This cross-sectional study was conducted at a nutritional rehabilitation unit (NRU) attached to the Department of Pediatrics, R. D. Gardi Medical College (RDGMC), Ujjain, MP, India from March 2015 to September 2016. Consecutive children who were admitted to the NRU during the study period were eligible to be included in the study. Between 6 and 59 monthsof age SAM was defined as [14]:

If a child fulfilled any of the above criteria an “appetite test” was performed [14] in which the child was offered a minimum amount of local ready-to-use therapeutic food (LRUTF) prepared from groundnut (25%), milk powder (30%), sugar (30%), and vegetable oil (15%) by weight [15]. The amount of LRTUF was determined by the weight of the child as per Indian Academy of Pediatrics (IAP) guidelines [14], which conform to the composition guideline provided by WHO in 2007 [16]. Only children failing the appetite test were admitted to NRU and were eligible to be included in the study after written informed consent was obtained from the children’s caregiver. Children having cerebral palsy, genetic neurological and neuromuscular conditions significant congenital diseases, and malignant diseases, and HIV positive children were excluded from the study. In the children who were admitted in the NRU, SAM was managed as per the IAP consensus statement and the Government of India guidelines [3, 14]. NRU is a public–private partnership between RDGMC and Government of MP. Children with SAM children are referred from aganwadi’s (early childhood care and education centres) from Ujjain district. The Government of MP pays daily wage compensation of up-to 200 Indian rupees for 14 days to the mother/caretaker of child admitted to NRU.

At admission, a detailed history of factors associated with SAM was obtained through a pre-designed questionnaire after obtaining written informed consent. The questionnaire included demographic factors, nutritional history, feeding patterns, comorbidities associated with SAM and relevant history. After admission, the findings of physical examination and anthropometry were recorded. Children with SAM were classified as underweight, stunted, wasted and stunted and wasted as per the WHO/Government of India criteria [3]. Weight was measured using a digital weighing scale (Accord Electronics, Malad, Mumbai, India) to the nearest 0.1 kg. Head circumference and MUAC were measured using a non-stretchable measuring tape. Length was measured using an infantometer (IndoSurgicalsPvt. Ltd., New Delhi, India) for children below 2 years and a stadiometer (IndoSurgicalsPvt. Ltd., New Delhi, India) for those above 2 years. All anthropometric measurements were supervised by a senior resident or faculty member. All children were clinically evaluated from their admission till discharge thrice daily by a team of junior and senior residents, and twice daily by a consultant professor (JM/AP). The relevant history was reported through parent interview. The birth weight was recorded from birth records directly. Development was assessed before discharge during the recovery phase of SAM when the children were gaining at least 10 g/kg weight per day, had a good appetite, and were playful and interactive. BSID (III) was used for developmental delay assessment [17]. The original scale (US norms) was not changed. The scale was forward translated from English to Hindi by two bilingual paediatricians. A panel consisting of two language experts resolved any discripencies in translation. Back translation was done by an independent language expert to see if original meaning was retained. BSID (III) was administered in a quiet, well-lit, and comfortable room by two trained paediatricians simultaneously in presence of mothers after informing them regarding the significance of identifying developmental delay for relevant interventions. For 12 months and younger and 13 months and older children, the administration time was, respectively, an average of 50 and 90 minutes. Development was assessed in five domains: gross motor, fine motor, expressive language, receptive language and cognition. Raw scores were converted to composite scores, as outlined in the manual [17]. To calculate the language composite-score, an average of the expressive and receptive language scores was obtained, and the motor composite score was calculated by averaging gross and fine motor scores. The average motor, language, and cognitive scores were compared with the scores in the standard score sheet in the manual [17]. For all children, developmental age equivalents were calculated for each of the five domains from raw scores [17]. An average developmental age was calculated from the developmental age equivalents of the five domains. To calculate the average difference in the developmental age, the chronological age of the child was subtracted from the developmental age equivalent. BSID (III) is not validated for Indian children.

Children exhibiting mild, moderate, and severe delay in all three domains, namely motor, language and cognitive domains according to the composite score on BSID (III) were classified as exhibiting global developmental delay which was the outcome variable.

For the study, the socioeconomic class was defined according to revised Kuppuswamy’s socioeconomic status scale [18]. Overcrowding was considered to be present if two persons, above 9 years of age, of opposite sexes, not husband and wife, had to sleep in the same room [19]. Low birth weight (LBW) was defined as a birth weight of less than 2500 g (up to and including 2499 g) [20]. Breastfeeding was considered exclusive if the infant has received only breast milk from his/her mother or a wet nurse, or expressed breast milk, and no other liquids or solids. Severe anemia was defined as hemoglobin less than 7gm/dL. Diarrhea was defined as three or more liquid or semi-liquid stools [i.e. able to take the shape of a container] during the past 24 hrs [21].

The sample size was calculated using Stata^®(Version 12.0 Statacorp Texas, USA) for one-sample comparison of the population proportion. To calculate the sample size, a prevalence rate of 2.5% for developmental delay was used based on a community study in south India [22].This study was used to calculate sample size as there was no other study reporting delay in all the developmental domains in SAM.Assuming a 10% confidence interval around the proportion of 2.5%, two-sided alpha of 0.05 and power of 90%, the estimated minimum sample size was 85.

Statistical analysis was done using Stata^® (Version 12.0 Statacorp Texas, USA). The Shapiro–Wilk test and normal quantile plots were used to determine whether continuous variables were normally distributed. Continuous variables are presented as means and SD, while categorical variables are presented as absolute and percentage values. Two-sample t tests and a Mann-Whitney rank-sum test (in the case of nonnormally distributed variables) were used. Pearson Chi-square or Fisher’s exact tests were used to assess differences in proportions.

