Timing of neonatal mortality and severe morbidity during the postnatal period: a systematic review

Abstract

Objective: 

The objective of this review was to determine the timing of overall and cause-specific neonatal mortality and severe morbidity during the postnatal period (1-28 days).

Introduction: 

Despite significant focus on improving neonatal outcomes, many newborns continue to die or experience adverse health outcomes. While evidence on neonatal mortality and severe morbidity rates and causes are regularly updated, less is known on the specific timing of when they occur in the neonatal period.

Inclusion criteria: 

This review considered studies that reported on neonatal mortality daily in the first week; weekly in the first month; or day 1, days 2-7, and days 8-28. It also considered studies that reported on timing of severe neonatal morbidity. Studies that reported solely on preterm or high-risk infants were excluded, as these infants require specialized care. Due to the available evidence, mixed samples were included (eg, both preterm and full-term infants), reflecting a neonatal population that may include both low-risk and high-risk infants.

Methods: 

MEDLINE, Embase, Web of Science, and CINAHL were searched for published studies on December 20, 2019, and updated on May 10, 2021. Critical appraisal was undertaken by 2 independent reviewers using standardized critical appraisal instruments from JBI. Quantitative data were extracted from included studies independently by 2 reviewers using a study-specific data extraction form. All conflicts were resolved through consensus or discussion with a third reviewer. Where possible, quantitative data were pooled in statistical meta-analysis. Where statistical pooling was not possible, findings were reported narratively.

Results: 

A total of 51 studies from 36 articles reported on relevant outcomes. Of the 48 studies that reported on timing of mortality, there were 6,760,731 live births and 47,551 neonatal deaths with timing known. Of the 34 studies that reported daily deaths in the first week, the highest proportion of deaths occurred on the first day (first 24 hours, 38.8%), followed by day 2 (24-48 hours, 12.3%). Considering weekly mortality within the first month (n = 16 studies), the first week had the highest mortality (71.7%). Based on data from 46 studies, the highest proportion of deaths occurred on day 1 (39.5%), followed closely by days 2-7 (36.8%), with the remainder occurring between days 8 and 28 (23.0%). In terms of causes, birth asphyxia accounted for the highest proportion of deaths on day 1 (68.1%), severe infection between days 2 and 7 (48.1%), and diarrhea between days 8 and 28 (62.7%). Due to heterogeneity, neonatal morbidity data were described narratively. The mean critical appraisal score of all studies was 84% (SD = 16%).

Conclusion: 

Newborns experience high mortality throughout the entire postnatal period, with the highest mortality rate in the first week, particularly on the first day. Ensuring regular high-quality postnatal visits, particularly within the first week after birth, is paramount to reduce neonatal mortality and severe morbidity.

Introduction

Although great strides have been made to improve neonatal outcomes, many newborns worldwide still face severe health outcomes within the first 28 days after birth. In 2019, there were an estimated 2.4 million neonatal deaths globally, suggesting that 6700 newborns died each day.¹ Neonatal death, defined as mortality among live-born infants during the first 28 days of life, can be further categorized as early neonatal deaths (ie, within the first 7 days after birth) and late neonatal deaths (ie, from the 8th to 28th day after birth).² The most common causes of neonatal mortality include sepsis, intrapartum trauma, and prematurity, which constitute nearly 75% of neonatal deaths.³ What remains largely unknown is a global perspective on the timing and causes of neonatal mortality within the first 28 days. Earlier work by Sankar et al.⁴ published in 2016 found that in low- and middle-income countries (LMICs), almost 60% of neonatal deaths occur within the first 3 days of life.

To explore the timing and causes of neonatal mortality globally, this review focuses on neonates who reflect generally healthy newborns. It must be recognized that differences in mortality timing and causes exist based on when a neonate is born. Previous work has established that the risk of mortality is high immediately after birth, particularly when associated with being born too early or too small,^5,6 indicating that there is variation in risk of mortality based on infant gestation at birth. Additionally, given the known risk factors around preterm infants, preterm or small-for-gestational-age infants are often admitted to a neonatal intensive care unit (NICU) to receive specialized care and improve their chances of survival. Therefore, this review focuses on generally healthy newborns (eg, community-based, non-NICU data) in order to shed light on the timing and causes of neonatal mortality from a broader perspective limiting bias, given that preterm infants tend to have higher mortality and morbidity, and present conditions specific to prematurity or low birth weight.

Although neonatal mortality remains an ongoing concern, rates of severe neonatal morbidity have been increasing with a growing concern across high-income countries and LMICs alike.^7,8 Severe neonatal morbidities that may occur in an apparently healthy newborn can include sepsis, acute respiratory infection/pneumonia, or seizures.^9-11 There are long-term repercussions associated with neonatal morbidities, especially related to early developmental outcomes, school performance, and future hospitalization.^10,12 The severity of complications that can occur during the postnatal period as a result of neonatal morbidity warrants further exploration related to postnatal care.

Despite substantial contextual differences between high-income countries and LMICs,¹³ synthesized evidence is needed on the cause and timing of death and severe morbidity among newborns globally within the first 28 days. In line with the Sustainable Development Goals,¹⁴ there have been improvements to neonatal care through enhanced training for health care providers and coverage for health care interventions for women and children (eg, immunizations, trained health care providers at birth, essential newborn care) in LMICs.^15,16 Still, coverage for essential interventions targeting the postnatal period (eg, support of breastfeeding initiation and maintenance, quality postnatal visits for mothers and newborns) is suggested to be insufficient.^13,17 Furthermore, coverage is threatened by disruptions of essential newborn care services due to the COVID-19 pandemic, which has been linked to increased adverse neonatal outcomes.¹⁸ Gaining further insight into when and why neonatal mortality and morbidity occur during the postnatal period will inform policy and recommendations for timely postnatal care.

To address this need, an evidence synthesis was necessary to identify the timing and causes of neonatal mortality and severe morbidity to inform the update of global recommendations related to postnatal care of the mother and newborn. In 2022, the World Health Organization (WHO) updated their 2013 recommendations on postnatal care of the mother and newborn.^19,20 Recommendations state that postnatal care should be provided within the first 24 hours after birth at a health facility or following a home birth. This is to be followed by a minimum of 3 postnatal contacts, with one occurring between 48 and 72 hours, one between days 7 and 14, and one at 6 weeks after birth.^19,20 For the postnatal care guideline update, information from this review was used as part of the evidence to ensure the timing recommendations for postnatal contact are aligned with periods when newborns are experiencing the greatest health challenges.

A preliminary search of PROSPERO, MEDLINE, the Cochrane Database of Systematic Reviews, and the JBI Evidence Synthesis was conducted, and no current or ongoing systematic reviews on the timing of overall or cause-specific newborn mortality and morbidity in the postnatal period were identified. Existing reviews focused on specific aspects, such as neonatal mortality timing in LMICs⁴ or maternal and perinatal mortality using institutional data in LMICs.²¹ Building on Sankar et al.'s⁴ review, which was conducted in 2012 and focused solely on LMICs, the current review includes all countries, with an updated search in 2021. There is variation in neonatal mortality and morbidity risks largely influenced by where a birth occurs (ie, high-, middle-, or low-income country), with the greatest risk known to occur in low-income countries.²² Despite substantial contextual differences across high-income countries and LMICs, the high mortality rates and growing morbidity rates for newborns continue to be a global priority, supporting the broad approach of the current review.

The objective of this review was to determine the timing of overall and cause-specific neonatal mortality and severe morbidity.

Review questions

What is the timing of overall and cause-specific neonatal mortality and severe morbidity in the postnatal period?

In particular:

• When are newborns dying within the first 28 days after birth (overall and cause-specific)?
• When are newborns experiencing severe morbidity within the first 28 days after birth (overall and cause-specific)?

Inclusion criteria

Participants

This review considered reports that included newborns (without known risk factors for complications) from birth to 28 days postnatal, consistent with current WHO definitions.² To be included, studies must have stated that they followed infants up to 28 days regardless of where they were born (ie, at home or at hospital). Studies that reported solely on preterm infants (ie, born before 37 weeks’ gestation) and high-risk infants (eg, malformations, small for gestational age, intrauterine growth restriction, multiples) were excluded from this review. While the original protocol aimed to include only healthy, low-risk neonates, due to the available evidence, mixed samples were included (eg, both preterm and full-term infants) and these are noted in the study characteristics table. Studies that reported data solely on preterm or high-risk infants or from NICUs were excluded because these infants require specialized care, and the timing of deaths and causes are known to vary from those of the general population of all newborns.

Condition

This review sought to locate existing evidence on the timing of overall and cause-specific neonatal mortality and severe neonatal morbidity during the postnatal period. Neonatal death was defined as deaths among live births during the first 28 completed days.² Neonatal morbidity only included severe morbidities identified after birth and before the end of the neonatal period. Causes of mortality and severe morbidity may have originated in the antenatal or intrapartum period but resulted in death or morbidity during the neonatal period. Considering that mixed neonatal populations were included (ie, sample contained both preterm and full-term newborns), the neonatal causes of death included birth asphyxia, congenital anomalies, prematurity, severe infection, diarrhea, and other/not specified. Causes were identified using the International Statistical Classification of Diseases 10^th Revision (ICD-10)^11,23 or as reported by study authors.

