Skip to main content
NCBI home page
Springer logoLink to Springer
. 2025 Sep 22;184(10):628. doi: 10.1007/s00431-025-06467-0

Causes of neonatal mortality in the European Economic Area and Switzerland: a WHO-based analysis and systematic review

Elizabeth E la Cour 1, Veronica N E Malange 1, Tasnim Mohaissen 1,2, Michael Christiansen 1,2,3, Jesper Sune Brok 4, Christina Gade 5,6, Ulrik Lausten-Thomsen 1,7, Paula L Hedley 1,4,✉
PMCID: PMC12450820  PMID: 40976833

Abstract

This study aimed to identify reported causes of neonatal death across the European Economic Area (EEA) member states and Switzerland as of 1 January 2010 (hereafter referred to as the European Region) and, where possible, examine the cause-specific distribution of neonatal deaths over time. We conducted a two-pronged analysis: (1) database analysis using WHO public datasets (2000–2021) for 28 countries in the European Region and (2) a systematic review and meta-analysis of pertinent records from PubMed, Embase, Web of Science, Scopus, Cochrane, and Google Scholar, which were comprehensively reviewed against inclusion criteria up to 10 August 2024, following PRISMA guidelines. The average neonatal mortality rate (NMRs) between 2000 and 2021 in WHO data was 2.63 per 1000 live births, with a significant decline (−0.074 per year) across the region. Prematurity (41.2%) and congenital anomalies (28.9%) were the most common registered causes of neonatal death; however, WHO data lacked etiological detail to reliably analyze cause-specific trends. The systematic review identified 41 eligible studies, of which 15 were included in meta-analyses. Pooled estimates showed that congenital anomalies and prematurity accounted for 30% (95% CI 17–46) and 31% (95% CI 18–49) of deaths, respectively. Among extremely preterm neonates, infections, respiratory, and cardiovascular disorders were the most common reported causes of death.

     Conclusion: NMRs across the European Region are declining, with prematurity and congenital anomalies being leading causes of neonatal death. Current reporting frameworks lack granularity, and a wider adoption of standardized classification systems is critical to improving surveillance and data comparability, especially for preterm infants.

What is Known:

• Neonatal mortality rates have declined across Europe over recent decades.

• There are considerable regional disparities in both causes and rates of neonatal death.

What is New:

• Prematurity and congenital anomalies each account for ~30% of neonatal deaths.

• Greater use of standardized classifications of neonatal death causes is urgently needed.

Open in a new tab

Supplementary Information

The online version contains supplementary material available at 10.1007/s00431-025-06467-0.

Keywords: Neonatal mortality, Europe, Prematurity, Congenital anomalies, ICD-PM, Cause of death

Introduction

The neonatal period accounts for the highest proportion of pediatric deaths worldwide [1, 2]. Consequently, reducing neonatal mortality is a global public health priority and an integral target under the United Nations’ Sustainable Development Goals (SDGs), adopted in 2015 [3]. Through focused global and regional efforts, the neonatal mortality rate (NMR) has declined substantially from 37 deaths per 1,000 live births in 1990 to 17 in 2023, globally [1]. In Europe, this this decline has been even greater, with NMR decreasing from 8 to 2 deaths per 1,000 live births over the same period [1].

Significant regional disparities persist in both NMRs and causes of death. Important causes include prematurity, intrapartum complications, congenital anomalies, and infections [1, 2]. Given that patterns of neonatal mortality vary depending on health system capacity, socioeconomic conditions, and clinical practices, it is essential to understand these patterns within the European context.

Broader regional analyses allow for improved insight into the causes of neonatal deaths, providing sufficient statistical power to detect patterns that would not be visible in smaller single country datasets, particularly when perinatal care is relatively uniform. However, the interpretation of neonatal mortality data is often hindered by non-specific cause-of-death classifications. For instance, “prematurity” is frequently recorded as a cause of death, despite being a risk factor rather than a definitive pathological mechanism [4].

To move beyond broad classifications and enable effective interventions, greater specificity in the categorization of neonatal deaths is essential. We aim to address that gap by systematically evaluating existing literature and publicly available databases to identify, classify, and synthesize the reported causes of neonatal death in the European region. A particular focus is placed on deaths occurring among extremely preterm infants, who represent a disproportionately vulnerable subgroup.

The primary objective of this systematic review is to improve understanding of the causes of neonatal death in the European Region by summarizing reports and databases on NMRs and causes.

Materials and methods

Study design and registration

This study comprises two components: 1) a database analysis using publicly available data from World Health Organization (WHO) to examine causes of death and NMRs from 2000 to 2021. 2) a systematic review of literature encompassing studies published between 1st January 2010 and 10th August 2024 was conducted in accordance with the PRISMA 2020 guidelines [5]. The review protocol was developed following PRISMA-P guidance and is registered with PROSPERO (CRD42024571819).

Study region

Analyses were restricted to Switzerland and all European Economic Area (EEA) member states as of 1 January 2010; hereafter referred to as the European Region.

Objectives and outcomes

The primary objective was to identify specific reported causes of neonatal death in light of country-specific changes in NMRs. A secondary objective was to examine the causes of neonatal death among extremely preterm neonates (< 28 weeks of gestation).

Database analysis outcomes

The primary outcome was the cause-specific distribution of neonatal deaths evaluated using WHO classification standards. Additionally, the NMR per 1,000 live births, analyzed by year and country.

Systematic review outcomes

The main outcome was the proportion of neonatal deaths attributed to specific causes. These were reported separately for studies which included neonates of all gestational ages, and a secondary subset of studies that focused on extremely preterm neonates (defined above). Studies that examined other gestational age categories (i.e. term, very preterm, or moderate to late preterm) were excluded from further analysis.

