A SPATIAL ANALYSIS OF CHILD HEALTH IN NORTH-EAST STATES, INDIA: EVIDENCE FROM NATIONAL FAMILY HEALTH SURVEY

Kh. Jitenkumar Singh 1 , Apoorva Nambiar 2 , Damini Yadav 3 , Swati Kadian 4 , Amandeep Kaur 4 , Kiran Saini 4 , Payal Kuchhal 4 and Diksha Kashyap 4 .  1. National Institute of Medical Statistics, ICMR, New Delhi. 2. M.Sc. Student, IIPS, Mumbai. 3. M.Sc. Student of Amity University, Noida. 4. M.Sc. Student, Kurukshetra University. ...................................................................................................................... Manuscript Info Abstract ......................... ........................................................................ Manuscript History


ISSN: 2320-5407
Int. J. Adv. Res. 5 (6), 1910-1921 1911 first month of life. Preterm birth, birth asphyxia and infections cause most newborn deaths. Child health status varies by both race and ethnicity, as well as by family income and related factors including educational attainment among household members [2,3]. Health risks to newborns are minimized by: quality care during pregnancy; safe delivery by a skilled birth attendant; and strong neonatal care: immediate attention to breathing and warmth, hygienic cord and skin care, and early initiation of exclusive breastfeeding. Globally, 30% of children under five are estimated to be stunted and 18% have low weight-for-height, and 43 million children are overweight. Optimal breastfeeding could save the lives of 1.5 million children under five every year. Nearly nine million children die every year from preventable diseases and infections. From one month to five years of age, the main causes of death are pneumonia, diarrhoea, malaria and measles. Over two million children die every year from preventable diseases in India. IMR in India is 63 deaths for every 1000 live births. Of these 47% of the deaths occur within the first week after birth. Malnutrition in childhood is also one of the biggest contributor in child mortality and estimated to contribute more than one third of all child deaths. Under-five year mortalities were 87.7% in Arunachal Pradesh, 85% in Assam and 70.5% in Meghalaya according to NFHS-3 (2005NFHS-3 ( -2006 [4] . Anaemic children perform worse on cognitive measures than non-anaemic children [5] . Measles is the largest cause of death among children which can be prevented by a vaccine. Tetanus in newborns remains a major problem Uttar Pradesh, Madhya Pradesh, Rajasthan, West Bengal, and Assam (UNICEF India). Data shows 24.5% of new mothers initiated breastfeeding in the hour after birth, 46.4% breastfed exclusively the first six months and 56.7% nursed beyond six months with the introduction of complementary food (NFHS-3). This results in more than 50% the child population below five being underweight. 44.9% of the children under three are stunted and 22.9% of children under three are wasted. Less than 50% of children receive full immunization during infancy. This number has been dropping significantly over the years indicating a problem with the universal immunization programme. Even though there are various government programmes and schemes to address the nutritional needs of school going children in the age group of 6-14 years for e.g. Mid-Day Meal (MDM) Scheme. The Integrated Child Development Scheme (ICDS), however, there is a need to examine these for their outcomes so as to ensure greater focus on child through improvements in existing schemes and suggest new holistic interventions. Some of the indicators of child health are full immunization, BCG, DPT, polio, measles, Hepatitis B, vitamin A, public health facility, stunting, wasted, underweight and child anaemia. The North-East states of India comprised of eight small states namely Assam, Arunachal Pradesh, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim and Tripura. The region is physically isolated from the rest of the country due to mountains terrain and poor infrastructure and inhabited by numerous tribal and ethnic groups, has diverse sociocultural practices. These indigenous people live mostly in rural areas and mountainous terrain in traditional thatched roof huts without basic amenities. The lack of basic amenities and diverse sociocultural practices is directly or indirectly responsible for the child health in North-East region. Geographical variations in the child health can reveal inequities between and within states. In North East states, immunization coverage is very poor and prevalence of underweight among children is still high. The purpose of this paper is to study the spatial patterns of child health in order to understand the prevalence and patterns of the various child health indicators. Hence, the aim of the study is to visualize the spatial patterns of child health across the districts in North-East states. And to identify areas with statistically significant clustering of high values (hot-spots) or low values (cold-spots), as well as spatial outliers.