Context

This review considered reports that identified neonates born in a health facility or at home. Although the significant burden of newborn mortality occurs in LMICs,²⁴ given that the Sustainable Development Goals focus on development for all countries,²⁵ no limits were placed on country. Due to the potential impact of the COVID-19 pandemic on neonatal mortality and severe morbidity, studies that reported on data collected solely after January 2020 were excluded.

Outcomes

The primary outcomes for this review were timing:

• Of neonatal mortality: overall
• Of neonatal mortality: cause-specific
• And type of severe neonatal morbidity.

Similar to the review by Sankar et al.,⁴ timing of neonatal mortality was considered at different time points. Estimates were calculated:

• Daily within the first week after birth (day 1 through day 7)
• By week within the first month (week 1 through week 4)
• By first day (day 1), days 2–7, and days 8–28 (this is an expansion from Sankar et al.⁴).

First-day mortality was defined as death that occurred within 24 hours of birth, which varied across studies (eg, sometimes described as “first day of life” or “within 24 hours”). Early mortality was typically defined as death that occurred between days 2 and 7. Given the high rate of mortality on the first day and the potential relation to antenatal and intrapartum causes,^4,26,27 early mortality was separately analyzed at 2 time points: first day (day 1) and days 2 and 7. Late mortality was defined as death that occurred between day 8 and up to 28 days after birth.²⁸

Types of studies

This review considered reports that provided prevalence or incidence rates for neonatal mortality and severe morbidity outcomes. This included, but was not limited to, population studies, facility-based studies, empirical studies (non-experimental), and/or civil registration vital statistics and population-based records as available through accessing ministry of health websites of the 193 WHO Member States²⁹ and WHO Mortality Database. Only quantitative studies reporting on prevalence or incidence data were included, excluding qualitative studies and modeling or estimate data (eg, Bayesian modeling, country-level estimates of mortality or morbidity). Relevant systematic reviews were searched to identify any additional original articles not previously captured in the search. Reports that did not define timing based on any of the above outcomes were excluded, such as Demographic and Health Survey data and Child Health Epidemiology Reference Group, which only reported early (days 1 to 7) and late (days 8 to 28) mortality.

Methods

This systematic review was conducted in accordance with JBI methodology for systematic reviews of prevalence and incidence.³⁰ An advisory panel with clinical expertise in the areas of neonatology and obstetrics was established to provide external consultation and guidance to the team throughout all stages of the review. This review was conducted in accordance with an a priori peer-reviewed, published protocol.³¹ Of note, the protocol included both maternal and neonatal outcomes. The maternal outcomes findings are reported separately.³² The methods section describes the approach used for this review, noting any deviations from the protocol. Due to lack of reporting in the included studies, the originally defined secondary outcomes in the protocol³¹ (timing of rehospitalization/readmission by cause and unscheduled use of health services) were not included in this review.

Search strategy

The search strategy aimed to locate both published and unpublished reports. An initial limited search of MEDLINE (Ovid) and CINAHL (EBSCOhost) was undertaken to identify articles on the topic. The text words contained in the titles and abstracts of relevant articles, and the index terms used to describe the articles were used by an experienced information specialist (RP) to develop a full search strategy for MEDLINE through Ovid (see Appendix I). The search strategy, including all identified keywords and index terms, was adapted for each included information source and peer-reviewed by a second information specialist.³³ No language limitations were applied to the searches. In order to capture the broad range of complications and conditions in this review, the search strategy included terms for the outcomes (eg, timing). However, to ensure complete capture, we attempted to identify any articles missed by the searches through the following approaches. Select gray literature sources were searched iteratively to fill gaps identified in the included studies, as were sources identified through stakeholder consultation. Gray literature was manually gathered to attain country-level reports on maternal and newborn health-related outcomes in the postnatal period from health ministry websites and the WHO Mortality Database. A Google Scholar search was conducted between July 2 and 6, 2020, and updated June 9–12, 2021, using each of the WHO Member States²⁹ and (maternal OR neonatal) AND (mortality OR morbidity) to further identify potential sources. The reference lists of all studies selected for critical appraisal were screened for additional studies.

Studies published in English, French, and Spanish were included. All reports published since 2000 that reported data after 2000 were considered for this review. This cut-off was selected to include recent evidence for updating the 2013 WHO Recommendations on Postnatal Care of the Mother and Newborn.¹⁹ Additionally, given the introduction of the Millennium Development Goals in 2000, there was a worldwide shift in measurement of mortality and morbidity, resulting in an increase in data quality and quantity after this period.³⁴ Studies that reported on data before 2000 were excluded. If data were reported separately by year, data older than 2000 were excluded.

The databases searched included MEDLINE ALL (Ovid), CINAHL with Full Text (EBSCOhost), Web of Science Core Collection (Web of Science), and Embase on December 20, 2019. These were updated by rerunning the searches on May 10, 2021. Searches were limited to publications since January 1, 2000. Sources of unpublished studies and gray literature included health ministry country websites and the WHO Mortality Database. Additional articles were identified through previous systematic reviews on maternal and neonatal mortality and morbidity that used data published from 2003 to 2012.^4,35 Due to the searching function limitations of the publicly available database interface, we were unable to complete the search in LILACS (BIREME/PAHO/WHO website) as stated in the protocol.

Study selection

Following the search, all identified citations were collated and uploaded into Covidence (Veritas Health Innovation, Melbourne, Australia)³⁶ and duplicates were removed through the Covidence automated duplicate identification tool. Titles and abstracts and full texts were then screened by 2 independent reviewers (JM, RO, RD, HDS, JSD, BH) for assessment against the inclusion criteria for the review. Reasons for exclusion of full-text studies that did not meet the inclusion criteria were recorded (see Appendix II). Any disagreements between reviewers at each stage of the study selection process were resolved with a third reviewer (JSD, BH, JC, MB) or through discussion. The results of the search are presented in a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram (Figure 1).³⁷

Assessment of methodological quality

Relevant studies were retrieved in full and their citation details were imported into the JBI System for the Unified Management, Assessment and Review of Information (JBI SUMARI; JBI, Adelaide, Australia).³⁸ Eligible studies were critically appraised by 2 independent reviewers (JSD, BR for the studies in English and MB, NR for the studies in French and Spanish) for methodological quality using standardized critical appraisal instruments from JBI, as appropriate.^39,40 Any disagreements that arose were resolved through discussion. The results of the critical appraisal are reported in narrative format and in tables. All studies, regardless of their methodological quality, were included in data extraction and synthesis.

Data extraction

Data were extracted from papers included in the review by at least 2 independent reviewers (BH, JM, RO, RD, HDS, MB, and NR) using a data extraction tool developed by the reviewers (see Appendix III). Data relevant to the review question and study objectives were extracted, including specific details about populations, study methods, and outcomes of interest (ie, overall timing of mortality, morbidity, and causes). The data extraction tool was modified and revised through piloting before full data extraction. Any disagreements between reviewers were resolved through a third reviewer (JSD) for the English studies and through discussion for the French/Spanish studies. Authors of 8 papers were contacted to request additional data for clarification, with 7 responding with the requested information. For the other paper, we were able to use the data as reported in the Sankar et al.⁴ study. For 3 of the included studies,^41-43 exact mortality data were not possible to extract from the original figures; thus, the data reported in Sankar et al.⁴ were used in this review.

Data synthesis

There were deviations from the protocol regarding data synthesis. First, we estimated missing data if sufficient data points were available, and second, STATA (Stata Corp LLC, Texas, USA) was used instead of RevMan 5.3 (Copenhagen: The Nordic Cochrane Centre, Cochrane). Detailed below are the specific steps taken in the revised data synthesis approach.

To be included in the review, studies must have reported data on a minimum of first-day mortality (day 1), days 2-7, and late mortality (days 8-28). For studies that provided incomplete mortality breakdown during the first week, the study must have provided data for at least 3 time points in the first week after birth to be used in pooling (eg, day 1, day 2, and days 3-7; days 1-3, days 4-7, and days 8-28). Studies that provided full data on the weekly breakdown in the first 28 days (ie, week 1, week 2, week 3, and week 4) were also included, but no pooling or extrapolation was done for weekly data. For pooling of the data, we followed the steps below:

Step 1: Studies that had data at each time point were used to obtain the summary estimate (proportion) for each time point.

Step 2: This estimate was used to calculate the proportion for the missing time points in other studies (extrapolation). For studies where missing data were estimated for daily deaths, the proportions per day estimated in step 1 were applied to each day with missing data. If data were reported for some of the days (eg, days 1 and 2, but then combined for days 4-6), the proportion for each day estimated in step 1 was applied to the sum of deaths across that group of days. For example, if a study provided data for days 1, 2, 3, and days 4-7, we split the data from days 4-7 into data for day 4, 5, 6, and 7 based on the proportion for each day obtained from actual data in step 1 for these time points. As another example, if a study provided data for days 1-2, and days 3-7, the data were extrapolated for days 1 and 2, based on the proportions obtained in step 1 for a sum of days 1 and 2. Subsequently, the data for days 3, 4, 5, 6, and 7 would be extrapolated from the days 3-7 sum from the pooled proportions for each day obtained from step 1.

Step 3: The data were pooled again, using the data from studies that had data for all time points, and the extrapolated data for which estimates of daily deaths were made based on steps 1 and 2 of the pooling processes.

