THE EFFECTS OF ELIMINATING COST-SHARING ON THE USE OF PREVENTIVE SERVICES FOLLOWING ACA IMPLEMENTATION

Transcription

1 THE EFFECTS OF ELIMINATING COST-SHARING ON THE USE OF PREVENTIVE SERVICES FOLLOWING ACA IMPLEMENTATION A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy In Public Policy By Meredith Freed, B.A. Washington, DC March 22, 2016

2 Copyright 2016 by Meredith Freed All Rights Reserved ii

3 THE EFFECTS OF ELIMINATING COST-SHARING ON THE USE OF PREVENTIVE SERVICES FOLLOWING ACA IMPLEMENTATION Meredith Freed, B.A. Thesis Advisor: Yuriy Pylypchuck, Ph.D. ABSTRACT One of the many goals of the Affordable Care Act is disease prevention. One strategy the ACA employs to promote prevention is the elimination of cost-sharing for preventive services, with the motivation being that reducing this barrier to access will increase the use of these services. This study seeks to determine whether this ACA provision is increasing the utilization of preventive services for those with private insurance. I analyze the receipt of three preventive services cholesterol screening, blood pressure screening, and flu vaccination before and after the implementation of the ACA. Results of the analysis suggest that for those with private insurance, the receipt of cholesterol screenings and flu vaccinations increased, but there was no effect for blood pressure screenings. The analysis suggests that this provision of the ACA is having the intended effect for some preventive services, but not all. iii

4 The research and writing of this thesis is dedicated to everyone who helped along the way. Many thanks, Meredith Freed iv

5 TABLE OF CONTENTS Introduction... 1 Institutional Background and Literature Review... 3 Methodological Approach and Underlying Conceptual Model Empirical Model and Estimation Strategy Description of Data Results Policy Implications Limitations Conclusions and Suggestions for Future Research Endnotes References v

6 LIST OF TABLES Table 1: Descriptive Statistics of Sample Population Table 2: Descriptive Statistics of Eligible Adults with Private Insurance Table 3: Descriptive Statistics for Receipt of Preventive Service Table 4: Changes in Use of Preventive Services After the Elimination of Cost-Sharing Following the Implementation of the Affordable Care Act Table 5: Regression Results: Receipt of Preventive Services Table 6: Difference-In-Difference Coefficients Estimated Separately by Wage, Education, Race, and BMI vi

7 INTRODUCTION In 2010, Congress passed the Patient Protection and Affordable Care Act, the largest overhaul of the United States health care system since Medicare and Medicaid were introduced in One of the many objectives of the Affordable Care Act is disease prevention through the increased use of preventive screenings and immunizations. 1 In 2012, more than 117 million Americans had at least one chronic condition. 86% of all health care spending was for people with one or more chronic conditions. 2 Research has shown that greater use of preventive services would save lives and would result in significant savings on U.S. health care spending. 3 With costs often cited as a barrier to care, the ACA attempts to allay some of the financial hardships associated with accessing health care with the goal of improving health outcomes while also defraying health care costs in future years. One of the provisions of the ACA is the requirement, beginning September 23, 2010, that plans in the group or individual market offer a range of preventive services without imposing cost-sharing requirements. 4 There are only a few studies that have looked at the effects of the elimination of cost-sharing on the receipt of those preventive services. Additionally, using data that includes more years after the implementation of this provision will mediate some of the effects of short-term factors affecting prior studies. This study evaluates whether the elimination of cost-sharing for certain preventive services following the implementation of the ACA in 2010 increased the receipt of those services. I am examining changes in the receipt of preventive services in 2008, before the implementation of the ACA cost-sharing provision, and in 2013, after the implementation of the ACA provision. This study uses data from the Medical Expenditure Panel Survey (MEPS) to analyze the receipt of three preventive services cholesterol screening, blood pressure screening, 1

8 and flu vaccination. Using a difference-in-difference approach, and controlling for a number of demographic characteristics, I am evaluating whether the policy change of eliminating costsharing for preventive services resulted in an increase in the receipt of those services. These results will suggest whether this provision of the Affordable Care Act is having its intended effect. 2

9 INSTITUTIONAL BACKGROUND AND LITERATURE REVIEW One of the primary goals of the Affordable Care Act is disease prevention. It is estimated that only 55% of Americans receive the recommended amount of preventive care. 5 Studies have shown that the reduction or elimination of cost-sharing requirements increases utilization of care. 6 In order to encourage individuals to use more preventive services, the ACA removes these financial barriers to access. 7 Section 2713 of the Affordable Care Act requires that nongrandfathered plans in the group or individual market offer a range of preventive services without imposing cost-sharing requirements. 8 In order to be considered a grandfathered plan, the plan must have been in existence prior to March 23, 2010, and cannot make significant coverage changes. These required preventive services include recommendations from the U.S. Preventive Services Task Force (USPSTF), the Advisory Committee on Immunization Practices (ACIP), the Health Resources and Services Administration s (HRSA s) Bright Futures Project, and HRSA and the Institute of Medicine (IOM) committee on women s clinical preventive services. The requirement that insurers cover preventive services without imposing cost-sharing went into effect on September 23, The coverage requirement for women s preventive services went into effect on August 1, Since this provision went into effect, it is estimated that 137 million people have received no-cost coverage for preventive services. 11 Some of these people, however, may already have had access to preventive services without cost-sharing prior to the Affordable Care Act. The no cost-sharing requirements in the ACA apply to plans in the individual and group markets, but further regulations have included public insurance plans as well. Medicare eliminated cost-sharing requirements for all recommended preventive services as of January 1, States must cover preventive services for adults newly eligible for Medicaid under the 3

