Patterns of preterm admission, morbidity, and mortality in a tribal-district neonatal intensive care unit: A retrospective cohort study

Abstract

BACKGROUND

Preterm birth remains the leading cause of neonatal morbidity and mortality worldwide. Data from tribal-district neonatal intensive care units (NICUs) in India are scarce, despite these populations bearing a disproportionate burden of adverse perinatal outcomes.

AIM

To evaluate morbidity patterns and identify predictors of mortality among preterm neonates in a tribal-district NICU.

METHODS

The present retrospective cohort study included total 1438 preterm neonates admitted between January 2022 and December 2024 to the tribal-district NICU. Clinical data were extracted from medical records. Mortality was analyzed by gestational age. Multivariable logistic regression, with multiple imputation for missing data, was used to identify independent predictors of in-hospital mortality.

RESULTS

Mortality demonstrated a steep inverse gradient with gestational age, decreasing from 49% among extremely preterm neonates to 4.6% among late preterm infants. The most common morbidities were respiratory distress syndrome (42%-55%) and neonatal sepsis (32%-45%). Early continuous positive airway pressure (CPAP) initiation (< 1 hour) was associated with reduced need for mechanical ventilation (P < 0.001). In multivariable analysis, increasing gestational age [adjusted odd ratios (AOR): 0.82 per week; 95%CI: 0.75-0.89], higher birth weight (AOR: 0.93 per 100 g; 95%CI: 0.89-0.97), culture-positive sepsis (AOR: 3.41; 95%CI: 1.98-5.89), delayed CPAP initiation (AOR: 1.94; 95%CI: 1.15-3.28), and outborn status (AOR: 1.67; 95%CI: 1.01-2.74) were independently associated with mortality. Discharge against medical advice rates remained high (9%-11%).

CONCLUSION

Lower gestational age, lower birth weight, infection, and delayed respiratory stabilization are associated with mortality. Strengthening early neonatal care may improve survival in tribal settings.

Key Words: Preterm neonate; Tribal health; Neonatal intensive care; Respiratory distress syndrome; Neonatal sepsis; Continuous positive airway pressure; Health equity; Neonatal mortality

Core Tip: Preterm neonates admitted to neonatal intensive care units in tribal districts face a substantial burden of morbidity and mortality. This 3-year retrospective cohort study demonstrates a clear inverse relationship between gestational age, birth weight, and survival. Culture-positive sepsis, delayed initiation of continuous positive airway pressure, and outborn status were independently associated with mortality. Strengthening early respiratory stabilization, improving antenatal care and infection control, and addressing socioeconomic barriers such as discharge against medical advice are essential to advancing neonatal survival and health equity in underserved tribal regions.

INTRODUCTION

Preterm birth, defined as delivery before 37 completed weeks of gestation, continues to be a major global public health challenge. According to the World Health Organization (WHO), approximately 15 million infants are born preterm each year, and complications associated with prematurity contribute to more than one million neonatal deaths worldwide, making it the leading cause of under-five mortality[1]. Preterm infants face a wide spectrum of acute and chronic health challenges, including respiratory distress syndrome (RDS), sepsis, intraventricular hemorrhage, feeding intolerance, hypothermia, and long-term neurodevelopmental impairment. The burden of preterm birth is disproportionately higher in low- and middle-income countries, where constraints in health systems limit the reach and effectiveness of essential newborn care services. India alone contributes the largest number of preterm births globally, with an estimated 3.5 million premature deliveries annually[2]. Despite major advancements in neonatal intensive care technologies-including the availability of continuous positive airway pressure (CPAP), surfactant therapy, improved thermal protection, and strengthened infection-control practices-the benefits of these innovations are unevenly distributed, and mortality remains significantly higher in rural, remote, and tribal regions[3].

Over the last 5 years, multiple studies have highlighted persistent maternal and neonatal health disparities in India’s tribal populations. Tribal communities, representing 8.6% of India’s population, consistently lag behind national averages across key reproductive and child health indicators. The National Family Health Survey-5 and state-level analyses reveal that tribal women have substantially lower antenatal care coverage, poor uptake of ≥ 4 antenatal care (ANC) visits, lower rates of institutional delivery, and limited access to antenatal supplements[4,5]. Recent analyses from Maharashtra, Odisha, and Gujarat demonstrate disproportionately high levels of maternal anemia, often exceeding 65% among tribal women, which substantially increases the risk of preterm labor and low birth weight[6]. Along with biological vulnerabilities, structural barriers such as long distances to health facilities, inadequate transport connectivity, and high out-of-pocket expenditure contribute to delayed access to obstetric and neonatal care[7]. Studies from 2020-2024 consistently reported that high-risk pregnancies from tribal districts often reached higher-level facilities late, which increases the probability of severe neonatal complications[8].