Data and Methods:-
The present study was based on a publicly available National Family Health Survey (NFHS-4), 2015-16 fact sheets data. NFHS-4, the fourth in the NFHS series, provides information on population, health and nutrition for India and each state/ Union territory. NFHS-4 was conducted under the stewardship of the Minister of Health and Family Welfare, coordinated by the International Institute for Population Sciences, Mumbai. The survey covered a range of health-related issues, including child health, maternal, fertility, malaria, reproductive health, infant and child mortality, non-communication diseases and HIV knowledge. The NFHS-4 fact sheets provide information on key indicators of all the districts. And helps to produce reliable estimates of most indicators for rural, urban and total of the districts as a whole [6] . We obtained fact sheet data for selected child health indicators of 87 districts of 8 states in North-East states viz., 27 in Assam, 16 in Arunachal Pradesh, 9 in Manipur, 7 in Meghalaya, 8 in Mizoram, 12 in Nagaland, 4 in Sikkim and 4 in Tripura, respectively.

Spatial Analysis:-
Spatial analysis refers to "a general ability to manipulate spatial data into different forms and extract additional meaning as a result". Specifically, spatial analysis comprises a body of techniques "requiring access to both the locations and the attributes of objects" [7] . Spatial statistics quantify geographic variation in geographic variables, 1912 and it can identify violations of assumptions of independence required by many epidemiological statistics; and measure how populations, their characteristics, covariates and risk factors vary in geographic space.
Spatial Autocorrelation:-Spatial autocorrelation analysis was applied to summarise the extent to which persons with a similar health status tend to occur next to each other i.e., form spatial clusters [8] . Spatial autocorrelation statistics depend on the definition of neighbourhood relationships through which the spatial configuration of the sampled subpopulation was defined prior to analysis. High or low values for a random variable tend to cluster in space (positive spatial autocorrelation) or location tend to be surrounded by neighbours with very dissimilar values (negative spatial autocorrelation). We used a binary weight matrix to assign weights to the neighbours. This binary weight matrix assigns a weight of unity for neighbours and zero for non-neighbours. The spatial patterns were investigated by global measures that allowed for spatial clustering tests. The present study used exploratory spatial data analysis (ESDA) techniques to measure the spatial autocorrelation among districts that are spatially contiguous. The first measure used in this study is global Moran's I, which gives an indication of the overall spatial autocorrelation of a dataset. The second measure is a local indicator of spatial association (LISA) measure of local Moran's I, which indicates the "presence or absence of significant spatial clusters or outliers for each location" in a dataset. High-high, low-low is positive autocorrelation and high-low, low-high is negative autocorrelation. Moran's I is defined as

Moran's I
where N is the number of spatial units indexed by I and j ; X is the variable of interest; ̅ is the mean of X; and is an element of a matrix of spatial weights.
The expected value of Moran's I under the null hypothesis of no spatial autocorrelation is ( ) Local Indicators of Spatial Association (LISA) statistics: The index used to observe spatial autocorrelation at local level is Anselin's LISA (Local Indicator of Spatial Autocorrelation), which can be seen as the local equivalent of Moran's-I. LISA essentially measures the statistical correlation between the value in subarea I and the values in nearby subareas. Univariate LISA statistics is used for the purpose which measures the extent of spatial nonstationary and clustering to its neighbourhood values.
∑ where observation z i , z j are in deviations from the mean from i th location to j th location and the summation over j such that only neighbouring values j  J i are included. And w ij is a spatial weight measuring the nearness of subareas i and j. For ease of interpretation, the weights w ij may be in row standardized form, though this not necessary and by convention, w ij =0.. LISA values close to zero indicate little or no statistical association among neighbouring values. There is a non-significant autocorrelation that is valued as (Moran's I =0.327). Full immunization has high-high spatial association in Dhemaji district of Assam and Imphal East, Imphal West and Thoubal district of Manipur whereas low-low in West and South district of Sikkim. Figure 2.1 depicts the district wise coverage among children aged 12-23 who had received BCG, where green color stands for the highest and red for the lowest coverage. In the coverage, the lowest percentage was observed in Upper Siang district of Arunachal Pradesh. There is a significant autocorrelation that is valued as (Moran's I = 0.290). BCG has high-high spatial association at Dhemaji and Sibsagar districts of Assam and Imphal West district of Manipur whereas low-low spatial association in West and North district of Sikkim. Figure 3.1 depicts the district wise coverage among children aged 12-23 who had received three doses of DPT, where green color stands for high and red color for low coverage. In the coverage, lowest percentage was observed in Upper Siang district of Arunachal Pradesh and Longleng district of Nagaland. There is non-significant spatial autocorrelation seen at the regional level (Moran's I= 0.305). DPT3 has high-high spatial association in Dhemaji distrct of Assam and Iimphal East and Thoubal districts of Manipur whereas low-low in West and South districts of Sikkim.

