Childhood malnutrition (undernutrition) is still a major global health problem, contributing to morbidity, mortality, and risk of disability. It refers to a combination of nutritional disorders that include underweight (mixed), wasting (acute), stunting (chronic), and micronutrient deficiencies. Severe Acute Malnutrition (SAM) is defined as weight-for-height < − 3 Z-scores of the median World Health Organization (WHO) growth standards or the presence of bilateral edema or mid-upper arm circumference < 115 mm for a child ≥ 6 months of age. It requires timely and appropriate management for the child to survive[1].

Any child with malnutrition is at increased risk of developing complications that may lead to severe illness and death [2]. Children with SAM often present with a medical condition or intercurrent illness, defined as complicated SAM, and require hospitalization and treatment[3]. Undernutrition contributes to nearly 45% of under-five child deaths, and the burden is relatively high in low- and middle-income countries. An estimated 19 million under-five children suffer from SAM, and this is estimated to account for approximately 400,000 child deaths each year [4].

More than 25% of children under five in the developing world are undernourished, accounting for approximately 143 million children. Of these 143 million malnourished children, almost three-quarters live in just 10 countries in the Sub-Saharan Africa region, where more than one-quarter of children under five are undernourished. Nigeria and Ethiopia alone account for more than 33% [5]. A meta-analysis of inpatient treatment for complicated severe acute malnutrition (SAM) in low- and middle-income settings shows a case fatality rate of 14% (range 5–30%). Several factors contribute to the high fatality rate among children admitted to inpatient care. These include hypoglycemia, infection, anemia, dehydration, hypothermia, electrolyte imbalance, HIV and TB infection, and age and sex [6]. Addressing comorbidities among severely malnourished children is a critical pillar of child survival strategies and is crucial for reducing the under-five mortality rate [6]. In Ethiopia, where the under-five child mortality rate is high (57 deaths per 1000 live births), malnutrition is a major underlying cause, accounting for an estimated 57% of child mortality [1]. Malnutrition, primarily due to SAM, costs approximately 16.5% of the national GDP. Severe acute malnutrition is the third leading cause of death for children under five (8.1%), after pneumonia and neonatal sepsis (15.3%). It accounts for 20% of pediatric hospital admissions in Ethiopia and 25–30% of deaths in many poor countries [7]. Although the mortality rate has significantly decreased in recent years, it remains high (55.2 deaths per 1000 live births in 2018).

According to the 2019 Ethiopian Mini Demographic and Health Survey report, 37% of children under five were stunted, 12% severely stunted, 7% wasted, 1% severely wasted, 21% underweight, and 6% severely underweight [8–10]. The minimum international standard for managing severe acute malnutrition (SAM), according to SPHERE standards, is a cure rate of at least 75% and a death rate of less than 10% [9]. However, case fatality rates in hospitals treating SAM in developing countries have remained high.

A study conducted at Hawassa University Referral Hospital in Southern Ethiopia showed a high death rate. Of 151 children under five admitted for SAM, 105 (70%) improved/were discharged, 23 (15.2%) defaulted, and 23 (15.2%) died [2]. An eight-year-old retrospective cohort study of 469 children aged 6–59 months with complicated severe acute malnutrition from Mekelle City's therapeutic feeding centers (TFCs) assessed treatment outcomes and identified risk factors for mortality; the death rate was 12.8% [11]. In a recently published systematic review and meta-analysis in Ethiopia the pooled mortality rate was found 8.32 per 1000 person days observation [12].

The war between the Federal Democratic Republic of Ethiopia and the Tigray regional government, which erupted in November 2020 and ended in November 2022 with a lasting peace agreement in Pretoria, South Africa, and the subsequent blockade and siege, resulted in a near collapse of the health system and infrastructure in the Tigray region [13–15]. During the war, the World Food Program reported that 5.2 million people (91% of the region's total population) were food insecure and required immediate humanitarian assistance; of these, 1.2 million women and children were to receive nutritionally fortified food from the WFP[13]. UNICEF revealed that in January 2021, the level of severe acute malnutrition (SAM) in Tigray was three times the global WHO emergency threshold, putting 70,000 children at risk. Furthermore, as of May 2021, Médecins Sans Frontières (MSF) found that 26.6% of 309 children screened in mobile clinics in Northwest Tigray were malnourished, and 6% were severely acutely malnourished, leading them to conclude that 'immediate action is warranted [13].

