Accepted: 12 April 2023 DOI: 10.1111/1471-0528.17518 S U P P L E M E N T A R T I C L E Neonatal mortality risk of vulnerable newborns: A descriptive analysis of subnational, population- based birth cohorts for 238 143 live births in low- and middle-i ncome settings from 2000 to 2017 Elizabeth A. Hazel1 | Daniel J. Erchick1 | Joanne Katz1 | Anne C. C. Lee2 | Michael Diaz1 | Lee S. F. Wu1 | Keith P. West Jr3 | Abu Ahmed Shamim4 | Parul Christian3 | Hasmot Ali5 | Abdullah H. Baqui6 | Samir K. Saha7 | Salahuddin Ahmed8 | Arunangshu Dutta Roy8 | Mariângela F. Silveira9 | Romina Buffarini9 | Roger Shapiro10 | Rebecca Zash11 | Patrick Kolsteren12 | Carl Lachat12 | Lieven Huybregts12,13 | Dominique Roberfroid14,15 | Zhonghai Zhu16 | Lingxia Zeng16 | Seifu H. Gebreyesus17 | Kokeb Tesfamariam18 | Seth Adu-A farwuah19 | Kathryn G. Dewey20 | Stephaney Gyaase21 | Kwaku Poku- Asante21 | Ellen Boamah Kaali21,22 | Darby Jack23 | Thulasiraj Ravilla24 | James Tielsch25 | Sunita Taneja26 | Ranadip Chowdhury26 | Per Ashorn27 | Kenneth Maleta28 | Ulla Ashorn29 | Charles Mangani28 | Luke C. Mullany1 | Subarna K. Khatry30 | Vundli Ramokolo31,32 | Wanga Zembe- Mkabile33,34 | Wafaie W. Fawzi10 | Dongqing Wang35 | Christentze Schmiegelow36 | Daniel Minja37 | Omari Abdul Msemo37 | John P. A. Lusingu37 | Emily R. Smith38 | Honorati Masanja39 | Aroonsri Mongkolchati40 | Paniya Keentupthai41 | Abel Kakuru42 | Richard Kajubi42 | Katherine Semrau43,44,45 | Davidson H. Hamer46,47 | Albert Manasyan48 | Jake M. Pry49 | Bernard Chasekwa50 | Jean Humphrey1 | Robert E. Black1 | Subnational Collaborative Group for Vulnerable Newborn Mortality | Vulnerable Newborn Measurement Core Group Correspondence Elizabeth A. Hazel, Department of Abstract International Health, Johns Hopkins Objective: We aimed to understand the mortality risks of vulnerable newborns (de- Bloomberg School of Public Health, Baltimore, MD, USA. fined as preterm and/or born weighing smaller or larger compared to a standard Email: ehazel1@jhu.edu population), in low- and middle- income countries (LMICs). Funding information Design: Descriptive multi- country, secondary analysis of individual- level study data Children's Investment Fund Foundation, of babies born since 2000. Grant/Award Number: 2004-0 4670 Setting: Sixteen subnational, population-b ased studies from nine LMICs in sub- Saharan Africa, Southern and Eastern Asia, and Latin America. Population: Live birth neonates. Methods: We categorically defined five vulnerable newborn types based on size (large- or appropriate- or small- for- gestational age [LGA, AGA, SGA]), and term Elizabeth A. Hazel and Daniel J. Erchick are joint first authors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. BJOG: An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd. BJOG. 2023;00:1–12. wileyonlinelibrary.com/journal/bjo | 1 2 | HAZEL et al. (T) and preterm (PT): T + LGA, T + SGA, PT + LGA, PT + AGA, and PT + SGA, with T + AGA (reference). A 10- type definition included low birthweight (LBW) and non- LBW, and a four- type definition collapsed AGA/LGA into one category. We per- formed imputation for missing birthweights in 13 of the studies. Main Outcome Measures: Median and interquartile ranges by study for the prevalence, mortality rates and relative mortality risks for the four, six and ten type classification. Results: There were 238 143 live births with known neonatal status. Four of the six types had higher mortality risk: T + SGA (median relative risk [RR] 2.8, interquartile range [IQR] 2.0– 3.2), PT + LGA (median RR 7.3, IQR 2.3–1 0.4), PT + AGA (median RR 6.0, IQR 4.4– 13.2) and PT + SGA (median RR 10.4, IQR 8.6– 13.9). T + SGA, PT + LGA and PT + AGA babies who were LBW, had higher risk compared with non- LBW babies. Conclusions: Small and/or preterm babies in LIMCs have a considerably increased mortality risk compared with babies born at term and larger. This classification sys- tem may advance the understanding of the social determinants and biomedical risk factors along with improved treatment that is critical for newborn health. K E Y W O R D S low- and middle- income countries, obstetrics and gynaecology, paediatrics: neonatal, preterm, small- for- gestational age 1 | I N TRODUC TION vulnerable babies to effectively target interventions, policies and programmes.8 In 2020, 2.4 million babies died during the first month In countries with complete and high- quality vital regis- after birth, and over three- quarters of these deaths oc- tration data systems, it is possible to estimate national-l evel curred in two regions – sub- Saharan Africa (1.1 million birth outcomes and associated neonatal mortality risks. In deaths) and Southern Asia (0.9 million deaths).1 Neonatal LMICs without these systems, we cannot empirically gen- deaths (deaths that occur within 28 days after birth) have erate national estimates, but we can use population- based decreased in the past decades, from an estimated 45.6 subnational studies that collected high- quality data on birth deaths in 1990 to 27.1 deaths per 1000 livebirths in sub- outcomes and neonatal mortality to estimate the associated Saharan Africa and 57.1 to 23.2 deaths per 1000 livebirths neonatal mortality risks. in Southern Asia.1 Despite this progress, the mortality In this paper, we describe the neonatal mortality risks as- risk for neonates is unacceptably high and inequally dis- sociated with four, six and ten vulnerable newborn type clas- tributed. The 2014 Every Newborn Action Plan set a mor- sifications based on combinations of size- for- gestational age, tality rate target of 12 or fewer neonatal deaths per 1000 delivery at term or preterm, and low or not- low birthweight livebirths and this target is now also a part of the 2030 (Table 1).8 The estimates presented in this analysis are intended Sustainable Development Goal (SDG).2,3 only to describe the data available by study and should not be There is elevated mortality risk associated with babies interpreted as global, regional or country- level estimates. ‘born too early’ (preterm) or ‘born too small’ (small- for- gestational age [SGA]) or both. In prior analysis, we found babies born SGA are twice as likely to die in the neonatal 2 | M ETHODS period, preterm babies have seven times the mortality risk, and babies born both preterm and SGA have up to 15 times This is a secondary analysis of individual participant data the risk.4 Sub- Saharan Africa and Southern Asia combined from multiple studies; women and newborns did not have include 81% of preterm and 72% of low birthweight (LBW) direct participation in this study (Table  S1). We identified babies born globally.3,5,6 population-b ased studies in LMICs that collected data on Historically, LBW (<2500 g) has been used to identify vul- birthweight and gestational age at delivery for newborns born nerable newborns. LBW is caused by being preterm, having since 2000. Studies were identified through systematic review fetal growth restriction (FGR) or a combination of the two. of peer-r eviewed literature databases, clinical trial registries As the underlaying aetiology of preterm and SGA is differ- and open data repositories and through professional net- ent, it is important to consider these separately because the works. Further details of the study identification methods outcomes and preventive interventions will differ as well. have been presented elsewhere.9 Principal investigators could Additionally, babies born large- for- gestational age (LGA) send their de- identified data for central processing or perform (>90th centile compared with a standard population) have the analysis themselves with standard statistical code to as- excess health risks.7 A more granular classification system sess the quality of the data, construct standardised study out- is needed to identify and understand the different risks for comes, and generate study-s pecific estimates. 