Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition.

BACKGROUND: Every year, over 4 million children are treated for severe acute malnutrition with varying program performance. This study sought to explore the predictors of time to recovery from and non-response to outpatient treatment of SAM.
METHODS: Children with weight-for-height z-score (WHZ) <-3 and/or mid-upper arm circumference (MUAC) <115 mm, without medical complications were enrolled in a trial (called MANGO) from outpatient clinics in Burkina Faso. Treatment included a weekly ration of ready-to-use therapeutic foods. Recovery was declared with WHZ ≥-2 and/or MUAC ≥125 mm, for two weeks without illness. Children not recovered by 16 weeks were considered as non-response to treatment. Predictors studied included admission characteristics, morbidity and compliance during treatment and household characteristics. Cox proportional hazard models were fitted and restricted mean time to recovery calculated. Logistic regression was used to analyse non-response to treatment.
RESULTS: Fifty-five percent of children recovered and mean time to recovery was eight weeks while 13% ended as non-response to treatment. Independent predictors of longer time to recovery or non-response included low age, being admitted with WHZ <-3, no illness nor anaemia at admission, illness episodes during treatment, skipped or missed visits, low maternal age and not practising open defecation. Eighty-four percent of children had at least one and 59% at least two illness episodes during treatment. This increased treatment duration by 1 to 4 weeks. Thirty-five percent of children missed at least one treatment visit. One missed visit predicted 3 weeks longer and two or more missed visits 5 weeks longer treatment duration.
CONCLUSIONS: Both longer time to recovery and higher non-response to treatment seem most strongly associated with illness episodes and missed visits during treatment. This indicates that prevention of illnesses would be key to shortening the treatment duration and that there is a need to seek ways to facilitate adherence.

Introduction

Severe acute malnutrition (SAM) is a condition that occurs when the food intake does not meet the nutritional requirements either as a consequence of poor intake or disease [1]. In children 6–59 months of age, SAM is diagnosed when a child presents with a weight-for-height z-score (WHZ) <-3, a mid-upper-arm circumference (MUAC) <125 mm or nutritional oedema [2]. While the overall prevalence of SAM is unknown, in 2020 it was estimated that 2% of all children below the age of 5 years presented a WHZ <-3 translating to more than 13.6 million children suffering from severe wasting at any time [3]. Children with SAM have a 11.6 increased risk of mortality compared to children with no nutritional deficits living in the same contexts [4].

Generally SAM arises in contexts with social, political and economic factors affecting food availability and where infections and inflammation are common [1]. This is also why no single intervention has been shown to reduce the incidence that requires a holistic package of interventions [5].

According to the World Health Organisation (WHO), children with SAM without medical complications are treated as outpatients with weekly check-up visits [6]. The treatment includes a systematic antibiotic regimen and ready-to-use therapeutic foods (RUTF), prescribed according to the weight of the child and continued until discharge [6]. RUTF are energy and nutrient dense pastes usually composed of peanut butter, milk powder, oil, sugar and a vitamin and mineral complex designed to fulfil the nutritional needs of children recovering from SAM [7]. Recovery is defined as having reached a WHZ ≥-2 for those admitted with a WHZ <-3 or a MUAC ≥125 mm for those admitted with a MUAC <115 mm for at least two weeks [6]. Children that never attain recovery within a maximum treatment time are called non-response to treatment [8].

In 2020, around 5 million children were treated for SAM globally [9]. With 13.6 million children suffering from severe wasting at any time [3] and applying of incidence correction factor of 3.5 [10] to account for all new cases arising in a year, this translates to 47.6 million episodes of severe wasting in a year. Thus, the current coverage of treatment is around 10%. This is when excluding the burden of MUAC cases which, if added, would translate to an even lower coverage of treatment. Such low coverage warrants reflection on how to improve it possibly by optimising and better targeting treatment in order to expand it to more cases.

Children treated for SAM in different contexts present with varying mean treatment time and proportion of non-response to treatment [11-15]. Longer time in treatment and high percent of non-response to treatment increase the cost of treatment and question the effectiveness of the current treatment protocol. Little is known about the factors influencing treatment duration and non-response to treatment. Understanding who requires longer treatment time and which children are at the greatest risk of non-response to treatment could help guide the optimisation of SAM treatment and explain differences in program performance observed between different contexts.

The current study seeks to explore the predictors of time to recovery from and non-response to treatment of SAM in a community based treatment setting. The study is based on data collected during a randomised controlled trial comparing the efficacy of a reduced dose of RUTF with a standard dose on the treatment outcomes of children with uncomplicated SAM managed in outpatient care. The trial showed no significant effect of the dose reduction on the weight gain velocity, length of stay in treatment, recovery percent or proportion of non-response to treatment [16].

Methods

Ethics

The MANGO study was performed in accordance with the principles in the Declaration of Helsinki. The research protocol was approved by the national ethics committee (Comité d’éthique pour la Recherche en Santé) and the clinical trials board (Direction Générale de la Pharmacie, du Médicament et des Laboratoires) of Burkina Faso and was registered at the IRSCTN registry http://www.isrctn.com/ISRCTN50039021. Caregivers of participating children gave their informed consent in a written form.

Study design and setting

This study is based on data collected prospectively as part of a randomized controlled trial (called MANGO), which compared the efficacy of a reduced RUTF dose to a standard RUTF dose in the management of SAM without medical complications in children 6–59 months of age in a non-inferiority design. The results from the randomized controlled trial have been published elsewhere [16-22]. The study recruited a total of 801 children which provides a good sample size for analysing predictors compared to previous studies looking at time to recovery with samples sizes starting from 200 and most around 400 children treated [23-34].

The trial was conducted in 10 health centres of the Fada N’Gourma health district in eastern Burkina Faso. Malaria was endemic with 69.3% of children in the region presenting a positive rapid test [35]. In 2016, the prevalence of severe wasting (WHZ <-3) and moderate wasting (WHZ between -3 and -2) was 2.4% and 8.6%, respectively [36]. Most households depend on small scale farming and livestock ownership [37] and 32% of the population lives more than 10 km away from nearest health post [38].

Study participants and treatment protocol

The participants included all children enrolled in the clinical trial for which the enrolment procedures have been described in detail elsewhere [16]. In short, participants were children presenting with WHZ <-3 and/or MUAC <115 mm but without medical complications, at the 10 participating health centres. Children with any grade of oedema or no appetite were referred to inpatient care.

Treatment followed the Burkina Faso national community-based management of acute malnutrition (CMAM) guidelines [39] in all aspects except the RUTF dose. Half of the children received a reduced dose from the third treatment week onwards: One sachet per day to children weighing <7 kg and two to children ≥7 kg, representing a reduction between 30 and 54% compared to standard dose depending on the weight category [16]. Co-morbidities diagnosed during SAM treatment were managed according to national protocol [39]. In case of medical complications, weight loss of over 5% at any point or stagnant weight defined as no more than 100g weight gained over 4 weeks, during treatment, children were referred to inpatient care, as per the Burkina national CMAM protocol [39]. Nutritional treatment was continued weekly until recovery.

