Abstract
Background
Neonatal mortality remains a critical public health challenge in Sub-Saharan Africa, with The Gambia reporting one of the highest rates in the region. Despite global progress, the neonatal mortality rate has shown limited decline, and context-specific data on delivery setting-related risk factors remain needed to inform health system strengthening in The Gambia. We aimed to identify risk factors for neonatal mortality at Kanifing General Hospital (KGH), and examine the association between place of delivery and survival outcomes.
Methods
We conducted a retrospective cohort study of 1,354 neonates admitted to KGH between January 2022 and December 2024. The primary outcome was neonatal death during admission. Kaplan-Meier survival analysis and Cox proportional hazards regression with robust standard errors clustered by facility were used to identify risk factors.
Results
During the study period, 137 deaths occurred (10.1% mortality; 95% CI: 8.6–11.9). Median follow-up was 96 h. Overall, survival at 7 days was 87.8%, but differed substantially between neonates delivered at KGH Labour Ward (91.3%) versus peripheral facilities (78.9%; log-rank p < 0.001). In multivariable analysis, delivery at KGH Labour Ward was strongly protective (adjusted hazard ratio [aHR] 0.41; 95% CI: 0.28–0.58; p < 0.001), representing a 59% reduction in mortality risk. Higher birth weight was protective (aHR 0.58 per kg; 95% CI: 0.37–0.90; p = 0.016), while birth asphyxia increased mortality risk (aHR 1.30; 95% CI: 1.11–1.53; p = 0.001). The model demonstrated acceptable discrimination (C-statistic = 0.713). E-value sensitivity analysis indicated robustness to unmeasured confounding (E-value = 4.36 for facility effect). Study limitations include the retrospective design and single-center setting.
Conclusions
Neonates referred from peripheral facilities experienced substantially higher mortality compared to those delivered at the referral hospital. Birth weight and birth asphyxia were independently associated with survival. These findings highlight the critical importance of strengthening referral hospital capacity and improving quality of care at peripheral facilities to reduce neonatal mortality in The Gambia.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12887-026-06569-0.
Keywords: Neonatal mortality, The gambia, Survival analysis, Peripheral facilities, Referral hospital
Introduction
The first 28 days of life, the neonatal period represent the most vulnerable time for child survival. Globally, an estimated 2.3 million neonates died in 2023, accounting for 48% of all under-five deaths [1, 2]. While substantial progress has been made in reducing child mortality over the past three decades, the decline in neonatal mortality has been slower than that of post-neonatal mortality, and the proportion of child deaths occurring in the neonatal period continues to increase [1, 3]. The Sustainable Development Goals (SDGs) call for all countries to reduce neonatal mortality to at least 12 per 1,000 live births by 2030; however, current trajectories suggest that more than 60 countries will fail to meet this target without accelerated action [3]. Sub-Saharan Africa bears a disproportionate burden of neonatal mortality, with rates approximately 50% higher than the global average [2, 2]. The region accounts for 43% of global neonatal deaths despite representing only 14% of global births [4]. The leading causes of neonatal death in Sub-Saharan Africa (SSA) include preterm birth complications, intrapartum-related events (including birth asphyxia), and neonatal sepsis, which are largely preventable with known, cost-effective interventions [4, 5]. However, weak health systems, limited access to quality care, and inadequate referral networks continue to impede progress [6, 7].
The Gambia, a small West African nation, faces significant challenges in neonatal health. According to the 2019–2020 Demographic and Health Survey, the neonatal mortality rate was 29 deaths per 1,000 live births, contributing to nearly half of under-five mortality [8]. This rate represents an increase from 22 per 1,000 reported in 2013, indicating a reversal of earlier gains in neonatal survival [8]. Common causes of neonatal death in The Gambia include neonatal sepsis, birth asphyxia, and complications of prematurity [8, 9]. Despite these challenges, epidemiological data on facility-level risk factors for neonatal mortality remain scarce, limiting evidence-based planning for health system strengthening.
Place of delivery and level of facility care have emerged as important determinants of neonatal survival in low- and middle-income countries [10, 11]. A large ecological study across 37 countries in SSA and South Asia found that regions with higher proportions of hospital deliveries (compared to lower-level clinics) had significantly reduced neonatal mortality [11]. Similarly, systematic reviews have demonstrated protective effects of health facility delivery on neonatal survival, with the magnitude of benefit varying by context and quality of care [12, 13]. However, these associations are complex; in settings with poor quality of care, facility delivery may not confer the expected survival advantage, and referral from peripheral facilities often indicates underlying complications that themselves increase mortality risk [11, 14]. Birth asphyxia remains a leading cause of neonatal mortality in SSA, accounting for an estimated 23% of neonatal deaths in the region [7, 15]. A recent meta-analysis reported that neonates with birth asphyxia in SSA had nearly four-fold higher odds of early neonatal mortality compared to those without asphyxia [16]. The burden of birth asphyxia is particularly high in settings with limited access to skilled birth attendance, emergency obstetric care, and neonatal resuscitation [7, 17]. Understanding the contribution of birth asphyxia to neonatal mortality in specific facility contexts is essential for developing targeted interventions.
