Health Spending For Low-, Middle-, And High-Income Americans, 1963–2012

For decades, US medical spending growth outpaced gross domestic product growth. Medical costs started slowing in 2004, ¹ and slower growth continued for a decade, ending in 2014. ² Previous studies have attributed a substantial portion of the slowdown (37–77 percent) to the lingering effects of the Great Recession (2007–09), with the remainder presumably accounted for by factors that predated the recession, such as rising copayments and deductibles, improved delivery system efficiency, the expiration of patents for several expensive drugs, and a decrease in costly medical innovations. ^3–5 These studies have not explored how Americans at different income levels experienced the slowdown and whether health expenditure trends differed according to need or income.

Need plays a greater role in determining the consumption of medical care than for most goods and services. Moreover, health insurance has insulated most patients from the cost of their care, and Medicaid and Medicare have heavily subsidized care for the poor and elderly, who tend to have the greatest health needs. ⁶ However, recent increases in copayments and deductibles for the privately insured, ⁷ along with flat income growth, ⁸ may have constrained health spending overall and skewed it toward wealthier Americans.

We used national survey data to analyze trends in health spending for different income groups over the past half-century, with particular attention to the recent spending slowdown.

Study Data And Methods

Data Sources

We analyzed individual-level data from twenty-two nationally representative surveys of health expenditures by and on behalf of the civilian noninstitutionalized US population conducted over the past fifty years: the Survey of Health Services Utilization and Expenditures (SHSUE), conducted in 1963 and 1970 ( $N=7,759$ and 11,619, respectively); the 1977 and 1980 National Medical Care Utilization and Expenditure Surveys ( $N=38,815$ and 17,123, respectively); the 1987 National Medical Expenditure Survey (NMES) ( $N=23,652$ ); and the 1996–2012 Medical Expenditure Panel Surveys (MEPS) ( $N$ = 21,571–37,418 per year), carried out annually by the Agency for Healthcare Research and Quality. All of these surveys tabulated health expenditures (including all out-of-pocket and third-party payments) based on respondents’ reports, which surveyors then verified with providers. The surveys also collected demographic information, including income and family size.

Analysis

Our main analysis examined trends in mean per capita health spending for each income quintile. We also explored income-related trends according to payer, type of service, and self-reported health status. Expenditure figures for all years were adjusted to 2012 dollars using the Consumer Price Index. ⁹

We divided the population in each survey year into income quintiles (fifths) based on family income as a percentage of poverty. We used the Census Bureau’s poverty measure (federal poverty level), which standardizes for age and family size (and, in 1963 and 1970 only, for farm versus nonfarm status). Because of limitations in the 1977 data, for that year we calculated poverty levels standardized for family size but not age. In 2012, income quintile cut points were at 125 percent, 230 percent, 361 percent, and 558 percent of the federal poverty level, corresponding to annual incomes for a family of three (two adults and one child) of $22,689, $41,820, $65,462, and $101,094.

In addition to total health expenditures, we calculated estimates for five subcategories of health services: inpatient care; outpatient care (including emergency care and outpatient lab tests and imaging); dental care; prescription medicines; and other (including home health care, vision aids, and medical supplies). Because expenditures for over-the-counter drugs were not included in MEPS and the 1987 NMES, we also excluded such expenditures in analyses of the earlier surveys.

We analyzed payments from six sources: private insurers; Medicare; Medicaid; other public payers (including workers compensation, the Department of Veterans Affairs, TRICARE [civilian coverage for military personnel and dependents], and other government sources); out of pocket; and unclassified.

We also analyzed income-related trends in self-reported health status (which correlates closely with more objective measures of health). ^10,11 MEPS respondents rated their health on a five-point scale: excellent, very good, good, fair, or poor. The earlier surveys used a four-point scale: excellent, good, fair, or poor. To harmonize the scales, we recorded “very good” responses in MEPS as halfway between “good” and “excellent.” We then calculated a mean health status score for each income quintile relative to the mean for the total population each year. We explored different definitions of this variable, which yielded substantially similar results.

