Physical activity level among children recovering from severe acute malnutrition

Abstract

Objective

To assess the level and predictors of physical activity at discharge among children recovering from severe acute malnutrition (SAM).

Methods

We conducted a prospective study among 69 children 6 - 59 months of age admitted with SAM for nutritional rehabilitation at Mulago National Referral Hospital, Uganda. Using hip-mounted tri-axial accelerometers, we measured physical activity expressed as counts per minute (cpm) during the last three days of hospital treatment. As potential predictors, we assessed socio-demographic and clinical characteristics, including duration to transition phase and duration of hospitalisation, serum C-reactive protein and whole-blood docosahexaenoic acid (DHA). Multiple linear regression analyses were used to identify predictors of physical activity.

Results

Median (IQR) age was 15.5 (12.6; 20.5) months. At discharge, the mean (SD) movement was 285 (126) cpm. Physical activity was 43 (19; 67) cpm higher for each unit increase in weight-for-height z-score (WHZ) at admission and 72 (36; 108) cpm higher for each centimeter increase in MUAC at discharge. Whole-blood DHA on admission was also a positive predictor of physical activity, whereas duration to transition phase and duration of hospitalisation were both negative predictors.

Conclusion

The level of physical activity at discharge among children treated for SAM was low. WHZ, MUAC and DHA on admission were positive predictors of physical activity whereas duration of stabilization and hospitalization were negative predictors of physical activity.

Introduction

Severe acute malnutrition (SAM) is defined as weight-for-height z-score (WHZ) <-3 or mid-upper-arm circumference (MUAC) ≤11.5 cm or bilateral pitting edema (1). SAM causes considerable morbidity and mortality among children in low-income countries (2). Physical activity is conventionally defined as any bodily movement produced by contraction of skeletal muscle that increases energy expenditure above a basal level (3). Malnourished children have low levels of physical activity (4,5) since they compensate for lack of dietary energy by decreasing energy expenditure through reduced physical activity (6). This same effect has been demonstrated in animal models (7). Physical activity requires not only adequate skeletal muscle mass, but also muscle contractility which may be impaired by inadequate docosahexaenoic acid (DHA) in skeletal muscle membranes (8,9).

Subjective methods, i.e. child self-report and parent proxy reports, have been used to quantify physical activity (10,11), but objective measures (i.e. direct assessment) such as heart rate monitors and motion sensors are now increasingly being used (12). Accelerometry has become a common method to assess physical activity in well-nourished children (12,13). Its use has been validated and found feasible in non-underweight toddlers (14). However, there is less objective data available in children with SAM.

Recovery from SAM is primarily assessed by weight gain, with no assessment of regain of function including physical activity. One study reported improved physical activity during rehabilitation of malnourished children (15), but that study was too small to allow analysis of key biological determinants of activity.

The purpose of this study was to assess the level and predictors of physical activity at the end of inpatient treatment among children recovering from SAM.

Methods

Study design

This was a prospective cohort study among children admitted with complicated SAM for in-patient treatment at Mwanamugimu Nutrition Unit, Mulago Hospital in Kampala, Uganda between October 2012 and March 2013. The study was approved by the Makerere University School of Medicine Research Ethics Committee, and Uganda National Council of Science and Technology, and a consultative approval was obtained from the Danish National Board of Research Ethics. Informed consent was provided by parents or guardians of the children before enrolment into the study.

Study site and standard of care

Mwanamugimu Nutrition Unit, Mulago Hospital in Kampala, Uganda, is the main national treatment centre for children with complicated SAM. All children received nutritional and medical care according to the Ugandan National Guidelines for in-patient treatment of SAM (16), which are based on recommendations from the WHO (1,17,18). This involved three stages: stabilization phase, transition and rehabilitation phase. In the stabilization phase, children were given pre-mixed, low-energy therapeutic formula, F-75 (Nutriset, France) orally or by naso-gastric tube when they were unable to feed orally. All children were given parenteral antibiotics, usually ampicillin and gentamycin. Children with dehydration were rehydrated with oral rehydration solution for malnutrition (ReSoMal, Nutriset, France). The children were given high-energy therapeutic formula, F-100 (Nutriset, France) when they entered transition phase, i.e. when they were clinically stable, were regaining appetite and their edema had started to resolve. When the children were clinically well, had a good appetite and no edema, they started rehabilitation phase and were discharged for out-patient treatment with ready-to use therapeutic food. In preparation for discharge, mothers were encouraged to give sensory stimulation and emotional support to the child.