Recovery was defined as having attained 1) a WHZ ≥-2 for those admitted with a WHZ <-3, or 2) a MUAC ≥125 mm for those admitted with a MUAC <115 mm, or 3) a WHZ ≥-2 and a MUAC ≥125 mm for those admitted with both WHZ<-3 and MUAC <115 mm, for 2 consecutive weeks and absence of illness. Children still not having reached the anthropometric recovery criteria by 16 weeks of treatment were declared as non-response to treatment. Other discharge categories included defaulting (defined as having missed 3 consecutive visits and confirmed alive), loss-to follow up (defined as having missed 3 consecutive visits and not confirmed alive), death, and false discharge (those discharged wrongly after verification of discharge anthropometrics).

Data collection

Upon admission, the child’s caregiver was interviewed regarding household socio-economic characteristics, care practices and recent morbidity of the child. Anthropometric measurements and a clinical examination were performed at each visit from admission to discharge. Weight was measured using an electronic scale (SECA 876) to the nearest 100 g, height using a wooden measuring board (locally made) to the nearest 1 mm, and MUAC using a non-stretchable colourless measuring tape to the nearest 1 mm. Z-scores were calculated using the WHO standards and STATA command zscore06 [40]. Haemoglobin (HemoCue) was measured at admission. All data were collected via tablets using the Open Data Kit (ODK1 software) and continuous data monitoring and cleaning was performed. Data monitoring included among other thing, checking duplicate entries and outliers, anthropometric decimal distributions and correct prescription of medication according to diagnosed conditions. Any potential data problem resulted in action. Data cleaning was based on double checks of electronic data against patient registries or therapeutic cards.

Outcomes

Two outcomes were studied: 1) time to recovery defined as days passed from admission to treatment until discharge as recovered, and 2) non-response to treatment defined as not reaching the anthropometric recovery criteria within 16 weeks. Recovery and non-response were dichotomised to recovered or non-recovered and response or non-response, respectively. For the study of time to recovery, non-recovered cases were right censored contributing to the analysis of time to recovery until exit from study. Patients referred to inpatient care were excluded from the analysis in order to limit potential bias that could be introduced due to their short length of stay in treatment.

Predictors

Predictors included in the analyses were variables describing 1) admission characteristics, 2) morbidity and compliance to treatment visits during treatment and 3) household characteristics. Admission characteristics included sex, age, WHZ, MUAC, height for-age z-score (HAZ), admission criteria (WHZ <-3, MUAC <115 mm or both), any illness, anaemia, breastfeeding status and low birth weight. Illness at admission was defined as any caregiver reported illness (including cough, diarrhoea, fever, vomiting, skin lesions) observed in the week prior to admission or diagnosed by study nurse upon admission. Anaemia was defined as a haemoglobin level <110 g/l [41]. Low birth weight (<2500 g) was confirmed from an official birth certificate or health card. Morbidity and compliance variables included an episode of malaria, acute respiratory illness (ARI) or diarrhoea during treatment and number of illness episodes as well as number of skipped and missed visits. Malaria episode during treatment was defined as an armpit temperature of >37.5°C, a positive malaria rapid diagnostic test (RDT) and a negative RDT at admission. ARI was defined as cough reported by caregiver in the past week or diagnosed by study nurse during visit. Diarrhoea included acute, persistent or dysenteric forms and was defined as 3 or more loose stools per day as reported by caregiver in the past week or diagnosed by study nurse. Illness episodes included any caregiver reported or nurse diagnosed illness in the past week. Skipped visits were those that were planned in advance and thus the caregiver was prescribed double dose of RUTF to cover 2 weeks of home treatment. Missed visits were unplanned and thus represent gaps in RUTF prescription. Household characteristics included caregiver’s age, education level, whether caregiver was the household head, number of children under 5 years of age in the household, water source, open defecation (the practice of defecating in nature instead of a toilet facility) or not, food security status, distance from health centre and urban or rural setting. Safe water source was defined as a protected well, pump or tap while unsafe water source included unprotected wells and rivers, lakes and ponds. Food security assessment was based on the Household Food Insecurity Access Scale (HFIAS) [42]. Distance from health centre was estimated by the caregivers as the time needed for a return trip with the available transportation means.

Data analysis

All children included in the MANGO trial regardless of their treatment arm (reduced RUTF or standard RUTF) were included in the analysis. Baseline characteristics of the study population are summarized as percent (n) or mean ±SD. Cox proportional hazard regression models were fitted to quantify effects of predictors on time to recovery with resulting hazard ratios describing the increased or decreased chance of recovery. Kaplan Meier plots were used to visualize results concerning age and admission categories. Restricted mean time to recovery was estimated for significant categorical predictors. Logistic regression was used to evaluate predictors of non-response to treatment. Both unadjusted and models adjusted for sex and age at admission were fitted. A p-value below 0.05 was used to declare statistical significance. All analyses were performed using STATA 15 (StataCorp, USA) and Kaplan Meier plots were produced in GraphPad Prism 8 (GraphPad Software, USA).

Results

In total, 801 children were enrolled in the RCT, of which 54% (n = 433) recovered with a median [IQR] time to recovery of 8 weeks [5-12]. Thirteen percent (n = 101) were considered non-response to treatment, 20% (n = 157) were referred to inpatient care due to stagnant weight, weight loss or medical complication, 10% (n = 83) defaulted, 3% (n = 24) were falsely discharged, 2 children died and 1 was lost to follow up. Excluding the 157 referrals, 644 children contributed to the analysis of time to recovery and all 801 to the analysis of non-response to treatment.

At admission, children were on average 13 months old, 86% were breastfed, 79% had an illness, and 80% had anaemia (Table 1). Most households (83%) had access to a safe drinking water source but 79% were practising open defecation. During treatment, 84% of children had at least one illness episode with 59% reporting at least two episodes. Up to 35% of children missed at least one treatment visit with 16% missing more than one visit.

Table 1

Characteristics of children included in the analysis of time to recovery.

Characteristics	Values
1. Admission characteristics
Male, % (n)	47 (305)
Age, months	13.4 ± 8.4
WHZ at admission	-3.0 ± 0.7
MUAC at admission, mm	113.1 ± 6.2
HAZ at admission	-2.4 ± 1.3
Admission criteria, % (n)
WHZ only	27 (171)
MUAC only	39 (252)
both WHZ and MUAC	34 (221)
Illness, % (n)	79 (507)
Anaemia, % (n)	80 (514)
Low birth weight, % (n)1	20 (84)
Breastfeeding, % (n)	86 (552)
2. Morbidity and compliance during treatment
Malaria episode, % (n)	16 (105)
ARI episode, % (n)	47 (303)
Diarrhoea episode, % (n)	27 (177)
Number of illness episodes, % (n)
none	16 (102)
one	25 (163)
two	18 (113)
three or more	41 (266)
Number of skipped* visits, % (n)
none	48 (309)
one	25 (163)
two or more	27 (172)
Number of missed* visits, % (n)
none	65 (418)
one	19 (120)
two or more	16 (106)
3. Household characteristics
Maternal age, years	27.9 ± 7.7
Mother has some formal education, % (n)	24 (154)
Caregiver is the household head, % (n)	3 (21)
Number of children under 5 in the household, % (n) 1
only index child	31 (201)
two	34 (220)
three or more	34 (218)
Using safe water source, % (n)	83 (533)
Open defecation, % (n)	77 (496)
Household is food secure, % (n)	89 (575)
>30 min return distance from health centre, % (n)	63 (407)
Urban setting, % (n)	14 (89)

Values are mean ± SD unless otherwise indicated

1 Birth weight was only available for 410 children and number of children in the household for 639 of all 644 children included in the time to recovery analysis

* skipped visits refer to those that were planned and thus benefitted from double RUTF prescription as opposed to missed visits that were unplanned

ARI, acute respiratory infection; HAZ, height-for-age z-score; MUAC, mid-upper arm circumference; WHZ, weight-for-height z-score.

