Global, regional, and national burden of neonatal sepsis and other neonatal infections, 1990–2019: findings from the Global Burden of Disease Study 2019

To provide an overview of the global, regional, and national incidence and mortality of neonatal sepsis and other neonatal infections (NS) and their change trends from 1990 to 2019, based on the data from the 2019 Global Burden of Disease study. This was a retrospective demographic analysis based on aggregated data. Annual incident cases, deaths, age-standardized incidence rate (ASIR), age-standardized mortality rate (ASMR) and their percentage changes of NS during 1990–2019 were collected from the 2019 Global Burden of Disease study. Globally, the incident cases of NS increased by 12.79% (from 5.59 million in 1990 to 6.31 million in 2019), and the deaths decreased by 12.93% (from 0.26 million in 1990 to 0.23 million in 2019). In the globe, the ASIR of NS per 100,000 population increased by 14.35% (from 85.21 in 1990 to 97.43 in 2019), and the ASMR decreased by 11.91% (from 3.97 in 1990 to 3.5 in 2019). Conclusion: Increasing trends in incidence and decreasing trends in mortality of NS were observed worldwide from 1990 to 2019. More robust epidemiological research and effective health strategies are urgently needed to reduce the disease burden of neonatal sepsis worldwide. What is Known: • Neonatal sepsis has significant impacts on neonatal health, but estimates on the global burden and trends of neonatal sepsis are scarce and existing findings vary considerably. What is New: • Globally, there were 6.31 million incident cases of neonatal sepsis and 0.23 million deaths due to neonatal sepsis. • Increasing trends in incidence and decreasing trends in mortality of neonatal sepsis were observed worldwide from 1990 to 2019, with the highest absolute burden in sub-Saharan Africa and Asia. What is Known: • Neonatal sepsis has significant impacts on neonatal health, but estimates on the global burden and trends of neonatal sepsis are scarce and existing findings vary considerably. What is New: • Globally, there were 6.31 million incident cases of neonatal sepsis and 0.23 million deaths due to neonatal sepsis. • Increasing trends in incidence and decreasing trends in mortality of neonatal sepsis were observed worldwide from 1990 to 2019, with the highest absolute burden in sub-Saharan Africa and Asia.


Introduction
Nearly 50% deaths of children younger than 5 years occurred in neonates (aged < 28 days), and the share continues to rise [1], underlining the need to improve understanding of neonatal disease burden and take interventions accordingly. Neonatal sepsis, a dysregulated host response to infection that causes life-threatening organ dysfunction in neonates less than 28 days of age, is the third most common cause of newborn death [2,3]. It is divided into early onset sepsis (EOS, occurring within 2-7 days after birth) and late onset sepsis (LOS, occurring after 3-8 days of birth up to 28 days) according to the age of onset [4,5]. As one of the main causes of neonatal morbidity and mortality, neonatal sepsis case mortality ranges from 11 to 19% [6]. Survivors of neonatal sepsis are more likely to have long-term neurologic impairments such as visual impairment, hearing loss, cerebral palsy, and cognitive delays [7,8].
World Health Assembly emphasized the need for robust data on the global burden of sepsis across all age groups in 2017 [9]. However, few comprehensive research on neonatal sepsis incidence has been conducted despite neonates being the most vulnerable group. Global burden of disease (GBD) 2019 offers the most recent evaluation of diseases and injuries for 204 countries and territories from 1990 to 2019 with standardized and improved methods [10]. Currently, the incidence and mortality of neonatal sepsis and other neonatal infections (NS) has not yet been systematically examined at global, regional, and national levels, as well as their association with socioeconomic status. Hence, the present study examined the data from GBD 2019 for global, regional and national incidence and mortality in terms of counts and age-standardized rate (ASR) from 1990 to 2019 and their associations with socioeconomic status to provide a comprehensive and comparable analysis of neonatal sepsis burden.

