demographic surveillance at both facility and community levels).

Eligibility criteria

For a study to be included in the narrative review it had to satisfy the following criteria: either 1) provide empirical data on the completeness and/or misclassification of maternal and/or pregnancy related deaths or 2) conduct an investigation to obtain the true number of maternal and/or pregnancy related deaths by triangulating data sources, and the results of the investigation could be compared against a CRVS observation obtained from the WHO Mortality Database [39]. 3) all included studies must report the definition used to classify the deaths (maternal, pregnancy-related).

Studies were included in the quantitative analysis if they provided the number of missed or misclassified maternal or pregnancy related deaths in a surveillance system that is routinely in place in the area or facility covered by the study. We excluded from the meta-analysis studies that only provided a percentage of under-estimation from which we could not obtain the number of missed or misclassified deaths.

Search strategy

We searched five bibliographic databases: Medline (OVID), EMBASE (OVID), EBSCO, Global Index Medicus, and Web of Science – Russian index (Russian-script). The searches covered publications indexed between September 2016 – when the searches for the 2019 Update were run, the results of this have been previously reported [31] – and March 2021. In addition to bibliographic databases, we searched National Statistics and Health Ministries’ official websites of the 194 UN Member States for any specialised studies that could satisfy the inclusion criteria described above. Finally, we screened the references of the studies that were included from the bibliographic search for additional sources. We used search terms relating to maternal and pregnancy related mortality, and under-reporting, incomplete recording or data verification related terms developed by Gülmezoglu et al, 2004 [17]. We only excluded articles published in Chinese, for lack of translators. The search terms are listed in Supplementary information 1.

Screening & data extraction

The identified citations were screened independently by two reviewers (SA, FA). The review process was completed in two stages. In the first stage, duplicate studies (n = 463) were removed, and title and abstracts of the studies were screened based on the inclusion criteria described above.

In the second stage, the full text for the identified studies was obtained and reviewed, and their usability was further examined based on the above criteria by two reviewers. A third reviewer was consulted to adjudicate on discordant opinions. The screening was managed using DistillerSR application [14].

For studies passing this stage, two reviewers extracted information on the study site and date, study design, coverage, method of identifying and ascertaining cause of death, the comparator, extent of incompleteness, specificity and/or sensitivity of death reporting, characteristics of misclassified and/or missed (incomplete) deaths and any contextual moderators influencing the registration of maternal/pregnancy related deaths.

For studies that did not compare directly to a registration system (eligibility criteria 2), two reviewers extracted the number of maternal deaths from the WHO mortality database for the year the study is reporting on.

In the case of non-English manuscripts, the extractors used a translation software, and a native speaker was sought to validate that the extraction was accurate.

Study risk of bias assessment

Risk of bias was assessed using a scale designed by the authors of this study, which evaluates the ability of the study to capture maternal deaths from four aspects. The first is whether the study methodology will enable it to identify specific deaths that have been reported in the literature to be prone to missingness or misclassification. These include early pregnancy deaths, indirect obstetric deaths (deaths due to a disease (other than HIV) aggravated by the effects of pregnancy and deaths occurring at home. For a study to identify these deaths, it should a) examine all deaths to women of reproductive age, including those occurring at home and in non-obstetric departments; and b) conduct its own review of cause of death using available data sources. The second aspect was the method of determining the cause of death. We considered a verbal autopsy or a medical record review to be inferior to a death certificate in a country with a complete CRVS – as defined in the MMEIG MMR estimation methodology [41] – and superior to a death certificate from a country with no or low completeness CRVS system. We considered a verbal autopsy equal to a medical review since it was difficult to determine the level of completeness of the records reviewed. A medical examiner or forensic report, or a combination of two or more of the above methods were considered the most robust.

Third was the percentage of records or deaths that were not reviewed or that did not have enough information to ascertain the cause of death.

The fourth and final aspect was the population coverage. We scored studies with nationally representative samples higher than those without, and studies that were population based, or facility based in a context where more than 95% of deliveries are attended by a skilled person were both scored higher than facility-based studies where less than 95% of were attended by a skilled birth attendant [39].

Studies scoring 1–2 were categorized as high risk of bias, 3–5 medium, and 6–8 low. Studies that scored high in the risk of bias were excluded from sub-group meta-analysis, except when we disaggregated by study risk of bias category. The scoring form is presented in Supplementary information 2.

This scale was chosen to evaluate criteria that are specific to maternal mortality identification and classification that wouldn’t be captured using standard risk of bias criteria.

Synthesis methods

Qualitative narrative synthesis was done for all studies included in this review. We extracted information about the methodology used in the studies, the factors associated with incompleteness or misclassification and any context-specific challenges that may have affected the reporting of maternal or pregnancy related deaths (as defined in ICD-11 [44].

