|Home | About | Journals | Submit | Contact Us | Français|
Prescription drug spending and pharmacy benefit design have changed substantially over the last decade, yet little is known about the financial impact these changes have had on consumers. We examined ten years of nationally representative data from the Medical Expenditure Panel Survey and describe trends in two measures of financial burden for prescription drugs: out-of-pocket drug costs as a function of family income and the proportion of all out-of-pocket health care costs accounted for by drugs. We found that although the percentage of people with high financial burden for prescription drugs increased from 1999 to 2003, it decreased from 2003 to 2007, with a slight increase in 2008. The decline is evidence of the success of strategies to lower drug costs for consumers, including the increased use of generic drugs. However, the financial burden is still high among some groups, notably those with public insurance and those with low incomes. For example, one in four nonelderly people devote more than half of their total out-of-pocket health care spending to prescription drugs. These trends suggest that the affordability of prescription drugs under the future insurance exchanges will need to be monitored, as will efforts by states to increase prescription drug copayments under Medicaid or otherwise restrict drug use to reduce public spending.
Spending for prescription drugs grew annually at double digit rates in the late 1990s and high annual growth rates continued into the early part of the last decade, outpacing overall health care spending growth.(1) This trend was attributable to a combination of factors including the introduction of new drugs, increased volume of drug use, and rising prices.(2)
Growth rates for drug spending slowed in the second half of the last decade, coinciding with substantial changes in pharmacy benefit design, the introduction of fewer new drugs, loss of patent protection for a number of blockbuster drugs, and increased use of generic drugs.(3-5) The introduction of tiered formularies and the use of utilization management tools (such as prior authorization) had a major effect on drug use and spending.(4,6-12)
Although these factors contributed to a dramatic slowing of growth rates in prescription drug spending,(3,5) little is known about their impact on consumers’ direct out-of-pocket costs. Furthermore, any changes in consumer spending would probably not be distributed equally across the population. The burden of overall health care costs falls heavily on low-income people without generous insurance, as well as on people with common chronic conditions such as diabetes, hypertension, and mental illness.(13-15) Less is known, however, about prescription drug costs and whether the patterns may be similar.
For our study, we used nationally representative data to examine trends in the out-of-pocket financial burden of paying for prescription drugs from 1999 to 2008. We examined how this financial burden differed by insurance and income status. Furthermore, we explored trends in the use of generic drugs during this period, because of the important role they play in lowering drug costs.
We used data from the 1999-2008 Medical Expenditure Panel Surveys (MEPS), a nationally representative household survey of the US civilian, noninstitutionalized population conducted by the Agency for Healthcare Research and Quality. Details of the survey design, and its limitations, have been previously published.(16-18)
The survey provides data on demographic characteristics, insurance coverage, and health care use and spending for all people in the sampled households. Household reports on use of services are supplemented with information on third-party payments and billing codes from medical providers’ billing records. In the case of outpatient medications, permission is obtained to contact all pharmacies.(18) For every reported purchase of a prescription drug, pharmacies are asked to provide information by source as well as National Drug Code. MEPS uses a comprehensive approach to address missing data and nonresponse bias in order to produce valid national estimates of health spending.(16)
We included all people surveyed in MEPS who were younger than age sixty-five. We followed the prior literature in analyzing out-of-pocket burdens separately for the elderly and nonelderly, because of the different policy implications, financing sources, and health care needs of the two groups. We chose to focus this analysis on the nonelderly.
For each person in the sample, we defined a family-level measure of financial burden for prescriptions drugs. The measure was calculated using total out-of-pocket drug costs for all family members in a given year divided by family income. We used family-level measures of financial burden assuming that family members shared the financial resources for health care among themselves.(13) Out-of-pocket drug costs for all family members were summed across the year, and divided by the total pre-tax family income.
We were unable to use after-tax family income because of privacy limitations in the public-use MEPS database.(19,20) When calculating family income, we followed the model of previous authors and imposed a $100 floor for family income to deal with underreporting or negative incomes, which could distort the calculated burden ratio.(13)
Premiums were excluded from the calculation of total drug spending because it is not possible to accurately calculate the proportion of these premiums going to drugs as opposed to other health spending. All spending and family income numbers were inflated to 2008 dollars using the consumer price index for urban areas. We chose to adjust for inflation using a general price index, as previous analyses have done, because we were measuring drug spending as a share of income and examining this ratio over time.