Data source
The GBD 2019 assessed global burden of 369 diseases and injuries as well as 87 risk factors across 204 countries/territories over the period 1990-2019 [10]. The GBD collaborators collect multiple relevant data sources from household surveys, civil registration, vital statistics and censuses, and so on. Data from different sources are obtained from systematic review of published studies, published reports, searches of government and international organization websites and primary data sources by GBD collaborators. The data is modeled using a meta-analysis tool named DisMod-MR model in the GBD 2019 project, which have been reported in detail in previous studies [10,11]. The incidence and mortality of NS from 1990 to 2019 were collected by gender, age, and location, from the Global Health Data Exchange query tool (http:// ghdx. healt hdata. org/ gbd-resul ts-tool). Neonates are classified into 2 groups in the GBD study according to age, including early neonates (< 7 days after birth) and late neonates (7-28 days after birth). Then, 204 countries and territories are categorized into 5 socio-demographic index (SDI) regions according to their socio-economic status, including low SDI regions, low-middle SDI regions, middle SDI regions, high-middle SDI regions, and high SDI regions. Moreover, there are 21 GBD regions worldwide based on geographical contiguity and epidemiological homogeneity. Neither animals nor human participants were involved in the study, therefore, no informed consent or ethics approval was required.

Definitions
The GBD 2019 study defined NS as codes A40.1, B95.1, P36-P36.9 and P38-P39.9 on the basis of the International Classification of Diseases, Tenth Revision (ICD-10) [10]. Both clinically diagnosed and laboratory-confirmed cases of neonatal sepsis were included in the GBD database [11]. ASR refers to an approach for processing statistics of the age-specific rates for each location, year and gender based on a GBD world standard population, which eliminates potential confounding of age structure and makes statistical indicators comparable [12]. The SDI, ranging from 0 to 1, is a composite indicator of social development status according to income per capita, average years of schooling for individuals ≤ 15 years older, and total fertility rate amongst those under the age of 25.

Statistical analysis
The annual numbers of incident cases, deaths, age standardized incidence rate (ASIR), age-standardized mortality rate (ASMR) (number per 100,000 population) and their corresponding 95% uncertainty intervals (UI) were collected to describe the global burden of NS. The percentage of relative changes in incident cases, deaths, ASIR and ASMR between 1990 and 2019 were also extracted from the GBD2019 database to assess the changing trends of NS. An increasing trend was determined if both 95% UI were > 0, and a decreasing trend was determined if 95% UI were < 0. Moreover, the correlations of ASIR and ASMR with SDI values in 21 regions and 204 countries were calculated by Pearson correlation analyses. The P-value less than 0.05 was considered statistically significant in the correlation analysis. All analyses were conducted using R 4.2.1.and GraphPad Prism 9.

Global level
According to the present study, the incident cases of NS increased by 12.79% from 5.59 (95% UI 4.0-7.55) million in 1990 to 6.31 (95% UI 4.51-8.50) million in 2019, and the overall ASIR of NS increased by 14.35% (from 85.21 per 100,000 in 1990 to 97.43 per 100,000 in 2019). In the globe, the number of deaths of NS decreased by 12.93% (from 0.26 million in 1990 to 0.23 million in 2019), and the ASMR of NS declined by 11.91% (from 3.97 per 100,000 in 1990 to 3.50 per 100,000 in 2019). Consistent with the global trend, the number of incident cases and ASIR increased and the number of deaths and ASMR decreased in both male and female neonates worldwide between 1990 and 2019, and the percentage increase of incident cases and ASIR in female neonates was twice as high as in male neonates (online supplemental Tables 1 and 2).

