Figures
Abstract
Background
Maternal mortality remains a major concern in resource-limited settings, particularly among critically ill obstetric patients requiring intensive care unit (ICU) admission.
Objective
To assess the time to maternal death and identify its predictors among obstetric patients admitted to the intensive care unit (ICU) in a resource-limited tertiary hospital in southern Ethiopia, over a ten-year period.
Methods
A retrospective cohort study was conducted among 378 obstetric patients admitted to the ICU between 2014 and 2023. Kaplan-Meier analysis estimated survival probability, multivariable Cox proportional hazards regression identified independent predictors of time to maternal death reported as adjusted hazard ratios (AHR) with 95% confidence intervals, and a Fine-Gray competing-risks model was additionally conducted with discharge alive as the competing event.
Results
Of 378 obstetric ICU admissions, 126 resulted in maternal death, with a median time to death of 2.71 days (95% CI: 2.13–3.44); 71.4% of deaths occurred within the first five days, and survival probability declined rapidly before stabilizing after ten days. Rural residency (AHR 1.56, 95% CI 1.03–2.37), shock (AHR 2.27, 95% CI 1.44–3.56), multi-organ failure (AHR 1.75, 95% CI 1.09–2.78), mechanical ventilation (AHR 1.82, 95% CI 1.11–2.98), and impaired consciousness (moderate GCS: AHR 2.94; severe GCS: AHR 5.43, both p < 0.001) were independently associated with higher hazard of death, with findings consistent across the Fine–Gray competing-risks model.
Conclusion
Maternal ICU deaths occurred early, with most fatalities within the first week of admission and survival probability declining sharply in the first ten days. Shock, multi-organ failure, invasive mechanical ventilation, impaired consciousness at admission, and rural residency were independent predictors of shortened time to maternal death. Interventions targeting these conditions must be initiated early and urgently, particularly within the critical first days of ICU admission, to improve survival in resource-limited settings.
Citation: Alemu TN, Shiferaw WG, Orsongo WE, Megule SM, Balcha WF (2026) Time to maternal death and its predictors among obstetric ICU patients in a resource-limited setting: A 10-year survival analysis. PLoS One 21(6): e0352904. https://doi.org/10.1371/journal.pone.0352904
Editor: Douglas Aninng Opoku, Kwame Nkrumah University of Science and Technology College of Health Sciences, GHANA
Received: July 26, 2025; Accepted: June 16, 2026; Published: June 30, 2026
Copyright: © 2026 Alemu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: This study is a secondary analysis based on data originally collected for our previously published research. The current analysis applied different statistical methods (survival analysis) to the same dataset to address a distinct research objective. The raw data were generated at Wolaita Sodo University Comprehensive Specialized Hospital. Because the dataset contains sensitive and potentially identifying information, it cannot be shared publicly. For this reason, the data cannot be made publicly available. Ethical restrictions on open sharing are mandated by the Institutional Review Board (IRB) of the College of Health Science and Medicine at Wolaita Sodo University, as well as by the hospital’s data governance policies. Requests should be directed to: Wolaita Sodo University Email address- wsu@wsu.edu.et P.O. Box 138, Sodo City, Ethiopia.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
According to the World Health Organization (WHO), maternal death is the death of a woman while pregnant or within 42 days of termination of pregnancy, from any cause related to or aggravated by the pregnancy or its management, but not from accidental or incidental causes [1]. Maternal near-miss and severe acute maternal morbidity have recently come to be recognized as indicators of the quality of obstetric treatment [2,3].
Obstetric patients with various medical and surgical emergencies are admitted to an intensive care unit (ICU), which also offers supportive treatment to patients experiencing obstetric problems [4,5]. Approximately 1–9 per 1,000 deliveries require admission to intensive care units (ICUs), with the majority of admission diagnoses being related to obstetric complications [6,7]. Complications such as postpartum hemorrhage, sepsis, eclampsia, and other life-threatening diseases that require sophisticated monitoring and therapeutic measures might receive specialized care in intensive care units [8].
Globally, nearly 800 women die each day from avoidable pregnancy‑related causes, corresponding to a maternal mortality ratio (MMR) of 223 deaths per 100,000 live births [9]. Most maternal mortality occurs in developing countries, with Sub-Saharan Africa (SSA) alone accounting for approximately 70% of these fatalities [10,11]. The financial burden of maternal deaths affects not only the healthcare system but also families and society as a whole [12]. Maternal mortality rates (MMR) in Ethiopia have dramatically decreased, from 871 per 100,000 live births in 2000 (95% CI: 705–1039) to 267 per 100,000 in 2020 (95% CI: 189–427) live births (95 percent CI, 189–427) in 2020 [5,13]. Even though it is trending downward, the maternal mortality ratio is still high in the country [14].To address this, Ethiopia set up the maternal death surveillance and response system (MDSR) with the goal of providing real-time data on the trends and patterns of avoidable maternal deaths [15].
In Ethiopian ICU settings, available evidence remains limited but indicates alarmingly high maternal mortality among obstetric ICU patients: 27% and 29.9% in Addis Ababa, and 17.6% in Mekelle [16–18], which are markedly higher than national averages [5,19], highlighting the critical need for improved care in ICU settings.
Several factors have been consistently associated with increased mortality, including hypertensive disorders, absence of antenatal care (ANC), early ICU discharge, disseminated intravascular coagulation (DIC), HELLP syndrome (Hemolysis, Elevated Liver enzymes, and Low Platelets), vasopressor use, multi-organ failure, preeclampsia, sepsis, advanced maternal age (≥35 years), pre-existing comorbidities, and low Glasgow Coma Scale (GCS) scores [16–18].