We evaluated time trends in health spending according to income quintile using linear regression. We chose 2004 as the start of the recent health spending slowdown based on prior national health spending tabulations. ¹² We conducted sensitivity analyses using cut points between 2002 and 2006 (data not shown); the choice of years did not substantially change our estimates.

To assess differences in health expenditures after adjustment for differences in age and health status among income groups, we used multiple linear regression with health status coded categorically on a five-point scale, age coded as a continuous variable, and the bottom income quintile’s health spending for each survey year as the reference group. Analyses with and without adjustment for age and health status yielded similar time trends; in most cases, we report the unadjusted figures. (Detailed age- and health status–adjusted estimates are available in online Appendix Exhibit 1.) ¹³

Analyses were conducted for each data year and income quintile. However, to smooth and simplify our visual presentation of the data, our graphs display two-year moving averages for 1996–2012 (the period for which annual data were available) and pool data for the middle-three quintiles (which followed similar trends).

Finally, to assess whether a small number of very-high-cost patients drove our findings, we repeated our analyses using quantile regression at the fiftieth, seventy-fifth, ninetieth, ninety-fifth, and ninety-seven and a half percentiles of expenditures.

Our estimates incorporated person-level weights that allowed extrapolation to the entire US noninstitutionalized population. The 1963 SHSUE survey was a simple random sample, with each respondent having equal weight. For all other surveys, we used SAS software survey procedures that adjust confidence intervals for the complex survey design.

Analyses were conducted using SAS, version 9.3.

Limitations

Several caveats apply to our findings. Although MEPS and its predecessors accurately reflect national trends in medical spending (and are widely used for analyses of expenditures for population and disease subgroups), they understate total expenditures as tabulated in the National Health Expenditure Accounts (NHEA) because of a number of factors including the exclusion of people in institutions (for example, nursing homes) and the military, alternative medicine (for example, acupuncture), and “non–patient care revenues”; and the underrepresentation of high-expenditure patients. ^14,15 Prior studies examining trends in health spending by age and sex adjusted for this discrepancy by scaling their estimates to data from the NHEA. ¹⁶ However, such adjustment was not possible in our analysis because income data are not available in the NHEA.

Second, the 1963, 1970, 1977, and 1980 surveys recorded charges instead of actual payments for medical services. Because charges often exceed actual payments, the pre-1987 surveys probably overestimate spending. However, this discrepancy would likely be similar across income groups and should not greatly distort income-related trends.

Study Results

Per capita health expenditures grew 549 percent (adjusted for inflation) between 1963 and 2012 ( Exhibit 1 ). As has been widely observed, spending rose rapidly between 1963 and 1987, surged again from 2000 through 2004, started slowing prior to the Great Recession, and slowed further during and after the recession.

SOURCES Authors’ analysis of data from the 1963 and 1970 Surveys of Health Services Utilization and Expenditures; the 1977 and 1980 National Medical Care Utilization and Expenditure Surveys; the 1987 National Medical Expenditure Survey; and the 1996–2012 Medical Expenditure Panel Surveys. NOTES Data represent two-year moving averages for years 1996–2012. Data before 1996 are shown for the survey years only; trends between data points are interpolated. The population was divided in each survey year into income quintiles; for simplification, the middle three quintiles were combined since they followed similar trends. Online Appendix Exhibit 8 displays confidence intervals for all estimates (see Note  13 in text).

Expenditures for the poorest group grew more rapidly than for other Americans from 1963 (before the implementation of Medicare and Medicaid) through 1987, when Medicare accounted for 24.6 percent and Medicaid accounted for 23.8 percent of the bottom quintile’s expenditures; expenditure growth differed little among the top four quintiles. By 1977, expenditures for the poorest quintile exceeded those for all other Americans by 23 percent.

