Quality and outcomes of maternal and perinatal care for 76,563 pregnancies reported in a nationwide network of Nigerian referral-level hospitals

Background: The WHO in collaboration with the Nigeria Federal Ministry of Health, established a nationwide electronic data platform across referral-level hospitals. We report the burden of maternal, foetal and neonatal complications and quality and outcomes of care during the first year. Methods: Data were analysed from 76,563 women who were admitted for delivery or on account of complications within 42 days of delivery or termination of pregnancy from September 2019 to August 2020 across the 54 hospitals included in the Maternal and Perinatal Database for Quality, Equity and Dignity programme. Findings: Participating hospitals reported 69,055 live births, 4,498 stillbirths and 1,090 early neonatal deaths. 44,614 women (58·3%) had at least one pregnancy complication, out of which 6,618 women (8·6%) met our criteria for potentially life-threatening complications, and 940 women (1·2%) died. Leading causes of maternal death were eclampsia (n = 187,20·6%), postpartum haemorrhage (PPH) (n = 103,11·4%), and sepsis (n = 99,10·8%). Antepartum hypoxia (n = 1455,31·1%) and acute intrapartum events (n = 913,19·6%) were the leading causes of perinatal death. Predictors of maternal and perinatal death were similar: low maternal education, lack of antenatal care, referral from other facility, previous caesarean section, latent-phase labour admission, operative vaginal birth, non-use of a labour monitoring tool, no labour companion, and non-use of uterotonic for PPH prevention. Interpretation: This nationwide programme for routine data aggregation shows that maternal and perinatal mortality reduction strategies in Nigeria require a multisectoral approach. Several lives could be saved in the short term by addressing key predictors of death, including gaps in the coverage of internationally recommended interventions such as companionship in labour and use of labour monitoring tool. Funding: This work was funded by MSD for Mothers; and UNDP/UNFPA/ UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), a co-sponsored programme executed by the World Health Organization (WHO). This cross-sectional study captured maternal and perinatal data in a nationwide network of 54 consenting tertiary hospitals, serving as referral centres for other health facilities in their environs (48 publicly funded and 6 privately funded), across the six geopolitical zones of Nigeria (Northcentral, Northeast, Northwest, Southeast, Southsouth, Southwest). All (n = 52) publicly funded referral-level hospitals providing in-patient services for obstetric and gynaecological admissions were targeted for inclusion in the study. Of these, 48 hospitals (92.4%) provided consent, participated and successfully implemented the study. In addition two referral-level privately funded facilities in each region were targeted for inclusion. Six privately funded hospitals located in Northwest (n = 2), Southwest (n = 2), and Southsouth (n = 2) regions consented and participated in the study. The study population comprised of all women (and their babies) who were admitted for delivery or on account of complications within 42 days of delivery or termination of pregnancy between 1 September 2019 and 31 August 2020. This population was chosen to account for all pregnancy-related complications that could result in severe morbidity or death in the participating hospitals as well as the standard denominators for estimating the burden of maternal, foetal, and neonatal mortality (live births and all births). Additionally, women who experienced a pregnancy loss or had an abortion were included, an often neglected population in national maternal and/or perinatal health databases. The 42-day time period for including women after giving birth or termination of a pregnancy ensured that all women whose deaths could be classified as a maternal death were enrolled. For the purpose of data entry, a woman was categorised as an obstetric admission if she was admitted for delivery at or after 28 weeks of gestation or if she was admitted within 42 days following delivery. A woman was categorised as a gynaecological admission if she was admitted with a pregnancy (<28 weeks) that ended in spontaneous, induced or missed abortion, molar pregnancy, ectopic pregnancy, or intrauterine foetal death. The scientific content of the study was approved by the WHO Human Reproduction Programme (HRP) Research Project Review Panel (protocol ID, A65930, 06 May 2018). WHO Ethics Review Committee (ID A65930, 05 June 2018) and the Nigerian National Health Research and Ethics Committee approved the study (ID NHREC/01/01/2007, 05 September 2018). Authorities of all participating hospitals granted written institutional approvals to participate in the programme's data collection, periodic analyses and reporting. Individual level written consent was not required as the study did not involve direct interaction with women or their babies, or interview of medical staff. The electronic data platform was developed by customising open-source District Health Information Software (DHIS-2), the approved Health Management Information System by