Strengths and limitations of the study
-
The study was carried out at multiple governmental hospitals, namely Tikur Anbessa Specialised Hospital, Zewditu Memorial Hospital, Yekatit 12 Hospital and Tirunesh Beijing Hospital, all of which have severe acute malnutrition stabilisation centres.
-
The inclusion of a 3-year record improves the ability to generalise the findings.
-
The study employed survival analysis to determine the time spent before the outcome and a Cox regression model to determine predictors of the desired outcome. This method of analysis allows us to identify factors (predictors) that influence the timing of the event of interest.
-
As a limitation, the study employed retrospective data follow-up, which may lead to the removal of incomplete data and result in selection bias. This selection bias can have several impacts, such as the overestimation or underestimation of results.
-
The study’s reliance on observational data raised the impact of unmeasured confounders on the results.
Background
Malnutrition is an energy and nutrient intake imbalance that adversely affects physical functions. It can result in undernutrition or overnutrition, with undernutrition further divided into acute, chronic and micronutrient deficiencies. In children, acute malnutrition is classified as moderate acute malnutrition, severe acute malnutrition (SAM) or global acute malnutrition.1–3
SAM is a severe form of acute malnutrition, characterised by indicators such as a weight for length Z-score (WHZ<−3), a mid-upper arm circumference (MUAC) less than 115 mm, bilateral pitting oedema or a combination of these symptoms.1 SAM can be classified as ‘complicated’ or ‘uncomplicated’, based on the presence of medical complications. Complicated cases require inpatient treatment following SAM management protocols in order to reduce mortality and aid in nutritional recovery and neurocognitive development.4 5
SAM, affecting 13.6 million children, contributes significantly to the 3.1 million malnutrition-related deaths among children under 5 years of age.4 5 Globally, SAM accounts for approximately 400 000 child deaths each year, with sub-Saharan Africa and Asia collectively bearing more than 80% of these fatalities.5 Ethiopia is the second-most malnourished country in sub-Saharan Africa and ranks among the top 20 nations worldwide with high malnutrition rates.6 The country experiences 3% of cases of SAM and a SAM-related mortality rate ranging from 6.67% to 28.67% among children aged under 5.7 8 A significant proportion of these fatalities occur within stabilisation centres in the country.9 10
Numerous studies have revealed several predictors of mortality among children aged under 5 with SAM. The age of the child and the residence area of the children have been found to contribute to this ongoing mortality.9–11 Additionally, the absence of oedema and a 1-cm increase in the percentage of weight for age (WFA) index have been found to decrease the risk of mortality in children aged under 5.12 13 On the contrary, several comorbidities, such as anaemia,10 12 14 tuberculosis (TB),8 congenital heart disease (CHD),15 HIV seropositivity,12 diarrhoea,14 altered body temperature (hypothermia)10 16 and sepsis16 were found to be significantly associated with SAM-related death among the population under study.
Moreover, the administration of intravenous fluids and antibiotics was found to be associated with an increased rate of mortality among children under 5 who were admitted for SAM.12 17 Similarly, children who were not recorded as receiving F-75 and F-100 therapeutic foods, vitamin A and folic acid supplementation demonstrated a higher risk of mortality from SAM compared with their counterparts who received these interventions.12 13 The number of admissions to a stabilisation centre and the history of vaccination were also found to be significantly associated with SAM-related mortality among children aged under 5.15 18
Although Ethiopia aims to decrease SAM-related mortality to under 5%, several healthcare facilities still experience rates exceeding 20%.19 Despite Ethiopia’s dedication to eliminating all forms of malnutrition by 2030, these high mortality rates persist due to inefficient management practices, which undermine progress towards sustainable development goals.20 Therefore, this study, focusing on Addis Ababa, the capital city of the country with multiple stabilisation centres and multiple governmental hospitals, aims to identify survival rates and predictors of mortality among children under five with SAM.