Regional level
For SDI regions, the percentage change in ASIR of NS increased in all five SDI regions from 1990 to 2019, with little increase in high-SDI regions (Fig. 1A, online supplemental Table 2). Among the five SDI regions, only low-and low-middle-SDI regions had an increase in incident cases (increased by 60.36% and 11.02%, respectively), whereas the remaining regions decreased (Fig. 2, online supplemental Table 1). The percentage change in ASMR between 1990 and 2019 for NS was reduced in all five SDI regions, notably reduced by almost 50% in high-SDI regions (Fig. 1B, online supplemental Table 2). Over the period 1990 to 2019, the number of NS deaths decreased in low-middle-, middle-, middle-high-, and high-SDI regions, but increased by 30.72% in low-SDI regions (Fig. 3, online supplemental Table 1). In 2019, the sums of incident cases and deaths in low-, low-middle-, and middle-SDI regions account for 78.90% and 93.91% of the world's total incident cases and deaths, respectively (online supplemental Table 1).
Across the 21 GBD regions, nearly half of the GBD regions had an increasing trend in both incident cases and ASIR of NS from 1990 to 2019, with the top two increase in incident cases in Central Sub-Saharan Africa (80.75%) and Western Sub-Saharan Africa (78.05%), and the largest increase in ASIR was in East Asia (70.83%). The combined number of incident cases of NS in East Asia (1.11 million) and South Asia (1.43 million) in 2019 accounted for about 40% of the world's total neonatal sepsis cases. It was estimated that Andean Latin America experienced the most severe threat of incidence of neonatal sepsis in 2019, with approximately 174.9 neonatal sepsis of 100,000 population, followed by Central Latin America (ASIR in 2019: 160.76 per 100,000) (online supplemental Tables 1 and 2).
In the 21 GBD regions, the number of deaths due to NS decreased in 66.67% of the regions (14 GBD regions) and ASMR decreased in 80.95% of the regions (17 GBD regions) from 1990 to 2019, with the largest decrease both in death cases (-78.02%) and ASMR (-65.85%) in Central Europe. In 2019, almost 75% of the world's NS deaths occurred in South Asia (55,810), Eastern Sub-Saharan Africa (46,260) and Western Sub-Saharan Africa (63,840). The highest ASMR occurred in Western Sub-Saharan Africa (ASMR: 8.18 per 100,000) in 2019, followed by Eastern Sub-Saharan Africa (ASMR in 2019: 6.88 per 100,000) (online supplemental Tables 1 and 2).

National level
For 204 countries and territories, the absolute number of incident cases of NS was highest in China (1.11 million) in Fig. 1 The ASIRs and ASMRs of neonatal sepsis by SDI regions, from 1990 to 2019. ASIR: age-standardized incidence rate; ASMR: agestandardized mortality rate; SDI: socio-demographic index 1 3 2019, followed by India (0.75 million) (online supplemental Table 3). The largest increase of incident cases of NS was in North Macedonia (increased by 300.44%) from 1990 to 2019, followed by Afghanistan (increased by 241.37%) (Fig. 4A). The highest ASIR was observed in Mauritius (244.26 per 100,000), followed by Bangladesh (231.66 per 100,000) and Dominican Republic (230.71 per 100,000) (Fig. 4B). Increasing trends in ASIR were observed in 109 countries from 1990 to 2019, with the most pronounced increase in North Macedonia (529.08%), followed by Bulgaria (285.91%) (Fig. 4C). Decreasing trends in ASIR were observed in 95 countries from 1990 to 2019, with the largest decrease in Serbia (-78.64%) (Fig. 4C).
Among 204 countries and territories, India suffered the highest number of deaths (36,933.26) from NS in 2019, followed by Nigeria with 27,201.80 cases (online supplemental Table 4). The country with the most pronounced increase in deaths of NS from 1990 to 2019   was North Macedonia (227.49%), followed by Bulgaria (216.09%), whereas the country with the greatest decrease was Greece (-93.00%) (Fig. 5A). The highest level of ASMR in 2019 was demonstrated in Mali (ASMR: 14.11 per 100,000), followed by Ghana (ASMR: 10.14 per 100,000) and Mozambique (ASMR: 9.65 per 100, 000) in 2019 (Fig. 5B). Between 1990 and 2019, ASMR exhibited downward trends in 167 countries, with the fastest decrease in Greece (-91.58%). Conversely, increasing trends in ASMR were observed in 35 countries and regions, with Bulgaria and Taiwan (province of China) showing the fastest increase (418.90%, 418.33%), followed by North Macedonia (412.14%). The ASMR remained stable in Austria, Bosnia and Herzegovina (Fig. 5C).