During the crisis in the Tigray region society, especially children, suffered from hunger due to war and siege. Furthermore, no recent studies have examined the outcomes and predictors of severe acute malnutrition (SAM) in Tigrayan children. Therefore, this study aimed to assess the outcomes and mortality predictors among under-five children with SAM admitted to two stabilization centers in Mekelle City. Because severe acute malnutrition is a major cause of under-five mortality, especially in developing countries, assessing outcomes and mortality predictors can improve the quality of care for SAM patients. Tigray's two-year blockade prevented humanitarian aid, making this study crucial for understanding the war's impact on SAM patients. This study will provide updated data on SAM patient outcomes and mortality predictors to clinicians, stakeholders, collaborators, and governors, addressing the data gap, particularly in resource-limited settings and war-affected populations.

The study was conducted in both stabilization centers of Mekelle General hospital and Ayder Comprehensive Specialized Hospital (ACSH) located in Mekelle City, the capital of the Tigray region, Northern Ethiopia. It was conducted from January to July 2022, while Tigray was under siege and blockade, following the war with Federal Democratic Republic of Ethiopia(FDRE). Currently, these hospitals provide various services ranging from primary to specialized care, serving urban and rural populations from nearby and distant districts. Ayder Comprehensive Specialized Hospital is a flagship hospital in the region, with 500 inpatient beds across four major departments and other specialties, and more than 150 specialists in various medical fields, as well as adequate numbers of other healthcare professionals.

The Department of Pediatrics and Child Health is a major specialty, with 22 specialists (5 of whom are subspecialists), two major wards, an eight-bed intensive care unit (ICU), a 30–40 bed neonatal intensive care unit (NICU), a general outpatient clinic, five subspecialty outpatient clinics, and an emergency room averaging 12 beds. One ward has a separate room (center) for treating malnourished children. Healthcare providers in this center follow updated, standardized treatment guidelines for severe acute malnutrition (SAM) management. According to these guidelines, all SAM patients with medical complications or who fail an appetite test are admitted to the malnutrition treatment center for inpatient follow-up and treatment.

An institution-based cross-sectional study with prospective data collection was employed. The study population comprised all pediatric patients admitted to the stabilization centers in Mekelle general hospital and ACSH. The target population consisted of children aged one month to five years with SAM. All children diagnosed with SAM and admitted to the stabilization centers were included as study participants.

All children aged one month to five years with Severe Acute Malnutrition (SAM) admitted to the stabilization centers of ACSH and Mekelle General Hospital were included. Neonates and children with moderate acute malnutrition were excluded.

All children under five years of age with Severe Acute Malnutrition (SAM) admitted to the stabilization centers of the two largest hospitals in Mekelle city who met the eligibility criteria were enrolled in the study. A meta-analysis conducted in sub-Saharan Africa, using a pre-specified protocol, was searched. This included studies from eight countries; inpatient mortality in Uganda was 14% [16].

n = Z²p(1-p)/w²

Where:

n = calculated sample size

Z = confidence interval

P = estimate of the population proportion

w = margin of error

When Z = 1.96, p = 0.14, w = 0.05:

n = (1.96)²(0.14)(1-0.14)/(0.05)² = 185

The dependent variable was mortality. Independent variables included: age, exclusive breastfeeding (EBF) status, bottle-feeding, presence of complications (pneumonia, dehydration, shock, hypoglycemia, severe anemia, hypothermia, and lethargy), presence of chronic medical illnesses (cardiac disease, neurologic disease, diabetes mellitus (DM), HIV status, and tuberculosis). Additional variables included appetite loss, use of a nasogastric (NG) tube for feeding, need for blood transfusion, and length of hospital stay.