Neonatal care infrastructure in tribal districts reflects similar disparities. Recent evaluations of special newborn care units in tribal and remote areas documented significant gaps in skilled manpower, inconsistent CPAP/high-flow nasal cannula (HFNC) availability, poor infection-control systems, and limited referral pathways[9,10]. Although essential respiratory interventions are standard in most urban neonatal intensive care units (NICUs), their timely use remains uncertain in tribal districts, thereby predisposing preterm neonates to worse outcomes. Data from the Sample Registration System (2023) revealed that neonatal mortality rates remained 30%-40% higher in tribal-majority districts than in non-tribal areas after adjusting for socioeconomic differences[11]. Furthermore, audits conducted between 2021 and 2023 in tribal districts reported discharge against medical advice (DAMA) rates 2 to 4 times higher than those of urban NICUs, driven by financial pressures, cultural beliefs, and difficulty remaining near the hospital for prolonged NICU care[12].

Despite substantial literature from urban tertiary NICUs across India, research from tribal districts, particularly NICU-level clinical datasets, is limited. Most available studies rely on aggregated district-level data, with little granularity on morbidity patterns, respiratory support practices, or detailed risk modeling. Very few investigations evaluate predictors of mortality among tribal preterm neonates using multivariate analysis, and even fewer explore the impact of early CPAP initiation or sepsis profiles in resource-constrained settings[13,14]. As a result, the ability to design targeted interventions for preterm infants in tribal populations remains limited.

This study addresses this critical gap by analyzing the clinical profile, morbidity patterns, respiratory support use, and short-term outcomes of 1438 preterm neonates admitted to the NICU of a tertiary-care hospital situated in a predominantly tribal district in Western India. This region faces intersecting biological, socioeconomic, and system-level challenges, including maternal malnutrition, inadequate antenatal surveillance, shortages of skilled neonatal nurses, inconsistent respiratory support availability, and high DAMA rates. By examining the burden of key morbidities such as RDS, sepsis, hypothermia, and apnea of prematurity, and assessing associations between gestational age, birth weight, sepsis, and mortality, this study offers clinically meaningful insights essential for improving neonatal outcomes in underserved tribal regions.

To our knowledge, this is the first large-scale NICU study from a tribal district in western India to describe detailed morbidity patterns, respiratory support practices, and predictors of mortality among preterm neonates. The findings have significant implications for neonatal health equity research and highlight the urgent need to strengthen maternal and newborn care services in tribal populations.

MATERIALS AND METHODS

Study design and setting

This retrospective cohort study was carried out in the level III NICU of a tertiary-care teaching hospital located in a predominantly tribal district of Western India. The district is geographically remote, characterized by hilly terrain, suboptimal transport connectivity, and widespread socioeconomic vulnerabilities that influence maternal and neonatal outcomes. The NICU serves as the primary referral center for institutional deliveries and outborn neonates from peripheral health facilities situated within a 70-100 km radius. It is equipped with 20 functional beds, radiant warmers, servo-controlled incubators, syringe pumps, bubble CPAP units, HFNC devices, conventional mechanical ventilators, and phototherapy units. Basic hematological and biochemical investigations are available onsite, while blood cultures are performed in-house. More advanced microbiological and metabolic investigations are outsourced to a regional reference laboratory due to infrastructural limitations.

Ethical considerations

The study protocol was reviewed and approved by the Institutional Ethics Committee (IEC) of the hospital (Approval No. NAMOMERI-SVBCH/IEC/2023-24/227 dated August 28, 2025). Given the retrospective nature of the study and use of anonymized medical records, the requirement for informed consent was waived in accordance with IEC guidelines and the principles of the Declaration of Helsinki. Strict measures were taken to ensure confidentiality; data were accessed only through password-protected hospital systems, and no personal identifiers were stored.

