Advertisement

Utilization Trends of a Government-Sponsored Health Insurance Program in South India: 2014 to 2018

  • Preeti Singh
    Correspondence
    Correspondence: Preeti Singh, MPH, Max Institute of Healthcare Management, Indian School of Business, Mohali, Knowledge City, Sector 81, Sahibzada Ajit Singh Nagar, Punjab, India 140306.
    Affiliations
    Max Institute of Healthcare Management, Indian School of Business, Sahibzada Ajit Singh Nagar, Punjab, India
    Search for articles by this author
  • Adam C. Powell
    Affiliations
    Max Institute of Healthcare Management, Indian School of Business, Sahibzada Ajit Singh Nagar, Punjab, India

    Payer+Provider Syndicate, Boston, MA, USA
    Search for articles by this author
Open ArchivePublished:November 26, 2021DOI:https://doi.org/10.1016/j.vhri.2021.02.007

      Highlights

      • The initial impact of the insurance program for low-income citizens, the Rajiv Aarogyasri Scheme (RAS), which was implemented in Andhra Pradesh, India, has been well-studied. Previous researchers have demonstrated that the RAS increased health service use. Researchers have found that most RAS claims are for care delivered at private hospitals, a finding which was confirmed in this article during a later period.
      • This article extends the literature on RAS by characterizing how the value of care delivered and the level of access to care changed from 2014 to 2018. Whereas the nominal mean claim amount was found to have significantly increased over time, in real terms, it decreased after accounting for inflation. Mortality rates significantly decreased over the period examined, suggesting that better outcomes are now being delivered at a reduced cost.
      • The article additionally identifies which districts have increased and decreased their ability to more fully serve the healthcare needs of their residents.

      Abstract

      Objectives

      To characterize the utilization trends associated with the Aarogyasri health insurance scheme in Andhra Pradesh, India.

      Methods

      This is a retrospective cross-sectional study including participants enrolled in the Aarogyasri health insurance scheme, with recorded claims pertaining to inpatient care from quarter 3, 2014 through quarter 2, 2018. The main outcome measure, was annual utilization by service category, trended to characterize changes in the mean claim amount and the median length of stay. Mortality by service category was also trended. Mann-Kendall correlation was used to evaluate trends. Additionally, interdistrict migration for care in 2014 versus 2018 was examined to evaluate changes in access to care.

      Results

      The distribution of claims by caste significantly shifted over time, with members of backward castes and scheduled tribes filing more claims, and members of other castes and scheduled castes filing fewer claims. The median age of patients significantly increased, rising from 44.0 years in 2014 to 46.0 years in 2018. The nominal mean claim amount in 2018 was 105.4% of the 2014 average, but the 2018 real mean claim amount was 90.3% of the 2014 average. The median length of stay significantly decreased from 5 to 4 days. Mortality rates after procedures significantly decreased from 2.4% to 2.1%. Interdistrict migration to access care remained high among beneficiaries from the districts YSR Kadapa and West Godaveri in 2014 and 2018.

      Conclusions

      Over time, the value delivered by Aarogyasri improved. More patients received care at lower real per claim cost, with a concurrent decline in mortality.

