Time to Recovery and Its Predictors among Children 6–59 Months Admitted with Severe Acute Malnutrition to East Amhara Hospitals, Northeast Ethiopia: A Multicenter Prospective Cohort Study

Background Malnutrition has been among the most common public health problems in the world, especially in developing countries including Ethiopia. Even though the Ethiopian government launched stabilization centers in different hospitals, there are limited data on how long children will stay in treatment centers to recover from severe acute malnutrition. This study aimed to assess the time to recovery and its predictors among children 6–59 months with severe acute malnutrition admitted to public hospitals in East Amhara, Northeast Ethiopia. Methods Institution-based, prospective cohort study was conducted in seven public hospitals in East Amhara and a total of 341 children were included in the study. The results were determined by Kaplan–Meier procedure, log-rank test, and Cox-regression to predict the time to recovery and to identify the predictors of recovery time. Variables having P value ≤0.2 during binary analysis were entered into multivarable Cox proportional hazards regression analysis. P value <0.05 was considered statistically significant. Results The nutritional recovery rate was 6.9 per 100 person-days with a median nutritional recovery time of 11 days (an interquartile range of 6). The independent predictors like using NG tube for feeding (AHR = 0.44, 95% CI: 0.27–0.71), not entering phase 2 on day 10 (AHR = 0.19, 95% CI: 0.12–0.29), and being admitted to referral hospitals (AHR = 0.52 95% CI: 0.37–0.73) were associated with longer periods of nutritional recovery time. Conclusion Both the recovery rate and the recovery time were within the acceptable minimum standards. But, special attention has to be given to children who failed to enter phase 2 on day 10, for those who needed NG tube for feeding, and for those admitted to referral hospitals during inpatient management.


Background
Malnutrition is a state in which the physiological functioning of the child's body is impaired due to either undernutrition or overnutrition [1]. It typically develops when the growth velocity and brain development are especially high during 6 to 18 months of age [2] and measured by stunting, wasting, and underweight [1].
Malnutrition can be acute or chronic while acute malnutrition is further classified into severe acute malnutrition (SAM) and moderate acute malnutrition (MAM) [3]. SAM is defined as very low weight for height (below -3z scores of the median World Health Organization (WHO) growth standards, or below 70% of the median of the National Centre for Health Statistics standard) and by the presence of nutritional edema [4]. In children 6-59 months of age, a middle-upper arm circumference (MUAC) less than 11.5 cm is also indicative of SAM [5].
SAM affects nearly 20 million under-five children; most of them are from Africa and Southeast Asia [2]. It is a significant factor in approximately one-third of the nearly 8 million deaths in children who are under 5 years of age worldwide [6]. Lower-middle income countries account for 76% of wasted children [7]. Children who are treated as inpatients because of complicated SAM have a reported mortality rate of 10 to 40% [8].
Due to weakened immunity, children with SAM are susceptible to infections [9]. ese infections in turn result in reduced appetite and prevent the body's normal absorption of food which worsens malnutrition and further results in impaired growth and increased risk of death particularly when wasting is severe [5]. e risk of death among children with SAM is nine times higher than that for a healthy child [10]. Most deaths in children with acute malnutrition are related to infections such as malaria, diarrhea, pneumonia, tuberculosis (TB), and HIV/AIDS [11].
Health professionals unknowingly use practices that are highly dangerous for severely malnourished children [8,11].
To solve such problems, WHO has developed guidelines for managing SAM [12] which was adopted later on to our context [13] and defined that inpatient management should include stabilization, transition, and phase 2 until full recovery at home.
According to the sphere standards, the recovery time of children hospitalized in stabilization centers should be less than four weeks and it is alarming if it takes longer than 6 weeks [14]. Even though sphere standards recommend shorter recovery times, most of the studies in Ethiopia showed that the median time to recovery was high and ranges from 14 days to 8.7 weeks [15,16], respectively. But, children, due to various factors, do not get cured on time. By understanding this problem, the FMOH of Ethiopia established the National Nutrition Program (NNP) to improve the nutrition service delivery in health facilities [17]. Most of the studies were retrospective and also lack the most important variables such as the wealth and food security status of the families and institutional factors which could affect the recovery time of children with SAM [18,19]. erefore, this study tried to assess time to recovery and its predictors among children 6-59 months admitted with SAM prospectively.