A meta-analysis of proportions was conducted for studies reporting 1) Incompleteness (the prevalence of under-reporting of maternal and/or pregnancy related deaths in a surveillance system operating in the study area), and for 2) the sensitivity of maternal cause of death registration in a surveillance system. Studies reporting on more than one study location where they provided numbers of total and missed/misclassified deaths for each location were treated as separate observations. We stratified incompleteness by cause of maternal death, place of death and by time of death relative to delivery due to data availability.

We finally calculated the pooled percentage by which maternal deaths were under-estimated either from incomplete recording, misclassification or both. The overall under-estimation was further stratified by the income level of the country and the surveillance system investigated.

We conducted a meta-analysis for the sensitivity, but not for the specificity due to the very small number of studies reporting it.

Statistical analysis was conducted using R studio, version 4.2.0. The variance of the proportions was used to weight estimates from each study and produce pooled estimates. The proportions from each study were transformed using a Freeman-Tukey type arcsine square-root transformation, and the DerSimonian-Laird random effects method was used to combine study estimates. Estimates were stratified based on the country income level, type of surveillance system, and the quality score of the study. We did not undertake sensitivity assessment for this model.

For studies that reported on incompleteness by the cause of maternal death (direct vs indirect), timing of death relative to pregnancy, and the place of death (home vs facility) we used the same methods above to pool the estimated incompleteness across the studies for each category, where possible.

To evaluate the consistency of the meta-analysis results we report H^2 and I^2. H^2 is a measure of heterogeneity in a meta-analysis, representing the ratio of observed variance to expected variance. A value greater than 1 suggests heterogeneity. I^2 quantifies the proportion of observed variance that can be attributed to differences between studies, rather than sampling error. An I^2 of 0% indicates no heterogeneity, 25% is low, 50% is moderate, and 75% or higher is high heterogeneity.


Search results

The results of our search strategy are presented in Fig. 1. In brief, 2872 records were identified through the searches of the bibliographic databases, of which 463 were duplicates. Six additional articles were identified from reviewing the reference lists of the eligible studies, and three more were obtained from searching government websites. After title-abstract screening, 285 sources were assessed for full-text screening and 29 were identified as eligible for inclusion. Of the 29 studies, 21 reported on the number of under-reported (incomplete) or misclassified deaths, and the total number of maternal deaths identified, thus were suitable for a meta-analysis, but one study [30] was excluded because their deaths were included in the aggregate of another included study [26], making the final number of studies included in the meta-analysis (n = 20). Two studies [10, 29] were excluded from subgroup analyses as they were considered to have high risk of bias.

Fig. 1
figure 1

PRISMA flow chart summarising the search strategy

Study characteristics

The characteristics of included studies are summarised in Table 1. All the studies had a cross-sectional design, apart from one that was a nested case control study. The latter allowed for comparable data to be extracted and thus included in meta-analysis. More than two thirds of the studies were subnational (n = 19), and six studies were facility-based. More than half of the studies were considered to be of low risk of bias (n = 15), eight were medium and four were high. Most prevalent issues were the high percentage (> 10%) – or non-reporting – of deaths for which a cause of death could not be established. Furthermore, only seven studies had complete coverage (national and investigates all deaths to women of the reproductive age). Two manuscripts were in Spanish, and the remaining 27 were retrieved in English.

Table 1 General study characteristics

Fourteen studies came from high income countries, five from upper-middle-income countries, and the remaining 10 were from low and lower-middle-income countries.

In general, studies from higher income countries more frequently had national coverage and lower risk of bias and tended to compare against a CRVS. Nine out of the 19 high and upper-middle-income studies had national coverage, they mostly investigated a CRVS (14/19) and only one had a high risk of bias score. In contrast, the ten Low and Lower Middle-Income Country studies were all subnational (n = 10), three were of high bias score, and only one made a comparison against a CRVS system.

Of the 20 studies eligible for quantitative synthesis, 15 were of low risk of bias, three were medium and two were high risk (supplementary information 3). One study reported on pregnancy related deaths, and the rest reported on maternal deaths. The two high risk of bias studies – which included the one reporting pregnancy related deaths – were only included when calculating the total under-estimation prevalence, and when we disaggregated under-estimation by risk of bias score.

Methods used to assess the validity of maternal/pregnancy related death data

Thirteen studies reviewed medical/clinical records, sometimes in addition to other sources (forensic reports, death certificates, and criminal reports) to identify maternal deaths. Twelve studies linked or triangulated data from multiple sources to identify incompletely recorded or misclassified deaths. In most cases, the sources used were death certificates, birth and fetal death registers, and/or hospital records often using a unique identification number.