Most Ethiopian and sub-Saharan African studies focus on static risk factors using cross-sectional or case-control designs and predominantly include non-ICU patients, despite the fact that the most severe obstetric cases are admitted to intensive care units, providing limited insight into when maternal deaths ensue [11,14,20–24]. Time-to-event (survival) analysis offers a dynamic approach, accounting for censored cases to estimate survival probabilities over time and identify predictors of both the risk and timing of death [25,26]. Survival analyses from Brazil, Nigeria, and South Africa indicate that complications such as shock, low GCS, multi-organ failure, sepsis, postpartum hemorrhage, acute kidney injury, and delayed ICU admission significantly increase mortality and reduce survival times among critically ill obstetric patients. [27–31]. These results highlight how crucial early detection and prompt, crucial interventions are to enhancing maternal outcomes in the critical care unit.
Despite nationwide improvements, maternal mortality among critically ill obstetric patients remains unacceptably high in low-resource settings such as Ethiopia [32]. We recently performed a case–control study of the same obstetric ICU cohort to determine factors associated with maternal mortality [33], which identified several static predictors of mortality, but the case–control design did not examine the temporal aspect of maternal death in the ICU. Survival analysis can be used to estimate time-to-event and to estimate how quickly death occurs and what factors contribute to the hazard of death over time. Such secondary analyses of existing datasets are widely encouraged to generate new scientific insights and maximize the value of previously collected clinical data [34].Therefore, the objective of this study was to use survival analysis to evaluate time to maternal death and its predictors among obstetric patients admitted to the ICU of a tertiary-level Hospital in Southern Ethiopia, over a ten-year period, providing important insights to guide future interventions and policy actions. Understanding when and why maternal death occur in critical care settings can inform resource allocation, optimize timely interventions, and ultimately improve maternal survival.
2. Method
2.1. Study design and setting
A retrospective cohort study was performed admitted to the intensive care unit of a tertiary-level Hospital in southern Ethiopia. The hospital provides a comprehensive range of outpatient and inpatient services, including emergency, medical, surgical, paediatric, ophthalmic, gynaecological and obstetric care [35]. The hospital encompasses a 10-bed adult Intensive Care Unit (ICU) that receives critically ill patients from all departments, including obstetric cases. It included secondary data of ICU-admitted obstetric patients over ten years.
This study analyzed the same cohort of obstetric ICU patients that was previously used in a case–control study investigating determinants of maternal mortality [33]. However, unlike the previous analysis, which applied a case–control design, the present study employed a survival analysis approach to assess time to maternal death and to identify predictors influencing the hazard of death during ICU admission.
2.2. Population and eligibility criteria
The source population comprised all obstetric patients admitted to the intensive care unit (ICU) during the ten-year study period (2014–2023 E.C.). The study participants included women who were pregnant (≥28 weeks of gestation) or within 42 days postpartum and admitted to the ICU due to obstetric or medical complications. Patients with more than 10% missing critical outcome data (admission/discharge status or survival time) were excluded due to inability to determine time-to-event outcomes. Patients referred to other hospitals before outcome determination were also excluded, as no follow-up information was available to ascertain final outcomes or survival time.
2.3. Sample size and sampling technique
A ten-year total population cohort of obstetric ICU admissions (n = 489) was identified from the hospital registry, and after review for eligibility and data completeness, Finally, 378 patients’ charts (81%) met the established inclusion criteria included in the final analysis (126 deaths and 252 censored cases; exclusions due to critical missing data (n = 111). The sample size was reduced, but it was still adequate for survival analysis using Cox proportional hazards regression. (See Fig 1 for more information).
2.4. Variables
2.4.1. Outcome variable.
The outcome variable was time to maternal death, measured in days from ICU admission to the occurrence of death. Patients who were discharged alive were considered censored observations.
2.4.2. Independent variables.
The study considered several factors that could influence maternal outcomes in the ICU. Socio-demographic characteristics included maternal age, residence (urban or rural), and marital status. Clinical factors encompassed gravidity, parity, pre-existing medical conditions, vital signs, and Glasgow Coma Scale (GCS) scores at admission. Interventions assessed during the ICU stay included the use of mechanical ventilation, blood transfusions, dialysis, antenatal care (ANC) follow-up, and other treatments administered in the intensive care unit. Admission-related factors included the primary diagnosis, mode of hospital presentation, and obstetric reason for ICU admission. In addition, complications that developed during the ICU stay, such as shock and multi-organ failure (MOF), were also evaluated as important determinants of maternal death.
2.5. Data collection procedure
Data were collected using literature-based checklists [28,36–39] and pretested on a subset of charts in the same hospital and excluded from final analysis. Data were accessed and extracted from ICU log books and patients’ reasonably complete (≥ 90 percent of variables) paper charts between 06/11/2023 and 30/12/2023. The checklists included socio-demographic data, ICU interventions, complications developed during ICU stay, and outcomes. The data were collected from selected participant charts. Two trained health professionals familiar with checklist, confidentiality, standard definitions and confidentiality procedures, performed data extraction with an overall completeness of the data of 96.4%. Quality control was ensured by double-checking a subset of records, and any discrepancies between data collectors were resolved by an assigned supervisor.
2.6. Operational definition
Maternal Death (Died): Death of a woman while pregnant or within 42 days of pregnancy termination from causes related to the pregnancy or its managements [1].
Censored (survived) Observations: Patients discharged alive from the ICU were considered censored, while those transferred to another facility were excluded and not censored.
Time to Death: Duration in days from ICU admission to the occurrence of maternal death within the ICU.
Survival Time: The total number of days a patient remained alive in the ICU, ending in either death or discharge.
Shock: Persistent hypotension requiring vasopressor support to maintain a mean arterial pressure (MAP) ≥ 65 mmHg (or equivalent) despite reasonable fluid resuscitation in ICU [40].
Multi-Organ Failures (MOFs): is the development of potentially reversible physiologic derangement involving two or more organ systems not involved in the disorder that resulted in ICU admission, and arising in the wake of a potentially life-threatening physiologic insult [41].
level of consciousness: Assessed on ICU admission using the Glasgow Coma Scale, with scores of 3–8 indicating severely decreased, 9–12 moderately decreased, and ≥13 mildly decreased consciousness [42].
