The united states experienced an unprecedented slowdown. Trends In Out-Of-Pocket Spending By Insured American Workers,

Transcription

1 Trends In Out-Of-Pocket Spending By Insured American Workers, The 1990s were kind to insured health care consumers lower out-of-pocket spending offset rising premiums. by Jon R. Gabel, Paul B. Ginsburg, Jeremy D. Pickreign, and James D. Reschovsky ABSTRACT: This paper examines trends in out-of-pocket spending for insured workers from 1990 to Data are from the Consumer Expenditure Survey conducted by the U.S. Bureau of Labor Statistics. The survey collects detailed quarterly data on all consumer spending from logs kept each year by more than 10,000 households with job-based health insurance. During the study period, total out-of-pocket spending in constant dollars remained unchanged. Spending for medical expenses, drugs, and supplies declined by 23 percent, but this decline was offset by rising employee contributions for health insurance premiums. The shift to managed care, whose benefit structure requires less cost sharing, may have played a role in reducing out-of-pocket spending. HEALTH CARE 47 TRENDS The united states experienced an unprecedented slowdown in health care costs in the 1990s, which went largely unreported in the media. Annual increases in job-based health insurance premiums fell from 18 percent in 1989 to 0.8 percent in For the three-year period from 1994 to 1997, inflation for employer-sponsored health insurance lagged behind overall inflation and increases in workers hourly earnings. 2 Less is known about trends in out-of-pocket spending by consumers for services typically covered by employer-sponsored plans and for the employee s share of premiums. Out-of-pocket spending is often seen as an indicator of the performance of the health care system, but the same statistics can support opposing conclusions. Some analysts see out-of-pocket payments as a measure of financial Jon Gabel is vice-president for health systems studies at the Health Research and Educational Trust (HRET) in Washington, D.C. Paul Ginsburg is president of the Center for Studying Health System Change (HSC) in Washington. Jeremy Pickreign is a statistician at HRET. James Reschovsky is a senior researcher at HSC Project HOPE ThePeople-to-PeopleHealthFoundation, Inc.

2 H e a l t h C a r e T r e n d s 48 OUT-OF- POCKET SPENDING barriers to care. 3 Others (mostly economists) view the percentage of costs borne by workers as a protection against moral hazard. 4 An increase in moral hazard will lead to increased use of services and less consumer resistance to inflationary pressures within the health care system. The American public often sees out-of-pocket payments as a measure of how well the health care system is working. Rather than viewing insurance as protection against catastrophic costs, many Americans perceive the function of insurance as prepayment; when families must pay for services out of pocket, Insurance isn t working. In this paper we examine trends in out-of-pocket health spending for households with employer-based health insurance for the years Study Methods Our study uses quarterly interview data from the Consumer Expenditure Survey (CES) conducted by the U.S. Bureau of Labor Statistics (BLS) for the years 1990, 1993, 1995, and The BLS conducts this national survey quarterly to construct the market basket of goods that urban households consume and is the basis for constructing the much-used Consumer Price Index (CPI). The BLS requests households participating in the CES to keep a log, using provided forms, of their consumer spending, including that for medical care. These logs also include information on reimbursements from public and private insurance plans, which are netted out from direct out-of-pocket payments for care. 6 The CES also collects information on socioeconomic characteristics. The CES is a rotating panel, with participants providing information for five consecutive quarters. From the random sample of households surveyed, approximately 10,000 11,000 respondents had employer-based health insurance. According to BLS staff, estimates based on calendar year and collection year (participants may start their interviews in any quarter of the year) do not differ greatly. We first aggregated the data to construct a quarterly series on medical care spending, and then we annualized the figures. We present data in terms of annual spending. The unit of analysis in the study is the individual household, termed consumer unit by the BLS. 7 Our interest is to examine out-of-pocket spending for medical care by households with employer coverage. We therefore selected households in which the reference person was under age sixty-five, at least one member was employed during the time of the interview, and the family was insured under a private plan through employment. 8 The four variables of interest include consumer spending for H E A L T H A F F A I R S ~ V o l u m e 2 0, N u m b e r 2