Study population and eligibility criteria

The study included all preterm neonates born before 37 completed weeks of gestation who were admitted to the NICU between January 1, 2022 and December 31, 2024. Place of birth was categorized as inborn (delivered at the study hospital) or outborn (referred from peripheral facilities). Gestational age was determined using first-trimester ultrasound when available, last menstrual period, and the New Ballard Score when antenatal dating information was unreliable. Neonates were eligible if their clinical records contained complete data regarding medical condition, management, and discharge outcomes. When multiple sources of gestational age estimation were available, a predefined hierarchy was applied. First-trimester ultrasonography was prioritized, followed by the last menstrual period. The New Ballard Score was used only when antenatal dating was unavailable or unreliable. In cases of discrepancy, the hierarchy-based estimate was retained.

Exclusion criteria included major congenital malformations incompatible with life, medical records with more than 20% missing information, and neonates transferred to another facility within the first six hours of birth. After applying these criteria, 1438 preterm neonates were included in the final dataset. This sample size adequately reflects the NICU caseload over 3 years and provides sufficient variance for multivariate statistical analysis.

Data collection procedures

Data were obtained from NICU admission registers, electronic medical records, nursing charts, laboratory reports, and discharge summaries. Two trained research assistants with prior experience in neonatal audits were responsible for data extraction. They received structured orientation on the study protocol, operational definitions, and data extraction tools prior to commencing work. To minimize ascertainment bias, the data collectors were not involved in the clinical management of the neonates during the study period and were blinded to outcome variables while extracting exposure-related information such as timing of CPAP initiation, infection status, and maternal variables.

Variables collected

Maternal variables included maternal age, parity, adequacy of antenatal care (defined as ≥ 4 visits), pregnancy-related complications such as pregnancy-induced hypertension, premature rupture of membranes, gestational diabetes mellitus, and documentation of antenatal steroid administration. Neonatal variables included gestational age, birth weight, sex, mode and place of delivery, Apgar scores at 1 minute and 5 minutes, and need for resuscitation at birth. Clinical morbidities were recorded based on the National Neonatology Forum (NNF) and WHO guidelines. These included RDS, neonatal sepsis (categorized as culture positive or clinical), hypothermia, hypoglycemia, neonatal jaundice requiring phototherapy, apnea of prematurity, necrotizing enterocolitis (NEC) (staged using modified Bell’s criteria), and intraventricular hemorrhage when cranial ultrasound was performed. Therapeutic interventions included type and timing of respiratory support (early CPAP defined as initiation within 1 hour of birth, delayed CPAP defined as initiation after 1 hour), HFNC therapy, mechanical ventilation, intravenous fluid therapy, parenteral nutrition, antibiotic initiation and escalation, initiation of enteral feeds, and use of phototherapy. Outcomes assessed included survival to discharge, DAMA, in-hospital mortality, and length of NICU stay. Neonates the DAMA were excluded from mortality regression analyses but included in descriptive outcome reporting.

Operational definitions

Operational definitions adhered to standard classifications. Extremely preterm neonates were defined as those born before 28 weeks of gestation, very preterm as 28-31⁺⁶ weeks, moderate preterm as 32-33⁺⁶ weeks, and late preterm as 34-36⁺⁶ weeks. Birth weight categories included extremely low birth weight (< 1000 g), very low birth weight (1000 g-1499 g), and low birth weight (1500 g-2499 g). Sepsis definitions followed WHO guidelines, and early CPAP was defined as initiation of CPAP within the first hour of life.

Data management and quality control

Data accuracy was ensured through a double-entry system, with cross-verification performed at each stage. Missing values constituted less than 5% of the dataset. To minimize bias and preserve statistical power, multiple imputation using chained equations was performed for variables with ≤ 5% missingness. Twenty imputed datasets were generated under the assumption of missing at random, and estimates were pooled using Rubin’s rules. Variables with > 10% missingness were not included in multivariable modeling. All data were stored in encrypted files accessible only to the study team.

Statistical analysis

Statistical analysis was performed using SPSS version 26. Continuous variables were summarized using mean ± SD, while categorical variables were expressed as n (%). Differences between groups were assessed using the χ² test for categorical variables and Student’s t-test or one-way ANOVA for continuous variables, depending on variance assumptions. Univariate logistic regression was performed to estimate crude odds ratios (ORs) for predictors of mortality, including gestational age (continuous), birth weight (continuous), culture-positive sepsis, hypothermia, 5-minute Apgar score, timing of CPAP initiation, and inborn/outborn status. Inborn/outborn status was included as a covariate in multivariable regression to account for potential confounding related to referral delay and stabilization. Covariates were selected a priori based on clinical relevance and biological plausibility rather than statistical significance in univariate analysis. The final multivariable model included gestational age (continuous), birth weight (continuous), culture-positive sepsis, delayed CPAP initiation, hypothermia, and inborn or outborn status.