Study Setting.
e study was conducted from February 28 to May 02, 2019, in East Amhara Hospitals. ere are a total of 34 hospitals in East Amhara, which provide services for an estimated population of 10 million. e study was conducted in seven public hospitals including Tefera Hailu Memorial Hospital of Sekota, Woldia Hospital, Mekaneselam Hospital, Hidar 11 hospitals, Dessie Referral Hospital, Kemise Hospital, and Debre Berhan Referral Hospital. e aim of the study was to determine the time to recovery and its predictors among 6-to 59-months children with severe acute malnutrition who were admitted to East Amhara Hospitals. Regarding the settings of the study, the study was conducted in pediatrics wards of the selected hospitals where children were admitted. e children were admitted in separate rooms during phase 1 of their treatment and they shared rooms with other children during the transition and phase 2 of their treatment.

Study
Design. An institution based prospective cohort study was conducted among children 6-59 months with SAM admitted to pediatrics wards of hospitals. e cases were followed for a maximum of 42 days. However, children who recovered earlier were only followed until recovery. Administration of the treatment was made by the frontline health workers, according to the national protocol without any direct intervention from the research team.

Study
Participants. All children 6-59 months of age who were admitted with severe acute malnutrition were included in the study. According to the national protocol, patients fulfilling the admission criteria are enrolled and given a daily F75 during phase 1 and F-100 during the transition and phase 2 of the treatment.

Sample Size and Sampling Procedure.
e sample size was calculated by using a sample size calculation for survival analysis in STATA 14 by taking the crude hazard ratio of 1.71, probability of withdrawals 0.4, probability of recovery rate 0.52 [18], alpha 0.05, and power of 80% which finally results in 325. By adding a 10% defaulter rate, the final sample size was 358. From 34 hospitals, 7 hospitals were selected by using a simple random sampling technique. e selected hospitals are Dessie Referral Hospital, Woldia Hospital, Kemise Hospital, Mekaneselam Hospital, Hidar 11 hospitals, Sekota Hospital, and Debre Berhan Referral Hospital. All children admitted to these hospitals were included in the study and those with congenital problems were excluded from the study.

Variables and Operational Definitions.
e dependent variable of this study is time to recovery, and recovery time is defined as the number of days it takes from admission until a child is discharged after being claimed recovered from SAM [18,20]. Recovery is reaching >80% of nutritional median weight for height and becoming free of acute infections within 42 days of treatment [20]. Death is a patient who died while he/she was being treated in the program in a facility [13], and defaulter is a SAM patient that was absent for two consecutive weighings [13]. A patient who could not meet the discharge criteria after six weeks of inpatient management is considered as a nonrespondent [13]. e independent variables considered in this study include sociodemographic variables: age and sex of the child, place of residence, age and sex of caretaker, marital status of caretaker, ethnicity, religion, educational status, and occupation; socioeconomic and household food security, routine medications, and complications: amoxicillin, vitamin A, ampicillin/gentamycin, folic acid, Albendazole/Mebendazole, pneumonia, diarrhea, HIV, TB, anemia, dermatosis, and CHF.

Data Collection Tools and Procedures.
For data collection, two types of questionnaires were developed from various studies. e first was to collect the daily follow-up characteristics of children and it was adopted from the SAM management protocol. On the other hand, a structured questionnaire was developed in English after reviewing relevant literature and translated to the local language (Amharic) and translated back to the English language to check for its consistency. is questionnaire included the sociodemographic, food security, and economic status of patients' families/immediate caretakers.
Data were collected by interviewing the parents of children with SAM and by doing a physical examination and collecting laboratory investigations. e admitted children were followed for changes in weight, edema, and clinical signs like vomiting, diarrhea, cough, and convulsions which were assessed on a daily basis. In addition to this, MUAC was taken three times a week. Anthropometric measurements were by using standardized techniques and equipment.
Data were collected by fourteen trained health professionals (Clinical Nurses), and one B.S. Nurse was assigned as a supervisor for each hospital. Since the study was an observational study, the research team did not involve in any aspect of the treatment of the children with SAM. In order to ensure confidentiality and avoid data loss, all questionnaires were stored in locked cabinets throughout the study and accessed only by the principal investigator.
To assure quality, data collectors and supervisors were trained prior to data collection for one day regarding technique, ethics of data collection, and data collection process which was carried out by the principal investigator. A 5% pretest was conducted in Kobo Primary Hospital before the actual data collection. Data collectors, at the end of the data collection session, checked the questionnaires for completeness.

