|Year : 2021 | Volume
| Issue : 2 | Page : 147-151
Effect of feeding practices on nutritional status of infant and young children residing in urban slums of berhampur: A decision tree approach
Durga Madhab Satapathy1, Nivedita Karmee2, Sanjaya Kumar Sahoo3, Sithun Kumar Patro4, Debasish Pandit4
1 Professor and HOD, Department of Community Medicine, MKCG Medical College, Berhampur, Odisha, India
2 Associate Professor, Department of Community Medicine, MKCG Medical College, Berhampur, Odisha, India
3 Assistant Professor, Department of Community Medicine, MKCG Medical College, Berhampur, Odisha, India
4 Senior Resident, Department of Community Medicine, MKCG Medical College, Berhampur, Odisha, India
|Date of Submission||17-Dec-2020|
|Date of Decision||18-Feb-2021|
|Date of Acceptance||04-May-2021|
|Date of Web Publication||14-Jun-2021|
Department of Community Medicine, MKCG Medical College, Berhampur, Odisha
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: Appropriate infant and young child feeding (IYCF) practices in the early years of life will ensure optimal growth and development of the child. However, many children are not fed in the recommended way. Objectives: To assess the risk of malnutrition as a result of various feeding practice patterns among the children with the application of the decision tree algorithm. Methods: It was a community-based cross-sectional study conducted in the urban slums of Berhampur Municipal Corporation in Ganjam District, Odisha, India, from January to December 2019. Among a sample of 360 children of 6–23 months, nutritional status and feeding practices were determined. Data were analyzed using R version 3.6.1 developed by R Foundation for Statistical Computing, Vienna, Austria. The effect of IYCF practices on nutritional status was explained with the decision tree method with the use of a Chi-squared automatic interaction detection algorithm. Results: The prevalence of children with early initiation of breastfeeding (EIBF), exclusive breastfeeding (EBF), minimum meal frequency (MMF), and minimum dietary diversity (MDD) was 62.2%, 59.7%, 41.9%, and 19.4%, respectively. The prevalence of wasting, stunting, and underweight among the participants was 36.4%, 31.1%, and 35.3%, respectively. The significant factors which classified and predicted wasting were EBF, EIBF, and MDD, for stunting factors were EBF, MMF, and MDD and for underweight, significant factors were EBF, EIBF, and MDD. Conclusion: With the decision tree approach, the probability of malnutrition in relation to various feeding practices patterns can be easily explained to the mothers and health workers as compared to interpreting odds ratio and strict adherence to IYCF guidelines can also be ensured.
Keywords: Decision tree, infants, infant and young child feeding, malnutrition, slums, young children
|How to cite this article:|
Satapathy DM, Karmee N, Sahoo SK, Patro SK, Pandit D. Effect of feeding practices on nutritional status of infant and young children residing in urban slums of berhampur: A decision tree approach. Indian J Public Health 2021;65:147-51
|How to cite this URL:|
Satapathy DM, Karmee N, Sahoo SK, Patro SK, Pandit D. Effect of feeding practices on nutritional status of infant and young children residing in urban slums of berhampur: A decision tree approach. Indian J Public Health [serial online] 2021 [cited 2021 Aug 2];65:147-51. Available from: https://www.ijph.in/text.asp?2021/65/2/147/318355
| Introduction|| |
The first 2 years of life are the “critical window of opportunity." The adaptation of correct infant and young child feeding (IYCF) practices from the beginning will ensure children's optimal growth and development. However, many children are not fed in the proper way. Although breastfeeding is now universally practiced, many mothers start breastfeeding lately or who initiate the breastfeeding satisfactory, often start complementary feeding early or breastfeeding stopped within a few weeks of life. Many children have sub-optimal complementary feeding in terms of meal frequency and dietary diversity.
Annually 45% of child deaths are due to undernutrition. Adequate nutrition during early childhood, especially in the first 2 years of life, results in a reduction in morbidity and mortality. It is also related to the cognitive development of the child. Optimal IYCF practice helps prevent the development of chronic diseases in adulthood.
The snapshot of IYCF practices in urban India is not promising. Only two-thirds of the children (67.6%) were breastfed within 1 h of birth, 64.3% of children were exclusively breastfed, and regarding complementary feeding only 9.4% of children receiving adequate diet. Similarly for under-five nutritional status, more than one-third of the world's malnutritional children live in India. The prevalence of wasting, stunting, and underweight in urban areas is 27.2%, 17.0%, and 26.2%, respectively.
