Treatment outcomes and associated factors of severe acute malnutrition among under-5 children in Jigjiga public hospitals, Somali region, Ethiopia: a retrospective cohort study

Abstract

Background

Severe acute malnutrition (SAM) affects about 20 million under-5 children and contributes to one million child deaths annually. Apart from the presence of clinical management protocols capable of reducing case fatality by 1%–5%, case fatality in hospitals in developing countries averages 20%–30% and has remained the same since the 1950s.

Objective

This study aimed to assess treatment outcomes and associated factors of severe acute malnutrition among under-5-year-old children admitted to Jigjiga city public hospitals.

Methods

A facility-based retrospective cohort study design was employed on patient records between 1 January 2020 and 31 December, 2021. A structured checklist was used for data extraction to collect data from patient record book. Cox proportional hazards model with a hazard ratio of 95% CI was used. The level of statistical significance was declared at a p<0.05.

Results

Overall median length of stay, recovery, death, defaulted and non-responder rate were 7 days, 257 (70.2%), 32 (8.7%), 58 (15.8%) and 19 (5.2%), respectively. Managing facility, tuberculosis (TB), pneumonia and nasogastric (NG) tube insertion were found to be significantly associated with treatment outcomes at a p<0.05.

Children who were managed at Jigjiga University Sheik Hassen Yabare Comprehensive Specialised Hospital were 57% less likely to recover from SAM than those managed at Karamardha General Hospital (adjusted hazard ratio (AHR)=0.437, 95% CI: 0.286 to 0.600). Children who did not have TB were almost three times more likely to recover than their counterparts (AHR=2.862, 95% CI: 1.604 to 5.107), and those without pneumonia were also 1.5 times more likely to recover than those with a diagnosis of pneumonia (AHR=1.509, 95% CI: 1.146 to 1.989). Furthermore, children without nasogastric tube insertion were about 1.5 times more likely to recover than their counterparts (AHR=1.472, 95% CI: 1.075 to 2.015).

Conclusions

The recovery and defaulter rates fell outside the acceptable targets set by SPHERE standards; however, the death rate was acceptable. The significant predicting factors of treatment outcome were treating facility, TB, pneumonia and NG tube insertion.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Although severe acute malnutrition treatment outcome can be improved and case fatality can be reduced to 1%–5% by proper treatment as per the guidelines, case fatality is as high as 20%–30% in developing countries.

• Factors including mismanagement, comorbidities like diarrhoea, malaria, pneumonia and tuberculosis (TB), and poor compliance with both medical and nutritional therapeutic options affect the outcome.

WHAT THIS STUDY ADDS

• This study retrospectively analysed secondary data of severely malnourished children treated in two public hospitals (one general hospital and one tertiary hospital) and examined the outcome.

• The general death rate was 8.7% but as low as 5.6% in the tertiary hospital, while significant predicting factors of treatment outcome were treating facility, TB, pneumonia and nasogastric (NG) tube insertion.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The death rate was relatively low and acceptable.

• In addition to comorbidities and NG tube insertion which may increase risk of infection, treating facility was a significant predictor of outcome in which those treated at the general hospital had a higher chance of recovery as defaulter rate was high in the tertiary hospital.

Introduction

Severe acute malnutrition (SAM) is defined as very low weight for height/length (below −3 z scores of the median referring to the WHO growth reference or <70% of the median of the National Center for Health Statistics (NCHS)) and by having nutritional pitting oedema bilaterally or mid-upper arm circumference (MUAC) <11.5 cm.¹ It is classified into oedematous (kwashiorkor) and non-oedematous (marasmus) and complicated and non-complicated SAM.² Furthermore, based on the type of cause, it can be classified as primary SAM, which is due to inadequate food supply caused by socioeconomic, political, and environmental factors, or secondary SAM, which is illness-related.³

According to the WHO treatment protocol, outcome is stated as recovered, defaulted, died, medical transfer and non-responding. Based on that, the recovery, death and default rates are considered acceptable when >75%, <10% and <15%, respectively, and alarming when <50%, >15% and >25%, respectively. Moreover, weight gain, length of stay and coverage are said to be acceptable when ≥8 g/kg/day, <4 weeks and >50–70%, respectively, and considered alarming when <8 g/kg/day, >6 weeks and <40%, respectively.⁴