Data Processing and Analysis.
Data were coded and entered using Epi-data Version 3.1 software. en, they were exported to STATA version 14 for further statistical analysis. Both descriptive and analytic analyses were executed. e patient follow-up characteristics were described in terms of mean (standard deviation) and median (interquartile range) for continuous data and frequency distribution for categorical data. Frequency tables, graphs, and cross-tabulation were used to present the findings of the study. e results were determined using the Kaplan-Meier procedure, logrank test, and Cox proportional hazard regression. e food security status of each child's family was assessed based on FANTA's Household Food Insecurity Access Scale (HFIAS) indicators [21]. After checking assumptions, Principal Component Analysis was performed to determine the socioeconomic status of the families of children with SAM. Based on the PCA result, the families of children with SAM were categorized into three groups, namely, poor, medium, and rich.
Recovery time from SAM was estimated using the Kaplan-Meier procedure. A log-rank test was also used to test if there was a significant difference in recovery time between different groups of predictor variables. ose who were died, transferred out, and defaulted were considered as censored. Variables having P value ≤0.2 during binary analysis were entered into the multivariable analysis. P value <0.05 was considered as statistically significant for time to recovery. e Cox proportional hazard assumptions were checked by using Schoenfeld residuals and log (−log) S (t) plots. Intracluster correlation coefficients were also checked for individual and group level random effects. Both crude hazard ratio (CHR) and adjusted hazard ratio (AHR) with 95% confidence interval (CI) were computed to show the strength of association and to identify the main predictors determining the nutritional recovery time.

Sociodemographic and Admission Characteristics.
Of the total of 341 study subjects admitted to the hospitals, 181 (52.9%) of children were males while 85% of children were in the age group of 6-23 months with a mean age of 15.15 months (SD � ± 8.63 months). Correspondingly, from the parents, most of them (99.7%) were between the ages of 20 and 49, and 314 (92%) of them were married (Table 1).

Socioeconomic and Household Food Security Status.
Of the total 341 families of children, 137 (40.18%) were poor, 111 (32.55%) were medium in socioeconomic status, and the rest 93 (27.27%) were rich. Regarding the household food security status of the families, 249 (73%) were food secure while the rest of them are food insecure. From families with food insecurity, 4 (1.17%), 80 (23.46%), and 8 (2.35%) were mildly, moderately, and severely food insecure.

Admission Characteristics.
From a total of 341 children admitted with SAM, 281 (82.4%) were admitted with a nonedematous form of malnutrition. During admission, around 294 (86%) of them have MUAC less than 11.5 cm while 175 (51.32%) have WFH below 70%.

Comorbidities/Complications and Routine Medications.
Pneumonia and diarrhea were the common comorbidities with a prevalence of 34.3% and 31.9% followed by anemia (18.8%). From routine medications, 307 (90%) of children with SAM received ampicillin/gentamycin. Likewise, from the study participants, 99 (29%) have taken amoxicillin while 31 (44.6%) from those who are eligible took deworming medications ( Table 2).

Treatment Outcomes of Children with Severe Acute
Malnutrition. From the total study participants, 74.49% of them were recovered from SAM while 5.28% of them died (Figure 1).

Follow-Up Characteristics.
From 58 children with edema, 12 (20.7%) failed to start to lose edema on day 4. On the other hand, from a total of 51 children, edema was present on day 10 in 10 (19.6%) of the cases. Eighty-seven children out of 341 (25.5%) failed to enter into phase 2 on day 10. In addition to this, from those children recovered from nonedematous SAM, 25.91% failed to gain more than 8 g/kg/day during phase 2.
irty-one (11.15%) of them failed to gain more than 5 g/kg/d for 3 successive days.

Incidence of Recovery.
e patients were followed for a minimum of 1 and a maximum of 42 days with 10 days [Q1, Q3, (9, 15)] median follow-up time. e overall nutritional recovery rate was 6.9 per 100 person-days (95 CI: 6.13-7.84) of observation. e incidence rate was found to be 6.8 (95% CI: 5.7, 8.0) and 7.13 (95% CI: 5.9-8.5) per 100 person-days in males and females, respectively. e death rate of children with SAM in this study was 2 per 100 person-days of observation.

Predictors of Time to Recovery.
e predictors of time to recovery were failing to enter phase 2 on day 10, being admitted to referral hospitals, and using NG tube for feeding. e recovery time was delayed by 48% (AHR = 0.52; 95% CI: 0.37, 0.73) among children admitted to referral hospitals as compared with children admitted to general hospitals. e rate of time to recovery among children who did not enter phase 2 on day 10 delayed recovery time by 81% from SAM than their counterparts (AHR = 0.19; 95% CI: 0.12, 0.29). Children who used NG tube for feeding were 56% less likely to recover in shorter days than their counterparts (AHR = 0.44; 95% CI: 0.27, 0.71) ( Table 3).