Polio 3:-
1914 Figure 4.1 depicts the district wise coverage among children aged 12-23 who had received three doses of Polio vaccine, where green color stands for highest and red for lowest coverage. In the coverage, lowest percentage was observed in Upper Siang district of Arunachal Pradesh. There is a non-significant positive spatial autocorrelation seen at the regional level (Moran's I=0.2869). Polio has high-high spatial association in Imphal East and Imphal West districts of Manipur and low-low spatial association in West and South districts of Sikkim.    There is non-significant positive spatial autocorrelation seen at the regional level (Moran's I= 0.299). Measles has high-high spatial association at Dhemaji district of Assam and Imphal East and Thoubal of Manipur whereas lowlow spatial association at West and South districts of Sikkim.     There is non-significant positive spatial autocorrelation seen at the regional level (Moran's I=0.377). Hepatitis B has high-high spatial association in Tamenglong, Imphal west, Imphal East and Thoubal districts of Manipur. Hepatitis B has low-low spatial association in South and West districts of Sikkim.    Wasted:- Figure 10.1 represents the district wise coverage among children under five years who are wasted, where green color stands for low and red color for highest coverage. In the coverage, the highest coverage was observed in South Garo Hills district of Meghalaya; Cachar district of Assam and Upper Siang district of Arunachal Pradesh. There is a nonsignificant positive spatial autocorrelation seen at the regional level (Moran's I= 0.219). Wasted has high-high spatial association in East Siang district of Arunachal Pradesh. Low-low association in Serchhip, Champhai districts of Mizoram; Senapati, Imphal East districts of Manipur and Kohima, Wokha of Nagaland.
Underweight:- Figure 11.1 depicts the district wise coverage of children under five years who are underweight, where green color stands for low and red color for high coverage. In the coverage, highest percentage was observed in five districts of Meghalaya; twelve districts of Assam and one district of Tripura and Arunachal Pradesh each. There is nonsignificant positive spatial autocorrelation seen at the regional level (Moran's I= 0.334). Underweight has high-high spatial association in Udalguri and Sonitpur districts of Assam and Tirap district of Arunachal Pradesh whereas lowlow association in Senapati and Imphal East of Manipur, Champai and Serchhip districts of Mizoram and in Wokha district of Nagaland.
Child Anaemia:- Figure 12.1 depicts the district wise coverage of anaemic children under five years, where green color stands for low and red color for highest coverage. In the coverage, the highest percentage was observed in South Garo Hills and East Garo Hills of Meghalaya; North and South districts of Sikkim and Changlang district of Nagaland. There is non-significant positive spatial autocorrelation seen at the regional level (Moran's I= 0.533). Child Anaemia has high-high spatial association in East Garo hills and South Garo Hills districts of Meghalaya whereas low-low in Kiphire, Zunheboto, and Wokha districts of Nagaland; Cachar district of Assam and Serchhip and Champhai districts of Mizoram.