At present, the Indian population is experiencing a triple state of transition, i.e., demographic, economic, and epidemiological transition. The transition phases resulted in rapid urbanization which gave rise to the establishment of slums. The poor living conditions imposed a great risk of malnutrition among children. Besides, poor feeding practices among infants and young children enhance the risk of malnutrition further.
Therefore, to address the problem, the present study was conducted to assess the risk of malnutrition as a result of various feeding practice patterns among the children residing in urban slums. The decision tree method was applied to classify the participants based on their feeding practices and weigh the risk of malnutrition among various sub-groups.
| Materials and Methods|| |
This was a community-based, cross-sectional study conducted in the urban slums of Berhampur Municipal Corporation (BeMC) which is located in Ganjam District, Odisha. BeMC has 174 notified slum areas distributed in 37 wards. The study was carried out for a period of 1 year from January to December 2019. The study participants were the children age 6–23 months residing in the urban slums for 1 year. The sample size was determined using the following formula Z2 PQ/l2, where, Z = 1.96 (Z value for 95% level of significance), P = 64.3% (Prevalence of exclusively breastfeeding in Odisha as per NFHS-4 survey), Q = 35.7% (100-P), and l = 5% (Absolute precession) and it was found to be 353.Finally a total of 360 participants were selected for this study.
The average number of children of aged 6–23 months residing in each slum was found to be 14 + 3.7. Hence, considering availability of 10 children of aged 6–23 months in each slum area, the participants were selected by two-stage simple random sampling. First, out of 174 notified slums areas, 36 slums were randomly selected and from each, 10 participants were selected randomly. Children whose parents were not willing to participate in the study and children with a congenital malformation, mental retardation, and critically ill were excluded from the study.
Sociodemographic characteristics and infant feeding practices were recorded using a predesigned and pretested modified IYCF questionnaire by 24 h of dietary recall method. The nutritional status of the children was assessed by calculating Z-scores for height for age, weight for age, and weight for height and comparing with WHO Child Growth Standards. The feeding practices of the infant and young child were assessed with the help of IYCF core indicators. Data were tabulated in Microsoft excel 2016 and analyzed using R version 3.6.1, R Foundation for Statistical Computing, Vienna, Austria QGIS version 2.18, OSGeo Foundation, Chicago, USA. The quantitative variable was expressed in mean and standard deviation (SD), and the qualitative variables were expressed in proportion and frequency. The effect of IYCF practices on nutritional status was explained with the decision tree method. The algorithm used was Chi-squared automatic interaction detection (CHAID). CHAID is a statistical method to create homogeneous groups based on the value of a specific outcome variable (wasting, stunting, and underweight) by splitting cases into two or more groups on the basis of designated predictor variables (Feeding practices). After each split, the resulting groups are evaluated separately to see if a further split on any of the predictor variables would create significantly more homogenous groups. When it is no longer possible to make the resulting groups more homogenous (at a significance level of a = 0.05), the program halts. These final groups are called “terminal nodes,” and represent the most homogenous groups that can be created given the predictor variables and specified a level. CHAID is a particularly useful technique when a study is exploratory rather than confirmatory in nature, involves the relationships between a number of independent variables and a single dependent variable, when the independent variables interact with each other, and when there is no strong theory concerning the relative importance of the independent variables in predicting the dependent variable. The CHAID analysis was run in duplicate with parent nodes defined as 20 participants, child node defined as 5 participants, and significance set at (αmerge, αsplit, and P < 0.05).
Ethics clearance was obtained from the Institutional Ethical Committee of M.K.C.G. Medical College and Hospital to conduct the study. The data collection was conducted only after obtaining written consent from the parents of the study participants.
| Results|| |
A total of 360 children were included in the study; among them 52.5% were male and 47.5% were female. The mean age of males was 13.2 ± 4.68 months and for females, it was 12.83 ± 4.51 months. The number of participants aged 16–23 months was 30.8% followed by 12–15 months (30.3%).