A structured data collection tool, developed in English, was used to collect data from patient interviews. The questionnaire consisted of five parts: sociodemographic characteristics; physical examination findings, including anthropometric measurements; routine laboratory tests; presence or absence of complications; and prescribed medications and outcome variables. The questionnaire was initially developed in English and then translated into Tigrinya, the native language. The principal investigator checked the questionnaire's completeness daily throughout the study.

To ensure data quality, the following measures were undertaken: A standard data collection instrument was used. Data abstraction was pretested in 5% of the sample population at Adi-Haki primary hospital in Mekelle city, and the clarity of each question was checked and discussed with the data collectors. The principal investigator also closely supervised the activity daily. At the end of each data collection day, the principal investigator checked the completeness of the completed questionnaires and the plausibility of the recorded information to ensure data quality.

Data coding, editing, cleaning, and verification were performed before entry into SPSS. The data were then entered into Epi Data version 4, exported to SPSS version 21, and analyzed statistically. Bivariate and multivariate logistic regression analyses were performed to identify predictors of mortality among patients with severe acute malnutrition (SAM). A p-value less than 0.05 was considered statistically significant.

This study included 185 children, achieving a 100% response rate. Of the participants, 112 (60.5%) were male. The median age was 11 months (IQR 5–24 months). One hundred thirty-two (71.7%) participants were under two years of age, and 107 (57.8%) were urban residents. One hundred forty-one (76.2%) participants had exclusive breastfeeding (EBF) during the first six months of life, while 46 (24.9%) had a history of bottle feeding. Of the 185 cases reviewed, 158 (85.4%) had non-edematous severe acute malnutrition (SAM), 127 (68.6%) had primary malnutrition, and 58 (31.4%) had secondary malnutrition. The most common secondary causes were cardiac diseases 24,(12.9%) followed by neurologic diseases 20 (10.8%). Thirty-six (19.5%) children were internally displaced due to war (Table 1).

Among the participants, 131 (70.8%) had severe wasting by weight-for-height (WFH), and 101 (54.6%) had severe wasting by mid-upper arm circumference (MUAC) on admission. Seventy-two (38.9%) had severe wasting by both WFH and MUAC, and 71 (38.4%) had severe stunting (Table 2).

Hemoglobin was determined in 93 (50.3%) participants; 20 (21.5%) had severe anemia, and 10 (10.7%) had a hemoglobin level < 4 g/dL. Urine analysis was performed for 96 (51.8%) participants; 6 (6.2%) had urinary tract infections (UTIs). Stool examination was performed in 93 (50.3%) participants. Provider-initiated counseling and testing for HIV infection (PICT) was performed in 123 (66.5%) cases; 2 (1.6%) were positive.

In addition, electrolyte determinations were performed in 71 of the participants. The results revealed that 8 (11.2%) had hypokalemia, 7 (9.8%) had hyperkalemia, and 4 (5.6%) had hyponatremia (Table 3).

Among the 185 study participants, 143 (77.3%) had medical complications and comorbidities on admission. The most common medical complications identified included pneumonia 61 (33%), dehydration 59,(31.9%), shock 24 (13%), severe anemia 20( 10.8%), hypoglycemia 6 (3.2%), lethargy (12, 6.5%), hypothermia (10, 5.4%), and tuberculosis 10,(5.4%). Of the 24 shock patients, 8 (33.3%) had septic shock and 16 (66.6%) had hypovolemic shock (Fig. 1).

Among the study participants, 176 (95.1%) received antibiotics; 148 (84.1%) received first-line intravenous antibiotics. In 47 (26.7%) cases, first-line antibiotics were revised to second-line antibiotics due to various reasons, such as hospital-acquired infection (HAI), deterioration, or lack of improvement. Of the total deaths, 16 (69.5%) occurred in patients who had received revised antibiotics. During phase I, 139 (75.1%) patients received F-75, and 46 (24.9%) received diluted F-100 (Table 4).

