Impact of maternal death reviews at a rural hospital in Zambia: a mixed methods study

Maternal mortality refers to the death of a woman from causes related to or exacerbated by pregnancy or childbirth. Most maternal deaths occur during childbirth or within 48 h postpartum, and is most commonly due to postpartum hemorrhage, puerperal sepsis, hypertensive disorders (e.g. eclampsia) or uterine rupture [1, 2].

Maternal mortality differs vastly between high-income countries and low- and middle-income countries; the latter account for 99% of maternal deaths, and sub-Saharan Africa alone accounts for two-thirds [3]. Decades of concerted global efforts on the part of the United Nations and the multi-institutional Countdown to 2015 partnership to improve maternal health have been met with limited success. Between 1990 and 2015, maternal deaths worldwide decreased by 45%, short of the 75% target set by the Millennium Development Goals [4]. In sub-Saharan Africa, maternal deaths decreased from 990 to 510 per 100,000 live births, remaining over thirty times higher than in developed regions [4]. The partial achievements of the Millennium Development Goals were succeeded by the United Nations Sustainable Development Goals and the Countdown to 2030 partnership, which aims to reduce maternal deaths in every country worldwide to < 140 per 100,000 live births by the end of this decade [5]. To achieve this, guaranteeing access to comprehensive, quality reproductive health services and skilled health professionals at the time of delivery is crucial, as are identifying causal factors surrounding maternal deaths and formulating informed responses [6, 7].

One means by which the latter is accomplished is through facility-based internal reviews of maternal deaths. Maternal death review (MDR) is a decades-old quality improvement approach centered on reviewing and deriving lessons from circumstances surrounding the deaths of individual mothers [8]. In 2004, the World Health Organization (WHO) established technical guidelines for carrying out MDRs to help health officials develop safe motherhood strategies and improve clinical and public health practice, and by 2013 approximately half of the Countdown countries had implemented community- and/or facility-based programs [7, 9]. Yet the evidence for the effectiveness of MDR in resource-limited settings is equivocal [10]; some studies identified MDR-associated improvements in maternal health services or reductions in maternal deaths [11‐17], whereas others found no impact [18‐20]. Although an old practice, MDR remains an essential component of WHO-recommended maternal death surveillance and response [9], and consolidating evidence of its impact on maternal morbidity and mortality in relevant contexts remains crucial to global safe motherhood strategies.

In 2007, the Ministry of Health of the Government of the Republic of Zambia developed a Reproductive Health Roadmap in effort to reduce maternal mortality [21]. The Zambian experience reflects that of sub-Saharan Africa at large, with maternal mortality of 591 deaths per 100,000 live births in the year the Roadmap was published, and more recent estimates by the WHO of 224 maternal deaths per 100,000 live births in 2015, translating to a lifetime risk of maternal death of 1 in 79 Zambian women [22]. The Reproductive Health Roadmap incorporated WHO recommendations for health facility-based reviews, and Zambia therein joined other Countdown countries in implementing a nationwide MDR program. However, to the best of our knowledge, the impact of MDR on maternal mortality in Zambia and the factors surrounding maternal deaths have not previously been evaluated in rigorous study designs.

We conducted a mixed methods convergent study to assess the impact of facility-based MDR on maternal mortality at a rural, district-level hospital in Zambia with the objective of informing policymakers and stakeholders in maternal health programs. In this setting, women often die of pregnancy-related complications because of delayed care-seeking behavior, or delays in reaching appropriate health facilities [23]. In-hospital maternal mortality is thereby worsened by late presentation to care, in addition to hospital related barriers to adequate obstetric care [23]. We hypothesized that MDR implementation would improve subsequent obstetric care and thereby reduce maternal mortality as problems were identified and MDR committee recommendations were applied to health facility workflow, provider practice, and resource utilization, affecting change across the local health system.

We identified 106 maternal deaths over the study period. All study subjects had complete MDR reports available. Thirty-nine deaths occurred in 2007, and 67 occurred during 2008–2011 when in-depth reviews were conducted. Characteristics of the latter group are listed in Table 1. The median age was 30 years (interquartile range: 23–44) and most of the women were multiparous (76%). Nearly a third of the women delayed seeking care (31%). A majority of the deaths occurred within 24 h of presentation (57%) and few (13%) had home delivery before reporting to the health facility. The most common cause of death was postpartum hemorrhage (21%) followed by puerperal sepsis (19%), uterine rupture (16%), and eclampsia (9%). Other causes included complications of human immunodeficiency virus (HIV) infection (15%) and malaria (8%). Thematic analysis of MDR committee reports revealed four general categories of opportunities and barriers to care: standards of practice, health systems, accessibility, and patient factors (Table 2).