Childhood undernutrition is not only a major global health problem which contributes greatly to child mortality, morbidity and decreased intellectual development but also contributes to decreased adult work capacity and increased adulthood diseases. Out of 7.5 million under-5 child deaths, about 35% is due to nutrition-related factors, and 4.4% is specifically attributable to severe wasting.^{5 6} SAM affects about 20 million under-5 children and contributes to one million child deaths annually.^{1 4} Children with SAM have 5–20 times higher likelihood of death than the well-nourished and cause death either directly or indirectly by increasing the fatality and sufferings of diarrhoea and pneumonia.⁷ Of the global 555 million under-5 children, 19 million are severely wasted. Above 90% of them live in developing countries, specifically sub-Saharan Africa and Southeast Asia.⁸

Apart from the presence of clinical management protocols capable of reducing case fatality by 1%–5%, case fatality in hospitals in developing countries averages 20%–30% and has remained the same since the 1950s.⁶ The mortality rate of children suffering from SAM, varying from 2.2% in India to as high as 42% in Malawi, has been observed, and likely causes of this are case mismanagement, comorbidities like diarrhoea, malaria, pneumonia and tuberculosis (TB) and poor compliance with both medical and nutritional therapeutic options.⁹

As per the Ethiopian Mini Demographic Health Survey 2019, about 37% of under-5 children are stunted, of which 12% are severely stunted, 7% wasted and 21% are underweight. In the Somali region, 30% (11% severe), 21% (6% severe) and 31% (10%) are stunted, wasted and underweight, respectively.¹⁰ Since the drought of the mid-80s, the pictures of severe drought and large-scale starvation have become highly linked with Ethiopia. Malnutrition can be best explained in Ethiopia as a year-round phenomenon which is not only a problem of dry seasons but a whole-year chronic problem for the majority of households across all regions of Ethiopia.¹¹

SAM still contributes to both mortality and morbidity of under-5 children in Ethiopia and specifically in the Somali region. Although health sector stakeholders invested in prevention and treatment of child malnutrition, the outcome of such efforts is not well known, especially in inpatient treatment of SAM in the eastern part of Ethiopia, where the few studies done focused on outpatient treatment. This study will contribute data on the management outcome of SAM in the study area and help improve inpatient management.

The outcome of inpatient treatment for SAM is influenced by sociodemographic factors, the presence of comorbidities, the type of malnutrition and the quality of care provided.

Objectives

General objective

To assess treatment outcomes and associated factors of SAM among under-5 children admitted to Jigjiga public hospitals between 1 January 2020 and 31 December 2021.

Specific objectives

To assess treatment outcome of SAM among under-5 children admitted to Jigjiga public hospitals.

To identify factors associated with treatment outcome of SAM among under-5 children admitted to Jigjiga public hospitals.

Methodology

Study design, area and period

Jigjiga is the capital city of the Somali regional state, which is located 627 km from Addis Ababa, the capital city of Ethiopia and 70 km from the border of Somalia. It lies between the coordinates of 9°21′N and 42°48′E and has an elevation of 1609 m above sea level. In terms of public health infrastructure, there are two public hospitals, Karamardha General Hospital and Jigjiga University Sheik Hassen Yabare Comprehensive Specialised (JJUSHYCS) Hospital, that give inpatient therapeutic care for severely acutely malnourished children. The city has a population of 884 666 people. The dominant ethnic groups are Somali (84.1%), followed by Amhara (13.9%), Oromo (0.5%) and other ethnic groups (2.5%).¹²

A health facility-based retrospective cohort was conducted using data from severely acutely malnourished children admitted to Jigjiga public hospitals between 1 January 2020 and 31 December 2021.

Population of the study

All severe acutely malnourished under-5 children who were admitted to Jigjiga city public hospitals were the source population, while those admitted between 1 January 2020 and 31 December 2021, were the study population.

Eligibility criteria

All severely acutely malnourished under-5-year-old children admitted between 1 January 2020 and 31 December 2021.