Multicollinearity among independent variables was evaluated using the variance inflation factor (VIF). A VIF value > 5 was considered indicative of moderate multicollinearity, and > 10 as unacceptable. Model performance was evaluated by assessing discrimination using the area under the receiver operating characteristic curve, calibration using the Hosmer-Lemeshow goodness-of-fit test, and overall predictive accuracy using the Brier score. Adjusted odds ratios (AOR) with 95% confidence intervals (CIs) were calculated, and a P value < 0.05 was considered statistically significant. Sensitivity analyses were performed treating DAMA cases under two extreme assumptions: All survived, and all died. Antenatal corticosteroid exposure was evaluated in univariate analysis and considered for multivariable modeling based on clinical relevance.

RESULTS

Table 1 presents the annual preterm admission burden and baseline characteristics of the cohort. Between 2022 and 2024, 1438 preterm neonates were admitted, constituting a substantial proportion of the 7788 NICU admissions. Preterm admissions increased progressively over time (431 in 2022, 492 in 2023, and 515 in 2024). Baseline characteristics were stable across years. The mean gestational age remained at approximately 32 weeks, and the mean birth weight at approximately 1.7 kg, with minimal interannual variation. Male neonates accounted for 54%-57% of admissions throughout the study period.

Table 1 Annual preterm admissions and baseline characteristics (n = 1438).

Variable	2022	2023	2024	Total
Preterm admissions	431	492	515	1438
Mean GA (weeks)	32.1 ± 3.4	32.0 ± 3.5	32.2 ± 3.4	32.1 ± 3.4
Mean birth weight (g)	1718 ± 506	1708 ± 518	1710 ± 514	1712 ± 512
Male (%)	54	55	57	55

GA: Gestational.

Table 2 outlines the year-wise distribution of preterm neonates by gestational age. Across the 3-year period, late preterm infants (34-36⁺⁶ weeks) constituted the largest subgroup (835/1438; 58.1%), followed by moderate preterm (324; 22.5%) and very preterm (210; 14.6%) neonates, while extremely preterm infants (< 28 weeks) accounted for a small but clinically significant proportion (69; 4.8%). The relative distribution of gestational age categories remained consistent from 2022 to 2024 (Figure 1).

Figure 1 Gestational age distribution of preterm neonates admitted to the tribal-district neonatal intensive care unit between 2022 and 2024. Illustrates the year-wise distribution of preterm neonates categorized by gestational age at admission. Late preterm infants (34-36⁺⁶ weeks) constituted the largest subgroup (835/1438; 58.1%), followed by moderate preterm (32-33⁺⁶ weeks; 324; 22.5%) and very preterm (28-31⁺⁶ weeks; 210; 14.6%) neonates. Extremely preterm infants (< 28 weeks) accounted for 69 admissions (4.8%). The distribution pattern remained stable across the three-year study period.

Table 2 Gestational age distribution.

Gestational category	2022	2023	2024	Total
< 28 weeks	23	18	28	69
28-32 weeks (28-31+6 weeks)	79	67	64	210
32-34 weeks (32-33+6 weeks)	96	101	127	324
34-37 weeks (34-36+6 weeks)	233	306	296	835
Total	431	492	515	1438

Table 3 summarizes gestational-age-specific mortality across the study cohort. Mortality demonstrated a marked inverse relationship with gestational age, declining from 49.3% among extremely preterm neonates (< 28 weeks) to 4.6% among late preterm infants (34-36⁺⁶ weeks). Very preterm (28-31⁺⁶ weeks) and moderate preterm (32-33⁺⁶ weeks) neonates exhibited intermediate mortality rates of 24.8% and 10.2%, respectively (Figure 2). Mortality rates were calculated excluding DAMA cases from the denominator.

Figure 2 Mortality gradient by gestational age among preterm neonates. Demonstrates gestational age-specific in-hospital mortality rates among preterm neonates. Mortality showed a marked inverse relationship with advancing gestational age, decreasing from 49.3% among extremely preterm neonates (< 28 weeks) to 24.8% in very preterm (28-31⁺⁶ weeks), 10.2% in moderate preterm (32-33⁺⁶ weeks), and 4.6% in late preterm infants (34-36⁺⁶ weeks). Mortality rates were calculated excluding neonates discharged against medical advice from the denominator.