Database analysis

Data sources

The database analysis utilized WHO indicators for neonatal mortality [6], causes of death [7], and live birth statistics from United Nations Economic Commission for Europe (UNECE) [8]. It should be noted that WHO neonatal mortality rates are reported as partially modelled estimates [9] in order to ensure consistency and comparability across countries, however, this may also smooth over local fluctuations.

Statistical analysis

Descriptive statistics were used to summarize cause-specific mortality distributions and NMR trends. Linear regression models were fitted separately for each country, with NMR as the dependent variable and calendar year as an independent variable.

To identify regional patterns in cause-specific mortality, a two-step cluster analysis was conducted. First, hierarchical clustering using Ward’s minimum variance method was applied to standardized cause-specific mortality proportions and their corresponding linear trends across the study period. These variables were used to construct a distance matrix. The resulting dendrogram informed the optimal number of clusters. Next, K-means clustering was performed using the selected number of clusters to assign countries into distinct mortality profile groups. Thus, cluster definitions incorporated both the relative burden of specific causes of neonatal death and their temporal trends over the two-decade study period.

Statistical significance was evaluated using p-values, with values < 0.05 considered statistically significant. All analyses were conducted in R version 4.4.2 [10], primarily using the packages “dplyr” and “tidyr” for data manipulation, “ggplot2” for data visualization, and “stats” and “cluster” for cluster analysis.

Systematic review and meta-analysis

Search strategy

A literature search was conducted on August 10, 2024, across PubMed, Embase, Web of Science, Google Scholar, Scopus, and the Cochrane Library (including CENTRAL). Prepublication servers (medRxiv, bioRxiv) were also searched. Additional references were identified through manual screening of bibliographies. Language translation, when required, was conducted using DocTranslator [11], and supported by language consultants as needed. A full search strategy is included in Supplementary Material.

Eligibility criteria

Eligible studies included cohort, case–control, cross-sectional, or interventional designs that reported on causes of neonatal death. Neonates were defined as per WHO criteria as infants under 28 days of age [12]; extremely preterm neonates were defined as those born before 28 weeks of gestation [13]. We defined neonatal death among preterm infants to be death within 28 days of the expected date of delivery. This definition was adopted to allow neonatal intensive care unit (NICU)-based death reports to serve as a proxy for neonatal deaths among preterm infants.

The review was restricted to causes of neonatal death within the European Region. To ensure relevance to contemporary neonatal care, we excluded studies published before 2010, and studies reporting on populations born prior to 1984.We included both peer-reviewed studies and preprints. However, grey literature sources such as governmental or national health authority reports were not systematically searched.

Data management and selection process

Search results were managed in Rayyan [14]. After de-duplication, two reviewers (EELC and VNEM) independently screened titles and abstracts against predefined eligibility criteria. Articles were classified as “included”, “excluded”, or “maybe”. Full-text reviews were performed for all “included” and “maybe” articles. Disagreements were resolved through discussion or arbitration by a third reviewer (PLH). The selection process is outlined in the PRISMA flow diagram (Supplementary Fig. 1) [5].

Quality assessment

The Newcastle–Ottawa Scale (NOS) [15] was used by two reviewers (EELC and PLH) to independently assess study quality. Discrepancies in scoring were resolved through consensus.

Data extraction

Two reviewers (EELC and PLH) independently extracted data from included studies, recording the author, publication year, study design, country, sample size, relevant findings, and calculated NOS scores for each included study. Studies reporting from overlapping datasets were cross-checked. Where overlap was identified, we retained studies if they addressed distinct populations (e.g., extremely preterm vs. all neonates) or contributed unique geographical information.

Carty et al. (2023) and Costeloe et al. (2012) presented data from overlapping time periods but Carty et al. examined all neonatal deaths [16] while Costeloe et al. examined extremely preterm only [17] and were consequently analyzed separately.

More complex overlaps occurred among Smith et al. (2010), Carty et al. (2023), and Tambe et al. (2015) [16, 18, 19]. Smith et al. and Carty et al. presented data from national English cohorts with partial overlap in 2005–2007 [16, 18], Tambe et al. reported UK-wide data without country disaggregation, overlapping with Smith et al. and Carty et al. in 2006–2008 [19]. As it was not possible to eliminate these duplications, we retained all three studies for their unique contributions, while acknowledging that this results in some double-counting.

Allanson et al. (2016) reported county-level data [4] nested within the national cohorts of Smith et al. and Carty et al., it was excluded from meta-analyses to avoid duplication.

Data synthesis and classification

Causes of neonatal death were grouped into ICD-10 categories [20]: Perinatal conditions (P00–P96), Congenital malformations (Q00–Q99), and Other/unspecified (i.e. any cause not in chapters P or Q). Meta-analyses were performed only for categories P and Q. The “Other or unspecified” group was excluded due to heterogeneity by default.

In studies with sufficient diagnostic detail, causes of neonatal death were further stratified using the WHO application of ICD-10 to deaths during the perinatal period (ICD-PM) framework to enhance classification specificity [21] (Supplementary Table 1). N3 (birth trauma) and N4 (intrapartum complications) were merged, as these categories were frequently reported together. N11 (unspecified causes) was excluded due to heterogeneity by default.

Statistical analysis

Heterogeneity was quantified using I2 and Tau statistics [22]. An I2 value > 50% was considered indicative of substantial heterogeneity and > 75% as considerable. Sensitivity analyses were performed, post-hoc, by iteratively excluding individual studies to assess the robustness of the pooled estimates.

Proportionate mortality for each cause of death was pooled across studies using a random-effects model due to high levels of heterogeneity. Meta-analyses were limited to studies reporting ICD-10 categories or ICD-PM categories with sufficient data. Categories with less than four studies or fewer than four deaths were excluded.

To explore publication bias, funnel plots were constructed for each cause-specific meta-analysis. Visual inspection of the funnel plots provided qualitative assessment of bias.