Discussion
is study tried to assess time to recovery and its associated factors among children 6-59 months with complicated, severe acute malnutrition admitted in public hospitals in East Amhara, Northeast Ethiopia. e nutritional recovery rate was 6.9 per 100 person-days with a median nutritional recovery time of 11 days (an interquartile range of 6). e independent predictors like using NG tube for feeding, not entering phase 2 on day 10, and being admitted to referral hospitals were associated with longer periods of nutritional recovery time.
is study revealed that the nutritional recovery rate is 74.49%, which was consistent with studies in Kenya [22] and South Wollo [23] and with the national minimum standards of the cure rate of 75% [24]. But, this recovery rate was better than findings from studies conducted in Ghana [25], Mekele [26], Gondar [27], Nekemte [28], and Bahir Dar [29] of  [20], Jimma [30], Wolaita [31], and Tigray [32]. is difference might be due to differences in settings, caseload, and severity of cases [33] and the availability of skilled and trained staff [34], socioeconomic status, and availability as well as the accessibility of therapeutic foods and medications [20]. e median nutritional recovery time which is 11 days in this study was similar to that in studies conducted in Debremarkos and Finoteselam [20], South Wollo [23], Gondar [27], Tigray region, and Northern Ethiopia [32] and the national minimum standards of the average length of stay [13]. However, it was longer than that in a study conducted in Zambia [35] and shorter when compared to studies conducted in Bahir Dar [18], Karat, and Fasha stabilization Centers in Southern Ethiopia [33] and Jimma [36]. ese disparities might arise from differences in treatment and caring practices, health care settings, and other socioeconomic factors among the study areas [34]. e time to recovery among children who did not enter phase 2 on day 10 delayed by 81% from SAM compared to their counterparts. is was supported by a study conducted in Bahir Dar [18]. e reason might be due to severity, treatment environment, failure to complete the multichart correctly, insufficient staff, poorly trained staff, inaccurate weighing machine, and food prepared or given incorrectly. Unless children enter into the transition phase or phase 2, they will not be given therapeutic foods (F-100 and plumy nut) which can promote weight gain. ose formulas are designed for patients to rapidly gain weight. erefore, children who enter phase 2 before the day of 10 will enjoy this advantage and recover early compared to their counterparts [12,17,18,32].
Children who used NG tube for feeding were 56% less likely to recover in shorter days than their counterparts. is was in line with a study conducted in Geode Zone, SNNPR, Ethiopia [37]. is may be due to the reason that the NG tube is required when the child is not able to feed because of altered consciousness, shock or serious infection, failure of appetite test, and complications like aspiration [38].
In this study, the recovery time was delayed by 48% among children admitted to referral hospitals as compared with children admitted to general hospitals, which is consistent with studies conducted in North Shoa, Karat, and Fasha and Sidama Zones of Southern Ethiopia [32,37,38]. is might be due to the reason that referral hospitals are referral centers for primary and general hospitals and cases at referral hospitals are more complicated with increased severity, increased number of patients, and settings [32,38].

Strengths and Limitations of the Study.
Since it is a prospective cohort study, it has tried to address many important variables, including the sociodemographic and socioeconomic aspects of the families and the health facilityrelated factors. Each datum was collected before the outcome is known which enhances reducing bias and establishing stronger temporal relationships since the outcome is recovery. Even though this is a prospective cohort study, it has failed to assess the perception of caregivers on SAM.

Conclusion
Generally, the recovery rate and the nutritional recovery time of children with SAM in the study area were within the recommended national standards. Additionally, it was  identified that being not able to enter phase 2 on day 10, using NG tube for therapeutic feeding, and being admitted to referral hospitals were associated with longer periods of recovery time. e Regional Health Bureau must strengthen and follow closely the routine activities of screening children for malnutrition, which can assist in early diagnosis and management of children with SAM. Basic and refreshment training should be provided for hospital staff and program coordinators which can assist in improving the quality of nutritional therapy and adherence to the national standard protocol of SAM management. Special attention should be given for children who need NG tube for feeding, children who failed to enter phase 2 on day 10, and children admitted to referral hospitals while providing care to a severely malnourished child. Future researchers are recommended to assess the perception of parents/caregivers on SAM. Data Availability e datasets used and/or analyzed for the current study are available from the corresponding author on reasonable request.

Ethical Approval
Ethical clearance was obtained from the Ethical Review Committee of the Institute of Public Health, College of Medicine and Health Science, University of Gondar. Permission to conduct the study was obtained from the Regional Health Bureau and respective hospitals prior to data collection.

Consent
Oral consent was obtained from the families of the study subjects.

Authors' Contributions
TKT developed the proposal; coordinated the data collection; carried out the data entry, statistical analysis, and interpretation of the study; and developed the manuscript. SM, MTH, and FGM conceived the study and participated in the revision of the proposal, statistical analysis, and developing the manuscript. All authors read and approved the final manuscript.