Table 3 Gestational-age-specific mortality.

Category	Deaths/total	Mortality, %
< 28 weeks	34/69	49.3
28 < 32 weeks	52/210	24.8
32 < 34 weeks	33/324	10.2
34 < 37 weeks	38/835	4.6

Table 4 summarizes the major morbidities observed among preterm neonates during the study period. Respiratory and infectious complications predominated, with RDS (49.5%) emerging as the most frequent morbidity, followed by neonatal sepsis (36%) and hypothermia (28%). Nearly half of the cohort required phototherapy for neonatal jaundice (45%), while apnea of prematurity (15%) and culture-positive sepsis (15%) were also common. NEC (≥ stage 2) occurred in a smaller but clinically important proportion (4%). Table 5 summarizes respiratory support strategies used in the study cohort. Early CPAP was the most frequently employed respiratory support modality, initiated within the first hour of life in 42.0% of preterm neonates, while 25.0% received delayed CPAP. HFNC was used in 22.0%, and mechanical ventilation was required in 16.0% of admissions overall. The need for invasive ventilation was markedly higher among extremely preterm infants, affecting 55.1% of neonates born before 28 weeks’ gestation. Table 6 summarizes short-term clinical outcomes among preterm neonates admitted during the study period. Overall, 79.1% of preterm infants survived to discharge, while 10.8% died during hospitalization. DAMA occurred in 10% of cases, reflecting significant socioeconomic and access-related challenges in the study population. The mean length of NICU stay was 13.4 ± 4.8 days, with longer hospitalization observed among survivors, particularly those born at lower gestational ages or requiring intensive respiratory support.

Table 4 Major morbidities across the cohort (n = 1438).

Morbidity	Cases, n (%)
RDS	712 (49.5)
Sepsis (any)	518 (36.0)
Culture-positive sepsis	216 (15.0)
Hypothermia	403 (28.0)
Jaundice needing phototherapy	647 (45.0)
Apnea	216 (15.0)
NEC (≥ stage 2)	58 (4.0)

RDS: Respiratory distress syndrome; NEC: Necrotizing enterocolitis.

Table 5 Respiratory support interventions.

Intervention	Cases, n (%)
Early CPAP (< 1 hour)	604 (42.0)
Delayed CPAP (> 1 hour)	360 (25.0)
HFNC	317 (22.0)
Mechanical ventilation	230 (16.0)
Mechanical ventilation in < 28 weeks	38/69 (55.1)

CPAP: Continuous positive airway pressure; HFNC: High-flow nasal cannula.

Table 6 Short-term clinical outcomes (n = 1438), n (%).

Outcome	Value
Survived to discharge	1138 (79.1)
Discharge against medical advice	144 (10.0)
In-hospital mortality	156 (10.8)
Length of NICU stay (LOS), day	13.4 ± 4.8

NICU: Neonatal intensive care unit; LOS: Late onset sepsis.

Table 7 presents the independent predictors of in-hospital mortality identified using multivariable logistic regression analysis. Increasing gestational age and higher birth weight were independently associated with reduced odds of mortality. Each additional week of gestation decreased the odds of death by approximately 18% (AOR: 0.82; 95%CI: 0.75-0.89; P < 0.001), while each 100 g increase in birth weight reduced mortality risk by 7% (AOR: 0.93; 95%CI: 0.89-0.97; P = 0.002). Culture-positive sepsis was strongly associated with increased mortality (AOR: 3.41; 95%CI: 1.98-5.89; P < 0.001). Delayed initiation of CPAP was independently associated with higher odds of death (AOR: 1.94; 95%CI: 1.15-3.28; P = 0.01). Although hypothermia showed increased odds, it did not retain statistical significance after adjustment (AOR: 1.28; 95%CI: 0.74-2.21; P = 0.37). Outborn status was associated with modestly increased mortality risk (AOR: 1.67; 95%CI: 1.01-2.74; P = 0.045). Antenatal corticosteroid exposure was evaluated in univariate analysis but was not independently associated with mortality after adjustment for gestational age and birth weight, and was therefore not retained in the final multivariable model. Table 8 summarizes the performance characteristics of the final multivariable logistic regression model. The model demonstrated good discrimination, with an area under the receiver operating characteristic curve of 0.83, indicating a strong ability to distinguish between survivors and non-survivors. Calibration was adequate, as reflected by a non-significant Hosmer-Lemeshow goodness-of-fit test (P = 0.41). The Brier score of 0.12 further suggests acceptable overall predictive accuracy.