Correlations of ASIR and ASMR with SDI
During the period 1990-2019, correlations between SDI and ASIR/ASMR in NS were examined at the GBD regional level. The results showed that SDI was negatively correlated with ASIR (R = -0.3, P < 0.001) and ASMR (R = -0.76, P < 0.001) (online supplemental Fig. 1A, B). Global ASIR of NS showed an increasing level during 1990-2019 and had higher than expected level in the recent 5 years. In the highincome super region, ASIR remained steady at low levels during 1990-2019, and yet they exceeded expectations in recent years except Australia. In the Latin-America superregion, only in South Latin America ASIR was decreased and below expected levels in the last 30 years. The ASIRs of Eastern Europe, Central Europe, and Southeast Asia showed overall decreasing trends from 1990 to 2019, but they remained higher than expected in the past twenty years. Finally, in the Sub-Saharan Africa super-region, Western Sub-Saharan Africa and Eastern Sub-Saharan Africa showed higher than expected ASIR during 1990-2019 (online supplemental Fig. 1A). In terms of ASMR, the globe and 11 GBD regions were above expected levels in the last decade, although they showed declining trends during this period (online supplemental Fig. 1B). At the national-level, we assessed the correlation between SDI in 2019 and ASIR/ ASMR of 204 countries in 2019 (online supplemental Fig. 2A-B). A significantly inverse association was found between SDI and ASIR (R = -0.3, P < 0.001) as well as ASMR (R = -0.79, P < 0.001) in 2019.

Discussion
The epidemiological findings of neonatal sepsis varied widely among studies, and there is a lack of precise global estimates. The GBD 2017 estimated 1.3 million cases of neonatal sepsis per year [11], while a recent meta-analysis estimated 3 million cases [6]. Neonatal sepsis incidence is estimated to be 2824 per 100,000 live births in the updated meta-analysis [4], significantly higher than 2202 per 100,000 live births reported in 2018 [6]. Although incidence of neonatal sepsis is disproportionately high in low-and middleincome countries (LMICs) [13,14], global estimates mentioned above were largely based on data from middle-and high-income countries [4,6]. Therefore, it is likely that the true burden of neonatal sepsis globally has been underestimated. Accurate assessment of the epidemiology of neonatal sepsis is critical to health decision-making, resources allocation, and contributes to the Sustainable Development Goal 3 (SDG) target of reducing neonatal mortality to at least 12 cases per 1000 live births by 2030 [15]. This is the first comprehensive study to assess the global burden of NS, as well as its relation to socioeconomic status, based on the 2019 GBD study. Globally, there were more than 6 million incident cases and 0.23 million deaths due to NS, which increased by 12.79% and declined by 12.93% from 1990, respectively. Additionally, it was found that the global ASIR and ASMR of NS increased by 14.35% and decreased by 11.91% from 1990 to 2019, respectively. Furthermore, it showed an inverse correlation between ASIR/ASMR and SDI in the past three decades at the GBD regional level, as well as an inverse correlation between ASIR/ASMR and SDI in 2019 at the national level. The data provide evidence of the significant yet under-recognized global impact of neonatal sepsis.
According to this study, the global incident cases of neonatal sepsis were more than twice as high as previously reported (6.31 million vs 3 million), likely due to different methodologies, such as varying data proceeding methods, sepsis definitions, and most importantly, different countries' data inputs. The previous study included data only from high-and middle-income countries, which precluded extrapolation of its findings since the disproportionately high neonatal sepsis burden in low-income countries [6]. Although it was observed a global increase in incident cases of neonatal sepsis, almost 90% of the increased cases came from low-SDI regions. Increased ASIR of neonatal sepsis globally could be attributed to improved survival of preterm newborns, increased hospitalizations and more use of invasive devices leading to more nosocomial LOS cases especially in high-SDI regions [16,17]. Increased reporting rates in low-and middle-SDI areas is another important possible reason [6]. The evolution of neonatal sepsis definition in recent years also might be a reason, which enabled more suspected neonatal sepsis not confirmed by laboratory tests included [11]. Decreasing trends of death caused by neonatal sepsis were demonstrated worldwide, whereas a significant increase in death cases in low-SDI regions. The GBD 2019 also showed that the highest absolute burden of neonatal sepsis in incidence and mortality was in 1 3