2.7. Data management and analysis
Data were first entered into Epi Data version 4.6 and then exported for statistical analysis using SPSS version 25 and Stata MP Version 17.0. Demographic and clinical characteristics of the study population were summarized by descriptive statistics (frequencies, medians, interquartile ranges). Time-to-event data were analyzed with survival analysis techniques; the outcome variable was time to maternal death in days from ICU admission until occurrence of death, and patients who discharged alive were censored at the time of discharge.
We estimated survival probabilities using Kaplan-Meier survival curves, and compared survival distributions between categorical predictors with the log-rank test. A life table analysis was additionally performed to summarize survival experience across predefined time intervals, providing estimates of the number at risk, number of events, censored observations, and interval-specific survival probabilities with corresponding standard errors and confidence intervals. We used a Cox proportional hazards regression model to identify independent predictors of maternal mortality: bi-variable Cox regression for each potential predictor, all variables with a p-value < 0.25 in the bi-variable analysis and clinically important variables included in the multivariable Cox regression model (reported as adjusted hazard ratios [AHR] with 95% confidence intervals [CI]). The proportional hazards assumption was assessed using log-minus-log survival plots and Schoenfeld residuals. The global Schoenfeld test indicated no violation of the proportional hazards assumption (p = 0.781), and it was found to be satisfied. Among the included participants, 96.4% had complete data. To account for the competing risk of discharge alive, a Fine-Gray sub distribution hazards model was additionally conducted using the same covariates, with results reported as sub distribution hazard ratios (SHR) with 95% confidence intervals. The remaining 3.6% of missing values were observed in selected covariates such as age, marital status, and other socio-demographic and clinical variables included in the regression analysis. These missing values were handled using single imputation, applying median imputation for continuous variables and mode imputation for categorical variables. Model adequacy was assessed by examining Cox-Snell residuals to determine whether the Cox proportional hazards model appropriately fit the data. A p-value < 0.05 was considered statistically significant in the final model.
3. Results
3.1. Demographic distribution of obstetric ICU admissions of the study
Out of 489 initial obstetrics ICU admissions over 10 years, the final cohort of 378 patients to be analyzed met the eligibility criteria and had complete medical records. The age of obstetric patients who died ranged from 18 to 45 years, while that of the survivors (censored cases) ranged from 18 to 48 years. The mean age for both cohorts was determined to be 29 years; however, the interquartile range (IQR) was more extensive for those who died (8.25 years) compared to the censored participants (5.75 years). A significant proportion of patients consisted of 103 (81.7%) of those who died and 218 (86.5%) of those who were censored, with the majority falling within the 18–34-year age range.
Most of the deceased patients resided in rural locations (88, 69.9%), while more of the censored participants were from urban areas (144, 57.2%). For both groups, about two-thirds of the participants were married, and around 3% were widowed. (See Table 1 for more information)
3.2. ICU admission characteristics and primary diagnoses of obstetric patients
The timing of ICU admission, length of stay, and types of diagnoses differed between the two groups (Died vs Censored).
Concerning admission timing, more maternal deaths happened after childbirth (postpartum); 65.9% of deaths occurred during this period, compared to 50.8% of survivors (censored). Conversely, 44.5% of censored patients were admitted during the pregnant period (antenatal), while only 29.4% of those who died were admitted during pregnancy.
The median time to death was 2.7 days (95% CI: 2.1–3.4), and 71.4% of deaths occurred within the first 5 days, but only 52.7% of survivors (censored) stayed fewer than five days. This shows that most deaths happened soon after ICU admission.
About admission diagnoses, obstetric hemorrhage was more frequent in patients who died (41.2%) than in those who survived (28.6%). Likewise, shock and multi-organ failure were reported more often in maternal deaths (46.0% and 20.6%, respectively) than in censored patients (15.1% and 4.8%, respectively).
On the other hand, hypertensive disorders of pregnancy were about the same in both groups, while heart problems were more frequent among survivors (34.1%) than in maternal deaths (23.0%). Severe anemia was common in both groups, affecting 54.0% of maternal deaths and 51.2% of censored patients. (See Table 2 for more information).
3.3. Management and clinical characteristics of ICU-admitted obstetric mothers: Comparison between died and censored groups
In our study of obstetric patients admitted to the ICU, most received antibiotics: 94.4% of those who died and 98.4% of those who survived (censored). Blood transfusions were also common, with almost three-quarters of those who died (73.8%) and about two-thirds of survivors (63%). Obstetric patients who died were much more likely to receive intensive life-saving interventions: CPR was performed on 14.2%, dialysis on 10.3%, and Invasive mechanical ventilation on 80% compared with lower rates for the survivors (censored), (11.1% CPR, 3.1% dialysis, and 43.2% mechanical ventilation).
Many of the patients who died showed signs of serious illness: about one third was tachycardic (heart rate, above 100 beats per minute) and hypertensive (blood pressure above 130/80 mmHg) was common. Nearly 40% were unconscious on admission, reflected by low scores on the Glasgow Coma Scale. (See Table 3 for more information).
Shock was the most frequent complication, affecting 65% of patients who died, mostly due to blood loss (hypotension), whereas fewer (33%) survivors (censored) experienced shock. Around one fourth of each group faced infections acquired during their ICU stay (Healthcare associated- infection), but multi-organ failure was much more frequent among non-survivor mothers (over 20%) than those who survived (5%). Respiratory dysfunction was the common disorder in ICU- admitted obstetric cases; those who died 36 (28.6%), and survivors had 59 (23.4%). (See Table 4 for more information).
3.4. Survival outcomes and time to maternal death of obstetric patients at the tertiary-level hospital of South Ethiopia
Among the 378 obstetric ICU admissions, 126 patients (33.3%) died, while 252 (66.7%) were discharged alive and considered censored cases in the survival analysis.
The mean survival time was estimated at 24.6 days (95% CI: 17.5–31.7). The median survival time estimate of 21 days (95% CI: 4.4–37.6). The median is lower than the mean because of right censoring: patients who were discharged alive contributed survival time only up to the date of discharge, but did not experience the event (death).
