FE | 9 minute read
10/6/26
10/6/26
Baseline report | Pilot Hospital
Neonatal mortality among low birthweight and preterm newborns receiving conventional care at a large tertiary hospital in Northern Nigeria
Reporting standard. This report presents observed data and makes no causal claims.
This is a baseline follow up conducted by an implementing organisation, not a research study conducted by an academic institution. It is not peer-reviewed. It is written to be as compliant as possible with STROBE guidelines for observational cohorts: https://www.strobe-statement.org/.
Summary
We conducted a one month prospective cohort study to assess baseline neonatal mortality among low birthweight and preterm newborns under conventional care at our pilot hospital. This precedes the introduction of an immediate Kangaroo Mother Care (iKMC) program here. Our aim was to enrolled every such newborn at birth or admission and follow them up to day 28 of life. Among 80 newborns, 21 deaths were recorded. Follow-up was completed for 55 (69%). Treating all newborns lost to follow-up as survivors, neonatal mortality was 26% (95% CI 18–37%). Across known outcomes mortality was 38% (95% CI 27–51%). The true rate likely lies between these rates. When estimating intervention effect size we will compare against the lower figure to be conservative. This is the prospectively defined pre-intervention comparator. The findings highlight the value of direct local measurement in low-resource settings where routine data are frequently incomplete.
1. Background
The true burden of neonatal mortality among the low birthweight and preterm newborn population in Nigeria is unclear. Limitations of existing survey-based estimates are discussed in detail in section 4. Estimated rates vary significantly between regions and even between newborns born in the community compared to in hospital.
The evidence that KMC reduces mortality is well established. A 2016 Cochrane meta-analysis of 21 randomised-controlled trials found a mortality reduction of 33% among low-birthweight newborns, relative to conventional incubator care (RR 0.67, 95% CI 0.48–0.95).[1] A five-country WHO trial, with Nigeria among the sites, found a 25% mortality reduction from initiating KMC immediately (iKMC) rather than after stabilisation (RR 0.75, 95% CI 0.64–0.89).[2] The forthcoming comparison is the same as the Cochrane meta-analysis.
The objective of this study was to measure the neonatal mortality rate among eligible low-birthweight and preterm newborns at our pilot hospital under conventional care. Based on calculations from available estimates we hypothesised this would be 20-25%. Standard care at the hospital during this baseline included incubators where available. The one-month window was the period immediately preceding iKMC implementation. Its length was set by operational constraints, making this a fixed-window enumeration.
The objective of this study was to measure the neonatal mortality rate among eligible low-birthweight and preterm newborns at our pilot hospital under conventional care. Based on calculations from available estimates we hypothesised this would be 20-25%. Standard care at the hospital during this baseline included incubators where available. The one-month window was the period immediately preceding iKMC implementation. Its length was set by operational constraints, making this a fixed-window enumeration.
2. Methods
Design and setting. Prospective observational cohort study at a tertiary hospital in Kano State, Nigeria. The enrolment window ran from 22nd January to 22nd February 2026. Each newborn was followed up to day 28 after birth.
Eligibility. WHO iKMC biological eligibility criteria: Birthweight ≤2.5 kg and/or gestational age of ≤37 weeks. Newborns were enrolled as soon as encountered after birth (inborn) or admission (referred). Multiple births were entered as separate records with birth order recorded.
Measurement. Birthweight was the newborn's first recorded weight, measured by clinical staff and either extracted from records (primarily) or directly observed. Vital status was the outcome. No other clinical variables were collected.
Procedure. A single enumerator reviewed the delivery register, theatre logs, and all neonatal ward registers daily. They identified eligible newborns, recording the medical record number (MRN) and two caregiver contacts. In hospital deaths, discharges, and transfers were logged daily from the ward logs, patient notes, and discharge books.
Vital status was established by telephone at day 7 and day 28. Non-responders received three call attempts, a message, and a call to the alternate contact before being defined as lost to follow-up.
Vital status was established by telephone at day 7 and day 28. Non-responders received three call attempts, a message, and a call to the alternate contact before being defined as lost to follow-up.
Definitions. Alive: confirmed by caregiver or clinician. Dead: confirmed by caregiver, clinician, or register, with date. Lost to follow-up: no contact after the full attempt protocol.
Verification. Each in-hospital death was triangulated against the ward death register and the patient file or nursing notes, with name, MRN, and date confirmed. Deaths were corroborated where a register entry was available.
