Morbidity patterns, mortality, and antimicrobial resistance among children hospitalized in neonatal and pediatric intensive care units in a rural area of Vietnam

Highlights

• •Respiratory infections dominated admissions (44.6%).
• •Infectious diseases prolonged hospital stay.
• •Worst outcomes (3.4%) were mainly driven by respiratory failure/asphyxia and prematurity.
• •High antimicrobial resistance burden: 61.5% multidrug resistance, 6.9% extensively drug-resistant, and 1.1% pandrug-resistant among bacterial isolates.
• •Extended-spectrum beta-lactamase (44.2%), carbapenemase-producing Enterobacteriaceae (19.2%), and methicillin-resistant Staphylococcus aureus (87.8%) were alarmingly prevalent.

Abstract

Objectives

To describe disease burden and antibiotic resistance among children hospitalized in the neonatal department (ND) and pediatric intensive care unit (PICU) in a rural area of Vietnam.

Methods

This was a retrospective study. Data on admissions to the ND and PICU from 2022 to 2024 were extracted and analyzed from electronic medical records.

Results

2359 patients were included. Respiratory tract infections accounted for 44.6% of all diagnoses, followed by respiratory failure or asphyxia (24.3%), neonatal jaundice (11.5%), and sepsis (9.9%). Cardiovascular, digestive, neurological diseases, COVID-19, and anemia each contributed 0.5-2.4%. Length of stay was shorter for non-infectious diseases compared with infectious diseases (median 6 vs 10 days, P < 0.001). The overall proportion of worst outcomes (death or severe illness with discharge at family request) was 3.4%. Respiratory failure or asphyxia accounted for 70.0% of these cases, followed by prematurity (26.2%) and respiratory tract infections (16.2%). Among 174 bacterial isolates, 61.5% were multidrug-resistant, 6.9% extensively drug-resistant, and 1.1% pandrug-resistant with 44.2% of extended-spectrum beta-lactamase Enterobacteriaceae and 19.2% of carbapenemase-producing Enterobacteriaceae. A total of 36 of 41 (87.8%) Staphylococcus aureus isolates were methicillin-resistant.

Conclusion

Improving survival in resource-limited settings requires better perinatal and respiratory care, stronger infection prevention, timely diagnostics, and antibiotic stewardship.

Introduction

Newborns and young children are vulnerable to life-threatening illnesses that require close monitoring and prompt management [1]. In rural hospitals in low- and middle-income countries, neonatal department (ND) and pediatric intensive care units (PICUs) play a key role in stabilizing and treating this population in the region. However, they often operate with limited staffing, diagnostic equipment, and treatment support, leading to poorer outcomes [[2], [3], [4]].

Previous studies have shown that the morbidity pattern in NDs mainly includes prematurity and extreme prematurity and their related complications, respiratory distress, perinatal asphyxia, hyperbilirubinemia, and congenital heart disease [1,5,6]. For extremely preterm infants, recent multicenter data showed that the risk of in-hospital mortality remains high despite advances in perinatal and neonatal care [7]. In addition, the severity of the illness, the need for invasive support, and comorbidities can prolong the length of hospital stay, increasing the risk of nosocomial infection and mortality in this population [8].

In PICUs, severe pneumonia, meningitis/encephalitis, septic shock, neurological emergencies, poisoning, and severe dehydration from diarrheal diseases, and sepsis remain major causes of organ dysfunction and death worldwide [9,10]. In addition, healthcare-related morbidity is common and has long-term physical, cognitive, and psychosocial consequences for children and their families [11,12].

Despite improvements in the quality of health care, mortality among critically ill infants and children remains significant, particularly in rural areas with limited resources and challenges in transporting patients to central levels [5,8,13]. Risk stratification based on morbidity and mortality patterns has therefore been recognized as important for refining prognosis and standardizing assessment across heterogeneous patient groups [14]. The COVID-19 pandemic has highlighted differential risks for severe illness and death among hospitalized children, underscoring the need to accurately identify high-risk groups and develop context-appropriate care pathways [15].

Another major emerging health concern is antimicrobial resistance (AMR). In remote health facilities, microbiological testing to provide evidence for antibiotic therapy is often unavailable due to the absence of bacterial culture services [16]. Moreover, in the acute settings described above, patients admitted to PICUs frequently present with infections that necessitate early empiric antibiotic treatment. Clinicians commonly initiate antibiotics without microbiological confirmation and may use prolonged or broad-spectrum regimens. This practice can promote the selection of multidrug-resistant organisms, including carbapenem-resistant Enterobacterales. Similar to a vicious cycle, AMR increases both morbidity and mortality in pediatric settings through higher disease burden, treatment failures, and prolonged hospital stays [[17], [18], [19]].