All children admitted with SAM received similar routine medical and nutritional treatment, regardless of participation in the study. Furthermore, all biological mothers were offered routine counseling and testing for HIV antibodies according to WHO guidelines, using rapid test (Determine™, 357 Matsuhidai, Matsudo-shi, Chiba, 270-2214, Japan) as first-line (19), and if the mother was infected, or absent, the child was tested.

Selection criteria

Children were all participants in the FeedSAM study, an observational cohort study among children aged 6 - 59 months admitted for in-hospital treatment of SAM, defined as WHZ <-3 of the WHO Growth Standard or MUAC ≤11.5 cm or bilateral pitting edema. Children were excluded from the FeedSAM study if they presented with shock on admission, severe respiratory distress requiring resuscitation, hemoglobin concentration <4 g/dl, body weight <4.5 kg or severe disability such as cerebral palsy.

Data collection

This was a prospective cohort study of children admitted for in-hospital treatment of SAM, between October 2012 and March 2013. Using a questionnaire, the child’s socio-demographic characteristics were recorded and physical examination performed. The age of the child was calculated from caretaker’s estimation of the child's date of birth. When available, the child's immunization card was used to confirm the date of birth. Children were considered breastfeeding if the mother reported they were still breastfeeding at admission and continued to breastfeed until discharge from hospital. Edema was diagnosed according to WHO guidelines (17). Length was measured using an infant length board (Infant/Child ShorrBoard®, Weigh and Measure, LLC 17802 Shortley Bridge Place, Olney, Maryland 20832-1671 USA) and MUAC using measuring tape (S0145620 MUAC, Child 11.5 Red/Pac-50), both to the nearest 1 mm. Body weight was measured to the nearest 100 g using a digital scale (Seca 874 U, Hammer Steindamm 9-25 22089 Hamburg, Germany). Anthropometric z-scores were computed using WHO Growth Standards (20), using the lowest weight recorded during admission to determine non-edema weight.

Blood sampling

On admission, hemoglobin was measured in venous blood collected in heparinized tubes (Vacutainer^®) Belliver Industrial Estate, Plymouth, PL6 7BP, UK) using HemoCue (Hb 201+, Ängelholm, Sweden).

Fatty acids were prepared from whole-blood by direct transesterification for analysis by gas-liquid chromatography as previously described (21,22). DHA values are expressed as % of total whole-blood fatty acids (FA%).

Plasma level of C-reactive protein (CRP) was measured by high sensitive kit on an ABX Pentra 400 (Horiba, France, no. A11A01611 and A11A01696).

Physical activity measurement

Children were clinically assessed daily for discharge readiness as indicated by resolution of complications and edema as well as regain of appetite and increasing weight. When children were almost ready for discharge, physical activity was measured over the next three days with a tri-axial accelerometer (ActiGraph GT3X+; ActiGraph, Pensacola, Florida) measuring accelerations in three orthogonal planes. The accelerometer was attached to an adjustable elastic belt and worn over the right hip. The parents/caretakers were instructed to have the child wear the device continuously throughout day and night over the three days. Monitors were programmed to sample data at 100 Hz and downloaded with the manufacturer’s software (ActiLife version 5.5.0; ActiGraph). This software was also used to reduce the raw physical activity data to 2-second epoch data. Monitor non-wear time was determined as time segments with ≥ 90 minutes of zero activity. We calculated the vector magnitude of the three axes and summarized average movement intensity (mean counts per minute [cpm]) by hour and further collapsed it by day and total monitoring period while minimizing diurnal bias caused by any imbalance in availability of valid wear data. Days with <22 hours of wear time as well as recordings with an average intensity of more than 10 000 cpm were excluded from all analyses.

Statistical analyses

Double data entry was done into Epidata (Odense, Denmark) with all further analyses performed in Stata 12 (StataCorp LP, College Station, Texas). Normally distributed data are presented as mean ±SD and data for variables that did not comply with a normal distribution are given as median (interquartile range (IQR). Activity was initially expressed as daily averages for the three days of monitoring. Using a mixed model for repeated measurements we compared level of activity between the three days taking into account repeated measurements per individual. Since we found no difference in level of activity between the three days, they were collapsed and reported as an overall daily average. Predictors of physical activity were identified using multiple linear regression, with adjustment for age and sex. Among possible predictors, child’s age, MUAC, WHZ, HAZ, hemoglobin, DHA, number of days before transition duration of admission and maternal age were analysed as continuous variables whereas child’s sex, HIV status, breastfeeding status and presence/absence of edema were analysed as categorical variables. Serum CRP was presented using median (IQR), and a value >10 mg/L was used to define inflammation when examining its association with physical activity. Maternal education was categorized as “primary school or less” and “secondary school or more”. Primary caretaker was categorized as the mother or any other caretaker. The number of siblings variable was presented as multi-category variable, but used as a continuous variable when examining its role as a potential predictor. Predictors measured at discharge (i.e. WHZ, MUAC and whole-blood DHA) were adjusted for admission status. Predictors were regarded as statistically significant at p <0.05.