Statistical analysis. Categorical variables are summarised as counts and percentages, birthweight as the median. Proportions carry 95% confidence intervals by the Wilson score method. This is suited to proportions near the bounds at small sample sizes. Mortality is given as two figures spanning a missingness range: a naïve rate treating all lost-to-follow-up newborns as survivors, and a complete-case rate restricted to known outcomes. This process aims to make our rate as accurate as possible.
No formal imputation, modelling, or significance testing was done. The study was descriptive and not powered to detect associations. The cohort included seven twin pairs, whose outcomes are not statistically independent.
Ethics. The study was approved by the Kano State Ministry of Health Department of Research and Statistics. Verbal caregiver consent for follow-up contact was obtained at enrolment using a standard script. Consent covered the day 7 and day 28 calls only.
3. Results
Participant flow. 80 eligible newborns were enrolled; none were excluded. We did not record the number screened to identify these 80, so the screening-to-eligible step is unquantified (see Limitations).
Characteristics. Birthweight was recorded for every newborn (median 1,750 g). Gestational age was recorded for only one. Composition was as follows:
Outcomes. Outcome at day 28 was established for 55 newborns (69%, 95% CI 57–77%), among whom 34 were alive and 21 were dead. The remaining 25 newborns (31%, 95% CI 23–43%) were lost to follow-up. We present mortality (28-day cumulative incidence) under two pre-specified missingness assumptions to reflect uncertainty.
These represent bounds under alternative assumptions about outcomes among newborns lost to follow-up. True mortality likely lies between these values, with direction dependent on the outcomes among the untraced newborns.
69% of newborns lost to follow up were discharged alive, and survival among their traced counterparts was 89%. This suggests that higher mortality estimates are less plausible.
69% of newborns lost to follow up were discharged alive, and survival among their traced counterparts was 89%. This suggests that higher mortality estimates are less plausible.
Deaths were observed in both inborn and referred newborns, distributed almost evenly between groups.
4. Comparison with external estimates
No direct, prospectively measured facility-based estimate of low-birthweight neonatal mortality for Kano has been published. National surveys are not able to provide one. The Nigeria Demographic and Health Survey (NDHS) and UNICEF's Multiple Indicator Cluster Surveys (MICS) do not record gestational age or routinely measured birthweight, instead relying on maternal perception of newborn size as a proxy. They ask mothers for a numerical birthweight only if it is recorded on an antenatal card. As a result, birthweight is recorded for only about a quarter of births[3], and neonatal deaths are substantially less likely to have associated birthweight data[4]. Organisations like the WHO adjust this data at aggregate level by combining numeric data with relative size assessments to correct for missing records and recall bias.
This approach means preterm newborns of normal size (who may still carry elevated mortality risk) are excluded and is subject to under-ascertainment of early neonatal deaths. Low-birthweight categorisation in these datasets is likely depleted of higher-risk newborns. Estimates reflect relative risk patterns rather than true mortality rates (perceived-small newborns have approximately twice the neonatal mortality)[5][6]. These figures should therefore be interpreted cautiously as they depend on many assumptions.
This approach means preterm newborns of normal size (who may still carry elevated mortality risk) are excluded and is subject to under-ascertainment of early neonatal deaths. Low-birthweight categorisation in these datasets is likely depleted of higher-risk newborns. Estimates reflect relative risk patterns rather than true mortality rates (perceived-small newborns have approximately twice the neonatal mortality)[5][6]. These figures should therefore be interpreted cautiously as they depend on many assumptions.
Nationally, Nigeria’s neonatal mortality rate is estimated at 39 per 1,000 live births (UNICEF)[7]. Evidence suggests low-birthweight newborns account for approximately 15% of births but around 60–80% of neonatal deaths[2]. While these figures are sometimes combined in back-of-the-envelope calculations, such decompositions rely on assumptions about stability of case-mix across settings, homogeneity of risk within birthweight strata, and completeness of death attribution in routine data. We treat them as descriptive relationships rather than a basis for precise estimation.
Single-centre Nigerian SCBU studies report overall neonatal mortality of approximately 8–23%, concentrated among preterm and outborn newborns[9]. Preterm delivery is associated with an adjusted odds ratio of around 4 for death, and reported birthweight-specific mortality reaches approximately 126 per 1,000 among preterm low-birthweight newborns [10].