the Federal Ministry of Health of Nigeria. The data platform collected information from patient medical records using an electronic case report form. The information obtained included women's sociodemographic data, past medical history, antenatal history, labour and delivery details (including the baby's clinical condition), and immediate postpartum observations. The data collected were used to report on the QED indicators specified for the multinational WHO QED initiative (Appendix II).6 There was a National Coordinating Unit, comprising of a national coordinator, a data manager, a statistician, six regional coordinators (each overseeing facilities in their region), Nigeria Ministry of Health representatives and WHO staff from the Nigeria country office and headquarters. Each of the participating hospitals had a team of one obstetrician, one neonatologist (designated as hospital coordinators) and two medical record officers (MROs) coordinating and entering data from patient medical records into the database. Each hospital had quarterly facility audits conducted by their hospital coordinators on available services, human resources and supplies. Trained MROs (two at each hospital) conducted daily surveillance of medical records in the obstetric ward, gynaecological emergency unit, birthing/delivery room, operating theatre and intensive care unit. For each woman, data entry was initiated at time of admission, updated during admission, and completed at time of discharge or death (whichever was earlier). Data were captured from patient medical records using a tablet-based case report form that was specifically developed for the project. A unique identifier was used to link the data between a woman and her newborn in the database. In the event of a maternal or perinatal death (stillbirth or early neonatal death), the local mortality audit team (led by an obstetrician and neonatologist) analysed and documented the primary cause of death (using International Classification of Diseases for maternal death [ICD-MM] and perinatal death [ICD-PM]),7,8 and the associated avoidable factors.9 In cases of re-admission of a woman (within 42 days of delivery) or neonate (within the first 7 days of life), a re-admission form was generated to record the reason(s) for the re-admission and the outcome. A re-admission was not considered a new entry and additional information from the re-admission was added to the previously recorded data for the woman or neonate. The data collected were synchronised in real-time from the internet-enabled tablet device to a secured central cloud-base server. The data record for each participant was closed 60 days after admission – allowing for correction of any data entry error by hospital coordinators, generation of a re-admission form (in case of a re-admission), and completion of maternal or perinatal death audit form where applicable. To ensure reliability of data and minimise heterogeneity in data collection across hospitals, several quality assurance procedures were put in place. A manual of standard operating procedures was developed and training provided to project teams from all facilities. The data platform had in-built validation rules, including the use of mandatory fields, to minimise data entry errors and ensure completeness of data. Before synchronising with the central server, data entered for each woman and her baby were verified by the hospital coordinators. The data manager at the National Coordinating Unit and a staff at the WHO Country Office conducted monthly checks of total admissions in relevant hospital registers against the admissions recorded in the database, and took necessary actions to address any discrepancies. In addition, approximately 5% of woman-infant records were randomly selected and scrutinised for any errors or inconsistencies on a weekly basis by the hospital coordinator. Any inconsistencies or errors identified were resolved before closure of the data record for each participant, ensuring only accurate data were included in the database. The availability of live-saving interventions (blood banking services, neonatal resuscitation facilities, operating theatre and anaesthetic machine), and availability of oxytocin, ergometrine, tranexamic acid, magnesium sulfate, intravenous fluids, and strength of the health care workforce were assessed using data obtained from quarterly facility-level audits. Key outcomes for women and their babies included the burden and causes of morbidity and mortality, avoidable factors contributing to mortality, and independent predictors of mortality. We assessed the burden of maternal morbidity according to the morbidity continuum adapted from Say et al.10 with the following categories: women with any complication regardless of severity, women with potentially life-threatening complication, women who survived life-threatening complication (severe acute maternal morbidity), and maternal death. The full description of these categories can be found in Appendix III. The burden of maternal mortality was determined by calculating the intrahospital maternal mortality ratio (defined as number of maternal deaths amongst all women admitted and managed in the hospital regardless where they gave birth, per 