Three studies used the capture-mark-recapture methodology, namely, Haiti, Indonesia and the Philippines [10, 16, 32]. This method is used in public health to determine the size of populations that are difficult to identify [23], and requires four critical assumptions to be met: 1) a population is fixed,2) individuals from the two sources can be linked,3) capture in the second sample is independent of capture in the first sample; and 4) the probability of capture does not differ between individuals. The study from Indonesia used the District Health System, and interviews with village informants and health volunteers to capture all maternal deaths [32]. In Haiti, the two sources were a register data capture form and the dossier data capture form [10]. In the Philippines study, they used vital registration and the second source was a Reproductive Age Mortality Survey [16].

The three studies concluded that no single data source was able to capture all deaths. In Indonesia and the Philippines, 49% and 44% of deaths were missed by one of the two sources respectively. In Haiti, where both sources were facility-based, only about a quarter of deaths were captured by each source. The study from Haiti however could not guarantee the first third, and fourth assumptions of the capture-mark-recapture method were met in their study.

A smaller subset of studies (n = 4) used active surveillance and notification to identify maternal deaths, and only one study followed all pregnant women to identify any deaths.

In total, 19 studies conducted their own independent review of cause of death to quantify misclassification. The studies mostly reviewed cause of death using either a verbal autopsy, review of medical records, or a combination of both (Table 2).

Table 2 List of studies that conducted a review of maternal cause of death, with the review methods, and result of the review

Incompleteness of maternal death recording

There was a wide range in the extent of incompleteness, from 0 to 85% across 16 studies, and a pooled proportion of 34% (95% CI: 28–48). We found high between-study heterogeneity across all pooled estimates of incompleteness (I^2 = 91.2%; P < 0.001) (Table 3).

Table 3 Random-effects meta-analysis of pooled prevalence of incompleteness and sensitivity of maternal deaths recording stratified by individual covariates (excluding two studies for high risk of bias score)

Across the six studies that stratified by cause of death, incompleteness for indirect deaths was higher than for direct deaths (42% and 22% respectively), though confidence intervals were very wide and overlapped substantially (10 -76% and 4—48% respectively). There was evidence of high between study variability in both categories (I2: 96.1% & 87.9% respectively; P < 0.001).

Among three studies stratifying by place of death, incompleteness was higher for death that occurred at home: 75% versus 27% incompleteness, albeit with a notable overlap in the confidence intervals (95%CI 20 to 100 & 6 to 58 respectively). There was strong evidence of between study variability (I^2 = 96.4; & 96.0 respectively) (Table 3).

Deaths occurring either during pregnancy or after 24 h postpartum had a higher incompleteness (52%) compared to deaths occurring during delivery or within 24 h postpartum (25%), with some overlap in the confidence intervals. Notably, there was no evidence of between study heterogeneity in the three categories (Table 3).

Only one study stratified unregistered deaths by maternal age: this study found that deaths at the extremes of maternal age (less than 20 and above 40) were more frequently missed; half of all maternal deaths among adolescents and more than half of all maternal deaths among women aged 40 and above were under-reported, while 28% were missed in the 20–39 age group [19].

Misclassification of maternal deaths

Sensitivity ranged from 10 to 86% across four studies, and the pooled estimate of sensitivity was 61% (95% CI 37–82) (Table 3). There was only one study reporting information about specificity and found it to be high (98.9%) [22].

Reported characteristics more prone to misclassification were the cause of maternal death being indirect, extremes of maternal age, the certifier being a physician rather than a coroner or medical examiner, and the deceased being a minority ethnic group (Lin et al., 2019b) [6, 11]. However, the number of deaths in these studies was too low to determine statistical significance.

Three studies (two from USA and one from China, Taiwan Province of China) looked at the impact of adding the pregnancy checkbox to the death certificate [12] [11, 26],they found it led to the identification of more maternal deaths and therefore an increase in the MMR (from 9 to 22 in the states it was implemented in in the US,from 55 to 82 in China, Taiwan Province of China, per 100,000 live births). However, they also found the checkbox led to an increase in the number of “false positives” and hence, may over-estimate maternal mortality if it is the sole reason for classifying a death as maternal.

Overall under-estimation of maternal deaths

Across 20 studies, underestimation ranged from 0% in Iceland to 85% in Mozambique, with a pooled proportion of 37%, due to incompleteness, misclassification (false negatives), or both. Heterogeneity between studies was high (I^2 = 93.3%; P < 0.001) (see Fig. 2). We found some evidence (P = 0.05) that studies with higher risk of bias score reported higher underestimation (9% 95%CI: 0—36) compared to medium and low risk of bias studies (28% 95%CI: 12—47 and 42% 95%CI: 33–52 respectively). When excluding the two studies with a high risk of bias score, pooled underestimation rose to 39% (95%CI: 30—48).