Between 2004 and 2012, per capita expenditures for the poorest quintile fell at a rate of $19.27 annually—3.7 percent over the eight-year period (Appendix Exhibit 2). ¹³ Meanwhile, health expenditures for the wealthiest group outpaced those of the three middle quintiles. While per capita expenditures rose at a rate of $106.04 annually (19.7 percent over eight years) for the wealthiest group, they increased only 12.5 percent during this period for the middle three quintiles. As a result, by 2012 the top income quintile, which until the early 2000s had among the lowest per capita health spending, had the highest expenditures of any income group.

As expected, individuals reporting worse health status had higher health expenditures (data not shown). The lowest income quintile had the worst health status, and the wealthiest the best throughout the study period (Appendix Exhibit 3), ¹³ and shifts in health status did not explain the recent divergence in health expenditures among income groups. The proportion of elderly people in the poorest quintile fell gradually over time, from 15.9 percent in 1963 to 10.9 percent in 2012. After adjustment for health status and age, health expenditures were higher for the top income quintile than for less affluent Americans throughout the fifty-year period ( Exhibit 2 ). In 2000 age- and health status–adjusted per capita health expenditures for individuals in the top quintile were $616 (18 percent) higher than for people in the lowest quintile; the difference increased to $1,743 (43 percent) in 2012 (Appendix Exhibit 1). ¹³

SOURCES Authors’ analysis of data from the 1963 and 1970 Surveys of Health Services Utilization and Expenditures; the 1977 and 1980 National Medical Care Utilization and Expenditure Surveys; the 1987 National Medical Expenditure Survey; and the 1996–2012 Medical Expenditure Panel Surveys. NOTES Data represent two-year moving averages for years 1996–2012. Data before 1996 are shown for the survey years only; trends between data points are interpolated. The population was divided in each survey year into income quintiles, and for simplification, the middle three quintiles were combined since they followed similar trends.

Expenditure trends differed markedly in people older and younger than age sixty-five. From 2004 to 2012, the elderly of all incomes experienced similar, flat expenditure growth, with the poorest fifth continuing to have the highest expenditures ( Exhibit 3 ). In contrast, the nonelderly population experienced a sharp income-based divergence in expenditure growth after 2004; spending grew rapidly in the top income quintile, modestly in the middle-three quintiles, and minimally among the poorest group ( Exhibit 4 ) (see also Appendix Exhibit 4). ¹³

SOURCES Authors’ analysis of data from the 1963 and 1970 Surveys of Health Services Utilization and Expenditures; the 1977 and 1980 National Medical Care Utilization and Expenditure Surveys; the 1987 National Medical Expenditure Survey; and the 1996–2012 Medical Expenditure Panel Surveys. NOTES Data represent two-year moving averages for years 1996–2012. Data before 1996 are shown for the survey years only; trends between data points are interpolated. The population was divided in each survey year into income quintiles, and for simplification, the middle three quintiles were combined since they followed similar trends.

SOURCES Authors’ analysis of data from the 1963 and 1970 Surveys of Health Services Utilization and Expenditures; the 1977 and 1980 National Medical Care Utilization and Expenditure Surveys; the 1987 National Medical Expenditure Survey; and the 1996–2012 Medical Expenditure Panel Surveys. NOTES Data represent two-year moving averages for years 1996–2012. Data before 1996 are shown for the survey years only; trends between data points are interpolated. The population was divided in each survey year into income quintiles, and for simplification, the middle three quintiles were combined since they followed similar trends.

Prescription drug spending grew similarly for all income groups after 2004 (Appendix Exhibit 2). ¹³ However, both inpatient and outpatient expenditures grew rapidly for the wealthiest quintile while remaining flat or actually declining for the poorest group. The income-related divergence in outpatient expenditures reflects both volume and price effects; wealthy individuals had both rising volumes of medical visits (Appendix Exhibit 5) ¹³ and increasing payments per visit. By 2012 the top income quintile made 40 percent more outpatient visits per capita than other Americans, and spending per visit was also higher ($303 versus $241).