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MORTALITY RISK FOR VN TYPES IN LMICS | 3 T A B L E 1 Key findings. inclusion in the mortality analysis. Studies were excluded 1. What was known? if (1) they had fewer than 20 neonatal deaths (the reduced Babies born preterm and/or small are at higher risk of dying during sample size impeded investigation of  mortality risk by the neonatal period. Previously neonatal mortality was estimated type categorisation) or (2) data missingness was greater for these conditions separately. However, these conditions can than 70% among neonatal deaths (combined gestational overlap and may have compounding mortality risks. Disease and mortality burden for preterm and/or small babies is age at delivery, birthweight and infant sex). As missing higher in low- and- middle income countries (LMICs), also where type was primarily driven by missing birthweight, we im- data availability is the lowest. puted birthweight for studies with missing birthweight 2. What was done that is new? ranging from 10% to 70%. We systematically searched and identified 16 studies from nine Data quality of the included studies was assessed LMICs that collected high- quality, population- based data on using proportion of missing or improbable birthweights, birth outcomes with follow- up through the neonatal period from gestational age and missing sex. We excluded missing 2000- 2017. Our pooled dataset of 238,143 livebirths provides the measured (or unable to impute birthweight due to miss- first multi-c ountry mortality estimates of these newborn types in ing covariates), gestational age, sex or a gestational age LMICs. +0 +6 We defined and described the neonatal mortality risks for vulnerable <22 weeks or >44 weeks for which it was not possi- newborn types categorized by preterm (PT) and term (T), size-f or- ble to assess size- for-g estational age. Birth records with gestational age (small (SGA), appropriate (AGA) and large (LGA)) implausible measured or imputed birthweights (<250 and low birthweight (LBW) and non- LBW (nLBW). or ≥6500 g) or implausible combinations of measured 3. What was found? or imputed birthweight and gestational age (defined as Preterm risks: All preterm types had high neonatal mortality risk birthweight >5 standard deviations above the mean birth- with PT + SGA as the highest risk (median relative risk (RR) 10.4, weight for gestational age and sex) were excluded. We also interquartile range (IQR): 8.6– 13.9 by study). investigated heaping of birthweight (measured only) as Risks for babies born at term: T + SGA had additional risk (median RR: a measure of the data collection quality. We calculated a 2.8, IQR: 2.0– 3.2) and also the greatest prevalence (median: 25.0%, IQR: 18.8%– 41.5%) of the vulnerable types, indicating the highest heaping index by study defined as the number of births population mortality burden. T + LGA babies had no additional reported at exactly 2500 g divided by the number with detected risk compared to T + AGA babies. 249 g below and above 2500 g. Lower values of this heap- Usefulness of LBW categorization: T + SGA babies who were also ing index indicate higher quality data collection and doc- LBW had greater mortality risk (median RR 4.9, IQR: 3.1–6 .4) umentation practices. compared to T + SGA babies who were nLBW (median RR 1.7, IQR: 1.4, 2.2). In settings with high T-S GA prevalence, it may be programmatically important to track LBW as well. 4. What next? 2.2 | Description of recalibration and Action in preventive programmes: This categorization of vulnerable imputation methods newborn types provides more granular detail on mortality risks, useful for improving measurement, understanding the disease We imputed birthweight at the study level to calculate size- aetiology and epidemiology, and improving clinical care and for- gestational age in 13 studies (Table S2). Eight of the 13 population- based interventions. studies included infants with ‘birthweight’ measured in Research gaps: High quality routine data systems that include gestational age, birthweight, and sex for every live- and stillbirth the early neonatal period. For these studies, we first recali- with linked neonatal mortality data are needed to adequately track brated all infant weights to weight at the time of delivery vulnerable newborn population level health. based on a longitudinal model of daily weight measure- ments on newborns in the first 10 days. The longitudinal dataset was collected on a subset of infants enrolled in a clinical trial of chlorhexidine newborn cleansing from 2002 2.1 | Inclusion and exclusion criteria to 2005 in rural Nepal.10,11 We then used these recalibrated birthweights multiply to impute missing birthweight based We defined study- and individual- level exclusion criteria. To on maternal education, age and parity, single or multiple be included, studies must have sampled more than 300 live pregnancy, infant sex, gestational age and neonatal sur- births, assessed gestational age at delivery through early ul- vival status. Additional details on the recalibration and trasound or timing of last menstrual period (LMP), collected imputation methods have been previously published by the data after the year 2000, and be population- based including authors.12 both home and facility births. Studies that sampled facility- level births were included if 80% or more of the population delivered in a health facility. Studies that sampled from ante- 2.3 | Exposure and outcome definitions natal care (ANC) clinics were considered population- based if 90% or more pregnant women received at least one ANC We categorised every included newborn based on gestational visit in the areas sampled. age at delivery (preterm birth <37 completed weeks [PT] or Studies compiled for the prevalence paper that followed term ≥37 weeks [T]) and size- for- gestational age defined as survival for at least 28 days after delivery were assessed for SGA <10th centile; or LGA >90th centile or AGA between 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 4 | HAZEL et al. 10th and 90th centile using a modified version (extended to deaths in Brazil to 45.1 deaths in Bangladesh per 1000 live- include GA from 22+0 to 44+6 weeks) of the INTERGROWTH- births. Loss to follow- up was minimal. In most studies, <5% 21st international newborn size for gestational age and were lost to follow- up during the neonatal period, and in one sex standards.13 Different combinations of these outcomes study (Tanzania 2) 7.2% (Table S2). This subset of studies generate six mutually exclusive newborn types: T + AGA had a similar heaping index to the full set used in the prev- (reference), T + SGA, T + LGA, PT + SGA, PT + AGA and alence analysis.9 The median heaping index was 6.6% (IQR PT + LGA. We examined a four- type classification that 1.6%– 32.3%) and over a third (42%) had a heaping index collapsed LGA/AGA: T + nonSGA (reference), T + SGA, >10% (data not shown). PT + nonSGA and PT + SGA. Finally, we also generated a Most of the mothers enrolled in the studies had primary more complex classification (including LBW) for ten types and lower secondary educations (median by study: 68.7%, including T + AGA + nonLBW (reference), T + LGA + non- IQR 47.6%–7 8.7%), a third were between 20 and 24 years LBW, T + AGA + LBW, T + SGA + nonLBW, T + SGA + LBW, of age (median: 33.0%, IQR 29.1%–3 9.8%) and a third had PT + LGA + nonLBW, PT + LGA + LBW, PT + AGA + non- no previous births (median: 29.0%, IQR 21.2%–4 0.6%) LBW, PT + AGA + LBW and PT + SGA + LBW. To esti- (Table 2). Most deliveries took place at a health facility, but mate neonatal mortality risk, infant survival status was this varied by study (median: 70.0%, IQR 43.3%– 88.0%). documented in each included study for the first 28 days Almost all babies were delivered vaginally (median: 94.2%, (0–2 7 days) after delivery. Infants who were lost to follow- up IQR 92.5%–9 7.1%) and were singletons (98.0%, IQR 97.2%– were censored. 