About 62.2% of the mothers had initiated breastfeeding within 1 h of birth i.e. early initiation of breastfeeding (EIBF). Similarly, 59.7% had practiced exclusive breastfeeding (EBF) for 6 months. Regarding complementary feeding, among the infants, 51.1% had received complementary food at the age of 6–8 months i.e. timely initiation of complementary feeding, 41.9% of the babies were taking minimum prescribed frequency of meal as per IYCF guideline minimum meal frequency (MMF), 46 and the dietary diversity of 4 or more food groups in their meals was found among 19.4% i.e. minimum dietary diversity (MDD). However, the practice of minimum acceptable diet was seen in only 9.7% of the participants [Table 1].
|Table 1: Prevalence of various feeding practices and malnutrition among the study participants (n=360)|
Click here to view
In the present study, 31.1% of the children were found to have Z-score of height <-'2 SD for their age (Stunting), 35.3% had a Z-score of weight <-'2 SD for their age (Underweight), and 36.4% children had a Z-score of weight <-'2 SD for their height (Wasting) [Table 1].
Decision tree evaluation
The decision tree algorithm was used to statistically classify the participants into various subgroups based on their feeding practices (EIBF, EBF, MMF, MDD, and TIBF) and predict the risk of occurrence of stunting, wasting, and underweight among the subgroups formed.
EBF is the most powerful factor (Chi-square = 45.60, P < 0.001) for classifying participants into subgroups and predicting wasting [Figure 1]. The probability of occurrence of wasting was maximum among non-EBF subgroups (i.e., 0.54), and those subgroups who were exclusively breastfed, EIBF further subgrouped and determined (Chi-square = 17.09, P < 0.001) the probability of wasting. The probability of wasting among EBF children but delayed initiation of breastfeeding was found to be 0.41. The subgroups with children having EIBF and EBF, MDD (Chi-square = 5.23, P < 0.02) is the classifying and determining factor to assess the probability of the wasting. The probability of wasting was the lowest in the subgroups with EBF, EIBF, and MDD (i.e. 0.04) as compared to those who did not have MDD in their meals.(i.e., 0.19).
|Figure 1: Decision Tree: Probability of Stunting among the study participants wrt to their feeding practices, MMF: Minimum Meal Frequency, EBF: Exclusive breastfeeding, MDD: Minimum dietary diversity.|
Click here to view
EBF is the most powerful factor (Chi-square = 11.64, P = 0.001) for classifying the participants into primary subgroups and for predicting stunting [Figure 2]. In the subgroup of children with non-EBF, MMF (Chi-square = 7.98, P = 0.005) helps in further classifying and determining the probability of stunting. The probability of stunting among non-EBF and non-MMF was 0.53 and MMF was 0.29. However, in subgroups of children with EBF, MDD helps in classifying further and determining (Chi-square = 4.72, P = 0.03), the probability of stunting. The probability of stunting among those with EBF and without MDD was 0.29 and those with MDD were 0.14.
|Figure 2: Decision Tree: Prediction of Underweight among the study participants wrt to their feeding practices, EIBF: Early initiation of breastfeeding, EBF: Exclusive breastfeeding, MDD: Minimum dietary diversity.|
Click here to view
EBF is the most powerful factor (Chi-square = 33, P < 0.001) for the classification of participants based on their feeding practices and prediction of underweight [Figure 2]. Further, in the subgroup with children who were nonexclusively breastfed, EIBF (Chi-square = 5.7, P = 0.017) determined the probability of underweight, and those with EBF, MDD and EIBF classified and determined the probability of underweight. The probability of underweight is the highest among children with non-EBF and delayed initiation of breastfeeding (i.e., 0.62) and is lowest among children having EBF, EIBF, and MDD (i.e., 0.1).
The timely initiation of complementary feeding indicator was not found significant while classifying the sample population and prediction of probability for wasting, stunting, and under-weight.
| Discussion|| |
Malnutrition in the first 2 years which is a crucial period of life can be prevented by early and EBF followed by adequate, safe, and age-appropriate complementary feeding.
Regarding IYCF practices, 62.2% of children were breastfed within 1 h of birth (EIBF) and 59.7% were exclusively breastfed up to 6 months (EBF). This finding was slightly lower as compared to the NFHS-4 survey for Odisha where the proportion of EIBF and EBF for the urban population was 67.6% and 64.3%, respectively. However, EIBF practices among our study participants were much higher than the study conducted by Davara et al. and Bhushan in a similar set up where the proportion of EIBF was 45% and 39.6%, respectively.
The finding from the study conducted by Mog and Datta showed the proportion of EBF in the urban slums of the Tripura district was similar to our study, i.e., 60.5%. However, a study conducted by Panigrahi and Sharma and Chakraborty et al. showed the EBF among the slum areas of Bhubaneswar and Durgapur was lower as compared to our study, i.e., 21.2% and 27.6%, respectively.