After 2004, private insurance expenditures for different income groups diverged strikingly, rising rapidly for the wealthiest quintile while falling for the poorest 20 percent (Appendix Exhibit 2). ¹³ This was true whether average private insurance expenditure was calculated per capita (Appendix Exhibit 2) ¹³ or per continuously enrolled privately insured nonelderly person (Appendix Exhibit 6). ¹³

We explored income-related trends for particular payer and age subgroups, although sample-size limitations preclude firm conclusions about differences. Per capita Medicare expenditure growth on behalf of the poor was slower than for other income groups after 2004, although interpretation of this finding is complex because many poor Medicare beneficiaries have supplemental Medicaid coverage. For those younger than age sixty-five, private insurance expenditures per enrollee fluctuated for the poorest group (these estimates are based on small numbers since relatively few in this group had private coverage), grew modestly for the middle three income quintiles, and sharply for the wealthiest. Medicaid spending per nonelderly recipient declined during this period (Appendix Exhibit 6), ¹³ although no trend was evident in the proportion of total spending for the poorest quintile that was attributable to Medicaid. In contrast, Medicare’s share of spending rose sharply during this period among the nonelderly poor (presumably because of expenditures for people with long-term disability or end-stage renal disease; data not shown).

To explore whether a small number of high-cost patients accounted for the income-based trends we observed, we repeated our analyses using quantile regression. Prior to 2004, the faster growth of expenditures for the poorest group (and slower growth for the wealthiest) was driven by the costliest (and presumably sickest) 10 percent of patients (Appendix Exhibit 7a). ¹³ After 2004, expenditures surged for both low-cost and high-cost affluent patients but fell for both low-cost and high-cost low-income patients (Appendix Exhibit 7b). ¹³

Discussion

The slowdown in health spending growth between 2004 and 2013 was widely reported and much celebrated. ^5,17–20 Our data suggest a sobering interpretation: Slower spending growth (at least through 2012) was concentrated among poor and middle-income Americans, leading to a growing disparity in health expenditures across income groups. It is unclear whether the recent acceleration of spending growth ² will reverse this trend.

The pattern of sharply rising spending for the wealthy and flat or slow growth for others mirrors the widening gap in the consumption of other goods ²¹ and could represent a shift from need-based to income-based receipt of medical care. We fear that it might presage deepening disparities in health outcomes.

Prior to the implementation of Medicaid and Medicare in 1966, the poor had the lowest health expenditures despite their greater medical need, while expenditures for the wealthy were nearly twice as high as those for the poor. Subsequent to these public investments, health spending tracked closer to medical need, with the poorest income quintile having the highest expenditures and the top quintile the lowest. (However, after adjustment for age and health status, the health expenditure gap between income groups was never fully reversed.)

Several factors probably account for the gradual, instead of sharp, upswing in spending for the poorest Americans after the passage of Medicaid and Medicare. First, Medicaid enrollment ramped up over time. Only about half of states implemented the program immediately; although almost all did so by 1971, enrollment did not level off until 1976. Second, many impoverished neighborhoods lacked doctors’ offices and clinics, and it took time to build up the capacity of neighborhood health centers and other clinics in underserved areas. Third, in 1972 Medicare was expanded to two small but expensive nonelderly groups, many of whom were poor: the chronically disabled and patients with end-stage renal disease. As a result, Medicare’s share of expenditures for the nonelderly bottom quintile rose from near zero in 1970 to 5.6 percent in 1977 and 8.0 percent in 1987.

The pattern of higher (unadjusted) expenditures for poorer people persists to this day for the elderly, virtually all of whom have public coverage through Medicare. Among the nonelderly, however, the income-related pattern reversed after 2004; wealthy individuals—who are the healthiest segment of the population—came to have the highest expenditures.

The shift of some lower-income families from private coverage to Medicaid, ⁴ which pays lower fees, could explain part of the widening gap in per visit payments, as well as the drop in out-of-pocket and private insurer expenditures for this group between 2004 and 2012. However, this would not explain the disparate trends in outpatient visit rates, the divergence between the wealthy and the middle class, or the fall in expenditures among poorer people with private insurance.