      Keywords

      Introduction

      In India, a large proportion of health expenditure is paid out of pocket (OOP) by households.
      • van Doorslaer E.
      • O’Donnell O.
      • Rannan-Eliya R.P.
      • et al.
      Catastrophic payments for health care in Asia.
      ,
      • van Doorslaer E.
      • O’Donnell O.
      • Rannan-Eliya R.P.
      • et al.
      Effect of payments for health care on poverty estimates in 11 countries in Asia: an analysis of household survey data.
      Over the years, OOP expenditure on healthcare in India has been progressive. As a proportion of nonfood expenditure, high-income people spend marginally more on healthcare OOP than low-income people.
      • Devadasan N.
      • Ranson K.
      • Van Damme W.
      • Criel B.
      Community health insurance in India: an overview.
      However, due to lack of resources, low-income people often either forgo treatment or risk financial catastrophe.
      • Devadasan N.
      • Ranson K.
      • Van Damme W.
      • Criel B.
      Community health insurance in India: an overview.
      In this context, a publicly funded health insurance scheme was created to reduce the OOP burden of healthcare on low-income people.
      In April 2007, the Chief Minister of Andhra Pradesh, a state in southeastern India, introduced a government-sponsored health insurance program for people living below the poverty line. Originally known as the Rajiv Aarogyasri Community Health Insurance Scheme (RAS), the program was renamed Dr YSR Aarogyasri and remains active today. The program enables participants to receive inpatient care at both private and government hospitals. Although RAS is specific to Andhra Pradesh, there are similar programs in other parts of the country, including Karnataka’s Vajpayee Arogyasri Scheme, Maharashtra’s Rajiv Gandhi Jeevandayee Arogya Yojana, and Tamil Nadu’s Chief Minister's Comprehensive Health Insurance Scheme in Tamil Nadu.
      • Prinja S.
      • Chauhan A.S.
      • Karan A.
      • Kaur G.
      • Kumar R.
      Impact of publicly financed health insurance schemes on healthcare utilization and financial risk protection in India: a systematic review.
      ,
      • Garg S.
      • Chowdhury S.
      • Sundararaman T.
      Utilisation and financial protection for hospital care under publicly funded health insurance in three states in Southern India.
      RAS has largely been consistent in its benefit design throughout its tenure. At the end of the study period, in 2018, the scheme covered 986 procedures and enabled patients to receive tertiary care without providing any cash to empaneled hospitals. The public and private hospitals providing RAS-financed services were preselected if they were to offer covered services at set prices for each procedure and met eligibility requirements. The scheme provides financial coverage up to Indian Rupee (INR) 5 lakhs
      Although the nominal maximum amount of coverage remained the same over the course of the period of the study when measured in INR, the equivalent value in USD fluctuated because of changes in the exchange rate. At the beginning of the observation period on July 1, 2014, the exchange rate was 60.0 INR/USD. At the conclusion, on June 30, 2018, the exchange rate was 68.4 INR/USD. Thus, INR 5 lakhs was equivalent to USD 8333 at the beginning of the observation period and USD 7310 at the conclusion.
      per family per annum and pays healthcare providers on a case-wise basis at a predefined rate.
      Andhra Pradesh State Government
      Frequently asked questions.
      ,
      Although the nominal maximum amount of coverage remained the same over the course of the period of the study when measured in INR, the equivalent value in USD fluctuated because of changes in the exchange rate. At the beginning of the observation period on July 1, 2014, the exchange rate was 60.0 INR/USD. At the conclusion, on June 30, 2018, the exchange rate was 68.4 INR/USD. Thus, INR 5 lakhs was equivalent to USD 8333 at the beginning of the observation period and USD 7310 at the conclusion.
      The maximum amount of coverage provided is not indexed to inflation and has remained constant over time.
      In 2009, about 85% of the total population of the state was covered by RAS (approximately 20 million families and 70 million beneficiaries). The beneficiaries are enrolled in the scheme using their ration cards—an identification document issued by state governments in India to below-poverty-line (BPL) households. The BPL status is either determined by a multidimensional means testing or designated by a relevant village authority. BPL households are eligible for a variety of benefits, such as publicly subsidized grain from the Public Distribution System, in concordance with the National Food Security Act.
      Ministry for Consumer Affairs, Food and Public Distribution
      Introduction.
      While other authors have studied RAS in specific regions within Andhra Pradesh and during other periods, a statewide study of the scheme from 2014 to 2018 has not previously been conducted.
      • Prinja S.
      • Chauhan A.S.
      • Karan A.
      • Kaur G.
      • Kumar R.
      Impact of publicly financed health insurance schemes on healthcare utilization and financial risk protection in India: a systematic review.
      ,
      • Garg S.
      • Chowdhury S.
      • Sundararaman T.
      Utilisation and financial protection for hospital care under publicly funded health insurance in three states in Southern India.
      ,
      • Shaikh M.
      • Woodward M.
      • Rahimi K.
      • et al.
      Major surgery in south India: a retrospective audit of hospital claim data from a large community health insurance programme.
      • Shaikh M.
      • Peters S.A.E.
      • Woodward M.
      • Norton R.
      • Jha V.
      Sex differences in utilisation of hospital care in a state-sponsored health insurance programme providing access to free services in South India.
      • Rao M.
      • Kadam S.
      • Sathyanarayana T.N.
      • et al.
      A rapid evaluation of the Rajiv Aarogyasri community health insurance scheme in Andhra Pradesh, India.
      • Nagulapalli S.
      • Rokkam S.R.
      Should Governments engage health insurance intermediaries? A comparison of benefits with and without insurance intermediary in a large tax funded community health insurance scheme in the Indian state of Andhra Pradesh.
      • Madaboyina G.
      An evaluative study of Rajiv Aarogyasri health insurance scheme in Vizianagaram District of Andhra Pradesh.
      As such, how the value delivered by the program evolved during the end of the 2010s is unexplored. Since the borders of Andhra Pradesh shifted in June 2014 when it was divided to form a new state, Telangana, this study focuses on care delivered after the split occurred. The objective of this study was to examine how demographics of the people filing claims and the nature of the claims filed have changed over time. In addition, the article also compares the interdistrict migration of the claimants in 2014 and 2018. Trends in claims expenditures, length of inpatient stay, and mortality are computed for the entire population and by category of service.

      Methods

      Study Participants

      This is a retrospective longitudinal study using publicly available deidentified hospital claim data for all surgical procedures conducted between July 1, 2014 and June 30, 2018 under a government-sponsored health insurance scheme (RAS) across all 13 districts of Andhra Pradesh. Each claim record included: (1) date of authorization, treatment, discharge, and death; (2) the amount approved and paid; (3) details of the surgery performed (name, International Classification of Disease code); and (4) characteristics of the hospital that filed the claim (name, location, and type [public/private]). Details on patients’ demographics, such as age, sex, caste, and residence, were also reported for each claim.

      Sample Selection

      Tertiary care claims from July 2014 to June 2018 from all people from Andhra Pradesh who were enrolled in RAS and used services covered by the scheme were included in the analysis (the scheme does not cover nontertiary care). Duplicate claims were removed from the sample after it was verified that they were included in the database by error.

      Outcomes and Analysis

      The analysis characterized trends in nominal claim amount, length of stay, and mortality for all surgical procedures in 25 different categories from 2014 to 2018. The length of stay was calculated as the difference between surgery and discharge dates. For the analysis to best represent the demand from the government’s perspective, each claim was treated as an independent event. If a person had multiple procedures in the same or different categories resulting in multiple claims in the same year, each was counted separately and the impact that they potentially had on one another was not considered. The date of claim for each surgery was used to calculate the annual number of beneficiaries who availed services under the scheme. Values were computed annually to reduce the impact of seasonal variation and were expressed as means (accompanied by the standard deviation), medians (accompanied by the interquartile range), or percentages, as appropriate. Chi-square tests were used to evaluate the significance of changes in the distributions of categorical data.
      Price trends for the claims were determined for each category of services. To understand the shift of prices over time, the prices of surgeries performed across 705 empaneled hospitals (public and private) per category were trended over time and were normalized to 2014 prices. It is possible that the types of procedures driving the price trends observed within a category shifted over time, therefore, an additional set of price trends were calculated for the most common surgery in each category similar to the third quarter of 2014. The length of stay trends and mortality postprocedure trends were also calculated. Mann-Kendall correlation coefficients were determined using the raw data (nonaggregated values) to assess whether each of the trends was monotonic and to determine the sign of the correlation between time and the measure (claim amount, length of stay, etc).
      Two measures of interdistrict migration of RAS beneficiaries were calculated for 2014 and 2018, and the results were plotted on maps. First, for each district, the number of claims for in-district residents for care at in-district facilities was divided by the overall number of claims for in-district residents for care at any facility. Second, for each district, the number of claims for in-district residents for care at in-district facilities was divided by the overall number of claims filed by in-district facilities for care delivered to all RAS beneficiaries.