In addition to calculating financial burden for prescriptions as described above, we calculated the proportion of people living in families in which out-of-pocket drug costs accounted for more than 50 percent of total out-of-pocket costs for all health care services, including prescriptions.
We estimated the number of nonelderly people living in families in which out-of-pocket costs for prescription drugs exceeded 10 percent of the family income, which we defined as high financial burden for drugs. We also examined the number of people living in families that spent more than 5 percent of the family income on prescription drugs. We avoided focusing on mean financial burden because of the potential bias from misreported income and outliers.(13,21,22) The literature often uses 10 percent and 20 percent thresholds for total out-of-pocket burden(13,20,22) and thus, 5 percent and 10 percent thresholds seemed reasonable when limiting the analysis to just one type of medical service.
In addition to presenting population estimates, we also examined financial burden by family income, insurance status, and health status (categorized based on the presence of specific chronic conditions and based on overall health spending). We used five income categories corresponding to the federal poverty level: poor is less than 100 percent of the federal poverty level; near poor is 100 to less than 150 percent; low income is 150 to less than 200 percent; middle income is 200 to less than 400 percent; and high income is 400 percent or more.
We assigned people to one of four mutually exclusive insurance categories: private group (employment-related), private nongroup (individual market), public, and uninsured. People were assigned to insurance categories based on the type of coverage they had for the most months during the year. People who lacked coverage for the entire year were classified as uninsured. We chose these categories to be consistent with prior MEPS analyses examining financial burden among the nonelderly, rather than focusing solely on prescription insurance coverage, which is measured less effectively in MEPS.
For health status, we examined financial burden for people living with one of three high-cost conditions: diabetes, hypertension, and mental illness. These conditions are self-reported by respondents in MEPS; professional coders recode verbatim responses into three-digit codes using the International Classification of Diseases, Ninth Revision. Mental illness is defined broadly to include mood disorders such as depression, cognitive disorders, and adjustment disorders, among others, and is not limited to serious and persistent mental illnesses such as schizophrenia.
Our second measure of health status was overall health spending. We divided the cohort into quartiles based on total health care spending and compared the trends in financial burden between those in the lowest and highest quartiles of spending.
We present similar analyses of the proportion of individuals living in families in which out-of-pocket drug costs accounted for more than 50 percent of total out-of-pocket health care costs, examining the overall trends and trends stratified by family income, insurance status, and health status. We also calculated the percentage of all prescriptions in each year that were for generic drugs (matching the National Drug Code to brand and generic indicators in the Multum Lexicon database) and the mean number of prescriptions per person overall, because of the potential effect of increasing generic drug use and increasing prescription use on financial burden.
For all analyses we used the appropriate weighting and survey design variables to obtain weighted results,(16,18) which represent population estimates for the noninstitutionalized US population younger than age sixty-five.
There are several limitations to our study given the study design and data sources. First, we focused on financial burden using pre-tax family income and were unable to adjust for differential tax burden among people. However, the use of after-tax income would clearly increase the proportion of people living in families with high out-of-pocket prescription drug burden, and thus, our analysis probably underestimates this proportion. Although MEPS income data are carefully edited, survey-reported incomes may be lower than what is found in tax records. Yet, we do not expect there has been any change over time in this potential bias.(17)
Second, there were modest improvements in the editing of the MEPS prescription drug cost variables starting in 2007 that affect relative prices for generic and brand name drugs.(23) These editing changes would not have affected the trends in the purchases of generic drugs, and we do not believe they would have a large effect on the trends seen in financial burden, although we are unable to test this.
Third, we did not account for health insurance premiums paid, because for people under age sixty-five those premiums are typically paid for both drug and nondrug insurance. Fourth, although we excluded those people over age sixty-five and thus most people with Medicare, a small percentage (less than 1 percent) of the cohort each year had Medicare-only coverage. In sensitivity analyses excluding this small group, the overall trends we report here were unchanged, although the proportion with high financial burden was slightly smaller among the publically insured when this group was excluded than when it was included. We elected to keep these people with Medicare-only coverage in our sample for our results to be fully generalizable to the under sixty-five age group.
Finally, our analysis is purposefully descriptive and does not attempt to identify the independent effects of each of the variables we discuss, nor does it attempt to make any inferences about the direct causal effects of these variables on financial burden.