Statistical significance was determined using p-values, with values < 0.05 considered statistically significant. All analyses were conducted in R version 4.4.2 [10], primarily using the packages “dplyr” and “tidyr” for data manipulation, “ggplot2” for data visualisation, and “metafor” and “meta” for meta-analyses.

Results

Database analysis

Causes of neonatal death and neonatal mortality rates in the European Region

Prematurity was reported as the leading cause (41.2%), followed by congenital anomalies (28.9%) (Fig. 1A). Both causes significantly declined over the study period (Fig. 1B). However, the use of aggregate cause of death categories (e.g., prematurity, congenital anomalies) meant that the WHO data lacked sufficient granularity to assess the cause-specific distribution of neonatal deaths over time within the European Region.

Fig. 1.

Fig. 1

Open in a new tab

A Causes of neonatal death in the European Region, 2000–2021. Proportionate mortality ratios for specific causes of neonatal death, based on combined data from WHO for the defined countries in the European region. B Top causes of neonatal death over time in the European Region, 2000–2021. Trends in neonatal mortality rates per 1000 live births for the three leading causes of death. Solid lines represent annual mortality rates. Dashed lines represent the linear regression models. Shaded areas show the 95% confidence interval for each regression model. All causes showed significant decline between 2000 and 2021 (P < 0.001). The rate of decline for each cause is labelled

Data on NMR from 2000–2021 were extracted from WHO databases for 28 European countries; Liechtenstein was excluded due to lack of data [6–8] (Supplementary Table 2 and Supplementary Fig. 2). All countries showed statistically significant declining NMRs, with the exception of France (p = 0.264). The average NMR across the European region was 2.63 deaths per 1,000 live births, with a significant decline of -0.074 per year (p < 0.001) (Supplementary Table 2).

Cluster analysis

Hierarchical clustering analysis grouped the Baltic countries into one cluster, while the remaining countries formed a second cluster (Supplementary Fig. 2 and Fig. 2). Key differentiators included lower rates of death due to prematurity and higher rates of death due to birth asphyxia in the Baltic countries, as well as differing trends in leading causes of death over time (Fig. 3). Consequently, most European countries share broadly comparable mortality profiles, while the Baltic states follow a somewhat distinct trajectory. Importantly, all countries represented in our systematic review fell within the larger cluster, supporting the comparability of the pooled data.

Fig. 2.

Fig. 2

Open in a new tab

Hierarchical clustering of countries in the European Region based on neonatal mortality causes and trends. The dendrogram illustrates the results of hierarchical clustering using Ward’s minimum variance method applied to standardized cause-specific mortality proportions and their linear trends across the study period. Two main clusters are identified: a smaller cluster (red) with Latvia, Estonia and Lithuania, and a larger cluster (blue) with the remaining countries. This model illustrates regional similarities and differences in mortality rates and causes of death

Fig. 3.

Fig. 3

Open in a new tab

A Causes of neonatal death divided in clusters of European countries, 2000–2021. Proportionate mortality ratios for specific causes of neonatal death, based on combined data from WHO for the defined clusters in the European region. Cluster 1 (blue): all countries with the exception of those in cluster 2 (red): Latvia, Estonia and Lithuania. B Top causes of neonatal death over time divided in clusters of European countries, 2000–2021. Trends in neonatal mortality rates per 1000 live births for the three leading causes of death. Solid lines represent annual mortality rates. Dashed lines represent the linear regression models. Shaded areas show the 95% confidence interval for each regression model. All causes showed significant decline between 2000 and 2021 (P < 0.001). Cluster 1: all countries with the exception of those in cluster 2: Latvia, Estonia and Lithuania

Systematic review and meta-analysis

Study selection and characteristics

Out of 4,644 identified articles, 1,320 duplicates were removed. After title and abstract screening of the remaining 3,324 articles, 601 underwent full-text review. Forty articles and two WHO databases met inclusion criteria, with one additional article included through reference list screening (Supplementary Fig. 1).

The 41 included studies covered 17 countries. The most frequently represented countries were the United Kingdom (n = 14), Denmark (n = 4), and Poland (n = 4). The study with the highest neonatal death incidence was Gatt et al., analyzing Maltese births from 1994–2013 (n = 84,821) [23]. No eligible studies were identified from the Baltic countries. Sample sizes ranged from 71 to over 6 million neonates. Neonatal death cohorts ranged from 4 to 13,077 cases. Reporting on multiple births was inconsistent, with only 39% of studies specifying whether the cases involved singleton of multiple births. Four studies were in non-English European languages and translated using an online tool [11]. Among the studies included, designs were cohort or cross-sectional; no case–control or interventional studies were identified. An overview of the included studies is provided in Supplementary Table 3.

Classification and quality of studies

Death cause classification methods varied and included ICD-PM [21], CODAC [24], Wigglesworth [25], INCODE-based methods [26], and the system used by Patel et al. [27]. Six studies focused on NICU deaths [13, 17, 28–31], while others such as Basu et al. [26], Troszynski et al. [32], and Miranda et al. [33] included only early neonatal deaths (< 7 days). Some studies offered detailed temporal or age-stratified classifications, e.g., Williams et al. [34], and Pantou et al. [35].

Seven studies used forensic assessments and autopsy-based classification [36–42]. Liu et al. [43] used modelled estimates based on UN IGME data [44] this study was not included in the meta-analyses. Thirty-one studies were cross-sectional and 10 were cohort studies; NOS scores ranged from 2 to 9 stars [15] (Supplementary Table 3).