Fig. 2
figure 2

Forest plot of the pooled under-estimated proportion of maternal deaths (from incomplete reporting, misclassification (false negatives), or both) by study quality: random effects model with DL transformation. Caption: Fig. 2 shows the proportion of maternal deaths under-estimated for each study, the pooled proportion for each risk of bias sub-group, and the overall pooled proportion for all sub-groups. The lines opposite each study indicate the 95% confidence interval of the D-L transformed proportion. The box indicates the point estimate and the size of the black box indicates the “weight” of the study, or how much the study contributes to the sub-group and overall pooled proportion. A study with a bigger box has more influence. The triangles indicate that the confidence interval of the study is wider than the x -axis scale, and the direction of the tip of the triangle indicates in which direction the confidence interval is wider than the x-axis

Under-estimation was higher in studies investigating District or Health Information Management Systems, compared to those investigating a CRVS or an HDSS (49%, 39% and 32% respectively), however there was no evidence of between group difference (p = 0.44), with a notable overlap in the 95% CI (Table 4). The under-estimation proportion was also higher among studies from low or lower middle-income settings (45%) compared to those from high/upper middle-income countries (36%), also with no evidence of between group heterogeneity (p = 0.36) (Table 4). Finally, under-estimation in studies where the mid-year of reporting is after 2010 was lower than those with mid-year between before 2000 and between 2000–2010, again with no evidence of heterogeneity between groups, and notable overlap in the confidence intervals.

Table 4 Pooled proportion of under-estimated maternal deaths, disaggregated by study covariates

Context specific challenges in classifying or registering maternal deaths

Broadly-speaking, there were three groups of challenges to recording maternal deaths. The first was lack of documentation and/or inadequate storage of medical records. One study from Haiti reported that out of 373 deaths to women of reproductive age, there was not sufficient information to determine cause of death for 56.3% of them, due to lack of documentation, or because medical records were damaged in storage [10]. In Switzerland, they could not determine cause of death for seven deaths (of 117 total) due to paucity of information available for reviewers [24].

Second was, challenges related to stigma, such as cultural beliefs about pregnancy and/or its termination. In Tanzania, induced abortion is illegal, and researchers identified this as a limitation to capturing resulting deaths [27].

Third, some studies identified issues arising from how the process of death notification/recording was organised. In, there were two separate electronic systems for recording deaths, and they were not sufficiently synchronised leading to one system not including any questions about pregnancy status while the second does [25]. Indonesia relies on a village midwife covering an area, urban or rural, for measuring maternal deaths, and with urban areas being more populated, the incompleteness was higher in urban areas compared to rural. Additionally, urban areas had more private clinics which may have led to midwives not being able to capture all deaths [32].


Our systematic review found substantial issues in the reporting of maternal deaths, with overall around a third of deaths not recorded in the studies examining different types of routine data systems. Sensitivity of maternal cause of death reporting was found to be 59%, and specificity from only one observation was 98%. These results align with the estimates of sensitivity and specificity produced by the Maternal Mortality Estimation Interagency Group’s Bayesian misclassification model (Sensitivity of 58% and Specificity of ~ 99%) [41]. The level of incompleteness and misclassification varied significantly between different contexts but was present across all settings. The combination of incompleteness and misclassification resulted in maternal deaths being underestimated by nearly 40%, though this is likely an under-estimate, since not all studies investigated both incompleteness and misclassification.

We found significant heterogeneity across most of the result, with the exception of incompleteness when disaggregated by cause of death (direct or indirect) and timing of death relative to delivery.

A higher proportion of maternal deaths were under-estimated in studies carried out in low and lower-middle-income countries (45%), compared to high and upper-middle- income countries (36%). A Health Management Information System appeared to miss more deaths (49%) than a CRVS or an HDSS (36% and 33% respectively). Additionally, studies with a low risk of bias reported less under-estimation than those with a high risk of bias score; these observations likely have common route causes.

We found the recording of indirect maternal deaths, deaths occurring at home and death further away from delivery (during pregnancy or after 24 h postpartum) to be less complete than their respective counterparts. The reasons for the latter two can be related to the prevalence of facility-based births in a country, coupled with the coverage of the surveillance systems to deaths which may occur outside of health facilities. Further, early pregnancy deaths maybe missed because the woman’s pregnancy status may not be known, or due to cultural barriers in countries where extramarital pregnancy is socially stigmatised. Deaths after 24 h postpartum will be more likely to happen at home or away from the obstetrics ward and thus missed.