Step 4: The pooled results for each day in the first week; each week in the first 4 weeks; and by first day, days 2-7, and days 8-28 are presented as bar graphs with the forest plots in Appendices IV to VI. To determine pooled estimates, analysis took place using the STATA v.14.0 (Stata Corp LLC, Texas, USA) metaprop command for binomial data. Random effects models were run to pool incidence proportion of overall maternal deaths for most analyses. When there were 0 deaths for a specific period (eg, in early mortality), random effects models using the Freeman-Tukey double arcsine transformation to compute the weighted pooled estimate, with a back-transformation on the pooled estimate, were carried out.

The proportional neonatal mortality ratio was defined and calculated as the number of neonatal deaths during a given period over the total number of neonatal deaths known during the postnatal period. For cause-specific analysis, number of deaths due to a specific cause was the numerator and the total number of newborns who died due to that cause in that period was the denominator (eg, number of deaths on day 1 related to infection divided by the total number of newborns who died due to infection).

Subgroup analysis for overall neonatal mortality was conducted based on high-, upper-middle, lower-middle, and low-income countries according to the World Bank.⁴⁴ Although not in the original protocol, an analysis was also conducted to compare studies that reported on data collected in or before 2010 and from 2011 onwards to reflect the changes in neonatal mortality that may have occurred over time. Due to an insufficient number of studies in each category, subgroup analysis on location of birth (facility vs. home) and type of study (population vs. facility-based) was not possible. Where statistical pooling was not possible, the findings are presented in narrative format, including tables and figures to aid in data presentation where appropriate.

Two reports provided data from multiple countries individually,^45,46 so findings are reported separately at the country level, whereas three reports^42,47,48 provided data from multiple countries combined, so findings are reported collectively. All other reports provided data from a single country. Hereafter, results of reports with multiple countries are referred to as “studies,” although data in multiple “studies” may have originated from a single published article.

Results

Study inclusion

Based on the combined search for maternal and neonatal outcomes, 27,673 records were identified through the original search strategy, and 23 records were identified through other methods (eg, Google Scholar, ministry of health websites). After duplicates were removed, 19,927 records were screened using titles and abstracts, after which 18,999 records were excluded. A total of 924 full-text records were retrieved with 894 excluded, leaving 30 records (see Figure 1).³⁷ Of the 23 records identified through website and citation searching, 17 were excluded and 6 included. In total, 36 reports were identified, some of which reported on multiple sites, resulting in a total of 51 unique study sites. Of these 51 studies, 48 reported on neonatal mortality outcomes and 3 reported on severe morbidity outcomes.

Figure 1: Search results and study selection and inclusion process.³⁷

Methodological quality

Articles that met the inclusion criteria were critically appraised for methodological quality as appropriate to their study design (see Tables 1 through 3 for full methodological quality assessment). Six articles were analytical cross-sectional studies, with critical appraisal scores ranging from 57% to 100%.^46,49-53 The greatest concerns were whether study subjects and settings were described sufficiently and whether appropriate statistical analysis was used. The most prominent type of article included was cohort study, with 19 studies having critical appraisal scores ranging from 55% to 100%.^{15,41-43,45,47,54-66} For this study type, the greatest concerns were whether confounding factors were identified and dealt with appropriately, as well as whether they used the appropriate statistical analysis. The remaining 11 articles were prevalence studies, with critical appraisal scores ranging from 56% to 100%.^48,67-76 For prevalence studies, the greatest concern was whether the study subjects and the setting were described in sufficient detail. The mean critical appraisal score was 84% (SD = 16%) and median was 88%.

Table 1 - Critical appraisal of analytical cross-sectional studies

Citation	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	%
Batieha et al. 201549	Y	Y	Y	Y	Y	Y	Y	Y	100
Bogale et al., 201750	Y	Y	Y	Y	Y	Y	Y	Y	100
Fottrell et al., 201546	U	Y	Y	Y	Y	Y	Y	U	75
Ivanova et al., 202051	Y	N	Y	Y	N	N/A	Y	U	57
Limaso et al., 202052	Y	Y	Y	Y	Y	Y	Y	Y	100
Upadhyay et al., 201353	Y	U	Y	Y	Y	Y	U	U	63
%	83	67	100	100	83	100	83	50

Y, yes; N, no; U, unclear; N/A, not applicable

JBI critical appraisal checklist for analytical cross-sectional studies

Q1. Were the criteria for inclusion in the sample clearly defined?

Q2. Were the study subjects and the setting described in detail?

Q3. Was the exposure measured in a valid and reliable way?

Q4. Were objective, standard criteria used for measurement of the condition?

Q5. Were confounding factors identified?

Q6. Were strategies to deal with confounding factors stated?

Q7. Were the outcomes measured in a valid and reliable way?

Q8. Was appropriate statistical analysis used?

Table 2 - Critical appraisal of cohort studies

Citation	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	Q9	Q10	Q11	%
AMANHI, 201845	Y	Y	Y	U	Y	Y	Y	Y	U	U	Y	73
Bapat et al., 201254	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Baqui et al., 200655	Y	U	Y	Y	U	Y	Y	Y	U	U	U	55
Belizan et al., 201242	Y	Y	Y	Y	U	Y	Y	Y	Y	N/A	U	80
Chowdhury et al., 201056	Y	Y	Y	U	Y	Y	Y	Y	U	N/A	Y	80
Diallo et al., 201157	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Edmond et al., 200858	Y	Y	Y	Y	U	Y	Y	Y	Y	N/A	Y	90
Jehan et al., 200959	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Khatun et al., 201241	Y	Y	Y	Y	U	Y	Y	Y	Y	N/A	U	80
Mengesha et al., 201660	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Mersha et al., 201915	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Mullany et al., 200961	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Munjanja, 200762	Y	Y	Y	U	U	Y	Y	Y	U	N/A	U	60
New Zealand MoH, 201263	Y	Y	Y	Y	U	Y	Y	Y	Y	N/A	U	80
Nga et al., 201243	Y	Y	Y	U	U	Y	Y	Y	Y	N/A	U	70
Niswade et al., 201164	Y	Y	Y	Y	Y	Y	Y	Y	Y	N/A	Y	100
Saleem et al., 201447	Y	Y	Y	U	U	Y	Y	Y	Y	N/A	Y	80
Waiswa et al., 201066	Y	Y	Y	U	N	Y	Y	Y	U	Y	Y	73
Welaga et al., 201365	Y	Y	Y	Y	Y	Y	Y	Y	Y	N/A	Y	100
%	100	95	100	68	53	100	100	100	74	78	68

Y, yes; N, no; U, unclear; N/A, not applicable

JBI critical appraisal checklist for cohort studies

Q1. Were the two groups similar and recruited from the same population?

Q2. Were the exposures measured similarly to assign people to both exposed and unexposed groups?

Q3. Was the exposure measured in a valid and reliable way?

Q4. Were confounding factors identified?

Q5. Were strategies to deal with confounding factors stated?

Q6. Were the groups/participants free of the outcome at the start of the study (or at the moment of exposure)?

Q7. Were the outcomes measured in a valid and reliable way?

Q8. Was the follow-up time reported and sufficient to be long enough for outcomes to occur?

Q9. Was follow-up complete, and if not, were the reasons to loss to follow-up described and explored?

Q10. Were strategies to address incomplete follow-up utilized?

Q11. Was appropriate statistical analysis used?

Table 3 - Critical appraisal of studies reporting prevalence

Citation	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	Q9	%
Al-Sheyab et al., 202067	N	Y	Y	Y	Y	Y	Y	Y	Y	89
Auger et al., 201568	Y	Y	Y	N	Y	Y	Y	Y	N/A	88
Guinsburg et al., 202169	Y	Y	Y	N	Y	Y	Y	Y	Y	89
Kulkarni et al., 200770	Y	Y	Y	N	U	Y	Y	U	U	56
Leonard et al., 201971	Y	Y	Y	N	Y	Y	Y	U	N/A	75
Parashar et al., 201772	Y	Y	Y	U	Y	Y	Y	U	Y	78
Puri et al., 202173	Y	U	U	N	U	Y	Y	Y	Y	56
Rasaily, 200874	Y	U	U	U	Y	U	Y	Y	Y	56
Saleem et al., 202048	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
WHO, 202175	Y	Y	Y	Y	Y	Y	Y	Y	Y	100
Yaya et al., 201476	Y	Y	N	Y	Y	Y	Y	Y	Y	89
%	91	82	73	36	82	91	100	73	89

Y, yes; N, no; U, unclear; N/A, not applicable

JBI critical appraisal checklist for studies reporting prevalence data

Q1. Was the sample frame appropriate to address the target population?

Q2. Were study participants sampled in an appropriate way?

Q3. Was the sample size adequate?

Q4. Were the study subjects and the setting described in detail?

Q5. Was the data analysis conducted with sufficient coverage of the identified sample?

Q6. Were valid methods used for the identification of the condition?

Q7. Was the condition measured in a standard, reliable way for all participants?

Q8. Was there appropriate statistical analysis?

Q9. Was the response rate adequate, and if not, was the low response rate managed appropriately?