Heterogeneity and publication bias

Considerable heterogeneity was observed across all pooled analyses (I2 > 75%). Funnel plot inspection suggested possible publication bias in most analyses, though interpretation is limited due to the small number of studies and high heterogeneity (Supplementary Figs. 3, 4, 5, and 6). Post-hoc sensitivity analyses excluding individual study outliers had minimal effect on I2 or Tau2, with the exception of the ICD-PM:N7 meta-analysis performed on the extremely preterm subgroup where the exclusion of Stensvold et al. [29] produced a substantial reduction in I2, indicating this study was a major source of heterogeneity. However, the pooled estimates were not materially altered.

Meta-analysis: neonates of all gestational ages

Fifteen studies were eligible for meta-analysis (Tables 1 and 2), yielding eleven general neonatal cohorts. Studies reporting only data on term neonates (e.g., Ali et al. [13], and Basu et al. [26]) were excluded, as was the study from Allanson et al. [4] which reported county level data which was entirely nested within Carty et al. [16] and Smith et al. [18]. Meta-analysis results showed that 59% of neonatal deaths were attributed to ICD-10 perinatal conditions (95% CI: 50–67%) (Supplementary Fig. 7) and 30% were attributed to congenital malformations (95% CI: 21–41%) (Supplementary Fig. 8). Among ICD-PM categories, the groups with highest mortality were N1 (congenital anomalies) at 30% (95% CI: 17–46%) and N9 (low birth weight/prematurity), at 31% (95% CI: 18–49%) (Supplementary Fig. 9). The other ICD-PM categories each accounted for < 9% of deaths (Supplementary Fig. 9). Due to insufficient data, no analysis was conducted for N2, N5, N7, N8, or N10 (defined in Supplementary Table 1).

Table 1.

Data extraction sheet for meta-analyses. ICD-10

Study
(NOS score)		 Period Duration months Country Level Singleton/Multiples Included GA GA group Death period Number of deaths Perinatal conditions
(P00-P96)		 Congenital malformations
(Q00-Q99)		 Other or unspecified

Allanson et al. (2016)

(★★★★☆☆☆)

 1997- 2010 156 UK County Both  > 20 + 0 all Neonatal deaths 4204 1614 1152 1438

Best et al. (2019)

(★★★★★★★)

 2014–2015 12 UK National Singletons  > 24 + 0 all Neonatal deaths 2345 1258 774 313

Carty et al. (2023)

(★★★★★★★★☆)

 2005–2014 108 England National Singletons  > 22 + 0 all Neonatal deaths 13,077 8398 4070 609

Costeloe et al. (2012) b

(★★★★★★★)

 1 March-31 December 1995 10 England National n.d 22 + 0—25 + 6 extremely preterm NICU deaths 400 176 6 218

Costeloe et al. (2012)

(★★★★★★★)

 2006 12 England National n.d 22 + 0—26 + 6 extremely preterm NICU deaths 522 356 4 162

Miranda et al. (2020) a

(★★★★★★☆)

 2008- 2017 108 Portugal NICU n.d  > 23 + 0 all Early Neonatal deaths 76 2 56 18

Nohr et al. (2021)

(★★★★★★★★★)

 1996–2002 72 Denmark National Singletons  > 22 + 0 all Neonatal deaths 226 81 91 54

Pantou et al. (2010) a

(★★★★★☆☆)

 1996–2004 96 Greece Regional & NICU n.d  > 22 + 0 all Early Neonatal deaths 41 33 5 3

Pantou et al. (2010)

(★★★★★☆☆)

 1996–2004 96 Greece Regional & NICU n.d  > 22 + 0 all Neonatal death 103 53 11 39

Smith et al. (2010)

(★★★★★☆☆)

 1997–2007 132 England National Singletons  > 24 + 0 all Neonatal death 17,981 13,169 4464 348

Stensvold et al. (2017)

(★★★★★★★★☆)

 2013–2014 12 Norway National Both 22 + 0—26 + 6 extremely preterm NICU deaths 66 32 2 32

Tambe et al. (2015)

(★★★★★☆☆)

 2006–2008 24 UK National n.d n.d all Neonatal deaths 8960 6817 1575 568

Tambe et al. (2015)

(★★★★★☆☆)

 2006–2008 24 Sweden National n.d n.d all Neonatal deaths 574 307 158 109

Troszynski et al. (2011) a

(★★★★☆☆☆)

 2007–2009 24 Poland Regional n.d n.d all Early Neonatal deaths 1412 683 435 294

van Beek et al. (2021) b

(★★★★★★★★☆)

 2007–2009 24 Netherlands National Both 24 + 0—26 + 6 extremely preterm NICU deaths 203 131 1 71

van Beek et al. (2021)

(★★★★★★★★☆)

 2011–2017 72 Netherlands National Both 24 + 0—26 + 6 extremely preterm NICU deaths 603 441 7 155

Verhagen et al. (2010)

(★★★★★★☆)

 2005- 2006 12 Netherlands NICU n.d  > 22 + 6 all NICU deaths 52 26 26 0
Open in a new tab

Data on neonatal mortality causes were classified into ICD-10 categories. NOS scores are indicated below the study name. n.d. = no data. a = indicates early neonatal death cohorts. b = indicates the older cohort in a study which compares two time periods

Table 2.