10 ACA, but this is not required for the adults enrolled in or eligible for traditional Medicaid prior to the ACA s expansion of the program. 13 In order to encourage states to expand coverage of preventive services without cost-sharing for Medicaid, as of January 1, 2013, states can receive a one percentage point increase in their federal Medicaid match rate for preventive services included in the USPSTF recommendations. 14 Many states have expanded or are currently expanding coverage, but there are still a number of states that still impose cost-sharing for preventive services. 15 For all types of health plans, it is expected that the elimination of the costsharing requirements will lead to increased utilization of preventive services with the goal that this will lead to the prevention of many chronic diseases and improve public health. Until this point, there have been few studies that evaluate the effects of the Affordable Care Act s no cost-sharing requirement on the utilization of preventive services. These studies show mixed results on the impact of the no cost-sharing provision. A study by Han, Yabroff, et al., titled Has recommended preventive service use increased after elimination of cost-sharing as part of the Affordable Care Act in the United States? seeks to assess the changes on the use of recommended preventive services before and after the implementation of the ACA s no cost-sharing provision. 16 The authors employ data from the Medical Expenditure Panel Survey (MEPS) from 2009 (before implementation of the no cost-sharing provision) and (after implementation) for those enrolled in private plans and Medicare plans, as well as those who are uninsured. The authors look specifically at certain preventive services including blood pressure check, cholesterol check, flu vaccination, and cervical, breast, and colorectal cancer screening. The authors control for a number of demographic characteristics including: age, gender, race/ethnicity, education, marital status, region, residence in a metropolitan statistical area, and number of chronic conditions. They 4

11 conclude that there was a statistically significant increase in the receipt of blood pressure check, cholesterol check and flu vaccination following implementation, but few statistically significant changes were observed in terms of the cancer screenings, with the increase largely confined to privately insured persons aged years. A study by Jensen, Salloum, et al., evaluates the use of preventive services for Medicare beneficiaries after the elimination of Medicare s cost-sharing requirement on January 1, 2011 as part of the ACA. 17 The authors use data from MEPS from and data from 2012, looking at preventive service utilization before and after the change went into effect. The authors examine eight preventive services including cholesterol check, blood pressure check, flu vaccination, endoscopy, and fecal occult blood test (FOBT) among all beneficiaries, prostate specific antigen test (PSA) among men, and breast examination and mammography among women. As part of their analysis, they look at preventive utilization rates for Medicare seniors overall, and for subgroups based on the nature of their total health insurance, such as those beneficiaries with supplemental private insurance, Managed Care beneficiaries, and beneficiaries with supplemental public coverage. The authors conclude that the ACA's enhancements to Medicare Part B coverage for preventive services have had only modest effects, if any, on the utilization of the eight preventive services examined. Compared to the period before ACA implementation, more Medicare beneficiaries in 2012 reported having a cholesterol check, but a smaller percentage reported having an endoscopy, and the percentage who reported having had a blood pressure check, flu vaccination, FOBT, PSA, breast examination, or mammogram was statistically unchanged. Furthermore, they find that for those in traditional Medicare, who had the most to gain from the elimination of cost-sharing, exhibited no statistical changes to their behavior. The authors do note, however, that this muted response is partly explained by the fact 5

12 that most seniors already had full or near-full insurance for recommended preventive services before the change went into effect. Moreover, this data is only from the first year following the change, so there could be longer-term effects not accounted for here. A study by Fedewa at el., examines the elimination of cost-sharing and receipt of screening for colorectal and breast cancer before and after the implementation of the ACA. 18 The authors use data from the National Health Interview Survey for privately and Medicareinsured adults from 2008 and from The authors conclude that colorectal cancer screening prevalence increased among low-income and lower educated individuals as well as Medicareonly respondents, but not for high-income, highly educated, and privately insured individuals. Despite increases in colorectal cancer screening for respondents with lower income, Medicare insurance, and lower educational attainment, gains in colorectal cancer screening were modest. Unlike colorectal cancer screening, there was no change in breast cancer screening prevalence noted among low-income women and those with lower educational attainment. The authors mention, however, this may be due to fewer pre-aca financial barriers because the cost of mammography is substantially lower than that of colonoscopy, and breast cancer screening is better supported by women s health initiatives. There has been a great deal of literature on whether reductions in cost-sharing increase use of preventive services. Rezayatmand, Pavlova, and Groot have done an extensive literature review on the impact of out-of-pocket costs and use of preventive services. 19 The results of their study indicate that empirical evidence of a causal relation between out-of-pocket payments and prevention is still lacking. Their results show that despite many studies focusing on the relationship between cost-sharing and prevention, the data used are not always collected with the objective of testing this relationship, particularly because most studies are based on claims, 6