98.5%) (Table  2, Table S4 by study). Median female sex of the infants was 48.9% (IQR 48.2%– 49.9%) and no intersex babies were reported in the studies. 2.4 | Analysis The missingness of newborn type was primarily driven by missing birthweight, especially among the neonatal deaths We calculated the proportion of births excluded from the (Table S5). In Tanzania study 2, Tanzania study 3 and India analysis and reason for exclusion (i.e. missing or improb- study 1, we were unable to perform the imputation due to able data) by neonatal survival status in each study and data access/availability, but more than 90% of birthweights described the demographic and obstetric characteristics. were measured in the first 24 hours after delivery and birth- We calculated type prevalence, neonatal mortality rate weight missingness was very low. China, Brazil, Burkina (NMR), defined as the number of neonatal deaths per 1000 Faso, Tanzania study 1 and Zambia study 2 had higher miss- livebirths, crude relative risk ratios (RR) and 95% confi- ingness (ranging from 37.8% to 12.2% among the neonatal dence intervals (95% CI). We reported these statistics for deaths) and more than 90% of birthweights were measured each study and then the overall median and interquartile in the first 24 hours. Our recalibration protocol does not im- range (IQR) by type. As these study level estimates were prove on weights measured in the first 24 hours after deliv- included in a global model of type prevalence and mortal- ery, so we did not perform the recalibration for these studies ity risks and the analytical aim is descriptive, we did not and instead used the measured weights to conduct the mul- perform meta-a nalyses.14 tiple imputation for the missing weights. For the other studies, we recalibrated the birthweights to time of delivery and used those for the multiple imputation. 3 | R E SU LTS In our recalibration model, twin/triplets, first- born infants and babies that later died during the neonatal period had a We identified 29 studies: five were excluded due to fewer lower estimated birthweight. Higher gestational age at de- than 20 neonatal deaths in the study, and five were excluded livery, higher maternal age and educational status, and male for other reasons, resulting in 19 studies (Figure 1).15–2 2 In six sex were associated with increased estimated birthweight studies from Burkina Faso, Malawi and one of the Tanzania (Table S6). We imputed a birthweight for 11 301/246 276 sur- studies, we pooled the data (two studies per country) that viving neonates (4.6%) and 2815/6636 (42.4%) of the neona- were carried out in the same site and by the same study tal deaths. After birthweight imputation, 5.2% of the deaths teams, giving us 16 studies. We assigned study ID based on and 5.9% of the surviving infants were excluded due to miss- country and timing of the data collection (Table S2). ing or improbable data, resulting in 238 046 live births of a This analysis includes data from 16 subnational datasets known type. from nine countries with data collected from 2000 to 2017 In these studies, T + AGA babies were the most prevalent (Table S3).23– 41 Seven studies were based in sub-S aharan (median: 52.0%, IQR 40.5%– 61.6%), followed by T + SGA Africa, seven in Southern Asia, one in Eastern Asia and one (median: 25.0%, IQR 18.8%–4 1.5%) and then PT + AGA (me- in Latin America. Gestational age of the pregnancy was cal- dian: 9.0%, IQR 7.8%– 11.3%) (Figure 2, Table S7 by study). culated from LMP collected during pregnancy for 11 studies, PT + LGA, T + LGA and PT + SGA had median prevalences one study collected LMP during pregnancy and at delivery, of <5%. T + AGA and T + LGA had similar neonatal mortal- two studies used ultrasound estimation, and two studies ity rates (median 7.8 deaths per 1000 livebirths, IQR 6.5– 13.0 used a combination of ultrasound/LMP (Table S2). Neonatal and median 5.7, IQR 0– 9.4, respectively). T + SGA had the mortality rates measured in the studies ranged from 8.8 next highest mortality rate (median 28.8, IQR 16.7–3 0.5), 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MORTALITY RISK FOR VN TYPES IN LMICS | 5 29 studies collected mortality data 10 studies excluded: 5: <20 deaths 1: type missingness among NN deaths was >70% 3: Data quality issues 1: High missingness, variables needed for imputation not collected 19 studies (16 studies after combining) 256,995 livebirths 4,083 lost-to-follow up 252,912 known neonatal status 246,276 alive/ 6,636 deceased 14,116 missing birthweight imputed 11,301 alive / 2,815 deceased 14,769 excluded: (14,425 alive/344 deceased) 665 gender missing (643 alive/22 deceased) 8837 gestational age missing (8615 alive/222 deceased) 2800 gestational age/birthweight improbable (2749 alive/51 deceased) 9615 birthweight missing after imputation (9333 alive/282 deceased) 238,143 known neonatal status and type F I G U R E 1 Flowchart of studies and live births included in the mortality analysis by type. followed by PT + AGA and PT + LGA (median 70.2, IQR 8.7–1 3.9). Compared with T + nonSGA babies, the me- 39.3–9 0.1 and median 76.2, IQR 22.1–1 05.6, respectively). dian RR for PT + nonSGA babies was 6.0 (IQR 4.1– 14.5) PT + SGA had the highest median mortality rate (median: (Table 3, Table S8 by study). 116.4, IQR 66.5–1 47.8). The collapsed T + AGA/LGA cat- Among the T + SGA, the median RR for babies who were egory (T + nonSGA) had a median of 57.2% by study (IQR also LBW was 4.9 (IQR 3.1–6 .4) and the median RR was 1.7 41.8%–6 5.3%) and neonatal mortality rate of 7.8 (IQR 6.4– (IQR 1.4– 2.2) for babies who were not LBW. PT + LGA + LBW 13.1). The PT + nonSGA median prevalence was 14.1% (IQR had a much higher median RR (23.1, IQR 16.2– 40.6) com- 11.8%–1 7.6%) and the median mortality rate was 76.3 (IQR pared with PT + LGA + nonLBW (RR 1.1, IQR 0.8– 2.0). 