The median length of hospital stay was 13 days (IQR 8–20 days). In response to treatment, 123 (66.5%) patients recovered, 22 (11.9%) were transferred out to other facilities, 17 (9.2%) defaulted, and 23 (12.4%, 95% CI: 7.6,17.8) died (Fig. 2).

Of the 24 patients who had shock, 8 (33.3%) died. Most deaths 16 (69.5%) occurred within the first two weeks of admission.

The association between baseline variables and the risk of mortality was analyzed using a binary logistic regression model. Variables such as bottle feeding, feeding by nasogastric tube, presence of diarrhea, dehydration, underlying cardiac illness, lethargy, severe anemia, and shock were statistically significant (p < 0.25) as predictors of child mortality. Bottle feeding increased the risk of mortality, with an unadjusted odds ratio (COR) of 4.12 (95% CI: 1.11,32.3, p < 0.05). On multivariable logistic regression, the presence of shock was an independent predictor of death during treatment (p < 0.05). Children with shock had more than four times the risk of mortality than those without shock: adjusted odds ratio (AOR) 4.40 (95% CI: 1.07,17.98, p = 0.041) (Table 5).

The study assessed the outcome and predictors of mortality among under five year old children in Mekelle city stabilization Centre amid the Tigray war. Accordingly, the recovery, defaulter and death rates were 123 (66.5%), 17 (9.2%), and 23 (12.4%), respectively. The presence of shock was an independent predictor of mortality.

The recovery rate was lower, while the death rate was higher than ,and the defaulted rate was within the standard (WHO minimum standards: recovery rate > 75%, death rate < 10%, defaulter rate < 15% [17]. This high death rate and low recovery rate may be due to delayed presentation to a stabilization center, particularly concerning the siege and blockade, which led to medication and therapeutic feeding scarcity and difficulty in strictly following WHO and national guidelines on SAM management. The crisis and the presence of medical complications contributed to this. Most patients 143 (77.3%) had at least one comorbidity. Common medical complications and comorbidities included pneumonia 61(33%), dehydration 59 (31.9%), shock 24 (13%), severe anemia 20 (21.9%), hypoglycemia 6 (6.4%), lethargy 12 (6.5%), hypothermia 10 (5.4%), and tuberculosis 10 (5.4%). This is supported by a 2020 Ugandan study in which 79.5% of participants had comorbidities [18]. Most deaths 16 (69.5%) occurred within the first two weeks of hospital admission. This aligns with a study conducted at Dilchora Hospital in eastern Ethiopia [1]. This may be because children in the early stages of treatment are more prone to fatal medical complications and mismanagement, potentially increasing the risk of death. The median length of stay of 13 days falls within the WHO's recommended minimum acceptable range for inpatient management of SAM patients (< 28 days).

The overall mortality finding was consistent with several studies from various countries, including Yemen (10%), Nigeria (7.7%), Uganda (10.7%), and four studies from Ethiopia (9%, 12.2%, 11.5%, and 12.8%) [1, 6, 7, 9, 11, 18, 19]. Similarly, a recently published systematic review and meta-analysis in Ethiopia showed the death rate 8.32 per 1000person days of observation [12]. However, this study result is higher than that of other studies conducted in various parts of Ethiopia, such as Ayder Referral Hospital (2014), in which the mortality rate was 3.6% [20]; Nekemte Referral Hospital (NRH), where it was 4.4% [5]; and Jimma University Medical Center, where it was 3.8%[21]. This gap might be due to various reasons, such as the war, siege, and blockade in the Tigray region, which contributed to difficulties in the proper implementation of SAM treatment guidelines due to scarcity of resources, late patient presentation, and delayed referral processes. Delays were caused by factors including a lack of fuel for ambulances, looting of ambulances, lack of access to other transportation, power outages, and the absence of communication channels such as telephones and internet access due to the blockade. In addition, financial constraints related to hospitalization and transportation resulted from the freezing of bank accounts and other financial institutions. Furthermore, differences in other study subject outcomes, specifically compared with Ayder's study, showed a significant difference in the defaulter rate (43.6% vs. 9.2%), which could contribute to the inconsistency.