13 administrative data, or surveys. Additionally, they contend the majority of studies fail to provide evidence of external validity of their results. Therefore, it is not always possible to interpret or analyze patient behavior in relation to payments. The results of their findings do, however, strongly support the notion that cost-sharing, as a financial barrier, decreases the use of preventive services. The relationship between cost-sharing and utilization is explicitly addressed in a study by Solanki and Schauffler. 20 The authors assess the relationships between utilization of preventive services and different forms of patient cost-sharing and how the effect is mediated by type of preventive service (counseling, blood pressure, Pap smear, and mammogram), type of costsharing (deductibles or copayments), and type of health plan. The authors find that there is significant variation among utilization based on the type of health plan and preventive service, but that cost-sharing has a negative effect on almost all combinations of type of preventive service and health plan. They conclude that the negative effect of patient cost-sharing was greatest on preventive counseling in PPO/indemnity plans (215%) and on mammograms in all health plan types (29% 10%). The effect on Pap smears was negative (28% 10%) for deductibles/coinsurance in PPO/indemnity plans and copayments in HMOs. The effect of costsharing on blood pressure was mixed. They also conclude that deductibles had a greater negative effect than copayments on utilization of preventive services. Other studies have looked at cost-sharing and prevention in relation to a specific type of insurance coverage. A study by Goodwin and Anderson looks at the effect of cost-sharing reductions on Medicare beneficiaries. 21 In 1997, Congress waived deductibles for mammograms and Pap Smears, decreasing the cost-sharing requirements of those procedures. The authors conclude that there was a 20 percent short-term increase and 25 percent long-term increase of 7

14 having a mammogram in the four years following the passage of the legislation. The likelihood of use of Pap Smears did not increase, however, following the passage of the legislation. The authors note that elimination of cost-sharing could be a way to increase use of preventive services, but there could be other factors that may influence its effectiveness such as the characteristics of the procedure and the disease type. A study evaluating the effects of costsharing in relation to Medicaid expansions for childless adults also concludes that those plans which saw increasing cost-sharing requirements did not expand use of preventive screenings compared to those which had a traditional cost-sharing structure. 22 The authors note that this study may not be generalizable because it refers to only childless adults, but it does reinforce existing studies that elimination of cost-sharing requirements has the potential to increase utilization of preventive services. In addition to cost-sharing requirements, several studies look at other barriers to access for preventive services. A study by Sambamoorthi and McAlpine examines the racial, ethnic, and socioeconomic disparities among women for preventive services. 23 The authors assess four different women s preventive services according to the USPSTF recommendations from They determine that the most consistent predictors of the receipt of preventive services were socioeconomic status in terms of both income and education. College educated women were significantly more likely to receive all four preventive services than those without a high school diploma. And those women with a high income were much more likely to receive all four of the services. They also conclude that use of preventive services varied by race and ethnicity. Even when controlling for insurance, usual source of care, and health status, socioeconomic disparities in rates of use of preventive services persisted. 8

15 Another study by McMorrow, Kenney, and Goin demonstrates the differences in the use of preventive services prior to the Affordable Care Act relative to socioeconomic factors. 24 The authors use data from MEPS from to measure the receipt of 8 preventive services prior to the implementation of the ACA no cost-sharing provision. The authors determine that there are significant differences in the use of preventive services depending on income. They find that those with higher incomes were more likely to receive every kind of preventive service examined than those with lower incomes, often as a result of insurance coverage status. They also find a great deal of variation of receipt of preventive services depending on the service itself. While differences in insurance coverage explain 25-40% of the variation in use of preventive services, they conclude that educational attainment, age, and health status are also important drivers of preventive service use. This review of literature demonstrates that while the majority of studies establish that reduced or elimination of cost-sharing often leads to more use of preventive services, this association is not conclusively demonstrated with the studies evaluating the effects of the no cost-sharing provision of Affordable Care Act. Moreover, this relationship can be shaped by the type of preventive service in question, type of insurance coverage, and socioeconomic and demographic factors. 9

16 METHODOLOGICAL APPROACH AND UNDERLYING CONCEPTUAL MODEL Health insurance is related to moral hazard, a situation in which individuals change their behavior because they are insured. This change in behavior can result in two often competing outcomes individuals use more health care than they otherwise would and they engage in more risky behavior than they otherwise would now that they are insured. These aspects of moral hazard are a major driver behind the elimination of the cost-sharing for preventive services. The reason that individuals use more health care than they otherwise would is because they are insured and do not experience the full cost of health care. Yet, even with insurance, health insurance premiums, deductibles, and out-of-pocket costs have become so high that people are often forgoing needed medical care. 25 As noted in the literature review, there are a number of studies that reveal that even when someone has health insurance, cost-sharing is a barrier to accessing care, and that the reduction or elimination of cost-sharing increases access to care. On the other hand, moral hazard in insurance markets leads individuals to take more risks because they are insured and do not bear the full cost of those risks. This relates to health insurance in that because individuals do not experience the full cost of medical care, they do not make as many healthy choices. Instead, because they are insured and can seek medical care in the future if they become seriously ill, they do not take preventive measures in the present. As mentioned previously, only 55% of Americans are receiving the recommended amount of preventive care. The primary goal of the ACA provision is to encourage more utilization of preventive services. The overriding belief is that the financial incentives will be enough to nudge individuals to use more of these services because they would no longer have to pay additional money out-of-pocket, despite the moral hazard associated with being insured. Conversely, the uninsured, which always have to pay out-of-pocket for healthcare, would not be 10

17 affected by an elimination of cost-sharing, and would be unlikely to alter their habits in the wake of this legislation. For this study, I am assessing whether eliminating cost-sharing for preventive services as part of the ACA increased the receipt of certain preventive services. The dependent variables are the receipt of three preventive services - blood pressure screening, cholesterol screening, and flu vaccination - in the recommended time frame according to the U.S. Preventive Services Task Force (USPSTF) and Advisory Committee on Immunization Practices (ACIP) guidelines. The receipt of these services will be based on whether one was either privately insured or uninsured. The independent variables control for a variety of demographic factors that would affect the use of those preventive services including age, gender, education, race/ethnicity, income, marital status, having a usual source of care, region, and BMI. The most important independent variable of interest is the year, specifically whether the receipt of the service was before or after the policy change. 11