Characteristics of included studies

Overall, 36 reports were identified that reported on a total of 51 study sites, with 16 studies that collected data in or before 2010, 25 studies that collected data in 2011 or after, and 10 studies that spanned both periods. Most of the studies were population-based (n = 46). In terms of country-level income classification, 11 studies were conducted in low-income countries, 32 from lower-middle-income countries, 5 from upper-middle-income countries, and 3 from high-income countries. When separate data by year were available, only the most recent data were included, which applied to Auger at al.⁶⁸ with only data from 2001–2012 included. Only one study provided data separately on healthy, full-term newborns.⁶⁹ The remaining articles included a mixed neonatal population in their sample (ie, full-term, preterm, and/or multiples). Not all studies reported on each outcome: 46 studies reported data on overall neonatal mortality timing outcomes, 10 studies reported on cause-specific timing outcomes, and 3 studies reported on neonatal morbidity outcomes. Among the studies that reported on mortality, there were 6,760,731 live births and 47,551 neonatal deaths with timing known. Characteristics of studies reporting neonatal mortality are presented in Table 4, and characteristics of studies reporting neonatal morbidity are presented in Table 5.

Table 4 - Characteristics of neonatal mortality studies

Study/country	Methods	Study population	Live births	Neonatal deaths	Summary of data collection	Results	Neonatal inclusion criteria	Limitations/ comments
Al-Sheyab et al.,67 Jordan	Prevalence study; August 2019-January 2020	Population based	10,226	144	National neonatal death surveillance system; health center data	Day 1: 36 (25%) Day 2: 28 (19.4%) Day 3: 23 (16.0%) Day 4: 4 (2.8%) Day 5: 10 (6.9%) Day 6: 8 (5.6%) Day 7: 5 (3.5%) Week 1: 114 (79.2%) Week 2: 23 (16.0%) Week 3: 5 (3.5%) Week 4: 2 (1.4%) Day 1: 36 (25%) Day 2-7: 78 (54.2%) Day 8-28: 30 (20.8%)	Any birth, stillbirth, and neonatal death that occurred within the 5 hospitals and with data entered into the surveillance system.	—
AMANHI45 Bangladesh, DRC, India, Pakistan, Ghana, Kenya, Tanzania, Zambia	Prospective study; July 2012-February 2016	Population based	Bangladesh: 26,295 India (H): 35,000 India (U): 37,813 Pakistan (M): 27,062 Pakistan (K): 17,189 DRC: 6145 Ghana: 23,640 Kenya: 30,992 Tanzania (I): 8128 Tanzania (P): 18,882 Zambia: 25,082	Bangladesh: 995 India (H): 1287 India (U): 1575 Pakistan (M): 1198 Pakistan (K): 803 DRC: 147 Ghana: 681 Kenya: 359 Tanzania (I): 218 Tanzania (P): 259 Zambia: 354	Verbal autopsy	Day 1 Bangladesh: 420 (42.2%) India (H): 36 (41.6%) India (U): 822 (52.2%) Pakistan (M): 497 (41.5%) Pakistan (K): 337 (42.0%) DRC: 80 (54.4%) Ghana: 308 (45.2%) Kenya: 172 (47.9%) Tanzania (I): 74 (33.9%) Tanzania (P): 131 (50.6%) Zambia: 124 (35.0%) Day 2 Bangladesh: 158 (15.9%) India (H): 140 (10.9%)) India (U): 91 (5.8%) Pakistan (M): 147 (12.3%) Pakistan (K): 86 (10.7%) DRC: 15 (10.2%) Ghana: 44 (6.5%) Kenya: 54 (15.0%) Tanzania (I): 39 (17.9%) Tanzania (P): 39 (15.1%) Zambia: 69 (19.5%) Day 3 Bangladesh: 91 (9.1%) India (H): 107 (8.3%) India (U): 111 (7.0%) Pakistan (M): 112 (9.3%) Pakistan (K): 74 (9.2%) DRC: 12 (8.2%) Ghana: 52 (7.6%) Kenya: 30 (8.4%) Tanzania (I): 39 (17.9%) Tanzania (P): 26 (10.0%) Zambia: 33 (9.3%) Day 4 Bangladesh: 41.0 (4.1%) India (H): 60.9 (4.7%) India (U): 99.8 (6.3%) Pakistan (M): 66.1 (5.5%) Pakistan (K): 46.2 (5.8%) DRC: 6.3 (4.3%) Ghana: 42.5 (13.8%) Kenya: 11.5 (3.2%) Tanzania (I): 2.2 (5.6%) Tanzania (P): 10.3 (3.9%) Zambia: 19.6 (5.5%) Day 5 Bangladesh: 28.9 (2.9%) India (H): 43.1 (3.3%) India (U): 70.5 (3.5%) Pakistan (M): 46.7 (3.9%) Pakistan (K): 32.6 (4.1%) DRC: 4.4 (3.0%) Ghana: 30.0 (9.7%) Kenya: 8.1 (2.3%) Tanzania (I): 8.6 (3.9%) Tanzania (P): 7.3 (2.8%) Zambia: 13.8 (3.9%) Day 6 Bangladesh: 20.9 (2.1%) India (H): 31.2 (2.4%) India (U): 50.9 (3.2%) Pakistan (M): 33.8 (2.8%) Pakistan (K): 23.6 (2.9%) DRC: 3.2 (2.2%) Ghana: 21.7 (7.0%) Kenya: 5.9 (1.6%) Tanzania (I): 6.2 (2.8%) Tanzania (P): 5.3 (2.0%) Zambia: 10.0 (2.8%) Day 7 Bangladesh: 20.0 (2.0%) India (H): 29.8 (2.3%) India (U): 48.8 (3.1%) Pakistan (M): 32.3 (2.7%) Pakistan (K): 22.6 (2.8%) DRC: 3.1 (2.1%) Ghana: 20.8 (6.8%) Kenya: 5.6 (1.6%) Tanzania (I): 5.9 (2.7%) Tanzania (P): 5.1 (1.9%) Zambia: 6 (2.7%) Day 1 Bangladesh: 420 (42.2%) India (H): 536 (41.6%) India (U): 822 (52.2%) Pakistan (M): 497 (41.5%) Pakistan (K): 337 (42.0%) DRC: 80 (54.4%) Ghana: 308 (45.2%) Kenya: 172 (47.9%) Tanzania (I): 74 (33.9%) Tanzania (P): 131 (50.6%) Zambia: 124 (35.0%) Day 2-7 Bangladesh: 360 (36.2%) India (H): 412 (32.0%) India (U): 472 (30.0%) Pakistan (M): 438 (36.6%) Pakistan (K): 285 (35.5%) DRC: 44 (29.9%) Ghana: 211 (31.0%) Kenya: 115 (32.0%) Tanzania (I): 111 (50.9%) Tanzania (P): 93 (35.9%) Zambia: 155 (43.8%) Day 8-28 Bangladesh: 215 (21.6%) India (H): 339 (26.3%) India (U): 281 (17.8%) Pakistan (M): 263 (22.0%) Pakistan (K): 181 (22.5%) DRC: 23 (15.6%) Ghana: 162 (23.8%) Kenya: 72 (20.1%) Tanzania (I): 33 (15.1%) Tanzania (P): 35 (13.5%) Zambia: 75 (21.2%	NR	Data extrapolated for first week analysis
Auger et al.,68 Canada	Prevalence study; 1981-2012	Population based	NR	2382	Data from birth and death registries	Day 1: 1227 (51.5%) Day 2-7: 569 (23.9%) Day 8-28: 586 (24.6%)	Live-born infants weighing greater than or equal to 500g	Birth location hospital only; data from after 2000 only
Bapat et al.,54 India	Cohort study; October 2005-September 2007	Population based	11,305	116	Verbal autopsy	Day 1: 40 (34.5%) Day 2: 12 (10.3%) Day 3: 11 (9.5%) Day 4: 8.9 (7.6%) Day 5: 6.3 (5.4%) Day 6: 4.5 (3.9%) Day 7: 4.3 (3.7%) Day 1: 40 (34.5%) Day 2-7: 47 (40.5%) Day 8-28: 29 (25%)	NR	Data extrapolated for first-week analysis
Baqui et al.,55 India	Cohort study, time not specified	Population based	NR	618	Data collected by data collectors Cause of death assigned by WHO verbal autopsy algorithm	Day 1: 197 (31.9%) Day 2: 48 (7.8%) Day 3: 62 (10.0%) Day 4: 46 (7.4%) Day 5: 25 (4.0%) Day 6: 28 (4.5%) Day 7: 31 (5.0%) Day 1: 197 (31.9%) Day 2-7: 240 (38.8%) Day 8-28: 181 (29.3%)	All live births that resulted in deaths on 0-27 postnatal days	Also reported on cause-specific mortality
Batieha et al.,49 Jordan	Analytical cross-sectional study; March 2011-April 2012	Facility based	22,330	327	Health center data and verbal autopsy	Day 1: 137 (41.9%) Day 2-7: 129 (39.4%) Day 8-28: 61 (18.7%)	NR	—
Belizan et al.,42 Argentina, Guatemala, Kenya, Zambia, India, and Pakistan	Cohort study; October 2009- March 2011	Population based	153,728	3882	Collected data from health centers, national data registries, and the Global Network for Women's and Children's Health Research Registry	Day 1: 2019 (52.0%) Day 2: 427 (11.0%) Day 3: 272 (7.0%) Day 4: 194 (5.0%) Day 5: 116 (3.0%) Day 6: 78 (2.0%) Day 7: 78 (2.0%) Week 1: 3184 (82.0%) Week 2: 349 (9.0%) Week 3: 233 (6.0%) Week 4: 116 (3.0%) Day 1: 2019 (52.0%) Day 2-7: 1165 (30.0%) Day 8-28: 698 (18.0%)	All births	Used data from Sankar et al.
Bogale et al.,50 Ethiopia	Analytical cross-sectional study; March 16-24, 2016, previous 18 months	Population based	NR	37	Verbal autopsy; Dabat Health and Demographic Surveillance System	Day 1: 19 (51.4%) Day 2-7: 9 (24.3%) Day 8-28: 9 (24.3%)	All neonatal deaths	Also reported on cause-specific mortality
Chowdhury et al.,56 Bangladesh	Cohort study; 2003-2004	Population based	11,291	365	Verbal autopsy; Health and Demographic Surveillance System	Day 1: 136 (37.3%) Day 2: 57 (15.6%) Day 3: 56 (15.3%) Day 4: 30 (8.2%) Day 5: 8 (2.2%) Day 6: 6 (1.6%) Day 7: 6 (1.6%) Day 1: 136 (37.3%) Day 2-7: 163 (44.7%) Day 8-28: 66 (18.1%)	All live-born infants who died within first 28 days of life
Diallo et al.,57 Burkina Faso	Cohort study; June 2006–May 2007	Population based	864	40	Verbal autopsy; monthly supervisory visits	Day 1: 8 (20%) Day 2-7: 15 (37.5%) Day 8-28: 17 (42.5%)	NR	Study nested within RCT
Edmond et al.,58 Ghana	Cohort study; January 2003-June 2004	Population based	19,621	590	Verbal autopsy	Day 1: 242 (41.0%) Day 2-7: 195 (33.1%) Day 8-28: 153 (25.9%)	NR	Study nested within RCT Also reported on cause-specific mortality
