Challenges in reducing maternal and neonatal mortality in Niger: an in-depth case study

Introduction Recent modelled estimates suggest that Niger made progress in maternal mortality since 2000. However, neonatal mortality has not declined since 2012 and maternal mortality estimates were based on limited data. We researched the drivers of progress and challenges. Methods We reviewed two decades of health policies, analysed mortality trends from United Nations data and six national household surveys between 1998 and 2021 and assessed coverage and inequalities of maternal and newborn health indicators. Quality of care was evaluated from health facility surveys in 2015 and 2019 and emergency obstetric assessments in 2011 and 2017. We determined the impact of intervention coverage on maternal and neonatal lives saved between 2000 and 2020. We interviewed 31 key informants to understand the factors underpinning policy implementation. Results Empirical maternal mortality ratio declined from 709 to 520 per 100 000 live births during 2000–2011, while neonatal mortality rate declined from 46 to 23 per 1000 live births during 2000–2012 then increased to 43 in 2018. Inequalities in neonatal mortality were reduced across socioeconomic and demographic strata. Key maternal and newborn health indicators improved over 2000–2012, except for caesarean sections, although the overall levels were low. Interventions delivered during childbirth saved most maternal and newborn lives. Progress came from health centre expansion, emergency care and the 2006 fee exemptions policy. During the past decade, challenges included expansion of emergency care, continued high fertility, security issues, financing and health workforce. Social determinants saw minimal change. Conclusions Niger reduced maternal and neonatal mortality during 2000–2012, but progress has stalled. Further reductions require strategies targeting comprehensive care, referrals, quality of care, fertility reduction, social determinants and improved security nationwide.


Quantitative data analysis
We computed the annual rate of reduction (ARR) in MMR between 2000 and 2017 and in NMR between  2000 and 2019 using an exponential growth formula with a constant negative rate of reduction.The trends in modeled estimates were used for comparison across countries in West Africa.We relied on measured empirical mortality for analysis.For NMR, we pooled birth history modules from the six national household surveys, restricted to births in the past ten years preceding each survey.We assessed the consistency of estimates of NMR across the surveys prior to pooling the data.The pooled data included a total of 152 876 births.We computed 95% confidence intervals (CIs) using the Jackknife non-parametric method.We computed NMR on three-year periods at national and five-year period at subnational levels (region and place of residence) and for characteristics that were unlikely to change over time for each birth (e.g., maternal education, place of birth, birth-risk factors, household wealth quintile).We analyzed inequalities in NMR by the same characteristics.The birth risk composition distinguished births with no risk, with unavoidable risk, and single or multiple risks (see definition in Appendix Table 2).
We assessed trends separately for 2000-2012 and 2013-2021.The year 2012 was used as the midway cutoff period because of the 2012 DHS.The analysis used the sampling weights to account for the sampling design. 22We computed standard errors around the mortality estimates using Jackknife non-parametric methods. 23Maternal mortality estimates were extracted from the country survey reports.Niger does not have empirical data on causes of neonatal and maternal deaths, besides a follow-up to the 2010 survey that collected child causes of death using verbal autopsy. 24 assessed changes in the coverage of standard MNH indicators and their disaggregation by similar stratifiers as for mortality (except for birth risks) using the reanalyzed database from the International Center for Equity in Health (ICEH). 25Absolute equity gaps, annual percentage points (pp) changes, and equity patterns were visualized using equiplots.
To assess changes in the quality of care, we computed facility ANC and delivery readiness using SARA 2015 and SARA 2019.The ANC readiness was computed as an arithmetic average of the availability of 22 essential items across five domains for ANC 26 : equipment, diagnostics, medicines and commodities, basic amenities, and human resources (Appendix Table 3).
We calculated the facility readiness score for delivery services from 20 basic emergency obstetric and newborn care (BEmONC) items across three domains: equipment and supplies, medicines and commodities, human resources and guidelines.We also generated a Comprehensive emergency obstetric and newborn care (CEmONC) readiness score covering 19 items, restricted to facilities providing C-section and blood transfusion services (Appendix Table 3).
We ecologically linked available health facility surveys with household surveys to assess readiness-adjusted coverage measures of interventions offered during antenatal care and childbirth that are not available in household surveys alone.The linking was possible only for delivery services for which place of delivery information was collected in the household surveys.Ecological linking was done by facility type and geographic region: first, we matched the facility types reported in the household survey to the corresponding facilities in the health facility survey.We then linked a woman's report of a facility delivery to the average labor and delivery readiness score of that facility type in the woman's region of residence, to compute readiness-adjusted delivery care estimates.For ANC, where linking was not feasible, we generated an ANC content score from six content interventions: urine and blood tests, counseling for pregnancy complications, blood pressure measurement, iron/folic acid supplementation, and intermittent preventive treatment of malaria (IPTp).Self-reported content of ANC was measured as the proportion of women with at least one ANC contact during their last pregnancy who reported receiving these key content interventions as part of ANC.We calculated an overall score out of 6 for the receipt of these 6 ANC content items, with one point assigned to each of the 6 content items, and the score was converted into a percentage.
We assessed the contribution of distal, intermediate, and proximate factors on maternal and neonatal mortality by conducting three sets of complementary analyses.First, we assessed the role of fertility changes on maternal and newborn lives saved and mortality decline between 2000 and 2017 using the proposed decomposition method by Jain. 27Fertility changes are key factors of change in maternal and neonatal mortality.Fertility decline implies changes in birth rates and in birth risk composition.Because both maternal mortality ratio and neonatal mortality rates are expressed per birth unit, decline in birth rates does not necessarily imply a decline in the risk of death.Thus, a decline in birth rates will only affect the number of maternal lives saved.However, when birth risk composition changes toward lower risk births, it can induce a decline in MMR and NMR.
We then analyzed the contribution of intervention coverage to changes in maternal and neonatal mortality using the lives saved tool (LiST), 10 a mathematical modeling tool that uses changes in coverage intervention coverage to estimate the number of lives saved.LiST also decomposes the contribution of specific interventions to the total lives saved.