44.0– 92.9). Finally, PT + AGA babies who were LBW also had a higher Compared with T + AGA babies, T + LGA had similar median RR (12.9, IQR 8.9–2 6.5) compared with babies born risk of death (median RR 0.9, IQR 0– 1.0) (Table 3, Table S7 non-L BW (RR 1.7, IQR 1.4– 1.9) (Table 3, Table S9 by study). by study). All other types had a higher risk of deaths: the risk of T + SGA babies dying in the neonatal period was 2.8 times higher (IQR 2.0– 3.2), the risk of PT + LGA and 4 | DISCUSSION PT + AGA babies dying was approximately seven times higher (median RR 7.3, IQR 2.3– 10.4 and median RR 6.0, We identified 16 subnational datasets from nine low- and IQR 4.4– 13.2, respectively) and the risk of PT + SGA ba- middle-i ncome countries with data collection from 2000 to bies dying was over 10 times higher (median RR 10.4, IQR 2017 to estimate the neonatal mortality risk of vulnerable 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 6 | HAZEL et al. T A B L E 2 Demographic characteristics of the included studies, median and interquartile range (IQR) of the included studies. T + AGA babies. We found no additional risk of LGA in this sample. For preterm babies, AGA and LGA mortality Median, % (Interquartile rates and relative risks were similar and T + LGA babies had range, %) equivalent mortality risks to T + AGA babies. Generally, the Years of education of mother RR for each type is lower than the national data from higher 9 No formal education (0 years) 25.7 (8.0– 40.8) income countries. This is due to the higher mortality risk of Primary and lower secondary 68.7 (47.6– 78.7) babies in the reference group (T + AGA, 7.8 deaths per 1000 (≤11 years) livebirths) compared with the national datasets in high in- Upper secondary and above 4.5 (2.7– 11.7) come countries (T + AGA, 0.6 deaths per 1000 live births); (≥12 years) this difference has been documented in other studies on 4 Missing 0.2 (0– 0.4) preterm and SGA mortality. There is considerable variation in our estimates by study, Age of mother related to the heterogeneity of the underlying populations. <15 years 0.1 (0–0 .5) These studies represent geographical variation in LMICs, 15– 19 years 16.5 (11.3–2 4.7) but temporal variation as well. For instance, we have three 20– 24 years 33.0 (29.1– 39.8) studies in different regions of Tanzania; the 2001– 2004 study 25– 29 years 24.9 (21.6– 27) in urban Dar es Salaam had a neonatal mortality rate of 28.5 30–3 9 years 20 (11.7– 25.4) deaths per 1000 livebirths and the 2012– 2013 study in Dar ≥40 years 1.4 (0.5–2 .2) es Salaam and Morogoro regions had 9.5 deaths per 1000 livebirths. Missing 0.3 (0– 0.5) This sub-s ample of studies that collected mortality data Place of delivery was similar to the set of studies used for birth type prev- Outside of facility 27.9 (11.8– 53.1) alence estimates, but there is a slightly higher proportion At facility 70.0 (43.3– 88.0) of vulnerable newborns.9 In the studies for the prevalence Missing 0.5 (0– 1.8.0) estimates, 58.5% were T + AGA versus 52% in the mortality Type of delivery sub- sample and 21.9% T + SGA versus 25% in the mortality Vaginal 94.2 (92.5–9 7.1) sample. The other vulnerable type prevalences were similar: 7.4% versus 9% for PT- AGA, 3.3% versus 2% for T + LGA, Caesarean 5.7 (1.7–7 .5) 1.7% versus 4% for PT + LGA, and PT + SGA was the same as Missing 0.8 (0– 1.2) the mortality sample. The higher proportion of SGA is likely Parity due to the study site locations; in this analysis almost half 0 29.0 (21.2– 40.6) of the studies were in Southern Asia, which has the highest 1 26.6 (23.1–3 0.5) regional prevalence of SGA. 42 2 17.7 (15.4– 19.1) We chose commonly used categorical definitions for preterm, SGA and LGA to define vulnerable newborns. 3 11.6 (7.3– 13.1) The 10th centile definition for SGA has been used since the ≥4 11.5 (4.5– 20.4) 1960s but further studies are needed to determine whether Missing 0.1 (0– 0.6) these definitions should be revised.43 Additionally, there is Number born evidence that revising the definition of LGA as >97th cen- Singleton 98.0 (97.2– 98.5) tile would better discriminate the vulnerable babies.44 We Multiples 2.1 (1.6–2 .8) also recognise the importance of capturing the risk of babies Infant gender born extremely early or post- term, but for simplicity in this initial examination of vulnerable newborn type risk, we re- Male 51.1 (50.1– 51.8) stricted ourselves to term and preterm. Finally, we used the Female 48.9 (48.2– 49.9) INTERGROWTH-2 1st international standard allowing for direct comparison across many studies. The four- or six-t ype definitions are less complex and newborn types. This analysis provides information on new- easier to interpret for programme and policy improvements born health in geographical settings where neonatal mortal- compared with the ten- type definition; however, there is ev- ity is the highest globally but data availability is the lowest. idence that the lower birthweights within preterm and SGA Newborns born at term or later and SGA, and preterm types confer higher mortality risk. T + SGA, PT + LGA and babies born either LGA or AGA had elevated mortality risk. PT + AGA babies who were also LBW had higher relative Preterm newborns born SGA had the highest risk; they were risk compared with their non- LBW counterparts, a find- 10 times more likely to die in the first month but had the ing reported in other analyses (reference: T + AGA + non- lowest prevalence (1%). Of all the vulnerable newborns with LBW). In an analysis from the CHERG study, babies born increased mortality, the highest prevalence was for T + SGA T + SGA + nonLBW had a RR of 1.89 of neonatal mortal- (25%), with 2.5 times the risk of mortality compared with ity, compared with 4.77 for T + SGA babies born LBW.45 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MORTALITY RISK FOR VN TYPES IN LMICS | 7 60 400.0 Prevalence Neonatal deaths per 1000 livebirths 350.0 50 300.0 40 250.0 30 200.0 150.0 20 116.4 100.0 10 76.2 70.2 50.0 28.8 0 7.8 5.7 0.0 T+5A2GA T+2LGA T+2S5GA PT+4LGA PT+9AGA PT+1SGA Prevalence Neonatal deaths per 1000 livebirths 60 300.0 50 250.0 40 200.0 30 150.0 116.4 20 100.0 76.3 10 50.0 28.8 0 7.8 0.0 T+nSGA T+SGA PT+nSGA PT+SGA F I G U R E 2 Median prevalence and neonatal mortality risk by study, four- and six- type categorisation. AGA, appropriate for gestational age; LGA, large for gestational age; PT, preterm; SGA, small for gestational age; T, term. Birthweights provide additional information on mortality our studies were community- based (n = 11 studies) and, for risk for each of the vulnerable types, even if just indicating the home deliveries, early neonatal deaths occurred before babies born at the lower centile of SGA. PT + LGA + LBW ba- the study team could arrive at the home to weigh the baby bies also had additional mortality risk compared with their (Table S3). Additionally, newborns typically lose weight in nonLBW counterparts, but this is likely a measurement ar- the first 2–3 days of life due to fluid losses until the estab- tefact. To be considered PT, LGA and LBW, boys must be lishment of breastfeeding. Weight measured in the 2– 3 days born <33 weeks and girls born <33+4 weeks gestational age, after delivery, at the nadir of early neonatal weight loss, in- so the mortality risk is likely associated with being born flates estimates of SGA and underestimates LGA.12,46 Using early, rather than LGA. only the measured birthweight to calculate mortality risk by In addition to the limitations presented in the subna- type would have underestimated the overall mortality rates tional prevalence paper in this series, the main limitation of and the mortality risk of certain vulnerable newborn types. this mortality analysis is the missing birthweights, especially We aimed to address this bias using the recalibrated birth- among neonatal deaths.9 We imputed almost half