Fottrell et al.,46 Bangladesh, Malawi, India, and Nepal	Analytical cross-sectional study; 2001-2011	Population based	Bangladesh: 42,241 India (E): 8819 Nepal (D): 15,299 Malawi: 22,563 India (S): 10,029 Nepal (M): 6735	Bangladesh: 1324 India (E): 518 Nepal (D): 528 Malawi: 730 India (S): 87 Nepal (M): 204	Verbal autopsy; site-specific surveillance systems	Day 1 Bangladesh: 418 (31.6%) India (E): 176 (34.0%) Nepal (D): 196 (37.1%) Malawi: 293 (40.1%) India (S): 30 (34.5%) Nepal (M): 58 (28.4%) Day 2 Bangladesh: 199 (15.0%) India (E): 64 (12.4%) Nepal (D): 66 (12.5%) Malawi: 120 (16.4%) India (S): 9 (10.3%) Nepal (M): 15 (7.4%) Day 3 Bangladesh: 159 (12.0%) India (E): 50 (9.7%) Nepal (D): 49 (9.3%) Malawi: 63 (8.6%) India (S): 5 (5.7%) Nepal (M): 14 (6.9%) Day 4 Bangladesh: 114 (8.6%) India (E): 21 (4.1%) Nepal (D): 51 (9.7%) Malawi: 50 (6.8%) India (S): 3 (3.4%) Nepal (M): 16 (7.8%) Day 5 Bangladesh: 59 (4.5%) India (E): 28 (5.4%) Nepal (D): 23 (4.4%) Malawi: 24 (3.3%) India (S): 2 (2.3%) Nepal (M): 8 (3.9%) Day 6 Bangladesh: 41 (3.1%) India (E): 19 (3.7%) Nepal (D): 13 (2.5%) Malawi: 25 (3.4%) India (S): 4 (4.6%) Nepal (M): 13 (6.4%) Day 7 Bangladesh: 34 (2.6%) India (E): 10 (1.9%) Nepal (D): 13 (2.5%) Malawi: 22 (3.0%) India (S): 2 (2.3%) Nepal (M): 4 (2.0%) Week 1 Bangladesh: 1024 (77.3%) India (E): 368 (71.0%) Nepal (D): 411 (77.8%) Malawi: 597 (81.8%) India (S): 55 (63.2%) Nepal (M): 128 (62.7%) Week 2 Bangladesh: 148 (11.2%) India (E): 74 (14.3%) Nepal (D): 57 (10.7%) Malawi: 77 (10.5%) India (S): 15 (17.2%) Nepal (M): 32 (15.7%) Week 3 Bangladesh: 85 (6.4%) India (E): 47 (9.1%) Nepal (D): 34 (6.4%) Malawi: 28 (3.8%) India (S): 10 (11.5%) Nepal (M): 25 (12.3%) Week 4 Bangladesh: 67 (5.1%) India (E): 29 (5.6%) Nepal (D): 26 (4.9%) Malawi: 28 (3.8%) India (S): 7 (8.0%) Nepal (M): 19 (9.3%) Day 1 Bangladesh: 418 (31.6%) India (E): 176 (34.0%) Nepal (D): 196 (37.1%) Malawi: 293 (40.1%) India (S): 30 (34.5%) Nepal (M): 58 (28.4%) Day 2-7 Bangladesh: 606 (45.8%) India (E): 192 (37.1%) Nepal (D): 215 (40.7%) Malawi: 304 (41.6%) India (S): 25 (28.7%) Nepal (M): 70 (34.3%) Day 8-28 Bangladesh: 300 (22.7%) India (E): 150 (29.0%) Nepal (D): 117 (22.2%) Malawi: 133 (18.2%) India (S): 32 (36.8%) Nepal (M): 76 (37.3%)	All neonatal deaths and stillbirths in study sites	Secondary analysis Also reported on cause-specific mortality
Guinsburg et al.,69 Brazil	Prevalence study; 2004-2013	Population based	5,285,112	12,589	Data registry, Civil Registry of São Paulo State	Day 1: 3921 (31.1%) Day 2: 1528 (12.1%) Day 3: 994 (7.9%) Day 4: 696 (5.5%) Day 5: 544 (4.3%) Day 6: 379 (3.0%) Day 7: 372 (3.0%) Week 1: 8434 (67.0%) Week 2: 1970 (15.6%) Week 3: 1268 (10.1%) Week 4: 917 (7.2%) Day 1: 3921 (31.1%) Day 2-7: 4513 (35.8%) Day 8-28: 4155 (33.0%)	All infants born, >400g and/or gestational age >22 weeks in São Paulo State to mothers residing in the state in 2004-2013	Data from gestational age 37-41 only
Ivanova et al.,51 Macedonia	Analytical cross-sectional study; 2011-2017	Facility based	36,733	912	Health center data	Day 1: 335 (36.7%) Day 2-7: 373 (40.9%) Day 8-28: 204 (22.4%)	Neonatal death of live-born neonates at the facility in the period of 0-28 days after delivery, with birth weight more than 500g and full 22 gestational weeks on the day of delivery	—
Jehan et al.59 Pakistan	Cohort study; September 2003-August 2005	Population based	1237	53	Health center data; clinician interview	Day 1: 18.1 (34.2%) Day 2-7: 21 (39.6%) Day 8-28: 14 (26.4%)	NR	Data extrapolated for first week analysis and day 0 and days 1-6
Khatun et al.,41 Bangladesh	Cohort study; January 2008-December 2009	Population based	NR	260	Verbal autopsy; house supervisory visits	Day 1: 94 (36.2%) Day 2: 31 (12.0%) Day 3: 23 (8.8%) Day 4: 16 (6.2%) Day 5: 8 (3.1%) Day 6: 10 (3.8%) Day 7: 10 (3.8%) Week 1: 192 (73.8%) Week 2: 39 (15%) Week 3: 17 (6.5%) Week 4: 12 (4.6%) Day 1: 94 (36.2%) Day 2-7: 98 (37.7%) Day 8-28: 68 (26.2%)	All deaths among children <5 years of age who were residents of the slums (based on program identification numbers) in the study area	Used data from Sankar et al.
Kulkarni et al.,70 India	Prevalence study; 2003-2005	Population based	NR	63	Health center data; house supervisory visits	Day 1: 31 (49.2%) Day 2-7: 21 (33.3%) Day 8-28: 11 (17.5%)	All perinatal deaths	—
Limaso et al.,52 Ethiopia	Analytical cross-sectional study January 2018-March 2018	Population based	584	24	Community supervisory	Week 1: 15 (62.5%) Week 2: 5 (20.8%) Week 3: 3 (12.5%) Week 4: 1 (4.2%)	All term pregnancies (≥37 weeks’ gestational age) who live in the study kebeles, neonates followed up for a total of 28 days	—
Mengesha et al.,60 Ethiopia	Cohort study; April-July 2014	Population based	1152	68	Verbal autopsy; community supervisory	Day 1: 15 (22.1%) Day 2-7: 35 (51.5%) Day 8-28: 18 (26.5%)	Neonates of mothers who gave live birth in the study hospitals or admitted within 6 hours	—
Mersha et al.,15 Ethiopia	Cohort study; April 2018-March 2019	Facility based	6769	52	Health center data; verbal autopsy	Day 1: 24 (46.2%) Day 2: 9.6 (18.5%) Day 3: 6.7 (12.9%) Day 4: 4.7 (9.0%) Day 5: 2.1 (4.0%) Day 6: 1.5 (2.9%) Day 7: 1.4 (2.7%) Day 1: 24 (46.2%) Day 2-7: 26 (50%) Day 8-28: 2 (3.8%)	Neonates born at the 6 study hospitals who died within 28 days of life	Data extrapolated for first week analysis
Munjanja,62 Zimbabwe	Cohort study; January-December 2006	Population based	44,242	506	Verbal autopsy; national data; birth and death registries	Day 1: 250 (49.4%) Day 2: 66 (13.0%) Day 3: 31 (6.1%) Day 4: 13 (2.6%) Day 5: 16 (3.2%) Day 6: 17 (3.4%) Day 7: 25 (4.9%) Day 1: 250 (49.4%) Day 2-7: 168 (33.2%) Day 8-28: 88 (17.4%)	All infants of women recruited to the study	—
New Zealand Ministry of Health,63 New Zealand	Cohort study; 2008-2009	Population based	128,618	385	National data	Day 1: 205 (53.2%) Day 2-7: 93 (24.2%) Day 8-28: 87 (22.6%)	All infant with a registered death in 2008-2009	Hospital births only
Nga et al.,43 Vietnam	Cohort study; 2008-2010	Population based	14,453	233	Verbal autopsy	Day 1: 136 (58.4%) Day 2: 15 (6.4%) Day 3: 14 (6.0%) Day 4: 6 (2.6%) Day 5: 4 (1.7%) Day 6: 6 (2.6%) Day 7: 7 (3.0%) Day 1: 136 (58.4%) Day 2-7: 52 (22.3%) Day 8-28: 45 (19.3%)	NR	Used data from Sankar et al. Also reported on cause-specific mortality
Niswade et al.,64 India	Cohort study; November 2006-October 2007	Population based	1103	36	Health center data; verbal autopsy; surveillance data	Day 1: 15 (41.7%) Day 2-7: 11 (30.6%) Day 8-28: 10 (27.8%)	NR	—
Parashar et al.,72 India	Prevalence study; July-September 2015	Population based	NR	24	Verbal autopsy	Day 1: 9 (37.5%) Day 2-7: 8 (33.3%) Day 8-28: 7 (29.2%)	All infants who died during the specified period and the parents or family members were available and gave consent	—
Rasaily,74 India	Prevalence study; January-July 2003	Population based	29,850	1521	Verbal autopsy; community surveys and supervisory visits	Day 1: 597 (39.3%) Day 2: 111 (7.3%) Day 3: 155 (10.2%) Day 4: 94 (6.2%) Day 5: 84 (5.5%) Day 6: 43 (2.8%) Day 7: 43 (2.8%) Week1: 1127 (74.1%) Week 2: 192 (12.6%) Week 3: 155 (10.2%) Week 4: 47 (3.1%) Day 1: 597 (39.3%) Day 2-7: 530 (34.8%) Day 8-28: 394 (25.9%)	All infants born during reference year	—
Saleem et al., 2014,47 Argentina, Guatemala, India, Kenya, Pakistan, Zambia)	Cohort study; 2010-2012	Population based	207,857	5230	Health center data; community supervisory	Day 1: 1804 (34.5%) Day 2: 755 (14.4%) Day 3: 508.2 (9.7%) Day 4: 353.1 (6.8%) Day 5: 249.5 (4.8%) Day 6: 180.5 (3.5%) Day 7: 172.6 (3.3%) Day 1: 1804 (34.5%) Day 2-7: 2219 (42.4%) Day 8-28: 1207 (23.1%)	All infants of women included in study	Data extrapolated for first-week analysis
Saleem et al. 2020,48 Kenya, Zambia, DRC, Pakistan, India, Guatemala	Prevalence study; January 2010-December 2018	Population based	382,635	4884	Maternal Newborn Health Registry (Global Network for Women's and Children's Health Research)	Day 1: 1787.1 (36.6%) Day 2: 586.9 (12.0%) Day 3: 495.1 (10.1%) Day 4: 343.9 (7.0%) Day 5: 243.1 (5.0%) Day 6: 175.8 (4.0%) Day 7: 168.1 (3.4%) Day 1: 1787.1 (36.6%) Day 2-7: 2013 (41.2%) Day 8-28: 1084 (22.2%)	All live-born infants weighing more or equal to 2500g	Data extrapolated for first week analysis
Upadhyay et al.,53 India	Analytical cross-sectional study; 2010	Population based	NR	50	Verbal autopsy	Day 1: 22 (44%) Day 2-7: 16 (32%) Day 8-27: 12 (24%)	All infant deaths	—
Waiswa et al.,66 Uganda	Cohort study; January 2005-December 2008	Population based	NR	64	Verbal autopsy; Health and Demographic Surveillance System	Day 1: 15 (23.4%) Day 2: 15 (23.4%) Day 3: 8 (12.5%) Day 4: 5 (7.8%) Day 5: 6 (9.4%) Day 6: 1 (1.6%) Day 7: 0 (0%) Day 1: 15 (23.4%) Day 2-7: 35 (54.7%) Day 8-28: 14 (21.9%)	Newborns at the time of the study	—
Welaga et al.,65 Ghana	Cohort study; January 2003-December 2009	Population based	17,751	424	Verbal autopsy	Day 1: 119 (28.1%) Day 2: 55 (13.0%) Day 3: 20 (4.7%) Day 4: 29 (6.8%) Day 5: 16 (3.8%) Day 6: 19 (4.5%) Day 7: 17 (4.0%) Week 1: 275 (64.9%) Week 2: 67 (15.8%) Week 3: 51 (12.0%) Week 4: 31 (7.3%) Day 1: 119 (28.1%) Day 2-7: 156 (36.8%) Day 8-28: 149 (35.1%)	NR	—
World Health Organization,75 Macedonia	Prevalence study; January 2019-December 2019	Population based	NR	97	National institutional database systems	Day 1: 31 (32.0%) Day 2-7: 42 (43.3%) Day 8-28: 24 (24.7%)	All stillbirths and neonatal deaths identified during 2019 were entered in a predesigned database	—
Yaya et al.,76 Ethiopia	Prevalence study; January 2006-December 2010	Population based	11,536	308	Community supervisory	Week 1: 143 (46.4%) Week 2: 72 (23.4%) Week 3: 63 (20.5%) Week 4: 30 (9.7%)	All pregnancies	—