Results

Of 120 children with SAM, physical activity data were available for 69 (57.5%) children (Figure 1). The median age of these 69 children was 15.5 (IQR, 12.6; 20.5) months, 62% of the children were boys and 68% had been admitted with edema (Table 1). On admission, mean WHZ was -3.3 ±1.2 and mean height-for-age z-score (HAZ) was -3.1 ±1.4. At discharge, the mean WHZ had improved to -1.9 ±1.1 at the time when the physical activity measurements were done. Nine (13%) of 67 children tested for HIV were infected. Thirty one (62%) of the 50 children tested for CRP had values > 10mg/L with a median of 16.3mg/L (IQR, 6.2; 30.1). Median duration of hospital admission was 17 (IQR, 14; 22) days and median duration from admission to transition was 5 (IQR, 4; 8) days. Thus, the median time from admission and transition to physical activity evaluation was 13 (IQR, 10; 18) and 8 (IQR, 4; 11) days, respectively.

Figure 1. Flow diagram of 120 children included in the FeedSAM study.

Table 1. Baseline characteristics of 69 children admitted with severe acute malnutrition.

Socio-demograghic characteristics	N
Age, months	69	15.5 (12.6; 20.5)
Male sex	69	43 (62.3)
Breastfeeding	65	12 (18.5)
Mother’s age, years	65	24.2 ± 4.4
Mother’s education	58
Primary school or less		29 (50)
Secondary school or above		29 (50)
Mother’s religion	65
Christian		50 (76.9)
Muslim		15 (23.1)
Primary care taker	68
Mother		62 (91.2)
Other		6 (8.8)
Patient has a sibling	69	66 (95.7)
Occupation of head of household	56
Unemployed		8 (14.3)
Agriculture		6 (10.7)
Domestic service, unskilled manual		14 (25.0)
Skilled manual, sales and service		24 (42.9)
Professional/technical/management		4 (7.1)
Clinical data
Weight-for-height z-score	69	-3.3 ± 1.2
Height-for-age z-score	69	-3.1 ± 1.4
Mid-upper arm circumference, cm	68	11.7 ± 1.3
Oedema	69	47 (68.1)
HIV infected	67	9 (13.4)
Haemoglobin, g/L	62	0.90 ± 0.22
Serum CRP >10mg/L	50	31 (62)
Docosahexaenoic acid, FA%	64	1.4 (1.0; 1.7)
Weight-for-height z-score at discharge	66	-1.9 ± 1.1
Duration from admission to transition, days	67	5 (4; 8)
Duration of admission, days	65	17 (14; 22)

Data are median, (25%; 75% IQR) or mean ± SD or number (%). N, number of children for whom information was available. All measurements were taken at admission unless otherwise stated.

Physical activity level and diurnal pattern of activity

Overall, the mean physical activity of the recovering children was 285 ±126 cpm, and there was no difference between boys and girls (291 vs 275 cpm, p= 0.550). On the assumption that the hours between 7 am and 7 pm represent awake time, mean awake physical activity was 381±173 cpm. In parallel, the uniaxial (Y-axis) estimates for total 24-hr and awake-only activity were 126 and 169 cpm, respectively. The total level of activity as well as the overall diurnal movement pattern was similar on all three days before discharge (daily data not shown). Children were most active between 5 am and 8 pm (Figure 2). Overall, there was a small reduction in activity from 10 am to 4 pm, followed by an evening peak at 6 pm from which point the activity level declined steadily towards zero at midnight.

Figure 2. Mean diurnal physical activity.

Predictors of physical activity

MUAC at discharge was a positive predictor of physical activity, i.e. for each one additional centimeter, physical activity was 72 cpm (95% CI 36; 108, p<0.001) higher (Table 2). WHZ at admission was associated with 43 (95% CI 19; 67, p=0.001) cpm higher physical activity per unit increase in WHZ. Similarly, whole-blood DHA on admission was associated with 76 (6; 147, p=0.034) cpm higher physical activity per FA%. For each extra day in the stabilization phase, physical activity was 9 cpm (95% CI 2; 16) lower, while each additional day of total hospital stay was associated with a 5 cpm (1; 10) lower physical activity at discharge. Age, sex, and HAZ were not associated with physical activity. Similarly, breastfeeding and serum CRP were not associated with physical activity level of the children at discharge.