18 EMPIRICAL MODEL AND ESTIMATION STRATEGY I am evaluating utilization of preventive services using a difference-in-difference approach. This approach is a method to isolate the treatment effects of a certain policy using non-experimental data. The study employs data from 2008 to evaluate the period before the ACA went into effect, and data from 2013 after the implementation of the provision, limiting the sample to those who are privately insured and those who are uninsured. Using data from 2013, the most recent data available, will account for some time lag in implementation of the provision. Those who have private insurance are serving as the treatment group because the policy change specifically affected this group. Those who are uninsured are serving as the control group because the implementation of the ACA provision did not affect them. While an ideal control group would be one very similar to the treatment group, there are some differences between the uninsured and privately insured populations. The uninsured are disproportionately more likely than the privately insured to be low-income, Hispanic, and young. 26 However, because the implementation of the ACA has been a piecemeal approach across all types of insurance coverage, the uninsured were the only group that remained stable before and after the implementation of the ACA, which is why they are serving as the control. Below is the difference-in-difference equation: Y it = β 0 + β 1 Private + β 2 Year13 + β 3 PrivateYear13 + β 4 age + β 5 gender + Β 6 race/ethnicity + β 7 education + β 8 wage + β 9 marital status + β 10 region + β 11 source of care + β 12 BMI Outcome Y it = receipt of the preventive service of the ith individual in year t Private = 1 if person has private insurance; = 0 if they are uninsured 12

19 Year13 = 1 if year is 2013; = 0 if year is 2008 Private Year13 = 1 if person has private insurance and year is 2013 Other studies on the subject have used adjusted prevalence or prevalence difference to examine the receipt of preventive service pre- and post- reform. By using a difference-indifference approach, the study will provide a clearer view of the implications of this policy decision, and will determine if increased screening and immunization is due to the ACA preventive care provision eliminating cost-sharing. 13

20 DESCRIPTION OF DATA The data comes from the Medical Expenditure Panel Survey (MEPS), which is a nationally representative survey of households in the U.S. MEPS is the most complete source of data on the cost and use of health care in the United States. The study sample includes those who are either privately insured or uninsured ages 18 to 65. The outcomes of interest are receipt of the following preventive services within the past year: cholesterol screening, blood pressure screening, and flu vaccination. The guidelines for cholesterol screening and blood pressure screenings are based on the A and B recommendations from the U.S. Preventive Services Task Force (USPSTF). The guidelines for flu vaccination are based on recommendations from the Advisory Committee on Immunization Practices (ACIP). While aspects of these guidelines have changed over time, the general screening recommendations have remained the same from 2008 to The study populations for each preventive service are defined separately to be consistent with the set of guidelines by the USPSTF and ACIP in order to be covered by the ACA preventive care provision. The USPSTF recommends cholesterol screenings for men age 35 years and older; men age 20 to 35 years if they are at increased risk for coronary heart disease; and women age 20 years and older if they are at increased risk of coronary heart disease. 27 Risk factors for high cholesterol include high blood pressure, smoking, diabetes, being overweight or obese, and not getting enough physical activity. While the USPSTF currently states that the optimal screening level for lipid disorders is uncertain, they recommend shorter intervals for those close to warranting therapy, so this study employs whether someone received a cholesterol screening within the past year as its screening timeframe. The USPSTF recommends screening for high blood pressure in adults age 18 years or older. The USPSTF recommends that those age 40 or 14

21 older or at higher risk for high blood pressure, such as those who are overweight, obese, or African American, should be screened annually. 28 The USPSTF also recommends that those between age 18 and 40 or those not at high risk for high blood pressure should be screened every 3 to 5 years. Those individuals ages 18 to 40 are excluded from the analysis because the screening interval would include time before the implementation of the ACA. ACIP recommends that adults age 18 years or older should receive the flu vaccination annually. 29 The outcome variables are determined by self-reported answers to questions on whether the populations consistent with the guidelines received the preventive service within the past year. The main independent variables of interest are year of survey (2013 vs. 2008) and insurance type (private vs. uninsured). Demographic variables include age (18-65 years old), gender, race (White, Black, Hispanic, Asian, Other); level of education (less than 8 th grade, no high school diploma, high school diploma, some college, college degree or higher); wage (low income: $0-$25,000, middle income: $25,000-$75,000, high income: $75,000 +), marital status (married or not married); region (Northeast, Midwest, South, and West); having a usual source of care; and BMI (underweight, normal weight, overweight, and obese). Both descriptive statistics and regression results were conducted using population survey weights. The regression results are calculated using a linear probability model. The differencein-difference estimates represent the percentage point effects of the policy change on the privately insured population. All analyses were conducted in STATA 13.1 and