3 medical services, drugs, medical supplies, and health insurance premiums (for example, the worker s share of the premium even if taken directly out of the paycheck). 9 We refer to the first three categories when combined as direct spending and refer to the total of all four as total consumer spending. Using the CPI for all items, we have adjusted all dollar amounts to constant 1990 dollars. Our analysis concentrates on direct spending, spending for medical services and for drugs individually, and consumer spending for premiums. 10 To validate the CES estimates, we compared consumer spending on premiums with an estimate from the 1997 Robert Wood Johnson Foundation (RWJF) Employer Health Insurance Survey. The two surveys produced very similar estimates. 11 In the subsequent analysis we examine trends in consumer spending and cross-sectional differences by plan type and income level. There are three categories for defining health plan enrollment: (1) the family is enrolled in a health maintenance organization (HMO) or point-of-service (POS) plan; (2) the family has an indemnity or preferred provider organization (PPO) plan; and (3) household members are enrolled in both HMO/POS and indemnity/ppo plans. BLS data do not allow one to distinguish indemnity from PPO enrollment, or HMO from POS enrollment. Hence, our analysis of the role of shifting enrollment to managed care plans is limited by not being able to capture movements from indemnity to PPO plans. From 1990 to 1997 PPO enrollment increased from 13 percent to 34 percent of those with work-based coverage, while indemnity enrollment fell from 62 percent to 18 percent. 12 Because of the small number of respondents reporting both HMO/POS and indemnity/ppo plans and the difficulty of interpreting trends in this category, we do not include a separate line for it in our exhibits. In our analysis, income is a three-level categorical variable: (1) low income (less than $20,000); (2) middle income ($20,000 $49,000); and (3) high income ($50,000 or more). The income categories are defined in 1990 dollars. The analysis provides a series of two-way tables for each variable of interest across time and by health plan enrollment status and income level. T-tests were conducted to compare results in 1990, 1993, and 1995 with the result in We also conducted regression analyses, and the results were consistent with our descriptive findings. These regressions indicated that changes in the demographic composition of the privately insured population did not explain trends in out-of-pocket payments over the study period. HEALTH CARE 49 TRENDS

4 H e a l t h C a r e T r e n d s 50 OUT-OF- POCKET SPENDING Tracking Out-Of-Pocket Spending n Trends in consumer spending, Direct spending by consumers on health care fell between 1990 and Average annual direct spending remained statistically unchanged from 1990 to 1993 and then steadily declined until 1997 (Exhibit 1). Its components did not move together, however. Drug expenses in 1997 were not statistically significantly different from their level in 1990, whereas medical expenses declined 28 percent. In contrast, data from national health care spending accounts indicate that spending under employment-based insurance for drugs in inflation-adjusted dollars more than tripled during the study years. Consumer spending for premiums, on the other hand, rose from 1990 to 1993, and then remained steady through Contributions for premiums in 1997 constituted 63 percent of out-of-pocket spending, up from 51 percent in This likely reflects different policies on the part of employers. In the early 1990s employers increased employees premium contribution share, especially for family coverage. From 1990 to 1995 the percentage of the premium paid by workers for family coverage increased from 28 percent to 34 percent. 13 This likely reflected a response to the large premium increases of the late 1980s and early 1990s and the economic recession of the early 1990s. In addition, employers were becoming more sensitive to the prevalence of two-earner families and wanted to avoid covering a disproportionate share of those families. n Household income level. Over the period direct spending declined most sharply for high-income households (Exhibit 2). In contrast, declines in direct spending for low- and middle-income households were not statistically significant. The difference is driven by the medical expense category, where spending by high-income households declined by 46 percent. At the end of EXHIBIT 1 Average Annual Out-Of-Pocket Health Care Spending By Workers With Employer- Based Coverage, In 1990 Dollars, Selected Years Total direct expense a Medical Drugs Premium expense = = = = $ 512 b 360 b 532 b $ 524 b 380 b 688 $ $ % Total expense c 1,040 1,208 b 1,152 b 1, SOURCE: Bureau of Labor Statistics, Consumer Expenditure Survey, 1990, 1993, 1995, and NOTE: N = 43,642. a Includes all consumer expenses paid directly to providers, including medical supplies, which are not reflected separately. b Significantly different from 1997 at the.05 level. c Includes medical expenses, drug expenses, medical supply expenses, and premium expenses. H E A L T H A F F A I R S ~ V o l u m e 2 0, N u m b e r 2