Kanifing General Hospital (KGH) is the largest referral hospital in the Greater Banjul Area of The Gambia, serving a catchment population of over 400,000 and receiving referrals from peripheral health facilities across the West Coast Region. The hospital’s neonatal unit provides both basic and comprehensive essential newborn care services. However, to date, no comprehensive survival analysis has examined risk factors for neonatal mortality at this facility or quantified the impact of referral patterns on neonatal outcomes. We conducted this study to: (1) describe temporal trends in neonatal mortality at KGH from 2022 to 2024; (2) identify demographic, clinical, and delivery setting-related risk factors for neonatal death using survival analysis methods; and (3) quantify the association between place of delivery (KGH Labour Ward versus peripheral facilities) and neonatal survival. These findings aim to inform evidence-based strategies for reducing neonatal mortality in The Gambia and similar resource-limited settings in West Africa.
Methods
Study design and setting
We conducted a retrospective cohort study of neonates admitted to Kanifing General Hospital (KGH), the largest referral hospital in the Greater Banjul Area of The Gambia, between January 2022 and December 2024. KGH serves as the primary referral center for the Kanifing Municipal Area and receives referrals from peripheral health facilities across the West Coast Region. The hospital’s neonatal unit provides both basic and comprehensive essential newborn care services. The Gambia’s health system is organized into three tiers: community-level health posts, health centers (minor and major), and hospitals. Peripheral health facilities in this study included 12 minor health centers, 4 major health centers, and one district hospital within KGH’s catchment area (Supplementary Table S1). Neonates delivered at home who were subsequently referred to KGH were included in the “peripheral facilities” category if the referring facility was documented; direct home-to-hospital admissions (n = 23, 1.7%) were also categorized under “peripheral facilities” for the primary analysis, as these neonates similarly lacked immediate access to hospital-level delivery care.
Study population
All neonates (aged 0–28 days) admitted to the KGH neonatal unit during the study period were eligible for inclusion. Neonates with missing data on survival time or vital status were excluded from the primary analysis. The final analytical sample comprised 1,354 neonates with complete data on the primary outcome and key covariates.
Data sources and variables
Data were extracted from hospital admission registers and patient case files. The primary outcome was neonatal death during hospital admission. Follow-up time was defined as hours from admission until death or discharge (administrative censoring), capped at 28 days (672 h) to capture the complete neonatal period. Covariates were selected a priori based on clinical relevance and prior literature, including: sex; birth weight (kg): For neonates delivered at KGH Labour Ward, birth weight was recorded immediately at delivery. For referred neonates, birth weight was obtained from referral documentation when available (78% of cases) or measured at KGH admission (22% of cases); gestational age (term, preterm, post-term); place of delivery (KGH Labour Ward versus peripheral health facilities); mode of delivery (spontaneous vaginal delivery, cesarean section, vacuum-assisted); and clinical diagnoses at admission (neonatal sepsis, birth asphyxia, prematurity/low birth weight, congenital abnormalities). A symptom severity score was calculated as the sum of presenting danger signs documented at admission, including respiratory distress, poor feeding, lethargy, convulsions, fever, hypothermia, jaundice, and cyanosis, with each symptom scored as 1 if present and 0 if absent (range: 0–8).
Statistical analysis
Baseline characteristics were summarized using frequencies and percentages for categorical variables and medians with interquartile ranges (IQR) for continuous variables, stratified by survival status. Differences between survivors and non-survivors were assessed using chi-square tests for categorical variables and Wilcoxon rank-sum tests for continuous variables.
Survival analysis
Kaplan-Meier methods were used to estimate survival probabilities at clinically relevant time points (24 h, 72 h, 7 days, 14 days, and 28 days). Survival curves were compared between groups using the log-rank test. Cox proportional hazards regression was used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for factors associated with neonatal mortality. Birth weight was modeled as a continuous variable (per kilogram increase) after confirming linearity using martingale residual plots. To account for potential correlation of outcomes within delivery facilities, we computed robust standard errors using the Huber-White sandwich estimator clustered at the facility level.
Variables associated with neonatal death at p < 0.25 in univariable analysis were considered for multivariable modeling. The final multivariable model included sex, birth weight, gestational age, place of delivery, mode of delivery, and clinical diagnoses (neonatal sepsis, birth asphyxia, prematurity). Model discrimination was assessed using Harrell’s concordance index (C-statistic). Statistical analyses were performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria) and Python version 3.12. Survival analysis was conducted using the survival and survminer packages in R and the lifelines package in Python. Publication-ready tables were generated using gtsummary and flextable. All statistical tests were two-sided with significance set at α = 0.05.
The proportional hazards assumption was evaluated using scaled Schoenfeld residuals with both global and variable-specific tests. Complementary log-log plots were visually inspected to confirm parallel hazard functions across exposure groups. Variables violating the proportional hazards assumption (p < 0.05 on the Schoenfeld residual test) were addressed through stratification.
Sensitivity analyses
We conducted several sensitivity analyses to assess the robustness of our findings: Complete case analysis: We restricted analysis to participants with complete data on all covariates to assess the impact of missing data. E-value calculation: To quantify susceptibility to unmeasured confounding, we calculated E-values for statistically significant associations. Time-varying analysis: We examined risk factors separately for the early neonatal period (0–7 days) and late neonatal period (7–28 days) using episode splitting.