Sample size determination

Sample size was estimated by using a single population proportion formula with the following assumptions: Recovery rate of SAM in under-5 children to be 67.7,⁹ with a confidence level of 95%, a margin of error (α=5%), a degree of precision (d)=0.05 and a 10% non-response rate. The formula is n= (zα/2)² p (1−p)/(d)², n=(1.96) 2×(0.323 × 0.677)/(0.05)² n=3.84×0.22/0.0025, n=338. With a 10% allowance for lost records, the final sample size obtained was 372.

Sampling technique and procedure

Since there are only two public health facilities that provide inpatient SAM treatment, both of them were selected purposefully. But patient medical records were selected randomly by using the registration logbook with a K-value of 3, and when a medical record with missing data was found, the next record was considered.

Data collection procedure

A structured checklist was used for data extraction to collect data from registration books and patient records. Patients’ medical record numbers were obtained from the registration books so that their records could be easily retrieved from the medical record unit. One nurse and two master of public health professionals who took training filled out the data extraction form from patient records. Information that was extracted included patient age, sex, residence, admission criteria, the number of admissions, death, defaulters, date of admission and discharge, length of stay, diagnosis, treatment and discharge condition.

Additionally, anthropometric measurements, including weight, the presence of bilateral oedema and MUAC, were also among the retrieved information.

Patient and public involvement

Patients and public had no involvement in the design, analysis and dissemination of this study.

Study variables

The dependent variable was treatment outcome, categorised as either ‘recovered’ or ‘not recovered’. The independent variables included sociodemographic factors such as sex, age, admission type and residence. The type of malnutrition—kwashiorkor, marasmus or marasmus-kwashiorkor—was also analysed. Additionally, comorbidities such as HIV, TB, malaria, anaemia, pneumonia, hypoglycaemia, shock, nasogastric (NG) tube insertion and diarrhoea were considered. Routine medications like vitamin A, deworming, folic acid, routine antibiotics, vaccination and antipyretics were also examined. Additionally, anthropometric measurements including weight, height and MUAC were also among the retrieved information.

Data quality control

The checklist was prepared, and a pretest of 2% was done before the actual study began to assess the conformity of the checklist with the objective of the study, and some modifications were made as needed. To improve the quality of the data, the data collectors were trained for 2 days, particularly in the proper filling of the checklist. The data collectors were closely supervised; each completed checklist was checked to ascertain that all categories were properly filled out and corrected by the supervisor. The information was rechecked in a randomly selected subsample of 5%.

Operational definitions

Comorbidities: children with SAM, who have TB, and/or HIV and/or malaria and/or anaemia coinfection at admission to stabilisation centre.⁷

Recovered/cured: those patients who have reached the discharge criteria.¹³

Defaulter: being absent for two consecutive days of weighing.⁷

SAM: is weight for height ratio which is <−3SD under the median WHO growth criteria or weight-for-height ratio of below 70% of the median NCHS reference or presence of nutritional oedema.¹⁴

Dead: total number of patient who lost their lives during the programme.¹⁵

Non-responder: patient who did not reach the discharge criteria after 40 days of inpatient treatment programme.¹⁵

Kwashiorkor: the existence of any two-sided pitting oedema.¹³

Marasmus: weight for height ≤ −3 z scores or ≤70% of the median NCHS reference.¹³

Special medication: intravenous fluid, intravenous antibiotic and blood transfusion.⁸

Data processing and analysis

All collected data were cleaned manually for incompleteness, coded and entered into EpiData V.3.1, then exported to SPSS V.20 for further analysis. Exploratory data analysis was carried out to check the levels of missing values, possible outliers and multicollinearity. To identify predictors associated with death rates, a Cox proportional hazards model with a hazard ratio of 95% CI was used. Variables at the p<0.25 level in the bivariable analyses and stepwise forward variable selection were computed so as to identify eligible variables in the final cox regression model to identify independent predictors of mortality. All statistical tests were considered significant at a p<0.05, and a Kaplan-Meier survival estimate was done to estimate the survival function.