of the neo- weights to generate imputed birthweight. The recalibrated natal deaths used in this analysis (42%, Figure 1). Many of weights were based on a longitudinal sample of singleton Prevalence (%) Prevalence (%) Neonatal deaths per 1000 livebirths Neonatal deaths per 1000 livebirths 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 8 | HAZEL et al. T A B L E 3 Relative risk of neonatal mortality for the six types (reference: Term + AGA), four types (reference: Term + nonSGA) and 10 For a subset of the studies, we compared the median types (reference: Term + AGA + nonLBW), median and interquartile range birthweight, and four- and six- type neonatal mortality rates (IQR) of the studies. and relative risks using (1) birthweights measured <72 hours after delivery with missing birthweight excluded, and (2) Relative risk of neonatal mortality birthweights using the recalibration and/or imputation Median (Interquartile range); number method (Tables S10– S12). of studies with sufficient data on type In the studies where only imputation (not recalibration) Six newborn types was used, the median birthweight using the imputation did T + AGA Reference not change or was <5 g different (China, Brazil and Zambia study 2) or increased (Burkina Faso). In studies where the T + LGA 0.9 (0– 1.0); n = 16 recalibration with imputation was used, the median birth- T + SGA 2.8 (2.0–3 .2); n = 16 weight increased for the Bangladesh study 3 (+50 g), India PT + LGA 7.3 (2.3– 10.4); n = 15 study 1 (+40 g), Nepal study 1 (+40 g), Nepal study 2 (+21 g) PT + AGA 6.0 (4.4–1 3.2); n = 16 and Zambia study 1 (78 g) studies, with the exception of PT + SGA 10.4 (8.6–1 3.9); n = 14 Malawi, where it essentially stayed the same. Four newborn types The recalibration protocol estimates a weight closer to the time of birth, and many of these studies had a significant T + nonSGA Reference portion of babies measured at the nadir of early neonatal T + SGA 2.7 (2.1–4 .0); n = 16 weight loss (estimated 1– 2 days after delivery). Therefore, PT + nonSGA 6.0 (4.1– 14.5); n = 16 for these five studies, the median birthweight was increased PT + SGA 10.4 (8.5–1 4.5); n = 14 slightly because an estimated birthweight (at time of deliv- Ten newborn types ery) was higher than that measured, as many of the infants T + AGA + nonLBW Reference were measured at the nadir. T + AGA + LBW 1.8 (0.2– 3.0); n = 10 As expected, the mortality risks of all types increased when using the recalibration and/or imputation method, T + LGA + nonLBW 0.7 (0– 0.9); n = 16 as a birthweight is now imputed for early neonatal deaths, T + SGA + nonLBW 1.7 (1.4–2 .2); n = 16 where previously they were excluded (Tables S11 and S12). In T + SGA + LBW 4.9 (3.1–6 .4); n = 16 the four- type categorisation, the median RR of T + SGA in- PT + LGA + nonLBW 1.1 (0.8– 2.0); n = 15 creased from 2.1 to 2.3 and the median RR of nonSGA + PT PT + LGA + LBW 23.1 (16.2– 40.6); n = 14 increased from 4.5 to 5.9 (Table S11). The median RR of PT + AGA + nonLBW 1.7 (1.4– 1.9); n = 15 SGA + PT decreased from 13.4 to 10.4. For the six-t ype categorisation, the median RR for T + LGA, T + SGA and PT + AGA + LBW 12.9 (8.9–2 6.5); n = 16 PT + AGA were similar using the measured and imputed PT + SGA + LBW 10.6 (8.8– 14.7); n = 14 birthweights (1.0 versus 1.0; 2.1 versus 2.3 and 5.7 versus 5.9, Abbreviations: AGA, appropriate for gestational age; LGA, large for gestational respectively). The median RR of PT + LGA increased from age; nonSGA, non- SGA (AGA and LGA combined); PT, preterm; SGA, small for 2.2 to 7.6 and the median RR of PT + SGA decreased from gestational age; T, term. 12.1 to 9.7 (Table S12). Our method estimated more neonatal deaths with a newborns in rural Nepal born between 2002 and 2005 who missing birthweight as PT + LGA using the six-t ype cat- survived at least 10 days. There is evidence that babies may egorisation (PT + nonSGA for the four- type) and fewer have different early neonatal growth patterns in different re- for PT + SGA. This could be due to an actual biological gions due to underlying population health or newborn feeding construct, measurement error with gestational age or our practices. A study in Tanzania of early neonatal weight change model overestimating birthweights for preterm babies. found an earlier nadir (27 hours for boys and 28 hours for girls) However, we consider this model an improvement on the than that measured in the Nepal study (2.1 days), although the measured birthweight data given we can include the neo- mean weight loss at the nadir was similar (4.7% in Tanzania natal deaths with missing birthweight, critical for this and 4.3% in Nepal).12,47 A study cohort of infants from Nepal, analysis on neonatal mortality risk. Most studies used Pakistan, Guinea-B issau and Uganda found a similar median LMP collected during pregnancy (n = 12) to calculate ges- nadir of two days, with an average mean weight loss of 5.9%, tational age. There could be measurement error that im- and babies born LBW had a slower growth trajectory over pacted the size- for- gestational age estimates. Although 30 days.48 Also, to be included in the rural Nepal longitudinal ultrasound measurement in the <24- week period is rec- sample of weights measured in the first 10 days, the baby must ommended by WHO for ascertainment of gestational age, have survived the early neonatal period. We included a covari- LMP is adequate in areas where access to ultrasound is ate adjusting for neonatal death for the multiple imputations limited.49– 51 We also used birthweight standard curves, in- (Table S6) but we do not have any information on how well stead of fetal weight standard curves, which underestimate our recalibration model estimates weight at time of delivery FGR for preterm babies, as the pathology that leads to FGR for early neonatal deaths. may also induce preterm births.52 However, there are also 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MORTALITY RISK FOR VN TYPES IN LMICS | 9 limitations with use of a universal fetal growth standard datasets was undertaken by DE, EH, MD and LSW. The in international settings. A study applying three differ- paper was drafted by EH with DE, JK, ACL, MD, and REB. ent fetal growth standards found important differences All authors helped revise the paper. All authors reviewed and in classification of SGA and LGA babies, indicating more agreed on the final version. work is needed on universal standards of fetal growth.53 A final limitation was that we presented crude mea- A F F I L I AT ION S sures of mortality risk for newborn types. Potential con- 1International Health Department, Johns Hopkins Bloomberg School of Public founders of neonatal mortality risk and newborn type Health, Baltimore, Maryland, USA2 range by socio- economic factors, underlying health of Pediatric Newborn Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA the maternal population, health system factors and many 3Department of International Health, Center for Human Nutrition, Johns Hopkins more exposures. There was limited information on spon- Bloomberg School of Public Health, Baltimore, Maryland, USA taneous versus vacuum- or forceps- assisted vaginal de- 4BRAC JP Grant School of Public Health, Dhaka, Bangladesh livery, emergency versus planned caesarean section, and 5JiVitA Maternal and Child Health Research Project, Rangpur, Bangladesh presentation of the newborn (i.e. breech). We were limited 6Department of International Health, Johns Hopkins Bloomberg School of Public by data collected in the studies and hope to address this in Health, Baltimore, Maryland, USA future research. 