Table 7 Independent predictors of in-hospital mortality identified by multivariable logistic regression.

Predictor	Adjusted odd ratio	95%CI	P value
Gestational age (per week)	0.82	0.75-0.89	< 0.001
Birth weight (per 100 g)	0.93	0.89-0.97	0.002
Culture-positive sepsis	3.41	1.98-5.89	< 0.001
Delayed CPAP	1.94	1.15-3.28	0.01
Hypothermia	1.28	0.74-2.21	0.37
Outborn status	1.67	1.01-2.74	0.045

Variables entered into the multivariable model included gestational age (continuous), birth weight (continuous), culture-positive sepsis, timing of continuous positive airway pressure initiation, hypothermia, and inborn/outborn status. CI: Confidence interval; CPAP: Continuous positive airway pressure.

Table 8 Model performance metrics.

Metric	Value
AUC	0.83
Hosmer-Lemeshow P value	0.41
Brier score	0.12

AUC: Area under the curve.

DISCUSSION

This retrospective cohort study provides a comprehensive, 3-year analysis of all 1438 preterm neonates admitted to a tertiary-level NICU located in a tribal district of Western India, making it one of the largest datasets from any tribal-region NICU in the country. The findings reaffirm established global and national trends in prematurity while bringing attention to the unique challenges posed by the sociocultural, geographic, and infrastructural realities of tribal communities. The predominance of moderate and late preterm births across all 3 years mirrors national preterm distribution patterns; however, the persistent annual presence of extremely preterm and very preterm neonates underscores gaps in maternal nutrition, antenatal care utilization, and high-risk pregnancy identification in this region. Similar concerns were emphasized by Shah et al[15], who described malnutrition and delayed antenatal registration as central contributors to preterm birth in tribal India.

The morbidity profile in our cohort reflects the well-documented spectrum of prematurity in resource-limited settings. RDS (49.5%) and neonatal sepsis (36.0%) were the leading complications, with prevalence showing a steep inverse relationship with gestational age. This aligns closely with the findings of Jain et al[16], who reported RDS rates of 42%-54% in rural NICUs, attributing them to lung immaturity and limited availability of delivery-room respiratory support. Neonatal sepsis, clinical or culture-confirmed, remained a major contributor to morbidity, consistent with the observations of Dramowski et al[17], who outlined high sepsis rates in low to middle-income countries (LMICs) NICUs due to suboptimal infection-control measures, prolonged labor, and delayed ANC intervention. The low yield of culture-positive sepsis in our cohort parallels the concerns raised by Silva et al[18], who reported that inadequate sample volumes, pre-culture antibiotic exposure, and laboratory limitations resulted in high rates of presumed but unconfirmed sepsis in peripheral neonatal units.

The prevalence of hypothermia across all years highlights critical gaps in thermal care at birth, during transport, and in early NICU stabilization. Tribal districts often lack warm-chain transport services and experience substantial delays in referral, which explains the patterns observed in our dataset. Kumar et al[19] documented similar findings in tribal newborn units, noting that inadequate temperature regulation contributed significantly to early metabolic instability and infection risk. Our results therefore, underscore the urgent need for systematic improvements in thermal care, from ensuring warm delivery rooms to strengthening neonatal transport pathways.

