Neonatal admissions and mortality: a statistical adjustment approach using birthweight-specific curves to address bias due to underreporting in 65 hospitals in four African countries
8 affiliations
- 1Department of Infectious Disease Epidemiology and International Health, London School of Hygiene & Tropical Medicine
- 2The University of Hong Kong
- 3Research Support Center, School of Public Health and Family Medicine, Kamuzu University of Health Sciences
- 4Health Systems Impact Evaluation and Policy Department, Ifakara Health Institute
- 5Rice360 Institute for Global Health Technologies, Rice University
- 6Kenya Medical Research Institute-Wellcome Trust
- 7Aga Khan University
- 8APIN Public Health Initiatives
Abstract
Key Findings WHAT WAS KNOWN? Underreporting of neonatal admissions and mortality is common in many countries, particularly for neonates <1000g, leading to underestimation of in-patient neonatal mortality rates (iNMRs). This has important implications for comparing between hospitals or over time, evaluation of interventions, as well as planning resource allocation and tracking progress towards SDG 3.2 for newborn survival. WHAT WAS DONE THAT IS NEW? Based on context-specific data, a simulation-based approach was used to develop birthweight-specific adjustment functions to account for underreporting of admissions and in-patient neonatal deaths. Various levels of underreporting were then introduced, and adjustment models were re-fitted to estimate the original simulated values. The approach was extended to adjust for underreporting of admissions and deaths in 65 hospitals partnering with NEST360 in four countries (Kenya, Malawi, Nigeria, and Tanzania). Adjustment curves were developed using data from 26 hospitals with plausible mortality rates (≥700 deaths per 1000 live births for neonates <1000g) to create robust birthweight-specific adjustment functions. These functions were then applied across data from all 65 hospitals to adjust for underreporting in a dataset of 172,809 baseline in-patient neonatal mortality admissions. WHAT WAS FOUND? Applying birthweight-specific adjustment functions to simulated data demonstrated the statistical approach to be robust in adjusting for underreporting, since adjusted neonatal iNMRs closely matched the original simulated values ("truth") even at varying levels of underreporting. Using a dataset of 172,809 admissions from 65 hospitals partnering with NEST360, unadjusted iNMRs were 121 deaths per 1000 admissions in Nigeria, 128 per 1000 in Tanzania, 132 per 1000 in Malawi, and 135 per 1000 in Kenya. Birthweight-specific adjustment functions were developed based on 26 hospitals with plausible mortality rates (≥700 deaths per 1000 live births for neonates <1000g). After applying adjustment curves, the iNMRs increased to 162 deaths per 1000 admissions in Nigeria, 183 per 1000 in Tanzania, 174 per 1000 in Malawi, and 172 per 1000 in Kenya. As expected, adjusted iNMRs were higher and reduced inter-hospital variability. WHAT NEXT? This adjustment method may be useful more widely, including in LMIC national routine health data, which is now the main source of neonatal data, given recent defunding of Demographic Health Surveys. Share learning to improve capture of all admissions and deaths in routine data systems, such as increased communication that all live-born babies are to be registered, irrespective of gestational age or weight, regular data audits, and educating/mentoring to promote accurate and complete reporting.
Related
- Associations between marital adjustment, motherhood perception, and body image perception during pregnancy: Statistical indirect associations involving perceived social support and depressive symptoms
- Epidemiology of pediatric intensive care unit admissions and mortality in low- and middle-income countries: a systematic review and meta-analysis
- Poor adjustment to nursing homes and 1-year mortality: a secondary analysis of the KASEHPAD cohort study
- Forecasting Ethiopia's Neonatal Mortality Rate Toward the Sustainable Development Goal Target 3.2 Using Statistical and Deep Learning Models: A Time-Series Forecasting Study.