3.5. Time-to-event of ICU-admitted obstetric patients at a tertiary-level hospital, South Ethiopia
The Kaplan-Meier survival curve showed that a significant proportion of patients experience the event of interest (died) early during their ICU stay. Specifically, the survival probability drops sharply within the first 10 days, with approximately 60% of patients experiencing the event during this early period. Beyond 30 days, the survival probability declines more gradually, stabilizing at around 20% by day 60. (See Fig 2, Fig 3 and Table 5 for more information).
3.6. Test for equality of survival function of the event of interest of the study participants
A Log-rank test was conducted to see the difference in survival time between different predictors. Accordingly, variables including baseline GCS score, shock, and multi-organ failure had significant survival differences.
The log-rank test (Mantel-Cox) indicated a statistically significant difference in survival distributions across the three GCS categories (χ² = 78.641, df = 2, p < 0.001). This suggests that the initial GCS score on ICU admission is strongly associated with survival outcomes, with patients in the lower GCS categories experiencing significantly poorer survival compared to those with higher scores.
There was also a significant survival difference between patients with and without shock (χ² = 32.659, df = 1, p < 0.001). Patients who developed shock after ICU admission had significantly lower survival probabilities compared to those without shock, indicating that shock is a strong predictor of mortality in critically ill patients.
Furthermore, a statistically significant difference in survival distributions based on the development of multi-organ failure (χ² = 29.339, df = 1, p < 0.001). Patients who developed multi-organ failure had significantly lower survival compared to those without, highlighting MOF as a critical predictor of mortality during ICU stay. (See Figs 4–6 for more information).
3.7. Independent predictors of maternal mortality among obstetric ICU patients: Multivariable Cox regression analysis and competing-risks regression analysis
Multivariable Cox proportional hazards regression was conducted to determine independent predictors of maternal death among obstetric ICU patients. Variables that had a p-value less than 0.25 in bivariate analysis were included in the model.
In the bivariate Cox regression analysis, several factors were significantly associated with maternal mortality, including age, residence, presentation at admission, obstetric hemorrhage, HELLP syndrome, amniotic fluid embolism, gestational diabetes mellitus, cardiac disorders, respiratory disorders, use of antibiotics, use of catecholamine, administration of anticoagulants, blood product transfusions, dialysis, invasive mechanical ventilation, decreased level of consciousness (GCS), healthcare-associated infections, shock, and multi-organ failure.
After adjusting for potential confounders, several variables were found to be significantly associated with time to maternal death. Rural residency was associated with a significantly higher risk of maternal death compared to urban residency (AHR = 1.563, 95% CI: 1.026–2.371, p = 0.036). Mechanical ventilation increased the hazard of maternal death by 81.9% (AHR = 1.819, 95% CI: 1.109–2.983, p = 0.018), while having multiple organ failure (MOF) after ICU admission was associated with a 74.6% higher hazard of death (AHR = 1.746, 95% CI: 1.094–2.786, p = 0.019).
One of predictors for decreased survival was low levels of consciousness upon admission. Compared to patients with mild GCS scores (13–5), those with moderate GCS scores (9–12) had nearly a threefold higher hazard of death (AHR = 2.942, 95% CI: 1.793–4.833, p < 0.001), and patients with severe GCS scores (<9) had nearly a five folds higher hazard of death (AHR = 5.431, 95% CI: 2.903–10.124, p < 0.001). Additionally, the presence of shock after ICU admission doubled the hazard of maternal death (AHR = 2.267, 95% CI: 1.421–3.559, p < 0.001).
To account for the competing risk of discharge alive, a multivariable Fine-Gray sub-distribution hazards model was conducted using the same covariates, with discharge alive specified as the competing event. Of 378 patients, 126 experienced ICU death and 252 were discharged alive, no observations were censored in the competing-risks model.
Four predictors retained independent significance in the Fine–Gray model. Rural residency (SHR 1.55, 95% CI 1.03–2.33, p = 0.035), severe impairment of consciousness at admission (GCS < 9; SHR 2.82, 95% CI 1.66–4.80, p < 0.001), shock (SHR 2.99, 95% CI 1.97–4.55, p < 0.001), and invasive mechanical ventilation (SHR 2.27, 95% CI 1.39–3.73, p = 0.001) were each independently associated with a higher sub-distribution hazard of ICU mortality. Moderate GCS impairment (9–12) showed a borderline non-significant trend (SHR 0.50, 95% CI 0.25–1.01, p = 0.053). Multi-organ failure did not retain significance (SHR 1.55, 95% CI 0.95–2.54, p = 0.079). (See Table 6 for more information).
4. Discussion
In our study, the median survival time was 21 days (95% CI: 4.4–37.6), and the Kaplan-Meier survival curve showed a steep decline during the early days of ICU admission, indicating that many maternal deaths occurred shortly after entering intensive care. Although a specific peak mortality period was not defined, the shape of the survival curve suggests a high risk of early death among critically ill obstetric patients. Similar patterns have been observed in other studies from sub-Saharan Africa, where early ICU mortality is often linked to delayed presentation and the limited capacity of critical care units. A study in Nigeria found that a substantial number of maternal deaths occurred soon after ICU admission, primarily among patients with shock and multi-organ failure [28]. Similarly, a study conducted in South Africa reported that septic obstetric patients admitted to the ICU had high early mortality, especially within the first three days [29]. An Ethiopian study also showed that most maternal ICU deaths happened within the first week of admission, further supporting the importance of timely care [18]. These findings highlight the need for early risk stratification, prompt transfer to the ICU, and aggressive management of critically ill obstetric patients during the initial phase of care in resource-limited settings.
The study also identified key independent predictors of maternal mortality among obstetric patients admitted to the ICU. The multivariable Cox regression analysis revealed that rural residency, invasive mechanical ventilation, developed multiple organ failure (MOF), poor Glasgow Coma Scale (GCS) scores upon admission, and having post admission shock were significant predictors of increased hazard of maternal death.