Fig. 5
Global trends in the mortality of neonatal sepsis in 204 countries and territories sub-Saharan Africa and Asia, which was consistent with the results of previous studies [18,19]. Several possible factors contributing to this disproportionate disease burden include rapid population growth, high maternal colonization rate, increasing antimicrobial resistance, poverty, and limited medical resources and so on [1,17,20]. For example, maternal Group B streptococcus (GBS) colonization, one of the leading pathogens causing EOS, has the highest absolute burden in sub-Saharan Africa and Asia, where there is a lack of routine prenatal GBS colonization testing and intrapartum antibiotic prophylaxis policy [21]. Notably, the predominant pathogens of EOS in South Asia are Gram negative (> 60%), suggesting that unhygienic practices in the delivery room and neonatal intensive care unit might contribute to the source of infection in EOS in this region. In addition, about 50-88% of common pathogens isolated from health facilities in South Asia are resistant to first line antibiotics, and 50-70% of the common Gram negative organisms are multidrug resistant now [18]. To change the unique pathogen profile and reduce the spiralling antimicrobial resistance in these regions, several measures could be considered including conducting routine prenatal GBS colonization testing, improving hand hygiene and aseptic routines, using vital signs as sepsis early warning system alert or risk score, diagnosing sepsis early using new nonculture-based techniques such as molecular methods and mass spectrometry, and introducing antimicrobial stewardship programmes [18,[22][23][24].
Despite the standardized and comprehensive assessment in GBD 2019, population-based epidemiology of neonatal sepsis still remains poorly understood particularly in LMICs [20]. Firstly, there is no uniform definition for neonatal sepsis that can be applied on a global scale, which hampers the comparability of sepsis epidemiology studies [25,26]. Secondly, a great deal of births taking place outside the hospital have greater risk of infection, but few sick babies receive healthcare particularly in LMICs. Lastly, the knowledge gaps on long-term neurodevelopmental impairment associated with neonatal sepsis also underestimate the disease burden [21]. Therefore, more epidemiological studies on neonatal sepsis including gestation age at time of illness, pathogen patterns, transmission sources of EOS and LOS etc., are urgently needed in all resource settings especially in LMICs where the incidence and mortality of neonatal sepsis are disproportionately high, and the trends in incident cases and deaths of neonatal sepsis are increasing. In addition to accurately identifying and quantifying burden of neonatal sepsis, it is also imperative to take effective measures including conducting inclusive research, strengthening public health systems, establishing evidence-based and context-specific management guidelines, advocacy, and promoting creative clinical interventions to decrease the considerable disease burden of neonatal sepsis [27].

Strength and limitations
Several limitations in this study should be noted. First, the limited data in less developed regions where populationbased studies and national surveillance were lacking might lead to bias. Second, the quality of available data impacted the accuracy and reliability of the estimates. For example, mis-classification and miscoding may cause original data less accurate. Third, we failed to analyze global burden and trends in neonatal sepsis by gestational age and causative pathogens, since no related data were available in the GBD study. Lastly, variations in neonatal sepsis definitions across time and regions might affect valid comparisons. Despite these limitations, this is the most comprehensive epidemiological reporting of the burden of neonatal sepsis for 204 countries and territories from 1990 to 2019. Although Ou et al. reported the changing trends of neonatal sepsis as one part of their estimates of neonatal disorders using the 2019 Global Burden of Disease Study, this study supplement and extend the previous study by reporting detailed incidence and mortality in terms of counts and ASR of different regions and countries, comparing with previous findings as well as analyzing possible reasons of differences, and providing suggestions for further works [28].

Conclusions
In summary, the global ASIR of neonatal sepsis increased and the global ASMR decreased from 1990 to 2019, with the highest absolute burden in sub-Saharan Africa and Asia. Neonatal sepsis remains a critical issue in the SDG, especially in low-and middle-income countries. Thus, efforts to reduce the disease burden of neonatal sepsis are urgently needed.
Authors' contributions JL contributed to the study conception and design. LLX, XSC, SYL and QS performed the material preparation, data collection and analysis. JL wrote the first draft of the manuscript. ZYH and XYC edited and revised the final manuscript. All authors read and approved the final version for publication.
Funding This study was supported by grants from the Clinical Research Project of Children's Hospital of Chongqing Medical University (YBXM 2019-007). The funding sources had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data availability
The datasets generated and/or analyzed during the current study are available in the GBD repository (https:// vizhub. healt hdata. org/ gbd-resul ts/).