Unadjusted and adjusted estimates of associations with time to recovery are presented in Table 2. When adjusted for sex and age at admission, independent predictors of longer time to recovery included low age, low WHZ at admission, being admitted with WHZ <-3 (compared to only MUAC <115mm), not having illness or anaemia at admission, having any co-morbidity episode during treatment, higher number of missed treatment visits, low maternal age and not practising open defecation. For example, for every 1 z-score increase in WHZ at admission, children have 26% higher chance of recovery.

Table 2

Predictors of time to recovery (days) among 644 children during outpatient treatment of severe acute malnutrition.

Predictors	n		Unadjusted			Sex and age adjusted2
	Event	Censored	HR1	95% CI	p-value	HR1	95% CI	p-value
1.Admission characteristics
Sex
Male	209	96	Ref			Ref
Female	224	115	0.86	0.71; 1.04	0.12	0.88	0.73; 1.07	0.20
Age, months	433	211	1.01	1.00; 1.03	0.004	1.01	1.00; 1.02	0.007
WHZ	433	211	1.12	0.98; 1.27	0.10	1.26	1.09; 1.46	0.002
MUAC, mm	433	211	1.01	1.00; 1.03	0.15	1.00	0.99; 1.02	0.85
HAZ	433	211	0.93	0.87; 1.00	0.055	0.95	0.88; 1.02	0.16
Admission criteria
WHZ only	102	69	Ref			Ref
MUAC only	185	67	1.31	1.03; 1.67	0.028	1.80	1.36; 2.38	<0.001
both WHZ & MUAC	146	75	1.05	0.82; 1.36	0.68	1.26	0.97; 1.65	0.086
Any illness
No	78	59	Ref			Ref
Yes	355	152	1.40	1.09; 1.79	0.008	1.35	1.05; 1.73	0.017
Anaemia
No	78	52	Ref			Ref
Yes	355	159	1.50	1.18; 1.92	0.001	1.60	1.25; 2.06	<0.001
Low birth weight
No	222	104	Ref			Ref
Yes	54	30	0.81	0.60; 1.09	0.16	0.80	0.60; 1.08	0.15
Breastfeeding
No	66	26	Ref			Ref
Yes	367	185	0.64	0.49; 0.84	0.001	0.70	0.48; 1.01	0.057
2.Morbidity and compliance during treatment
Malaria episode
No	382	157	Ref			Ref
Yes	51	54	0.42	0.32; 0.57	<0.001	0.43	0.32; 0.58	<0.001
ARI episode
No	248	93	Ref			Ref
Yes	185	118	0.44	0.37; 0.54	<0.001	0.45	0.37; 0.55	<0.001
Diarrhoea episode
No	342	125	Ref			Ref
Yes	91	86	0.43			0.44	0.35; 0.56	<0.001
Number of illness episodes	433	211	0.57	0.53; 0.61	<0.001	0.57	0.53; 0.61	<0.001
Number of skipped* visits	433	211	0.67	0.61; 0.73	<0.001	0.66	0.60; 0.73	<0.001
Number of missed* visits	433	211	0.60	0.53; 0.67	<0.001	0.59	0.52; 0.67	<0.001
3.Household characteristics
Maternal age, years	433	211	1.02	1.00; 1.03	0.029	1.01	1.00; 1.03	0.048
Maternal education
No formal education	333	157	Ref			Ref
Some formal education	100	54	0.84	0.67; 1.05	0.13	0.84	0.67; 1.05	0.13
Number of children under 5 in the household	432	207	1.04	0.97; 1.11	0.31	1.05	0.98; 1.12	0.16
Caregiver is the household head
No	416	207	Ref			Ref
Yes	17	4	1.51	0.93; 2.45	0.098	1.36	0.83; 2.23	0.22
Safe water source
No	62	49	Ref			Ref
Yes	371	162	1.27	0.97; 1.66	0.082	1.24	0.95; 1.62	0.12
Open defecation
No	90	58	Ref			Ref
Yes	343	153	1.30	1.03; 1.65	0.025	1.32	1.04; 1.67	0.021
Food insecure household
No	383	192	Ref			Ref
Yes	50	19	0.92	0.69; 1.24	0.59	0.86	0.64; 1.16	0.33
Return time from health centre
≤30 min	156	81	Ref			Ref
>30 min	277	130	0.97	0.80; 1.19	0.80	1.01	0.83; 1.23	0.93
Setting
Rural	375	180	Ref			Ref
Urban	58	31	0.92	0.70; 1.22	0.58	0.89	0.67; 1.17	0.40

1 HR>1 means faster recovery

2 when analysing sex as a predictor, only age was included as adjustment and when analysing age as a predictor only sex was included as adjustment

*skipped visits refer to those that were planned and thus benefitted from double RUTF prescription as opposed to missed visits that were unplanned

ARI, acute respiratory infection; HR, Hazard Ratio; MUAC, mid-upper arm circumference; WHZ, weight-for-height z-score.

The mean time to recovery for significant categorical variables are presented in Table 3. Children <12 months of age required 13 more days to reach recovery compared to children ≥12 months of age (Fig 1). Being admitted with only MUAC criteria and thus a WHZ ≥-3 predicted faster recovery compared to those admitted with WHZ only (Fig 2) with an average of 15 days shorter time to recovery compared to WHZ only or 10 days shorter compared to those with both admission criteria (Table 3). Illness and malaria at admission were associated with 6 and 14 days faster recovery, respectively. On the contrary, having one, two or more than two illness episodes during treatment was associated with 1, 4 and 8 weeks longer time to recovery, respectively, compared to those with no illness episodes during treatment. One skipped visit increased time to recovery by 2 weeks and one missed visit by 3 weeks. Open defecation was associated with 1 week faster recovery. Including children referred to inpatient care in the analysis of different factors with time to recovery resulted in similar associations.

Table 3

Estimated restricted mean time (days) to recovery from SAM by significant predictor characteristics.