The finding that rural residency was associated with lower survival aligns with previous Ethiopian studies, which attribute poorer outcomes to limited access to timely and quality obstetric care in rural areas [43]. In rural settings, delays in recognizing complications and reaching healthcare facilities significantly contribute to increased maternal mortality [44]. This disparity underscores the ongoing need for improved referral systems and enhanced critical care accessibility for rural populations.
The increased hazard of death associated with invasive mechanical ventilation reflects the severity of illness among those requiring this intervention. Mechanical ventilation is commonly used as a marker of critical respiratory failure or multisystem involvement in ICU patients, and it has been consistently reported as a predictor of mortality in obstetric ICU cohorts worldwide [45,46]. These findings highlight the importance of early identification and management of respiratory compromise to improve outcomes.
In the study, low GCS scores (< 9) upon admission also found to be predictor of mortality, underscoring the prognostic value of neurological status in critically ill obstetric patients. A low GCS typically reflects underlying cerebral hypoxia, severe metabolic derangements, or intracranial pathology, all of which compromise systemic organ perfusion and increase vulnerability to death [47]. Similar findings have been reported in Ethiopian tertiary hospitals and other resource-limited settings where delayed management of neurological complications leads to poor outcomes [48–51]. These findings emphasize the need for early neurological assessment and continuous monitoring. In low-resource settings, constraints such as limited neuroimaging, trained personnel, and delayed referrals impede timely intervention, contributing to higher mortality [8,52]. Strengthening training, resources, and protocol-driven care could markedly improve outcomes for critically ill obstetric patients.
Shock in ICU, predominantly hemorrhagic in origin, more than doubled the hazard of death in this cohort. This is consistent with other obstetric ICU studies, which identify shock as a critical determinant of maternal mortality due to its rapid progression and high fatality if not promptly managed [19,53]. Effective management of shock, including blood transfusion and hemodynamic support, remains a cornerstone of improving survival [54]. In resource-limited ICUs, delayed recognition, limited availability of blood products, and restricted access to advanced monitoring often exacerbate outcomes [55]. Similarly, developed multi-organ failure (MOF) in ICU significantly increases the risk of mortality, as shown in both the Ethiopian and the global ICU studies, reflecting the cumulative effects of serious conditions such as sepsis, hemorrhage, and shock [56–59]. In resource-limited settings, delayed recognition and inadequate availability of advanced supportive care, such as vasopressors, renal replacement therapy, and invasive monitoring, further worsen outcomes [8,60]. Shock causes tissue hypo-perfusion and cellular hypoxia, leading to metabolic acidosis and organ dysfunction [61]. Persistent hypoxia and systemic inflammation trigger multi-organ failure, impairing vital organ function. Without timely intervention, these processes culminate in irreversible organ failure and death [62,63]. Both findings highlight the common underlying problem of insufficient critical care capacity and highlight the need for protocol-based management, early detection of fragility, and investment in ICU infrastructure and staff training to improve maternal outcomes in resource-poor settings.
Our findings are consistent with our previous study conducted using the same cohort [33], which identified developed Shock, Multi-organ failure, decreased level of consciousness (GCS) and invasive mechanical ventilation as key predictors of maternal mortality. While the earlier study examined static associations, the present survival analysis incorporates the temporal dimension by evaluating time to maternal death. Assessing the timing of death provides additional insight into how rapidly these clinical conditions influence mortality risk [64,65]. The consistency of predictors across both analytical approaches strengthens the robustness of these findings among obstetric ICU patients [34].
The application of the Fine–Gray competing-risks model largely corroborated the Cox model findings, with rural residency, severe impairment of consciousness, shock, and mechanical ventilation retaining significance across both models, strengthening confidence in these associations. However, Multi-organ failure remained a significant predictor in the cause-specific Cox model but lost statistical significance in the Fine–Gray competing-risks model (SHR = 1.55, 95% CI: 0.95–2.54; p = 0.079). This difference likely reflects the influence of the competing event of discharge alive on the cumulative incidence of mortality. While the Cox model estimates the instantaneous risk of death among patients remaining at risk in the ICU, the Fine–Gray model incorporates the probability of discharge alive, which may attenuate the observed effect of multi-organ failure on mortality. These findings suggest that competing-risk methods provide additional insight into ICU outcomes in settings where discharge alive is a common and clinically meaningful event [66].
Overall, these findings illustrate that despite contextual differences, the determinants of maternal ICU mortality are largely consistent worldwide, while the magnitude of risk is intensified in low-resource environments. Addressing these challenges requires system-level interventions, including expanding ICU capacity, implementing protocol-driven management for obstetric emergencies, ensuring blood product availability, and enhancing early detection of critical illness. Strengthening referral linkages and investing in ICU staff training, particularly in rural and secondary facilities, are essential steps toward reducing preventable maternal deaths.
5. Strengths and limitations
This study is the first cohort analysis of obstetric ICU patients in the study area, providing important insights into maternal survival and its predictors using a 10-year dataset. Survival analysis strengthened the methodology by appropriately handling time-to-event data. The inclusion of a Fine -Gray competing-risks model alongside the Cox proportional hazards model is a key strength, as it accounts for discharge alive as a competing event and provides more robust estimates of mortality risk.
As a retrospective study, the findings may be affected by information bias due to incomplete or inaccurate medical records and limited availability of some clinical variables. Missing data were addressed using single imputation, which may underestimate variability and introduce bias if data were not missing completely at random. In addition, exclusion of patients with incomplete outcome data may have contributed to selection bias.
The single-center design limits external validity and generalizability to other settings. Despite these limitations, the study addresses a critical gap in obstetric critical care in low-resource settings and provides evidence to inform future research and practice.
6. Conclusion and recommendations
The study showed a high early maternal mortality among obstetric patients admitted to the ICU, with over two-thirds of deaths occurring in the first five days of admission and a short median time to death. Independent clinical factors associated with an increased hazard of maternal death included complications such as shock and organ failure, the need for invasive mechanical ventilation, rural residency, and low consciousness at admission, highlighting the importance of early recognition, rapid resuscitation, and proactive critical care for high-risk obstetric patients.