7Child Health Research Foundation, Dhaka, Bangladesh This analysis is possible due to the generous collabora- 8Projahnmo Research Foundation, Dhaka, Bangladesh tion of our co-a uthors and represents what is achievable 9Post-G raduate Program in Epidemiology – Federal University of Pelotas, Pelotas, with increased data availability and sharing. As health Brazil10 data systems improve in completeness and quality, coun- Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA11 tries will be able directly to track the health of vulnerable Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA12 newborns but, until then, the global health community Department of Food Technology, Safety and Health, Ghent University, Ghent, Belgium relies on research data. The authors support continued 13Poverty, Health and Nutrition Division, International Food Policy Research openness and availability of de- identified, individual- level Institute, Washington, District of Columbia, USA study data. 14Namur University, Namur, Belgium Babies in low- and middle- income settings who are 15Belgian Health Care Knowledge Centre, Brussels, Belgium preterm or growth- restricted have considerable mortality 16Department of Epidemiology and Biostatistics, School of Public Health, Xi'an risk compared with full term and not growth-r estricted ba- Jiaotong University Health Science Centre, Xi'an, China bies born in the same location. All preterm types had higher 17Department of Nutrition and Dietetics, School of Public Health, Addis Ababa neonatal mortality risks compared with the term types and University, Addis Ababa, Ethiopia18 there was compounding risk of preterm with SGA. Term Department of Food Technology, Safety, and Health, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium SGA babies have lower risk compared with preterm babies 19Department of Nutrition and Food Science, University of Ghana, Accra, Ghana but are the most prevalent vulnerable newborn type. Four- 20Department of Nutrition, Institute for Global Nutrition, University of California, or six-t ype definitions were less complex to calculate and in- Davis, California, USA terpret, especially the four- type definition, as we did not find 21Kintampo Health Research Centre, Kintampo, Ghana evidence of differential risk between AGA and LGA babies in 22Research and Development Division, Ghana Health Service, Accra, Ghana this sample. The ten- type definition shows that babies with 23Columbia University's Mailman School of Public Health, New York, New York, USA LBW have higher risks but, as an population-l evel indicator 24Aravind Eye Hospital, Madurai, India of neonatal health, this is difficult to calculate and interpret, 25George Washington University Milken Institute School of Public Health, and some categories are measurement artefacts such as the Washington, District of Columbia, USA PT + LGA + LBW, which only captures early preterm babies 26Centre for Health Research and Development, Society for Applied Studies, New (<33 weeks for boys and <33+4 weeks for girls), likely indicat- Delhi, India27 ing the risk of early preterm rather than LGA or LBW status. Faculty of Medicine and Health Technology, Tampere University and Tampere University Hospital, Tampere, Finland This study provides critical information on vulnerable 28School of Global and Public Health, Kamuzu University of Health Sciences, newborn health in areas where the burden is the highest but Blantyre, Malawi data availability is the lowest. The classification of births as 29Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland preterm and/or SGA may assist in the understanding of the 30NNIPS, Kathmandu, Nepal social determinants and biomedical risk factors that are im- 31HIV and Other Infectious Diseases Research Unit, South African Medical portant to design and implement preventive interventions, Research Council, Cape Town, South Africa as well as improved management of vulnerable newborns. 32Gertrude H Sergievsky Center, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, USA 33 AU T HOR C ON T R I BU T ION S Health Systems Research Unit, South African Medical Research Council, Cape Town, South Africa The Vulnerable Newborn Measurement Collaboration was 34South African Research Chair in Social Policy at College Graduate of Studies, planned by JEL and REB. This analysis was designed by University of South Africa, Pretoria, South Africa DE, EH, JK and ACL with REB. All authors contributed to 35Department of Global and Community Health, College of Public Health, George the study protocol and analysis. Descriptive analysis of the Mason University, Fairfax, Virginia, USA 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 10 | HAZEL et al. 36Centre for Medical Parasitology, Department of Immunology and Microbiology, University of Copenhagen, and Department of Infectious Diseases, Copenhagen Anne C. C. Lee  https://orcid.org/0000-0003-2654-9862 University Hospital, Copenhagen, Denmark Christentze Schmiegelow  https://orcid. 37National Institute of Medical Research, Tanga, Tanzania org/0000-0002-9360-9741 38Department of Global Health, Milken Institute School of Public Health, Katherine Semrau  https://orcid. Washington, District of Columbia, USA org/0000-0002-8360-1391 39Ifakara Health Institute, Dar es Salaam, Tanzania Jake M. Pry  https://orcid.org/0000-0001-6312-4420 40ASEAN Institute for Health Development, Mahidol University, Salaya, Thailand 41College of Medicine and Public Health, Ubon Ratchathani University, Ubon R E F E R E N C E S Ratchathani, Thailand 42 1. United Nations Inter- agency Group for Child Mortality Estimation Infectious Diseases Research Collaboration, Kampala, Uganda (UN IGME). Levels and trends in child mortality. New York: United 43Ariadne Labs, Brigham and Women's Hospital and Harvard T.H. Chan School of Nationas Children's Fund; 2021. Public Health, Boston, Massachusetts, USA 2. World Health Organization. Every newborn: an action plan to end 44Division of Global Health Equity, Department of Medicine, Brigham and preventable deaths. Geneva: World Health Organization; 2014. Women's Hospital, Boston, Massachusetts, USA 3. Lawn JE, Blencowe H, Oza S, You D, Lee ACC, Waiswa P, et al. Every 45Department of Medicine, Harvard Medical School, Boston, Massachusetts, USA newborn: progress, priorities, and potential beyond survival. Lancet. 46Department of Global Health, Boston University School of Public Health, Boston, 2014;384(9938):189–2 05. Massachusetts, USA 4. Katz J, Lee AC, Kozuki N, Lawn JE, Cousens S, Blencowe H, et al. 47Section of Infectious Diseases, Department of Medicine, Boston University Mortality risk in preterm and small- for- gestational-a ge infants in Chobanian & Avedisian School of Medicine, Boston, Massachusetts, USA low- income and middle-i ncome countries: a pooled country analysis. 48University of Alabama at Birmingham, Birmingham, Alabama, USA Lancet. 2013;382(9890):417– 25. 49 5. Chawanpaiboon S, Vogel JP, Moller AB, Lumbiganon P, Petzold M, Centre for Infectious Disease Research in Zambia, Lusaka, Zambia Hogan D, et al. Global, regional, and national estimates of levels of 50Zvitambo Institute for Maternal and Child Health Research, Harare, Zimbabwe preterm birth in 2014: a systematic review and modelling analysis. Lancet Glob Health. 