Data extraction sheet for meta-analyses. ICD-PM

Study
(NOS score)		 Period Country Level Singleton/
Multiples		 Included
GA		 GA
group		 Death
period		 Number
of deaths		 N1 N2 N3 &
N4		 N5 N6 N7 N8 N9 N10 N11

Allanson et al. (2016)

(★★★★☆☆☆)

 1997- 2010 UK County Both  > 20 + 0 all Neonatal deaths 4204 1152 13 84 25 67 286 105 1328 55 1097

Best et al. (2019)

(★★★★★★★)

 2014–2015 UK National Singletons  > 24 + 0 all Neonatal deaths 2345 774 286 38 203 73 75 660 238

Carty et al. (2023)

(★★★★★★★★☆)

 2005–2014 England National Singletons  > 22 + 0 all Neonatal deaths 13,077 4070 1494 726 6178 609

Costeloe et al. (2012) b

(★★★★★★★)

 1 Mar-31 Dec 1995 England National n.d 22 + 0—25 + 6 extremely preterm NICU deaths 400 6 157 119 29 99

Costeloe et al. (2012)

(★★★★★★★)

 2006 England National n.d 22 + 0—26 + 6 extremely preterm NICU deaths 522 4 1 258 110 97 52

Miranda et al. (2020) a

(★★★★★★☆)

 2008- 2017 Portugal NICU n.d  > 23 + 0 all Early Neonatal deaths 76 56 22 1 7 13 14 34 5

Nohr et al. (2021)

(★★★★★★★★★)

 1996–2002 Denmark National Singletons  > 22 + 0 all Neonatal deaths 226 91 26 23 32 54

Pantou et al. (2010) a

(★★★★★☆☆)

 1996–2004 Greece Regional & NICU n.d  > 22 + 0 all Early Neonatal death 41 5 3 5 25 3

Pantou et al. (2010)

(★★★★★☆☆)

 1996–2004 Greece Regional & NICU n.d  > 22 + 0 all Neonatal death 103 11 9 34 37 4 6 2

Smith et al. (2010)

(★★★★★☆☆)

 1997–2007 England National Singletons  > 24 + 0 all Neonatal death 17,981 4464 1295 1677 8238 348

Stensvold et al. (2017)

(★★★★★★★★☆)

 2013–2014 Norway National Both 22 + 0—26 + 6 extremely preterm NICU deaths 66 2 2 19 5 26 13 59 6

van Beek et al. (2021) b

(★★★★★★★★☆)

 2007–2009 Netherlands National Both 24 + 0—26 + 6 extremely preterm NICU deaths 203 1 41 52 111 2 19

van Beek et al. (2021)

(★★★★★★★★☆)

 2011–2017 Netherlands National Both 24 + 0—26 + 6 extremely preterm NICU deaths 603 7 55 128 157 3 27
Open in a new tab

Data on neonatal mortality causes were classified into ICD-PM categories, described in Supplementary Table 1. NOS scores are indicated below the study name. n.d. = no data. a = indicates early neonatal death cohorts. b = indicates the older cohort in a study which compares two time periods

Meta-analysis: extremely preterm neonates

Five cohorts from three studies focused on extremely preterm neonates (Tables 1 and 2). Meta-analyses showed that 61% of deaths were due to ICD-10 perinatal conditions (95% CI: 50–70%) (Supplementary Fig. 10) and ICD-PM categories N6 (Infection), N7 (Respiratory and cardiovascular disorders), and N8 (Other neonatal conditions) each accounted for 21–26% of deaths (Supplementary Fig. 11).

Discussion

This study aimed to explore the causes of neonatal deaths in the European Region and examine the cause-specific distribution of neonatal deaths over time, with a particular focus on extremely preterm neonates. Based on the WHO data [6], the leading causes were reported to be prematurity (41.2%) and congenital anomalies (28.9%). All the major causes of death were declining over the study period.

The WHO database is a key global resource for neonatal mortality surveillance. However, its global structure may not suit the specific analytical needs of high-income regions such as Europe. Several cause-of-death categories (e.g., HIV/AIDS, malaria, diarrhoeal diseases, tetanus, tuberculosis), rarely reported in European countries due to high levels of hygiene and universal healthcare access [13, 45], may obscure regionally relevant conditions [2]. Another limitation is that cause-specific mortality data from WHO, with the exception of the specific infectious diseases mentioned above, are only reported in broad categories without etiological detail. Additionally, WHO NMRs are partly modelled estimates, rather than direct national statistics [46]. While this improves internal consistency and comparability across countries, it may also obscure short-term or country-specific variations in mortality.

Cluster analysis demonstrated that, with the exception of the Baltic states (Latvia, Estonia, and Lithuania) most European countries share broadly similar neonatal mortality patterns. In contrast, the cluster comprising the Baltic states was characterized by lower mortality from prematurity and higher mortality from birth asphyxia as compared to the larger cluster. None of the studies identified in our systematic review originated from the Baltic states, which may reflect their smaller populations or different research infrastructures. This highlights the broad comparability across most European countries and the importance of recognizing sub-regional differences.

The systematic review was performed with the aim of providing some of the granularity missing in the WHO data. We identified 41 relevant studies from an initial pool of 4,644 citations. Many were excluded due to broad outcome measures, such as “infant mortality” or “perinatal death”, that precluded meaningful neonatal-specific analysis. NICU-based death reports were retained, acknowledging that many preterm infants may be chronologically older than 28 days at death but still fall within the neonatal period [17, 28]. A major limitation of the existing literature is the inconsistency in classification systems. Studies often used vague or ambiguous categories such as “neonatal” or “prematurity” without further detail. For instance, Allanson et al. reported that 77.3% of neonatal deaths in the UK occurred in preterm infants, but without more precise coding, actionable insights remain limited [4]. The absence of standardized classification frameworks creates heterogeneity across studies and countries. Variations in live birth definitions, abortion policies, prenatal diagnostics, and ethical standards regarding the withdrawal of care in cases of severe complications may influence how deaths are registered and interpreted [29, 31, 47]. Without standardized approaches, producing reliable cross-country comparisons and pooled estimates remains a challenge.