Understanding the burden of disease in neonatal units and PICUs is essential for guiding appropriate resource allocation in settings with limited capacity [20]. However, studies in Vietnam remain limited. Existing research often examines single components, such as infection rates, mortality, or length of stay, without integrating disease burden, mortality, and antibiotic resistance patterns within a unified framework [21,22]. Moreover, many studies focus on central hospitals rather than provincial or rural facilities [21,23]. This limits the ability to develop policies and quality-improvement strategies that reflect local disease patterns and resource constraints. Our study aims to fill these gaps by describing the triple threat of disease patterns, mortality, and antibiotic resistance among children admitted to the ND and PICU in a rural area of Vietnam.

Methods

Study design and location

This was a retrospective study, conducted at Thai Binh Pediatric Hospital (TBPH), the provincial referral children’s hospital for Thai Binh, Vietnam. Before its merger with Hung Yen to form Hung Yen Province in July 2025, Thai Binh was a coastal province in the Red River Delta with approximately 2 million residents, low urbanization, and a relatively older age structure; children aged 0-14 years accounted for about 20% of the population.

TBPH is located at 02 Ton That Tung Street and functions as the main pediatric referral center in the province. Recent administrative report indicate a hospital scale of 600 beds, including an ND with 45 beds and a PICU with 30 beds. A previous 5-year review from TBPH described high admission volumes and predominantly infectious disease cases, supporting its role as the provincial pediatric hospital [24].

Study population

All hospitalizations of neonates and children admitted to the ND and PICU from 2022 to 2024 were included. Neonate were defined as patients aged <28 days at admission, and children as ≥28 days to <16 years (hospital standard). We excluded day-care/short-stay encounters (<24 hours) and admissions without a recorded discharge disposition. Repeated admissions of the same individual were analysis separately.

Data collection and definition

Data were extracted from the hospital electronic medical record and administrative databases, including demographics (age, sex), admission and discharge dates, units, diagnoses at discharge (International Classification of Disease, Tenth Revision codes), and microbiology results. Diagnoses were grouped into clinically coherent categories for non-infectious and infectious diseases.

AMR phenotypes were determined according to routine laboratory procedures using Clinical and Laboratory Standards Institute (CLSI)/European Committee for Antimicrobial Susceptibility Testing (EUCAST) breakpoints. Multidrug-resistant (MDR) organisms were defined as isolates non-susceptible to at least one agent in ≥3 antimicrobial categories. Extensively drug-resistant (XDR) organisms were defined as isolates non-susceptible to at least one agent in all but ≤2 antimicrobial categories. Pandrug-resistant (PDR) organisms were defined as isolates non-susceptible to all agents in all antimicrobial categories [25]. Based on the antibiogram, extended-spectrum beta-lactamase (ESBL) production was defined as resistance of Enterobacteriaceae to ceftazidime and cefotaxime. Carbapenemase-producing Enterobacteriaceae (CPE) were defined as resistance of Enterobacteriaceae to carbapenems. Vancomycin-resistant Enterococci (VRE) referred to Enterococci that were no longer susceptible to vancomycin. Methicillin-resistant Staphylococcus Aureus (MRSA) was defined as S. aureus resistant to methicillin.

Length of stay (LOS) was calculated as the number of days between admission and discharge. In-hospital mortality was defined as any death occurring before discharge.

Statistical analysis

RStudio ver. 2026.01.0-392 was used for statistical analysis. All data were de-identified before analyzing. Our main outcomes were: (i) morbidity patterns (distribution of diagnosis groups and organ-support use); (ii) worst outcomes; and (iii) AMR profiles among culture-confirmed infections. The secondary outcomes included LOS. In this study, worst outcomes were defined as a composite endpoint including both in-hospital deaths and children with severe illness who were discharged at the family’s request because of poor prognosis. This definition was used to reflect local clinical and cultural practice in Vietnam, where some families prefer terminally ill children to return home rather than die in hospital.

We performed descriptive analyses only. Categorical variables are summarized as counts and percentages. Mortality is presented as an overall proportion and by key diagnosis groups. LOS distributions are presented as medians (interquartile range) and compared descriptively across strata. The Wilcoxon rank-sum test was used to compare differences in median of LOS. No analytical or multivariable analyses were performed. The findings are intended to describe patterns of disease burden rather than to identify independent predictors or make inferential causal interpretations.

Results

Characteristics of included population

A total of 2359 pediatric patients were included in the study. The majority were male (57.8%, n = 1364), while females accounted for 42.2% (n = 995). Neonates aged <28 days old represented 57.5% (n = 1357) of the study population. Patients aged 28 to <60 days accounted for 26.1% (Table 1).

Table 1.