      Ethics

      As publicly available, deidentified claims were retrospectively reviewed, informed consent from participants or their families was not obtained, and it was not feasible for the authors to obtain. All data used in the study are available on the Dr YSR Aarogyasri Health Care Trust website
      Andhra Pradesh State Government
      Dr. YSR Aarogyasri health care trust.
      and were not collected by the authors.

      Results

      A total of 917 474 unique ration cards were found in the dataset, corresponding to 938 404 uniquely identifiable beneficiaries. During the years for which a complete year of data are available (2015-2017), the number of individuals filing claims with the program steadily increased. Whereas 186 083 individuals filed claims in 2015, 275 772 did so in 2017 (Table 1). It should be noted that many people were insured by the program but did not file claims because the program only covers inpatient care, and people often do not require inpatient care each year. For each of the categorical variables featured in Table 1, the significance of the change in the distribution of the population across categories is presented. The distribution of claims across sex and age categories (women 0-14 and > 15 years and men 0-14 and > 15 years) changed significantly (P<.001) over time, with women accounting for a greater proportion of claims in 2018 than in 2014. The median age of patients seeking care significantly (P<.001) increased over time, rising from 44.0 years in 2014 to 46.0 years in 2018. The caste distribution of the individuals significantly changed over time, with the percentage of individuals from a backward caste steadily rising from 49.7% in 2014 to 52.4% in 2018. Most claims were for care at private facilities, ranging from a low of 71.5% in 2018 to a high of 76.0% in 2016. As actual charges do not always reach the preauthorized maximum, the average amount claimed was INR 26 698.60, whereas the average amount preauthorized was INR 29 545.80.
      Table 1Baseline descriptive statistics for sample.
      2014
      Years 2014 and 2018 both only consist of 2 quarters of data.
      (N = 104 596)
      2015 (N = 186 083)2016 (N = 232 432)2017 (N = 275 772)2018
      Years 2014 and 2018 both only consist of 2 quarters of data.
      (N = 139 521)
      P
      P is from Student’s t test of significance for continuous variables and the Chi-square test of significance for categorical variables.
      Sex and Age
       Female (> 15 y)36 082 (34.5%)61 177 (32.9%)81 430 (35.0%)93 027 (33.7%)48 316 (34.6%)<.001
       Female (0-14 y)4745 (4.5%)9377 (5.0%)8631 (3.7%)12 923 (4.7%)6553 (4.7%)
       Male (> 15 y)55 743 (53.3%)99 593 (53.5%)127 309 (54.8%)148 323 (53.8%)73 903 (53.0%)
       Male (0-14 y)8026 (7.7%)15 936 (8.6%)15 062 (6.5%)21 499 (7.8%)10 749 (7.7%)
       Age, median (IQR)44.0 (28.0, 56.0)44.0 (28.0, 57.0)45.0 (30.0, 58.0)45.0 (29.0, 59.0)46.0 (30.0, 60.0)<.001
      Caste
       BC51,980 (49.7%)94 385 (50.7%)119 129 (51.3%)143 941 (52.2%)73 167 (52.4%)<.001
       Minorities6676 (6.4%)11 824 (6.4%)13 887 (6.0%)16 659 (6.0%)8395 (6.0%)
       OC24 739 (23.7%)42 562 (22.9%)54 188 (23.3%)61 553 (22.3%)30 263 (21.7%)
       Others76 (0.1%)146 (0.1%)222 (0.1%)220 (0.1%)83 (0.1%)
       SC18 023 (17.2%)31 309 (16.8%)38 005 (16.4%)44 971 (16.3%)23 119 (16.6%)
       ST3102 (3.0%)5857 (3.1%)7001 (3.0%)8428 (3.1%)4494 (3.2%)
      Hospital
       Private76 557 (73.2%)138 753 (74.6%)176 725 (76.0%)207 569 (75.3%)99 782 (71.5%)<.001
       Government28 039 (26.8%)47 330 (25.4%)55 707 (24.0%)68 203 (24.7%)39 739 (28.5%)
      Note: N equals the number of beneficiaries, whereas “%” means column percentages.
      BC indicates backward caste; IQR, interquartile range; OC, other caste (which includes Brahmins, Kshatriyas, and other socially privileged groups); SC, scheduled caste; ST, scheduled tribe (geographically isolated tribal communities).
      Years 2014 and 2018 both only consist of 2 quarters of data.
      P is from Student’s t test of significance for continuous variables and the Chi-square test of significance for categorical variables.
      As is shown in Table 2, the nominal mean claim amount significantly (P<.001) increased over time. The mean claims were INR 25 215 in 2014, rose to a peak of INR 27 652 in 2017, and then declined to INR 26 574 in 2018. The most expensive category was the Others category, which included some expensive procedures, such as cochlear implant surgery, and had a mean claim amount of INR 175 385 in 2018. The second most expensive category was cardiac and cardiothoracic surgery, which had a mean claim amount of INR 80 757 in 2018. Overall, 14 categories of claims significantly increased, 9 categories of claims significantly decreased, and 2 categories did not significantly change. After normalizing the mean claim amounts on the basis of their 2014 levels (Table 2), the overall mean claims in 2018 were found to be 105.4% of overall mean claims in 2014. Although the mean claims for most categories increased from 2014 to 2018, this was not universally the case, with nominal declines of 5% or more seen in claims for the following: (1) endocrinology; (2) ears, nose, and throat surgery; (3) general surgery; (4) gynecology and obstetrics surgery; (5) nephrology; (6) pediatric surgeries; and (7) pediatrics.
      Table 2Normalized trend and mean nominal claim amount (in INR) across tertiary care health services under the Rajiv Arogyasri Scheme from 2014 to 2018.
      CategoriesNominal claim amount, Normalized trend (mean ± SD)Taub