22 RESULTS TABLE 1: DESCRIPTIVE STATISTICS OF SAMPLE POPULATION Characteristic Gender Male Female 50.99% 49.01% 50.55% 49.45% Age Race White Black Asian Hispanic Other Education Less than 8 th Grade No HS Diploma HS Diploma Some College College Degree or More Education Missing Wage Low Income Middle Income High Income Has Usual Source of Care Yes No BMI Underweight Normal Weight Overweight Obese BMI Missing Region Northeast Midwest South 13.20% 11.17% 10.24% 11.18% 11.65% 11.52% 31.04% 67.66% 10.79% 5.16% 14.55% 1.83% 3.75% 9.94% 29.26% 25.20% 31.28%.57% 45.78% 43.93% 10.29% 71.59% 28.41% 1.75% 35.00% 34.18% 26.85% 2.22% 17.59% 21.91% 36.92% % 10.68% 10.73% 10.68% 10.12% 10.66% 32.79% 64.92% 10.31% 6.39% 16.29% 2.10% 3.05% 11.15% 21.26% 31.90% 32.09%.55% 46.03% 41.70% 12.27% 71.42% 28.58% 1.81% 35.50% 32.46% 28.01% 2.22% 17.60% 21.43% 37.47%

23 West Region Missing Marital Status Married Not Married Health Insurance Private Insurance Uninsured 23.59% 0.00% 57.54% 42.46% 78.11% 21.89% 21.43% 2.07% 56.12% 43.88% 77.19% 22.81% TABLE 2: DESCRIPTIVE STATISTICS OF ELIGIBLE ADULTS WITH PRIVATE INSURANCE Gender Male Female Age Race White Black Asian Hispanic Race Other Education Less than 8 th Grade No HS Diploma HS Diploma Some College College Degree or More Education Missing Wage Low Income Middle Income High Income % 50.69% 11.19% 10.01% 10.21% 11.53% 11.91% 11.78% 33.37% 73.30% 9.85% 5.47% 9.59% 1.79% 1.55% 7.14% 27.30% 26.35% 37.30%.36% 36.44% 51.03% 12.53% 48.81% 51.19% 13.69% 9.49% 10.11% 10.52% 10.25% 10.93% 35.01% 71.09% 9.34% 6.62% 10.73% 2.22% 1.18% 8.41% 19.00% 32.39% 38.74%.29% 36.52% 48.07% 15.41% Has Usual Source of Care Yes 79.56% 79.91% 17

24 No 20.44% 20.09% BMI Underweight Normal Weight Overweight Obese 1.59% 35.62% 33.85% 27.01% 1.72% 35.76% 32.86% 27.81% Region Northeast Midwest South West Marital Status Married Not Married 19.04% 23.30% 34.65% 23.02% 62.55% 37.45% 19.26% 22.69% 35.52% 22.52% 61.42% 38.58% Table 1 provides descriptive statistics for each of the independent variables that serve as controls in the analysis. The sample population contains all adults age 18 to 65 and those who are either privately insured or uninsured. The sample populations were mostly consistent across years, though the population in 2013 was slightly more racially diverse. Table 2 provides descriptive statistics of just the privately insured population. There are some differences between the privately insured population compared to the population as a whole. Compared to the entire sample population, the privately insured population was more proportionately white % of the privately insured population was white, while 64.92% of the entire sample was white in Unsurprisingly, there was a higher proportion of Hispanics in the entire population versus just the privately insured population. In 2013, 16.29% of the sample population was Hispanic, while 10.73% of the privately insured population was Hispanic. There were also a higher number of those with college educations in the privately insured population versus the entire sample in 2013, at 38.74% and 32.09%, respectively. The reverse holds true for those who are low-income. In the sample as a whole, 46.03% of the individuals were low-income, while in the privately insured-only population, 36.52% of the individuals were 18

25 low-income in These differences in the sample populations reflect some of the differences between the privately insured and uninsured. However, the rates for BMI, and those with high school diplomas and some college were very similar across the two sample populations. TABLE 3: DESCRIPTIVE STATISTICS FOR RECEIPT OF PREVENTIVE SERVICE Cholesterol Screening Blood Pressure Screening Flu Vaccination Health Insurance Private Insurance Uninsured Gender Male Female Age Race White Black Asian Hispanic Other Education Less than 8 th Grade No HS Diploma HS Diploma Some College College Degree or More Wage Low Income Middle Income High Income Has Usual Source of Care Yes No 34.07% 4.88% 21.08% 17.87%.96%.18% 2.55% 3.89% 4.94% 5.73% 19.08% 27.24% 4.78% 1.53% 4.68%.73% 1.28% 2.53% 11.24% 10.21% 13.53% 14.50% 19.37% 5.08% 33.39% 5.73% 40.03% 5.67% 22.59% 23.11% 2.04% 2.85% 3.72% 4.57% 5.10% 6.02% 21.39% 30.26% 5.48% 2.78% 6.31%.73% 1.21% 3.20% 9.45% 14.64% 16.97% 16.60% 21.75% 7.34% 38.76% 7.04% 42.77% 6.55% 23.19% 26.13% % 9.09% 26.13% 36.03% 5.33% 1.98% 5.09%.89% 1.53% 3.15% 14.39% 12.52% 17.46% 18.36% 24.16% 6.79% 42.47% 7.24% 42.01% 6.93% 22.98% 25.96% % 8.22% 27.80% 34.54% 5.30% 2.54% 5.74%.81% 1.36% 3.41% 10.16% 15.33% 18.44% 17.27% 23.12% 8.55% 41.62% 7.59% 24.52% 2.79% 11.78% 15.53%.18% 2.05% 1.86% 2.61% 2.92% 3.20% 12.87% 20.44% 2.23% 1.52% 2.55%.57%.61% 1.55% 6.95% 6.93% 11.09% 10.70% 12.75% 3.87% 23.39% 4.01% 32.71% 3.96% 15.35% 21.32% 3.75% 3.24% 3.55% 3.92% 3.35% 3.87% 14.99% 25.73% 3.19% 2.42% 4.67%.65%.76% 2.77% 6.15% 11.71% 15.14% 14.24% 16.13% 6.29% 30.93% 5.90% 19