Our finding of indirect maternal death reporting being less complete than for direct deaths requires attention, given the current trend of increasing maternal age and obesity globally, which may lead to an increase in the proportion of maternal deaths that are indirect. This trend forms the basis of the obstetric transition, in which countries either have, or are shifting from high to lower maternal mortality, and from a predominance of direct causes to indirect ones [36].

When the cause of death is not directly related to pregnancy, it can occur in non-obstetric departments/referral centres and likely outside the intrapartum period. Individuals classifying the cause of death, hence, may need to be prompted to check if any Women of Reproductive Age is pregnant or postpartum at the time of death via a pregnancy checkbox in the death certificate. The usefulness of this intervention was demonstrated in our review, where studies reporting on the validity of the pregnancy checkbox found that it did indeed lead to the identification of more maternal deaths.

The pregnancy checkbox is also present in the WHO’s International Form of Medical Certificate of Cause of Death, last updated in 2016, to guarantee the recording of minimum information required to code cause of death consistently across countries [44]. The WHO certificate has been adopted in some countries, but still not routinely used in others, and where it is used, it is sometimes not filled correctly or to completion [18]. Needless to say, the pregnancy check-box should not be the sole reason for classifying a death as maternal, to avoid over-estimation.

We found fewer studies reporting on misclassification compared to incompleteness, likely because identifying misclassified deaths requires additional information on the correct cause of death, and a wider sample frame. This was particularly true for specificity; studies tended to investigate deaths with a maternal cause of death or with evidence of pregnancy, allowing them to identify true positives and false positives, and some false negatives. They rarely however identified true negatives, as these would, in addition to correct cause of death, require investigating all deaths to women of the reproductive age, including those with no evidence of pregnancy. As a result, we could not evaluate over-estimation of maternal deaths. We argue, however, that in most contexts, under-estimation of maternal mortality is often more prevalent than over-estimation [3, 33], as demonstrated by the low sensitivity of maternal deaths compared to sensitivity both in this review and MMEIG Bayesean model [31].

Our review had a number of strengths. Firstly, we implemented a comprehensive search strategy, with broad search terms and no language restrictions to insure the identification of a good number of eligible studies. The literature was also supplemented by searching of government websites and reference lists, and also using CRVS maternal mortality estimates from the WHO mortality database to provide a comparator for studies which did not provide a comparison in their report. Secondly, the narrative synthesis we conducted enabled us to contextualise and better interpret the results obtained from the quantitative synthesis and meta-analysis. Thirdly, we were able to allow for more flexibility in estimating incompleteness and misclassifications in systems that would not be evaluated in the current MMEIG maternal mortality estimates, due to representativeness concerns. This allowed to us to compare the quality of reporting between different surveillance systems.

However, there were also some important limitations. First, the low number of studies reporting on incompleteness and more so on misclassification, especially in lower income settings. The search conducted was limited to studies indexed between 2016 – 2021, which may have reduced the number of potential studies. This was mitigated slightly by searching the reference lists of identified studies, allowing us to include older relevant sources. This limitation is further exacerbated by the low number of maternal deaths as an event, hindering our ability to make conclusive inferences on when and which deaths are incomplete or misclassified.

Secondly, we noted substantial heterogeneity in contexts and surveillance systems. These systems in general serve similar purposes, but how they are organised in a given country/context, their inadequacies, and challenges vary substantially. This is clear from the high estimates of heterogeneity (> 60%) throughout most of our results. This was also demonstrated in our synthesis of contextual challenges reported in the literature, where some studies identified issues specific to the system or country they validated. The heterogeneity means that our results cannot be generalised and must be interpreted with caution.

The methodologies employed in the included studies can provide useful indications on how to improve identification and classification of maternal deaths. One common technique was the linking of records from several sources to identify missed maternal deaths. The importance of this approach was highlighted in the three studies using the capture-mark-recapture methodology, where all three studies found not one single source was able to capture all deaths [16,33, 10] Another important element is the expert review of cause of death, which is an essential element of confidential inquiries into maternal deaths. The aim of confidential enquiries is, in addition to identifying an accurate cause of death, to highlight missed opportunities and to prevent the reoccurrence of similar deaths in the future [26]. Finally, we highlight the importance of investigating all deaths to WRA, and not only those with evidence of pregnancy.

We consider a CRVS to be superior to other surveillance systems investigated here, primarily due to its national-level coverage. Nonetheless, DHIS and HDSS systems can potentially provide valuable information on maternal mortality in LMIC, if data emerging from them is interpreted in the light of their limitations. HDSS data seem to be of relatively better quality than DHIS, however they are often sub-national, covering a relatively small area, and hence do not provide a national estimate of maternal mortality. HDSS data is a valuable resource that has been used to validate different mortality estimation methods in low- and middle-income settings, and can be used to inform and monitor the development and quality of routine sources, including a CRVS.