Sociodemographic characteristics of patients.

Characteristics	N = 2359	%
Gender
Female	995	42.2
Male	1364	57.8
Age
<28 days	1357	57.5
28 to <60 days	615	26.1
2 to <12 months	128	5.5
12 to <36 months	116	4.9
36 to <60 months	62	2.6
≥5 years	81	3.4

Morbidity of pediatric patients in neonatal and ICU departments

The burden of disease was predominantly attributable to infectious conditions. Among these, respiratory tract infections were the most common, accounting for 44.6% of all diagnoses. This was followed by respiratory failure and asphyxiation (24.3%), neonatal jaundice (11.5%), and sepsis (9.9%). Cardiovascular, digestive, neurological diseases, conjunctivitis, COVID-19, and anemia, comprised smaller proportions, each ranging from 0.5-2.4% of the patient population (Figure 1).

Figure 1.

Burden of morbidity among patients admitted at neonatal and pediatric intensive care departments.

LOS and mortality of pediatric patients in ND and PICUs

Patients admitted with non-infectious diseases had a shorter LOS compared with those with infectious diseases (median = 6 days vs 10 days, respectively, P < 0.001) (Figure 2).

Figure 2.

Length of stay of the patients.

IQR, interquartile range.

Regarding outcomes, the majority of neonatal patients were discharged home (96.0%), with a mortality rate of only 0.5% (n = 9). Sixty-two (3.4%) patients were referred to national hospitals, and one patient (0.1%) was discharged at family request. Among PICU patients, 386 (71.2%) were discharged, while 86 (15.9%) were transferred to national referral centers. Mortality was notably higher, occurring in 7.7% (n = 42) of cases, and 28 patients (5.2%) were discharged at family request.

Among children with worst outcomes, the main cause was respiratory failure and asphyxia (70.0%), followed by premature birth (26.2%) and respiratory tract infection (16.2%) (Figure 3).

Figure 3.

Causality of worse outcomes (mortality and worsening, discharged from hospital at family request) in patients.

Bacteria culture and antibiotic resistance of isolated pathogens

Among 179 culture-positive isolates from 1845 patients who were sampled for bacterial culture, 58.1% originated from the PICU. Specimen types included pharyngeal fluid (42.0%), blood (32.8%), and endotracheal aspirates (20.1%) (Supplementary Table 1).

In terms of microbiological classification, Gram-negative bacilli were the most common (49.2%), followed by Gram-positive cocci (34.6%) and Gram-negative cocci (11.2%). The most frequently identified pathogens were S. aureus (22.9%), Pseudomonas aeruginosa (17.9%), and Escherichia coli (14.0%) (Supplementary Table 2).

Analysis of AMR patterns revealed that 61.5% of isolates were classified as MDR, 6.9% as XDR, and 1.1% as PDR. A proportion of 44.2% and 19.2% of ESBL and CPE, respectively. A total of 36 of 41 (87.8%) S. aureus isolates were MRSA. No case of VRE was reported (Table 2).

Table 2.

Phenotypic antibiotics resistance of 174 isolated bacteria^a.

Classification of antimicrobial resistance	Number	%
No	53	30.5
Multidrug resistance	107	61.5
Extensively drug-resistant	12	6.9
Pandrug-resistant	2	1.1
Extended-spectrum beta-lactamase Enterobacteriaceae (N = 52)	23	44.2
Carbapenemase-producing Enterobacteriaceae (N = 52)	10	19.2
Vancomycin-resistant Enterococci (N = 2)	0	0
Methicillin-resistant Staphylococcus aureus (N = 41)	36	87.8

^aFive isolates of Candida spp were not included.

The Supplementary Figures 1 to 5 show the antibiotic resistance profile of the five most frequently isolated bacteria, including E. coli, Haemophilus influenzae, Klebsiella pneumoniae, P. aeruginosa, and S. aureus.

Discussion

This study provides a comprehensive picture of the burden of disease and mortality in pediatric patients admitted to neonatal and ICUs at a provincial hospital in Vietnam. Infectious diseases were the most common cause of admission and prolonged hospital stay. However, worse outcomes were mainly due to non-communicable diseases such as prematurity, neonatal respiratory distress, and underlying chronic diseases (including cardiovascular diseases, cerebral palsy, and congenital diseases).

The high proportion of neonates among admitted patients reflects the early vulnerability of this population and is consistent with global trends in low- and middle-income countries, where neonatal complications account for a substantial proportion of pediatric admissions [[26], [27], [28]]. In a recent Indian study [27], at least one neonatal complication occurred in 33% of newborns, and mortality rate was 2.8% of live births.