Standardized protocols for prompt identification and aggressive management of life-threatening complications such as neurological impairment, hemodynamic instability, and respiratory failure; optimization of mechanical ventilation; strengthening of triage systems; and ensuring equal access to intensive care, particularly for patients from rural areas, are the key priorities to improve outcomes. Given the short time to death, rapid assessment at ICU admission and immediate initiation of lifesaving interventions should be prioritized for all high-risk obstetric patients. Early identification of deterioration using structured monitoring tools can substantially reduce preventable deaths. These findings highlight the need for future prospective, multicenter studies that incorporate competing-risks analysis as a primary analytical approach when discharge alive is a relevant competing event for ICU mortality. Such studies should also evaluate targeted interventions addressing the identified key predictors and generate stronger evidence to inform clinical guidelines and national strategies aimed at reducing maternal mortality in low-resource settings.
Acknowledgments
The authors would like to express their sincere appreciation to all academic advisors and mentors whose guidance has contributed meaningfully to our broader research and academic development. Although they were not directly involved in the design, analysis, or preparation of this particular study, their prior mentorship and scholarly support have been instrumental in shaping the authors’ research skills and critical thinking. Their dedication to academic excellence continues to inspire and influence our ongoing work.
References
- 1.
Organization WH. International statistical classification of diseases and related health problems: 10th revision (ICD-10). 1992.http://www.whoint/classifications/apps/icd/icd
- 2. Mir E, Khan SA, Fareed P, Ashraf H, Wani FJ. Maternal demography, clinical characteristics, & outcomes at an obstetric intensive care unit of a tertiary-care teaching maternity hospital in the Kashmir Valley. Indian J Med Res. 2025;161(3):278–86. pmid:40347505
- 3. Singh V, Barik A. Maternal near-miss as a surrogate indicator of the quality of obstetric care: a study in a tertiary care hospital in Eastern India. Cureus. 2021;13(1):e12548. pmid:33564542
- 4. Sodhi K, Bansal V, Shrivastava A, Kumar M, Bansal N. Predictors of mortality in critically ill obstetric patients in a tertiary care intensive care unit: a prospective 18 months study. J Obstet Anaesth Crit Care. 2018;8(2):73.
- 5. Tesfay N, Tariku R, Zenebe A, Habtetsion M, Woldeyohannes F. Place of death and associated factors among reviewed maternal deaths in Ethiopia: a generalised structural equation modelling. BMJ Open. 2023;13(1):e060933. pmid:36697051
- 6. Beza Z, Tadesse R, Teshome H, Tadele G, Siferih M. Admission indications, initial diagnoses, Interventions, and patient outcomes within the sole obstetric high-dependency unit in Ethiopia. BMC Women’s Health. 2024;24(1):329.
- 7. Rocha FR, Gonçalves TN, Xavier-Ferreira MI, Laranjeira F, Magalhães GM, Lopes MI, et al. Obstetric intensive care admissions and neonatal outcomes: 15 years of experience from a single center. Medicina (Kaunas). 2024;60(12):1937. pmid:39768818
- 8. Abie A, Getie Mehari M, Eseyneh Dagnew T, Mebrat Delie A, Melese M, Workie Limenh L, et al. Obstetric admission and maternal mortality in the intensive care unit in Africa: a systematic review and meta-analysis. PLoS One. 2025;20(4):e0320254. pmid:40238732
- 9. Khalil A, Samara A, O’Brien P, Coutinho CM, Quintana SM, Ladhani SN. A call to action: the global failure to effectively tackle maternal mortality rates. Lancet Glob Health. 2023;11(8):e1165–7. pmid:37474218
- 10. Ceylan Y, Doğan Y, Yurtçu E, Yıldırım Köpük Ş, Medişoğlu MS. Decreasing and Maintaining Low Maternal Mortality Rate and Near Miss in Kocaeli District. Anatolian J Obstet Gynecol Res. 2024.
- 11. Yuya M, Tura AK, Girma S, Ahmed R, Knight M, van den Akker T. Factors associated with maternal mortality in eastern Ethiopia: a multicenter case-control study. Int J Gynaecol Obstet. 2025;169(2):630–8. pmid:39644178
- 12. Nam JY. How much can we reduce delivery-related medical costs associated with maternal mortality? A nationwide cohort study from 2003 to 2021. Front Public Health. 2025;13:1411534. pmid:40226323
- 13. Tesfay N, Zenebe A, Dejene Z, Tadesse H, Woldeyohannes F, Gebreyesus A, et al. Implementation status of maternal death surveillance and response system in Ethiopia: evidence from a national-level system evaluation. PLoS One. 2024;19(12):e0312958. pmid:39625947
- 14. Shiferaw MA, Bekele D, Surur F, Dereje B, Tolu LB. Maternal death review at a tertiary hospital in Ethiopia. Ethiop J Health Sci. 2021;31(1):35–42. pmid:34158750
- 15. Tesfay N, Tariku R, Zenebe A, Woldeyohannes F. Critical factors associated with postpartum maternal death in Ethiopia. PLoS One. 2022;17(6):e0270495. pmid:35749471
- 16. Tasew A, Melese E, Jemal S, Getachew L. Obstetrics mortality and associated factors in intensive care unit of Addis Ababa public hospital in, 2020/21: a hospital based case control study. Ann Med Surg (Lond). 2022;81:104458.
- 17. Jegora AA, Ahmed EH, Mohamed MS, Laytin AD, Negussie AT. Utility in the obstetric high dependency unit and intensive care unit in tertiary medical center in Ethiopia: a comparative cross-sectional study. Int J Reprod Contracept Obstet Gynecol. 2024;13(4):803–11.
- 18. Yohannes T, Yibrah B, Mache T. Obstetric ICU admissions and their outcomes in Ayder Comprehensive Specialized Hospital: Institution based retrospective study. East Afr J Health Sci. 2020;2(1):250–62.