26 BMI Underweight Normal Weight Overweight Obese Region Northeast Midwest South West Marital Status Married Not Married.17% 6.59% 16.47% 15.18% 7.67% 8.39% 14.77% 8.12% 26.55% 12.40%.31% 10.26% 17.53% 16.95% 8.35% 9.85% 17.94% 9.55% 30.82% 14.88%.42% 12.70% 18.45% 16.81% 9.39% 11.20% 17.88% 10.85% 34.78% 14.54%.48% 12.56% 17.91% 17.32% 8.93% 10.95% 18.73% 10.33% 34.33% 14.61%.39% 8.86% 9.01% 8.36% 5.22% 6.62% 9.39% 6.08% 18.35% 8.97%.57% 12.45% 12.35% 10.80% 6.97% 8.05% 13.19% 8.46% 23.48% 13.19% For each preventive service, only populations with age range, gender, and risk factors consistent with USPSTF and ACIP recommendations associated with the ACA no cost-sharing provision were included. Table 3 provides descriptive statistics of the receipt of each preventive service according to the USPSTF and ACIP guidelines. The receipt of each service is measured across years for each of the independent control variables. In 2008, 34.7% of those with private insurance received a cholesterol screening, while in 2013, 40.03% of those with private insurance received the screening. In 2008, 24.52% of those with private insurance received a flu vaccination, while in 2013, 32.71% of those with private insurance received the vaccination. While these preventive services increased after ACA implementation, the receipt of blood pressure screenings was essentially flat with 42.77% receiving a blood pressure screening in 2008 and 42.01% receiving a screening in These trends were consistent across almost each of the independent control variables. For the uninsured, the number of those who received screenings did not increase substantially. In 2008, 4.88% of the uninsured received a cholesterol screening, while in 2013, 5.67% of the uninsured received the screening. In 2008, 6.55% of the uninsured received a blood pressure screening, while in 2013, 6.93% of the uninsured received the screening. In 2008, 2.79% of the uninsured received a flu vaccination, while in 2013, 3.96% of 20

27 the uninsured received the vaccination. Further analysis will show whether these changes from 2008 to 2013 for the privately insured are statistically significant. TABLE 4: CHANGES IN USE OF PREVENTIVE SERVICES AFTER THE ELIMINATION OF COST-SHARING FOLLOWING THE IMPLEMENTATION OF THE AFFORDABLE CARE ACT Private Insurance Uninsured D-in-D Cholesterol Screening Blood Pressure Screening Flu Vaccination Table 4 provides the difference-in-difference estimates for each of the preventive services without including any of the control variables. These initial results show that for those with private insurance in 2013, there was an increase of 5.17 percentage points in their receipt of cholesterol screenings as a result of the policy. Those with private insurance in 2013 also saw an increase of 7.02 percentage points in their receipt of flu vaccinations as a result of the policy. This is what we would expect as the result of the implementation of the policy. In contrast, the results show that there would be a 1.14 percentage point decrease in the receipt of blood pressure screenings. However, these results do not indicate whether the difference-in-difference estimates are significant, as Table 5 shows. TABLE 5: REGRESSION RESULTS: RECEIPT OF PREVENTIVE SERVICES Cholesterol Screening 21 Blood Pressure Screening Flu Vaccination D-in-D Coefficient.0671** (.0134).0139 (.0118).0687** (.0141) Year 2008 (ref group) (.0112) (.0099).0382** (.0110) Health Insurance Private Insurance.1068** (.0096).0797** (.0076).0836** (.0105) Uninsured (ref group) Gender Male Female (ref group).0220** (.0059) ** (.0045) ** (.0068)

28 Age (ref group) Race White (ref group) Black Asian Hispanic Race Other Education Less than 8 th Grade No HS Diploma HS Diploma Some College College Degree or More (ref group) Wage Low Income (ref group) Middle Income High Income Has Usual Source of Care Yes No (ref group) BMI Underweight Overweight Obese Normal Weight (ref group) Region Midwest (ref group) Northeast South West Marital Status Married Not Married.0578** (.0119).0940** (.0132).1617** (.0138).2239** (.0138).2857** (.0132).3780** (.0100).0882** (.0094).0511* (.0124).0571** (.0088).0189 (.0209) * (.0165) ** (.0114) ** (.0091) (.0089).0107 (.0078).0177 (.0135) ** (.0048) ** (.0056).4349** (.0122).6952** (.0111).7132** (.0096).7702** (.0077).0286** (.0059) ** (.0088).0015 (.0015).0069 (.0152) ** (.0138) ** (.0092) ** (.0065) * (.0058).0085 (.0055).0201* (.0097).0056 (.0130) (.0139).0198 (.0154).0061 (.0140).1332* (.0136).4503** (.0118) ** (.0100).0182 (.0151).0026 (.0081) (.0279) ** (.0146) ** (.0144) ** (.007) ** (.0109) (.0089).0598** (.0138).1616** (.0078).1165** (.0067).1107** (.0089) (.0182).2160** (.0084).2704** (.0082).0387** (.0110).0077** (.0101) (.0116).0120 (.0126).0929** (.0058).1200** (.0065).0111 (.0070).0170** (.0060) (.0069) (.0269).0235** (.0082).0420** (.0093).0038 (.0168).0007 (.0123) (.0125).0284** (.0071).0022 (.0053).0278** (.0082) **p-value <.01 *p-value.01<p<.05 For each preventive service, only populations with age range, gender, and risk factors consistent with USPSTF and ACIP recommendations associated with the ACA no cost-sharing provision were included. 22