      Coefficients are from the Mann-Kendall test of correlation.
      2014
      Years 2014 and 2018 both only consist of 2 quarters of data.
      2015201620172018
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Mean claim amountNormalized claim amountNormalized claim amountNormalized claim amountNormalized claim amount
      ( ± SD)(mean ± SD)(mean ± SD)(mean ± SD)(mean ± SD)
      Cardiac and Cardiothoracic Surgery66 419.8 (43 591.9)1.04 (68 841 ± 37 557.5)1.24 (82 495.6 ± 34 387.6)1.24 (82 487.1 ± 34 334.5)1.22 (80 756.5 ± 31 541.8)0.13
      Cardiology25 989.7 (16 694.2)1.03 (26 836.3 ± 15 744.2)1.05 (27 200.1 ± 12 289.4)1.03 (26 731.9 ± 12 201.4)1.04 (26 953.4 ± 11 581.4)0.03
      P<.001.
      Critical care38 095.4 (19 440.3)0.95 (36 359 ± 17 683.2)1.20 (45 745.7 ± 19 566.5)1.32 (50 185.2 ± 18 981.9)1.19 (45 464.1 ± 16 718.1)0.16
      P<.001.
      Endocrinology21 278.1 (18 191.7)0.85 (18 027.1 ± 5060.5)0.60 (12 828.8 ± 6036.2)0.70 (14 875.4 ± 5725.6)0.71 (15 003.4 ± 5926.8)-0.09
      P<.001.
      ENT surgery15 629.7 (2607.4)0.96 (15 061 ± 2080.6)0.96 (14 984.5 ± 2110.3)0.96 (14 938.1 ± 2189.5)0.93 (14 469.5 ± 2419.0)-0.09
      P<.001.
      Gastroenterology20 586.9 (13 006)1.04 (21 339.3 ± 9670.1)1.09 (22 459.4 ± 1,1376.3)1.16 (23 798.1 ± 11 701.3)1.09 (22 437.8 ± 10 904.3)0.04
      P<.001.
      General medicine21 543.4 (12 511)0.96 (20 638.5 ± 10 577.8)0.89 (19 148.7 ± 8623.8)1.01 (21 841.0 ± 11 251.7)1.02 (21 997.5 ± 9960.8)0.06
      P<.001.
      General surgery30 272.5 (11 875.7)0.95 (28 743.5 ± 10 352)0.94 (28 436.9 ± 10 095.0)0.93 (28 142.8 ± 9794.6)0.92 (27 829.5 ± 9907.9)-0.04
      P<.001.
      Genitourinary surgeries26 928.7 (12 509)0.97 (26 253.5 ± 13 579.7)1.04 (27 953 ± 12 784.4)1.07 (28 709.7 ± 12 937.5)1.05 (28 349.6 ± 12 283.0)0.07
      P<.001.
      Gynecology and Obstetrics surgery24 794.3 (6376.2)0.96 (23 878.7 ± 6403.2)0.94 (23 379.6 ± 5793.3)0.93 (23 180.2 ± 6035.1)0.93 (23 013.4 ± 6096.8)-0.07
      P<.001.
      Medical oncology5394.70 (7647.90)1.06 (5713.3 ± 7815.3)1.20 (6465.4 ± 8359.5)1.26 (6820.0 ± 8887.8)1.33 (7149.7 ± 9280.1)0.07
      P<.001.
      Nephrology13 211.5 (4742.9)1.01 (13 314.6 ± 4827.2)1.02 (13 480.8 ± 5114.0)1.02 (13 525.0 ± 5558.2)0.93 (12 259.9 ± 6058.3)-0.20
      P<.001.
      Neurology17 974.3 (6067.2)1.03 (18 568.3 ± 6755.0)1.07 (19 214.6 ± 7973.0)1.10 (19 810.4 ± 8264.4)1.10 (19 811.3 ± 7776.8)0.03
      P<.001.
      Neurosurgery49 722.8 (19 210.3)1.00 (49 695.8 ± 19 644.9)1.01 (50 256.3 ± 18 522.2)1.01 (49 993.4 ± 18 902.7)0.98 (48 543.4 ± 17 646.5)-0.01
      P<.050.
      Ophthalmology surgery16 646 (7619.90)0.97 (15 917.3 ± 7926.9)1.02 (16 653.6 ± 10 490.6)0.97 (15 848.6 ± 10 112.8)0.97 (15 884.5 ± 10 771.5)-0.08
      P<.001.
      Orthopedic surgery18 699.6 (9560.5)1.01 (18 891 ± 9576.8)1.23 (23 046.9 ± 10 922.4)1.25 (23 292 ± 10 934.1)1.24 (23 226.5 ± 10 397.5)0.14
      P<.001.
      Others
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      147 834.9 (218 711.8)1.23 (181 503.7 ± 234 549.7)0.97 (143 663.2 ± 216 083.8)0.83 (123 155.6 ± 200 949.9)1.19 (175 385.3 ± 231 008.8)0.00
      Pediatric surgeries41 545.4 (14 694.3)0.82 (33 935.3 ± 16 640)0.80 (33 284.4 ± 16 866.3)0.77 (32 061.5 ± 16 926.5)0.79 (32 622.8 ± 16 892.7)-0.10
      P<.001.
      Pediatrics24 270.5 (11 116.8)0.99 (24 041.3 ± 12 097.7)1.01 (24 412.4 ± 12 354.8)1.02 (24 837.6 ± 13 605.7)0.89 (21 559.4 ± 13 779.6)-0.05
      P<.001.
      Plastic surgery36 250 (18 214.4)1.00 (36 268.1 ± 19 182.9)0.93 (33 887.4 ± 17 092.6)1.03 (37 437.2 ± 17 055.1)0.99 (36 059.1 ± 15 723.5)0.01
      Poly trauma27 054.7 (9417.7)0.99 (26 865.7 ± 9243)1.04 (28 074.6 ± 8820.6)1.04 (28 247.8 ± 8781.0)1.04 (28 048.9 ± 8623.1)0.03
      P<.001.
      Pulmonology23 138.9 (9948.1)1.02 (23 646.9 ± 8329.7)1.08 (24 971.8 ± 8596)1.11 (25 631.1 ± 9237.8)1.07 (24 670 ± 8011.5)0.07
      P<.001.
      Radiation oncology29 646.4 (26 286.3)1.06 (31 317.2 ± 27 429.8)1.07 (31 587.6 ± 27 429.6)1.07 (31 808.7 ± 26 506.0)1.21 (35 835 ± 24 375.7)0.09
      P<.001.
      Surgical gastroenterology50 959.6 (27 402.9)1.01 (51 613.2 ± 27 173.1)1.09 (55 361.0 ± 28 116.4)1.16 (59 249.6 ± 29 035.0)1.06 (53 871.8 ± 26 988.6)0.07
      P<.001.
      Surgical oncology33 998.5 (17 775)1.02 (34 836.6 ± 17 289.3)1.21 (41 287.7 ± 17 511.3)1.25 (42 502.5 ± 17 985.2)1.26 (42 852.5 ± 18 565.7)0.16
      P<.001.
      Overall25 205.02 (25 215.1)1.00 (25 320.7 ± 24 825.6)1.09 (27 393.1 ± 25 487.2)1.10 (27 652.2 ± 25 951.7)1.05 (26 574.2 ± 25 031.3)0.02
      P<.001.
      ENT indicates ears, nose, and throat; INR, Indian Rupee.
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Coefficients are from the Mann-Kendall test of correlation.
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      § P<.001.
      P<.050.
      To assess whether price trends experienced by categories were driven by changes in the distribution of use across procedures within the categories, the price trends for the most frequent procedure in each category in the third quarter of 2014 were determined. As presented in Table 3, the price trends for the most frequent procedure in each category, except endocrinology, others, and surgical oncology, were significant (P<.050) and monotonic. The directions of the price trends were not consistent across categories—15 had a positive, significant trend, and 8 had a negative, significant trend.
      Table 3Monotonicity of trends in the claim amounts of the most common types of surgeries in the third quarter of 2014.
      CategoryMost frequent surgery in the third quarter of 2014Taub