Nonetheless, the finding in this study was lower than rates reported in studies conducted at a university teaching hospital in Zambia (30%) [22], Mulago referral and teaching hospital in Uganda (24%)[23], and Gondar University teaching hospital in Ethiopia (18%)[24]. This difference may be attributable to different methodological approaches and differences in healthcare systems.

Regarding predictors of mortality, variables such as bottle feeding, nasogastric tube feeding, diarrhea, dehydration, cardiac illness, lethargy, severe anemia, and shock were statistically significant (p < 0.25) as predictors of child mortality in bivariate analysis. However, in multivariable logistic regression, the presence of shock was an independent predictor of death during treatment (p < 0.05). Children with shock had more than four times the odds of mortality compared to those without shock( Table 5).

Similarly, a study conducted at St. Paul’s Hospital in Addis Ababa showed that children with shock were more likely to die than those without this complication [25]. In a systematic review and meta-analysis conducted by Adugna A, et al. the hazard of death among children who had shock was 4.2 times higher than those who had not shock (HR: 4.20, 95% CI:2.92 to 6.04) [12] ,which agrees with this finding.

This was also supported by a study conducted in Southern Nigeria[6] and a systematic review and meta-analysis conducted in Sub-Saharan countries[16]. Children with severe acute malnutrition (SAM) are highly at risk of shock secondary to severe infections due to depressed humoral and cellular immunity and diarrheal diseases, which can cause hypovolemic or septic shock. Shock indicates a critical patient with end-organ dysfunctions (such as low cardiac reserve) and a high risk of complications from management, such as fluid overload, further increasing the risk of death. In this study, 24 (12.9%) patients experienced shock; 8 (33.3%) cases were septic, and 16 (66.6%) were hypovolemic. Of the total deaths, 26.1% were due to septic shock, and 4.3% were due to hypovolemic shock.

Several authors reported that comorbidities such as diarrhea, pneumonia, anemia, HIV, and dehydration were significantly associated with the risk of mortality [1, 2, 16, 26, 27], unlike in this study. This could be due to several reasons, such as the small sample size in the current study and difficulties in conducting some routine investigations, including organ function tests and electrolyte determinations, particularly for patients who could not afford private care. For example, hemoglobin was not determined in 92 participants (49.7%), and an HIV test was not performed in 62 (33.5%), which is believed to have contributed to 8 (34.7%) and 4 (17.4%) of the deaths, respectively. The possible reasons for these missing determinations were a lack o f resources, including essential chemicals and reagents, and power outages due to the war, siege, and blockade in the region.

However, although medical complications such as hypoglycemia, hypothermia, infection, severe dermatosis, diarrhea, and dehydration were not statistically significant, they should be considered clinically significant contributors to death. Exclusive breastfeeding (EBF) for the first 6 months of life plays a major role in preventing childhood mortality, reducing under-five mortality by 13%. Breastfeeding reduces mortality from diseases such as diarrhea and provides immunoglobulins that promote faster recovery from illness.

The multicenter design is a strength of the study. However, there are also some important limitations. Firstly, the sample size was small. Secondly, routine laboratory investigations were not performed for various reasons (as mentioned above). The scarcity of routine medications and other materials was significant.

In conclusion, the overall mortality during the study period was high. In addition, the recovery rate was below whereas the defaulted rate was within the minimum SPHERE standard. According to the study, the presence of shock was independent predictor of death. We recommend early identification and prompt treatment of shock to decrease mortality related to SAM, particularly in resource-limited and war-affected populations. Moreover, with the growth of armed conflicts in various parts of the world, future, appropriately designed research is needed to examine the outcomes and predictors of SAM.