In this study, the proportion of children who had MMF as per their age was found to be 41.9% and the proportion having dietary diversity in their meal was found to be 19.4%. Similar findings were observed in the study conducted by Tegegne et al. where the MMF among the study participants was found to be 49.6%; however, the study conducted by Chaudhary et al. in the urban slums of Ahmadabad, the MMF was found to be 64.3%.
The nutritional status of the children was assessed and it was found that the proportion of stunting, wasting, and underweight in the urban slums of Berhampur were 31.1%, 36.4%, and 35.3%, respectively. The proportion of malnutrition in our study was higher as compared to the NFHS-Odisha survey where the proportion of stunted, wasted, and underweight children in the urban areas was 27.2%, 17.0%, and 26.3% respectively. However, a study conducted by Houghton et al. in the urban slums of Delhi found that among the children residing in slums areas, 39% were stunted, 31% underweight, and 10% wasted and another study conducted by Lohia and Udipi in the urban slums of Mumbai found that among participants, nearly 51.3% were stunted, 26.7were underweight, and 41.7% were wasted.
Proper child feeding practices, especially breastfeeding and complementary feeding are two important interventions that can reduce acute malnutrition along with stunting in the early years of childhood. In our study “decision tree” approach, a classification-based model was used to subgroup the participants based on their feeding practices and predict the probability of undernutrition such as wasting, stunting, and under-weight among the sub-groups formed. The benefit of using this model is that the tree formed is easy to interpret and make others understand the impacts of feeding practices on the nutritional status of children at the village level.
EBF, EIBF, and MDD were the significant factors that classified the study participants statistically (P < 0.05) and predicted the probability of wasting in the [Figure 3] of the present study. A study conducted by Bentley et al. also showed the odds of wasting were lower among those with EIBF and MMF and MDD in their meals. Another study conducted by Chaudhary et al. found the significant negative association between wasting and EBF, EIBF, and MMF but no association with MDD. Sheikh et al. in their study found that the odds of having wasting were 0.22 times less likely for a child who received MDD and MMF, respectively. In our study, EBF was the principal factor that determined the probability of wasting among our study participants. The probability of wasting among the non-EBF children was found to be 0.54, whereas the subgroups with EBF but non-MDD the probability was 0.41, and subgroups with EBF and MDD the probability was minimal, i.e., 0.19. This indicates the better the nutritional practices as per WHO IYCF guideline, the lesser will be the risk of wasting among the children.
|Figure 3: Decision Tree: Probability of wasting among the study participants wrt to their feeding practices, EIBF: Early initiation of breastfeeding, EBF: Exclusive breastfeeding, MDD: Minimum dietary diversity.|
Click here to view
Similarly, EBF, MMF, and MDD were the significant factors (P < 0.05) which classified the participants into two subgroups and predicted the probability of stunting among them. EBF was the primary node to determine the stunting and subsequent nodes were MMF and MDD in different groups. The probability of stunting was the highest among the subgroups with non-EBF and non-MMF (0.53) and lowest for the children who were EBF and had MMD in their diets. The study conducted by Uwiringiyimana et al. found the significant negative association of stunting with EBF. Similarly, the study conducted by Appiah et al. reported a significant relationship between stunting and MMF among the children aged 6-'23 months.
For underweight, EBF, EIBF, and MDD were the significant factors (P < 0.05) which classified and predicted the probability of underweight among the study participants. A study conducted by Hashmi et al. reported that there was a significant association between underweight and MMF in their study participants. In our study, EBF was the principal node that classified the participants into two sub-groups. The sub-group with non-EBF, EIBF was the determinant factor whereas the sub-group with EBF, both EIBF and MDD were the determinant factors for underweight. The probability of underweight was the highest among those having delayed initiation of breastfeeding and nonexclusively breastfed children (0.62) and lowest for the sub-group having EIBF, EBF, and MDD practices (0.1).
| Conclusion|| |
With the decision tree approach, the outcome and probability of malnutrition concerning the subgroups formed by various feeding practices can be easily explained. This method of presentation will help the mothers to understand the pattern of IYCF practices and the possibility of malnutrition in their baby as compared to interpreting the odds ratio. The risk of malnutrition among the children of urban slums can be addressed by strict adherence to the IYCF guidelines and continuous monitoring of the feeding practices by health workers.