Single-centre Nigerian SCBU studies report overall neonatal mortality of approximately 8–23%, concentrated among preterm and outborn newborns[9]. Preterm delivery is associated with an adjusted odds ratio of around 4 for death, and reported birthweight-specific mortality reaches approximately 126 per 1,000 among preterm low-birthweight newborns [10].
For Kano specifically, State Ministry of Health estimates report a neonatal mortality rate of 109 per 1,000 live births[11]. A study of low-birthweight deliveries in a northern Nigerian tertiary hospital (2011) reported an 11.3% prevalence of low birthweight but provided no mortality estimates[11].
Historical data from our facility reported 40 low-birthweight or preterm neonatal deaths over 10 months. In contrast, we observed 21 such deaths in one month. Given the magnitude of inconsistency across multiple routine indicators and observed outcomes, we judged this data unsuitable as a comparator due to internal inconsistency and misalignment with both observed and external evidence. This discrepancy is consistent with likely undercounting of deaths in routine records, particularly due to incomplete tracking of post-discharge outcomes.
Historical data from our facility reported 40 low-birthweight or preterm neonatal deaths over 10 months. In contrast, we observed 21 such deaths in one month. Given the magnitude of inconsistency across multiple routine indicators and observed outcomes, we judged this data unsuitable as a comparator due to internal inconsistency and misalignment with both observed and external evidence. This discrepancy is consistent with likely undercounting of deaths in routine records, particularly due to incomplete tracking of post-discharge outcomes.
Taken together, the available evidence implies that low-birthweight and preterm newborns experience substantially higher mortality than term newborns. Plausible relative risks are on the order of 10–20 times higher under standard decomposition assumptions.
Overall, the spread of estimates underscores the limitations of existing data sources and persistent evidence gaps in low income settings such as ours.
Overall, the spread of estimates underscores the limitations of existing data sources and persistent evidence gaps in low income settings such as ours.
5. Limitations
There are a number of important limitations of this study of unclear magnitude.
5.1 Missing outcomes (post-enrolment loss to follow-up)
A day 28 outcome was not obtained for 25 of 80 enrolled newborns (31%). Missingness may be non-random (missing-not-at-random) and reflects loss after enrolment and discharge. This introduces uncertainty into mortality estimation, as outcomes for this group are unknown and likely differ systematically from those with complete follow-up. Mortality estimates are therefore sensitive to assumptions about this group, and we present a range reflecting this uncertainty.
A day 28 outcome was not obtained for 25 of 80 enrolled newborns (31%). Missingness may be non-random (missing-not-at-random) and reflects loss after enrolment and discharge. This introduces uncertainty into mortality estimation, as outcomes for this group are unknown and likely differ systematically from those with complete follow-up. Mortality estimates are therefore sensitive to assumptions about this group, and we present a range reflecting this uncertainty.
5.2 Missing pre-enrolment information and baseline characteristics
The number of eligible newborns screened prior to enrolment was not recorded, meaning the denominator of all potentially eligible newborns is unknown. As a result, enrolment completeness cannot be assessed and some newborns may have died before being captured in registers or enrolled, implying potential under-ascertainment of early neonatal deaths.
The number of eligible newborns screened prior to enrolment was not recorded, meaning the denominator of all potentially eligible newborns is unknown. As a result, enrolment completeness cannot be assessed and some newborns may have died before being captured in registers or enrolled, implying potential under-ascertainment of early neonatal deaths.
Gestational age was recorded for only one newborns, preventing stratification by prematurity and precluding a preterm-specific mortality estimate. Cause of death was not recorded, limiting interpretation of mortality mechanisms.
5.3 Measurement and Ascertainment
Outcome ascertainment relied on hospital records and telephone follow-up. Approximately 75% of deaths were identified via caregiver report rather than direct observation, particularly after discharge. This introduces potential misclassification, including under-reporting of deaths due to recall or social desirability bias, which would bias mortality downward. In addition, deaths occurring outside the hospital or follow-up system without caregiver reporting would not be captured.
Outcome ascertainment relied on hospital records and telephone follow-up. Approximately 75% of deaths were identified via caregiver report rather than direct observation, particularly after discharge. This introduces potential misclassification, including under-reporting of deaths due to recall or social desirability bias, which would bias mortality downward. In addition, deaths occurring outside the hospital or follow-up system without caregiver reporting would not be captured.