DRC, Democratic Republic of Congo; H, Haryana; I, Ifakara; K, Karachi; M, Matiari; NR, not reported; P, Pemba; RCT, randomized controlled trial; U, Uttar Pradesh

Table 5 - Characteristics of studies reporting solely on neonatal morbidity

Study/country	Methods	Study population	Live births	Neonatal deaths	Summary of data collection	Morbidity focus	Neonatal inclusion criteria	Limitations/ comments
Leonard et al.,71 United Kingdom	Prevalence study; January 2010-December 2016	Population based	1,598,069	NR	Health center data	Neonatal group A streptococcus infection	All laboratory-confirmed severe group A streptococcus cases in neonates in London and the Southeast of England with a date of onset within 28 days of birth	—
Mullany et al.,61 Tanzania	Cohort study (secondary analysis of RCT) September 2004-December 2005	Population based	1,653	NR	Community supervisory visits	Umbilical cord infection	All live-born babies born to women enrolled in a cluster-randomized community-based trial	—
Puri et al.73 Democratic Republic of the Congo, Kenya, and Nigeria	Prevalence study; time period NR	Population based	84,759	237	Community supervisory	Infection, signs of infection	All births	Secondary analysis

NR, not reported

Review findings

Overall neonatal mortality

Based on data from 34 studies, the highest proportion of neonatal deaths within the first week occurred on day 1 (first 24 hours), followed by day 2 (see Figure 2). This is consistent across country income levels. The overall proportion of deaths on each day of the first week was consistent in studies in or before 2010 and in or after 2011. No studies from high-income countries reported on daily deaths within the first week.

Figure 2: Proportion of neonatal deaths in the first week after birth, overall and by country income level (n = 34 studies).

Based on data from 16 studies, the highest proportion of deaths was within the first week, followed by the second week (see Figure 3). This was consistent across country income levels, although lower-middle-income countries had a slightly higher proportion of neonatal mortality in the first week, and low-income countries had a higher proportion than other country income levels in the second week. No studies from high-income countries reported on deaths by week. The proportion of deaths by week was similar in studies conducted in or before 2010 and in or after 2011.

Figure 3: Proportion of neonatal deaths by week, overall and by country income level (n = 16 studies).

Based on data from 46 studies, the highest proportion of deaths occurred on day 1 (39.5%), followed closely by days 2-7 (36.8%, see Figure 4). High-income countries had the highest proportion of mortality on day 1 (51.8%) and the lowest proportion on days 2-7 (23.9%), compared to the other country-level income groups. On the other hand, upper-middle-income countries had the lowest proportion of mortality on day 1 (33.7%) and the highest proportion on days 2-7 (41.6%). The proportion of mortality in the late period (days 8-28) is similar regardless of income classification. The overall proportion of neonatal deaths by day 1, days 2-7, and days 8-28 was similar in studies conducted in or before 2010 and in or after 2011.