Ethical considerations
This study was approved by Ministry of Health under FETP program (ref #: DHS/AD/2025/3/(MTN)). The study used de-identified secondary data from routine health facility records; individual informed consent was waived given the retrospective nature and minimal risk. All data were handled according to the Gambian Data Protection Act and international standards for data security. Patient information was confidential and de-identified using numeric codes. This study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Results
Study population and baseline characteristics
Between January 2022 and December 2024, 1,393 neonates were admitted to KGH. After excluding 39 (2.8%) with missing data on survival time or vital status, 1,354 neonates were included in the analysis (Table 1). The median age at admission was 24 h (IQR: 4–72), and the median birth weight was 2.80 kg (IQR: 2.36–3.30). Males comprised 59.4% (n = 804) of the cohort. The majority of neonates (74.3%, n = 1,006) were delivered at the KGH Labour Ward, while 25.7% (n = 348) were referred from peripheral health facilities. Among neonates with documented gestational age, 78.8% were term (n = 767), 18.9% were preterm (n = 184), and 2.4% were post-term (n = 23). Low birth weight (< 2.5 kg) was present in 29.5% of neonates. The most common admission diagnoses were neonatal sepsis (47.3%), birth asphyxia (28.4%), and prematurity/low birth weight (14.9%). Neonates who died had significantly lower birth weight (median 2.50 vs. 2.90 kg), were more likely to have low birth weight (49.6% vs. 27.3%), and were more frequently referred from peripheral facilities (47.4% vs. 23.3%) compared to survivors. The median age at admission was lower among non-survivors (19.5 h) than survivors (24 h), though this difference was not statistically significant (Wilcoxon rank-sum p = 0.312). This pattern likely reflects confounding by indication, whereby neonates with severe illness at birth (such as severe birth asphyxia) require urgent admission and have inherently higher mortality risk regardless of admission timing.
Table 1.
Baseline characteristics of neonates admitted to Kanifing general Hospital, the Gambia (2022–2024), by survival status
| Characteristic | Survived (n = 1,217) | Died (n = 137) | Total (N = 1,354) |
|---|---|---|---|
| Demographics | |||
| Male sex, n (%) | 732 (60.1) | 72 (52.6) | 804 (59.4) |
| Female sex, n (%) | 485 (39.9) | 65 (47.4) | 550 (40.6) |
| Clinical measurements | |||
| Birth weight (kg), median (IQR) | 2.90 (2.40–3.30) | 2.50 (2.00–2.85) | 2.80 (2.36–3.30) |
| Low birth weight (< 2.5 kg), n (%) | 332 (27.3) | 68 (49.6) | 400 (29.5) |
| Age at admission (hours), median (IQR) | 24.0 (4.0–72.0) | 19.5 (4.0–72.0) | 24.0 (4.0–72.0) |
| Gestational age, n (%) | |||
| Term (≥ 37 weeks) | 710 (79.9) | 57 (67.1) | 767 (78.8) |
| Preterm (< 37 weeks) | 162 (18.2) | 22 (25.9) | 184 (18.9) |
| Post-term (> 42 weeks) | 17 (1.9) | 6 (7.1) | 23 (2.4) |
| Place of delivery, n (%) | |||
| KGH Labour Ward | 934 (76.7) | 72 (52.6) | 1,006 (74.3) |
| Peripheral health facilities | 283 (23.3) | 65 (47.4) | 348 (25.7) |
| Mode of delivery, n (%) | |||
| Spontaneous vaginal delivery | 839 (71.4) | 92 (74.2) | 931 (71.7) |
| Cesarean section | 304 (25.9) | 28 (22.6) | 332 (25.6) |
| Vacuum-assisted delivery | 32 (2.7) | 4 (3.2) | 36 (2.8) |
| Clinical diagnoses, n (%) | |||
| Neonatal sepsis | 575 (47.2) | 65 (47.4) | 640 (47.3) |
| Birth asphyxia | 334 (27.4) | 51 (37.2) | 385 (28.4) |
| Prematurity/LBW | 171 (14.1) | 31 (22.6) | 202 (14.9) |
| Congenital abnormalities | 18 (1.5) | 7 (5.1) | 25 (1.8) |
| Admission outcomes | |||
| Symptom severity score, median (IQR) | 1 (1–1) | 1 (1–2) | 1 (1–1) |
| Duration of admission (hours), median (IQR) | 96.0 (72.0–120.0) | 48.0 (13.0–72.0) | 96.0 (72.0–120.0) |
Symptom severity score calculated as sum of presenting danger signs (respiratory distress, poor feeding, lethargy, convulsions, fever, hypothermia, jaundice, cyanosis); each symptom scored 1 if present, 0 if absent
Abbreviations: IQR interquartile range, KGH Kanifing General Hospital, LBW low birth weight
Neonatal mortality and survival estimates
During the study period, 137 neonatal deaths occurred, yielding an in-hospital mortality rate of 10.1% (95% CI: 8.6–11.9). The median duration of admission was 96 h (IQR: 72–120) among survivors and 48 h (IQR: 13–72) among non-survivors. Most deaths (60.6%, n = 83) occurred within the first 24 h of admission. Kaplan-Meier survival estimates at key time points were: 95.0% at 24 h (95% CI: 93.8–96.2), 92.0% at 72 h (95% CI: 90.4–93.5), 87.8% at 7 days (95% CI: 85.9–89.6), 78.1% at 14 days (95% CI: 75.5–80.8), and 61.4% at 28 days (95% CI: 57.3–65.7). Survival differed substantially by place of delivery (Fig. 1A). Neonates delivered at KGH Labour Ward had significantly higher 7-day survival (91.3%, 95% CI: 89.4–93.1) compared to those referred from peripheral facilities (78.9%, 95% CI: 74.5–83.5; log-rank p < 0.001). Survival also varied by birth weight category (Fig. 1B), with 7-day survival of 63.2% for neonates weighing < 1.5 kg, 85.1% for 1.5–2.5 kg, and 93.4% for > 2.5 kg (log-rank p < 0.001). Among neonates delivered at KGH Labour Ward, 47 (4.7%) were initially discharged home and subsequently readmitted to the neonatal unit within the first 7 days of life. Mortality among these readmitted neonates was significantly higher than among those admitted directly from the Labour Ward (8/47 [17.0%] vs. 64/959 [6.7%]; chi-square p = 0.007), suggesting potential gaps in discharge assessment or postnatal care practices (Supplementary Table S2).