5 EXHIBIT 2 Average Annual Out-Of-Pocket Health Care Spending By Workers With Employer- Based Coverage, By Annual Income In 1990 Dollars, Selected Years Sample size Total direct expense a Medical Drugs Premium expense Total expense c 9,633 1,891 4,768 2,974 $ 504 b b 356 b b b b 600 b 1, b 1,332 7,510 1,807 4,597 1,106 $ 484 b b 620 b 352 b b 464 b b b 1, ,204 b 1,276 8,422 1,599 4,149 2,674 $ b , ,092 1,404 9,682 1,966 4,721 2,995 $ , ,072 1,284 SOURCE: Bureau of Labor Statistics, Consumer Expenditure Survey, 1990, 1993, 1995, and NOTES: N = 35,247. Sample size and expenses may differ from Exhibit 1 because of missing data on family income. : under $20,000. : $20,000 $49,999. : $50,000 and higher. a Includes all consumer expenses paid directly to providers, including medical supplies, which are not reflected separately. b Significantly different from 1997 at the.05 level. c Includes medical expenses, drug expenses, medical supply expenses, and premium expenses. the period, high-income households still had higher spending for medical services than other households did, but it was only 17 percent greater in 1997 compared with 101 percent greater in The large difference in trends between high-income households and others is surprising. To attempt to explain this, we first examined whether the former were more likely to switch from indemnity/ppo to HMO/POS plans, but this was not the case. 14 A minor factor is that the boundary points between the income categories were adjusted for overall inflation. As real family income grew in the 1990s, more middle-income families moved into the high-income category, which thus became somewhat more like the middleincome category over time (if we use these labels to describe relative rather than absolute incomes). Another reason, but a speculative one that would be difficult to test with data, is that high-income households are more likely to use providers of care who charge higher fees. With the switch to managed care plans that do not allow balance billing for network providers, this may have reduced

6 H e a l t h C a r e T r e n d s out-of-pocket spending for medical services more for high-income households than for those with lower incomes. This same spending pattern by household income does not show up in consumer spending for insurance premiums. In this case, the increase for high-income households is similar to that for middleincome households, and the increase for both is much greater than that for low-income households. For all income categories, spending for premiums increased from 1990 to 1995 and then leveled off. Premium spending for high-income households was much higher than spending for other households throughout the period. 15 n Type of insurance coverage. Direct expense for HMO/POS enrollees was essentially unchanged over the period, while spending for indemnity/ppo enrollees declined from 1993 to 1997 (Exhibit 3). This difference is likely attributable to the fact that the BLS does not distinguish indemnity coverage from PPO coverage and that plan enrollment within the indemnity/ppo category changed markedly during the period. In 1993 PPOs accounted for 29 percent of indemnity/ppo enrollment. 16 By 1997 they were 65 percent. 17 Since PPOs have lower cost sharing than traditional indemnity plans for withinnetwork use, this shift would lead to lower direct spending by EXHIBIT 3 Average Annual Out-Of-Pocket Health Care Spending For Workers With Employer- Based Coverage, In 1990 Dollars, By Type Of Plan, Selected Years Sample size Total direct expense a Medical Drugs Premium expense Total expense c 11,008 1,957 8,429 $ 512 b b b 544 b 516 b 1, ,092 10,600 2,234 7,748 $ 524 b b 380 b b ,208 b 920 1,272 10,099 2,624 6,835 $ ,152 b 976 1,196 11,935 5,826 5,114 $ , ,192 SOURCE: Bureau of Labor Statistics, Consumer Expenditure Survey, 1990, 1993, 1995, and NOTES: N = 43,642. means worker has a health maintenance organization (HMO) or point-of-service (POS) plan only or an HMO and some other plan that is not indemnity or preferred provider organization (PPO). means worker has an indemnity or PPO plan only or one of these plans and some other plan that is not an HMO. a Includes all consumer expenses paid directly to providers, including medical supplies, which are not reflected separately. b Significantly different from 1997 at the.05 level. c Includes medical expenses, drug expenses, medical supply expenses, and premium expenses. H E A L T H A F F A I R S ~ V o l u m e 2 0, N u m b e r 2