The slowdown in medical spending growth between 2004 and 2013 was the sum of disparate trends: flat spending for the elderly and poor, slow growth for the nonelderly middle class, and exuberant growth for the nonelderly wealthy. Personal health care expenditures totaled $2.3793 trillion in 2012; ¹⁸ if all income groups had experienced the same growth as the wealthiest quintile between 2004 and 2012, personal health spending would have been approximately $2.5370 trillion in 2012—an increase of $157 billion in that year alone.

Much discussion of the current slowdown in health spending has focused on whether it is primarily caused by structural or cyclical factors. ^{3–5,12,17–20,22} Although cyclical (that is, recession-related) factors could result in the observed income-related trends in health spending, the divergence in spending among the quintiles appears to predate the Great Recession by several years. ^1,5

Some structural changes, such as improved efficiency of providers or slower diffusion of expensive medical technology and drugs, would be expected to affect all income groups uniformly. Other structural changes—particularly increased cost sharing for the privately insured—would be expected to have a greater impact on low- and middle-income people, consistent with our findings. Cost sharing and depressed income among the poor and middle class might explain why spending by the privately insured dropped after 2008 ¹⁹ and why regions of the United States most affected by the Great Recession experienced the slowest growth in health spending for the privately insured between 2007 and 2011. ³

A mix of cyclical and structural factors—the lingering effects of the Great Recession on nonwealthy households and increased cost sharing—seems the best explanation for the rising income-based spending disparities we observed. Wages for most workers have been slow to rebound from the recession. Meanwhile, the percentage of privately insured workers with individual-plan deductibles of at least $2,000 has increased sixfold since 2006. ⁷ Such high-deductible plans cause particularly large drops in health care use among the highest-cost (that is, sickest) enrollees ²³ and disproportionately affect low-wage workers.

The trend toward higher copayments and deductibles seems likely to continue under the Affordable Care Act (ACA). A typical individual silver-tier plan sold through the insurance exchanges carries a $2,907 deductible. ²⁴ While the ACA’s Medicaid expansion will boost expenditures for the eight million newly enrolled, they constitute only 13 percent of the poorest quintile, and several states are imposing cost sharing on Medicaid enrollees, ²⁵ which might dampen usage increases.

The rising income-based disparity in spending suggests a shift from allocation of health care according to need to allocation by willingness (and ability) to pay. It is unclear whether this shift arises from the underuse of needed care among the poor or overuse of unnecessary care by the wealthy. The sharp spending increase among the nonelderly top income group merits further study and could be caused by the widening gap in cost-sharing requirements in private insurance plans for employees of small versus large firms ⁶ (the latter of which tend to pay higher wages), the rise of concierge medical practices, or supply-induced demand. ^26–28 Irrespective of the cause, the pattern suggests that the efficiency of medical spending is declining, with an increasing share of medical resources devoted to people with the least medical need.

Conclusion

Increasing income inequality has drawn much attention in recent years. Our findings suggest that inequality in health care spending is also on the rise: Expenditures for the poorest (and sickest) segment of the population are actually falling, while those for the wealthy are growing rapidly and now exceed those for other Americans. This pattern, which has not been seen since before Medicare and Medicaid were introduced, could portend a widening of disparities in health outcomes. ^29,30

ACKNOWLEDGMENTS

This study was presented at the national meeting of the Society of General Internal Medicine, April 24, 2015, in Toronto, Ontario. Samuel Dickman received funding support from the Center for Primary Care at Harvard Medical School. Benjamin Smith provided valuable feedback on early drafts of the article.