We conducted a mixed methods convergent study of an MDR program in a rural Zambian hospital to evaluate its impact on inpatient maternal mortality. We hypothesized that health system improvements and other changes affected by MDR implementation would lead to reductions in hospital-based maternal deaths and high-risk features of cases. Our results showed fewer inpatient maternal deaths following MDR implementation, identified contributory factors to maternal mortality, highlighted inherent limitations to the program, and captured general and context-specific recommendations rendered by the MDR committee.

Maternal deaths were 35% lower after MDR was implemented compared to the period before its implementation, similar to results of other before-and-after studies of MDR in sub-Saharan Africa [11, 12]. Inpatient maternal deaths fell from 23 per thousand live births to 8 per thousand. Note that the current study was restricted to inpatient deliveries at a district-level referral hospital, whereas public health statistics cited elsewhere in this report included deliveries at local health centers where the large majority of live births occur, which are typically uncomplicated and result in far fewer deaths.

MDR programs have previously been linked to improvements in maternal outcomes, and prior studies offer insights as to how. In an extensive review, Ivers et al. found slight but significant improvements in professional practice and health outcomes after the introduction of audit- and feedback-based approaches [27]. MDR programs have led to shorter wait times, lower rates of severe complications (e.g. uterine rupture), and higher attainment of the six components of basic emergency obstetric care while organizational restructuring and resource mobilization spurred by MDR-guided feedback has led to lower overall maternal mortality [10, 14‐16, 28, 29].

In our series of 106 maternal deaths, the major direct causes were postpartum hemorrhage (21%), puerperal sepsis (19%), uterine rupture (16%), and eclampsia (9%) while the two most common secondary causes were HIV (15%) and malaria (8%). These findings resemble contemporaneous maternal mortality statistics elsewhere in Zambia and sub-Saharan Africa; leading causes of maternal death nationally in Zambiawere postpartum hemorrhage (34%), puerperal sepsis (13%), obstructed labor/uterine rupture (8%), and eclampsia (5%) with 10–11% attributed to each HIV and malaria [21, 22, 30]. The relatively low prevalence of malaria in the sample was somewhat surprising; Nchelenge District is among the most malarious areas in Zambia [31, 32] and although intermittent presumptive treatment of malaria in pregnancy is indicated according to national guidelines, uptake in the community remains low overall [33]. The apparent discrepancy appears to be explained by the fact that malaria in pregnancy is most life-threatening in primigravidae [34], whereas most of the mothers in our study (76%) were multiparous.

In Zambia and across sub-Saharan Africa, postpartum hemorrhage is the most common cause of maternal mortality [22, 30], so the apparent ineffectiveness of MDR to lessen its toll warrants special attention. Postpartum hemorrhage requires prompt treatment, therefore delays in reaching a first-level health facility (e.g. due to distance or delayed care-seeking) or second-level facility (e.g. due to delayed or unavailable ambulance) contribute to poor outcomes [35]. The inefficacy of MDR to reduce postpartum hemorrhage deaths signals a need for material investments and capacity building beyond MDR. When a mother who recently gave birth or underwent a Cesarean section has major internal bleeding, prompt diagnosis and rapid response by a skilled surgical team are necessary but insufficient if, for example, the mother cannot reach a facility in time, or if the facility has no blood for transfusion. Policies with monetary backing that build capacity for basic emergency obstetric services, ensure that blood banks are stocked, and adequately equip operating theaters are equally crucial and should follow hand-in-hand with the upskilling of healthcare workers that MDR accommodates [36].

The causes of delay, challenges to care, and medical errors identified by the MDR committee encompassed individual, institutional, and structural elements that echoed findings from the earliest MDR program assessments done as far away and far back as England and Wales in the 1950s [8] to more recent accounts in neighboring sub-Saharan Africa countries [12‐17]. Prior studies of MDR programs in sub-Saharan Africa documented many of the same factors contributing to maternal death, and similarly point to root causes that are structural, beyond the sweep of MDR committees or their regional governments [20, 37, 38]. These factors can be considered according to the Thaddeus paradigm of delays in care (e.g. transportation delays, delayed care-seeking, late diagnosis leading to an in-hospital delay) [23] and they are one and the same as the structural barriers that accompany unalleviated poverty: shortages of life-saving supplies and skilled staff, competing priorities that vie for scarce resources, whole districts without hospitals.