This review adhered to PRISMA guidelines [5] and implemented a comprehensive literature search. Dual independent reviewers performed screening, data extraction, and quality appraisal using the NOS scale [15]. However, methodological challenges were encountered. To standardize data, all study-reported causes were recoded into ICD-PM categories [21], which may have introduced misclassification despite best efforts. Given the low incidence of neonatal death in the included countries, many studies were based on small sample sizes. While exclusion based on study size was avoided to minimize selection bias, smaller cohorts contributed to substantial heterogeneity in the meta-analyses. Additionally, several large English cohorts [16, 18, 19] reported partially overlapping data, and therefore some neonatal deaths are duplicated. Although our meta-analyses pool study-level proportions rather than summing deaths across all cohorts, this duplication nonetheless means that deaths captured in overlapping datasets have a disproportionate influence on pooled estimates. As these overlapping cohorts are from England, this may have biased our pooled estimates somewhat toward the English cause-of-death distribution. However, given the inclusion of studies from 10 other countries and the results of our cluster analysis indicating broadly similar neonatal mortality patterns across these countries, the overall impact on the principal findings is likely limited. A Further limitation is that the review did not include grey literature sources such as national health authority reports. The accessibility, indexing, and use of heterogeneous definitions make it difficult to apply systematic and comparable review methods across all included countries. Similarly, multi-country initiatives such as EURO-PERISTAT provide harmonized indicators of perinatal health, but do not consistently present cause-specific neonatal mortality distributions [48]. Future work integrating governmental and multi-country reports with peer-reviewed studies may help to strengthen surveillance and cross-country comparisons.

A significant concern is the dominant use of the term “prematurity” as a cause of death. As a classification, “prematurity” lacks specificity, analogous to using “old age” in adult death certificates and fails to capture the underlying pathologies that lead to neonatal death in preterm infants. This is particularly relevant for extremely preterm infants, among whom twin pregnancies are overrepresented (~ 10%). The distinct physiological profiles of these infants and elevated risk of complications may influence the types and frequencies of underlying causes of death recorded. Given that most neonatal deaths occur among preterm infants [49], this lack of detail restricts our understanding and limits the clinical utility of the data for guiding interventions.

National policies around abortion and neonatal viability thresholds affect reported mortality distributions [18]. Differences in the proportion of neonatal deaths attributed to congenital malformations and prematurity across countries may, in part, reflect underlying disparities in prenatal care, pregnancy termination practices, and obstetric decision-making. Severe congenital anomalies are often detectable through routine antenatal screening programs [50] and in countries where pregnancy termination is both legally permitted and culturally accepted, such cases may be electively terminated, thereby reducing the number of neonates born with fatal anomalies. Conversely, in settings where abortion is legally restricted or socially discouraged, these pregnancies may continue to term, resulting in a higher proportion of neonatal deaths from congenital malformations. Similarly, the timing of preterm birth is not solely determined by spontaneous physiological processes but is also shaped by clinical practice. National obstetric guidelines influence when and under what conditions high-risk pregnancies, such as those complicated with fetal growth restriction, pre-eclampsia or other maternal–fetal indications, are electively delivered prematurely. Variability in these guidelines and the availability of perinatal resources likely contributes to inter-country differences in the distribution and timing of preterm births, and subsequently, to the observed patterns of neonatal mortality.

Improved specificity in neonatal death reporting is necessary to enable targeted prevention strategies. Future studies should adopt standardized, detailed classification systems and explicitly define inclusion criteria for neonatal age brackets (e.g., early vs. late neonatal death), as well as multi-fetal vs singleton pregnancies. Vague umbrella categories like “preterm birth” or “other” should be avoided. More frequent use of autopsies could also substantially improve diagnostic accuracy. For example, Liebrechts-Akkerman et al. demonstrated that autopsy findings provided critical insight into causes of death not captured by standard certification [39]. Yet, autopsy rates have declined in countries such as Denmark [51], and re-expanding their use in clinical practice will require addressing ethical, logistical, and financial barriers. Establishing standardized mortality classifications across Europe would strengthen the validity of national datasets and improve the reliability of cross-border comparisons. Harmonized registration practices, integrated with robust classification systems such as ICD-PM, would enhance the depth of epidemiological analysis and inform health policy.

Conclusions

This study demonstrated that prematurity and congenital anomalies were the leading causes of neonatal death in a frame of decreasing NMRs across the European Region between 2000 and 2021. However, broad cause of death classifications reflects the limited granularity of current data sources, particularly the WHO database, which lacks diagnostic detail suited to high-income settings. The systematic review of the literature revealed considerable heterogeneity in classification methods and outcome reporting across studies. This variability underscores the urgent need to use standardized, detailed frameworks, such as ICD-PM, for neonatal mortality reporting.

A consistent and granular approach to classifying neonatal deaths, particularly among preterm neonates, is essential to advance the accuracy of epidemiological data, enable meaningful international comparisons, and guide more effective clinical and public health interventions.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 6.08 MB) (6.1MB, docx)