Predictors	mean (days)	Unadjusted			Sex and age adjusted2
		Difference	95%CI	p-value	Difference	95%CI	p-value
1.Admission characteristics
Age
<12 months	75.7
≥12 months	62.3	-13.5	-18.9; -8.1	<0.001	-13.4	-18.8; -8.0	<0.001
Admission criteria
WHZ only	73.8
MUAC only	66.0	-7.8	-14.8; -0.9	0.027	-15.1	-22.6; -7.6	<0.001
WHZ & MUAC	73.5	-0.2	-7.4; 6.9	0.95	-4.8	-12.1; 2.5	0.20
Any illness
No	76.5
Yes	68.9	-7.6	-14.4; -0.9	0.026	-6.2	-12.9; 0.6	0.075
Anaemia
No	81.4
Yes	67.7	-13.7	-20.1; -7.4	<0.001	-14.5	-21.1; -7.9	<0.001
Breastfeeding
No	57.8
yes	72.6	14.7	7.1; 22.4	<0.001	8.9	-1.3; 19.1	0.086
2.Morbidity and compliance during treatment
Malaria episode
no	66.2
yes	91.5	25.3	19.2; 31.4	<0.001	24.2	18.0; 30.3	<0.001
ARI episode
no	58.0
yes	83.0	24.9	20.0; 29.9	<0.001	24.0	19.0; 29.1	<0.001
Diarrhoea episode
no	63.6
yes	87.9	24.4	19.0; 29.7	<0.001	23.0	17.5; 28.4	<0.001
Number of illness episodes
none	39.2
one	46.6	7.4	1.6; 13.1	0.012	6.9	1.2; 12.6	0.017
two	66.5	27.3	21.0; 33.7	<0.001	26.9	20.6; 33.2	<0.001
more than two	92.2	53.0	47.8; 58.2	<0.001	52.7	47.4; 58.0	<0.001
Number of skipped1 visits
none	56.1
one	71.4	15.3	9.1; 21.5	<0.001	16.2	10.2; 22.2	<0.001
more than one	91.1	35.0	29.6; 40.3	<0.001	35.6	30.2; 41.0	<0.001
Number of missed1 visits
none	59.1
one	83.0	23.9	17.8; 30.1	<0.001	23.2	16.8; 29.6	<0.001
more than one	98.7	39.6	34.5; 44.7	<0.001	39.4	34.1; 44.6	<0.001
3.Household characteristics
Open defecation
no	75.7
yes	68.9	-6.7	-13.2; -0.2	0.042	-7.0	-13.5; -0.5	0.036

1 skipped visits refer to those that were planned and thus benefitted from double RUTF prescription as opposed to missed visits that were unplanned

2 when analysing age categories as a predictor only sex was included as adjustment

ARI, acute respiratory infection; MUAC, mid-upper arm circumference; OR, odds ratio; WHZ, weight for height z-score.

Fig 1

Kaplan Meier plot of cumulative recovery from SAM by age category during outpatient treatment.

Fig 2

Kaplan Meier plot of cumulative recovery from SAM by admission criteria during outpatient treatment.

Factors associated with non-response to treatment were largely similar to those associated with time to recovery in that factors predicting a slower recovery also predicted non-response to treatment and factors predicting faster recovery also predicted not ending up non-response (see Table 4).

Table 4

Predictors of non-response to outpatient treatment of SAM in 801 patients without medical complications at admission.

Predictors	n		Unadjusted			Sex and age adjusted2
	Non-response	Response	OR	95% CI	p-value	OR	95% CI	p-value
1.Admission characteristics
Sex
Male	41	355	Ref			Ref
Female	60	345	1.51	0.99; 2.30	0.058	1.46	0.95; 2.24	0.082
Age	101	700	0.97	0.94; 1.00	0.039	0.97	0.94; 1.00	0.051
WHZ	101	700	0.96	0.72; 1.30	0.80	0.80	0.58; 1.09	0.16
MUAC	101	700	0.98	0.95; 1.01	0.30	1.00	0.96; 1.04	0.98
HAZ	101	700	1.11	0.94; 1.30	0.22	1.05	0.89; 1.24	0.58
Admission criteria
WHZ only	30	179	Ref			Ref
MUAC only	33	276	0.71	0.42; 1.21	0.21	0.41	0.22; 0.75	0.004
WHZ & MUAC	38	245	0.93	0.55; 1.55	0.77	0.66	0.38; 1.14	0.14
Any illness
no	31	143	Ref			Ref
yes	70	557	0.58	0.37; 0.92	0.020	0.61	0.38; 0.97	0.038
Anaemia
no	29	144	Ref			Ref
yes	72	556	0.64	0.40; 1.03	0.065	0.63	0.39; 1.01	0.057
Low birth weight
no	54	338	Ref			Ref
yes	19	91	1.31	0.74; 2.31	0.34	1.25	0.70; 2.23	0.44
Breastfeeding
no	9	103	Ref			Ref
yes	92	597	1.76	0.86; 3.61	0.12	1.19	0.48; 2.94	0.71
2.Morbidity and compliance during treatment
Malaria episode
no	65	603	Ref			Ref
yes	36	97	3.44	2.17; 5.46	<0.001	3.41	2.14; 5.42	<0.001
ARI episode
no	23	380	Ref			Ref
yes	78	320	4.03	2.47; 6.56	<0.001	3.81	2.33; 6.23	<0.001
Diarrhoea episode
no	50	507	Ref			Ref
yes	51	193	2.68	1.75; 4.09	<0.001	2.57	1.68; 3.95	<0.001
Number of illness episodes	101	700	1.95	1.72; 2.21	<0.001	1.95	1.72; 2.21	<0.001
Number of skipped1 visits	101	700	1.95	1.65; 2.29	<0.001	1.94	1.65; 2.29	<0.001
Number of missed1 visits	101	700	2.01	1.69; 2.38	<0.001	2.04	1.71; 2.42	<0.001
3.Household characteristics
Maternal age	101	700	1.00	0.97; 1.02	0.74	1.00	0.97; 1.03	0.92
Mother has received some formal education
no	70	536	Ref			Ref
yes	31	164	1.45	0.92; 2.29	0.113	1.45	0.91; 2.29	0.12
Number of children under 5 in the household	101	700	0.86	0.72; 1.03	0.103	0.83	0.69; 1.00	0.054
Safe water source
no	18	122	Ref			Ref
yes	83	578	0.97	0.56; 1.68	0.92	1.06	0.61; 1.84	0.83
Open defecation
no	34	158	Ref			Ref
yes	67	542	0.57	0.37; 0.90	0.016	0.58	0.37; 0.91	0.018
Food insecure household
no	90	613	Ref			Ref
yes	11	87	0.86	0.44; 1.67	0.66	0.97	0.50; 1.91	0.94
Return time from health centre
≤30min	41	263	Ref			Ref
>30min	60	437	0.88	0.58; 1.35	0.56	0.82	0.53; 1.27	0.37
Setting
rural	80	608	Ref			Ref
urban	21	92	1.73	1.02; 2.94	0.041	1.93	1.13; 3.31	0.017

1skipped visits refer to those that were planned and thus benefitted from double RUTF prescription as opposed to missed visits that were unplanned

2 when analysing sex as a predictor, only age was included as adjustment and when analysing age as a predictor only sex was included as adjustment

ARI, acute respiratory infection; MUAC, mid-upper arm circumference; OR, odds ratio; WHZ, weight-for-height z-score.