Methods
Study design, period and setting
A retrospective cohort study design was conducted from 1 April to 30 April 2023, using records of children under the age of 5 with SAM who were admitted to the stabilisation centres in Addis Ababa public hospitals from 1 January 2020 to 31 December 2022. Addis Ababa, the capital city of Ethiopia, had a population of approximately 4 million as of 2017, with current estimates surpassing 5 million.21 The city is divided into 11 subcities and houses various government health facilities, including 13 hospitals and 98 public health centres.22 Among the six hospitals under the administration of Addis Ababa city, four hospitals provide inpatient services for SAM. These four hospitals, namely, Tikur Anbessa Specialised Hospital, Zewditu Memorial Hospital, Yekatit 12 Hospital and Tirunesh Beijing Hospital, were deliberately selected based on the presence of a stabilisation centre and their length of providing services for children aged under-5.
Study variables
Dependent variables
-
Time to death.
Independent variables
-
Comorbidities: anaemia, pneumonia, HIV, TB, sepsis, age and CHD.
-
Sociodemographic variables: age, gender and residence.
-
Types of severe acute malnutrition: Kwashiorkor, Marasmus and Marasmic kwashiorkor.
-
Anthropometric baseline measurements: weight, height and MUAC.
-
Treatment: treatments, supplements and therapeutic feeding.
-
Admission types: new admission and readmission.
-
Immunisation status: fully Immunised, partially immunised and non-immunised.
Inclusion criteria and exclusion criteria
Children who had been admitted to the stabilisation centres of the selected governmental hospitals with SAM between three consecutive years of 1 January 2020 and 31 December 2022, were included in the study, while those with incomplete medical records were excluded.
Sample size determination
The sample size was determined using the double population proportion formula, employing the Kelsey formula within V.7 of the Epi Info computer programme. The calculations took into account a 95% CI, 80% power and an unexposed to exposed ratio of 1:2. A 10% allowance was made for incomplete records, and an important variable of the study was the presence of pneumonia (P1).23
Pneumonia was considered an important variable in the study because it played a crucial role in securing the maximum possible final sample size. The exposed group (P1) was defined as children who had pneumonia on admission, while the unexposed group (P2) consisted of children who were not diagnosed with pneumonia on admission. The expected case mortality rate (incidence) of SAM among children exposed to pneumonia and those unexposed to pneumonia was taken into account.
The specific values used for the important variable of pneumonia were a prevalence rate of 14.2% (P1) and an adjusted HR (AHR) of 2.43. Consequently, the final required sample size for the study was determined to be 435 individuals.
The calculated sample size was proportionally allocated to the four stabilisation centres for each year.
Sampling technique
After determining the proportional allocation of sample sizes to the stabilisation centres in each study site (Tikur Anbessa Spcialized Hospital, Zewditu Memorial Hospital, Yekatit 12 Hospital and Tirunesh Beijing Hospital) for each year, a sampling frame was created using the institutional list. The study units were then identified using a systematic random sampling procedure, with a calculated interval of four (k=4).
We used an interval of four because the proportion of available medical records was similar across all four hospitals. The selection process began by randomly choosing the first medical record from the sampling frame. Then, every fourth medical record was reviewed in chronological order over the course of three consecutive years. In cases where the fourth record was missing, the subsequent fourth record was reviewed until the required sample size was reached.
The systematic random sampling procedure was chosen to guarantee an equal opportunity for all medical records to be included in the study. As a result, this method ensured a representative sample across the four hospitals throughout the 3-year duration.
Data collection tool and procedure
The data were collected using a structured data abstraction checklist. The process checklist development involved reviewing medical records and creating a data collection form based on various sources, including the inpatient therapeutic feeding registration book, individual follow-up charts from the stabilisation unit, the SPHERE standard for SAM management, the Ethiopian management protocol for SAM24 25 and previous studies on SAM.11 13–15
Data collection was conducted by two nurses with a Bachelor of Science (BSc), working at the SAM stabilisation centre of the selected hospitals through structured checklist using the Open data kit collect app on the data collector’s smart phone or tablet.
Data quality assurance
All data collectors underwent a 1-day training session to learn how to properly complete the structured data collection checklist. To ensure the accuracy of our data, we conducted a pretest on 5% of the sample at Minilik-II Hospital. During the pretest, we compared the questionnaire with the current registration book to collect all relevant data.
Operational definitions
-
Event (death): Under-5 children who died while receiving treatment in the health facility.
-
Recovered: Under-5 children who have overcome medical problems and satisfied the requirements for discharge.
-
Censored Discharged against medical advice or transferred to other health institutions without knowing the outcome.