Pearson Chi-square test was used to test for each independent variables association with the outcome-global developmental delay. Stepwise multivariate logistic regression with backward elimination was used to develop the final model. In bivariate analyses, a P value of <0.1 was considered significant for entry into the model. The analysis was started with the full model, then the variable with highest P value was removed, and model was revised and refitted with remaining predictors. The procedure was repeated till the P value of all the predictor variables was less than 0.1 except age and sex; we designated this as final model. Adjusted odds ratios (aOR) and their respective 95% confidence intervals (CI) were then calculated from the final model and a P value of <0.05 was considered significant. We used the Bonferroni-Holm method to correct for multiple testing [23]. For the final model, model discrimination was done using C-statistics-receiver-operating-characteristics (ROC) curve, while model calibration was done using Hosmer-Lemeshaw “goodness-of-fit” test [24–26].

The study was approved by the Institutional Ethics Committee of R.D. Gardi Medical College, Ujjain (Approval number-458/2014). A written informed consent was obtained from the parents of the children included in the study.

During the study, the BSID (III) was done by two trained paediatricians simultaneously. The inter-rater reliability coefficient was excellent (intraclass correlation coefficient (ICC) = 0.85–0.92). High internal validity was achieved by Cronbach’s alpha coefficient for BSID (III) of 0.87. ICC and Cronbach’s alpha coefficient were obtained because BSID (III) is not validated for Indian children.

Out of the 104 children available for inclusion, 100 were analyzed. A flow chart of children included in the study is shown in Fig 1.

Admitted children met the following inclusion criteria: weight for height/length <-3 SD (n = 87), MUAC < 11.5 cm (n = 57), visible wasting (n = 18), and bipedal edema (n = 7). All children exhibiting visible wasting (n = 18) fulfilled at least one other inclusion criteria. Of the 100 children with SAM included in the study, 90% were between 6 and 24 months, and 52% were girls with an equal proportion from rural and urban areas. Most (95%) mothers were illiterate or received only primary grade education, whereas 90% fathers received education till primary and middle school. Majority (58%) of mothers were nonworking, and majority (87%) of fathers were unskilled or semiskilled workers. The socioeconomic status according to modified Kuppuswamy classification showed that most (67%) children belonged to the upper-lower class (Table 1). No children belonged to upper and upper middle classes, which are the top two tiers of the socioeconomic status.

Approximately one-third (36%) of the admitted children weighed less than 2.5 kg at birth (LBW). The feeding pattern of majority (78%) showed that they were exclusively breastfed for the first 6 months of age. Complimentary feeds were initiated by 6 months of age only in 45% along with continued breastfeeding. The quantity and frequency of complimentary feeding was ≤3 feeds per day in approximately60% of children with SAM (Table 2).

The examination and anthropometry findings revealed that 87% of patients were underweight, 39% were stunted, and 87% were wasted, and 18% exhibited visible wasting. General examination revealed severe anemia (Hemoglobin less than 7g/dL) in 67% children, hair changes in 56%, skin changes in 25%, and edema in 7% (Table 3).

Developmental delay assessment of children with SAM revealed that 75% exhibited delay in motor, 75% in language, and 63% in cognitive domains according to the composite scores (Fig 2). The overlap between the three domains is shown in Fig 2.

A total of 42 children exhibited delay in all three domains. The mean and 95% CI of the motor, language, and cognitive composite scores were 78.06 (95% CI 75.29–80.82), 78.22 (95% CI 75.89–80.54), and 83.97 (95% CI 81.87–86.06), respectively. The average difference in the overall developmental age ranged from +3 to –14 months. Overall developmental delay assessment revealed that 63% children exhibited delay by an average of 4–7 months in the total developmental age. A total of 60% of children exhibited delay by an average of 6–9 months inexpressive language, 49% by an average of 4–6 months in receptive language, 48% by an average of 5–7 months in gross motor, 62% by an average of 3–6 months in fine motor, and 54% by an average of 3–5 months in cognitive domain. Thus, most children exhibited delay in the fine motor domain, but the quantum of delay was the highest in expressive language. No statistically significant difference was observed in the developmental scores of SAM children with or without edema.

Tables 1, 2 and 3 displays bivariate analysis results. Table 4 shows the outcome of multivariate logistic regression outcome of independent risk factors associated with delay in all the domains. An increased risk for global developmental delay was observed with LBW (aOR:18.06), having working mother (aOR:17.54), weight-for-age less than three SD (aOR:6.09) and presence of severe anemia (aOR:16.34).

ROC of the final model was 0.9382 showing an excellent model fit. The Hosmer-Lemeshaw test revealed that chi-square was 9.70 (P = 0.1378) showing good model calibration.

The study has some limitations. BSID (III) used in the present study is not been validated for Indian children. Therefore, we report ICC. High internal validity for the scale was achieved by Cronbach’s alpha coefficient of 0.87. The use of US normed scale in Indian children could potentially skew the scores toward the lower side. The sample size of the study was insufficient to ascertain independent risk factors for each of the domain (motor, language, and cognitive) separately. Certain environmental factors including water, sanitation, and hygiene characteristics have been shown to be associated with SAM, but information on these factors was not was not collected in the present study. Assessment at the time of admission to NRUs would have helped in comparing of the BSID (III) before–after scores but was not performed because of ethical issues.