Discussion

In this study, only 55% of children admitted to outpatient treatment of SAM recovered while 13% were discharged non-response to treatment. The mean time to recovery was 8 weeks and was most strongly associated with illness episodes and missed and skipped visits during treatment. Up to 59% of children had at least two illness episodes during treatment which increased the treatment duration by nearly 4 weeks with two episodes and over 7 weeks with more than two episodes. Up to 35% of children missed at least one visit which increased the time to recovery by 3 weeks with one and over 5 weeks with two or more missed visits. Similarly, non-response to treatment seemed most strongly associated with illness episodes and missed visits during treatment increasing the odds of non-response to 1.95 for each additional illness episode and 2.04 for each additional missed visit.

The mean time to recovery was longer than reported by most previous studies conducted in outpatient settings [14, 15, 27, 31, 32, 34, 43, 44]. Most of the variability in length of stay in treatment between different studies and contexts can probably be ascribed to differences in discharge criteria: in many studies recovery was declared when children reached a weight-for height >85% of WHO median growth reference regardless of being admitted with low WHZ or MUAC [14, 24–26, 30–32, 43, 45]. Additionally, often the recommendation [6] of presenting the anthropometric recovery criteria for at least 2 weeks is not followed [14, 24–26, 30, 44]. These deviations from the WHO issued protocol have consequences on the program performance indicators [46].

Illness episodes during treatment were the strongest predictor of recovery and non-response. Few outpatient studies have looked at co-morbidities occurring during treatment. Yebyo et al. (2013) found a significant association between co-morbidities and lower recovery rate but did not distinguish between illnesses diagnosed at admission or during treatment [14]. The relatively long treatment duration in the current study probably contributed to the observation that illnesses diagnosed during treatment were more predictive of recovery than those diagnosed at admission. Children were systematically treated for the diagnosed illnesses and the observation that these episodes were still associated with remarkably longer treatment time is noteworthy. It highlights the importance of preventing the conditions as it seems that once children catch diarrhoea, malaria or ARI during SAM treatment, the illnesses substantially slow their recovery trajectory.

One missed visit increased the time to recovery by more than 3 weeks. The reason for missing a visit was in most cases related to time or accessibility impediments to travel to the health centre to receive treatment. This calls for seeking more flexible and possibly more closely available services; flexible in the sense of providing services on several week days instead of just one and more closely available potentially in the form of community health worker delivered treatment [47]. Interestingly, even when children had benefitted from double RUTF prescription, one skipped visit still increased the time to recovery by more than 2 weeks. This suggests that increased visit spacing could result in longer treatment time. This said, part of the increased time (1 week) can be explained by missed and skipped visits occurring towards the end of treatment artificially lengthening treatment time when recovery can only be declared upon reaching discharge criteria during two consecutive visits.

We observed that higher age was associated with faster recovery. Previous studies in outpatient settings have found positive [27, 34], negative [32] and no [14, 31, 33, 43] associations between age and recovery. This could partly be due to the use of different cut-offs for defining age groups ranging from 12 months to 36 months. Worth noting though, in our study only 10% of children were ≥24 months old at admission, meaning that in a context with an older SAM population, the association could be different.

Being admitted based on MUAC only (ie. MUAC <115 mm and WHZ ≥-3) was associated with 2 weeks quicker recovery and 0.41 the odds of non-response than being admitted with WHZ only (ie. WHZ <-3 and MUAC ≥115 mm). This indicates either that children presenting with a low WHZ potentially have a different pathophysiological status that requires a longer treatment time or that reaching a WHZ ≥-2 takes longer than reaching a MUAC ≥125 mm. This observation is to be considered when observing success rates from programs implementing MUAC only admission criteria as based on the current study excluding children with WHZ <-3 but MUAC > 115mm would result in better recovery rate.

Illness at admission were associated with faster recovery, contrary to results from most previous studies showing slower recovery in inpatient [24, 25, 28–30, 45, 48] and outpatient settings [14] among those admitted with co-morbidity. Interestingly however, one study from Gambia showed that higher cortisol at admission (indicating acute illness) predicted higher WAZ gain during treatment [49]. The authors suspected that this was due to children with high cortisol being more sick and when treated for their condition responding faster by also gaining weight. In somewhat similar lines, another study found that children that failed the appetite test at admission, possibly indicating sub-clinical illness, had higher weight gains during treatment compared to children having passed the appetite test [50]. Our observation offers support to this hypothesis in that it would seem that children with co-morbidities respond fastest to the treatment. It could be that their malnutrition is a secondary condition related to a primary co-morbidity that when managed correctly allows a rapid return to a normal health and nutrition status. Consequently, children with no apparent co-morbidities at admission possibly have a different causal pathway to malnutrition and maybe a different pathophysiological state leading to slower response to treatment. It is also possible that they present with some underlying chronic hard-to-detect pathology such as environmental enteropathy [51-53] or congenital heart diseases [54] that increase the nutrient needs and could affect the time to recovery.

Anaemia at admission was associated with faster recovery. This is in contrast to previous studies in inpatient settings reporting slower recovery rates among children with anaemia at admission [24, 25, 30, 45]. It could be that children admitted to inpatient care because of medical complications and in addition presenting with anaemia have a different pathophysiologic profile to children in outpatient care with anaemia but who don’t present medical complications and that would explain a slower response to treatment. It could also be that the inpatient treatment with therapeutic feeds such as F75 and F100 that do not contain iron [55] would slow down the recovery of anaemic patients. Regardless, anaemia is a condition driven by multiple factors including infections and nutritional deficiencies [56]. Similarly to the possible explanation for the faster recovery among children with illnesses at admission, it could be that these children, when managed properly via medical and nutritional treatment, respond quickly to treatment.

Open defecation was associated with slightly faster recovery. This association was contrary to what we expected. It could be that due to poor hygiene conditions these children entered treatment after an enteric illness and weight loss and subsequently responded fast to treatment. However, controlling for diarrhoea at admission did not reduce the association.

Living in a more urban setting was associated with higher non-response rate compared to a more rural setting and we do not have a hypothesis for why this should be. Among previous studies looking into factors influencing time to recovery two have reported no association between residence and recovery [31, 34] and one reported a quicker recovery among those coming from a more urban setting [27].

We found no association between HAZ at admission and time to recovery or non-response to treatment. Categorising HAZ into <-2 and ≥-2 did not reveal an association either. This is worth noting as there has been some interest in looking at the treatment outcomes of concurrently stunted and wasted children. Previously there has even been concern that short wasted children would not respond adequately to treatment although this has been shown not to be the case [57]. Our findings are consistent with this observation.

This study has several strengths including that it was done prospectively as part of a clinical trial with few missing data. The data quality was high with nurse diagnosed morbidity and strict respect of recovery criteria to investigate time to recovery and non-response to treatment.

The study also has limitations. The population studied was very young (mostly <24 months old) so some of the associations could be different if the full age range of 6–59 months were present. Also, the use of strict referral criteria identifying children with stagnant weight gain and weight loss led to high referral rate (20%) and many children being excluded from the time to recovery analysis. However, including these children in the analysis did not change the associations. Additionally, the findings could be context specific as the causal pathways and the response to treatment may be different in other settings.