Furthermore, the unsteadiness of parameter estimates among variables in the final fitted model was checked using the variance inflation factor, and the goodness of fit of the final model was checked by the Nelson Aalen cumulative hazard function against the Cox-Snell residual.

Results

Sociodemographic characteristics of severely malnourished under-5 children

A total of 366 were included in the review, and there were 6 lost patient charts. Among them, 196 (53.6%) were male, and more than half of them, 204 (55.7%) aged between 6 and 24 months, while 35.2% were under 6 months, with a mean and median age of 12 and 9 months, respectively. The majority (72.7%) of the children whose records were reviewed were from urban settings. There were only 20 (5.5%) readmitted severely malnourished children, while the rest, 346 (94.5%), were new admissions. As far as the source of referral is concerned, the majority (314 or 85.8%) of the admitted children came spontaneously, followed by those sent from health centres, constituting 38 (10.3%) of the total number, and the rest were from hospitals (table 1).

Table 1. Sociodemographic characteristics of SAM under-5-year children admitted to Jigjiga public hospitals between 1 January 2020 and 31 December 2021.

	Frequency (n=366)	%
Age (months)
<6	129	35.2
6–24	204	55.7
25–59	33	9.0
Sex
Male	196	53.6
Female	170	46.4
Residence
Urban	266	72.7
Rural	100	27.3
Type of admission
New	346	94.5
Readmission	20	5.5
Source of referral
Spontaneous	314	85.8
Health centre	38	10.3
Hospital	14	3.8

SAM, severe acute malnutrition.

Comorbidities of severely malnourished under-5 children

Almost all (97%) admitted severely malnourished children had at least one comorbidity, of which diarrhoea accounted for 216 (59%), followed by fever (43.2%), pneumonia (42.1%), anaemia (21.9%), TB (14.2%) and shock (9%). Malaria and superficial infection were also reported but were less common.

Types of clinical malnutrition

The predominant form of malnutrition in this study was marasmus 312 (85.2%), where oedematous malnutrition accounted for only 5 (1.4%), and the rest 49 (13.4%) were kwashiorkor-marasmus. Out of 312 (85.2%) marasmic cases, 228 (73.1%) recovered, 12 (3.8%) died, 53 (17%) defaulted and 19 (6.1%) were non-responders. Of the 5 oedematous cases, 2 (40%) were cured, 2 (40%) died and 1 (20%) was a non-responder (figure 1).

Figure 1. Types of malnutrition among acute severely malnourished under-5 children admitted to therapeutic units of Jigjiga public hospitals.

Treatment outcome of SAM in accordance with type of malnutrition in under-59-month children

When outcome is glimpsed against type of malnutrition, marasmus dominated the desired outcome (recovered) where among the 17 children who died, 2 (40%) were kwashiorkor, 12 (3.8%) were marasmic and the remaining 3 (6.1%) were kwashiorkor-marasmus. In addition, marasmus had the highest chance of defaulting from treatment.

Medication provision and mineral supplementation

As far as medication provision is concerned, the majority, 350 (95.6%) of the admitted children took intravenous antibiotics, and more than half, 206 (56.3%) were given amoxicillin. The other therapeutic and supplementary drugs or minerals provided were acetaminophen 187 (51.1%), intravenous fluids 48 (13.1%), albendazole 25 (6.8%) and vitamin A 19 (5.2%).

Treatment outcome in reference to Sphere standards

From the 366 children whose records were reviewed, 257 (70.2%) recovered, 32 (8.7%) died, 58 (15.8 %) defaulted and 19 (5.2%) were found to be non-responders. The median length of stay in the hospitals was 7 days. These outcomes of admitted severely malnourished children in both facilities were not acceptable compared with Sphere standards except the length of hospital stay and death rate, where both were in an acceptable range (33)(table 2). The survival pattern of children dropped in the first weeks (figure 2).

Table 2. Outcome indicators of children under-5 years of age admitted to therapeutic units of Jigjiga public hospitals between 1 January 2020 and 31 December 2021 compared with Sphere standards (n=366).