Qualitative data analysis
A total of the 31 key informant interviews conducted between August 9 and November 29, 2021, and 25 could be transcribed and the other six were subject to written summaries.

Coding of responses
The first task was to create general categories corresponding to the major themes contained in the interview guide.Each category was then subdivided into sub-categories or sub-themes, which we then added to as we processed the data.Each sub-theme is subdivided in turn, according to the desired level of detail.This results in sub-categories that are finer and more numerous than the initial categories from which they were derived, and that express increasingly precise aspects of the survey.This process of splitting up the interviews allows us to put the data in order and thus prepare for the data analysis work.

Data analysis
The analysis was carried out in two stages.The first stage grouped the data according to their affinity and thus classified the points of view into "families", distinguishing between those that converge and those that diverge on the different questions of the survey.The second step consisted in triangulating the data by considering these points of view in relation to each other.This cross-reading of the data made it possible, depending on the case, to confirm, qualify or refute the points expressed by one or another interviewee.
The triangulation was not limited to the data collected in the survey.In fact, the information obtained from the literature search allowed us to better understand the general problem of maternal and neonatal health and, consequently, to take a step back from the positions defended by our interlocutors.In other words, the literature search provided a reading grid that considerably facilitated the analysis of the data.

Litterature review
The literature review focused on the evolution of the health system policy and program development and implementation in Niger between 2000 and 2020 to address several themes including the organization of care, infrastructure, human resources, financing, drugs, governance, etc.In most cases, these documents were the subject of reading notes that made it possible to identify the content of the policies and programs discussed, the context in which they were formulated, the mechanisms for their implementation, their results and their implications for maternal and neonatal health (MNH).The literature review thus allowed us to develop a general description of the organization of care for the period 2000-2020 and to develop a detailed chronology of MNH health policies in Niger for the same period.Certain themes, such as health financing and family planning, were the subject of in-depth analyses.These analyses are included in the appendix of this study.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

Appendix Figure 5: Trends in coverage equity in key maternal and newborn health (MNH) indicators by household wealth , 1998 -2021
Note: Q1 are the poorest households and Q5 is the wealthiest Abbreviations: mDFPS: Demand for family planning satisfied by modern contraceptive methods (modern methods include pills, condoms (male and female), intrauterine device, sterilization (male and female), injectables, implant, diaphragm, spermicidal agents, patch and emergency contraception); ANC1+: One or more visits of antenatal care; ANC4+: Four or more visits of antenatal care; ideliv: Birth occurred at a health institution/health facility; PNC-mom: women received a postnatal check-up within two days post-delivery; SBA: skilled birth attendant; csection: cesarean section; EBF: Infants less than one month of age received only breastmilk in the previous 24 hours; TT: baby was born protected from tetanus toxoid infection; Early BF: Baby was breastfed in the first hour after delivery; PNC-baby: Baby received a postnatal check-up within two days post-delivery.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) of fertility decline to maternal lives saved Total maternal lived saved in 2017 = 3,998 Maternal deaths saved by decline in birth rate Maternal deaths saved by safe motherhood programs Maternal deaths saved by changes in birth risk composition of fertility change to declined in maternal mortality ratio, Niger Total %decline in MMR 2000-2017 = 37% Maternal deaths saved by changes in birth risk composition MMR decline due to safe motherhood of fertility decline to neonatal lives, Niger Total neonatal lives saved in 2017, 26,071 Neonatal deaths saved by to fertility decline Neonatal deaths saved by newborn health programs programs Neonatal deaths saved by overlap of both (age-parity changes) of fertility change to observed declined in neonatal mortality rate, Niger Total %decline in NMR 2000-2019 = 56% Kante AM placed on this supplemental material which has been supplied by the author(s) Kante AM placed on this supplemental material which has been supplied by the author(s) placed on this supplemental material which has been supplied by the author(s) BMJ Glob Health doi: 10.1136/bmjgh-2023-011732 :e011732.9 2024; BMJ Glob Health , et al. doi: 10.1136/bmjgh-2023-011732 :e011732.9 2024; BMJ Glob Health , et al.
Appendix Figure 13: Readiness adjusted basic emergency obstetric and newborn care (BeEmONC) services by region, 2015 Appendix Figure 14: Readiness adjusted basic emergency obstetric and newborn care (BEmONC) services by region, 2021 Contribution of fertility decline (decline in birth rates and changes in birth risk composition) to maternal lives saved and decline in maternal mortality in Niger.Appendix Figure 16: Contribution of fertility decline (decline in birth rates and changes in birth risk composition) to neonatal lives saved and decline in neonatal mortality in Niger.

Table 3 :
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Current use of contraceptive by type: no method, folkloric method, traditional method, or modern method.Variable was dichotomized to modern method or the rest.Antenatal care, BeMONC and CeMONC readiness items by domain.