Background
Publicly financed health insurance schemes have been adopted by several low- and middle-income countries (LMICs) including India over the last decade, as one of the means of overcoming the challenges of the limited supply of publicly provided services and moving towards the goal of Universal Health Coverage (UHC) [
1‐
3].
The Ayushman Bharat- Pradhan Mantri Jan Arogya Yojana (PM-JAY), is India’s most recent national-level publicly financed health insurance scheme and was launched towards the end of 2018. PM-JAY has replaced the earlier existing Rashtriya Swasthya Bima Yojana (RSBY) and aims to expand financial protection to approximately 7000 USD per year per family, consequently also expanding the benefits package to include more secondary and tertiary treatments. The scheme also removed the earlier limit on the number of family members who could avail the benefits of the scheme and claims an eligible population of about 500 million beneficiaries from the lowest socio-economic strata [
4].
In assessing the effectiveness of RSBY and other earlier existing publicly financed health insurance schemes, mixed findings have been reported. Outcomes such as access to care, as measured through hospitalization rates, appear to have improved in populations covered under these schemes [
5,
6]. However, the impact on financial protection as measured through out-of-pocket expenditure and catastrophic headcounts was not significant [
6‐
10]. Studies exploring the institutional and governance arrangements of these schemes indicate challenges that may have limited the achievement of scheme objectives [
11‐
14]. Several of these challenges relate to the institutional arrangements for healthcare purchasing adopted in these schemes.
The trust and insurance models are two alternative institutional arrangements for healthcare purchasing that have been adopted in the publicly financed health insurance schemes in India, including PMJAY. In the trust model, the public payer, termed the State Health Agency (SHA) is registered as a not-for-profit trust and purchases services directly from empanelled providers. Implementation support agencies (ISAs) may be contracted to support the scheme administration functions of the trust [
4]. In the insurance model, the state health agency can contract an insurance company to insure beneficiaries and pay providers for the pre-defined list of services, in return for a fixed premium per beneficiary family unit covered. Here, insurers are responsible for authorizing treatments, processing claims and paying providers. Consequently, fraud detection and overall financial risk management are undertaken by the insurance company. Insurance companies may also further contract third party administrators (TPAs) to support their functions.
Studies on earlier schemes have indicated that when insurance company profits are linked to claim pay-outs, there are incentives to unnecessarily reject claims or empanel a limited number of hospitals, or hospitals with insufficient capacities to provide services [
12‐
14]. Such misalignments of incentives between insurance companies and the government, suggest that healthcare purchasing through an insurance company may limit the achievement of programme objectives that seek to expand access. On the other hand, arguments have also been made about the inefficiencies of government functionaries in administering large financing schemes within a trust, due to the lack of required expertise, inability to control supplier-induced demand, or to absorb risk [
13,
15], as also the fixed costs of setting up the state health agency.
The for-profit motives of private sector insurance companies to minimize claims payouts and maximize profit, are in conflict with the objectives of such schemes, which were to increase healthcare access and financial protection for the beneficiary population [
11,
14]. In analyzing the contractual arrangements between the public payer (central and state governments) and private entities (insurers who functioned as purchasers and the healthcare providers), both, design and implementation indicated limited oversight and stewardship of the scheme by government [
11,
12,
14]. Strategic purchasing frameworks emphasize the necessary stewardship of purchasing arrangements by governments, to ensure that purchasing decisions reflect public health priorities [
16,
17]. Whether purchasing arrangements include only state entities, or a mix of public and private agencies contracted to increase capacities, experiences of other LMICs, as well as countries such as China and Thailand that have made rapid strides in their UHC journey, indicate that the capacities and incentives of purchasing agencies, and their oversight, can be important determinants to effective purchasing [
18‐
21].