In conclusion, illness episodes during treatment were common and also the strongest predictor of non-recovery and non-response, in addition to missed visits. Over half of the children suffered at least 2 illness episodes during treatment increasing time to recovery by nearly 4 weeks with two episodes and over 7 weeks with more than two episodes. This indicates that while correct diagnosis and management of co-morbidities is crucial, prevention would be key to shortening the treatment duration. Missing a visit during treatment also increased treatment time considerably by 3 weeks which calls for reflecting the service delivery to better accommodate clients’ needs to rearrange appointments and potentially further decentralise health services. In general, we call for more data on children with SAM, both prior and during treatment, in order to better understand causal pathways and pathophysiology of children admitted to care and subsequently their specific needs for recovery.

19 Apr 2021

PONE-D-20-35912

Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition

PLOS ONE

Dear Dr. Kangas,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jun 03 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:  http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Seth Adu-Afarwuah

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

2. Thank you for stating the following in the Competing Interests section:

'STK was previously employed by Nutriset, a producer of RUTF. HF has received research grants from ARLA Food for Health Centre, and also has research collaboration with Nutriset. Other authors declare no financial relationships

with any organisations that might have an interest in the submitted work in the previous five years, and declare no other relationships or activities that could appear to have influenced the submitted work.'

a. Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these.

Please note that we cannot proceed with consideration of your article until this information has been declared.

b. Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

3. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

4. Please amend your list of authors on the manuscript to ensure that each author is linked to an affiliation. Authors’ affiliations should reflect the institution where the work was done (if authors moved subsequently, you can also list the new affiliation stating “current affiliation:….” as necessary).

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The primary analysis used was the proportional hazard regression models which were fitted to quantify effects of predictors on time to recovery. The Kaplan Meier plots were used to visualize results concerning age and admission categories. Restricted mean time to recovery was estimated for significant categorical predictors. Logistic regression was used to evaluate predictors of non-response to treatment. Both unadjusted and adjusted models for sex and age at admission were fitted. The analysis appears reasonable.

The data is from the MANGO study . The study design and some protocol information is noted on page 4. The authors note on page 7 that excluding the 157 referrals, 644 children contributed to the analysis of time to recovery and all 801 to the analysis of non-response to treatment. The study design lacks statistical justification in the requires sample size from a power perspective. Given the large number of variables considered in the time to event and non response analyses and the large number of p-values presented, some rationale for the sample size is appropriate, despite the descriptive nature of the study. The sample size certainly appears adequate. However, the authors should comment on the statistical aspects of its adequacy.

Reviewer #2: This study was an offshoot study from the Mango trial looking at the effect of a nutrient supplement on recovery from severe acute malnutrition

Firstly, the paper contains a number of grammatical errors that needs to be addressed.

Abstract/Introduction

More details need to be added on what SAM is, what is known about the causes of SAM etc.

What is not clear is if SAM can be caused by socioeconomic factors such as lack of food availability. This needs to be addressed as it ties into some of the speculations made in the discussion section in relation to categorizing the participants that present with illness vs those that don’t and their recovery.

4 million children where? Globally or locally?

Methods/results

It is unclear whether the data analysed are regarding standard RUTF or reduced RUTF

Also what is RUTF? A Nutrition label table should be added.

Equations of WHZ and MUAC should be included

Line 81: what does subsist mean?

Clearer detail on the study protocol should be included: Number of visits, when treatment was administered, dose, information collected, etc—in a figure.

Line 93: a reference needs to be included for the “national protocol”

Details of manufacturer for the kits/equipment used needs to be included in the “data collection” section

Line 108-109 how was data monitoring and cleaning performed?

Not clear on how the predictor values were calculated in terms of hazards ratio, REF and odds ratio. Needs to be clarified.

What is open defecation?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

30 May 2021

Dear Editor and Reviewers,

Many thanks for the careful review of our manuscript on the “Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition”. We truly appreciated the thoughtful feedback provided and have now worked to improve our manuscript addressing the specific points raised. Please find below an explanation of the improvements made to the manuscript. Our responses are highlighted in bold italics.

Editors comments:

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response: Thank you for the orientation. We have now revised the formatting to correspond to the Plos One requirements.

2. Thank you for stating the following in the Competing Interests section:

'STK was previously employed by Nutriset, a producer of RUTF. HF has received research grants from ARLA Food for Health Centre, and also has research collaboration with Nutriset. Other authors declare no financial relationships

with any organisations that might have an interest in the submitted work in the previous five years, and declare no other relationships or activities that could appear to have influenced the submitted work.'

a. Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If there are restrictions on sharing of data and/or materials, please state these.

Please note that we cannot proceed with consideration of your article until this information has been declared.

Response: Thank you for this guidance. We have now revised the conflict of interest statement to include the sentence “The declared competing interests do not alter our adherence to PLOS ONE policies on sharing data and materials”.

b. Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Response: Thank you, this has been done.

Please know it is PLOS ONE policy for corresponding authors to declare, on behalf of all authors, all potential competing interests for the purposes of transparency. PLOS defines a competing interest as anything that interferes with, or could reasonably be perceived as interfering with, the full and objective presentation, peer review, editorial decision-making, or publication of research or non-research articles submitted to one of the journals. Competing interests can be financial or non-financial, professional, or personal. Competing interests can arise in relationship to an organization or another person. Please follow this link to our website for more details on competing interests: http://journals.plos.org/plosone/s/competing-interests

3. We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository. Please add a citation to support this phrase or upload the data that corresponds with these findings to a stable repository (such as Figshare or Dryad) and provide and URLs, DOIs, or accession numbers that may be used to access these data. Or, if the data are not a core part of the research being presented in your study, we ask that you remove the phrase that refers to these data.

Response: Thank you for this clarification. We have thus pursued to suppressing this part of the discussion: More children with both admission criteria reached MUAC ≥125 mm before reaching a WHZ ≥-2 (data not shown). It would thus seem that reaching a WHZ ≥-2 is more challenging for recovering young children and maybe unreachable for some as indicated by the high proportion of non-response to treatment.

4. Please amend your list of authors on the manuscript to ensure that each author is linked to an affiliation. Authors’ affiliations should reflect the institution where the work was done (if authors moved subsequently, you can also list the new affiliation stating “current affiliation:….” as necessary).

Response: Thank you, this has now been done.

Reviewers' comments:

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The primary analysis used was the proportional hazard regression models which were fitted to quantify effects of predictors on time to recovery. The Kaplan Meier plots were used to visualize results concerning age and admission categories. Restricted mean time to recovery was estimated for significant categorical predictors. Logistic regression was used to evaluate predictors of non-response to treatment. Both unadjusted and adjusted models for sex and age at admission were fitted. The analysis appears reasonable.

The data is from the MANGO study . The study design and some protocol information is noted on page 4. The authors note on page 7 that excluding the 157 referrals, 644 children contributed to the analysis of time to recovery and all 801 to the analysis of non-response to treatment. The study design lacks statistical justification in the requires sample size from a power perspective. Given the large number of variables considered in the time to event and non response analyses and the large number of p-values presented, some rationale for the sample size is appropriate, despite the descriptive nature of the study. The sample size certainly appears adequate. However, the authors should comment on the statistical aspects of its adequacy.