      Coefficients are from the Mann-Kendall test of correlation.
      Cardiac and Cardiothoracic SurgeryCoronary balloon angioplasty with stent (00.45)0.30
      P<.001.
      CardiologyManagement of acute myocardial infarction with angiogram0.01
      P<.050.
      Critical careMedical management of poisoning requiring ventilatory assistance0.12
      P<.001.
      EndocrinologyMedical management of pyelonephritis in uncontrolled diabetes mellitus-0.04
      ENT surgeryTympanoplasty-0.33
      P<.001.
      GastroenterologyMedical management of acute pancreatitis (mild)0.05
      P<.001.
      General medicineMedical management of tuberculous meningitis-0.02
      P<.050.
      General surgeryLaparoscopic appendicectomy0.15
      P<.001.
      Genitourinary surgeriesUreteroscopic lithotripsy-0.37
      P<.001.
      Gynecology and obstetrics surgeryVaginal hysterectomy with pelvic floor repair (70.79)0.13
      P<.001.
      Medical oncologyChemotherapy for cervical cancer with weekly cisplatin0.18
      P<.001.
      NephrologyMaintenance hemodialysis for chronic renal failure-0.30
      P<.001.
      NeurologyMedical management of ischemic strokes-0.02
      P<.001.
      NeurosurgerySpinal fusion procedure0.06
      P<.001.
      Ophthalmology surgeryVitrectomy, membrane peeling, endolaser, silicon oil or gas, with or without belt buckling0.29
      P<.001.
      Orthopedic surgerySoft tissue reconstruction procedures for joints/osteotomy0.05
      P<.001.
      Others
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      Medical management of systemic lupus erythematosus0.03
      Pediatric surgeriesStage 1 procedure for hypospadias-0.15
      P<.001.
      PediatricsMedical management of term baby severe hyperbilirubinemia clinically evident septicemia0.14
      P<.001.
      Plastic surgeryPostburn contracture surgeries, severe-0.20
      P<.001.
      PolytraumaSurgical correction of long bone fracture-0.01
      P<.001.
      PulmonologyMedical management of acute respiratory failure (without ventilator)0.16
      P<.001.
      Radiation oncologyBrachytherapy intracavitary ii. HDR per application0.16
      P<.001.
      Surgical gastroenterologyLaparoscopic cholecystectomy0.09
      P<.001.
      Surgical oncologyMastectomy any type in malignant conditions0.03
      ENT indicates ears, nose, and throat; HDR, high dose rate.
      Coefficients are from the Mann-Kendall test of correlation.
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      P<.001.
      § P<.050.
      Overall, the median length of stay shortened from 5 to 4 days between 2014 and 2018 (Table 4). There were significant, negative trends in the length of stay for 15 categories of procedures, and significant, positive trends in the length of stay for 8 categories of procedures. Nephrology procedures had, by far, the longest lengths of stay, with a median of 30 days in 2018. The median medical oncology procedure had a length of stay of 0 days.
      Table 4The median length of hospital stay (in days) across tertiary care health services under Rajiv Arogyasri Scheme from 2014 to 2018.
      CategoriesLength of hospital stay (in days), median (IQR)Taub
      Coefficients are from the Mann-Kendall test of correlation.
      2014
      Years 2014 and 2018 both only consist of 2 quarters of data.
      2015201620172018
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Cardiac and cardiothoracic surgery4 (2,7)4 (2,7)3 (2,6)3 (2,6)3 (2,5)-0.06
      P<.001.
      Cardiology5 (3,7)5 (4,7)5 (4,7)5 (3,7)5 (3,7)-0.01
      P<.001.
      Critical care6 (1,10)7 (1,11)9 (5,12)10 (7,12)10 (7,12)0.17
      P<.001.
      Endocrinology7 (4,11)8 (6,12)2 (0,7)3 (0,7)3 (0,7)-0.15
      P<.001.
      ENT surgery6 (3,7)3 (2,6)2 (1,4)2 (1,3)2 (1,4)-0.18
      P<.001.
      Gastroenterology1 (0,6)5 (1,8)5 (2,8)5 (2,8)6 (3,8)0.10
      P<.001.
      General medicine3 (0,8)4 (0,9)3 (0,9)5 (1,10)5 (3,9)0.10
      P<.001.
      General surgery8 (6,11)7 (4,10)6 (4,9)6 (3,9)6 (4,9)-0.05
      P<.001.
      Genitourinary surgeries3 (2,5)3 (2,5)3 (2,4)3 (2,4)3 (2,4)-0.08
      P<.001.
      Gynecology and obstetrics surgery8 (6,10)8 (4,10)7 (4,10)7 (4,10)7 (4,10)-0.06
      P<.001.
      Medical oncology0 (0,2)0 (0,2)0 (0,2)0 (0,2)0 (0,1)-0.03
      P<.001.
      Nephrology31 (13,35)31 (21,35)31 (21,35)31 (21,35)30 (21,35)-0.001
      Neurology4 (0,8)5 (0,9)5 (1,8)6 (2,9)5 (2,8)0.04
      P<.001.
      Neurosurgery8 (6,10)8 (5,10)7 (5,10)8 (5,11)7 (5,10)-0.01
      P<.050.
      Ophthalmology surgery1 (1,2)1 (1,2)1 (1,2)1 (1,2)1 (1,1)-0.07
      P<.001.
      Orthopedic surgery5 (4,7)5 (4,7)5 (3,7)4 (3,6)4 (3,7)-0.07
      P<.001.
      Others
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      1 (0,3)5 (1,9)5 (1,10)5 (1,11)2 (1,9)0.07
      P<.050.
      Pediatric surgeries8 (6,10)7 (4,9)7 (4,9)6 (4,9)6 (4,9)-0.05
      P<.001.
      Pediatrics6 (2,10)6 (3,10)6 (2,9)6 (3,11)6 (4,11)0.02
      P<.001.
      Plastic surgery7 (4,13)7 (5,13)7 (4,13)7 (5,13)6 (4,13)-0.01
      Polytrauma5 (4,8)5 (4,7)5 (4,7)5 (3,7)4 (3,7)-0.08
      P<.001.
      Pulmonology4 (0,9)7 (3,9)7 (4,9)7 (3,10)7 (4,10)0.05
      P<.001.
      Radiation oncology2 (0,39)11 (0,40)12 (0,40)12 (0,40)7 (0,39)-0.001
      Surgical gastroenterology7 (3,11)7 (4,11)7 (4,11)7 (5,11)7 (4,10)0.02
      P<.050.
      Surgical oncology7 (5,9)7 (5,9)7 (5,9)7 (4,9)6 (4,9)-0.03
      P<.001.
      Overall5 (1,9)5 (2,10)4 (2,8)4 (2,10)4 (1,9)-0.01
      P<.001.
      ENT indicates ears, nose, and throat.
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Coefficients are from the Mann-Kendall test of correlation.