In conclusion, maternal mortality is substantially under-estimated in almost all contexts, but to varying degrees. Efforts to implement well-functioning CRVS systems are key to ensuring that all maternal deaths in a country are recorded and accurately classified. Where CRVS is instituted and functional, countries should continuously evaluate the completeness and accuracy of maternal deaths recording in the CRVS, perhaps through confidential enquiries. The WHO has produced guidance on improving the measurement of maternal mortality that is aimed at national level professionals [45]. This guidance provides valuable insights that address the measurement challenges identified in this review.

Availability of data and materials

Data used in the quantitative analysis is publicly available from reviewed studies manuscripts, and from the WHO mortality database.


  1. Abalos E, Duhau M, Escobar P, Fasola ML, Finkelstein JZ, Golubicki JL, Krupitzki H, Marconi É, Santoro A, Vinacur J, EstudioColaborativoArgentino EORMM. Omisión de registros de causas maternas de muerte en Argentina: estudio observacional de alcance nacional. Rev Panam Salud Pública. 2019;43(1):1–10.

    Article  Google Scholar 

  2. Abouchadi S, Zhang WH, De Brouwere V. Underreporting of deaths in the maternal deaths surveillance system in one region of Morocco. PLoS One. 2018;13(1):1–15.

    Article  CAS  Google Scholar 

  3. Agampodi S, Wickramage K, Agampodi T, Thennakoon U, Jayathilaka N, Karunarathna D, Alagiyawanna S. Maternal mortality revisited: The application of the new ICD-MM classification system in reference to maternal deaths in Sri Lanka. Reprod Health. 2014;11(1):1–5.

    Article  Google Scholar 

  4. AIHW. Australia’s mothers and babies, Maternal deaths - Australian Institute of Health and Welfare.2020

  5. Anwar J, Torvaldsen S, Sheikh M, Taylor R. Under-estimation of maternal and perinatal mortality revealed by an enhanced surveillance system: Enumerating all births and deaths in Pakistan. BMC Public Health. 2018;18(1):1–14.

    Article  Google Scholar 

  6. Baeva S, Saxton DL, Ruggiero K, Kormondy ML, Hollier LM, Hellerstedt J, Hall M, Archer NP. Identifying maternal deaths in texas using an enhanced method, 2012. Obstet Gynecol. 2018;131(5):762–9.

    Article  PubMed  Google Scholar 

  7. Berdzuli N, Lomia N, Staff AC, Kereselidze M, Lazdane G, Jacobsen AF. Maternal mortality in Georgia: Incidence, causes and level of underreporting: a national reproductive age mortality study 2014. Int J Women’s Health. 2020;12:277–86.

    Article  Google Scholar 

  8. Bess Constantén S, Martínez Morales MÁ, Fernández Viera MR, Mazorra Ramos V, Alonso Alomá I, López Nistal LM, Gran Álvarez MA, Álvarez Fumero R, Piloto Padrón M. Calidad de las estadísticas de mortalidad materna en Cuba, 2013. Rev Panam Salud Pública. 2018:1–9.

  9. Boutin A, Cherian A, Liauw J, Dzakpasu S, Scott H, Van den Hof M, Cook J, Blake J, Joseph KS. Database Autopsy: An Efficient and Effective Confidential Enquiry into Maternal Deaths in Canada. J Obstet Gynaecol Can. 2021;43(1):58-66.e4.

    Article  PubMed  Google Scholar 

  10. Boyd AT, Hulland EN, Grand’Pierre R, Nesi F, Honoré P, Jean-Louis R, Handzel E. Use of Rapid Ascertainment Process for Institutional Deaths (RAPID) to identify pregnancy-related deaths in tertiary-care obstetric hospitals in three departments in Haiti. BMC Pregnancy Childbirth. 2017;17(1):1–10.

    Article  Google Scholar 

  11. Catalano MA, Davis NL, Petersen EE, Harrison C, Kieltyka L, You MM, Elizabeth J, Ewing AC, Callaghan WM. Pregnant? Validity of the Pregnancy Checkbox on Death Certificates in Four States, and Characteristics Associated with Pregnancy Checkbox Errors. Am J Obstet Gynecol. 2021;222(3):1–15.

    Article  Google Scholar 

  12. Davis NL, Hoyert DL, Goodman DA, Hirai AH, Callaghan WM. Contribution of maternal age and pregnancy checkbox on maternal mortality ratios in the United States, 1978–2012. Am J Obstet Gynecol. 2017;217(3):352.e1-352.e7.