Fig. 1.
Kaplan-Meier survival curves by facility type and birthweight. A By place of delivery (KGH Labour Ward vs. peripheral facilities). B By birth weight category (< 1.5 kg, 1.5–2.5 kg, > 2.5 kg). Shaded areas represent 95% confidence intervals
Risk factors for neonatal mortality
In univariable Cox regression (Table 2), factors significantly associated with neonatal mortality included male sex (HR 0.71, 95% CI: 0.51–0.99; p = 0.044), higher birth weight (HR 0.56 per kg, 95% CI: 0.44–0.71; p < 0.001), low birth weight (HR 2.13, 95% CI: 1.52–2.99; p < 0.001), and delivery at KGH Labour Ward versus peripheral facilities (HR 0.41, 95% CI: 0.29–0.57; p < 0.001). In multivariable Cox regression with robust standard errors clustered by facility, three factors remained independently associated with neonatal mortality: Higher birth weight was protective, with each kilogram increase associated with a 42% reduction in the hazard of death (adjusted HR [aHR] 0.58, 95% CI: 0.37–0.90; p = 0.016). Delivery at KGH Labour Ward was strongly protective compared to referral from peripheral facilities (aHR 0.41, 95% CI: 0.28–0.58; p < 0.001), representing a 59% reduction in mortality risk. Birth asphyxia was significantly associated with increased mortality (aHR 1.30, 95% CI: 1.11–1.53; p = 0.001). Sex, gestational age, cesarean delivery, neonatal sepsis, and symptom severity were not independently associated with mortality after adjustment. The model demonstrated acceptable discrimination with a concordance index of 0.713 (Fig. 2).
Table 2.
Cox proportional hazards regression of factors associated with neonatal mortality at Kanifing general Hospital, the Gambia (2022–2024)
| Variable | HR (95% CI) | P-value | aHR (95% CI) | P-value |
|---|---|---|---|---|
| Univariable | Multivariable | |||
| Male sex | 0.71 (0.51–0.99) | 0.044 | 0.76 (0.54–1.06) | 0.106 |
| Birth weight (per kg increase) | 0.56 (0.44–0.71) | < 0.001 | 0.58 (0.37–0.90) | 0.016 |
| Preterm (< 37 weeks) | 0.98 (0.62–1.55) | 0.919 | 0.68 (0.39–1.18) | 0.166 |
| Low birth weight (< 2.5 kg) | 2.13 (1.52–2.99) | < 0.001 | 1.56 (0.87–2.81) | 0.139 |
| KGH Labour Ward (vs. peripheral) | 0.41 (0.29–0.57) | < 0.001 | 0.41 (0.28–0.58) | < 0.001 |
| Cesarean delivery | 0.81 (0.53–1.23) | 0.318 | 1.17 (0.75–1.81) | 0.495 |
| Neonatal sepsis | 1.10 (0.79–1.55) | 0.561 | 1.10 (0.76–1.59) | 0.629 |
| Birth asphyxia | 1.23 (0.99–1.52) | 0.059 | 1.30 (1.11–1.53) | 0.001 |
| Prematurity diagnosis | 1.36 (0.91–2.03) | 0.138 | 0.75 (0.42–1.35) | 0.345 |
| Symptom severity score | 0.97 (0.92–1.01) | 0.155 | 0.96 (0.91–1.01) | 0.145 |
Bold p-values indicate statistical significance (p < 0.05). Multivariable model adjusted for all variables shown with robust standard errors clustered by facility. n = 1,354; 137 events. Concordance index = 0.713
Abbreviations: aHR adjusted hazard ratio, CI confidence interval, HR hazard ratio, KGH Kanifing General Hospital
Fig. 2.
Forest plot of adjusted hazard ratios from multivariable Cox proportional hazards regression. Red points indicate statistically significant associations (p < 0.05). Horizontal lines represent 95% confidence intervals
Model diagnostics and sensitivity analyses
The global test of the proportional hazards assumption was not statistically significant (p = 0.124), indicating no strong evidence of violation. Visual inspection of scaled Schoenfeld residual plots (Fig. 3) supported the proportional hazards assumption for all covariates, with no systematic departures from horizontal patterns over time.
Fig. 3.
Scaled Schoenfeld residual plots for proportional hazards assessment. Each panel displays the scaled Schoenfeld residuals plotted against time (hours) for the corresponding covariate. The red line represents a LOWESS smoothed trend. A horizontal pattern around zero suggests the proportional hazards assumption is met. The global test of the proportional hazards assumption was not statistically significant (p = 0.124)
Time-varying analysis stratified by neonatal period (Table 3) confirmed that the protective effect of KGH delivery was consistent in both the early neonatal period (0–7 days: aHR 0.38, 95% CI: 0.26–0.56; p < 0.001) and late neonatal period (7–28 days: aHR 0.78, 95% CI: 0.21–2.86; p = 0.705), though the latter estimate was imprecise due to fewer events (n = 12). Birth asphyxia showed a stronger association with mortality in the early neonatal period (aHR 1.29, 95% CI: 1.10–1.51; p = 0.001) compared to the late period (aHR 0.93, 95% CI: 0.14–6.02; p = 0.936), consistent with the acute nature of asphyxia-related mortality.