2019;7(1):e37– 46. 6. Blencowe H, Krasevec J, de Onis M, Black RE, An X, Stevens GA, AC K NOW L E D G E M E N T S et al. National, regional, and worldwide estimates of low birthweight We would like to acknowledge Dr Birgitte Bruun Nielsen for in 2015, with trends from 2000: a systematic analysis. Lancet Glob the STOPPAM and FOETALforNCD studies. We thank all Health. 2019;7(7):e849–6 0. relevant funders for their investments to enable the input 7. Mendez-F igueroa H, Truong VTT, Pedroza C, Chauhan SP. Large for data for each of the included studies. Most importantly, we gestational age infants and adverse outcomes among uncomplicated pregnancies at term. Am J Perinatol. 2017;34(7):655–6 2. thank all women and families included in subnational data- 8. Ashorn P, Black RE, Lawn JE, Ashorn U, Klein N, Hofmeyr J, et al. sets as well as all members of the subnational study teams. The lancet small vulnerable newborn series: science for a healthy start. Lancet. 2020;396(10253):743– 5. F U N DI NG I N FOR M AT ION 9. Erchick DJ, Hazel EA, Katz J, Lee ACC, Diaz M, Wu LSF. Vulnerable The Children's Investment Fund Foundation, grant 2004- newborn types: analysis of subnational, population-b ased birth co-horts for 541,285 live births in 23 countries, 2000 to 2021. BJOG. 04670. The funders had no role in the study design, data col- 2023;in press. lection, analysis or interpretation of the paper. 10. Tielsch JM, Darmstadt GL, Mullany LC, Khatry SK, Katz J, LeClerq SC, et al. Impact of newborn skin-c leansing with chlorhexidine on C ON F L IC T OF I N T E R E S T S TAT E M E N T neonatal mortality in southern Nepal: a community- based, cluster- None declared. randomized trial. Pediatrics. 2007;119(2):e330– 40. 11. Mullany LC, Darmstadt GL, Khatry SK, Katz J, LeClerq SC, Shrestha S, et al. Topical applications of chlorhexidine to the umbilical cord DATA AVA I L A BI L I T Y S TAT E M E N T for prevention of omphalitis and neonatal mortality in south- Data sharing and transfer agreements were jointly devel- ern Nepal: a community- based, cluster- randomised trial. Lancet. oped and signed by all collaborating partners. The pooled 2006;367(9514):910– 8. summary table data generated during the current study have 12. Hazel EA, Mullany LC, Zeger SL, Mohan D, Subedi S, Tielsch JM, et al. Development of an imputation model to recalibrate birth been deposited online with data access subject to approval at weights measured in the early neonatal period to time at delivery and https://doi.org/10.17037/ DATA.00003095. assessment of its impact on size-f or-g estational age and low birth- weight prevalence estimates: a secondary analysis of a pregnancy co- E T H IC S A PPROVA L hort in rural Nepal. BMJ Open. 2022;12(7):e060105. The Vulnerable Newborn Measurement Collaboration was 13. Villar J, Cheikh Ismail L, Victora CG, Ohuma EO, Bertino E, Altman DG, et al. International standards for newborn weight, length, granted ethical approval by the Institutional Review Boards and head circumference by gestational age and sex: the newborn of the London School of Hygiene & Tropical Medicine (ref: cross- sectional study of the INTERGROWTH-2 1st project. Lancet. 22858) and Johns Hopkins University (ref: 16439). All col- 2014;384(9946):857– 68. laborators received local ethical permission for their data 14. Lawn JE, Ohuma EO, Bradley E, Idueta LS, Hazel EA, Okwaraji YB. where relevant. Small babies, big risks: global estimates of prevalence and mortailty for vulnerable newborns to accelerate change and improve counting. Lancet. 2023;19– 31. ORC I D 15. Tesfamariam K, Argaw A, Hanley-C ook GT, Gebreyesus SH, Elizabeth A. Hazel  https://orcid. Kolsteren P, Belachew T, et al. Multiple mycotoxin exposure during org/0000-0002-9176-3278 pregnancy and risks of adverse birth outcomes: a prospective cohort Daniel J. Erchick  https://orcid.org/0000-0002-2852-280X study in rural Ethiopia. Environ Int. 2022;160:107052. 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MORTALITY RISK FOR VN TYPES IN LMICS | 11 16. Adu-A farwuah S, Lartey A, Okronipa H, Ashorn P, Zeilani M, 33. Semrau KEA, Herlihy J, Grogan C, Musokotwane K, Yeboah- Peerson JM, et al. Lipid- based nutrient supplement increases the Antwi K, Mbewe R, et al. Effectiveness of 4% chlorhexidine umbil- birth size of infants of primiparous women in Ghana. Am J Clin Nutr. ical cord care on neonatal mortality in Southern Province, Zambia 2015;101(4):835– 46. (ZamCAT): a cluster-randomised controlled trial. Lancet Glob 17. Mulowooza J, Santos N, Isabirye N, Inhensiko I, Sloan NL, Shah S, Health. 2016;4(11):e827– 36 et al. Midwife- performed checklist and ultrasound to identify obstet- 34. Tembo T, Koyuncu A, Zhuo H, Mwendafilumba M, Manasyan A. ric conditions at labour triage in Uganda: a quasi-e xperimental study. The association of maternal age with adverse neonatal outcomes Midwifery. 2021;96:102949. in Lusaka, Zambia: a prospective cohort study. BMC Pregnancy 18. Baqui AH, Arifeen SE, Williams EK, Ahmed S, Mannan I, Rahman Childbirth. 2020;20(1):684. SM, et al. Effectiveness of home- based management of newborn in- 35. Klemm RDW, Merrill RD, Wu L, Shamim AA, Ali H, Labrique A, fections by community health workers in rural Bangladesh. Pediatr et al. Low- birthweight rates higher among Bangladeshi neonates Infect Dis J. 2009;28(4):304– 10. measured during active birth surveillance compared to national sur- 19. Isaranurug S, Mo- suwan L, Choprapawon C. A population- based vey data. Matern Child Nutr. 2015;11(4):583– 94. cohort study of effect of maternal risk factors on low birthweight in 36. West KP, Shamim AA, Mehra S, Labrique AB, Ali H, Shaikh Thailand. J Med Assoc Thai. 2007;90(12):2559– 64. S, et al. Effect of maternal multiple micronutrient vs iron-f olic 20. Zash R, Holmes L, Diseko M, Jacobson DL, Brummel S, Mayondi G, acid supplementation on infant mortality and adverse birth out- et al. Neural-t ube defects and antiretroviral treatment regimens in comes in rural Bangladesh: the JiVitA- 3 randomized trial. JAMA. Botswana. N Engl J Med. 2019;381(9):827–4 0. 2014;312(24):2649– 58. 21. Jack DW, Asante KP, Wylie BJ, Chillrud SN, Whyatt RM, Ae- 37. Saha SK, Schrag SJ, El Arifeen S, Mullany LC, Shahidul Islam M, Ngibise KA, et al. Ghana randomized air pollution and health study Shang N, et al. Causes and incidence of community- acquired serious (GRAPHS): study protocol for a randomized controlled trial. Trials. infections among young children in south Asia (ANISA): an observa- 2015;16:420. tional cohort study. Lancet. 2018;392(10142):145–5 9. 22. Humphrey JH, Mbuya MNN, Ntozini R, Moulton LH, Stoltzfus RJ, 38. Mitra DK, Mahmud A, Begum N, Rafiqullah I, Roy A, Moin SMI, et al. Tavengwa NV, et al. Independent and combined effects of improved Implementation of the ANISA protocol in Sylhet, Bangladesh: chal- water, sanitation, and hygiene, and improved complementary feed- lenges and solutions. Pediatr Infect Dis J. 2016;35(5 Suppl 1):S55– 9. ing, on child stunting and anaemia in rural Zimbabwe: a cluster- 39. Rahmathullah L, Tielsch JM, Thulasiraj RD, Katz J, Coles C, Devi S, randomised trial. Lancet Glob Health. 