- Early erythropoiesis‐stimulating agents in preterm or low‐birthweight infants
BibTeX
@article{malla2026neonatal,
title = {{Neonatal admissions and mortality: a statistical adjustment approach using birthweight-specific curves to address bias due to underreporting in 65 hospitals in four African countries}},
author = {Lucas Malla and Siu Nam Wong and Samuel Ngwala and Evelyn Zimba and Msandeni Chiume and Norman Lufesi and Kondwani Kawaza and Honorati Masanja and Josephine Shabani and Prosper Mshana and Robert Tillya and Nahya Salim and John Wainaina and Julius Thomas and Junwei Chen and Franklin Okech and George Okello and William Macharia and Olabisi Dosunmu and Chinyere Ezeaka and Opeyemi Odedere and Morris Ogero and James H. Cross and Rebecca E. Penzias and Christine Bohne and Rebecca Richards-Kortum and Maria Oden and Eric O. Ohuma and Joy E. Lawn and with NEST360 Collaborative authorship group},
journal = {Population Health Metrics},
year = {2026},
month = jul,
doi = {10.1186/s12963-026-00496-z},
url = {https://doi.org/10.1186/s12963-026-00496-z},
note = {CC BY},
}
RIS
TY - JOUR
TI - Neonatal admissions and mortality: a statistical adjustment approach using birthweight-specific curves to address bias due to underreporting in 65 hospitals in four African countries
AU - Lucas Malla
AU - Siu Nam Wong
AU - Samuel Ngwala
AU - Evelyn Zimba
AU - Msandeni Chiume
AU - Norman Lufesi
AU - Kondwani Kawaza
AU - Honorati Masanja
AU - Josephine Shabani
AU - Prosper Mshana
AU - Robert Tillya
AU - Nahya Salim
AU - John Wainaina
AU - Julius Thomas
AU - Junwei Chen
AU - Franklin Okech
AU - George Okello
AU - William Macharia
AU - Olabisi Dosunmu
AU - Chinyere Ezeaka
AU - Opeyemi Odedere
AU - Morris Ogero
AU - James H. Cross
AU - Rebecca E. Penzias
AU - Christine Bohne
AU - Rebecca Richards-Kortum
AU - Maria Oden
AU - Eric O. Ohuma
AU - Joy E. Lawn
AU - with NEST360 Collaborative authorship group
PY - 2026
DA - 2026/07/01
JO - Population Health Metrics
DO - 10.1186/s12963-026-00496-z
UR - https://doi.org/10.1186/s12963-026-00496-z
AB - Key Findings WHAT WAS KNOWN? Underreporting of neonatal admissions and mortality is common in many countries, particularly for neonates <1000g, leading to underestimation of in-patient neonatal mortality rates (iNMRs). This has important implications for comparing between hospitals or over time, evaluation of interventions, as well as planning resource allocation and tracking progress towards SDG 3.2 for newborn survival. WHAT WAS DONE THAT IS NEW? Based on context-specific data, a simulation-based approach was used to develop birthweight-specific adjustment functions to account for underreporting of admissions and in-patient neonatal deaths. Various levels of underreporting were then introduced, and adjustment models were re-fitted to estimate the original simulated values. The approach was extended to adjust for underreporting of admissions and deaths in 65 hospitals partnering with NEST360 in four countries (Kenya, Malawi, Nigeria, and Tanzania). Adjustment curves were developed using data from 26 hospitals with plausible mortality rates (≥700 deaths per 1000 live births for neonates <1000g) to create robust birthweight-specific adjustment functions. These functions were then applied across data from all 65 hospitals to adjust for underreporting in a dataset of 172,809 baseline in-patient neonatal mortality admissions. WHAT WAS FOUND? Applying birthweight-specific adjustment functions to simulated data demonstrated the statistical approach to be robust in adjusting for underreporting, since adjusted neonatal iNMRs closely matched the original simulated values ("truth") even at varying levels of underreporting. Using a dataset of 172,809 admissions from 65 hospitals partnering with NEST360, unadjusted iNMRs were 121 deaths per 1000 admissions in Nigeria, 128 per 1000 in Tanzania, 132 per 1000 in Malawi, and 135 per 1000 in Kenya. Birthweight-specific adjustment functions were developed based on 26 hospitals with plausible mortality rates (≥700 deaths per 1000 live births for neonates <1000g). After applying adjustment curves, the iNMRs increased to 162 deaths per 1000 admissions in Nigeria, 183 per 1000 in Tanzania, 174 per 1000 in Malawi, and 172 per 1000 in Kenya. As expected, adjusted iNMRs were higher and reduced inter-hospital variability. WHAT NEXT? This adjustment method may be useful more widely, including in LMIC national routine health data, which is now the main source of neonatal data, given recent defunding of Demographic Health Surveys. Share learning to improve capture of all admissions and deaths in routine data systems, such as increased communication that all live-born babies are to be registered, irrespective of gestational age or weight, regular data audits, and educating/mentoring to promote accurate and complete reporting.
SN - 1478-7954
ER -