Table 3.
Time-varying analysis of risk factors by neonatal period
| Variable | Early neonatal (0–7 days) | Late neonatal (7–28 days) | ||||
|---|---|---|---|---|---|---|
| aHR | 95% CI | P-value | aHR | 95% CI | P-value | |
| Male sex | 0.81 | (0.57–1.16) | 0.249 | 0.32 | (0.09–1.22) | 0.095 |
| Birth weight (per kg) | 0.53 | (0.33–0.85) | 0.009 | 1.17 | (0.24–5.62) | 0.842 |
| Preterm (< 37 weeks) | 0.54 | (0.29–1.01) | 0.053 | 2.28 | (0.40–12.92) | 0.352 |
| Low birth weight | 1.44 | (0.78–2.66) | 0.241 | 2.89 | (0.28–30.19) | 0.376 |
| KGH Labour Ward | 0.38 | (0.26–0.56) | < 0.001 | 0.78 | (0.21–2.86) | 0.705 |
| Cesarean delivery | 1.21 | (0.76–1.91) | 0.419 | 0.57 | (0.09–3.50) | 0.544 |
| Neonatal sepsis | 1.14 | (0.78–1.68) | 0.505 | 0.32 | (0.05–2.15) | 0.239 |
| Birth asphyxia | 1.29 | (1.10–1.51) | 0.001 | 0.93 | (0.14–6.02) | 0.936 |
| Prematurity diagnosis | 0.79 | (0.43–1.45) | 0.442 | 0.39 | (0.05–3.38) | 0.395 |
| Symptom severity | 0.96 | (0.91–1.01) | 0.147 | 0.93 | (0.48–1.82) | 0.835 |
Bold p-values indicate statistical significance (p < 0.05). Models adjusted for all variables shown. Wide confidence intervals in the late neonatal period reflect the smaller number of events (n = 12)
Cox proportional hazards regression stratified by early neonatal period (0–7 days; n = 1,354, 125 events) and late neonatal period (7–28 days; n = 148, 12 events). The late neonatal analysis was restricted to neonates who survived beyond 7 days
Abbreviations: aHR adjusted hazard ratio, CI confidence interval, KGH Kanifing General Hospital
E-value sensitivity analysis for unmeasured confounding (Table 4) indicated that the observed associations were reasonably robust. For the protective effect of KGH delivery (aHR 0.41), an unmeasured confounder would need to be associated with both the exposure and outcome by a risk ratio of at least 4.36 (E-value for point estimate) or 2.81 (E-value for confidence interval limit) to fully explain the association. For birth asphyxia (aHR 1.30), the corresponding E-values were 1.93 and 1.46, respectively. These findings suggest that substantial unmeasured confounding would be required to nullify the observed effects.
Table 4.
E-value sensitivity analysis for unmeasured confounding
| Variable | aHR (95% CI) | P-value | Direction | E-value | E-value (CI) |
|---|---|---|---|---|---|
| Birth weight (per kg increase) | 0.58 (0.37–0.90) | 0.016 | Protective | 2.86 | 1.45 |
| KGH Labour Ward (vs. peripheral) | 0.41 (0.28–0.58) | < 0.001 | Protective | 4.36 | 2.81 |
| Birth asphyxia | 1.30 (1.11–1.53) | 0.001 | Harmful | 1.93 | 1.46 |
E-values quantify the minimum strength of association that an unmeasured confounder would need to have with both the exposure and outcome, conditional on measured covariates, to fully explain the observed association. Higher E-values indicate greater robustness to unmeasured confounding
Abbreviations: aHR adjusted hazard ratio, CI confidence interval. E-value (point) corresponds to the point estimate; E-value (CI) corresponds to the confidence interval limit closest to the null. For protective associations (aHR < 1), E-values are calculated using 1/aHR. Interpretation: an unmeasured confounder would need to be associated with both exposure and outcome by a risk ratio of at least the E-value magnitude to explain away the observed association
Interpretation
The E-value analysis suggests that the observed associations are reasonably robust to unmeasured confounding:
KGH Labour Ward effect: An unmeasured confounder would need to be associated with both place of delivery and neonatal mortality by a risk ratio of 4.36 to fully explain the protective effect (E-value for CI limit = 2.81). This suggests substantial robustness to unmeasured confounding.
Birth weight effect: An unmeasured confounder would need risk ratio associations of 2.86 (point estimate) or 1.45 (CI limit) to explain the protective effect of higher birth weight.
Birth asphyxia effect: The harmful association with birth asphyxia has E-values of 1.93 (point) and 1.46 (CI limit), indicating moderate robustness to unmeasured confounding.
Discussion
In this retrospective cohort study of 1,354 neonates admitted to KGH, we observed an in-hospital mortality rate of 10.1%, with most deaths (60.6%) occurring within the first 24 h of admission. Three factors were independently associated with neonatal mortality: delivery at the referral hospital (protective), higher birth weight (protective), and birth asphyxia (harmful). Most notably, neonates referred from peripheral health facilities experienced a 59% higher hazard of death compared to those delivered at KGH Labour Ward, even after adjustment for clinical characteristics. This facility-level disparity persisted across both early and late neonatal periods and was robust to sensitivity analyses for unmeasured confounding.