    Article  PubMed  Google Scholar 

  13. Deneux-Tharaux C, Berg C, Bouvier-Colle MH, Gissler M, Harper M, Nannini A, Alexander S, Wildman K, Breart G, Buekens P. Underreporting of pregnancy-related mortality in the United States and Europe. Obstet Gynecol. 2005;106(4):684–92.

    Article  PubMed  Google Scholar 

  14. DistillerSR: Systematic Review Software version 2.35. 2021. (n.d.). Retrieved May 3, 2022, from

  15. Donati S, Maraschini A, Lega I, D’Aloja P, Buoncristiano M, Manno V, Alberico S, Antonelli A, Asole S, Basevi V, Cetin I, Chiodini P, Dardanoni G, Di Lallo D, Dubini V, Germinario C, Giangreco M, Gnaulati L, Loverro G, Voller F. Maternal mortality in Italy: results and perspectives of record-linkage analysis. Acta Obstet Gynecol Scand. 2018;97(11):1317–24.

    Article  PubMed  Google Scholar 

  16. Garces RG, Sobel HL, Pabellon JAL, Lopez JM, De Quiroz Castro M, Nyunt-U S. A comparison of vital registration and reproductive-age mortality survey in Bukidnon, Philippines, 2008. Int J Gynecol Obstet. 2012;119(2):121–4.

    Article  Google Scholar 

  17. Gülmezoglu AM, Say L, Betrán AP, Villar J, Piaggio G. WHO systematic review of maternal mortality and morbidity: Methodological issues and challenges. BMC Med Res Methodol. 2004;4:1–8.

    Article  Google Scholar 

  18. Hoffman RA, Venugopalan J, Qu L, Wu H & Wang MD. (n.d.). Improving Validity of Cause of Death on Death Certificates. Computational Biology and Health Informatics, 18, pages.

  19. Horon IL. Underreporting of maternal deaths on death certificates and the magnitude of the problem of maternal mortality. Am J Public Health. 2005;95(3):478–82.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Institute of Health Metrics and Evaluation. Global Burden of Disease (GBD). 2019.

  21. Kodan LR, Verschueren KJC, van Roosmalen J, Kanhai HHH, Bloemenkamp KWM. Maternal mortality audit in Suriname between 2010 and 2014, a reproductive age mortality survey. BMC Pregnancy Childbirth. 2017;17(1):1–9.

    Article  Google Scholar 

  22. Kodio B, Bernis L. De, Ba M, Ronsmans C, Pison G. Levels and causes of maternal mortality in Senegal. Trop Med Int Health. 2002;7(6):499–505.

    Article  PubMed  Google Scholar 

  23. Laska EM. The use of capture-recapture methods in public health. Bull World Health Organ. 2002;80(11):845.

    PubMed  PubMed Central  Google Scholar 

  24. Laura P, Roland Z & C, Q. L. K. Maternal mortality in Switzerland 2005 – 2014. 2020:1–8.

  25. Lin CY, Tsai PY, Wang LY, Chen G, Kuo PL, Lee MC, Lu TH. Changes in the number and causes of maternal deaths after the introduction of pregnancy checkbox on the death certificate in Taiwan. Taiwan J Obstet Gynecol. 2019;58(5):680–3.

    Article  PubMed  Google Scholar 

  26. MBRRACE-UK. Lessons learned to inform maternity care from the UK and Ireland Confidential Enquiries into Maternal Deaths and Morbidity 2016–18. 2020;31(1).

  27. Mswia R, Lewanga M, Moshiro C, Whiting D, Wolfson L, Hemed Y, Alberti KGMM, Kitange H, Mtasiwa D, Setel P. Community-based monitoring of safe motherhood in the United Republic of Tanzania. Bull World Health Organ. 2003;81(2):87–94.

    PubMed  PubMed Central  Google Scholar 

  28. Murray CJL, Rajaratnam JK, Marcus J, Laakso T, Lopez AD. What can we conclude from death registration? Improved methods for evaluating completeness. PLoS Med. 2010;7(4):1–3.

    Article  Google Scholar 

  29. Mwaniki BK, Edwards JK, Kizito W. How complete were maternal death reviews in Central Kenya 2015–2018? Afr J Reprod Health. 2020;24(4):122–31.

    Article  PubMed  Google Scholar 

  30. O’Hare MF, Manning E, Corcoran P, Greene RA, Ireland MDE. Confidential Maternal Death Enquiry Ireland, Report for 2016–2018. 2020(4):1–4). Cork: MDE Ireland. ISSN 2009–7298

  31. 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 year. Stat Med. 2022;41(14):2483–96.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Qomariyah SN, Sethi R, Izati YN, Rianty T, Latief K, Zazri A, Besral Bateman M, Pawestri EA, Ahmed S, Achadi EL. No one data source captures all: a nested case-control study of the completeness of maternal death reporting in Banten Province, Indonesia. PLoS One. 2020;15(5):1–13.