There were 20,771 people younger than age sixty-five included in MEPS in 1999, a statistical sample representing 241,735,177 nonelderly people in the United States. The mean annual family income was $65,382, and the mean family out-of-pocket cost for prescription drugs was $294 (all amounts are in 2008 dollars). Sample sizes for each of the years, along with mean family income and mean out-of-pocket drug costs can be seen in Exhibit 1.
In 1999, there were 7.1 million people (2.9 percent) living in families spending more than 10 percent of their family income on prescription drugs (high drug-cost burden). That same year, there were 64.5 million people (26.7 percent) who lived in families in which out-of-pocket drug costs accounted for more than half of all out-of-pocket health care costs (Exhibit 1).
By 2003, out-of-pocket drug costs were higher, such that 10.8 million people (4.3 percent) had high drug-cost burden and 85.2 million (33.6 percent) had out-of-pocket drug costs accounting for more than half of all out-of-pocket health care costs. By 2008, these numbers had decreased to 8.3 million people (3.1 percent) living in families with high drug-cost burden, and 67.1 million (25.4 percent) with out-of-pocket drug costs accounting for more than half of all out-of-pocket health care costs.
The likelihood of high drug-cost burden differed substantially among the insurance and income groups. In 2008, the percentage of people living in families with high drug-cost burden was 7.5 percent among those with public insurance (down from 11.1 percent in 1999), 4.5 percent among those with private nongroup (nonemployer-related) insurance, and only 1.2 percent among those with private group (employer-related) insurance (Exhibit 2). In each insurance group, the financial burden of prescription drugs declined between 2004 and 2007 with a slight increase in 2008; this increase in 2008 was larger for the uninsured.
Similar patterns were seen for people in different income groups (Appendix Exhibit 1).24 The financial burden for drugs was highest for the poor. In 1999, 16.5 percent of poor people had high drug-cost burdens, and theses burdens peaked in 2003 when more than one out of five poor people (21.2 percent) lived in families with high drug-cost burdens. In 2008 this proportion had decreased to 13.6 percent.
The proportion of people living in families in which out-of-pocket drug costs accounted for more than half of all out-of-pocket health care costs followed similar trends as financial burden, although without the slight increase in 2008 (Exhibit 3). In 2008, 43.6 percent of people with public insurance had out-of-pocket drug costs that accounted for more than half of all out-of-pocket health care costs, compared to 21.5 percent in private group insurance and 15.3 percent in private nongroup insurance. Low-income people were more likely to live in families in which out-of-pocket drug costs accounted for more than half of all health care costs (Appendix Exhibit 2).24
People with chronic conditions were much more likely than the overall population to have a high financial burden of drug costs. (Exhibit 4). In 1999, more than 13 percent of people with diabetes had high drug-cost burden. That proportion had increased to 16.4 percent in 2003, and decreased back to 13.1 percent in 2008. A majority of people with diabetes lived in families in which drug costs accounted for more than half of all out-of-pocket health care spending. Similar trends were seen for patients with hypertension and mental disorders (Exhibit 4).
People in the highest quartile of total health care spending were more likely to live in families with high financial burdens for drugs than those in the lowest quartile (5.4 percent compared to 1.1 percent in 2008), despite their higher family income (Appendix Exhibit 3).24 People in the highest quartile were also more likely than those in the lowest quartile to live in families in which drug costs accounted for more than half of all out-of-pocket costs (27.8 percent compared to 14.9 percent in 2008). Overall trends were similar, with a decrease in financial burden since 2003.
Between 1999 and 2003, 36 percent to 39 percent of prescriptions were filled using generic drugs (Exhibit 5). The share increased slightly in 2004 (40 percent), and in 2008, 56 percent of all prescriptions were filled with generic drugs. In 2008, people with private group insurance had the lowest rates of generic drug use (54 percent) and people with public insurance and the uninsured had the highest rates (61 percent and 59 percent, respectively). There was a steady increase in the mean number of prescriptions per person over time for nonelderly people in MEPS, from 9.5 in 1999 to 11.9 in 2003 to 13.1 in 2008 (data not shown).
This study provides the first nationally representative estimates of changes in the financial burden of out-of-pocket prescription drug costs for people younger than age sixty-five, from 1999 to 2008. The percentage with high financial burden for prescription drugs peaked in 2003, then decreased between 2003 and 2007, with a slight increase in 2008. This overall decline occurred during a period when drug use was on the rise.
