Making Datasets Talk To Each Other

 

Objective
National averages obscure geographic variation in program performance. We determined Parliamentary Constituency (PC)-wise estimates of TB notification to guide political engagement.

Methods
We extracted district-level TB notification data from the 2018 annual TB report. We derived PC-level estimates by building a ‘cross-walk’ between districts and PCs using boundary shapefiles. We described the spatial distribution of the PC-wise estimates of Total Notification Rate and percentage of Private Sector Notification.

Results
The median PC-wise Total Notification Rate was 126.24/100 000 (IQR: 94.86/100 000, 162.22/100 000). The median PC-wise Percentage Private Sector Notification was 18.03% (IQR: 9.56%, 26.84%). Only 16 (2.94%) PCs met the target of 50% private sector notification. Most of high notification rates in PCs were driven by high notification in public sector. There was geographic – both interstate and within state inter-PC – variation in the estimates of these indicators. The study identified some geographic patterns of notification – high positive outlier PCs with adjoining PCs in lower deciles of notification rates, intra-state differences in PC performance, and similarities in notification rates of adjoining PCs in different states.

Conclusion
In addition to regional inequality, the study identified geospatial patterns that can aid in the formulation of suitable interventions. These include decongestion of overburdened facilities by strengthening poorly performing units. The PCs with a high percentage Private Sector Notification can act as role models for neighbouring PCs to improve private sector engagement. MPs can play a crucial role in mobilising additional resources, creating awareness, and establishing inter-PC and inter-state collaboration to improve TB program performance.

 

There are emerging opportunities to assess health indicators at truly small areas with increasing availability of data geocoded to micro geographic units and advanced modeling techniques. The utility of such fine-grained data can be fully leveraged if linked to local governance units that are accountable for implementation of programs and interventions. We used data from the 2011 Indian Census for village-level demographic and amenities features and the 2016 Indian Demographic and Health Survey in a bias-corrected semisupervised regression framework to predict child anthropometric failures for all villages in India. Of the total geographic variation in predicted child anthropometric failure estimates, 54.2 to 72.3% were attributed to the village level followed by 20.6 to 39.5% to the state level. The mean predicted stunting was 37.9% (SD: 10.1%; IQR: 31.2 to 44.7%), and substantial variation was found across villages ranging from less than 5% for 691 villages to over 70% in 453 villages. Estimates at the village level can potentially shift the paradigm of policy discussion in India by enabling more informed prioritization and precise targeting. The proposed methodology can be adapted and applied to diverse population health indicators, and in other contexts, to reveal spatial heterogeneity at a finer geographic scale and identify local areas with the greatest needs and with direct implications for actions to take place.

In India, districts serve as central policy unit for program development, administration and implementation. The one-size-fits-all approach based on average prevalence estimates at the district level fails to capture the substantial small area variation. In addition to district average, heterogeneity within districts should be considered in policy design. The objective of this study was to quantify the extent of small area variation in child stunting, underweight and wasting across 36 states/Union Territories (UTs), 640 districts (and 543 PCs), and villages/blocks in India. We utilized the 4th round of Indian National Family Health Survey (NFHS-4) conducted in 2015–2016. The study population included 225,002 children aged 0–59 months whose height and weight information were available. Stunting was defined as height-for-age z-score below 2 SD from the World Health Organization child growth reference standards. Similarly, underweight and wasting were each defined as weight-for-ageþinspace}< -2 SD and weight-for-heightþinspace}< -2 SD from the age- and sex-specific medians. We adopted a four-level logistic regression model to partition the total variation in stunting, underweight and wasting. We computed precision-weighted prevalence of child anthropometric failures across districts and PCs as well as within-district/PC variation using standard deviation (SD) measures. For stunting, 56.4% (var: 0.237; SE: 0.008) of the total variation was attributed to villages/blocks, followed by 25.8% (var: 0.109; SE: 0.030) to states/UTs, and 17.7% (Var: 0.074; SE: 0.006) to districts. For underweight and wasting, villages/blocks accounted for 38.4% (var: 0.224; SE: 0.007) and 50% (var: 0.285; SE: 0.009), respectively, of the total contextual variance in India. Similar findings were shown in multilevel models incorporating PC as a geographical unit instead of districts. We found high positive correlations between mean prevalence and SD for stunting (rþinspace}=þinspace}0.780, pþinspace}<þinspace}0.001), underweight (rþinspace}=þinspace}0.860, pþinspace}<þinspace}0.001), and wasting (rþinspace}=þinspace}0.857, pþinspace}<þinspace}0.001) across all districts in India. A similar pattern of correlation was found for PCs. Within-district and within-PC variation are the primary source of variation for child malnutrition in India. Our results suggest the importance of considering heterogeneity within districts and PCs when planning and administering child nutrition policies.