29 Table 5 includes the regression results and difference-in-difference estimates for each preventive service. The results were mixed as to the ACA preventive care provision s effect on the use of preventive services. The results illustrate that there was a statistically significant increase (p-value <.01) in the receipt of cholesterol screenings and flu vaccinations after the policy change, but that there was not a statistically significant change for blood pressure screenings. The elimination of cost-sharing resulted in a 6.71 percentage point increase in the receipt of cholesterol screenings for those with private insurance. Similarly, the elimination of cost-sharing resulted in a 6.87 percentage point increase in the receipt of flu vaccinations for those with private insurance. For blood pressure screenings, there does not appear to be a statistically significant change in the receipt of these screenings following the implementation of this policy. These findings suggest that the ACA preventive care provision is having the intended effect for some preventive screenings, but not all. TABLE 6: DIFFERENCE-IN-DIFFERENCE COEFFICIENTS ESTIMATED SEPARATELY BY WAGE, EDUCATION, RACE, AND BMI Wage Low Income Middle Income High Income Education Less than a College Degree College Degree or More Race White Black Hispanic Difference-in-Difference Coefficients When Limiting Sample to Each Sub-Category Cholesterol Screening.0034 (.0101).0548** (.0092).0090* (.0039).0447** (.0123).0241** (.0064).0586** (.0120).0038 (.0043) (.0064) Blood Pressure Screening (.0088).0289** (.0095).0008 (.0043).0103 (.0109).0031 (.0075).0120 (.1220).0018 (.0440) (.0070) Flu Vaccination (.0106).0578** (.0082).0155** (.0039).0221 (.0021).0473** (.0076).0628** (.0628).0066 (.0036) (.0069) BMI Overweight or Obese.0197 (.0128) (.0108).0345** (.0112) **p-value <.01 *p-value.01<p<.05 23

30 For each preventive service, only populations with age range, gender, and risk factors consistent with USPSTF and ACIP recommendations associated with the ACA no cost-sharing provision were included. Table 6 provides difference-in-difference estimates for each of the screenings when the sample is limited to certain demographic sub-populations. Those who were middle-income and received the screenings saw a statistically significant increase in the receipt of cholesterol screenings, blood pressure screenings, and flu vaccinations. Holding all other factors constant, the policy resulted in a 5.48 percentage point increase for cholesterol screenings, a 2.89 percentage point increase in blood pressure screenings, and a 5.78 percentage point increase flu vaccinations for those privately insured and middle-income in These results are in the direction one would expect for the elimination of cost-sharing. However, there was no statistically significant change for those who were low-income. There was also a statistically significant increase in cholesterol screenings and flu vaccinations for those with high incomes, though there was no statistically significant change for blood pressure screenings. This trend also applied to those who were college educated and white, with statistically significant increases in in cholesterol screenings and flu vaccinations, but no statistically significant change in blood pressure screenings. There was a statistically significant increase in cholesterol screenings for those with less than a college degree though there was no statistically significant effect for blood pressure screenings or flu vaccinations. Holding all other factors constant, the policy resulted in a 4.47 percentage point increase in cholesterol screenings for those with less than a college degree. There was a statistically significant increase in flu vaccinations for those who were overweight or obese though there was no statistically significant effect for cholesterol screenings or blood pressure screenings. Holding all other factors constant, the policy resulted in a 3.45 percentage point increase in flu vaccinations for those who were overweight or obese. Lastly, there was no significant effect of the policy on those who were black or Hispanic. 24

31 POLICY IMPLICATIONS These results highlight that the policy is working for some preventive screenings, though not all. Overall, there was a statistically significant effect of the policy with increases in the receipt of cholesterol screenings and flu vaccinations, but not blood pressure screenings. The policy also had a significant positive effect for certain demographic sub-groups, such as those who were middle-income and high-income, white, and college educated. On the other hand, the policy did not have a statistically significant effect on some of the more vulnerable populations in which we might hope to see such changes such as those who are low-income, Hispanic, or black. This could be because even with insurance coverage, these populations are less likely to seek care. Although not all three preventive screenings increased, these results are promising, and 2013 was still early on in the timeline of the implementation of the ACA. Since 2013, millions more people have become insured as a result of the exchanges and Medicaid expansions. 30 The ACA s impact on the utilization of preventive services may be greater in the next few years as the ACA continues to become more established and embedded in the U.S. health care system. However, the cost-sharing provision itself has suffered from implementation challenges that could affect overall use of preventive services. For example, an office visit and preventive service might be billed separately, but the insurer may impose costsharing for the visit itself. Additionally, if the primary reason for the visit was not a preventive service, patients may still have to pay cost-sharing for the visit. 31 These issues will hopefully be addressed as implementation of the ACA continues. Moreover, there still may be a general lack of awareness that it is possible to receive these preventive screenings without cost-sharing, so some individuals may still be postponing care. If the rate of preventive screenings do continue to 25

32 increase, it will be interesting to see whether there are improvements in public health as well as if the government experiences any cost savings. 26