Response: Thank you for raising this question. We studied a total of 25 predictors for a total of 433 events resulting in ratio of 17 events per predictor (EPP). Often 10 EPP is considered sufficiently robust for predictor analysis. When calculating the power it seems we have a 64% power to detect a 20% increase or reduction in the hazard (HR<0.8 or HR>1.25). While this post hoc power calculation can be done it does not seem like good statistical practice (Hoenig & Heisey, 2001 ). So instead in the manuscript we have now given reference to the sample size of similar studies that, in most cases, seem to include a sample of around 400 children treated. This is how it states now in the manuscript: “The study recruited a total of 801 children, which was an adequate sample size for analyzing predictors as judged by previous studies looking at time to recovery with samples sizes starting from 200 and most around 400 children treated [16-27].” (LINE 83-90 in final version without trackchanges)

Reviewer #2: This study was an offshoot study from the Mango trial looking at the effect of a nutrient supplement on recovery from severe acute malnutrition

Firstly, the paper contains a number of grammatical errors that needs to be addressed.

Response: Thank you for this note. We have now revised the manuscripts English thoroughly.

Abstract/Introduction

More details need to be added on what SAM is, what is known about the causes of SAM etc.

What is not clear is if SAM can be caused by socioeconomic factors such as lack of food availability. This needs to be addressed as it ties into some of the speculations made in the discussion section in relation to categorizing the participants that present with illness vs those that don’t and their recovery.

Response: Thank you for this comment. We have now included a paragraph on the causes of SAM in the Introduction section as follows: “Severe acute malnutrition (SAM) is a condition that occurs when the food intake does not meet the nutritional requirements either as a consequence of poor intake or disease [1]. Generally SAM arises in contexts with social, political and economic factors affecting food availability and where infections and inflammation are common [1]. This is also why no single intervention has been shown to reduce the incidence that requires a holistic package of interventions [2].” (LINE 43-47 in final version without trackchanges)

4 million children where? Globally or locally?

Response: Thank you for this comment. We now added the precision of “globally” in the following sentence: “Every year, more than 4 million children are treated for SAM globally [6] with a varying mean treatment time and proportion of non-response to treatment [7-11].” (LINE 57-58 in final version without trackchanges)

Methods/results

It is unclear whether the data analysed are regarding standard RUTF or reduced RUTF

Response: Thanks for the question. The data were analysed for all children treated regardless of their treatment arm. This has now been more clearly stated in the Methods section as follows: “All children included in the MANGO trial regardless of their treatment arm (reduced RUTF or standard RUTF) were included in the analysis.” (LINE 157-158 in final version without trackchanges)

Also what is RUTF? A Nutrition label table should be added.

Response: Thanks for the question. We have now added the following to the introduction section: “RUTF are energy and nutrient dense pastes usually composed of peanut butter, milk powder, oil, sugar and a vitamin and mineral complex designed to fulfil the nutritional needs of children recovering from SAM [4].” (LINE 51-53 in final version without trackchanges)

Equations of WHZ and MUAC should be included

Response: Thanks for the helpful comment. We have now added these precisions to the methods section under Data collection: “Weight was measured using an electronic scale (SECA 876) to the nearest 100 g, height using a wooden measuring board (locally made) to the nearest 1 mm, and MUAC using a non-stretchable colourless measuring tape to the nearest 1 mm. Z-scores were calculated using the WHO standards and STATA command zscore06 [21].” (LINE 115-118 in final version without trackchanges)

Line 81: what does subsist mean?

Response: Thanks for noting. We have changed the word to “depend” (LINE 100 in final version without trackchanges)

Clearer detail on the study protocol should be included: Number of visits, when treatment was administered, dose, information collected, etc—in a figure.

Response: Thanks for pointing out. We have added some clarifications for these points in the Methods section under the Study participants and treatment protocol and Data collection sub-sections. They detail when and what treatment was administered (specifically at admission and then during the weekly follow ups) and the type of data collected at different time points.

Line 93: a reference needs to be included for the “national protocol”

Response: Thank you. This has been done.

Details of manufacturer for the kits/equipment used needs to be included in the “data collection” section

Response: Thank you. These details have been included and the following additions/edits made to the sections: “Weight was measured using an electronic scale (SECA 876) to the nearest 100 g, height using a wooden measuring board (locally made) to the nearest 1 mm, and MUAC using a non-stretchable colourless measuring tape to the nearest 1 mm. Z-scores were calculated using the WHO standards and STATA command zscore06 [21].” (LINE 115-118 in final version without trackchanges)

Line 108-109 how was data monitoring and cleaning performed?

Response: Thank you for the comment. We have now added the following section to the Data collection sub-section of the Methods: “Data monitoring included among other thing, checking duplicate entries and outliers, anthropometric decimal distributions and correct prescription of medication according to diagnosed conditions. Any potential data problem resulted in action. Data cleaning was based on double checks of electronic data against patient registries or therapeutic cards.” (LINE 120-123 in final version without trackchanges)

Not clear on how the predictor values were calculated in terms of hazards ratio, REF and odds ratio. Needs to be clarified.

Response: Thank you for mentioning this shortcoming. We have now clarified this in the Data analysis sub-section of the Methods as follows: “Cox proportional hazard regression models were fitted to quantify effects of predictors on time to recovery by means of hazard ratios, describing the increased or decreased rate of change in likelihood of recovery.” (LINE 159-161 in final version without trackchanges)

What is open defecation?

Response: Thank you for the question. We have added the definition of open defecation in the Methods now as follow: “open defecation (the practice of defecating in nature instead of a toilet facility)” (LINE 150 in final version without trackchanges)

Submitted filename: Response_to_Reviewers_FINAL_new.docx

Click here for additional data file.

9 Feb 2022

5 Mar 2022

Dear Editor and Reviewers,

Many thanks for the careful review of our manuscript on the “Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition”. We truly appreciated the thoughtful feedback provided and have now worked to improve our manuscript addressing the specific points raised. Please find below an explanation of the improvements made to the manuscript.

Editors comments:

Line 33. Suggest changing to ..<-3, anaemia or no illness at admission…’ as it is now not clear whether the ‘no’ refers only to illness or also to anaemia.

Response: Thanks for the observation. We have now changed the sentence as follows:

LINE 33: “…WHZ <-3, no illness nor anaemia at admission,..”

Line 169-173. Can the authors explain the difference between ‘non-response to treat-ment’, which to me would indicate lack of weight gain, and ‘referred to inpatient care due to stagnant weight or weight loss’ (I leave out here the medical complications, which are different). Would the results have been different if children referred to inpa-tient care because of lack of weight gain would have been included in the analysis of time to recovery?