A particularly important finding across the full 1438-infant cohort is the clear benefit observed with early CPAP initiation, which consistently reduced the need for mechanical ventilation across all 3 years. Outborn status may contribute to delayed respiratory stabilization and increased infection risk due to transport-related instability and referral delays. The observed association between delayed CPAP initiation and mortality should not be interpreted as causal. Delayed CPAP may reflect underlying illness severity, referral delays, or transport-related instability rather than solely delayed intervention. Evidence from randomized trials such as the COIN and SUPPORT studies supports the clinical benefits of early CPAP in preterm infants; however, the association observed in our cohort should be interpreted cautiously given its observational design[20,21]. Despite this evidence, delayed CPAP initiation remained common, especially among outborn infants, reflecting infrastructure constraints in peripheral centers, limited availability of delivery-room CPAP, and delays in early stabilization. Strengthening regional perinatal networks and implementing standardized delivery-room CPAP protocols may thus substantially improve outcomes in tribal NICUs. The mortality patterns across all 1438 preterm neonates reveal a characteristic gradient, with survival improving dramatically with increasing gestational age. Mortality was highest among extremely preterm infants (< 28 weeks), exceeding 45% in most years, while late preterm mortality remained consistently below 6%. This trend echoes the viability curves described by Manuck et al[22], who documented similar steep declines in mortality with advancing gestational age across LMIC neonatal populations. While the overall mortality in our cohort (10.8%) appears lower than previously reported rates of 16%-22% in tribal NICUs, it remains higher than mortality reported from urban tertiary centers (8%-10%), reflecting systemic inequities in antenatal care, transport, and neonatal support. Yang et al[23] similarly noted that disparities in antenatal steroid administration, staff competencies, and NICU infrastructure were major contributors to the mortality gap between tribal and urban facilities. Direct comparison with urban tertiary NICUs should be interpreted cautiously due to differences in case-mix, referral patterns, resource availability, and baseline risk profiles.

A distinctive contextual finding in our study is the persistently high incidence of DAMA, averaging 9%-11% across the 3-year population. This is considerably higher than DAMA rates in urban NICUs and reflects the precarious socioeconomic circumstances of families in tribal communities. High DAMA rates may result in underestimation of true mortality, as post-discharge outcomes remain unknown. Policy-level interventions such as caregiver accommodation support, financial assistance, and community follow-up mechanisms may mitigate this inequity.

Caregivers frequently face barriers such as daily wage loss, long-distance travel, limited accommodation options, and inadequate understanding of illness severity. Sharma et al[24] reported similar DAMA patterns in tribal districts, noting that financial constraints and sociocultural perceptions often prompt early withdrawal from care. Importantly, DAMA contributes to outcome misclassification because post-discharge deterioration or mortality remains undocumented. Thus, the DAMA patterns observed in this study underscore the need for culturally sensitive counseling, financial support systems, and community-based follow-up programs.

When viewed within the global LMIC context, our findings are consistent with established patterns. Salam et al[25] reported RDS prevalence between 35%-60% and neonatal mortality between 12%-20% in LMICs, which range closely aligned with our large-scale dataset. Urban Indian NICUs, such as those described by Aggarwal et al[26], continue to outperform tribal facilities due to better infrastructure, comprehensive staffing, improved transport mechanisms, and higher ANC uptake. These comparisons reinforce the importance of implementing standardized NICU protocols, particularly early CPAP initiation, improved sepsis prevention, and enhanced thermal care, to narrow the survival gap between tribal and urban neonatal populations.

Strengths and limitations

This study’s strengths include the complete inclusion of all 1438 preterm admissions across 3 years, representing one of the most extensive tribal NICU datasets published to date. The use of WHO/NNF definitions strengthens comparability, while multivariate modeling provides robust insights into mortality predictors. To minimize overadjustment, correlated categorical constructs were avoided, and gestational age and birth weight were modeled as continuous variables. The contextual interpretation of findings adds further value to neonatal health equity research in underserved populations. The study’s limitations stem primarily from its retrospective design. Neuroimaging was inconsistent, which may have underestimated IVH prevalence; however, major clinical outcomes such as respiratory failure and sepsis were well documented. Blood culture sensitivity was suboptimal due to pre-labor antibiotic exposure and small-volume sampling, leading to possible underestimation of culture-positive sepsis. The incomplete documentation of maternal steroid timing, transport conditions, and surfactant administration limited detailed analysis of these variables. DAMA cases could not be tracked post-discharge, possibly leading to underestimation of true mortality. Although multiple imputation reduces bias compared to single imputation, the validity of this approach depends on the assumption of missing at random, which cannot be fully verified in a retrospective dataset. Despite these limitations, consistency across the 3-year dataset, robust sample size, and alignment with external evidence strengthen the overall validity of our findings.

CONCLUSION

This 3-year analysis shows that preterm neonates in tribal districts carry a high burden of respiratory, infectious, and thermoregulatory complications, with mortality concentrated among extremely preterm and extremely low birth weight infants. These outcomes reflect both biological vulnerability and systemic inequities, including suboptimal antenatal care, delayed stabilization, referral delays, and high DAMA rates. Improving survival requires coordinated, equity-focused strategies such as strengthening delivery-room CPAP, enhancing antenatal steroid coverage, ensuring thermal and infection control, improving neonatal transport, and supporting families through community-based interventions.