Outcome indicators	JJUSHYCS hospital	Karamardha hospital	Both facility	Sphere standards
				Over all	Acceptable	Alarming
Recovery rate	65.4%	77%	70.2%	77.9%	75%	<50%
Death rate	5.6%	13.2%	8.7%	5.5%	10%	>10%
Defaulter rate	21%	8.6%	15.8%	12.3%	<10%	>25%

JJUSHYCS, Jigjiga University Sheik Hassen Yabare Comprehensive Specialised Hospital.

Figure 2. Survival function of severely malnourished under-5 children admitted to therapeutic units of Jigjiga public hospitals.

Factors associated with recovery of severely malnourished children

Bivariate analysis

After bivariate analysis was carried out, facility name, TB, pneumonia, shock, fever, NG tube insertion, intravenous fluid, paracetamol and amoxicillin administration were the associated factors with treatment outcome at p<0.25.

Multivariate analysis

Managing facility, tuberculosis (TB), pneumonia and NG tube insertion were found to be significantly associated with treatment outcomes at a p<0.05. Severely malnourished children who were managed at Jigjiga University Sheik Hassen Yabare Referral Hospital were 57% less likely to recover from SAM than those managed at Karamardha General Hospital (adjusted hazard ratio (AHR)=0.437, 95% CI: 0.286 to 0.600). Children who did not have TB were almost three times more likely to recover than their counterparts (AHR=2.862, 95% CI: 1.604 to 5.107), and those without pneumonia were also 1.5 times more likely to recover than those with a diagnosis of pneumonia (AHR=1.509, 95% CI: 1.146 to 1.989). Furthermore, children without NG tube insertion were about 1.5 times more likely to recover than their counterparts (AHR=1.472, 95% CI: 1.075 to 2.015) (table 3).

Table 3. Bivariate and multivariate analysis (Cox regression) of factors associated with treatment outcome of SAM among under-5-year-old children admitted to therapeutic units of Jigjiga public hospitals between 1 January 2020 and 31 December 2021 (n=366).

	Recovered		CHR (95% CI)	P value	AHR (95% CI)	P value
	Yes	No
Treatment facility
Karamardha	117	35	1		1	0.00
JJUSHYCS	140	74	0.44 (0.34 to 0.57)	0.00	0.437 (0.30 to 0.62)
Paracetamol
Yes	129	58	1	0.126	1	0.26
No	128	51	1.212 (0.948 to 1.549)		0.84 (0.62 to 1.133)
Pneumonia
Yes	90	64	1	0.00	1	0.003
No	167	45	1.69 (1.31 to 2.20)		1.509 (1.146 to 1.989)
Amoxicillin
Yes	139	67	1	0.002	1	0.36
No	118	42	0.681 (0.532 to 0.873)		1.175 (0.829 to 1.667)
Nasogastric tube
Yes	64	32	1	0.00	1	0.016
No	193	77	1.73 (1.29 to 2.31)		1.472 (1.075 to 2.015)
Shock
Yes	19	14	1	0.178	1	0.85
No	235	95	1.38 (0.864 to 2.205)		0.945 (0.524 to 1.706)
TB
Yes	15	37	1	0.00	1	0.00
No	242	72	2.87 (1.67 to 4.96)		2.862 (1.604 to 5.107)
Fever
Yes	98	60	1	0.001	1	0.936
No	159	49	1.534 (1.191 to 1.975)		1.013 (0.742 to 1.383)

AHR, adjusted hazard ratio; CHR, crude hazard ratio; CI, confidence interval; JJUSHYCS, Jigjiga University Sheik Hassen Yabare Comprehensive Specialised Hospital; SAM, severe acute malnutrition; TB, tuberculosis.