In the context of these observations, it is important to understand the experience of the trust and insurance models of purchasing services in the newly launched PMJAY, in which states have been provided the choice of model to be adopted for these purchasing actions. Very limited peer-reviewed literature is available on the performance of the scheme since its launch [
22‐
25]. This study originated based on requests from the policy designers and implementers of PMJAY, to provide early insights into the experiences of states adopting either model. Based on policymaker concerns around how states with either model were performing on the empanelment and claim management functions in this scheme, in light of adverse experiences with similar schemes as described above, we focussed on these two key purchasing functions. Our objectives were; a) To provide a descriptive analysis of the institutional agencies for provider contracting and claim management in either model b) To assess performance on provider contracting c) To assess performance on claim management d) To understand implementation costs; during the initial phases of implementation.
These findings would serve as guidance to policy makers on further decisions regarding purchasing functions. The study also adds to the existing literature on the experiences of LMICs with different purchasing arrangements under health financing schemes, targeted at large sections of the population.
Discussion
In this study, we aimed to provide insights into the trust and insurance models of healthcare purchasing in PMJAY, with respect to provider contracting, claim management and implementation costs, during the early phase of the scheme. We discuss here our main findings and offer some recommendations for policy and further research.
Our first main observation was that irrespective of the model adopted, many contracted resources form a part of the institutional structures implementing the scheme at the state and district levels. Key functions such as processing treatment pre-authorization requests and processing provider claims were carried out by insurance companies, implementation support agencies and resources provided to the government through development partners organizations, as technical support for these operations. While such arrangements provide the much-needed additional capacity for implementing large schemes, they also cause concerns about institution building, and the ability of the system to develop and retain the expertise required for sustaining operations of such financing schemes. Long-term partnerships and continuous oversight by the state health agency, would be required to continue with such arrangements.
One of the key elements of health protection schemes is the network of providers empanelled within the scheme. Under PMJAY, we did not find any significant differences in the criteria and processes for contracting providers adopted by the states with either model. A key observation with respect to provider contracting was that irrespective of the model, the State was the decision-making authority on contracting hospitals. This was in contrast to the erstwhile RSBY, where the insurance company was primarily responsible for empanelment.
We also observed that a fresh round of inspections had been conducted in these two states for contracting providers, rather than continuing with the RSBY empaneled hospitals. However, public hospitals were deemed empanelled, and this was likely to impact their readiness and ability to provide services under the scheme. Data on infrastructure and specialties of public hospitals had also not been maintained in the empanelment database. While good processes had been put into place for contracting private hospitals, there were some limitations. Due to the urgent need to empanel sufficient hospitals in a very little amount of time at the start of the scheme, some compromises might have been made on the quality of these hospitals. However, states were in the process of reviewing empanelled hospitals, to ensure that they continued to meet the necessary criteria and some had been consequently de-empaneled. This process needs to be periodically carried out to ensure adequate oversight of hospitals, by state health agencies.
Consequently, the observed variations in distribution and characteristics of empanelled private hospitals appeared to reflect the available infrastructure in the two states. Uttar Pradesh had a higher density of empaneled hospital beds per beneficiary family, as well as specialist services as compared to Jharkhand, with large variations in available bed strength across districts in both states. Although as per guidelines, states could relax some empanelment criteria for specialist services, we did not find either state had done this. More guidance to states on this aspect would be helpful, to enable them to address the skewed distribution of providers across districts. Complete data on empanelled public hospitals would serve to further assess the adequacy of empanelled hospitals in the scheme.
Both states empanelled a higher number of private hospitals, as compared to public hospitals. This possibly reflects the higher number of private hospitals estimated to be available in both states, as compared to public hospitals [
28,
29]. These trends are reflected at the national level, with both models showing a similar distribution of public–private mix of hospitals [
30]. Reliable data on private sector infrastructure in India is however lacking, rendering it challenging to estimate the effectiveness of the scheme in drawing participation from the large private sector that otherwise provides the major proportion of in-patient and out-patient care in India [
28,
31]. Mid-sized hospitals formed a larger proportion of all empaneled hospitals in UP, as compared to Jharkhand wherein small hospitals formed half of all empaneled hospitals. In terms of quality of care, overall, very few accredited hospitals were found to be participating in the scheme. These findings, therefore, appear to highlight the available health infrastructure in the states. Both Uttar Pradesh and Jharkhand report a scarcity of specialist healthcare providers in rural areas [
32]. These issues are not confined to the states we studied, but possibly highlight the persisting supply-side gaps in many states of India [
29,
33].