Many of the mothers in our study (39%) were referred from outside the district, and most (76%) received at least some antenatal care, highlighting the opportunity for upstream maternal health interventions. Reducing maternal deaths due to complications of obstructed labor and eclampsia in particular demands skilled antenatal care and early diagnosis and treatment at pre-referral centers [14, 39]. Our study also described patient preferences for ethnomedical care and home delivery, which carry previously well-characterized risks due to inadequate pharmacotherapies and unskilled management of labor complications [28, 40, 41]. Logistical barriers due to transportation were prevalent in the study, which have an extensive literature documenting their contribution to maternal deaths by delaying urgently needed care [42, 43]. These upstream factors anticipate an opportunity for maternal health and patient education-based interventions beyond hospital-based MDR [44].

Recommendations proposed by the MDR committee ranged from pragmatic solutions that could be enacted at the point of care, to farther-reaching appeals to the district, provincial, and national health offices. More proximal recommendations sought to address staff performance, coordination of care, and delegation of responsibility, in line with interventions previously shown to reduce maternal deaths [10, 16]. These included recommendations to train staff on current guidelines, improve documentation, promote teamwork and clear communication, devise clear work schedules that are disseminated effectively, and to assign skilled tasks to the appropriate provider (e.g. wound dressing changes, sterilization of surgical equipment). Community-level recommendations included engaging traditional birth attendants, for which there is mixed evidence to support reductions in maternal mortality [45‐47], and community sensitization via radio announcements to increase antenatal care clinic attendance.

Several of the MDR committee recommendations implicated chronic understaffing. Understaffing is associated with lower quality of care and higher maternal mortality, and was previously identified as a major barrier to achieving the Millennium Development Goals [48, 49]. MDR recommendations instructed personnel to take up duties outside their typical purviews, which can lead to worse maternal health outcomes and job dissatisfaction on the part of providers [48]. Midwives were recommended for post-operative wound care, nurses from non-obstetric wards for obstetric emergencies, surgical nurses for supervising sterilization procedures typically performed by trained environmental staff, and laboratory technicians for ensuring blood supplies despite having no capacity for blood collection, atop frequent stock-outs in the nearest blood bank located 240 km away. Finally, all of the MDR committee recommendations have a prerequisite for effective leadership at all levels, a recognized necessity for the success of any maternal health intervention [50]. From staff performance (e.g. adherence to guidelines, appropriate partograph use) to the delegation of duties and fostering channels of communication, several of the contributing factors and the recommendations they spurred can be traced to, or rest upon, good leadership.

There are limitations to this study that are inherent to its before-and-after design, such as the lack of rigorous control group or ability to show causation, and others that are particular to the study sample and context. Anecdotally, there were no obvious socioeconomic or other secular trends over the comparison periods to account for improved outcomes, although we did not measure indices of poverty, literacy, education, or other potential explanatory factors. Similarly, we cannot discount the possibility of decreased referrals of complicated cases resulting in a shift of maternal deaths from the hospital to peripheral centers, although casual observations from the field did not support this. Despite fewer maternal deaths seen in the before-and-after analysis, we found no clear patterns in the before-and-after analysis of potential explanatory factors, outlined in Table 4, to suggest the drivers of the change. There were downward trends in transportation delays, missed diagnoses, and medication and supply stock-outs. The analysis was hindered by the lack of a more stringent comparator groups (e.g. mothers with the same diagnoses who survived) and small sample size. Our study included only mothers who presented or were transferred to the district hospital, therefore pre-hospital deaths were not captured, and unlike some other MDR programs ours did not include near misses.

Our analysis of more than one hundred MDR reports portrays the complex web of health care delivery that interlinks mothers, health providers, clinics and hospitals, local communities, policymakers, and budgetary officers. In Nchelenge, their interrelationships and commitment to the common goal of reducing maternal mortality are confounded by economic underdevelopment and its attendant hardships. Infrastructure that is hard-worn or nonexistent, such as empty blood banks, impassable roads or want of vehicle, and shortages of essential equipment impedes the efforts of MDR committees.

This study contributes evidence of MDR’s effectiveness in reducing maternal deaths, but also unveiled a critical gap. While maternal deaths from three of the top four causes fell after implementation of MDR, deaths due to postpartum hemorrhage—the most common cause of maternal death in Zambia and the rest of sub-Saharan Africa—remained unchanged. Future research should focus on causal analysis and interventional studies to better understand and ultimately close this gap. MDR programs are impactful but alone are insufficient to eliminate the vast disparities in obstetric care between low and high income countries. In Zambia and similar resource-deprived settings, progress must continue to rely on financial and material provisioning by global aid, regional and national governments, and other holders of the purse strings while fuller economic development awaits realization.