Figure 4: Proportion of neonatal deaths by day 1, days 2-7, and days 8-28, overall and by country income level (n = 46 studies).

Cause-specific neonatal mortality

Ten studies reported on cause-specific timing of neonatal mortality.^{43,46,50,55,58} Of the total 48,606 neonatal deaths identified, causes were available for 4419 deaths (9.1%). Timing of neonatal mortality was grouped by i) birth asphyxia, ii) congenital anomalies, iii) prematurity, iv) severe infection, v) diarrhea, and vi) other/not specified. As stated previously, these deaths occurred in the neonatal period, but the onset of the cause leading to death may have been during the antenatal/intrapartum period.

As seen in Figure 5, neonatal mortality varied widely in the total number of deaths per cause, ranging from 17 deaths from diarrhea to 1562 deaths from severe infection. For the first analysis, we considered the number of deaths by cause and by time over the total number of cause-specific deaths stratified by cause (eg, number of day 1 deaths due to prematurity/total deaths by prematurity). The causes with the highest proportion of deaths on day 1 were birth asphyxia (68.1%) followed by congenital anomalies (58.2%). Between days 2 and 7, causes with the highest proportion were severe infection (48.1%) and prematurity (33.7%). Between days 8 and 28, causes with the highest proportion were diarrhea (62.7%) and severe infection (46.2%). Due to the low number of studies in each outcome, no subgroup analysis was possible. When considering the number of deaths by cause over all cause-specific deaths at each time point (Figure 6), it is clear that birth asphyxia is the most common cause of death on day 1 (52.0%) with severe infection the most common cause of death on days 2-7 (44.4%) and days 8-28 (64.2%).

Figure 5: Proportion of neonatal deaths on day 1, days 2-7, and days 8-28 by causes: (A) birth asphyxia (n = 10 studies, 1326 deaths); (B) congenital anomalies (n = 9 studies, 157 deaths); (C) prematurity (n = 10 studies, 968 deaths); (D) severe infection (n =10 studies, 1562 deaths); (E) diarrhea (n = 4 studies, 17 deaths); (F) other/not specified (n = 10 studies, 389 deaths)

Figure 6: Cumulative proportion of neonatal mortality on day 1, days 2-7, and days 8-28 by causes at each time point.

Neonatal morbidity

Three studies report on neonatal morbidity outcomes. Because of the difference in outcomes, this is reported narratively. Leonard et al.⁷¹ reported on infants with severe group A streptococcal infections born in London and the Southeast of England between 2010 and 2016. For infants diagnosed with severe group A streptococcal disease, the median onset time was 12 days after delivery (IQR of 7-15 days).⁷¹ Multiple clinical presentations were noted in 50% of the sample: sepsis (41%), pyrexia, (29%), respiratory distress (12%), infections of the umbilicus or ear (12%), or hypoxic ischemic encephalopathy (6%).⁷¹

Mullany et al.⁶¹ reported on umbilical cord infection in Zanzibar, Tanzania, as part of a secondary analysis of a randomized controlled trial between September 2004 and December 2005. The authors developed 6 sign-based definitions from 4 possible signs of cord infection (pus, redness, swelling, and foul odor), with infection status defined as one or more of the possible signs, with a range of severity (none, mild, moderate, severe).⁶¹ Among the 1653 infants in the study, the mean onset ranged from 3.0 days (± 2.1 days) for moderate/severe redness to 4.2 days (± 1.8 days) for pus and foul odor alone. Additional signs had a mean onset of 3.2 days (± 1.6) for presence of moderate/severe redness and pus or foul odor, 3.3 days (± 1.8) for moderate/severe redness with pus, 3.5 days (± 1.5) for any redness with pus and foul odor, and 3.7 days (± 1.8) for any redness with pus. Of note, less than 5% of assessments for pus, redness, swelling, and foul odor were positive in the first 48 hours after birth, with 90% of occurring by day 7.⁶¹

Puri et al.⁷³ reported on serious bacterial infections across 3 countries (Democratic Republic of the Congo, Kenya, and Nigeria) as part of a secondary analysis of the African Neonatal Sepsis Trial. Local infections were identified, including umbilical, skin, eye, and mixed infections, with day 7 (14 per 1000 infants) and day 14 (12 per 1000 infants) being the highest frequency, and day 28 (3 per 1000 infants) and day 1 (4 per 1000 infants) being the lowest frequency. They also reported on signs of systemic infection, with fast breathing being the highest on day 3 (17 per 1000 infants) and day 7 (16 per 1000 infants), and high body temperature being the highest on day 3 (10 per 1000 infants) and day 1 (9 per 1000 infants). All other signs of systemic infections (eg, severe chest indrawing, stopped feeding well, no movement, multiple signs) were less common, fewer than 8 per 1000 at each time point.⁷³

Discussion

This review provides insight into the timing of overall and cause-specific neonatal mortality daily within the first week; weekly in the first month; and comparing day 1, days 2-7, and days 8-28. Across all analyses, the first day after birth (day 1) had the highest number of neonatal deaths. High-income countries had the highest proportion of mortality on day 1 and the lowest proportion on days 2-7, whereas upper-middle-income countries had the lowest proportion of mortality on day 1 and the highest proportion on days 2-7. Similar proportions of mortality on day 1 and days 2-7 were found in lower-middle- and low-income countries. In terms of causes, birth asphyxia, prematurity, and congenital anomalies were the predominant causes of neonatal mortality on day 1, which is expected given their association with intrapartum events. Severe infection is the leading cause of mortality for days 2-7 and days 8-28. Due to heterogeneity, no specific conclusions could be drawn relating to severe neonatal morbidity.

Overall neonatal mortality

Looking at neonatal mortality, most newborns died on the first day, with over one-third of infants who died in the neonatal period dying on the first day, followed by 12.3% on day 2 and 9.0% on day 3. Looking at early mortality within the first week, 39.9% of infants died on day 1 and 36.4% died between days 2 and 7, which means that more than 75% of infants who died did so in the first week. When looking at country income level differences, high-income countries had a higher proportion of mortality on day 1 (51.8%) and a lower proportion on days 2-7 (23.9%), compared to the other country income level classifications, yet this difference was not apparent between days 8 and 28, which had a relatively low proportion of deaths on average (23.1%). Overall, the findings are in line with Sankar et al.'s review⁴ as well as Oza et al.⁷⁷ who also found that the first day and the first week were the most significant in terms of risk of neonatal mortality. Interestingly, Oza et al.⁷⁷ found that developed regions (similar to high-income countries in our study) had a lower risk of death on day 1, whereas our review found that high-income countries had a greater proportion on deaths on day 1 versus other country income level groups. This may be related to the differences in data collection, analysis approach, and the limited number of high-income countries included in this review.

It is also interesting to note that there were some variations in terms of timing of neonatal mortality rates within countries as well as between countries. Additionally, due to the breadth of timing of the included studies (2000–2020), there were variations in neonatal mortality rates, although we did not find any significant variation when comparing data collected in or before 2010 or in or after 2011. However, a systematic review exploring the trends on neonatal mortality between 1990 and 2017 found that neonatal mortality rates have been declining since 1990, with the greatest reduction between 2000–2017 compared with 1990–2000.²² Therefore, it is important to reflect on differences in findings, which could be related to the inclusion criteria of neonates (eg, community sample containing high-risk and low-risk neonates, low-risk hospital births) or differences in data collection related to follow-up or record-keeping of timing of neonatal mortality. Furthermore, there has been a global increase in the focus on improving neonatal outcomes since the establishment of the Millennium Development Goals in 2000 and the subsequent Sustainable Development Goals in 2015.^14,78

Nevertheless, the first few days and the first week are a critical time for neonatal survival, which makes access to high-quality perinatal health care, education, and counseling essential. Greater emphasis and access to care and support may be required for the whole early neonatal period, beyond just the first day. Early community-based interventions—defined as a multiple-intervention approach offered through a variety of strategies, such as community support groups; women's groups; timely, quality antenatal care; skilled birth attendants; community mobilization; home visitation; and training community health workers—have been found to be effective at reducing neonatal mortality and severe morbidity.⁷⁹ Ensuring access to timely, evidence-based interventions may be an important way to reduce neonatal mortality in the postnatal period.

Neonatal cause-specific mortality

Looking at the causes of neonatal mortality, most newborns died due to infection after the first day, with deaths due to birth asphyxia, prematurity, and congenital malformations occurring predominantly on day 1. These findings are similar to others showing that the risk of sepsis increased over time in the first week after birth.⁴ Oza et al.⁸⁰ similarly found that preterm birth and intrapartum complications accounted for most early neonatal deaths (days 1-7), whereas infections caused nearly one-half of neonatal deaths occurring between days 8 and 28. The finding that infection is of greatest risk to newborns who survive the first day is important to note in targeting areas for improved care across the whole neonatal period. Bottlenecks related to neonatal infection include the following: a shortage of health workforce personnel with adequate knowledge, a lack of access to quality antenatal care and assessment, challenges with access to antibiotics, as well as delays in receiving care, either through delayed access due to cultural norms (ie, laying in practices) or challenges being transferred to higher-level facilities.⁸¹

Additionally, Sankar and colleagues⁴ note that while the proportion of deaths due to birth asphyxia, prematurity, and congenital malformations declined after the first day, there were still a number of deaths that occurred up to the first week, suggesting that ongoing monitoring and regular assessment of newborns by a skilled health care provider is needed in the week after birth. In this review, we excluded studies that reported solely on preterm infants (ie, born before 37 weeks’ gestation) and high-risk infants (eg, malformations, small for gestational age, intrauterine growth restriction, multiples). Almost all of our samples were mixed, generally reflecting population-level estimates, suggesting that even among the community population, it is important to consider birth asphyxia, prematurity, and congenital malformations as key outcomes during the postnatal period, as there is opportunity to improve outcomes for newborns in the postnatal period if adequate, quality care is available.