Provider payments were processed differently in the two models due to the involvement of the insurance company in Jharkhand, versus the implementation support agencies working with the state health agency in Uttar Pradesh. The oversight of the contracted agency was found to be much higher by the state health agency in Uttar Pradesh, wherein it conducted audits of all claims that were initially processed by the implementation support agencies, and levied significant financial penalties in cases of wrongful claim approvals by the agencies. This scrutiny was further reflected in longer turn-around times for provider reimbursement and repeated queries on the claims raised. Higher claim rejection rates were also reported by the trust in Uttar Pradesh than the insurance company in Jharkhand. These observations indicate two important issues.
Firstly, the higher claim rejections, queries and oversight of the implementation support agencies by Uttar Pradesh, reflect that the trust had the capacity to be vigilant with the implementation support agencies and hospitals, and thereby prevent wrongful claim approvals. This is in contrast to earlier experiences with public–private partnerships in health that have reported instances of the inability of government bodies to monitor their partners in the private sector, including ensuring rightful implementation of the contract terms and service delivery output [
12,
34]. We could not conclusively determine whether the higher vigilance was due to a cautionary approach taken by state officials, based on their earlier experiences of corruption reported under RSBY [
35,
36].
However, the second possible implication was that long turn-around times, numerous queries and high rejections due to incorrect documentation for claim reimbursements, also indicated weak capacities of hospitals in complying with these processes. We did not have any evidence of differential capacities of hospitals in Uttar Pradesh and Jharkhand to conclusively determine the reasons for higher rejection rates in Uttar Pradesh, as compared to Jharkhand. However, these observations merit further study, in order to prevent these challenges for hospitals and to improve claim processing efficiencies in the state.
We also found that the claim rejection rate in Uttar Pradesh was comparable to that of the trust in the Rajiv Arogyashri scheme of Andhra Pradesh in its initial year of implementation (4%), however, the rejection rate in the Rajiv Arogyashri scheme increased subsequently up to about 10% [
13,
37]. The trust in Karnataka’s publicly financed health insurance scheme reported a rejection rate of 12% [
37]. These rates, as reported by trusts in other states, were however always lower than those reported by insurance companies (in the range of 16%), which was in contrast to our observations [
13,
37].
Consequently, our observations on lower claim rejection rates by the insurance company in Jharkhand was unexpected since insurance companies have been previously seen to hold back payments to hospitals, as a means to minimize claim payouts, and thereby increase their profits [
6,
13]. However, the contract terms between the state health agency and insurance company in PMJAY imposed a ceiling on the administrative costs that insurance companies could retain that was linked to the claim ratios achieved. Unspent balances were to be returned to the state health agency. Insurance companies were allowed to retain a maximum proportion of premiums received as administrative costs if they achieved claims ratios of 70 to 85%, which was possibly incentivizing the insurance company, in this case, to achieve claims ratios in this range.
Jharkhand reported a claims ratio of 32.4% in the first six months of the scheme. Scheme utilization was higher in Jharkhand, as compared to Uttar Pradesh, despite Uttar Pradesh having double the number of eligible beneficiary families. The implementation support agencies in Uttar Pradesh had fixed payment terms, independent of the claim volumes processed. Several factors other than the models and purchasing agencies are likely to influence the utilization of the scheme in both states. A comprehensive assessment of these factors was beyond the scope of the objectives of this study. Our findings, however, indicate that the claim approving behavior of the insurance company could have been influenced by the incentives created by the payment terms in the contract, partially affecting utilization in Jharkhand. Such practices by insurance companies have been earlier reported in Rajiv Arogyashri wherein claims ratios somehow remained at the level at which administrative fees retained, were at the highest proportion permissible (these contracts had similarly designed payment terms) [
13]. However, these observations from PMJAY are early in the scheme implementation phase and would need to be repeated after the completion of one or more policy periods, in order to better understand the claim approving behaviour of the insurance company. A comprehensive assessment of the factors contributing to higher scheme utilization in Jharkhand as compared to Uttar Pradesh, also merits assessment.