NOTES

• 1 Roehrig C , Turner A , Hughes-Cromwick P , Miller G . When the cost curve bent—pre-recession moderation in health care spending . N Engl J Med . 2012 ; 367 ( 7 ): 590 – 3 . Crossref, Medline, Google Scholar
• 2 Martin AB , Hartman M , Benson J , Catlin A . National health spending in 2014: faster growth driven by coverage expansion and prescription drug spending . Health Aff (Millwood) . 2016 ; 35 ( 1 ): 150 – 60 . Go to the article, Google Scholar
• 3 Dranove D , Garthwaite C , Ody C . Health spending slowdown is mostly due to economic factors, not structural change in the health care sector . Health Aff (Millwood) . 2014 ; 33 ( 8 ): 1399 – 406 . Go to the article, Google Scholar
• 4 Henry J. Kaiser Family Foundation . Assessing the effects of the economy on the recent slowdown in health spending [Internet]. Menlo Park (CA) : KFF ; 2013 Apr 22 [cited 2016 Jun 6 ]. Available from: http://kff.org/health-costs/issue-brief/assessing-the-effects-of-the-economy-on-the-recent-slowdown-in-health-spending-2/ Google Scholar
• 5 Cutler DM , Sahni NR . If slow rate of health care spending growth persists, projections may be off by $770 billion . Health Aff (Millwood) . 2013 ; 32 ( 5 ): 841 – 50 . Go to the article, Google Scholar
• 6 Banks J , Marmot M , Oldfield Z , Smith JP . Disease and disadvantage in the United States and in England . JAMA . 2006 ; 295 ( 17 ): 2037 – 45 . Crossref, Medline, Google Scholar
• 7 Claxton G , Rae M , Panchal N , Damico A , Bostick N , Kenward K , et al. Employer health benefits: 2014 annual survey [Internet]. Menlo Park (CA) : Henry J. Kaiser Family Foundation ; 2014 [cited 2016 Jun 9 ]. Available from: https://kaiserfamilyfoundation.files.wordpress.com/2014/09/8625-employer-health-benefits-2014-annual-survey6.pdf Google Scholar
• 8 DeNavas-Walt C , Proctor BD , Smith JC . Income, poverty, and health insurance coverage in the United States: 2011 . Washington (DC) : Census Bureau ; 2012 Sep . (Report No. P60-243). Google Scholar
• 9 Crawford MChurch JAkin B , editors. CPI detailed report: data for December 2015 [Internet]. Washington (DC) : Department of Labor ; 2015 [cited 2016 Jun 9 ]. Available from: http://www.bls.gov/cpi/cpid1512.pdf Google Scholar
• 10 Schnittker J , Bacak V . The increasing predictive validity of self-rated health . PLoS One . 2014 ; 9 ( 1 ): e84933 . Crossref, Medline, Google Scholar
• 11 Mavaddat N , Parker RA , Sanderson S , Mant J , Kinmonth AL . Relationship of self-rated health with fatal and non-fatal outcomes in cardiovascular disease: a systematic review and meta-analysis . PLoS One . 2014 ; 9 ( 7 ): e103509 . Crossref, Medline, Google Scholar
• 12 Blumenthal D , Stremikis K , Cutler D . Health care spending—a giant slain or sleeping? N Engl J Med . 2013 ; 369 ( 26 ): 2551 – 7 . Crossref, Medline, Google Scholar
• 13 To access the Appendix, click on the Appendix link in the box to the right of the article online.
• 14 Aizcorbe A , Liebman E , Pack S , Cutler DM , Chernew ME , Rosen AB . Measuring health care costs of individuals with employer-sponsored health insurance in the U.S.: a comparison of survey and claims data . Stat J IAOS . 2012 ; 28 ( 1–2 ): 43 – 51 . Medline, Google Scholar
• 15 Selden TM , Levit KR , Cohen JW , Zuvekas SH , Moeller JF , McKusick D , et al. Reconciling medical expenditure estimates from the MEPS and the NHA, 1996 . Health Care Financ Rev . 2001 ; 23 ( 1 ): 161 – 78 . Medline, Google Scholar
• 16 Meara E , White C , Cutler DM . Trends in medical spending by age, 1963–2000 . Health Aff (Millwood) . 2004 ; 23 ( 4 ): 176 – 83 . Go to the article, Google Scholar