7 Trends in out-of-pocket spending for premiums did not differ as much by plan type as direct expenses did. consumers. On the other hand, HMO-type plans include traditional HMOs and POS plans, and the latter allow employees to use out-ofnetwork providers with greater cost sharing. In 1993 POS plans constituted 29 percent of HMO-type enrollment, and that figure grew to 31 percent by Changes in the pattern of risk selection over time may have played a role, but it is likely to have been overwhelmed by the effects of switches from indemnity plans to PPOs. Our analysis of the CES data showed no change in the demographics of the indemnity/ppo and HMO/POS groups over the seven-year period. 19 Trends in out-of-pocket spending for premiums did not differ as much by plan type as direct expenses did. Spending for HMO premiums increased by 20 percent, while spending for indemnity/ppo premiums increased by 33 percent. Since households in HMO/POS plans have lower direct and premium expenses than those in indemnity/ppo plans have, trends in plan enrollment have played a role in the trend in consumer spending. During the study period, indemnity/ppo enrollment declined from 77 percent of responding households to 43 percent, whereas HMO plan enrollment rose from 18 percent to 48 percent. 20 To assess the effect of this shift, we calculated what the level of expenses would have been if plan enrollment patterns had not changed between 1990 and In other words, we calculated how much higher consumer spending would have been if workers had not moved from indemnity/ppo to HMO/POS plans. 22 We found that direct spending would have declined 8 percent over the period, in contrast to the 23 percent decline actually experienced (Exhibit 4). Total consumer spending would have risen 11 percent rather than the 4 percent that actually occurred. These calculations likely understate the role of shifts in plan type because they do not reflect the shift from traditional indemnity to PPO plans. In addition, they do not reflect market-level changes in provider reimbursements or practice patterns that resulted from managed care plan s becoming a more important factor in the marketplace. Discussion And Policy Implications Contrary to the impression left by the media, the period turned out to be an excellent one for consumers concerning their out-of-pocket spending for medical care. 23 With declining spending HEALTH CARE 53 TRENDS

8 H e a l t h C a r e T r e n d s EXHIBIT 4 Percentage Change In Out-Of-Pocket Expenses, , With 1990 And 1997 Plan Market Shares Used As Market Basket In 1997 Total direct expense a Medical Drugs b Premium expense $ $ $ % % Total expense b,c 1,040 1,084 1, SOURCE: Bureau of Labor Statistics, Consumer Expenditure Survey, 1990, 1993, 1995, and NOTES: Estimated 1997 out-of-pocket expense with 1990 health maintenance organization (HMO) market share rates was compared with actual 1997 out-of-pocket expense market shares refers to the distribution of enrollment in HMO, fee-for-service, and preferred provider plans in market shares refers to the distribution of enrollment in a Includes all consumer expenses paid directly to providers, including medical supplies, which are not reflected separately. b Changes in drug expenses and total expenses were statistically insignificant. c Includes medical expenses, drug expenses, medical supply expenses, and premium expenses. 54 OUT-OF- POCKET SPENDING for medical services offsetting increased spending for the employee share of premiums for health insurance, total spending for health services in constant dollars was flat. Of course, statistics on declining out-of-pocket expenses depict central tendency. Like the proverbial person who drowned in a river averaging six inches in depth, not all U.S. workers shared this experience. The typical U.S. household may have experienced a decline in out-of-pocket spending, but that may not have been the experience for families with severely handicapped children or members with serious mental health conditions who need a specialized provider from outside the network. There are no data on the experience of the severely ill. Moreover, costs are not the only issue that is relevant to consumers experiences with health care. Those whose employers increased the proportion of the premium paid by employees in most cases saw a sharper increase in their payments for premiums than the average reflected in this paper. Consumers who obtained lower cost sharing by switching to managed care plans had to adjust to less choice of providers and to administrative restrictions, such as gatekeeping requirements, to benefit from the different benefit structure. Some consciously made this trade-off, while others, whose employers eliminated traditional plans, did not have a choice. To the experts consultants, employee benefit managers, and health economists the decline in the relative importance of out-ofpocket payments would be of little surprise. In 1990 the modal health insurance plan was an indemnity plan with a $250 deductible and 20 percent coinsurance, once the patient exceeded the deductible. 24 By 1997 the modal health plan was an HMO or POS plan with H E A L T H A F F A I R S ~ V o l u m e 2 0, N u m b e r 2