Discussion

This study found overall median length of stay, recovery, death, defaulted and non-responder rates of 7 days; 257 (70.2%); 32 (8.7%); 58 (15.8%) and 19 (5.2%), respectively. The median length of stay in this study is lower and acceptable according to the SPHERE standard and studies done in Bahir Dar and Axum.^{16 17} The recovery rate is 257 (70.2%), which does not meet the minimum Sphere standard that recommends >75% of recovery.¹⁸ Similarly, it is below one study done in Minia, Egypt, which shows 90% recovery and also lower than studies done in Hawasa and Enderta, Ethiopia, which showed 78% and 76.8%, respectively.^{13 19} This might be due to older and better primary healthcare in Egypt and differences in management setups. But the recovery rate of this study is higher than those from Sanglah General Hospital, Indonesia, and the University of Gondar Comprehensive Specialised Hospital, Ethiopia, which show 57.5% and 67.7%, respectively.^{9 20} The study’s findings suggest that secondary malnutrition, particularly in older participants, may hinder recovery in Sanglah General Hospital. On the other hand, the study’s recovery rate is in line with one study done in Hawassa, southern Ethiopia, which showed 69.4% of recovery.⁸

This study found a death rate of 32 (8.7%) which is in an acceptable range of SPHERE standard and also lower when compared with that of Sanglah general hospital and Gonder university comprehensive specialised hospital.^{9 20} Among the treatment centres, the death rate is higher at Karamardha hospital (13.2%), which could be a quality of care issue but needs further study. Where in Jigjiga University Sheik Hassan Yabare Referral Hospital, the defaulter rate is high (21%), which could be due to public perception issues of the teaching hospitals or lack of counselling of the caregivers.

As far as associated factors are concerned, treating facility, TB, pneumonia and NG tube insertion were found to be significantly associated with treatment outcome of this study. Children treated at Jigjiga University Sheik Hassan Yabare Referral Hospital were 57% less likely to recover than those treated at Karamardha Hospital (AHR=0.437, 95% CI: 0.307 to 0.623), which may be due to the fact that non-responder and defaulter rates were higher at JJUSHYCS Hospital and more complicated cases might be referred to JJUSHYCS, which in turn may increase the non-responder rate. This is consistent with a study done in Southern Ethiopia where the recovery rate was different among therapeutic feeding centres in which severely malnourished children treated at Fasha were 1.4 times more likely to recover than their counterparts treated at Karat therapeutic feeding centre.²¹ Coming to comorbidities, children diagnosed with TB are 2.8 (AHR=2.862, 95% CI: 1.604 to 5.107) times less likely to recover than their counterparts. TB is known to have a significant metabolic impact. Active TB can lead to increased energy expenditure due to the heightened body immune response, which in turn worsens malnutrition. The inflammatory pathways activated in TB infection can also divert essential nutrients away from recovery processes, slowing down healing and recovery, and a similar association has been reported in a study done in Hawassa, Ethiopia.⁸

The study also shows children with a diagnosis of pneumonia are 1.5 (AHR=1.509, 95% CI: 1.146 to 1.989) times less likely to recover than those without pneumonia, which may be due to the synergistic relationship of pneumonia and SAM. Similar findings were reported in studies carried out in Zambia and Yekatit hospital, Addis Ababa.^{22 23} On the other hand, the study shows children with NG tube insertion are 1.4 times less likely to recover than their counterparts (AHR=1.472, 95% CI: 1.075 to 2.015) which is in line with a similar study done in Gedeo zone, southern Ethiopia.²⁴ This might be explained by the fact that NG tube increases risk of infection and children who need tube insertion are those who are severely ill, unconscious, poorly feeding and have severe metabolic malnutrition.

The strength of this study stems from its cohort study design, which allows for a temporal relationship between associated factors and the outcome. This study can also depict the general picture of inpatient SAM treatment at the main hospitals in the eastern part of Ethiopia. However, its limitation lies in the fact that the data were secondary, collected retrospectively, and there was no control over the quality of measurements recorded during hospitalisation. It was also impossible to assess the socioeconomic characteristics of the parent/caretaker, which may also influence treatment outcome.

Conclusions

The death rate was acceptable; however, the recovery and defaulter rates deviated from the acceptable targets set by SPHERE standards. The significant predicting factors of treatment outcome were treating facility, TB, pneumonia and NG tube insertion. Strengthening routine immunisation and TB screening would reduce the acquisition of tuberculosis (TB) and pneumonia. Lastly, we suggest a prospective study to look at the reasons behind the differences in treatment outcomes across treatment facilities and the characteristics associated with non-compliance with inpatient SAM treatment, particularly the high defaulter rate.

Data availability statement

Data are available on reasonable request.