The author would like to acknowledge the study participants who spend their valuable times and for full cooperation. The author would also like to acknowledge the ASHAs, USHAs, and ANMs of BeMC who coordinate the participants during the study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
WHO. Infant nutrition. In: Infant Young child Feed Model Chapter Textb Med students allied Heal Prof. WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland WHO; 2012. p. 624-36.
Sethi V, de Wagt A, Bhanot A, Singh KD, Agarwal P, Murira Z, et al.
Levels and determinants of malnutrition among India's urban poor women: An analysis of Demographic Health Surveys 2006 and 2016. Matern Child Nutr 2020;16:1-4.
Park K. Textbook of Preventive and Social Medicine. 23rd
ed. Jabalpur: Banarsidas Bhanot; 2015.
WHO. Indicators for assessing infant and young child feeding practices. World Heal Organ 2007;2007:1-9.
Biggs David, De Ville Barry SE. A method of choosing multiway partitions for classification and decision trees. J Appl Stat 1991;18:49-62.
Assink M, van der Put CE, Stams GJ. The development and validation of an actuarial risk assessment tool for the prediction of first-time offending. Int J Offender Ther Comp Criminol 2016;60:847-64.
Davara K, Pandya C, Chavda P, Solanki D, Mehta K, Shringarpure K. Feeding practices by mothers having children under 6 months of age – A community based study in urban slum areas of Vadodara City. Natl J Community Med 2019;10:86-90.
Bhushan DR. Breastfeeding practices in slums of Delhi, India: A cross-sectional study. J Med Sci Clin Res 2019;7:72-9.
Mog C, Datta A. Prevalence and determinants of exclusive breastfeeding among lactating mothers in an urban slum, West Tripura: A cross sectional study. Int J Res Med Sci 2019;7:2818.
Panigrahi A, Sharma D. Exclusive breast feeding practice and its determinants among mothers of children aged 6–12 months living in slum areas of Bhubaneswar, eastern India. Clin Epidemiol Glob Heal 2019;7: 424-8. [doi: 10.1016/j.cegh. 2018.11.004].
Chakraborty SN, Roy SK, Rahaman MA, Ghose G. A community based study on pattern of breast feeding of under 6 months aged children in slums of an Eastern Industrial City of India. IOSR J Dent Med Sci 2015;14:61-4.
Tegegne M, Sileshi S, Benti T, Teshome M, Woldie H. Factors associated with minimal meal frequency and dietary diversity practices among infants and young children in the predominantly agrarian society of Bale zone, Southeast Ethiopia: A community based cross sectional study. Arch Public Health 2017;75:53.
Chaudhary SR, Govil S, Lala MK, Yagnik HB. Infant and Young Child Feeding Index and its association with nutritional status: A cross-sectional study of urban slums of Ahmedabad. J Family Community Med 2018;25:88-94.
Houghton LA, McIntosh DR, Trilok-Kumar G, Haszard JJ, Gibson RS. Suboptimal feeding and caring practices among young Indian children ages 12 to 24 mo living in the slums of New Delhi. Nutrition 2020;69:110553.
Lohia N, Udipi SA. Infant and child feeding index reflects feeding practices, nutritional status of urban slum children. BMC Pediatr 2014;14:1-11.
Bentley A, Das S, Alcock G, Shah More N, Pantvaidya S, Osrin D. Malnutrition and infant and young child feeding in informal settlements in Mumbai, India: Findings from a census. Food Sci Nutr 2015;3:257-71.
Sheikh N, Akram R, Ali N, Haque SR, Tisha S, Mahumud RA, et al.
Infant and young child feeding practice, dietary diversity, associated predictors, and child health outcomes in Bangladesh. J Child Health Care 2020;24:260-73.
Uwiringiyimana V, Ocké MC, Amer S, Veldkamp A. Data on child complementary feeding practices, nutrient intake and stunting in Musanze District, Rwanda. Data Brief 2018;21:334-42.
Appiah CA, Mensah FO, Hayford FE, Awuuh VA, Kpewou DE. Predictors of undernutrition and anemia among children aged 6–24 months in a low-resourced setting of Ghana: A baseline survey. J Heal Res 2021;35:27-37. [doi: 10.1108/jhr-05-2019-0095].
Hashmi AH, Nyein PB, Pilaseng K, Paw MK, Darakamon MC, Min AM, et al.
Feeding practices and risk factors for chronic infant undernutrition among refugees and migrants along the Thailand-Myanmar border: A mixed-methods study. BMC Public Health 2019;19:1-6.
[Figure 1], [Figure 2], [Figure 3]