Our finding that delivery at the referral hospital was strongly protective is consistent with ecological evidence from SSA. Gage et al. [10] reported that a 10-percentage point increase in the proportion of facility deliveries occurring in hospitals (versus lower-level clinics) was associated with 1.2 fewer deaths per 1,000 births across 37 countries. Our effect estimate (aHR 0.41) suggests an even stronger association in this single-facility context, which may reflect the substantial quality differential between KGH and peripheral facilities in The Gambia. The referral hospital has continuous access to skilled personnel, resuscitation equipment, and emergency interventions that may be intermittently available at peripheral sites.
The protective effect of referral hospital delivery likely operates through multiple mechanisms. First, neonates born at KGH benefit from immediate access to skilled birth attendants and neonatal resuscitation if required [17]. Second, complications identified during labor can be managed promptly without the delays inherent in referral processes. Third, the quality of postnatal monitoring and early intervention for deteriorating neonates is likely superior at the referral hospital. However, our findings also reflect selection effects: neonates referred from peripheral facilities may have experienced delays in recognition, referral, and transport that contributed to their poorer outcomes independent of facility quality [18, 19]. The inability to fully disentangle quality effects from referral selection represents an important limitation of observational studies in this context.
Our findings contrast with some studies from settings with poor quality of facility care, where facility delivery does not consistently confer a survival advantage. Lohela et al. found that in Malawi, increasing distance to a health facility was paradoxically associated with lower neonatal mortality in some areas, potentially reflecting poor quality of care at accessible facilities [14]. Similarly, Fink et al. observed that the protective effect of facility delivery was attenuated or absent in regions with weak health systems [20]. These contrasting findings underscore that the benefits of hospital delivery depend critically on quality of care, not merely location of delivery, and suggest that KGH may provide a meaningful quality advantage over peripheral facilities in our setting.
The association between birth asphyxia and neonatal mortality (aHR 1.30) aligns with regional evidence. Systematic reviews have consistently identified birth asphyxia as a major risk factor for neonatal death in SSA, with pooled odds ratios ranging from 2.6 to 3.9 [16, 21]. Our more modest effect estimate may reflect the specific hospital-based population, where asphyxiated neonates receive immediate resuscitation and supportive care. The stronger association observed in the early neonatal period (aHR 1.29 for 0–7 days) compared to the late period (aHR 0.93 for 7–28 days) is biologically plausible, as severe asphyxia-related mortality occurs predominantly in the first days of life [15, 22]. These findings underscore the need for strengthened intrapartum care and immediate newborn resuscitation capacity, particularly at peripheral facilities where birth asphyxia may go unrecognized or inadequately managed [7, 17].
The protective effect of birth weight (42% reduction in mortality per kilogram increase) is well-established in the literature. Low birth weight, whether due to preterm birth or intrauterine growth restriction, is the single strongest predictor of neonatal mortality globally [23]. Our finding that 49.6% of neonates who died had low birth weight compared to 27.3% of survivors highlights the vulnerability of this population. Interestingly, after adjustment for birth weight as a continuous variable, the categorical low birth weight variable and prematurity diagnosis were no longer significantly associated with mortality, suggesting that the continuous measure captures most of the prognostic information. This has implications for risk stratification in resource-limited settings where gestational age assessment may be unreliable [24].
Strengths and limitations
This study has several strengths. First, we employed rigorous survival analysis methods with appropriate handling of censored observations and time-varying effects. Second, the use of robust standard errors clustered by facility accounted for potential correlation of outcomes within delivery locations. Third, comprehensive sensitivity analyses, including E-value calculations and time-varying models, demonstrated robustness of our principal findings. Fourth, the study population represents the largest referral hospital in The Gambia, providing insights relevant to health system planning.
This study has some limitations for consideration. First, the retrospective design relies on routinely collected clinical data, which may be subject to measurement error and missing information. Second, as a single-center study, findings may not be generalizable to other facilities in The Gambia or the region. Third, we could not assess quality of care at referring facilities or details of the referral process (timing, transport conditions), which may confound the facility-mortality association. Fourth, unmeasured confounders such as maternal HIV status, socioeconomic factors, and antenatal care quality may have influenced our estimates. Additionally, we could not assess facility-specific risk factors at individual peripheral health facilities, as our data were limited to place of delivery categorization (KGH vs. peripheral facilities as a group) without information on quality-of-care indicators at referring sites. Fifth, birth weight for referred neonates was sometimes measured at admission rather than at birth, which may introduce measurement error due to physiological weight loss in the first days of life; however, sensitivity analysis restricted to neonates admitted within 24 h confirmed the robustness of our findings. However, E-value analysis suggested that substantial unmeasured confounding (risk ratio > 4.3) would be required to explain away the facility effect. Finally, we could not distinguish between neonates who were delivered at peripheral facilities and referred versus those born at home and subsequently referred, which may represent distinct risk profiles.