Response: Thanks for the observation. Indeed we would expect that children ending up as non-responders would present with no weight gain during treatment. However, in our cohort we saw that some children continued to gain weight but with a very slow rate which meant that they were not classified as stagnant weights (where the definition was no more than 100g gained over 4 weeks) but continued their treatment. These children could also be gaining in height meaning that their WHZ was not improving which meant they were not reaching the recovery cut-off of WHZ-2. Also, to clarify the implication on the results of excluding the children referred to inpatient care we firsts added a sentence in the outcome definition section on line 142 and then a sentence on the sensitivity analysis results when including these children (which led to similar results).

Thus, we have now added the following sections:

LINE 101-103: “In case of medical complications, weight loss of over 5% at any point or stagnant weight defined as no more than 100g weight gained over 4 weeks, dur-ing treatment, children were referred to inpatient care, as per the Burkina national CMAM protocol.”

LINE 141-144: “For the study of time to recovery, non-recovered cases were right censored contributing to the analysis of time to recovery until exit from study. Pa-tients referred to inpatient care were excluded from the analysis in order to limit potential bias that could be introduced due to short length of stay in treatment.”

LINE 210: “Including children referred to inpatient care in the analysis of different factors with time to recovery resulted in similar associations.”

In table 4. How can sex and age still be in the model when the model was adjusted for these predictors (last columns)

Response: Thanks for the question. The odds ratios for these variables (sex and age) in the sex and age adjusted model represent the OR adjusted for the other variable. Meaning, when looking at sex as a predictor, the adjusted model adjusts for age to look at the age independent predictive power of sex. And the inverse in the case of age. We have now made this clearer in all the tables (2, 3 and 4) adding a footnote indicating that “when analysing sex as a predictor, only age was included as adjustment and when analysing age as a predictor only sex was included as adjustment”.

6. Review Comments to the Author

Reviewer #3: The study provides relevant information on a subject with limited literature. The publication of the study results could contribute to the current discussion on the need to further decentralise essential health services to reach a broader number of peo-ple in need.

Some aspects of this work could have been more clearly presented and explained, as detailed below:

Introduction

The authors should extend the rationale and explain why the topic treated is particularly crucial. This could be done by better explaining the absolute magnitude of the problem and, on the other hand, by better underling the implications of the study's results on current SAM treatment and prevention programs.

Response: Thanks for the suggestion. We added the following sections highlighting the magnitude of the problem (in the 1st paragraph of the introduction section) and the subsequent coverage of treatment (4th paragraph of the introduction section):

LINE 43 – 49: “Severe acute malnutrition (SAM) is a condition that occurs when the food intake does not meet the nutritional requirements either as a consequence of poor intake or disease [1]. SAM is diagnosed when a child presents with a weight-for-height z-score (WHZ) <-3, a mid-upper-arm circumference (MUAC) <125 mm or nutritional oedema [2]. While the overall prevalence of SAM is unknown, in 2020 it was estimated that 2% of all children below the age of 5 years presented a WHZ <-3 translating to more than 13.6 million children suffering from severe wasting at any time [3]. Children with SAM have a 11.6 increased risk of mortality compared to children with no nutritional deficits living in the same contexts [4].”

LINE 62-66: “In 2020, around 5 million children were treated for SAM globally [10]. With 13.6 million children suffering from severe wasting at any time and applying of incidence correction factor of 3.5 [11] to account for all new cases arising in a year, this translates to 47.6 million episodes of severe wasting in a year. Thus, the current coverage of treatment is around 10%. This is when excluding the burden of MUAC cases which, if added, would translate to an even lower coverage of treatment. Such low coverage warrants for reflection on how to improve it possibly by optimising and better targeting treatment in order to expand it to more cases.”

In the background, the authors mentioned the last figure available for coverage without mentioning the global burden of wasted children 6-59months.

Response: Thanks for the observation. We added the following sections highlighting the magnitude of the problem (in the 1st paragraph of the introduction section) and the subsequent coverage of treatment (4th paragraph of the introduction section):

LINE 43 – 49: “Severe acute malnutrition (SAM) is a condition that occurs when the food intake does not meet the nutritional requirements either as a consequence of poor intake or disease [1]. SAM is diagnosed when a child presents with a weight-for-height z-score (WHZ) <-3, a mid-upper-arm circumference (MUAC) <125 mm or nutritional oedema [2]. While the overall prevalence of SAM is unknown, in 2020 it was estimated that 2% of all children below the age of 5 years presented a WHZ <-3 translating to more than 13.6 million children suffering from severe wasting at any time [3]. Children with SAM have a 11.6 increased risk of mortality compared to children with no nutritional deficits living in the same contexts [4].”

LINE 62-66: “In 2020, around 5 million children were treated for SAM globally [10]. With 13.6 million children suffering from severe wasting at any time and applying of incidence correction factor of 3.5 [11] to account for all new cases arising in a year, this translates to 47.6 million episodes of severe wasting in a year. Thus, the current coverage of treatment is around 10%. This is when excluding the burden of MUAC cases which, if added, would translate to an even lower coverage of treatment. Such low coverage warrants for reflection on how to improve it possibly by optimising and better targeting treatment in order to expand it to more cases.”

The comparison between global burden and treatment coverage shows how coverage is still impressively low. Thus, how it is essential to reach more children affected by SAM. As mentioned in the paper, a shorter treatment length would eventually decrease costs per individual treated. Reduced cost per child treated could subsequently help extend the coverage of the programs.

Response: Indeed! To make this clear in the manuscript, we added the following paragraph in the Introduction section:

LINE 62-66: “In 2020, around 5 million children were treated for SAM globally [10]. With 13.6 million children suffering from severe wasting at any time and applying of incidence correction factor of 3.5 [11] to account for all new cases arising in a year, this translates to 47.6 million episodes of severe wasting in a year. Thus, the current coverage of treatment is around 10%. This is when excluding the burden of MUAC cases which, if added, would translate to an even lower coverage of treatment. Such low coverage warrants for reflection on how to improve it possibly by optimising and better targeting treatment in order to expand it to more cases.”

The authors should also consider explaining that global data reported are only for children 6-59 months and eventually why this target group is so critical.

Response: Thanks for the comment. We added the following sentence to the 1st paragraph of the introduction section to clarify the importance of targeting children with SAM due to their high risk of mortality:

LINE 49: “Children with SAM have a 11.6 increased risk of mortality compared to children with no nutritional deficits living in the same contexts [4].”

Methods

Statistical analysis is clearly described.

Response: Thanks!

Discussion

Findings are clearly resumed and discussed.

Response: Thanks!

Minor points:

-Consider eventually modifying the abstract according to the modifications done in the introduction

Response: Due to very limited space in abstract, we did not add any of the further justifications now included in the introduction section.

- line 150, it would be better to specify what acronym ARI is the first time it is used. The authors specified it only a few lines below.

Response: Thanks for the observation. We defined ARI now at the 1st appearance on LINE 150.

Submitted filename: Response_to_Reviewers_5March2022.docx

Click here for additional data file.

12 Apr 2022

Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition

PONE-D-20-35912R2

Dear Dr. Kangas,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Frank Wieringa, M.D., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

20 May 2022

PONE-D-20-35912R2

Predictors of time to recovery and non-response during outpatient treatment of severe acute malnutrition

Dear Dr. Kangas:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Frank Wieringa

Academic Editor

PLOS ONE