100,000 live births); and pre-discharge maternal mortality ratio (defined as number of deaths amongst women who delivered in the hospital prior to discharge per 100,000 hospital live births). While the former was based on standard definition of maternal mortality ratio at the hospital level, the latter had been specifically developed as one of the QED indicators. The burden of perinatal mortality was estimated by calculating the QED indicators institutional stillbirth rate and pre-discharge neonatal mortality rate. Institutional stillbirth rate was disaggregated by antenatal and intrapartum stillbirths and was defined as the number of babies born in a health facility with no signs of life at birth, per 1000 facility births, and pre-discharge neonatal mortality rate was defined as the number babies born live in a facility who died prior to discharge per 1000 facility live-births. Avoidable factors contributing to maternal and perinatal deaths were assigned by the hospital mortality audit team using the National Maternal and Perinatal Death Audit Tool that was developed and published by the Nigerian Federal Ministry of Health.9 Avoidable factors included patient-orientated and facility-level factors as defined in the National Maternal and Perinatal Death Audit Tool (Appendix IV). Avoidable factors experienced by women and their babies prior to their arrival at the hospital and upon arrival at hospital were based on evaluation of patient medical records and were ascertained by the hospital mortality audit team. The quality of care provided in the participating hospitals was assessed using a list of indicators that were developed by a consultative group of experts to standardize quality assessment across the nine countries participating in the WHO QED initiative.6 The definitions and measurement of these QED indicators are presented in Appendix II. To complement this assessment, the overall quality of care performance for specific maternal conditions were also assessed using cause-specific case fatality rates. Cause-specific case fatality rate was defined as the proportion of women who died from a specific condition amongst all women with the condition. Cause-specific case fatality rates were calculated at the national and regional levels. The following sociodemographic and clinical variables were separately analysed as independent predictors of maternal and perinatal deaths: maternal age, marital status, highest education level attained, woman's occupation, husband/partner's occupation, previous miscarriage(s), previous caesarean section(s), parity, antenatal care, referral status, cervical dilatation at admission, mode of birth, companionship in labour (for obstetric admissions), continuous labour monitoring with a partograph (in women who underwent labour) and use of uterotonic for the prevention of postpartum haemorrhage. We conducted descriptive analysis on the available human resources and supplies at the facility level, as well as the sociodemographic and clinical characteristics of the study population at the individual level. We performed descriptive analysis of the proportion of women in each category of the maternal morbidity continuum. We calculated intrahospital maternal mortality ratio and pre-discharge maternal mortality ratio, the live births and stillbirths (by place of birth), based on the above definitions, for all facilities, and for publicly-funded and privately-funded facilities separately. Descriptive statistics were used to examine the causes of maternal and perinatal deaths and avoidable factors contributing to deaths. Independent predictors for maternal death, and for perinatal death were separately explored using logistic regression models. Each variable was first entered in a univariable logistic regression model with death as the binary outcome for unadjusted estimates. Variable levels were aggregated where appropriate. Multilevel mixed-effects logistic regression models were used to determine the sociodemographic and clinical characteristics that were associated with maternal death. A similar model was used to determine the variables associated with perinatal death. Random effects were adjusted for at the hospital level. A backward stepwise variable selection procedure was used to progressively remove the least useful predictors until the most parsimonious model was achieved. Unadjusted and adjusted odds ratios (OR) with corresponding 95% confidence intervals (CI) and p-values are reported. Statistical significance was accepted at p<0·05. All analyses were conducted with STATA Version 16 (College Station, TX: StataCorp LLC).11 Variables included in the univariable and multivariable models with >5% missing data were checked for type of missingness (missing not at random, missing completely at random, missing at random).12 Where necessary, sensitivity analysis was conducted by comparing complete case analyses with models including missing data. The use of multiple imputation for missing values was explored where appropriate. Role of funding source: The funders did not play any role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. JT, TL, MB, OO had access to the data. The decision to submit the paper for publication was taken by all authors.