5.4 Inference and Generalisability
This is a single-site, one-month descriptive cohort and should be interpreted as a fixed-period snapshot rather than a population-representative estimate. The sample size is small due to operational constraints.
This is a single-site, one-month descriptive cohort and should be interpreted as a fixed-period snapshot rather than a population-representative estimate. The sample size is small due to operational constraints.
Overall mortality reflects newborns successfully enrolled into the cohort; however, some eligible newborns may have died before enrolment and therefore remain uncounted, meaning the true mortality burden among eligible newborns is likely higher than reported. Findings may not generalise beyond this facility due to variation in referral patterns, staffing, and neonatal care availability, and temporal variation is not captured within the study window.
6. Conclusions
The absence of reliable newborn mortality data for low birthweight and preterm populations is a recognised constraint in newborn health. Expert discussions suggest this a reason it is neglected within global health. The lack of credible local counterfactuals limits work and funding because the effects of interventions for these populations (iKMC) cannot be accurately estimated without it. Here, direct local measurement may help address this by providing a more reliable basis for estimating mortality burden and calculating intervention effects.
One in four low birth-weight and preterm newborns enrolled at our pilot facility died under conventional care. This is consistent with estimates of low birth weight prevalence in Nigeria and the contribution of such newborns to the neonatal mortality burden. This cohort serves as the facility pre-intervention comparator for our pilot program. Effect sizes will also be triangulated against regional and national estimates for the population we intend to treat, noting limitations.
One in four low birth-weight and preterm newborns enrolled at our pilot facility died under conventional care. This is consistent with estimates of low birth weight prevalence in Nigeria and the contribution of such newborns to the neonatal mortality burden. This cohort serves as the facility pre-intervention comparator for our pilot program. Effect sizes will also be triangulated against regional and national estimates for the population we intend to treat, noting limitations.
Funding. The study was funded by First Embrace as part of its core programme. No funding was restricted to this study. No funder had any role in any part of the study.
References
- Conde-Agudelo A, DÃaz-Rossello JL. Kangaroo mother care to reduce morbidity and mortality in low birthweight infants. Cochrane Database of Systematic Reviews 2016;8:CD002771. (21 RCTs; mortality at latest follow-up RR 0.67, 95% CI 0.48–0.95.)
- WHO Immediate KMC Study Group. Immediate "Kangaroo Mother Care" and Survival of Infants with Low Birth Weight. New England Journal of Medicine 2021;384:2028–2038. (RCT, 3,211 infants 1.0–<1.8 kg, 5 countries incl. Nigeria; neonatal mortality 12.0% vs 15.7%, RR 0.75, 95% CI 0.64–0.89. Premise: LBW/preterm ≈15% of neonates, ≈70% of neonatal deaths.)
- Nigeria Demographic and Health Survey 2018. Numerical birthweight recorded for ~24% of births.
- Abubakar A et al. Birthweight data completeness and quality in population-based surveys: EN-INDEPTH study. 2021. (Neonatal deaths substantially less likely to have been weighed, aOR 0.19, 95% CI 0.16–0.24.)
- Ezeh OK et al. Determinants of neonatal mortality in Nigeria: evidence from the 2008 Demographic and Health Survey. 2014. (Perceived smaller-than-average size, neonatal mortality HR 2.10, 95% CI 1.77–2.50.)
- Adewuyi EO et al. Trends in neonatal mortality in Nigeria and effects of bio-demographic and maternal characteristics. 2015. (Pooled 1990/2013 NDHS; small birth size HR 1.72, 95% CI 1.39–2.14.)
- A five-year review of the morbidity and mortality pattern in the SCBU of a private tertiary hospital in Nigeria. Pan African Medical Journal 2025. (Preterm case fatality 18.6%; preterm delivery aOR 4.10, 95% CI 2.06–8.16. Related Nigerian SCBU series report overall mortality 8–23%.)
- Onwuanaku CA et al. The effects of birth weight and gender on neonatal mortality in north central Nigeria. 2011. (Birthweight-specific mortality 126 per 1,000 in preterm LBW vs 5 per 1,000 at term.)
- Kano State Ministry of Health data
- Hassan, O. and Muhammad, Z. (2011). Trends in the incidences of low birth weight deliveries in a tertiary hospital, in northern Nigeria. Sahel Medical Journal, 3(2): 100–107.
Give a newborn the first embrace they need