- 19. Geleto A, Chojenta C, Taddele T, Loxton D. Magnitude and determinants of obstetric case fatality rate among women with the direct causes of maternal deaths in Ethiopia: a national cross sectional study. BMC Pregnancy Childbirth. 2020;20(1):130. pmid:32106814
- 20. Tessema GA, Laurence CO, Melaku YA, Misganaw A, Woldie SA, Hiruye A, et al. Trends and causes of maternal mortality in Ethiopia during 1990-2013: findings from the Global Burden of Diseases study 2013. BMC Public Health. 2017;17(1):160. pmid:28152987
- 21. Izedonmwen I, Izedonmwen JO. Unveiling maternal mortality challenges in a resource limited setting, Ethiopia: a systematic literature review. Br J Multidiscip Adv Stud. 2023;4(5):33–51.
- 22. Tesfay N, Hailu G, Tariku R, Firde H, Woldeyohannes FH. Inequality in maternal delays related to maternal death at home and en route to a health facility in Ethiopia: insights from national mortality surveillance data. BMJ Open. 2025;15(2):e083962. pmid:39933803
- 23. Musarandega R, Nyakura M, Machekano R, Pattinson R, Munjanja SP. Causes of maternal mortality in Sub-Saharan Africa: a systematic review of studies published from 2015 to 2020. J Glob Health. 2021;11:04048. pmid:34737857
- 24. Onambele L, Ortega-Leon W, Guillen-Aguinaga S, Forjaz MJ, Yoseph A, Guillen-Aguinaga L, et al. Maternal Mortality in Africa: Regional Trends (2000-2017). Int J Environ Res Public Health. 2022;19(20):13146. pmid:36293727
- 25.
Kleinbaum DG, Klein M. Survival analysis a self-learning text. Springer; 1996.
- 26. Bauserman M, Thorsten VR, Nolen TL, Patterson J, Lokangaka A, Tshefu A, et al. Maternal mortality in six low and lower-middle income countries from 2010 to 2018: risk factors and trends. Reprod Health. 2020;17(Suppl 3):173. pmid:33334343
- 27. Saintrain SV, Oliveira JGRD, Saintrain MVDL, Bruno ZV, Borges JLN, Daher EDF. Fatores associados à morte materna em unidade de terapia intensiva. Revista Brasileira de Terapia Intensiva. 2016;28(4):397–404.
- 28. Adeniran AS, Bolaji BO, Fawole AA, Oyedepo OO. Predictors of maternal mortality among critically ill obstetric patients. Malawi Med J. 2015;27(1):16–9. pmid:26137193
- 29. Lafon J, Buga E, Nethathe G. Characteristics and outcomes of obstetric patients with maternal sepsis requiring admission to a South African intensive care unit: a retrospective review. South Afr J Obst Gynaecol. 2021;26(3):89.
- 30. V VM, T JM, B PR. Severe obstetric morbidity (near miss). J Clin Obstet Gynecol Infertil. 2019;3(2):1043.
- 31. Kumar R, Gupta A, Suri T, Suri J, Mittal P, Suri JC. Determinants of maternal mortality in a critical care unit: A prospective analysis. Lung India. 2022;39(1):44–50. pmid:34975052
- 32. Jegora AA, Ahmed EH, Mohamed MS, Laytin AD, Negussie AT. Utility in the obstetric high dependency unit and intensive care unit in tertiary medical center in Ethiopia: a comparative cross-sectional study. Int J Reprod Contracept Obstet Gynecol. 2024;13(4):803–11.
- 33. Alemu TN, Shiferaw WG, Daga WB, Sisay Y, Balcha B. Determinants of maternal mortality among obstetric patients admitted to intensive care unit of Wolaita Sodo comprehensive specialized hospital, Southern Ethiopia: unmatched case-control study. BMC Pregnancy Childbirth. 2025;25(1):878. pmid:40849460
- 34. Waithira N, Kestelyn E, Chotthanawathit K, Osterrieder A, Mukaka M, Lang T, et al. Investigating the secondary use of clinical research data: protocol for a mixed methods study. JMIR Res Protoc. 2023;12:e44875. pmid:36877564
- 35. Mamo TF, Zenebe NH, Jemberie Getnet Y, Mada Mazga P, Letta Desisa S, Argeta Hailemariam H, et al. Prevalence and associated factors of prolonged emergency department length of stay among adults patients attending University Hospital, South Ethiopia: a cross-sectional study. Int J Afr Nurs Sci. 2026;24:100958.
- 36. Surekha T, Neha G, Poonam S, Dinesh B, Apurva R, Himanshu B, et al. Role of obstetric high dependency and intensive care unit in improving pregnancy outcome and reducing maternal mortality-a study in rural central India. Int J Crit Care Emerg Med. 2018;4(2).
- 37. Tasew A, Melese E, Jemal S, Getachew L. Obstetrics mortality and associated factors in intensive care unit of Addis Ababa public hospital in, 2020/21: A hospital based case control study. Ann Med Surg (Lond). 2022;81:104458. pmid:36147061
- 38.
Mideksa T, Mekonnen T, Mengiste B. Outcomes and Associated Factors of Mothers Admitted to Intensive Care Unit During Pregnancy and Postpartum at Saint Paul’s Hospital Millennium Medical College, Addis Ababa, Ethiopia. 2022.
- 39. Oliveira Neto AF, Parpinelli MA, Cecatti JG, Souza JP, Sousa MH. Factors associated with maternal death in women admitted to an intensive care unit with severe maternal morbidity. Int J Gynaecol Obstet. 2009;105(3):252–6. pmid:19342049
- 40. Asfar P, Radermacher P, Ostermann M. MAP of 65: target of the past? Intensive Care Med. 2018;44(9):1551–2.
- 41.
Holzheimer RG, M J. The multiple organ dysfunction syndrome. In: Holzheimer RG, editor. Surgical Treatment: Evidence-Based and Problem-Oriented. Munich: Zuckschwerdt; 2001.