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      § P<.001.
      P<.050.
      Mortality rates following procedures generally improved over time (Table 5). While 2.4% of procedures were followed by death in 2014, only 2.1% were followed by death in 2018—a significant (P<.001) 12.5% decline in the mortality rate. The biggest decline occurred in critical care, the category with the highest mortality rate. The mortality rate for critical care decreased from 29.4% to 21.5%, a significant decrease (P<.001). The only category with a significant increase in its mortality rate was neurology, which increased from 5.4% to 7.7% (P<.001).
      Table 5Mortality across tertiary care health services under Rajiv Arogyasri Scheme from 2014 to 2018.
      CategoryMortality rates, % (N)Taub
      Coefficients are from the Mann-Kendall test of correlation.
      2014
      Years 2014 and 2018 both only consist of 2 quarters of data.
      2015201620172018
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Cardiac and cardiothoracic surgery2.4 (264)2.0 (424)2.2 (457)2.3 (588)1.5 (183)-0.00
      P<.001.
      Cardiology6.1 (212)7.1 (579)4.4 (346)4.6 (760)4.1 (271)-0.02
      P<.001.
      Critical care29.4 (160)32.6 (295)26.7 (198)22.1 (422)21.5 (208)-0.09
      P<.001.
      Endocrinology1.4 (1)0.5 (1)0.5 (1)1.5 (7)1.6 (6)0.02
      ENT surgery0.0 (0)0.0 (0)0.0 (0)0.0 (0)0.0 (0)0.00
      Gastroenterology3.0 (19)4.2 (64)3.9 (53)3.4 (109)5.5 (98)0.00
      General medicine11.0 (24)7.1 (46)4.8 (30)6.5 (81)6.5 (53)-0.01
      General surgery2.3 (103)1.6 (199)1.0 (214)0.8 (208)0.9 (132)-0.02
      P<.001.
      Genitourinary surgeries0.1 (17)0.0 (16)0.0 (17)0.0 (24)0.0 (10)-0.00
      Gynecology and obstetrics surgery0.1 (2)0.0 (0)0.0 (0)0.0 (3)0.0 (1)-0.00
      Medical oncology0.3 (30)0.8 (69)0.7 (102)0.7 (81)0.8 (54)-0.00
      Nephrology6.2 (462)6.8 (446)6.7 (686)6.7 (553)6.1 (240)-0.00
      Neurology5.4 (173)7.0 (596)6.2 (394)7.4 (956)7.7 (474)0.01
      P<.001.
      Neurosurgery5.9 (254)5.7 (453)5.4 (489)5.9 (566)6.2 (334)0.00
      Ophthalmology surgery0.0 (0)0.0 (0)0.0 (0)0.0 (0)0.0 (1)0.00
      Orthopedic surgery0.0 (1)0.0 (5)0.1 (8)0.0 (4)0.0 (1)-0.00
      Others
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      0.6 (1)2.4 (7)2.2 (3)1.4 (5)1.6 (2)-0.01
      Pediatric surgeries4.7 (66)2.8 (82)2.6 (90)2.2 (84)2.4 (43)-0.02
      P<.001.
      Pediatrics4.9 (302)4.4 (627)3.7 (397)3.5 (719)2.8 (299)-0.03
      P<.001.
      Plastic surgery14.5 (119)15.4 (171)12.8 (219)12.8 (185)5.7 (52)-0.07
      P<.001.
      Polytrauma0.7 (169)0.8 (354)0.7 (390)0.7 (410)0.8 (246)-0.00
      Pulmonology13.6 (128)8.5 (178)10.2 (166)10.9 (400)11.6 (251)-0.00
      Radiation oncology1.5 (52)2.4 (79)1.8 (98)1.8 (58)1.8 (36)-0.00
      Surgical gastroenterology3.0 (16)2.8 (21)2.5 (25)2.1 (21)2.2 (12)-0.01
      Surgical oncology0.9 (32)0.8 (44)1.0 (70)1.1 (75)1 (41)0.00
      Overall2.4 (2607)2.5 (4756)1.9 (4453)2.2 (6319)2.1 (3048)-0.001
      P<.001.
      Note: Percentages are presented as percent of the total number of surgeries for each category per year.
      ENT indicates ears, nose, and throat.
      Years 2014 and 2018 both only consist of 2 quarters of data.
      Coefficients are from the Mann-Kendall test of correlation.
      Others include cochlear implant surgery, dermatology, infectious diseases, prostheses, and rheumatology.
      § P<.001.
      Aarogyasri claims are concentrated in urban areas and the densely populated coast, with most hospitals located in district headquarters and disproportionately in the largest districts of Visakhapatnam and Krishna. This mismatch between where people live and where hospitals are located causes patients to travel significant distances to seek care. The percentage of the total demand served locally was found to be the lowest in YSR Kadapa (30.9% in 2014 and 45.2% in 2018), followed by West Godaveri (33.8% in 2014 and 49.3% in 2018), implying that most residents of these districts had to migrate to access care. In contrast, Nellore, Chittoor, and Visakhapatnam had the highest percentage of total demand served locally (100.0%, 90.1%, and 91.1% in 2014 and 100.0%, 94.7%, and 92.7% in 2018, respectively), implying that the lowest percentage of their residents migrated to other districts to seek care. Over the 5 years, Kurnool saw the maximum decrease in its residents migrating to other districts to seek care, with its percentage of demand served locally doubling from 48.3% in 2014 and 89.8% in 2018. In contrast, Vizianagram saw a decrease in the percentage of total demand served locally (41.8% in 2014 and 38.1% in 2018) (Appendix Figure 1A, B in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2021.02.007).
      Similarly, the percentage of care provided by district facilities to RAS beneficiaries residing in the facility’s district was the highest in YSR Kadapa (94.9% in 2014 and 100.0% in 2018) and West Godaveri (83.9% in 2014 and 100.0% in 2018). This implies that the number of migrants accessing care from these districts decreased to 0 in 2018. The percentage of care provided by district facilities to RAS beneficiaries residing in the facility’s district was the lowest in Visakhapatnam, Krishna, and Chittoor (54.7%, 71.7%, and 67.8% in 2014 and 54.1%, 64.0%, and 67.8% in 2018, respectively) as the facilities of these districts addressed the needs of most migrants. West Godaveri showed an increase in the percentage of care provided to home district residents from 2014 to 2018 (Appendix Figure 1C, D in Supplemental Materials found at https://doi.org/10.1016/j.vhri.2021.02.007).

