Predicting resource-dependent maternal health outcomes at a referral hospital in Zanzibar using patient trajectories and mathematical modeling

Abstract

Poor intra-facility maternity care is a major contributor to maternal mortality in low- and middle-income countries. Close to 830 women die each day due to preventable maternal complications, partly due to the increasing number of women giving birth in health facilities that are not adequately resourced to manage growing patient populations. Barriers to adequate care during the ‘last mile’ of healthcare delivery are attributable to deficiencies at multiple levels: education, staff, medication, facilities, and delays in receiving care. Moreover, the scope and multi-scale interdependence of these factors make individual contributions of each challenging to analyze, particularly in settings where basic data registration is often lacking. To address this need, we have designed and implemented a novel systems-level and dynamic mathematical model that simulates the impact of hospital resource allocations on maternal mortality rates at Mnazi Mmoja Hospital (MMH), a referral hospital in Zanzibar, Tanzania. The purpose of this model is to provide a rigorous and flexible tool that enables hospital administrators and public health officials to quantitatively analyze the impact of resource constraints on patient outcomes within the maternity ward, and prioritize key areas for further human or capital investment. Currently, no such tool exists to assist administrators and policy makers with effective resource allocation and planning. This paper describes the structure and construct of the model, provides validation of the assumptions made with anonymized patient data and discusses the predictive capacity of our model. Application of the model to specific resource allocations, maternal treatment plans, and hospital loads at MMH indicates through quantitative results that medicine stocking schedules and staff allocations are key areas that can be addressed to reduce mortality by up to 5-fold. With data-driven evidence provided by the model, hospital staff, administration, and the local ministries of health can enact policy changes and implement targeted interventions to improve maternal health outcomes at MMH. While our model is able to determine specific gaps in resources and health care delivery specifically at MMH, the model should be viewed as an additional tool that may be used by other facilities seeking to analyze and improve maternal health outcomes in resource constrained environments.

Reducing Maternal Mortality: A Qualitative Study of Health Workers’ Expectation in Urban Area, Indonesia

“…Midwives are the obstetrician's right hand to carry out initial treatment for maternal referrals according to instructions because obstetricians do not stand by 24 h at the referral hospital, while pregnant women who get referrals need urgent action to prevent delays inadequate service. Nadkarni et al explained that in their research, doctors did not stay by every shift, increasing the number of maternal mortality by tenfold [49]. This is consistent with Ghana's research because the delay in intervention by obstetricians to pregnant women at referral hospitals is one of the main factors in maternal mortality [50], [51].…”

“…Sharan et al mentioned that in CEmOC and non-CEmOC referral hospitals in Eritrea, most obstetricians take the optimal day and night shifts to provide direct referral care to pregnant women to minimize morbidity and mortality [54]. Other studies describe secondary facilities as referral places that must have complete staff in pregnant women at risk, resulting in overall maternal health [49]. The importance of a 24-h stand-by obstetrician in referral services at the hospital to provide care and direct case management for pregnant women at risk is proven to provide adequate care, resulting in maternal and infant health outcomes.…”

Data Science Methods for Nursing-Relevant Patient Outcomes and Clinical Processes

“…Two studies reported predictive models formed from ML methods, and one study was a report of NLP used to transform clinical notes into assessment forecasting. The Nadkarni 165 research group used a stepwise, iterative, object-oriented program written with workflow and treatment processes in mind in a sample of 343 patients with potentially life-threatening complications and 2,285 uncomplicated mothers in a Tanzanian hospital. Aimed at providing decision makers with a tool to analyze the impact of resource limitations on maternal inpatient complications, key variables included treatment efficacy, severity distribution, number and frequency of nurse visits, nurse staffing at the shift level, deterioration rate, and maternal near-misses.…”

“…Two studies 29,162 reported predictive models formed from ML methods, and one study 163 was a report of NLP used to transform clinical notes into assessment forecasting. Each had retrospective designs using secondary data, two used a single tertiary care setting 29,162 (tertiary care, maternal care), and one 163 used two matched psychiatric settings. Diverse samples were used, with one study 162 using over 2500 maternal inpatients, another 29 using over 800 medical/surgical inpatients, and the remaining study 163 sampling the admission encounters of over 5000 psychiatric patients.…”

“…We identified three publications 29,162,163 for staffing but none for scheduling or workload. Data science methods were reportedly used in these studies aimed at estimating the antecedents or consequences of nurse staffing.…”