Policy implications
Our findings have important implications for health policy in The Gambia and similar settings. First, the substantial survival disadvantage among neonates referred from peripheral facilities highlights the urgent need to strengthen neonatal care capacity at lower-level health facilities. This includes ensuring availability of basic resuscitation equipment, training health workers in Helping Babies Breathe and essential newborn care, and improving early recognition of danger signs [25]. Second, strengthening referral systems including communication between facilities, transport infrastructure, and pre-referral stabilization may help mitigate the mortality risk associated with referral [26, 27]. Third, the strong association with birth asphyxia underscores the importance of quality intrapartum care, including partograph use, fetal monitoring, and timely intervention for prolonged labor [9].
At the referral hospital level, our findings support continued investment in neonatal intensive care capacity at KGH, which serves as the safety net for complicated cases from across the region. The concentration of early neonatal deaths in the first 24 h suggests that interventions targeting the immediate postnatal period including enhanced monitoring, early feeding support, and prompt sepsis recognition may yield the greatest impact [28]. Quality improvement initiatives focused on standardized admission protocols, handover procedures from referring facilities, and real-time mortality surveillance could help identify modifiable factors contributing to early deaths.
The higher mortality observed among neonates readmitted after initial discharge from KGH Labour Ward (17.0% vs. 6.7%; Supplementary Table S2) warrants attention to postnatal discharge practices. Readmitted neonates presented later (median 72 h vs. 6 h) and more commonly with neonatal sepsis (66.0% vs. 45.0%) and jaundice (38.3% vs. 9.3%) (Supplementary Table S3), suggesting that conditions developing after initial discharge were the primary reasons for readmission. This finding suggests that some neonates may be discharged without adequate assessment or that danger signs are missed during the immediate postnatal period. Strengthening pre-discharge newborn examinations, enhancing maternal education on danger signs, and implementing structured postnatal follow-up within the first 48–72 h could help identify at-risk neonates before clinical deterioration.
Conclusions
This study demonstrates that neonates referred from peripheral health facilities to KGH experience substantially higher mortality compared to those delivered at the referral hospital. Birth weight and birth asphyxia are independently associated with neonatal survival. These findings highlight critical gaps in the continuum of care for newborns in The Gambia and point to opportunities for intervention at multiple levels of the health system. Achieving the SDG target for neonatal mortality will require concurrent investments in peripheral facility strengthening, referral system optimization, and sustained quality improvement at referral hospitals. Future research should examine specific components of the referral process, quality of care indicators, and community-level factors that may further explain facility-level disparities in neonatal outcomes.
Supplementary Information
Acknowledgements
The authors thank MOH and KGH for their support throughout this study.
Informed consent statement
This is not application to this study as we only review the hospital records between 2022 to 2024.
Authors’ contributions
AB, KKK and KSD conceptualized, designed, and conducted the analysis and wrote the background, methods, and results. AB, KKK and KSD critically reviewed and edited the manuscript for its intellectual content. All authors have read and agreed to submit this manuscript for publication.
Funding
No funding was received to support this study.
Data availability
The datasets are available upon request from the corresponding author.
Declarations
Ethics approval and consent to participate
This secondary analysis was conducted under the ethical approvals obtained from Ministry of Health (ref #: DHS/AD/2025/3/(MTN)) and Kanifing General Hospital accordingly. This study adhered to the guidelines of the Declaration of Helsinki.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.United Nations Inter-agency Group for Child Mortality Estimation (UN IGME). Levels and trends in child mortality: report 2023. New York: UNICEF; 2024. [Google Scholar]
- 2.World Health Organization. Child mortality and causes of death. 2025. https://www.who.int/data/gho/data/themes/topics/topic-details/GHO/child-mortality-and-causes-of-death. Accessed 25 Dec 2025.
- 3.World Health Organization. World Health Statistics 2023: Monitoring Health for the SDGs, Sustainable Development Goals. 1st ed. Geneva: World Health Organization; 2023. [Google Scholar]
- 4.Lawn JE, Blencowe H, Oza S, You D, Lee AC, Waiswa P, et al. Every newborn: progress, priorities, and potential beyond survival. Lancet. 2014;384:189–205. 10.1016/S0140-6736(14)60496-7. [DOI] [PubMed] [Google Scholar]
- 5.Bhutta ZA, Das JK, Bahl R, Lawn JE, Salam RA, Paul VK, et al. Can available interventions end preventable deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet. 2014;384:347–70. 10.1016/S0140-6736(14)60792-3. [DOI] [PubMed] [Google Scholar]
- 6.Kinney MV, Kerber KJ, Black RE, Cohen B, Nkrumah F, Coovadia H, et al. Sub-Saharan Africa’s mothers, newborns, and children: where and why do they die? PLoS Med. 2010;7:e1000294. 10.1371/journal.pmed.1000294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Usman F, Imam A, Farouk ZL, Dayyabu AL. Newborn mortality in sub-Saharan Africa: why is perinatal asphyxia still a major cause? Ann Glob Health. 2019;85:112. 10.5334/aogh.2541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gambia Bureau of Statistics, ICF. The Gambia Demographic and Health Survey 2019-20. Banjul, The Gambia and Rockville, Maryland, USA: GBoS and ICF; 2021. [Google Scholar]
- 9.UNICEF Gambia. Child survival and development. 2025. https://www.unicef.org/gambia/child-survival-and-development. Accessed 24 Oct 2025.
