Only in an alternate universe would one have thought of Indian governmental institutions as lean business enterprises underpinned by interoperable information systems and robust business processes. However, the times are changing. The inexorable march of data and technology has transformed almost every enterprise in its way. Amongst these, government agencies are indeed the most recent and significant beneficiaries. Owing to this shift, governments have gained enough ground in transitioning from Luddites to reluctant embracers to, finally, cause champions.
As a result, the entire gamut of data ranging from transactional to analytical data is being digitised, collected, and carefully stored within public run enterprises. Granular data describing daily operations and events are records of transactions that are carried out, which are essential to the formation of a detailed database. This, put together with pre-identified key metrics, measurements, and indicators used in assessing and optimising institutional decisions are of particular importance. The relevance of such data collection is more pressing than ever in a scenario where streamlining key processes, measuring impact and improving policy outcomes are pertinent.
The buck, however, does not stop there. Processing the data to glean insights, trends, and predictions is where the utility of data really shines through. The critical mass required to conquer the last bastion of the data lifecycle is still wanting in government organisations. Leveraging on this sort of a foregone opportunity cost has been Ank Aha’s primary raison d’être as an organisation, and to this end, in July 2017 we kick-started our engagement with one such quintessential establishment – Department of Information, Technology and Communication (DoIT&C) at Jaipur, Rajasthan.
Unknown to many, the Government of Rajasthan has been extremely proactive in digitizing policy processes and has put in place a robust virtual infrastructure to aid evidence-backed policy making. In collaboration with the Health Department, the DoIT&C designed the online portal for the Bhamashah Swasthya Bima Yojana (BSBY), a state-run health insurance scheme aimed at minimising out-of-pocket expenditure and revolutionising healthcare in rural areas. Budgeted at Rs. 370 crores per year on a floater basis, the scheme provides free-of-cost hospital in-patient care (IPD) to 1 crore families across the state, identified under the National Food Security Act (NFSA) and RSBY scheme, at empanelled public and private hospitals and clinics.
However, despite collecting granular transactional data for this flagship health insurance scheme, the role played by data is a passive one. Thus, on cue from the senior leadership at DoIT&C, along with my colleague Sreelakshmi, our work for this project entailed using smart and sophisticated data analytics to assess the impact and diagnose gaps in public health-related policies implemented by the state bureaucracy.
Our data analysis started with collating and cleaning the voluminous transactional data collected over the past 20 months. Following that, given the wealth of data and the limitless scope of analysis possible, our goal was to tackle an urgent yet relevant sub-issue within the health domain. After a few rounds of meetings and discussion with health officials and experts at the government, it was unanimously agreed upon to focus on maternal and neo-natal healthcare. Once the data was cleaned, the topic of analysis narrowed in on, and time reference frame selected (owing to implementation complexities), we were eager to explore the data and let it reveal beyond the anecdotal and the intuitive.
One of the main objectives of the BSBY is stated to be the provision of private sector facilities in rural Rajasthan. On that basis, it was imperative to study the distribution of care-giving burden between government and private hospitals across the state – made up of 33 districts that are clustered into 7 divisions. We found out that across all districts across Rajasthan, barring a few outliers, government hospitals were predominantly responsible for providing a much higher proportion of services than private hospitals.
In addition to Primary Health Centres (PHC), Community Health Centres (CHC) form the bedrock of a critical healthcare network in impoverished rural areas. The urgent nature of care required in maternal and neo-natal healthcare renders geographical proximity of the catchment population serviced and the healthcare facility as extremely important.
Based on this premise, we decided to explore the status and package uptake trends among CHCs for all the districts. However, our analysis revealed that there were a few glaring aberrations to this expectation: there were 32 CHCs across Rajasthan where not a single maternal and neo-natal healthcare package was availed over the timeframe considered. While, a direct implication of this finding necessitates further investigation of the on-ground realities at those CHCs, it also signals an implementation gap for both the BSBY and the centrally sponsored Janani Suraksha Yojana (JSY) due to beneficiary and benefit overlaps. Compounded by the absence of clear directives and guidelines for each scheme, it is very likely that institutional deliveries, which form a substantial portion of maternal health packages availed, were reported under JSY in lieu of BSBY – adding to erroneous data collection.
We also attempted to carry out innovative analytical enquiries aimed to study and demystify the interplay of complexmechanisms affecting the beneficiary level behaviour and institutional level patterns. Akin to the Gini coefficient which is the gold standard in quantifying income inequality, we calculated the burden of maternal and neo-natal healthcare-providing on the top percentile of hospitals in the division. Named as the burden-sharing co-efficient, this metric measured package distribution at an aggregate level, where a higher number signifies a more iniquitous distribution.
Furthermore, based on the performance of the top hospitals, we set out to examine the healthcare seeking patterns of beneficiaries with respect to their domicile geographic location. This analysis – Migration Analysis – was aimed at understanding the movement of beneficiaries across regions in order to seek maternal and neo-natal healthcare.
Beneficiary movement patterns for maternal and neo-natal healthcare unearthed not just the hotspot districts and hospitals where the population at large prefers availing healthcare, but also revealed and confirmed the spatial aspect of such distributional iniquities : remote districts tend to have sparse population and the distribution of healthcare facilities is heavily skewed. Thus, on this basis, by creating spatial interaction models that further explore the nuances associated with the manner in which beneficiaries seek healthcare can assist in making healthcare more accessible while keeping it affordable.
The intersection of Big Data and public service delivery is a niche that we at Ank Aha are constantly aiming to explore and push the boundaries of. From the wealth of data that was made accessible to us, the development of a machine learning predictive analytics tool to aid in optimal decision-making and allocation of resources is an exciting work in progress.
Look out for the next blog in this series to know more about how we are getting computer science closer to the policy space.
Aruj Shukla is a Data Associate at Swaniti Initiative