• 17 Orszag P . Economy can’t be all that’s slowing health costs . Bloomberg View [serial on the Internet]. 2013 Sep 24 [cited 2016 Jun 6 ]. Available from: http://www.bloomberg.com/news/2013-09-24/economy-can-t-be-all-that-s-slowing-health-costs.html Google Scholar
• 18 Hartman M , Martin AB , Lassman D , Catlin A National Health Expenditure Accounts Team . National health spending in 2013: growth slows, remains in step with the overall economy . Health Aff (Millwood) . 2015 ; 34 ( 1 ): 15 – 60 . Go to the article, Google Scholar
• 19 Ryu AJ , Gibson TB , McKellar MR , Chernew ME . The slowdown in health care spending in 2009–11 reflected factors other than the weak economy and thus may persist . Health Aff (Millwood) . 2013 ; 32 ( 5 ): 835 – 40 . Go to the article, Google Scholar
• 20 Chandra A , Holmes J , Skinner J . Is this time different? The slowdown in healthcare spending [Internet]. Cambridge (MA) : National Bureau of Economic Research ; 2013 Dec [cited 2016 Jun 6 ]. (Working Paper No. 19700). Available from: http://www.nber.org/papers/w19700 Google Scholar
• 21 Cynamon BZ , Fazzari SM . Inequality, the Great Recession, and slow recovery [Internet]. Rochester (NY) : Social Science Research Network ; 2014 Oct 24 [cited 2016 Jun 6 ]. Available from: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2205524 Google Scholar
• 22 Levine M , Buntin M . Why has growth in spending for fee-for-service Medicare slowed? [Internet]. Washington (DC) : Congressional Budget Office ; 2013 Aug [cited 2016 Jun 6 ]. (Working Paper No. 2013-06). Available from: http://www.cbo.gov/sites/default/files/cbofiles/attachments/44513_MedicareSpendingGrowth-8-22.pdf Google Scholar
• 23 Brot-Goldberg ZC , Chandra A , Handel BR , Kolstad JT . What does a deductible do? The impact of cost-sharing on health care prices, quantities, and spending dynamics [Internet]. Cambridge (MA) : National Bureau of Economic Research ; 2015 Oct [cited 2016 Jun 6 ]. (Working Paper No. 21632). Available from: http://www.nber.org/papers/w21632 Google Scholar
• 24 Coleman K , Geneson J . Deductibles, out-of-pocket costs, and the Affordable Care Act [Internet]. Mountain View (CA) : Health Pocket ; 2013 Dec 12 [cited 2016 Jun 6 ]. Available from: http://www.healthpocket.com/healthcare-research/infostat/2014-obamacare-deductible-out-of-pocket-costs#.VJXx9ABo Google Scholar
• 25 Dickson V . Medicaid cost-sharing could reduce enrollment, experts warn . Mod Healthc . 2014 ; 44 ( 38 ): 12 . Google Scholar
• 26 Fisher ES , Wennberg JE . Health care quality, geographic variations, and the challenge of supply-sensitive care . Perspect Biol Med . 2003 ; 46 ( 1 ): 69 – 79 . Crossref, Medline, Google Scholar
• 27 Enterline PE , Salter V , McDonald AD , McDonald JC . The distribution of medical services before and after “free” medical care—the Quebec experience . N Engl J Med . 1973 ; 289 ( 22 ): 1174 – 8 . Crossref, Medline, Google Scholar
• 28 Beck RG , Horne JM . Utilization of publicly insured health services in Saskatchewan before, during, and after copayment . Med Care . 1980 ; 18 ( 8 ): 787 – 806 . Crossref, Medline, Google Scholar
• 29 National Academies of Sciences, Engineering, and Medicine . The growing gap in life expectancy by income: implications for federal programs and policy responses . Washington (DC) : National Academies Press ; 2015 . Google Scholar
• 30 Case A , Deaton A . Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century . Proc Natl Acad Sci USA . 2015 ; 112 ( 49 ): 15078 – 83 . Crossref, Medline, Google Scholar