- 10.Gage AD, Fink G, Ataguba JE, Kruk ME. Hospital delivery and neonatal mortality in 37 countries in sub-Saharan Africa and South Asia: an ecological study. PLoS Med. 2021;18:e1003843. 10.1371/journal.pmed.1003843. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Moyer CA, Dako-Gyeke P, Adanu RM. Facility-based delivery and maternal and early neonatal mortality in sub-Saharan Africa: a regional review of the literature. Afr J Reprod Health. 2013;17:30–43. [PubMed] [Google Scholar]
- 12.Bezie MM, Fente BM, Asmare ZA, Asnake AA, Asebe HA, Negussie YM, et al. Impact of health facility delivery and antenatal care on neonatal mortality in Sub-Saharan Africa: a propensity score matching analysis. BMC Pregnancy Childbirth. 2025;25:440. 10.1186/s12884-025-07565-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tura G, Fantahun M, Worku A. The effect of health facility delivery on neonatal mortality: systematic review and meta-analysis. BMC Pregnancy Childbirth. 2013;13:18. 10.1186/1471-2393-13-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lohela TJ, Campbell OMR, Gabrysch S. Distance to care, facility delivery and early neonatal mortality in Malawi and Zambia. PLoS ONE. 2012;7:e52110. 10.1371/journal.pone.0052110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lee AC, Kozuki N, Blencowe H, Vos T, Bahalim A, Darmstadt GL, et al. Intrapartum-related neonatal encephalopathy incidence and impairment at regional and global levels for 2010 with trends from 1990. Pediatr Res. 2013;74:50–72. 10.1038/pr.2013.206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Techane MA, Alemu TG, Wubneh CA, Belay GM, Tamir TT, Muhye AB, et al. The effect of gestational age, low birth weight and parity on birth asphyxia among neonates in sub-Saharan Africa: systematic review and meta-analysis: 2021. Ital J Pediatr. 2022;48:114. 10.1186/s13052-022-01307-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wall SN, Lee AC, Niermeyer S, English M, Keenan WJ, Carlo W, et al. Neonatal resuscitation in low-resource settings: what, who, and how to overcome challenges to scale up? Int J Gynecol Obstet. 2009;107 Supplement:S47–64.. 10.1016/j.ijgo.2009.07.013. [DOI] [PMC free article] [PubMed]
- 18.Mishra N, Saini SS, Jayashree M, Kumar P. Quality of referral, admission status, and outcome of neonates referred to pediatric emergency of a tertiary care institution in North India. Indian J Community Med. 2022;47:253–7. 10.4103/ijcm.ijcm_889_21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Salam RA, Mansoor T, Mallick D, Lassi ZS, Das JK, Bhutta ZA. Essential childbirth and postnatal interventions for improved maternal and neonatal health. Reprod Health. 2014;11:S3. 10.1186/1742-4755-11-S1-S3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fink G, Ross R, Hill K. Institutional deliveries weakly associated with improved neonatal survival in developing countries: evidence from 192 demographic and health surveys. Int J Epidemiol. 2015;44:1879–88. 10.1093/ije/dyv115. [DOI] [PubMed] [Google Scholar]
- 21.Moxon SG, Lawn JE, Dickson KE, Simen-Kapeu A, Gupta G, Deorari A, et al. Inpatient care of small and sick newborns: a multi-country analysis of health system bottlenecks and potential solutions. BMC Pregnancy Childbirth. 2015;15:S7. 10.1186/1471-2393-15-S2-S7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ahmed I, Ali SM, Amenga-Etego S, Ariff S, Bahl R, Baqui AH, et al. Population-based rates, timing, and causes of maternal deaths, stillbirths, and neonatal deaths in South Asia and sub-Saharan africa: a multi-country prospective cohort study. Lancet Global Health. 2018;6:e1297–308. 10.1016/S2214-109X(18)30385-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Katz J, Lee AC, Kozuki N, Lawn JE, Cousens S, Blencowe H, et al. Mortality risk in preterm and small-for-gestational-age infants in low-income and middle-income countries: a pooled country analysis. Lancet. 2013;382:417–25. 10.1016/S0140-6736(13)60993-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Blencowe H, Cousens S, Oestergaard MZ, Chou D, Moller A-B, Narwal R, et al. National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since 1990 for selected countries: a systematic analysis and implications. Lancet. 2012;379:2162–72. 10.1016/S0140-6736(12)60820-4. [DOI] [PubMed] [Google Scholar]
- 25.Morris SM, Fratt EM, Rodriguez J, Ruman A, Wibecan L, Nelson BD. Implementation of the Helping Babies Breathe Training Program: A Systematic Review. Pediatrics. 2020;146(3):e20193938. 10.1542/peds.2019-3938. [DOI] [PubMed]
- 26.Kiputa M, Salim N, Kunambi PP, Massawe A. Referral challenges and outcomes of neonates received at Muhimbili National Hospital, Dar es Salaam, Tanzania. PLOS ONE. 2022;17(6):e0269479. 10.1371/journal.pone.0269479. [DOI] [PMC free article] [PubMed]
- 27.Okai E, Fair F, Soltani H. Neonatal transport practices and effectiveness of the use of low-cost interventions on outcomes of transported neonates in Sub-Saharan Africa: A systematic review and narrative synthesis. Health Sci Rep. 2024;7(3):e1938. 10.1002/hsr2.1938. [DOI] [PMC free article] [PubMed]
- 28.Baqui AH, Mitra DK, Begum N, Hurt L, Soremekun S, Edmond K, et al. Neonatal mortality within 24 hours of birth in six low- and lower-middle-income countries. Bull World Health Organ. 2016;94(10):752–758B. 10.2471/BLT.15.160945. [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets are available upon request from the corresponding author.