- 42. Teasdale G, Jennett B. Assessment of coma and impaired consciousness. A practical scale. Lancet. 1974;2(7872):81–4. pmid:4136544
- 43. Melberg A, Teklemariam L, Moland KM, Aasen HS, Sisay MM. Juridification of maternal deaths in Ethiopia: a study of the Maternal and Perinatal Death Surveillance and Response (MPDSR) system. Health Policy Plan. 2020;35(8):900–5. pmid:32594165
- 44. Habte A, Wondimu M. Determinants of maternal near miss among women admitted to maternity wards of tertiary hospitals in Southern Ethiopia, 2020: a hospital-based case-control study. PLoS One. 2021;16(5):e0251826. pmid:33999941
- 45. Bekele TG, Melaku B, Demisse LB, Abza LF, Assen AS. Outcomes and factors associated with prolonged stays among patients admitted to adult intensive care unit in a resource-limited setting: a multicenter chart review. Sci Rep. 2024;14(1):13960. pmid:38886468
- 46. Asmare TB, Gobezie NZ, Wubet HB, Diress GM, Abuhay AG, Melesew AA, et al. Severity and factors associated with pain in patients on mechanical ventilators in Amhara region, North-West Ethiopia: a multi-center prospective observational study. Sci Rep. 2025;15(1):22135. pmid:40592995
- 47. Fadiloglu E, Bulut Yuksel ND, Unal C, Ocal S, Akinci SB, Topeli A, et al. Characteristics of obstetric admissions to intensive care unit: APACHE II, SOFA and the Glasgow Coma Scale. J Perinat Med. 2019;47(9):947–57. pmid:31603858
- 48. Debebe F, Goffi A, Haile T, Alferid F, Estifanos H, Adhikari NKJ. Predictors of ICU mortality among mechanically ventilated patients: an inception cohort study from a tertiary care center in Addis Ababa, Ethiopia. Crit Care Res Pract. 2022;2022:7797328. pmid:36533249
- 49. Endeshaw AS, Tarekegn F, Bayu HT, Ayalew SB, Gete BC. The magnitude of mortality and its determinants in Ethiopian adult intensive care units: a systematic review and meta-analysis. Ann Med Surg (Lond). 2022;84:104810. pmid:36582907
- 50. Kebede F, Mosisa G, Yilma M. Incidence and predictors of mortality among patients admitted to adult intensive care unit at public hospitals in Western Ethiopia: a retrospective cohort study. Front Med (Lausanne). 2024;11:1370729. pmid:39635586
- 51. Adeniran AS, Bolaji BO, Fawole AA, Oyedepo OO. Predictors of maternal mortality among critically ill obstetric patients. Malawi Med J. 2015;27(1):16–9. pmid:26137193
- 52. Vasco M, Pandya S, Van Dyk D, Bishop DG, Wise R, Dyer RA. Maternal critical care in resource-limited settings. Int J Obstet Anesth. 2019;37:86–95.
- 53. Rudakemwa A, Cassidy AL, Twagirumugabe T. High mortality rate of obstetric critically ill women in Rwanda and its predictability. BMC Pregnancy Childbirth. 2021;21(1):401. pmid:34034687
- 54. Desta M, Ferede AA. Mortality rate and predictors among women with obstructed labor in a tertiary academic medical center of Ethiopia: a retrospective cohort study. SAGE Open Nurs. 2023;9. pmid:37101828
- 55. Njolomole SE, Sachidanandan RF, Mandere G, Jenny A, Muula AS, M’baya B, et al. Meeting demand-Obstetric hemorrhage and blood availability in Malawi, a qualitative study. PLoS One. 2022;17(8):e0273426. pmid:36001581
- 56.
Liyew AD, Abebaw N, Geremew MA, Ahmed L, Simegne YN, Kettema WG. Utilization of non-pneumatic anti-shock garment and associated factors for postpartum hemorrhage management among health care providers in Sub-Saharan Africa: systematic review and meta-analysis. 2024.
- 57. Vasquez DN, Estenssoro E, Canales HS, Reina R, Saenz MG, Das Neves AV, et al. Clinical characteristics and outcomes of obstetric patients requiring ICU admission. Chest. 2007;131(3):718–24. pmid:17356085
- 58. Agarwal M, Bhushan D, Singh S, Singh S. Determinants of survival in obstetric sepsis: retrospective observational study. J Obstet Gynaecol India. 2022;72(Suppl 1):159–65. pmid:35928076
- 59. Pérez A, Acevedo O, Tamayo F del C, Oviedo R. Characterization of obstetric patients with multiple organ failure in the intensive care unit of a Havana teaching hospital, 1998 to 2006. MEDICC Rev. 2010;12(2):27–32. pmid:20486411
- 60. Soares FM, Pacagnella RC, Tunçalp Ö, Cecatti JG, Vogel JP, Togoobaatar G, et al. Provision of intensive care to severely ill pregnant women is associated with reduced mortality: Results from the WHO Multicountry Survey on Maternal and Newborn Health. Int J Gynaecol Obstet. 2020;150(3):346–53. pmid:32464683
- 61. Vincent J-L, De Backer D. Circulatory shock. N Engl J Med. 2013;369(18):1726–34. pmid:24171518
- 62. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801–10. pmid:26903338
- 63. Iba T, Helms J, Connors JM, Levy JH. The pathophysiology, diagnosis, and management of sepsis-associated disseminated intravascular coagulation. J Intensive Care. 2023;11(1):24. pmid:37221630
- 64. Merdad L, Ali MM. Timing of maternal death: Levels, trends, and ecological correlates using sibling data from 34 sub-Saharan African countries. PLoS One. 2018;13(1):e0189416. pmid:29342157
- 65. Dol J, Hughes B, Bonet M, Dorey R, Dorling J, Grant A, et al. Timing of maternal mortality and severe morbidity during the postpartum period: a systematic review. JBI Evid Synth. 2022;20(9):2119–94. pmid:35916004
- 66. Putter H, Schumacher M, van Houwelingen HC. On the relation between the cause-specific hazard and the subdistribution rate for competing risks data: The Fine-Gray model revisited. Biom J. 2020;62(3):790–807. pmid:32128860