Respiratory tract infections were the leading cause of morbidity overall, comprising 44.6% of the total case burden, followed by sepsis, jaundice, and respiratory failure. These findings confirm the persistent burden of preventable and treatable infections in hospital settings. This observation aligns with previous studies emphasizing the dominance of infectious causes in pediatric inpatient morbidity [[29], [30], [31]]. In our previous study at the same hospital [24], infectious diseases accounted for more than 60% of all hospitalization, with respiratory and gastrointestinal infections being the most frequent. Studies from pediatric wards and ICUs in other low- and middle-income countries have also reported that respiratory infections and sepsis are the most frequent admission diagnoses [32,33]. However, unlike morbidity, mortality and serious outcomes, such as discharge at family request due to clinical deterioration, were largely driven by non-infectious conditions.

This pattern is broadly consistent with the global transition in child health, in which the overall burden of under-5 mortality has shifted from classic infectious causes toward neonatal disorders and chronic conditions, while infections remain a major cause of morbidity. Global Burden of Disease analyses have shown that neonatal disorders, including complications of prematurity and intrapartum events, are now the leading causes of death in children under 5 years, followed by lower respiratory infections and diarrheal diseases [9].

Several hospital-based studies in neonatal units support the dominant contribution of prematurity and respiratory distress to mortality. For example, Muhe et al. [5] reported that among preterm infants in low- and middle-income countries, respiratory distress syndrome accounted for 45% of deaths, with sepsis and pneumonia as the second leading group of causes. Other studies have similarly identified prematurity, neonatal respiratory distress, the need for mechanical ventilation, and adverse obstetric conditions as major predictors of neonatal death [4,13,26,34,35]. In line with these findings, the majority of deaths in our cohort occurred in preterm infants or those with severe respiratory compromise, whereas infectious diseases in this age group more often contributed to prolonged hospitalization than to immediate death. This suggests that infection is common but not always the proximate cause of death, reflecting a dual burden of disease. In practice, infections trigger hospitalization and exacerbate underlying vulnerability, but mortality is strongly shaped by gestational age, baseline organ function, and the presence of complex chronic conditions.

The presence of AMR further complicates clinical management. In our setting, MDR organisms constituted a significant proportion of bacterial isolates. This finding aligns with cohort data from neonatal units in Vietnam, where MDR is common, particularly among Gram-negative pathogens [21]. Evidence from NICUs in other countries has similarly shown that MDR infections are associated with higher risks of treatment failure, longer hospital stays, and increased consumption of health care resources [36]. In 2019 systematic analysis, six bacterial pathogens accounted for roughly three-quarters of all deaths linked to AMR worldwide. Ranked from highest to lowest mortality burden, these included E. coli, S. aureus, K. pneumoniae, S. pneumoniae, Acinetobacter baumannii, and P. aeruginosa [37]. These are also predominant isolated bacteria in our population. As resistance among pathogens causing neonatal infections continues to rise, ongoing monitoring of neonatal antibiotic use and regular evaluation of antimicrobial susceptibility patterns are essential [19].

This study has several limitations. First, it was conducted in a single provincial hospital, which may not represent the diversity of pediatric care settings across Vietnam. Second, microbiological testing was not systematically performed for all patients with infections, and cultures were more likely obtained from critically ill children, potentially inflating the observed prevalence of resistant pathogens. Third, detailed information on previous antibiotic exposure and the timing of culture collection was unavailable, limiting causal interpretation. Fourth, we did not assess genotypic resistance mechanisms, which could have provided additional insights into resistance patterns and transmission. Fifth, our definition of worst outcomes was a composite measure rather than in-hospital mortality alone. Although this approach reflects local clinical and cultural practice and may better capture severe terminal events, it may reduce direct comparability with studies using mortality as the sole endpoint. Specifically, the inclusion of discharge at family request among critically ill children may result in a higher estimate of adverse outcomes than studies restricted to documented in-hospital deaths only. Finally, post-discharge outcomes, particularly for patients who were transferred or discharged at family request, were not captured. This may have led to an underestimation of overall mortality and long-term morbidity.

Conclusion

Despite these limitations, our findings have important implications for clinical practice and health policy in comparable resource-limited settings. Reducing mortality will require sustained investment in high-quality perinatal care, prevention of preterm birth, and improved respiratory support for preterm and critically ill newborns, in line with global recommendations for managing neonatal disorders. Because infectious diseases remain a major reason for hospitalization and prolonged LOS, targeted improvements are needed. Strengthening infection prevention and control practices is essential. Ensuring timely access to microbiological diagnostics will support appropriate treatment decisions. Implementing effective antibiotic stewardship programs in neonatal and PICUs is also critical to reducing the impact of resistant pathogens.

Declaration of competing interest

The authors have no competing interests to declare.