9 High-income households saw the greatest reduction in their out-of-pocket spending for medical care. no deductible and a $10 copayment when using network providers. 25 Benefits in HMO plans are richer than those in indemnity plans, particularly as to preventive and prescription drug benefits. 26 Perhaps because of the sporadic frequency of contact with the health care system, consumers may have little idea of trends in outof-pocket spending and even less notion about the percentage of the total bill paid by the insurer. The lack of transparency in cost increases to consumers may have provided fertile ground for the backlash against managed care that is so pervasive today. If the main attraction of managed care to many is its lower costs, but costs are not a problem for consumers at the moment, then there is little to restrain people from taking actions to minimize what is perceived as the downside of managed care: restrictions on choice of providers and on physicians discretion to prescribe and deliver services that they feel are in the best interest of patients. Private actions have involved favoring managed care products that are less restrictive. Public initiatives entail restrictions on how managed care plans conduct business, such as patients bill of rights legislation. Thus, it is not surprising that high-income Americans have the greatest antipathy toward HMOs. 27 Unlike low-income households, financial barriers to care have not been a major problem for highincome households. This study showed that high-income households saw the greatest reduction in their out-of-pocket spending for medical care. Cost management techniques, however, employed by managed care plans, such as networks of providers, primary care gatekeepers, and formularies, impose barriers to care that the upper middle class has not faced before. In this journal we recently reported that premiums rose 8.3 percent from spring 1999 to spring 2000, the highest increase in seven years. 28 Health care inflation has returned at a time when managed care plans, in response to the managed care backlash, have reduced many of their more unpopular cost management techniques, such as the use of prospective utilization management and primary care gatekeepers, while expanding the size of their provider networks. At some point, these increased premiums will flow through to consumers. Yet employees contributions for single coverage were lower in nominal dollars in 2000 than they were in 1996, and deductibles, coinsurance, and copayments have changed little. So far, the change has been concentrated on prescription drugs, where three-tier cost HEALTH CARE 55 TRENDS

10 H e a l t h C a r e T r e n d s sharing is being used much more widely. However, the vibrant economy and tight labor markets appear to be postponing the day when employers again ask their employees to share more fully the growing costs of health care. The question for employee benefit managers, managed care executives, and policymakers is self-evident. If health plans reduce the use of cost management techniques, and employees continue to face diminishing out-of-pocket expenses when using services, how will costs be controlled? The authors gratefully acknowledge the Robert Wood Johnson Foundation for its financial support; Wolf Webber at the Bureau of Labor Statistics for his assistance; and Jennifer Rabideaux for her administrative support. 56 OUT-OF- POCKET SPENDING NOTES 1. J. Gabel et al., Employer-Sponsored Health Insurance in America, Health Affairs (Fall 1990): ; and Henry J. Kaiser Family Foundation and Health Research and Educational Trust, Employer Health Benefits 2000 (Menlo Park, Calif.: Kaiser Family Foundation, 2000), KPMG Peat Marwick, Health Benefits in 1998 (Montvale, N.J.: KPMG Peat Marwick, 1998), For a recent reexamination of the consequences of financial barriers, see L. Duchon et al., Listening to Workers (New York: Commonwealth Fund, 2000). 4. Moral hazard is the increase in demand that results from the presence of insurance. In the case of health care, because the price of services declines when services are covered, patients buy more services. See M. Pauly, The Economics of Moral Hazard, American Economic Review 58, no. 3 (1968): The last year for which data are available in public-use format is BLS analysts indicated that payments from IRS Section 125 accounts, commonly referred to as cafeteria plans or reimbursement accounts, are not treated consistently in the study. In the majority of cases, these payments are regarded as insurance payments, but some households may not report these payments in their logs. 7. Consumer units, the standard unit of comparison in the CES, are defined as (1) members of a household who are related by blood, marriage, adoption, or other legal arrangement; (2) a person who lives alone or shares a household with others and is financially independent; or (3) two or more persons living together who use their income to make joint expenditures. 8. In a few cases, a person in a household covered by job-based coverage may be covered by a public plan, such as Medicare or Medicaid. We believe that such dual coverage is rare and should not affect trends over the study period. 9. Premium expense includes payments for indemnity, PPO, and HMO coverage as well as Medicare supplement and other premium payments, excluding dental insurance. Medical expense includes spending for physician services, lab and x-ray, other medical professionals, hospital services, nursing home services, medical services in a retirement community, and other medical services such as blood donation, ambulance, or emergency room. Drug expense includes prescription drugs and medicines. Medical supplies expense includes medical and surgical equipment for general use, supportive or convalescent medical equipment, hearing aids, and rental of equipment. Total health ex- H E A L T H A F F A I R S ~ V o l u m e 2 0, N u m b e r 2