      Discussion

      Between 2014 and 2018, there was a modest but significant nominal increase in the cost per RAS claim, accompanied by a significant decrease in mortality. During the years for which complete data are available, there was steady growth in the number of individuals availing services under the scheme; increasing from 186 083 in 2015 to 275 772 in 2017. The length of stay overall declined, potentially increasing the capacity of hospitals to treat patients.
      Throughout the study period, most beneficiaries availing care were men. This is consistent with a previous study of sex disparities on the use of care funded by the RAS. A previous study of RAS claims from 2008 to 2012 in Andhra Pradesh found that men accounted for most claimants, bed-days, and care expenditures.
      • Shaikh M.
      • Peters S.A.E.
      • Woodward M.
      • Norton R.
      • Jha V.
      Sex differences in utilisation of hospital care in a state-sponsored health insurance programme providing access to free services in South India.
      In addition, the study found that men accessed the majority of care for sex-neutral conditions, suggesting that the disparity observed is being driven by factors other than a differing, underlying need for medical care.
      Although there were nominal increases in costs, they occurred during a period of inflation. According to the Indian government, the cost inflation index was 240 in the fiscal year of 2014 to 2015 and 280 in the fiscal year of 2018 to 2019.
      Income Tax Department – Government of India Notified cost inflation index under Section 48. Explanation (V).
      While the costs increased by 16.7% in the broader Indian economy, the 5.4% increase in claims costs observed represents a real decrease in the cost of care borne by the government after inflation is taken into consideration. Thus, in 2018, claims were 90.3% of what they had been in 2014 when all claims were expressed in real rather than nominal amounts. Only categories of service with normalized nominal mean claim amounts in 2018 exceeding 1.17 experienced a real increase in costs; there were 7 such categories, as is shown in Table 2.
      The finding on the preference for private hospitals is consistent with the findings of previous researchers. A study of people with RAS insurance within the Vizianagaram District of Andhra Pradesh found most patients surveyed preferred to receive care at private hospitals.
      • Madaboyina G.
      An evaluative study of Rajiv Aarogyasri health insurance scheme in Vizianagaram District of Andhra Pradesh.
      Another pan-Andhra Pradesh study of claims data likewise concluded that between 2008 and 2012, most RAS claims were for care delivered at private hospitals.
      • Shaikh M.
      • Woodward M.
      • Rahimi K.
      • et al.
      Major surgery in south India: a retrospective audit of hospital claim data from a large community health insurance programme.
      Furthermore, a study of publicly funded health insurance schemes in Andhra Pradesh, Karnataka, and Tamil Nadu in 2014 concluded that most use occurred at private hospitals in all 3 states.
      • Garg S.
      • Chowdhury S.
      • Sundararaman T.
      Utilisation and financial protection for hospital care under publicly funded health insurance in three states in Southern India.
      Collectively, the findings of this study suggest that RAS delivered increasing value over the period observed. The number of patients that filed claims grew and the mortality rates experienced by patients declined. While costs increased in nominal terms, after accounting for inflation, they declined on a per claim basis in real terms as well. Since value is defined as outcomes relative to costs, improving outcomes and declining costs suggest improvements in value.
      • Porter M.E.
      What is value in health care.
      The interdistrict migration among claimants could be because of various reasons. First, it is possible that some of the local facilities were empaneled only for a set of services under the RAS. Second, the perceived quality of care among claimants in local facilities might be poor. Finally, it is possible that the district facilities were not adequately staffed with specialists owing to the unavailability of human resources.
      Although the findings of this study are fully representative of the population studied, they may not be reflective of trends experienced in other parts of India or for people with higher incomes that are ineligible for RAS. In India, state governments play a substantial role in shaping their local health insurance policy and in determining how insurance schemes are implemented. Furthermore, there are vast differences in the income per capita, degree of transportation connectivity, and even life expectancies in different states. Andhra Pradesh falls in the middle of the pack, with a life expectancy at birth of 68.5 years from 2010 to 2014 compared with an all-India average of 67.9 years. Life expectancy ranges from a low of 63.9 years in Assam to a high of 74.9 years in Kerala.
      Office of the Registrar General & Census Commissioner, India
      Abridged life tables – 2010-2014.
      Although not measured in this analysis, outpatient department care plays a substantial role in the healthcare system. Technology increasingly makes it possible for care that had formerly been delivered by inpatient departments to be delivered by outpatient departments. As outpatient claims were not in the database provided by the government, it is not possible to observe how outpatient use shifted over time.