Abbreviations

EEA

European Economic Area

ICD-10

International Classification of Diseases, 10th Revision

ICD-PM

The WHO application of ICD-10 to deaths during the perinatal period

LB

Live births

Lm coeff

Linear regression coefficient

NMR

Neonatal mortality rate

NOS

Newcastle-Ottawa Scale

SD

Standard deviation

SDGs

United Nations’ Sustainable Development Goals

UNECE

United Nations Economic Commission for Europe

WHO

World Health Organization

Authors' contribution

Elizabeth E. la Cour, Michael Christiansen, Ulrik Lausten-Thomsen, and Paula L. Hedley contributed to the study conception and design. Material preparation, data collection and analysis were performed by Elizabeth E. la Cour, Veronica N. E. Malange and Paula L. Hedley. The first draft of the manuscript was written by Elizabeth E. la Cour and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1. Sharrow D, Hug L, Liu Y, Lindt N, Fell GW, Nie W, You D (2025) Levels and trends in child mortality report 2024. UNICEF, New York, USA
  • 2. WHO (2024) Newborn mortality. https://www.who.int/news-room/fact-sheets/detail/newborn-mortality Accessed Date: 17/03/2025
  • 3. UN (2023) The sustainable development goals report. Geneva, Switzerland
  • 4.Allanson ER, Vogel JP, Tuncalp Ӧ, Gardosi J, Pattinson RC, Francis A, Erwich J, Flenady VJ, Froen JF, Neilson J, Quach A, Chou D, Mathai M, Say L, Gulmezoglu AM (2016) Application of ICD-PM to preterm-related neonatal deaths in South Africa and United Kingdom. BJOG 123:2029–2036 [DOI] [PubMed] [Google Scholar]
  • 5.Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, Shamseer L et al (2021) The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372:n71 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. WHO (2024) Neonatal mortality rate (per 1000 live births). https://data.who.int/indicators/i/E3CAF2B/A4C49D3 Accessed Date: 29/01/2025
  • 7. WHO (2025) Maternal, newborn, child and adolescent health and ageing. https://platform.who.int/data/maternal-newborn-child-adolescent-ageing Accessed Date: 15/01/2025
  • 8. UNECE (2024) Total live births by sex. https://w3.unece.org/PXWeb2015/pxweb/en/STAT/STAT__30-GE__02-Families_households/01_en_GEFHTotLiveBirths_r.px/ Accessed Date: 29/01/2025
  • 9. WHO (2025) The global health observatory: explore a world of health data. https://www.who.int/data/gho/data/indicators/indicator-details/GHO/neonatal-mortality-rate-(per-1000-live-births) Accessed Date: 19/08/2025
  • 10. R Core Team (2024) R: a language and environment for statistical computing. Version 4.4.2 “Pile of Leaves”
  • 11. DocTranslator DocTranslator. onlinedoctranslator.com Accessed Date: 12/03/2025
  • 12. WHO (2024) Newborn health. https://www.who.int/westernpacific/health-topics/newborn-health Accessed Date: 10/04/2024
  • 13.Ali F, Chant K, Scales A, Sellwood M, Gallagher K (2024) Neonatal organ donation: retrospective audit into potential donation in a single neonatal unit. Nurs Crit Care 29:532–535 [DOI] [PubMed] [Google Scholar]
  • 14.Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A (2016) Rayyan-a web and mobile app for systematic reviews. Syst Rev 5:210 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Wells GA, Shea B, O'Connell D, Peterson J, Welch V, Losos M, Tugwell P (2019) The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp Accessed Date: 20/01/2020
  • 16.Carty L, Grollman C, Plachcinski R, Cortina-Borja M, Macfarlane A (2023) Neonatal mortality in NHS maternity units by timing and mode of birth: a retrospective linked cohort study. BMJ Open 13:e067630 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Costeloe KL, Hennessy EM, Haider S, Stacey F, Marlow N, Draper ES (2012) Short term outcomes after extreme preterm birth in England: comparison of two birth cohorts in 1995 and 2006 (the EPICure studies). BMJ 345:e7976 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Smith LK, Manktelow BN, Draper ES, Springett A, Field DJ (2010) Nature of socioeconomic inequalities in neonatal mortality: population based study. BMJ 341:c6654 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Tambe P, Sammons HM, Choonara I (2015) Why do young children die in the UK? A comparison with Sweden. Arch Dis Child 100:928–931 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. WHO (2015) International statistical classification of diseases and related health problems, 10th revision, Fifth edition, 2016. World Health Organisation, Geneva, Switzerland
  • 21. WHO (2016) The WHO application of ICD-10 to deaths during the perinatal period: ICD-PM. World Health Organisation, Geneva Switzerland
  • 22.Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21:1539–1558 [DOI] [PubMed] [Google Scholar]
  • 23.Gatt M, England K, Grech V, Calleja N (2015) Contribution of congenital anomalies to neonatal mortality rates in Malta. Paediatr Perinat Epidemiol 29:401–406 [DOI] [PubMed] [Google Scholar]
  • 24.Froen JF, Pinar H, Flenady V, Bahrin S, Charles A, Chauke L, Day K, Duke CW, Facchinetti F, Fretts RC, Gardener G, Gilshenan K, Gordijn SJ, Gordon A, Guyon G, Harrison C, Koshy R, Pattinson RC, Petersson K, Russell L, Saastad E, Smith GC, Torabi R (2009) Causes of death and associated conditions (Codac): a utilitarian approach to the classification of perinatal deaths. BMC Pregnancy Childbirth 9:22 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Wigglesworth JS (1980) Monitoring perinatal mortality. A pathophysiological approach. Lancet 2:684–686 [DOI] [PubMed] [Google Scholar]
  • 26.Basu MN, Johnsen IBG, Wehberg S, Sorensen RG, Barington T, Norgard BM (2018) Causes of death among full term stillbirths and early neonatal deaths in the region of Southern Denmark. J Perinat Med 46:197–202 [DOI] [PubMed] [Google Scholar]