2019;7(1):e132–4 7. et al. Impact of supplementing newborn infants with vitamin A on 23. Zeng L, Dibley MJ, Cheng Y, Dang S, Chang S, Kong L, et al. Impact early infant mortality: community based randomised trial in south- of micronutrient supplementation during pregnancy on birth ern India. BMJ. 2003;327(7409):254. weight, duration of gestation, and perinatal mortality in rural west- 40. Mazumder S, Taneja S, Bhatia K, Yoshida S, Kaur J, Dube B, ern China: double blind cluster randomised controlled trial. BMJ. et al. Efficacy of early neonatal supplementation with vitamin 2008;337:a2001. A to reduce mortality in infancy in Haryana, India (Neovita): 24. Hallal PC, Bertoldi AD, Domingues MR, da Silveira MF, Demarco a randomised, double- blind, placebo-c ontrolled trial. Lancet. FF, da Silva ICM, et al. Cohort profile: the 2015 Pelotas (Brazil) birth 2015;385(9975):1333– 42. cohort study. Int J Epidemiol. 2018;47(4):1048–8 h. 41. Summers A, Visscher MO, Khatry SK, Sherchand JB, LeClerq SC, 25. Roberfroid D, Huybregts L, Lanou H, Henry MC, Meda N, Menten Katz J, et al. Impact of sunflower seed oil versus mustard seed oil J, et al. Effects of maternal multiple micronutrient supplementation on skin barrier function in newborns: a community- based, cluster- on fetal growth: a double- blind randomized controlled trial in rural randomized trial. BMC Pediatr. 2019;19(1):512. Burkina Faso. Am J Clin Nutr. 2008;88(5):1330– 40. 42. Lee ACC, Katz J, Blencowe H, Cousens S, Kozuki N, Vogel JP, et al. 26. MISAME Study Group. Prenatal food supplementation forti- National and regional estimates of term and preterm babies born fied with multiple micronutrients increases birth length: a ran- small for gestational age in 138 low-i ncome and middle-i ncome coun- domized controlled trial in rural Burkina Faso. Am J Clin Nutr. tries in 2010. Lancet Glob Health. 2013;1(1):e26– 36. 2009;90(6):1593–6 00. 43. Zeve D, Regelmann MO, Holzman IR, Rapaport R. Small at birth, 27. Ashorn P, Alho L, Ashorn U, Cheung YB, Dewey KG, Harjunmaa but how small? The definition of SGA revisited. Horm Res Paediatr. U, et al. The impact of lipid- based nutrient supplement provision to 2016;86(5):357–6 0. pregnant women on newborn size in rural Malawi: a randomized 44. Xu H, Simonet F, Luo ZC. Optimal birth weight percentile cut- controlled trial. Am J Clin Nutr. 2015;101(2):387–9 7. offs in defining small- or large- for- gestational- age. Acta Paediatr. 28. Luntamo M, Kulmala T, Mbewe B, Cheung YB, Maleta K, 2010;99(4):550– 5. Ashorn P. Effect of repeated treatment of pregnant women with 45. Katz J, Lee ACC, Kozuki N, Black RE. Mortality risk among term and sulfadoxine-p yrimethamine and azithromycin on preterm deliv- preterm small for gestational age infants. In: Embleton ND, Katz J, ery in Malawi: a randomized controlled trial. Am J Trop Med Hyg. Ziegler EE, editors. Low-b irthweight baby: born too soon or too small 2010;83(6):1212– 20. (Nestlé Nutr Inst Workshop Ser; vol. 81). Basel: S. Karger AG.; Vevey: 29. Fawzi WW, Msamanga GI, Urassa W, Hertzmark E, Petraro P, Willett Nestec Ltd.; 2015. p. 29–3 5. WC, et al. Vitamins and perinatal outcomes among HIV- negative 46. DiTomasso D, Cloud M. Systematic review of expected weight women in Tanzania. N Engl J Med. 2007;356(14):1423– 31. changes after birth for full- term, breastfed newborns. J Obstet 30. Hjort L, Møller SL, Minja D, Msemo O, Nielsen BB, Christensen DL, Gynecol Neonatal Nurs. 2019;48(6):593–6 03. et al. FOETAL for NCD— foetal exposure and epidemiological transi- 47. Mtove G, Abdul O, Kullberg F, Gesase S, Scheike T, Andersen FM, tions: the role of Anaemia in early life for non-c ommunicable diseases et al. Weight change during the first week of life and a new method for in later life: a prospective preconception study in rural Tanzania. BMJ retrospective prediction of birthweight among exclusively breastfed Open. 2019;9(5):e024861. newborns. Acta Obstet Gynecol Scand. 2022;101(3):293–3 02. 31. Schmiegelow C, Scheike T, Oesterholt M, Minja D, Pehrson C, 48. Flaherman VJ, Ginsburg AS, Nankabirwa V, Braima da Sa A, Medel- Magistrado P, et al. Development of a fetal weight chart using serial Herrero A, Schaefer E, et al. Newborn weight change and predictors trans- abdominal ultrasound in an east African population: a longitu- of underweight in the neonatal period in Guinea-B issau, Nepal, dinal observational study. PLoS One. 2012;7(9):e44773. Pakistan and Uganda. Matern Child Nutr. 2022;18(4):e13396. 32. Masanja H, Smith ER, Muhihi A, Briegleb C, Mshamu S, Ruben J, 49. World Health Organization. WHO antenatal care recommendations et al. Effect of neonatal vitamin A supplementation on mortality in for a positive pregnancy experience: maternal and fetal assessment infants in Tanzania (Neovita): a randomised, double-b lind, placebo- update: imaging ultrasound before 24 weeks of pregnancy. Geneva: controlled trial. Lancet. 2015;385(9975):1324–3 2. World Health Organization; 2022. 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 12 | HAZEL et al. 50. Deputy NP, Nguyen PH, Pham H, Nguyen S, Neufeld L, Martorell R, et al. Validity of gestational age estimates by last menstrual period A PPE N DI X 1 and neonatal examination compared to ultrasound in Vietnam. BMC Pregnancy Childbirth. 2017;17(1):25. 51. Macaulay S, Buchmann EJ, Dunger DB, Norris SA. Reliability Subnational Collaborative Group for Vulnerable and validity of last menstrual period for gestational age estima- Newborn Mortality* tion in a low- to- middle- income setting. J Obstet Gynaecol Res. 2019;45(1):217–2 5. 52. Ehrenkranz RA. Estimated fetal weights versus birth weights: Rolf D. W. Klemm, Allan B. Massie, Maithilee Mitra, Sucheta should the reference intrauterine growth curves based on Mehra, Kerry J. Schulze, Alfred Sommer, Md. Barkat Ullah, birth weights be retired? Arch Dis Child Fetal Neonatal Ed. Alain B. Labrique, Mabhubur Rashid, Saijuddin Shaikh, 2007;92(3):F161– 2. Nazma Begum, Nabidul Haque Chowdhury, Md. Shafiqul 53. Grantz KL. Fetal growth curves: is there a universal reference? Obstet Gynecol Clin North Am. 2021;48(2):281– 96. Islam, Rasheda Khanam, Dipak Kumar Mitra, Abdul Quaiyum, Modiegi Diseko, Joseph Makhema, Yue Cheng, Meselech Roro, Bilal Shikur Endris, Charles D. Arnold, Rajiv SU PP ORT I NG I N FOR M AT ION Bahl, Nita Bhandari, Jose Martines, Sarmila Mazumder, Additional supporting information can be found online Lotta Hallamaa, Juha Pyykkö, Willy Urassa, Phillippe in the Supporting Information section at the end of this Deloron, Ib Christian Bygbjerg, Sofie Lykke Moeller, Thor article. Grundtvig Theander, Alfa Muhihi, Ramadhani Abdallah Noor, Moses R. Kamya, Miriam Nakalembe, Godfrey Biemba, Julie M. Herlihy, Reuben K. Mbewe, Fern Mweena, Kojo Yeboah-A ntwi, Andrew Prendergast. How to cite this article: Hazel EA, Erchick DJ, Katz J, Lee ACC, Diaz M, Wu LSF, et al. Neonatal mortality risk of vulnerable newborns: A descriptive analysis of Vulnerable Newborn Measurement Core Group subnational, population-b ased birth cohorts for 238 143 live births in low- and middle- income settings LSHTM: Joy E. Lawn; Hannah Blencowe; Eric Ohuma; Yemi from 2000 to 2017. BJOG. 2023;00:1–12. https://doi. Okwaraji; Judith Yargawa; Ellen Bradley; Lorena Suarez Idueta org/10.1111/1471-0528.17518 *Individuals involved in multiple studies on this list are only named once. 14710528, 0, Downloaded from https://obgyn.onlinelibrary.wiley.com/doi/10.1111/1471-0528.17518 by University of Ghana - Accra, Wiley Online Library on [06/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License