11 pense is the aggregate of these items. 10. We have deemphasized a discussion about trends in medical supply expenses because these expenses constitute an exceedingly small portion of total outof-pocket health spending. 11. According to the RWJF survey, workers contributed on average $651 per year toward premiums in 1997 (in 1990 dollars). This compares closely with the corresponding $688 estimate shown in Exhibit 2. The remaining difference between the two estimates is likely explained by the fact that the employer survey estimate is for workers, while the CES estimate is for households, which in some cases may have multiple workers who contribute to health insurance premiums. 12. Data for 1990 are from C.B. Sullivan and T. Rice, The Health Insurance Picture in 1990, Health Affairs (Summer 1991): Data for 1997 are from M.S. Marquis and S. Long, Trends in Managed Care and Managed Competition, , Health Affairs (Nov/Dec 1999): Authors calculation from G. Jensen et al., The New Dominance of Managed Care: Insurance Trends in the 1990s, Health Affairs ( Jan/Feb 1997): ; and Sullivan and Rice, The Health Insurance Picture in In 1990, 16 percent of high-income families were enrolled in an HMO, and that figure increased to 45 percent in The corresponding figures for lowincome families were 18 percent in 1990 and 54 percent in For middleincome families, the corresponding figures were 17 percent in 1990 and 50 percent in One explanation for higher spending by high-income families is that when employees have a choice of plans, high-income families are more likely than low-income families are to choose a high-cost plan. 16. J. Gabel et al., The Health Insurance Picture in 1993: Some Rare Good News, Health Affairs (Spring I 1994): Marquis and Long, Trends in Managed Care and Managed Competition. 18. Ibid.; and Gabel et al., The Health Insurance Picture in This includes changes in the income distribution of HMO and indemnity/ppo plans, age, sex, family size, race, and family composition. 20. The residual enrollment is households that had both indemnity/ppo and HMO/POS coverage. 21. Unfortunately, BLS data do not distinguish PPO from indemnity enrollment, making it impossible to determine the effect of the movement from indemnity to PPO plans on out-of-pocket payments. 22. In technical terms, we calculated costs in 1997 using 1990 weights for plan enrollments. This is a price index using a fixed market basket of goods, the familiar Laspeyres index. 23. J.B. White and R.L. Rundle, Big Companies Fight Health Plan Rates, Wall Street Journal, 19 May 1998, A2; Higher Health Care Inflation, New York Times, 24 April 1998, Business Digest; and M. Freudenheim, Health Care Costs Edging Up and a Bigger Surge Is Feared, New York Times, 21 January Sullivan and Rice, The Health Insurance Picture in KPMG Peat Marwick, Health Benefits in 1997 (Montvale, N.J.: KPMG Peat Marwick, 1997), Ibid. 27. Washington Post/Kaiser Family Foundation/Harvard School of Public Health, Issue Poll Number 2, Health Care, 5 18 July J. Gabel et al., Job-Based Health Insurance in 2000: Premiums Rise Sharply while Coverage Grows, Health Affairs (Sep/Oct 2000): HEALTH CARE 57 TRENDS