Checks and balances for the prevention of supplier-induced demand are also dependent on appropriate validation of all pre-authorization and claim requests, while processing payments. For-profit providers are likely to indulge in such behavior irrespective of the type of purchaser, when the type of payment remains the same (a type of case-based payment referred to as a package rate in PMJAY) [
38]. The concern over the occurrence of supplier-induced demand remains, irrespective of whether the purchaser is a private insurance company or a public agency [
7,
13,
39]. While our study does not explore its occurrence, we found the lack of adequately qualified doctors on the teams of the contracted agencies processing claims in both models was likely to influence the effectiveness of processing provider payments, while preventing unnecessary treatments. This was reflected in our interviews with implementation support agencies, who expressed their inability to question treating doctors over their chosen treatment pathways. In Uttar Pradesh, the double auditing of claims by qualified doctors in the state health agency fulfilled this gap within implementation support agencies. However, national guidelines do not necessitate that state health agencies audit all claims; hence this may not be the case in other states adopting the trust model. Contracted agencies in both models, therefore, could not adequately evaluate the clinical decision-making of hospitals. These observations call for the timely uptake of standard treatment guidelines for treatment conditions available under the scheme. These guidelines would also help to improve the capacities of hospitals and prevent high claim rejections. Standard treatment guidelines have been recently notified by NHA and will have to be studied once sufficiently operationalized, with regards to their scope in supporting claim management processes [
40].
These observations also reiterate the vital role of oversight of purchasing agencies by the state, in both models. Enforcing contract terms to build accountability remains important to ensure that processes are followed, and is an important aspect of the governance of strategic purchasing in such schemes [
16,
17]. We found this oversight to be better ensured in the functioning of the trust model in our case studies. However, these observations need to be further explored in a larger number of states, and at later phases of scheme implementation. Such studies will remain important as the scheme grows and expands. Experiences in Nigeria and Malawi indicate that limitations in capacities of the purchasers, as well as inadequate oversight of these agencies by the state, contributed to ineffective purchasing in their publicly financed health insurance schemes [
19,
20]. Experiences with strategic purchasing in Thailand’s universal coverage scheme also report that penalties and incentives are one of the components contributing to the effectiveness of purchasing [
21]. Although China has made significant progress in health financing reforms, the country still faces problems with the capacities of the third-party purchaser in ensuring cost-effective care under its social health insurance system [
18].
Our study of initial implementation costs is limited in some ways, but provides useful early insights into possible implications of adopting either model in the scheme. We found the unit administrative costs in the trust model studied to be lower than in the insurance model. However, cost-effectiveness showed mixed results. Uttar Pradesh showed slightly better administrative cost-effectiveness in terms of beneficiaries enrolled, but Jharkhand reported better cost-effectiveness in terms of claims generated. These observations reflected Jharkhand’s higher utilization of the scheme, as discussed earlier. These observations exacerbate the need for a more comprehensive assessment of factors driving utilization in states, and the resulting cost-efficiencies of the models adopted. We could not compare our findings on the cost-effectiveness of the two models to others, due to differences in definitions and methods [
13]. However, costs would have to be assessed again when utilization stabilizes and in a larger number of states, to determine the cost-effectiveness of either purchasing model.
Our study is limited in that one state with either model was included. The study was commissioned at an early phase of the scheme, and hence several unknown implementation factors other than those reported could have been at play, to affect outputs reported in this study. Our cost data were also limited by the availability of complete data, as well as by the fact that we used partially estimated costs, rather than all actual expenditures due to the timing of this study. However, the states included are large and populous states, that are accorded policy priority in India. Insights into their early experiences with provider empanelment, claims management and indicative implementation costs were important to inform further policy and implementation decisions on an ambitious scheme such as PMJAY. The behaviors of institutions such as trusts and insurance companies in more states, with either model, need to be assessed at later stages of scheme stabilization to further inform scheme design, going forward. Until then, in both models, the state health agencies will have to ensure that capacities are built and retained, and that the incentives for the supporting agencies are continually assessed and re-designed, if needed, to ensure that scheme objectives are achieved.