-
Defaulters: Those who stop receiving treatment before the child is recovered or lost and whose condition is unknown.
-
Comorbidity: Those children under the age of 5 who were admitted with/developed medical diseases in addition to SAM.
Data analysing process
For analysis purposes, all confirmed data were transferred to STATA V.17. Exploratory data analysis techniques, such as examining levels of missing values, identifying significant outliers, assessing multicollinearity, assessing normality and evaluating proportionality of hazards over time, were employed. There were no missing values in the dataset, but cases with incomplete data were excluded from the analysis. This decision was made to avoid introducing potential bias into the results, as including incomplete data could have compromised the validity of the findings. To verify the normality of the data, both graphic and statistical techniques, including the Kolmogorov-Smirnov test, were used. Additionally, residual plots, specifically the Cox-Snell residual plot, were graphed to assess the fitness and suitability of the Cox regression model to the data.
To find associations between dependent and independent variables, bivariate analysis was used. A life table was created to calculate the likelihood of mortality for both exposed and unexposed groups at 60 days. Bivariate analysis was done to identify associations between dependent and independent variables. HR, 95% CI and p value were used to assess the strength of the association and statistical significance. In order to find independent predictors of mortality, the final Cox regression analysis included factors significant at p<0.25 level in the bivariate analysis was done.
Ethical consideration
Ethical clearance was granted from the Addis Ababa University (AAU) College of Health Sciences, School of Nursing and Midwifery, under protocol number of 45/22/SNM, as well as the Addis Ababa City Administration Health Bureau (AACAHB). Additionally, the hospital’s board granted approval for the study. To uphold privacy and confidentiality, robust passwords were used for electronic devices, and data abstraction forms were kept anonymous.
Patient and public involvement
Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Results
Sociodemographic characteristics and anthropometric measurement
Of 435 traced medical cards, 422 (97%) were verified as complete. Among the participants, 218 (51.7%) were female, and 344 (81.5%) resided in urban areas. Among the enrolled under-5 children, 374 (88.6%) were under 24 months old. Age in months ranges from 1 month to 58 months, with a mean of 13.15 (±SD 10.72). A total of 58.8% of the participants had a MUAC measurement of less than 11.5 cm at the time of admission, while 24.9% were not eligible for MUAC measurement due to their age. The weight of the participants ranged from 1.7 to 15, with a mean of 5.58 (± SD 2.25) kg. Regarding their height, the mean height/length was 66.6 (±SD 11.68) cm. The majority of the children, 365 (86.5%), had a weight for height (WFH) Z score of less than −3, and 334 (79.1%) had a WFA Z score of less than −3 on admission. More than half of the children, 352 (83.4%), were admitted with marasmus (table 1).
Sociodemographic characteristics and anthropometric measurement of severe acute malnutrition children admitted to stabilisation centre from 1 January 2020 to 31 December 2022, in Addis Ababa, Ethiopia (n=422)
Comorbidities and treatments for the comorbidities
The most frequently observed diseases among the under-5 children were diarrhoeal disease 165 (39.1%), pneumonia 160 (37.9%) and fever 121 (28.7%) (table 2).
Comorbidities of severe acute malnutrition children admitted to stabilisation centre from 1 January 2020 to 31 December 2022, in Addis Ababa, Ethiopia (n=422)
Regarding the therapeutic interventions, supplemental intravenous antibiotics were the most prescribed drugs, 363 (86%), followed by F-75 297 (70.4%) and F-100 286 (67.8%). On the contrary, deworming 24 (5.7%) and vitamin A 42 (10%) were the least prescribed drugs (table 3).
Treatment provided for severe acute malnutrition children admitted to Stabilisation Centre from 1 January 2020 to 31 December 2022 in Addis Ababa, Ethiopia (n=422)
Survival status
In this study, the analysis time was recorded until its maximum capacity and subsequently censored. Among the 422 study subjects, a total of 44 individuals (10.4%) died, while 27 (6.4%) were censored. The remaining 351 subjects (83.2%) were discharged. From those censored, six (1.42%) were transferred to other health institutions, and the rest, 21 (4.97%), were defaulters.
