Table 2. Methodological characteristics and findings of the studies with a comparison group.

Study Author &Year	Study design	Source of data and Methodology	Time period after implementation of the scheme	Quality of the study	Impact on Utilization	Impact on Financial risk protection*	Health impact
Rao et al., 2014 [24]	Quasi experimental design (Pre and post design)	Primary survey in the states Andhra Pradesh and Maharashtra and comparison with the findings of NSS@ 2004–05 round.	3 years	Strong	Utilization increased in both states; more increase in Andhra Pradesh than Maharashtra	Inpatient expenditure, large expenditure (proxy for catastrophic health expenditure) increased over the time period with more increase in Maharashtra than Andhra Pradesh.
Selvaraj et al., 2012 [25]	Quasi experimental design (Pre and post design)	National Sample Survey rounds for the year 2004–05 and 2009–10 were compared	< 3 years	Moderate		OOP$ inpatient spending, catastrophic headcount ratio and OOP spending as a proportion of overall spending increased over the time.
Amicus Advisory Pvt. Ltd. [29]	Cross sectional	Primary survey in 10 villages of Jaunpur district, Uttar Pradesh	< 3 years	Weak		Eligible and users of the scheme incurred less expenditure than non-users.
Aiyar et al., 2013 [26]	Cross sectional	Two rounds of data collection from the 2 districts of Karnataka, 2 years apart.	< 3 years	Weak	Incidence of hospitalization increased among insured than non-insured.	OOP expenditure and catastrophic health expenditure increased in both insured and non-insured households.
Sunny et al. [27]	Cross sectional	Primary data collected from the insured and non-insured hospitalised cases in the state of Kerala.	< 3 years	Moderate		There was similar amount of expenditure incurred by both insured and non-insured cases.
GIZ, 2012 [28]	Cross sectional	Primary survey conducted across three states of Bihar, Uttrakhand and Karnataka.	>3 years	Moderate		90% of the insured households did not spend any money on hospitalization.
Fan et al., 2012 [33]	Quasi experimental design (Pre and post design with a DID* based analysis)	National Sample Survey rounds for the year 1999–2000, 2004–05 and 2007–08 were compared.	< 3 years	Strong		Initial reduction in OOP expenditure and catastrophic health expenditure, followed by an increase in inpatient expenditure.
Bergkvist et al., 2014 [30]	Quasi experimental design (Pre and post design with a DID based analysis)	Primary survey in the states of Andhra Pradesh and Maharashtra; results compared with NSS 2004–05 round.	3 years	Strong	Increased rate of utilization, with faster increase among both the poor and the better off in Andhra Pradesh than Maharashtra.	Smaller growth in OOP expenditure in Andhra Pradesh compared to Maharashtra and mainly concentrated among the richest 60%.
Dhanaraj et al., 2014 [31]	Quasi experimental design (Pre and post with analysis based on panel logit model)	Panel longitudinal dataset of Young Lives project of rounds 2002, 2006, and 2009 was compared for the state of Andhra Pradesh.	< 3 years	Strong		No significant effect in reduction of OOP expenditure over the time period.
Katyal et al., 2015 [32]	Quasi experimental design (Pre and post design with a DID based analysis)	A primary survey undertaken in the states Andhra Pradesh and Maharashtra and the results was compared with the findings of NSS 2004–05 round.	3 years	Strong	Utilization of private hospitals increased in Andhra Pradesh and decreased in Maharashtra. Utilization of public facilities declined in both the states with more decline in Andhra Pradesh.	OOP increased both in public and private facilities, with greater increase in Maharashtra than Andhra Pradesh.
Mitchell et al., 2011 [34]	Cross sectional	Primary household survey conducted in two districts of Andhra Pradesh	< 3 years	Weak		Households with insurance reported higher OOP expenses than those without insurance.
Philip et al., 2012 [35]	Cross sectional	Primary survey conducted in the state of Tamil Nadu.	< 3 years	Strong	Utilization was significantly high among insured as compared to non-insured.	Mean OOP expenses among insured was significantly higher than uninsured households.
Sood et al., 2014 [36]	Quasi experimental design (Geographic discontinuity design with analysis based on logit model)	Primary surveys conducted between the communities where scheme has and has not been implemented in the state of Karnataka	< 3 years	Strong	Insured households were more likely to use the facilities as compared to non-insured.	There was reduction in OOP expenditures among insured as compared to non-insured families.	Enrolled households had relatively lower mortality rate from conditions covered by the scheme.
Ghosh et al., 2014 [37]	Cross sectional	Primary survey conducted in the state of Maharashtra.	5 years	Moderate	Utilization was higher among the insured than non-insured families.

* DID: difference in difference

^@ NSS: national sample survey

^$ OOP: out-of-pocket.