TABLE 2—

Reviews That Considered the Health Impacts of Health Care Policy

Author	Critical Appraisala	Review Qualityb	Included Study Quality	Context	Summary of Insights
Kesselheim et al.65	2–4, 6,c 7–10, 11c	High	Low	United States	There was an absence of evidence on the impact of changes to insurance coverage for prescription drugs on health in the United States.
Gopalan et al.66	1,c 2, 3,c 4, 5, 7,c 8–10	Moderate	High	India	The Janani Surakha Yojana conditional cash transfer for skilled birth attendance in India was associated with a change of −14.2 (95% CI = −2.7, −31) perinatal deaths per 1000 pregnancies and a reduction of 6.2 (95% CI = −8.1, 20.4) neonatal deaths per 1000 live births.
Sumah et al.67	2–4, 6,c 7, 8,c 9–11	Moderate	High	Spain and Canada	There was evidence that inequalities in self-rated health were smaller in Spain and Canada in association with decentralization of health care governance.
Liang et al.68	2–6, 7,c 8–11	Moderate	Mixed	Rural China	The health impact of the Chinese NCMS health care insurance scheme varied widely across the available studies and so the overall impacts were unclear.
Yuan et al.69	2–4, 5,c 6,c 7,c 8,c 9–11	Moderate	Not reported	India and Philippines	There was an absence of evidence in relation to the impact of conditional cash transfers and health insurance policies on maternal mortality.
Acharya et al.70	2, 4, 5, 9, 10, 11c	Low	Not reported	Wide global coverage of low- and middle-income countries	Health care insurance schemes in the informal sector had uncertain impacts on health.
Hadley71	2,c 4, 5, 7,c 8,c 9	Low	Unclear	Not clearly reported, but many from the United States	A change to becoming health care insured was associated with a reduction in mortality ranging between 4% and 25% across studies and improved SRH.
Murray et al.72	1–5, 7,c 9, 10, 11c	Low	High	India, Nepal, Mexico, and Tanzania	Different studies of conditional cash transfers have been associated with an 11% reduction in maternal mortality, no change in neonatal mortality, and 17% and 2% reductions in 2 studies of infant mortality. Some studies have associated payments to offset the costs of health service access with declines in neonatal mortality, while others have shown no change. Maternity service vouchers were associated with a 1 percentage point decline in stillbirths but no effect on neonatal deaths compared with comparison areas.
Ciccone et al.73	2, 3,c 9, 10, 11c	Low	Not reported	Mostly Africa, with some from Asia and 2 in South America	Different aspects of governance in low- and middle-income countries were associated with health outcomes, but the exposure and outcome measures and contexts were highly variable as was the degree of association and extent to which the effect was mediated through other factors.
Kruk et al.74	2–5, 9, 10, 11 c	Low	Not reported	Wide range—Caribbean, Latin America, Central America, sub-Saharan Africa, and Asia	There was some evidence that primary care programs in middle- and low-income countries have reduced child mortality and in some cases wealth-based mortality inequalities.
Singh75	2,c 3,c 4, 5, 8,c 9, 10, 11c	Low	Not reported	United States	There was consistent evidence that public health spending in the United States was associated with better population health outcomes.

Note. CI = confidence interval; NCMS = New Co-operative Medical Scheme; SRH = self-rated health.

^aCritical appraisal criteria: 1 = an a priori design for the review was provided; 2 = a comprehensive search was undertaken (including relevant search terms and at least 2 databases); 3 = studies were selected for inclusion by at least 2 independent researchers; 4 = there were clear inclusion and exclusion criteria; 5 = the status of publication (e.g., gray literature) was ignored in the inclusion and exclusion criteria; 6 = the data were extracted independently by at least 2 researchers; 7 = the scientific quality of the included studies was assessed and documented; 8 = the scientific quality of the included studies was used appropriately in formulating conclusions; 9 = the methods used to combine the findings of studies was appropriate; 10 = the likelihood of publication bias was assessed (if possible); 11 = there were no important conflicts of interest that may have had an impact on the conclusions.

^bQuality assessment: all (very high); at least 2, 4, 7, 8, 9 (high); at least partially 2, 4, 7, 8, 9 (moderate); all others (low).

^cDenotes a partially fulfilled criterion.