422 children under the age of five were observed between 1 January2020 and 31 December 2022, for a maximum of 60 days and a minimum of 1 day. The overall time at point was 4250 days, with an incidence rate of 10.3 per 1000 person-days (95% CI: 7.71 to –13.9) with 8 (95% CI: 8.00 to −9.00) days of median length of stay in the hospital, the cumulative survival probability on the first, seventh, 14th and 21st days was 99.03%, 92.56%, 85.99% and 80.22%, respectively. The overall survival probability by the end of 59 days was 73.23%, with a SE of 0.07 (95% CI: 55.38 to 84.85) (figure 1).
Overall Kaplan-Meier survival estimate of severely malnutrition children under the age of 5, 1 January 2020 to 31 December 2022 in Addis Ababa, Ethiopia (n=422).
Estimation of the Kaplan-Meir median survival time
With the Kaplan-Meier survival estimate of time to death with various covariates, a significant mean survival time difference was observed among under-5 children with HIV, 19 days (95% CI:12.210 to 26.318) and non-reactive were 53 days (95%CI: 49.875 to 55.420) (online supplemental figure 1) and children with pneumonia had a 41-days mean survival time (95% CI: 34.623 to 47.804), whereas those without pneumonia had 53 days (95% CI: 49.883 to 56.756) (online supplemental figure 1). Furthermore, other predictors such as presence of shock, Down syndrome and global developmental delay were shown to have a mean survival time of 17 days, 25 days and 22 days among children with the disease, respectively.
Supplemental material
Predictors of mortality among severely malnourished children
To identify the significant predictors of mortality of severe acute malnutrition in this study, the variate Cox proportional hazard regression was run for every predictor and transferred those with a p<0.25 to the multivariate Cox proportional hazard regression. Accordingly, variables with p<0.05 at 95% CI were presented as they have significant association with the dependent variable.
From the comorbidities the study identified, children with HIV/AIDS had a two times higher mortality rate than HIV-negative under-5 children (AHR: 2.2, 95% CI: 1.00 to 1.65). Similarly, children with pneumonia had a two times higher mortality rate than their counterparts (AHR: 2.2, 95% CI: 1.085 to 2.275). Additionally, children with shock had almost four times higher mortality rate than those children without shock (AHR: 3.5, 95% CI: 1.451 to 8.321).
Regarding the treatment and the supplements provided during the hospital stay, in comparison to their counterpart, children who had received intravenous fluid had almost fourfold increased risk of dying than children who did not take intravenous fluid (AHR; 3.7, 95% CI: 1.525 to 8.743). Likewise, children whoo got F-75 had an 80% lower mortality rate than those who did not (AHR: 0.2, 95% CI: 0.062 to 0.651). Similarly, children who were fully immunised had an 80% lower mortality rate than those who had not (AHR: 0.2, 95% CI 0.088 to 0.583) (table 4)
Predictors of mortality among under five children admitted with SAM admitted to stabilisation centres from 1 January 2020 to 31 December 2022 in Addis Ababa, Ethiopia (n=422)
Discussion
In this research study conducted in Addis Ababa, Ethiopia, our main objective was to determine the survival status and factors influencing mortality among children under the age of 5 who were diagnosed with SAM. We monitored 422 children retrospectively during the study, with the follow-up period ranging from a minimum of 1 day to a maximum of 2 months. However, the final sample size of 422 participants rather than the planned 435 may impact the results by reducing the study’s ability to detect potential effects, and there is a possibility of bias in the effect estimates. Of the 422 children, death was occurred in 44 children (10.4%), with an incidence rate of 10.3 per 1000 person-days. Full vaccination, feeding practices (F75), intravenous fluid administration, presence of HIV, pneumonia and occurrence of shock were identified as significant predictors of mortality.
When comparing the recovery rate of the children in our study to the guidelines set by the SPHERE project, we found that it falls within an acceptable range, as it exceeded the minimum requirement of 75%. Additionally, our study’s recovery rate was either on par or higher when compared with two previous studies conducted in Ethiopia.9 10 14 It is worth considering that the observed differences in recovery rates between studies may be attributed to various factors. For instance, the hospitals involved in our study were referral centres, which may have had an impact on the medical supplies available and the expertise of the staff. Moreover, differences in sociodemographic factors and sample sizes between studies might also contribute to these variations.