  • 27.Patel RM, Rysavy MA, Bell EF, Tyson JE (2017) Survival of infants born at periviable gestational ages. Clin Perinatol 44:287–303 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Costa S, Rodrigues M, Centeno MJ, Martins A, Vilan A, Brandao O, Guimaraes H (2011) Diagnosis and cause of death in a neonatal intensive care unit–how important is autopsy? J Matern Fetal Neonatal Med 24:760–763 [DOI] [PubMed] [Google Scholar]
  • 29.Stensvold HJ, Klingenberg C, Stoen R, Moster D, Braekke K, Guthe HJ, Astrup H, Rettedal S, Gronn M, Ronnestad AE, Norwegian Neonatal N (2017) Neonatal morbidity and 1-year survival of extremely preterm infants. Pediatrics 139:e20161821 [DOI] [PubMed] [Google Scholar]
  • 30.van Beek PE, Groenendaal F, Broeders L, Dijk PH, Dijkman KP, van den Dungen FAM, van Heijst AFJ, van Hillegersberg JL, Kornelisse RF, Onland W, Schuerman F, van Westering-Kroon E, Witlox R, Andriessen P (2021) Survival and causes of death in extremely preterm infants in the Netherlands. Arch Dis Child Fetal Neonatal Ed 106:251–257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Verhagen AA, Janvier A, Leuthner SR, Andrews B, Lagatta J, Bos AF, Meadow W (2010) Categorizing neonatal deaths: a cross-cultural study in the United States, Canada, and the Netherlands. J Pediatr 156:33–37 [DOI] [PubMed] [Google Scholar]
  • 32.Troszynski M, Maciejewski T, Wilczyfiska A, Banach B (2011) Causes of stillbirths and perinatal death in Poland between 2007–2009. Ginekol Pol 82:598–601 [PubMed] [Google Scholar]
  • 33.Miranda M, Costa S, Soares H, Barbosa J, Flor-de-Lima F, Rodrigues M, Guimaraes H (2020) A Importância da Autópsia na Morte Neonatal Precoce em Portugal. Acta Med Port 33:811–818 [DOI] [PubMed] [Google Scholar]
  • 34.Williams EJ, Embleton ND, Bythell M, Ward Platt MP, Berrington JE (2013) The changing profile of infant mortality from bacterial, viral and fungal infection over two decades. Acta Paediatr 102:999–1004 [DOI] [PubMed] [Google Scholar]
  • 35.Pantou K, Drougia A, Krallis N, Hotoura E, Papassava M, Andronikou S (2010) Perinatal and neonatal mortality in Northwest Greece (1996–2004). J Matern Fetal Neonatal Med 23:1237–1243 [DOI] [PubMed] [Google Scholar]
  • 36.Borowska-Solonynko A, Krajewski P (2011) Causes of perinatal deaths in children delivered out of hospital in material collected by chair and department of forensic medicine, medical University of Warsaw. Arch Med Sadowej Kryminol 61:139–145 [PubMed] [Google Scholar]
  • 37.Galante N, Blandino A, Disegna M, Franceschetti L, Casali MB (2024) Intentional child and adolescent homicides in Milan (Italy): a 30-year interdisciplinary study. Leg Med (Tokyo) 68:102433 [DOI] [PubMed] [Google Scholar]
  • 38.Gheorghe A, Banner J, Hansen SH, Stolborg U, Lynnerup N (2011) Abandonment of newborn infants: a Danish forensic medical survey 1997–2008. Forensic Sci Med Pathol 7:317–321 [DOI] [PubMed] [Google Scholar]
  • 39.Liebrechts-Akkerman G, Bovee JV, Wijnaendts LC, Maes A, Nikkels PG, de Krijger RR (2013) Histological findings in unclassified sudden infant death, including sudden infant death syndrome. Pediatr Dev Pathol 16:168–176 [DOI] [PubMed] [Google Scholar]
  • 40.Makhlouf F, Rambaud C (2014) Child homicide and neglect in France: 1991–2008. Child Abuse Negl 38:37–41 [DOI] [PubMed] [Google Scholar]
  • 41.Ptaszynska-Sarosiek I, Niemcunowicz-Janica A, Filimoniuk M, Oklota M, Wardaszka Z, Szeremeta M, Sackiewicz A (2011) The analysis of neonatal deaths based on autopsy protocols of the Department of Forensic Medicine in Bialystok in the years 1955–2009. Arch Med Sadowej Kryminol 61:367–372 [PubMed] [Google Scholar]
  • 42.Schulte B, Rothschild MA, Vennemann M, Banaschak S (2013) Examination of (suspected) neonaticides in Germany: a critical report on a comparative study. Int J Legal Med 127:621–625 [DOI] [PubMed] [Google Scholar]
  • 43.Liu L, Oza S, Hogan D, Chu Y, Perin J, Zhu J, Lawn JE, Cousens S, Mathers C, Black RE (2016) Global, regional, and national causes of under-5 mortality in 2000–15: an updated systematic analysis with implications for the sustainable development goals. Lancet 388:3027–3035 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44. IGME (2025) Child mortality, stillbirth, and causes of death estimates. https://childmortality.org/ Accessed Date: 19/08/2025
  • 45. WHO (2019) Water, sanitation and hygiene: burden of disease. https://www.who.int/data/gho/data/themes/topics/water-sanitation-and-hygiene-burden-of-disease Accessed Date: 18/03/2025
  • 46. Hug L, Alexander M, You D, Alkema L, Estimation UNI-aGfCM (2019) National, regional, and global levels and trends in neonatal mortality between 1990 and 2017, with scenario-based projections to 2030: a systematic analysis. Lancet Glob Health 7:e710-e720
  • 47.Farrah K, Young K, Tunis MC, Zhao L (2019) Risk of bias tools in systematic reviews of health interventions: an analysis of PROSPERO-registered protocols. Syst Rev 8:280 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. PERISTAT E (2022) European perinatral health report: core indicators of the health and care of pregnant women and babies in Europe from 2015 to 2019.
  • 49. WHO (2023) Preterm birth. https://www.who.int/news-room/fact-sheets/detail/preterm-birth Accessed Date: 16/03/2025
  • 50. WHO (2025) Congenital disorders. https://www.who.int/health-topics/congenital-anomalies#tab=tab_1 Accessed Date: 07/052025
  • 51.Sunekaer K, Hansen SH, Banner J (2022) Trends in infant mortality: an evaluation of forensic autopsied infants in Eastern Denmark over 39 years. Int J Legal Med 136:169–178 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary file1 (DOCX 6.08 MB) (6.1MB, docx)

Data Availability Statement

No datasets were generated or analysed during the current study.

Articles from European Journal of Pediatrics are provided here courtesy of Springer

RESOURCES