      Conclusions

      Inpatient claims suggest that there have been improvements in the value of care delivered to patients with RAS insurance coverage between 2014 and 2018. The length of stay and mortality have declined, the number of people treated has expanded, and although the cost of claims has increased in nominal terms, after accounting for inflation, there was a real decline overall in the cost of care on a per claim basis. The growing availability of data on health insurance schemes and other government programs, fostered by the Digital India initiative, is increasing the transparency with which the public may observe program performance. As more cashless transactions occur, the quantity and quality of information available about Indian society will only increase.
      • Athique A.
      A great leap of faith: the cashless agenda in Digital India.
      This analysis demonstrates the granularity with which the public can observe program performance and suggests that RAS has improved the value that it delivers over time.

      Article and Author Information

      Author Contributions: Concept and design: Singh, Powell
      Acquisition of data: Singh
      Analysis and interpretation of data: Singh, Powell
      Drafting of the manuscript: Singh, Powell
      Critical revision of the paper for important intellectual content: Singh, Powell
      Statistical analysis: Singh
      Supervision: Powell
      Conflict of Interest Disclosures: Dr Powell is employed by Payer+Provider Syndicate and owns stock in Amazon, Berkshire Hathaway, HCA Healthcare, Payer+Provider Syndicate, and Tenet Healthcare Corp. outside of the submitted work. Although Dr Powell’s conflicts relate to the health insurance and health services industries, none relate to India or the program analyzed. No other disclosures were reported.
      Funding/Support: Support was provided by the Max Institute of Healthcare Management at the Indian School of Business. No direct financial support was provided.
      Role of the Funder/Sponsor: No organization had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

      Acknowledgment

      The authors thank Jagmanjot Singh, Bachelor of Computer Science and Engineering, Punjab Engineering College, Chandigarh, India for his assistance in creating maps for this article.

      Supplemental Materials

      References

        • van Doorslaer E.
        • O’Donnell O.
        • Rannan-Eliya R.P.
        • et al.
        Catastrophic payments for health care in Asia.
        Health Econ. 2007; 16: 1159-1184
        • van Doorslaer E.
        • O’Donnell O.
        • Rannan-Eliya R.P.
        • et al.
        Effect of payments for health care on poverty estimates in 11 countries in Asia: an analysis of household survey data.
        Lancet. 2006; 368: 1357-1364
        • Devadasan N.
        • Ranson K.
        • Van Damme W.
        • Criel B.
        Community health insurance in India: an overview.
        Econ Polit Wkly. 2004; 39: 3179-3183
        • Prinja S.
        • Chauhan A.S.
        • Karan A.
        • Kaur G.
        • Kumar R.
        Impact of publicly financed health insurance schemes on healthcare utilization and financial risk protection in India: a systematic review.
        PLoS One. 2017; 12e0170996
        • Garg S.
        • Chowdhury S.
        • Sundararaman T.
        Utilisation and financial protection for hospital care under publicly funded health insurance in three states in Southern India.
        BMC Health Serv Res. 2019; 19: 1004
        • Andhra Pradesh State Government
        Frequently asked questions.
        • Ministry for Consumer Affairs, Food and Public Distribution
        Introduction.
        https://dfpd.gov.in/index.htm
        Date accessed: November 15, 2019
        • Shaikh M.
        • Woodward M.
        • Rahimi K.
        • et al.
        Major surgery in south India: a retrospective audit of hospital claim data from a large community health insurance programme.
        Lancet. 2015; 385: S23
        • Shaikh M.
        • Peters S.A.E.
        • Woodward M.
        • Norton R.
        • Jha V.
        Sex differences in utilisation of hospital care in a state-sponsored health insurance programme providing access to free services in South India.
        BMJ Glob Health. 2018; 3e000859
        • Rao M.
        • Kadam S.
        • Sathyanarayana T.N.
        • et al.
        A rapid evaluation of the Rajiv Aarogyasri community health insurance scheme in Andhra Pradesh, India.
        BMC Proc. 2012; 6: O4
        • Nagulapalli S.
        • Rokkam S.R.
        Should Governments engage health insurance intermediaries? A comparison of benefits with and without insurance intermediary in a large tax funded community health insurance scheme in the Indian state of Andhra Pradesh.
        BMC Health Serv Res. 2015; 15: 370
        • Madaboyina G.
        An evaluative study of Rajiv Aarogyasri health insurance scheme in Vizianagaram District of Andhra Pradesh.
        Arts Soc Sci J. 2016; 7: 231
        • Andhra Pradesh State Government
        Dr. YSR Aarogyasri health care trust.
      1. Income Tax Department – Government of India Notified cost inflation index under Section 48. Explanation (V).
        (Published September 2019. Accessed December 23, 2019)
        • Porter M.E.
        What is value in health care.
        N Engl J Med. 2010; 363: 2477-2481
        • Office of the Registrar General & Census Commissioner, India
        Abridged life tables – 2010-2014.
        • Athique A.
        A great leap of faith: the cashless agenda in Digital India.
        New Media Soc. 2019; 21: 1697-1713