    Article  CAS  Google Scholar 

  33. Said A, Malqvist M, Pembe AB, Massawe S, Hanson C. Causes of maternal deaths and delays in care: Comparison between routine maternal death surveillance and response system and an obstetrician expert panel in Tanzania. BMC Health Serv Res. 2020;20(1):1–14.

    Article  Google Scholar 

  34. Sesmero JR de M, Cacho PM, Solano AM, Feu JMP, Gómez MG, Prieto AP, González NLG, Vicens JML. Maternal mortality in Spain from 2010 to 2012 : Results of Spanish society of OBSTETRICIA Y GINECOLOG ´ ˜ a en el periodo 2010-2012 : Prog Obstet Ginecol, January 2018. 2016.

  35. Songane FF, Bergström S. Quality of registration of maternal deaths in Mozambique: A community-based study in rural and urban areas. Soc Sci Med. 2002;54(1):23–31.

    Article  PubMed  Google Scholar 

  36. Souza JP, Tunçalp Ö, Vogel JP, Bohren M, Widmer M, Oladapo OT, Say L, Gülmezoglu AM, Temmerman M. Obstetric transition: the pathway towards ending preventable maternal deaths. BJOG. 2014;121(Suppl):1–4.

    Article  PubMed  Google Scholar 

  37. Vangen S, Bødker B, Ellingsen LIV, Saltvedt S, Gissler M & Geirsson RT. Maternal deaths in the Nordic countries. 2017;96:1112–1119

  38. World Health Organization. (n.d.-a). Births attended by skilled health personnel (%). 2022. Retrieved June 9, 2022, from

  39. World Health Organization. (n.d.-b). WHO Mortality Database . Retrieved April 28, 2022, from

  40. World Health Organization. Strategies toward ending preventable maternal mortality ( EPMM ). 2015.

  41. World Health Organization. Maternal mortality: level and trends 2000 to 2017. In Sexual and Reproductive Health. 2019.

  42. World Health Organization. Score for health data technical package: Essential Interventions. 2020.

  43. World Health Organization. Ending Preventable Maternal Mortality (EPMM) - A renewed focus for improving maternal and newborn health and well-being (Issue April). 2021a.

  44. World Health Organization. International Classification of Diseases 11th revision (ICD-11). 2021b.

  45. World Health Organization. Maternal Mortality measurement: Guidance to improve national reporting. 2022

  46. World Health Organization. Trends in maternal mortality 2000 to 2020: estimates by WHO, UNICEF, UNFPA, World Bank Group and UNDESA/Population Division. 2023.

  47. Wu TP, Huang YL, Liang FW, Lu TH. Underreporting of maternal mortality in Taiwan: a data linkage study. Taiwan J Obstet Gynecol. 2015;54(6):705–8.

    Article  PubMed  Google Scholar 

  48. Zakariah AY, Alexander S, Van Roosmalen J, Buekens P, Kwawukume EY, Frimpong P. Reproductive age mortality survey (RAMOS) in Accra. Ghana Reproductive Health. 2009;6(1):1–6.

    Article  Google Scholar 

Download references


We acknowledge the contributions of Farida Abudulai (FA) who acted as the second reviewer for the current systematic review. We would also like to acknowledge Kavita Kothari from the WHO library, for managing the literature search.


This study was funded by USAID (GHA-G-00–09-00003) and the UNDP-UNFPA-UNICEF-WHO-World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP). The funder had no role in this review.

Author information

Authors and Affiliations



S.A and JC planned the methodology and analysis. S.A conducted the analysis, prepared figures and wrote the main manuscript text. All authors reviewed the manuscript.

Corresponding author

Correspondence to Sahar M. A. Ahmed.

Ethics declarations

Ethics approval and consent to participate

Not applicable – systematic review.

Consent for publication

Not applicable – systematic review.

Competing interests

The authors declare no competing interests.

© World Health Organization 2022

The authors are staff members of the World Health Organization. The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the decisions or policies of the World Health Organization.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Supplementary information 1.

Search terms used in the bibliographic search

Additional file 2: Supplementary information 2.

Study risk of bias scoring form

Additional file 3: Supplementary information 3.

Study risk of bias score for included studies

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ahmed, S.M.A., Cresswell, J.A. & Say, L. Incompleteness and misclassification of maternal death recording: a systematic review and meta-analysis. BMC Pregnancy Childbirth 23, 794 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: