Menu
Menu
Menu
Menu
Background Sustainable Development Goal (SDG) 3.1 target is to reduce the global maternal mortality ratio (MMR) to less than 70 maternal deaths per 100 000 live births by 2030. In the Ending Preventable Maternal Mortality strategy, a supplementary target was added, that no country has an MMR above 140 by 2030. We conducted two cross-sectional reproductive age mortality surveys to analyse changes in Zimbabwe’s MMR between 2007-2008 and 2018-2019 towards the SDG target. Methods We collected data from civil registration, vital statistics and medical records on deaths of women of reproductive ages (WRAs), including maternal deaths from 11 districts, randomly selected from each province (n=10) using cluster sampling. We calculated weighted mortality rates and MMRs using negative binomial models, with 95% CIs, performed a one-way analysis of variance of the MMRs and calculated the annual average reduction rate (ARR) for the MMR. Results In 2007-2008 we identified 6188 deaths of WRAs, 325 pregnancy-related deaths and 296 maternal deaths, and in 2018-2019, 1856, 137 and 130, respectively. The reproductive age mortality rate, weighted by district, declined from 11 to 3 deaths per 1000 women. The MMR (95% CI) declined from 657 (485 to 829) to 217 (164 to 269) deaths per 100 000 live births at an annual ARR of 10.1%. Conclusions Zimbabwe’s MMR declined by an annual ARR of 10.1%, against a target of 10.2%, alongside declining reproductive age mortality. Zimbabwe should continue scaling up interventions against direct maternal mortality causes to achieve the SDG 3.1 target by 2030.
Findings of the 2007–2008 Zimbabwe Maternal and Perinatal Mortality Survey (ZMPMS) prompted the Zimbabwe Ministry of Health and Child Care (MoHCC) to implement a raft of interventions to reduce maternal mortality. A maternal and neonatal health roadmap was developed to address the direct and indirect causes of maternal mortality.19 Family planning services, prevention of mother-to-child transmission (PMTCT) of HIV services and community mobilisation for safe motherhood were scaled up through community health workers. Basic and comprehensive emergency obstetric and newborn care was rolled out in primary care and secondary/tertiary health facilities, respectively.20–22 The UK’s Royal College of Obstetricians and Gynaecologists and the Liverpool School of Tropical Medicine and Bristol University conducted training of trainers for 120 doctors and nurse–midwives who trained over 700 other doctors and nurses nationwide on the management of obstetrical and neonatal emergencies.20 Maternity waiting homes, which the MoHCC started establishing in the 1980s, were expanded in different districts, allowing women to stay at maternity facilities from the third trimester until delivery,23–26 increasing access to antenatal care and reducing home deliveries. The government developed guidelines for maternal and perinatal death surveillance and response system,27 and instituted maternal and perinatal death audits.21 Through a US$235 million health transition fund (2012–2015) and US$682 million health development fund (2016–2020), the government introduced free maternity services and doctors’ and nurses’ retention allowances in the rural provinces and supplied health facilities with essential commodities, among other initiatives, to improve maternity, neonatal and child healthcare.11 12 28 29 The 2007–2008 survey found that HIV was the major cause of maternal mortality, contributing 26% of maternal deaths; meaning that HIV interventions would significantly impact maternal mortality. In 2007–2008, Zimbabwe had high adult (15–49 years) HIV mortality.30–33 Antiretroviral therapy (ART) roll-out was in the early phases at this time.34 ART was available in only 5.2% (86/1643) of the health facilities by December 2007 and 17% (282/1643) by December 2008; mostly in secondary and tertiary hospitals that are less accessible to communities.34 In 2008 only 24% (148 144/596 965) of individuals needing ART received it and the need for ART was defined by a CD4 count below 350 cells/µ of blood at the time.34 35 However, HIV programmes received significant funding over the years, exceeding US$400 million annually.36 37 Consequently, ART was rolled out to 91% (1566/1722) of all health facilities in the country by 2017, and in 2019, 97% of adults with known HIV-positive status received ART.38 39 Adult HIV mortality substantially declined from an estimated 83 000 deaths in 2009 to 14 000 in 2018.33 35 36 In PMTCT, Zimbabwe rolled out WHO 2010 (‘Option A’) and 2013 (‘Option B+’) guidelines.40 41 Option ‘A’ was rolled out to 85% (1320/1560) MNCH facilities in 9 months and Option ‘B+’ to 88% (1385/1560) facilities in 5 months.41 All HIV-infected pregnant and breastfeeding women were initiated on lifelong ART, under the ‘Option B+’ guidelines, irrespective of disease stage. The combined impact of the ART and PMTCT programmes was that by 2019, 88% of adult women (15–49 years) living with HIV had known HIV-positive status, of which 98% were on ART.39 In 2018, 94% of HIV-positive pregnant women received ART for PMTCT.36 Without ART, pregnant women can die from AIDS-related complications including pneumonia, tuberculosis and meningitis. With a weakened immune system, HIV-infected pregnant women also have a higher risk of mortality from pregnancy-related sepsis, haemorrhage and other direct causes.42–44 The ART and PMTCT interventions should have contributed to the 91% reduction in HIV-related maternal mortality found in the causes of death analysis from this study.18 A before-and-after analysis was performed using data from the two RAMOS conducted in 2007–2008 and 2018–2019. The surveys collected births and deaths among women in the reproductive ages (WRAs) 12–49 years, including maternal deaths, to analyse changes in Zimbabwe’s MMR in the context of the interventions described. The sampling method was designed for the first survey in 2007 and maintained in the second survey for comparability of the study findings. Two-stage cluster sampling was applied in the two surveys. In the first stage, the study population was clustered into the 10 provinces of the country, and one district was simple-randomly selected from each province. Two districts were selected from Harare province because of its large population and that several provinces refer complicated maternal cases to two central hospitals in the province. In the second stage, all births and deaths among WRAs 12–49 years from the selected districts were included in each survey. Sample sizes of births required to calculate the MMR were calculated for each survey. In the first step, simple random samples were calculated using the Wald Test for a one-sample proportion (treating the MMR as a proportion).45 46 The recent MMRs from the Zimbabwe Demographic and Health Survey (ZDHS) of 2005–200647 and 2015–2016 respectively were the expected proportions.48 Power of 80% and the z-value for two-sided 95% CI, continuity correction for normal approximation of the expected proportion and 2.5% error margin for the alternative hypothesis of MMR outside the 95% CI of the expected proportion were applied. In the second step, the random sample sizes were multiplied by the design effect (DE) to obtain the final sample sizes. The DE for the 2007–2008 survey was calculated from the pilot study and the DE for the 2018–2019 survey was calculated from the 2007–2008 survey (online supplemental table S2). The detailed procedure for calculating the sample size is also described in the study protocol.17 The two surveys required sample sizes of 45 000 and 71 500 births respectively. Figure 1 shows the 11 study districts. Nkulumane (Bulawayo province), Western and South-Eastern districts (Harare province) are urban districts, while Mutare (Manicaland province), Bindura (Mashonaland East province) and Kwekwe (Midlands province) are semiurban districts and the rest are rural districts. Map of study districts for Zimbabwe maternal and perinatal mortality study 2007–2008 and 2018–19. For each death (including pregnancy-related deaths) among WRAs, we collected location information (province, district and place of residence—urban or rural), age (in completed years), pregnancy status (pregnant or not) and cause of death (as stated on medical records and death certificates). For pregnancy-related deaths, we also collected information on parity, gestational age, antenatal care, pregnancy and delivery complications, referrals to other health institutions, delivery outcome and place of death (home or institutional). The first survey collected data for the period 1 May 2007 to 15 June 2008 and the second for the period 1 May 2018 to 15 June 2019. Data for the first survey were collected prospectively during the study period and data for the second survey were collected retrospectively from 1 May 2020 to 31 July 2020 and from 3 May 2021 to 20 July 2021. The 2007–2008 survey collected data from civil registration and vital statistics (CRVS) records at the government Registrar General (RG)’s offices, health facilities and the community. Study nurse–midwives trained on the study protocol and supervised by the investigators collected the data. They collected data on deaths of WRAs from RG’s offices and data for live births and pregnancy-related deaths in health facilities and the community for eligible women. Data for health facility deaths occurring during pregnancy or post-delivery were collected from medical records in the labour ward, theatre, high dependency and intensive care units, medical and surgical female wards, mortuaries and police posts. The survey was approved to collect identified data, hence, data collectors used the women’s personally identifying information (PII)—name, address, age and national identity numbers, to link individual women across these records. Deliveries and deaths of WRAs occurring outside health institutions were enumerated in the community. Village health workers and village heads recorded them in study-provided register books. The research nurses followed up on every delivery and death recorded in the registers and interviewed the mothers (for births) and relatives (husband, mother, sister, aunty) for deaths, using study questionnaires. They collected additional data (dates when the death occurred, pregnancy status and signs and symptoms of sickness at death) for deaths of WRAs using a verbal autopsy (VA) form adapted from WHO.49 The data collectors used the women’s PII to cross-check and de-duplicate deaths identified in health facilities, the community and CRVS records. A group of six obstetrician–gynaecologists reviewed the data collection and VA forms for all pregnancy-related deaths, classified the deaths as maternal and non-maternal and assigned the causes of death. In the 2018–2019 survey another group of nurse–midwives collected the data from the RG’s offices and health facility records (as in the 2007–2008 survey) and maternal death notification forms at the MoHCC’s district, provincial and national reproductive health offices. They cross-checked and de-duplicated the deaths using PII (as above). Live births data were collected from the MoHCC’s District Health Information System V.2 (DHIS2), a database system for health indicators.50 Nurses in the MNCH units recorded all institutional and home births presented at health facilities in birth registers. The nurses summarised the data on a standard monthly report form and submitted the reports to health information officers who entered the data into the DHIS2 database. Given the health system structure where deliveries occur in private and public health institutions, and health centres (rural and urban) refer complicated maternal cases to district hospitals, which refers to provincial hospitals, which also refers to central hospitals in Harare and Bulawayo, we counted the live births for women referred from the study districts to private, provincial and central hospitals, and added them to the DHIS2 births. We collected population data for WRAs for the study districts from the Zimbabwe National Statistics Agency (ZimStat).51 During the collection of 2018–2019 data, 2007–2008 deaths were verified in the CRVS records at the RG’s offices and health facility records (same sources as above). All questionnaires and VA forms for 2007–2008 deaths were reviewed by a new group of obstetrician–gynaecologists in 2020, the causes of death re-assigned and maternal deaths confirmed using the International Classification of Diseases V.10 manual for deaths during pregnancy, childbirth and puerperium (ICD-10 MM). The database for 2007–2008 deaths was cleaned for completeness and accurate data entry. The 2018–2019 data were collected in two rounds to ensure that all deaths in the source records were identified and correctly captured in the study. In Zimbabwe legislation regulates CRVS and medical records. The birth and death registration act mandates the registration and issuance of certificates for all births and deaths.52 The legislation mandates parents, health workers at institutions where the birth occurs or community leaders (for community births) to notify the RG’s office of the birth. Similarly, for persons who died at home, relatives or village heads are required by the law to notify the RG’s office, for the creation of a death record and issuance of a death certificate. Deaths that occur in health institutions get a medical death certificate signed by the doctor or nurse who attended the death. Home or community deaths attended to by the police are taken to hospitals where a doctor conducts a postmortem and issues another medical death certificate, stating the cause of death. The medical death certificates are deposited at the local RG’s office, where a record is created, and a civil death certificate is issued. The RG’s offices file birth and death records by date and year of registration and store them in secure record rooms. The public health act guides the recording, collection, storage, access, use, protection and confidentiality of health data.53 Reports presenting estimates of the MMR for Zimbabwe 2000–2019 were reviewed, including the ZDHS for 2000, 2005–2006, 2010–2011, 2015–201647 48 54; the Multiple Cluster Indicator Survey (MICS) in 2014 and 201955 56; population census in 2002 and 2012,54 57 the Maternal and Perinatal Death Surveillance Response in 2018 and 2019,27 and the United Nation’s Maternal Mortality Estimation Inter-Agency Group (MMEIG) MMR estimates for 2000 to 2017.6 58–60 WRAs are women aged 15–49 years, but children aged 12–14 years were included as some gave birth and died from maternal causes at this age. Pregnancy-related deaths were female deaths in which the woman was pregnant or within 42-days of termination of pregnancy, irrespective of the cause of death. Maternal deaths were deaths of women during pregnancy or within 42-days of pregnancy termination, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management, but not from accidental or incidental causes.61 We adjusted the number of maternal death for 2018–2019 for missed community deaths, using the 2007–2008 data as a standard, which comprehensively collected community deaths.17 62 The number of community deaths missed in 2018–2019 was estimated by equating the proportion of community to institutional deaths in the two surveys (online supplemental table S3). We triangulated the total number of live births identified with the expected number of live births and estimated pregnancies obtained from ZimStat,51 to assess the completeness of the former. We calculated the expected number of live births by multiplying the 2018 populations of WRAs for each district with general fertility rates (GFRs) from the 2019 MICS survey,55 using the rural or urban GFR as applicable to each district. WHO recommends that when calculating MMRs using data from CRVS or health records, the live births must be corrected for missed births.63 As such, we calculated correction factors (expected/identified births), ranging from 1.0 to 1.3 (online supplemental table S4), and used them to correct the number of live births for each district for missed births. We performed a before-and-after analysis of mortality using data from the two surveys calculating mortality incidence rates (IRs) of WRAs (number of deaths/1000 women) for each survey and the incidence rate ratios (IRRs) (mortality rate 2018–2019/mortality rate 2007–2008) and 95% CIs by district, age group and totally. Similarly, we computed the MMRs (number of maternal deaths/100 000 live births) and their IRRs for each district and totally. IRRs applied because the total person-years for each district cluster equalled the district population in 1 year. Half person-years were assigned to women who died during the year. We calculated the IRs using negative binomial models in Stata (V.17.0) immediate commands,64 treating the two surveys as cohorts, to use IRRs to estimate the magnitude of change in mortality levels between the two surveys. Stata immediate commands were employed because of the aggregate live births data in the IR denominators. Overall the IRs, MMRs and IRRs were weighted using the location variable (district) (online supplemental tables S5 and S6), to account for the clustering of the deaths within districts in the pooled samples. MMR 95% CIs were calculated using sampling errors (SEs) for the location variable. We calculated the SEs using the Jackknife repeated replication method used in the DHS48 (online supplemental table S5). We also performed repeated measures’ one-way analysis of variance (ANOVA),65 for before-and-after comparison to confirm the statistical significance of the changes in the MMRs (online supplemental table S7). Using the WHO online calculator,66 we calculated the annual ARR for the country’s MMR from the study and the ARR needed to achieve the SDG target of 140 maternal deaths per 100 000 live births by 2030 from a 2015 ZDHS baseline MMR of 651 and 2019 MICS estimate of 46255 (online supplemental table S8). There was no patient or public involvement in this study.
N/A