Table 2. Predictors of physical activity in children admitted with severe acute malnutrition at the end of in-patient treatment.

Socio-demographic data	Regression coefficient	p-value
Age, months	-1.4 (-5.7; 3.0)	0.528
Male sex	-19.1 (-82.6; 44.4)	0.550
Mother’s age, years	-0.1 (-2.4; 2.2)	0.924
Mother’s education ≤ primary level	-25.1 (-94.1; 44.0)	0.470
Mother as care taker	-19.8 (-137.7; 98.1)	0.738
Number of siblings	3.7 (-21.6; 29.0)	0.772
Clinical data
Breastfeeding	69.8 (-13.2; 152.7)	0.098
Oedema	-5.0 (-75.1; 65.0)	0.886
Weight-for-height z-score
Admission	43.0 (19.3; 66.7)	0.001
Discharge	-0.1 (-37.9; 37.8)	0.998
Height-for-age z-score	10.7 (-16.0; 37.5)	0.427
Mid-up-arm circumference, cm
Admission	20.0 (-4.1; 44.1)	0.102
Discharge	72.2 (36.2; 108.3)	<0.001
Haemoglobin, g/L	65.7 (-90.0; 221.5)	0.402
Serum C-reactive protein >10 mg/L	58.4 (-18.5; 135.2)	0.133
HIV infected	-46.0 (-142.0; 50.0)	0.342
Docosahexaenoic acid, FA%
Admission value	76.2 (6.0; 146.5)	0.034
Discharge value	49.0 (-41.9; 139.8)	0.284
No of days before transition	-8.9 (-16.1; -1.7)	0.016
Duration of admission, days	-5.2 (-9.5; -0.9)	0.018

Data are regression coefficient (95% confidence interval) after adjusting for age and sex. Predictors measured at discharge were adjusted for admission values. All predictors were measured at admission unless otherwise stated.

Discussion

In this study among 69 children admitted with SAM, we found an overall low level of physical activity at discharge which was associated with various factors such as admission WHZ and DHA, discharge MUAC as well as total time of hospital admission.

Physical activity was generally low but still displaying a diurnal pattern, being lowest between 9 pm to 4 am. This corresponds well with the fact that children sleep for most of that time. The observed decrease in levels of activity between 10 am and 4 pm may correspond to the period when most children are napping (26).

Low activity levels in SAM children were also observed in a small (n=13) Ethiopian study, reporting a mean activity of 204 cpm measured after only 10 days of rehabilitation, compared to 17 days in the present study) (15); although the Ethiopian children were a bit younger and treatment regimes may differ slightly between countries, it is possible that the extra week of hospital treatment in our study may have been incrementally beneficial for restoring natural activity levels. These data in SAM children are complimented by recent observations in children from Burkina Faso with moderate acute malnutrition who accumulated on average 707 cpm of activity (10-second epoch), ie more than twice that of SAM children (24), and further by the even higher levels of physical activity reported in non-wasted toddlers from Malawi (307 counts/15 seconds ~1228 cpm) (23), a level four times as high. These differences in triaxial movement levels are not due to differences in epoch settings, as rerunning analyses in our data with 10- and 14-sec epochs before calculating vector magnitude yielded very similar (3-5% lower) estimates; rather, differences are indicative of a positive dose-response relationship between nutritional status and physical activity. In further support of that notion, activity levels also appear to be higher in healthy children from countries where malnutrition is rare; 455 cpm was noted in Australian 19-months old children (K. Hesketh, personal communication (27)) and 692 cpm was reported in 3-year old Scottish children (25), both studies using uniaxial accelerometers worn during day-time hours only, thus comparable to the methodologically equivalent uniaxial estimate of 169 cpm in our study. It is, however, also important to bear in mind that physical activity in the present study was measured in a hospital setting which may have restricted the children’s natural behavior and movement.

In terms of associations within the present study, we found no sex differences in activity levels which is in line with findings from healthy Australian pre-school children (27) although pre-school boys were reported more active than girls elsewhere (28). We observed non-significantly higher physical activity in breastfed children, which may partly be due to passive movement when the mother picks up the child for feeding and also breastfed children being carried more around by their mothers. Hence passive movement by mothers may contribute to the slightly higher physical activity level of breastfed children, although an actual effect of breastfeeding on subsequent spontaneous activity cannot be ruled out.