It is important to acknowledge that causes of neonatal mortality may vary both between and within countries, and only 9% of total neonatal deaths could be attributed to a cause in our analysis. Therefore, our findings related to the timing of cause-specific neonatal mortality must be interpreted with caution. For instance, in an analysis of neonatal mortality trends in India, it was found that not only did the rate of neonatal mortality vary between districts, but the causes of death varied between districts.⁸² Furthermore, because our sample included both high- and low-risk neonates, this may have influenced our findings related to the causes of neonatal mortality, reflecting a mixed-sample population. Future work is needed to further understand the causes of neonatal mortality within the first 28 days to explore both within and between country differences in order to develop appropriate interventions.

Neonatal morbidity

Studies reporting on severe morbidity were heterogeneous, which did not allow for pooling of data, and thus made it difficult to identify any trends or patterns in the results. It does appear that onset of severe neonatal morbidity occurs early in the neonatal period, with 63% of infants with severe infection diagnosed between day 3⁶¹ and day 7.⁷³ Given the common early onset, it is important that newborns have access to health care providers for early detection and management of infections across the postnatal period. Enhanced discharge counseling and education to ensure families are advised on preventive measures, such as clean cord care, as well as danger signs and symptoms of infection, such as changes in respiratory rates, may be a benefit. However, because the data on morbidity are scarce and heterogeneous, limited conclusions can be drawn about the timing of morbidity. Further work is needed to expand on the knowledge around timing of onset of neonatal morbidity before specific conclusions can be drawn.

Limitations

While this study has several strengths, there are some limitations that must be acknowledged. First, there are several countries that had data from more than one study from the same geographical area, which may have influenced the incidence rate. However, data were collected from different studies at different time points and with different purposes, and thus it was important to include all studies to provide a broad overview of timing of mortality and morbidity.

Another limitation was the inability to focus on newborns born at low risk of complications, as most studies included a mixed sample that included both preterm and full-term infants and/or multiples. Part of this challenge is that many infants born in LMICs are considered small for gestational age or are born preterm, particularly in countries with high representation in our data (ie, India, Pakistan, and Bangladesh).^83,84 Despite the mixed population in our sample, the findings provide insight into the timing of neonatal mortality and severe morbidity around the world in a community-based sample.

A limitation of the cause-specific mortality data is that causes were not always linked with timing, and only 9.1% of all neonatal deaths reported in the included studies were able to be included in the cause-specific analysis. Studies would often report on the causes of mortality but in reference to the whole neonatal period, without the breakdown by day 1, days 2-7, and days 8-28. Additionally, not all studies reported on causes of death when reporting timing of death. This could partly be due to the challenges in determining causes of death in low-resource areas where autopsies may not be available or feasible.⁸⁵ Furthermore, we reported on the causes of death as reported in studies; however, the number of deaths due to these conditions may only be an estimate, as an infant could have more than one condition, with one cause of death assigned as the primary.

Not all studies reported on each time point, requiring extrapolation, which may not accurately reflect the actual deaths at each time point. Due to limitations in data collection, it is possible that first-day deaths were misclassified as fresh stillbirths and vice versa, as previous studies have acknowledged the challenges of correctly classifying early neonatal mortality in LMICs.^86,87 There is a need for improved data systems for newborns to ensure data accuracy, including neonatal mortality audits/reviews.

Additionally, while there is potential for facility-based factors to influence the neonatal mortality rates, all six facility-based studies followed newborns up to 28 days, similar to all population-based studies. Three of the studies were in countries where most of the births occur in health facilities,^49,51,67 suggesting good coverage of births that would occur in the population. However, due to the limited number of studies and how data were reported, we were unable to compare whether birth in a facility or at home impacted neonatal morbidity or mortality. This is an important consideration because newborns born at home may face more complications, and timing of events may not be as clear or well-documented.

It is also important to consider the confounding factor of country income level classification that may influence the level of risk for neonatal mortality when interpreting these findings, such as proximity to health centers, access to primary health care for postnatal assessment, and postnatal advice on neonatal danger signs. Another potential confounding factor is the influence of time on mortality outcomes, as the number of neonatal deaths has been declining over time. However, due to challenges with a lack of data on high-income countries after 2010 and limited country representation of lower-middle-income countries, further analysis was limited.

Conclusions

This is the first review examining the timing of neonatal mortality and morbidity, both overall and cause-specific, comparing across country income level classification, including high-income countries. Newborns are most likely to die during the first day and the first week, with less than one-fourth of all newborns dying between days 8 and 28. On the first day, the most common causes of death were birth asphyxia, prematurity, and congenital malformations, with infection being the most common cause after day 1. It is important to increase focus on improving access to care throughout the entire postnatal period as an essential way to improve neonatal outcomes and achieve the Sustainable Development Goals of reducing the neonatal mortality rate.¹⁴

Recommendations for practice

It is important that newborns continue to receive health care by a skilled health care provider within the first 24 hours after birth, regardless of whether they are born at a health facility or at home (Grade A recommendation⁸⁸). Ready access to quality health care providers who are trained in essential newborn care during this period could potentially decrease the number of deaths that occur on the first day. Early interventions have been found to reduce mortality,⁸⁹ including skin-to-skin contact,^90,91 exclusive breastfeeding,^92,93 and education and training of health care providers.²⁶ Furthermore, the provision of quality antenatal care and assessment, as well as having the birth assisted by a skilled attendant, are known to improve both maternal and neonatal outcomes.^94-97 Therefore, given the important burden that mortality and morbidity represent across the first 28 days for newborns, including the late neonatal period, there is a need to ensure continuity in access to and use of postnatal care services, including through effective service delivery models such as midwife-led continuity of care models.⁹⁸

A second recommendation for practice is to continue with regular postnatal follow-up visits within the postnatal period as recommended by the WHO¹⁹ (Grade A recommendation⁸⁸). Because the incidence of mortality between days 2 and 7 is almost as high as first day mortality, it is important that newborns continue to have access to care during this period. Additionally, given the continued risk of mortality during the late neonatal period, consideration of continuing care during this period is essential.

In addition to the number of postnatal follow-up visits, health care providers should be well-trained, knowledgeable, and have access to life-saving equipment and medication to provide a high quality of care across the high-risk time points.^81,99 In a systematic review and meta-analysis conducted by Langlois and colleagues,¹³ they reported that postnatal care services in LMICs are inequitable and vary depending on socioeconomic status and between urban and rural locations. Thus, investment must be made to strengthen the global health care system with an increased focus on the postnatal period, enhancing not only the quantity of postnatal care follow-up visits, but also the quality in order to deliver high-quality care for newborns, as well as identify newborns left behind.

Recommendations for research

The reporting period on neonatal mortality should be standardized. For instance, there was a range of definitions of “first day” throughout the studies, from first 24 hours, day 0, and day 1. For future studies and increased ease of meta-analysis, first day mortality should be defined as death that occurs within the first 24 hours. Additionally, reporting early mortality separately from first-day mortality is recommended. Given the high number of deaths that occur on the first day, it is important to tease apart improvements in reducing early neonatal death from the first week and the first day, as some first day mortality cases may be related to antenatal or intrapartum events and outcomes. Another recommendation is to report, in the full study or in a supplementary file, the number of deaths that occur on each day of the neonatal period. This will again allow further insight into when newborns die and when follow-up visits should be scheduled to detect danger signs and provide timely treatment to minimize the number of deaths in the neonatal period.

Further research is also needed on the causes of mortality and severe morbidity linked to timing of death and onset. This area was considerably weaker than the overall timing of mortality data, making recommendations for focused care difficult. Future research is needed to know why newborns are dying in order to provide targeted care and education on those issues.

Additionally, consideration of assessment of place of birth and/or presence of a skilled attendant at birth are recommended. Due to an insufficient number of studies in each category, subgroup analysis on location of birth (facility/home) and type of study (population vs. facility-based) were not possible despite our intention. Therefore, further studies should consider comparing timing of neonatal mortality based on these outcomes as well.

A final recommendation is for more studies on timing of mortality conducted in high-income countries and countries not typically represented in published literature (eg, Middle East, Asia and Pacific). There were few studies that reported on data from countries in these areas, while much work has been conducted in exploring mortality in LMICs. Although data from LMICs help provide a picture of the timing of neonatal mortality and morbidity, many other countries are not represented in this meta-analysis.

Funding

This project was supported with funding in part by the Canadian Institutes of Health Research (CIHR) under the Strategy for Patient-Oriented Research (SPOR) initiative through the SPOR Evidence Alliance and the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.