Although the death rate was not in the alarm figure (>15%), it was slightly above the acceptable value and marginally greater than the SPHERE project’s acceptable figure (<10%) and other previous studies.13 17 24 However, this result was less than a couple of studies from Ethiopia.18 23 26 This discrepancy mainly occurred because of the study areas. Previous studies focused on one study area, a specific district. In this case, the study was conducted in Addis Ababa, the capital city, with multiple governmental hospitals for SAM treatment.
Unlike other previous studies,10 13 26 the current study evidenced a significant association between immunisation status and death rate. Children who had received all recommended vaccinations for their age had about an 80% lower mortality rate than those who had not, which is nearly equal to the findings of a previous study in Ethiopia, 84%.15 Unvaccinated children are more susceptible to contracting serious childhood diseases that can be readily prevented with immunisation. In particular, having SAM increased the chance of mortality from different infections like sepsis, measles, meningitis and tuberculosis.27
Death can result from fluid overload and sodium retention, which can later lead to heart failure and other complications.27 In light of this, our study indicated that the risk of mortality is around 3.6 times higher in children aged under 5 who had received intravenous fluid, which is approximately equal to the findings of previous studies whose AHRs were 3.24 and 3.2, respectively.14 26 At the same time, a study from Uganda revealed almost double the mortality risk that we had found, at 7.2.17
Compared with children who did not receive F-75 treatment, F-75 recipients had 80% lower mortality rate than their counterparts.14 23 This may be due to not following up the WHO and national SAM management protocol which can effectively reduce the mortality rate if implemented strictly. F-75 is given for SAM children to provide the body requirement of nutrients without overwhelming the bodies system, since they can’t tolerate too much protein and sodium. Thus, it decreases the risk of death by preventing fluid overload.5 28
This study revealed that being HIV infected increases the mortality rate more than two times than those who were not infected, which approximately similar to previous study from Ethiopia, 2.8 (26). In the contrary, our finding was much lower than the findings of another previous study from, Eastern Ethiopia which revealed a 11-fold increased risk of death among the HIV positive children.23 This may be due to malnutrition and HIV/AIDS are inextricably linked, with one fueling the other and raises the chance of occurrence of opportunistic diseases such as diarrhoea, TB pneumonia, kala-azar, meningitis and malaria.29
According to our finding, the presence of shock increases the risk of death by 3.5 times. This was in line with other previous studies which reported 3.8 (28) and 3.1 (32) times mortality risk among children with shock. Since severe diarrhoea, septicaemia and hypovolemia can result into shock and the managements are still controversial, and the fluid administration can result in precipitated heart failure and increase the mortality of the child.28 30 Moreover, the risk of death was more than two times higher when pneumonia is present. This is likely due to the synergetic effect of pneumonia and the inflammatory effect of pneumonia, which may aggravate the malnutrition condition by further putting the children in malnutrition-related disease and masking early clinical manifestation of pneumonia.31 32
This study, possesses a few limitations worth acknowledging. First, it relied on retrospective data follow-up, which may have led to the exclusion of incomplete medical records, thus introducing a potential selection bias. Additionally, since our study solely relied on observational data, there exists the possibility that our results could be influenced by unmeasured confounding variables. Nevertheless, these limitations serve as reminders for future research endeavours.
Conclusion
This study aimed to investigate survival status and factors contributing to mortality in children afflicted by SAM. Notably, death was occurred in 44 children (10.4%), with an incidence rate of 10.3 per 1000 person-days, slightly exceeding the acceptable threshold of 10%. Our comprehensive analysis revealed significant predictors of mortality, including immunisation status, the administration of therapeutic fluid (F-75) and intravenous infusion. Moreover, comorbidities and clinical markers such as shock, pneumonia and HIV reactivity substantially increased mortality risk in children aged under 5 with SAM. To mitigate this risk, a collaborative effort involving all stakeholders is crucial, emphasising the need to address these identified predictors and adhere to SAM management protocols.
Data availability statement
Data are available upon reasonable request. All the data used on this study will be available from the corresponding author up on reasonable request.
Ethics statements
Patient consent for publication
Acknowledgments
We would like to acknowledge stabilization centers from the four hospitals for the cooperation in providing medical records of the children under study.
This post was originally published on https://bmjopen.bmj.com