33 LIMITATIONS There are some limitations to this study. First, some of the private plans are still considered grandfathered meaning they were not subject to the elimination of cost-sharing, though the number of these plans continues to decrease. 32 This grandfathered status would impact the receipt of preventive services, most likely understating the effect of the policy change. Second, these results are taken from MEPS, which is a self-reported survey, and people often do not remember exactly when they last received screenings and immunizations. However, this recall error would apply across survey years, so it is unlikely it would appreciably change the results. Third, this preventive care provision and overall ACA implementation occurred in the period after the financial crisis and during the ensuing recovery, which could have impacted the utilization of health care, though the receipt of screenings remained relatively stable for the uninsured so those effects may not be considerable. Lastly, as mentioned previously, the uninsured as the control group may have impacted the results because there are some differences in the populations and their rate of screenings were considerably different at the outset. However, by examining certain sub-populations that are similar across groups, the effects of these differences have been mitigated. 27

34 CONCLUSIONS AND SUGGESTIONS FOR FUTURE RESEARCH This paper extends the existing literature on the use of preventive services following the elimination of cost-sharing as part of the implementation of the ACA. It continues to confirm that there have been mixed results on the utilization of preventive services, suggesting there have been increases in cholesterol screenings and flu vaccinations, but no increases for blood pressure screenings. Furthermore, by employing a difference-in-difference approach, it attributes these increases to the policy change itself instead of solely looking at the rate of screenings pre-post analysis. Further analysis will indicate whether this policy change will have an effect on the use of preventive services years into the future and whether the United States can stem the tide of chronic disease. If the rate of receipt of recommended preventive screenings continues to increase, it would be interesting to examine the impact the screenings have on actually preventing chronic disease and what cost-savings these preventive screenings can achieve. 28

35 ENDNOTES 1 Howard K. Koh and Kathleen G. Sebelius, Promoting Prevention Through the Affordable Care Act, New England Journal of Medicine, September 30, CDC, Chronic Disease Overview, August 26, Michael V. Maciosek1, Ashley B. Coffield, Thomas J. Flottemesch, Nichol M. Edwards, and Leif I. Solberg, Greater Use Of Preventive Services In U.S. Health Care Could Save Lives At Little Or No Cost, Health Affairs 29 (2010): CFR , 29 CFR , 45 CFR Elizabeth A. McGlynn, Steven M. Asch, John Adams, Joan Keesey, et al., The Quality of Health Care Delivered to Adults in the United States, New England Journal of Medicine. January 26, J.P., Newhouse, Free for All?: Lessons From the Rand Health Insurance Experiment (Cambridge, Mass: Harvard University Press, 1993). 7 Koh and Sebelius, Promoting Prevention Through the Affordable Care Act, 8 The Patient Protection and Affordable Care Act, Sec. 2713, Coverage of Preventive Health Services. 9 Kaiser Family Foundation, Preventive Services Covered by Private Health Plans under the Affordable Care Act, August Ibid. 11 ASPE, The Affordable Care Act is Improving Access to Preventive Services for Millions of Americans, May CMS, Reminder Beneficiary Cost-Sharing for Medicare-Covered Preventive Services Under the Affordable Care Act, May 8, MLN/MLNMattersArticles/downloads/SE1129.pdf. 13 Sara E. Wilensky and Elizabeth A. Gray, Existing Medicaid Beneficiaries Left Off The Affordable Care Act s Prevention Bandwagon, Health Affairs. July 2013: Vol Kaiser Family Foundation, Coverage of Preventive Services for Adults in Medicaid, November Ibid. 16 Xuesong Han, K. Robin Yabroff, Gery P. Guy Jr., Zhiyuan Zheng, et al. Has recommended preventive service use increased after elimination of cost-sharing as part of the Affordable Care Act in the United States? Preventive Medicine. July 23, Gail A. Jensen, Ramzi G. Salloum, Jianhui Hu, Nasim Baghban Ferdows, and Wassim Tarraf, A slow start: Use of preventive services among seniors following the Affordable Care Act's enhancement of Medicare benefits in the U.S., Preventive Medicine. April 18, Stacey A. Fedewa, Michael Goodman, W. Dana Flanders, Xuesong Han, et al., Elimination of Cost-Sharing and Receipt of Screening for Colorectal and Breast Cancer, Cancer. September Reza Rezayatmand, Milena Pavlova, and Wim Groot, The impact of out-of-pocket payments on prevention and health-related lifestyle: a systematic literature review, European Journal of Public Health. April 2012: Vol. 23, No. 1, Geetesh Solanki and Helen Halpin Schauffler, Cost-Sharing and the Utilization of Clinical Preventive Services, American Journal of Preventive Medicine. 1999: Vol. 7 No Suzanne M. Goodwin and Gerard F. Anderson, Effect of Cost-Sharing Reductions on Preventive Service Use Among Medicare Fee-for-Service Beneficiaries Preventive Service Use, Medicare & Medicaid Research Review. 2012: Volume 2, Number Gery P. Guy Jr.; The Effects of Cost Sharing on Access to Care among Childless Adults, HSR: Health Services Research. December 2010: 45:6, Part I. 23 Usha Sambamoorthi and Donna D. McAlpine, Racial, ethnic, socioeconomic, and access disparities in the use of preventive services among women, Preventive Medicine. 2003: Vol. 37; Stacey McMorrow, Genevieve M. Kenney, and Dana Goin, Determinants of Receipt of Recommended Preventive Services: Implications for the Affordable Care Act, American Journal of Public Health. December 2014: Vol. 104, No