We found lower physical activity in children with longer duration before transition and longer total duration of hospital admission. It is plausible that children who spend a long time in the stabilization phase are the sickest with severe complications and thus stay longer in hospital. Our experience is that caretakers sometimes ask for early discharge before the child fully recovers due to other demands like work or the need to take care of other children at home. The results therefore indicate that such children may have been discharged before adequate recovery following the prolonged period of hospital stay.

On admission, we found two biological predictors of physical activity, namely whole-blood DHA and WHZ – the better the status the higher the level of activity. Such association with WHZ is consistent with findings from Indian children in whom wasting was associated with low physical activity (5). In severe wasting, muscle contractility may be weak resulting in very low physical activity, and children with low energy intake are supposedly more sedentary to preserve energy (6). The effect of stunting on physical activity is not fully understood, but reduced body size and muscle mass associated with stunting have been linked to reduced muscle strength. A study among 193 Mexican infants found reduced physical activity in stunting (4). Thus, we would have expected higher physical activity among children with higher HAZ, but did not find such association. Admission MUAC also did not predict physical activity while discharge levels were associated with higher levels of physical activity. Similar findings of poor physical activity with low MUAC were reported in malnourished Indian children (5). MUAC is determined by arm muscle and sub-cutaneous fat (29) and is correlated with muscle mass in healthy children (30), thus one would expect higher physical activity in children with higher MUAC.

Much as physical activity is associated with nutritional status measured at one time point, nutritional recovery (change from admission) is also important. Since change in nutritional status from admission to discharge reflects recovery, our linear regression models assessed discharge nutritional status as predictor for physical activity with adjustment for admission values to assess impact of recovery per se.

While admission values of DHA were a positive predictor of physical activity, there was no association between physical activity and change in whole-blood DHA at discharge. DHA is important for brain development and functioning and DHA deficiency has been associated with psychomotor and cognitive deficits (31). Studies in animal models have also demonstrated that low levels of DHA induce abnormalities in dopaminergic and serotonergic neurotransmission systems which are closely involved in the modulation of attention, motivation and emotion (31, 32). Admission DHA values in the present population may reflect a better nutritional status also of other nutrients. In Canadians, DHA intake was associated with protein intake (33, 34) possibly from animal products. In addition, skeletal muscle has been demonstrated to have a high DHA content (36) and it is the largest tissue reserve of DHA due to its relatively large mass (37). Therefore, it is possible that blood DHA could potentially be a biomarker of diet intake and/or muscle mass status, but these relationships with protein intake and muscle status need to be examined further. Higher protein intakes and/or increased muscle mass could both explain the positive association with physical activity.

One limitation to this study is its observational nature, which limits causal inference between low level of physical activity and the predictors. Furthermore, since the aim was to explore predictors of physical activity at discharge, many statistical tests were performed, which may have increased the risk of chance findings. Physical activity is associated with nutritional status at the time when it is measured, thus physical activity measured at discharge may not reflect association with admission nutritional status. Further still, it is difficult to make direct comparisons with other studies that used different wear protocols and epoch settings although the latter can be approximately dealt with by scaling. Another limitation is that we only could measure physical activity at discharge of 69 children out of a total 83 with discharge data in the whole cohort. The outcome variable of interest in our current investigation was overall volume of activity; this encompasses all movements, irrespective of intensity which is a sensible first analysis in this understudied population but subsequent analyses may examine the underlying intensity distribution, as well as activity bout patterns. Furthermore, we were not able to assess active (child-driven) vs passive (eg carrying) movements in the children. However, this study is among the first to describe physical activity in a larger group of children with SAM, and to explore correlations with clinical characteristics.

Conclusion

Among Ugandan children admitted with SAM, the level of physical activity at discharge was very low. In this study, levels of physical activity were associated with degree of muscle wasting in terms of MUAC at discharge as well as WHZ and DHA values at admission. The use of accelerometers as an objective method to measure physical activity as an outcome of nutritional rehabilitation should be encouraged in research settings as a tool to monitor treatment programs. Furthermore, more research is needed to determine the impact of stimulating physical activity by caregivers and staff, together with optimal nutrition on spontaneous physical activity and muscle mass.

List of abbreviations

CI

confidence interval

CPM

counts per minute

CRP

C-reactive protein

DHA

docosahexaenoic acid

HAZ

height-for-age z-score

MUAC

mid-upper arm circumference

SAM

severe acute malnutrition

SD

standard deviation

WHO

World Health Organization

WHZ

weight-for-height z-score