This decrease in financial burden is important evidence of the success of strategies already in place to lower drug costs for consumers, including the shift to increased use of generic drugs that is clearly evident in our analysis. However, the financial burden is still high among some groups, notably those with public insurance and those with low incomes, which has important implications for future expansions in insurance coverage.
Although several prior studies have examined how changes in cost sharing have affected the financial burden for patients in paying for health care,(13,15,19,20,22,25,26) the studies have not focused on the financial burden attributable to prescription drug costs. Across the board, these prior analyses point to a rising burden of overall out-of-pocket health costs over the past decade, with millions more people each year added to the roles of the underinsured.
A large nationally representative survey in 2007 reported that an estimated twenty-five million insured people under age sixty-five were underinsured (spent more than 10 percent of their income on out-of-pocket medical care), representing a 60 percent increase from 2003.(22) Our findings point to a different trend for prescription drugs costs, one in which out-of-pocket burden for drugs has decreased since 2003. Importantly, had it not been for this decrease, it is likely that the increase in overall health care cost burden since 2003 would have been even greater.
The overall decrease in financial burden for prescription drugs should not, however, obscure the important role that prescription drug spending plays in consumers’ overall health budgets. Although overall prescription drug spending accounts for a small portion (10 percent) of total health care spending from all payment sources,(1) out-of-pocket costs for drugs take up a much larger portion of each household's health budget. We find that one in four nonelderly people in the United States devote more than half of their total out-of-pocket health care costs to prescription drugs, with a substantially larger percentage among low-income people with public insurance and chronic conditions such as diabetes.
We found that the financial burden for prescription drugs was higher for people in the nongroup insurance market compared with those with private group insurance. Although this difference may be because people with group insurance have higher incomes than people with nongroup insurance, we did find slightly higher out-of-pocket costs per capita for drugs in the nongroup market than in the group insurance market. It is important to note that others have also found higher out-of-pocket health care spending (overall) for those with nongroup insurance as compared to group insurance.(19)
These differences between private nongroup and group insurance are worth exploring further, considering the provisions in the Affordable Care Act to expand coverage to twenty-four million people through new health insurance exchanges that build on the nongroup market.(27) There are reasons to suspect, however, that future nongroup policies will have better, more generous drug coverage than existing policies.
For example, policies sold on the insurance exchange will be mandated (by law) to include prescription drug coverage, which is not the case currently. Although drug coverage will be part of the minimum essential benefits, it is unknown how plans will set limits on covered services, or set limits at all. Recent work from the Institute of Medicine defining the minimum essential benefit provided no additional guidance on this important point, other than assuring that plans balance competitiveness and affordability.(28) The affordability of prescription drugs under the insurance exchanges will need to be monitored, especially for people with low incomes and high health care needs.
The trends in financial burden differed slightly for people insured by public programs, who experienced a relatively steady rate of financial burden between 1999 and 2006, and a pronounced decline after 2006. Part of the decline in burden may reflect the introduction of Medicare Part D in 2006, although only a small minority of people under age sixty-five had Medicare-only coverage in each of the years (less than 1 percent). The trends likely reflect changes made to the Medicaid programs across the country in 2004-05, with the addition of preferred drug lists and increased use of prior authorization.(4,7,8)
The trends we outline in this article have other potential implications for prescription drug policy. As most states and the federal government enter a period of prolonged budgetary constraint, state Medicaid agencies may increase prescription drug copayments or place other restrictions on drug use to reduce public outlays.
Although these changes may be inevitable, their impact on the poor and those with public insurance should not be ignored. Almost one in seven people with family income less than the federal poverty level lived in families with high drug-cost burden in 2008, and more than one in three spent more out-of-pocket for drugs than for any other aspect of health care. Monitoring the affordability of prescriptions in insurance programs, especially public insurance programs, may be just as important as monitoring the medical loss ratio or quality metrics such as cancer screening and patient satisfaction.
In conclusion, although the overall financial burden of prescription drug costs increased from 1999 to 2003, it decreased over the next five years until 2008, during a time when use of prescription drugs increased. The lower financial burden of drug costs is probably, in part, a result of the increased use of generic drugs. Upward pressure on out-of-pocket drug spending in the near future is possible, with the market entry of new expensive drugs such as biologics, continued increases in the prevalence of chronic conditions, and state and federal budgetary constraints. Whether the public or private sectors will find new innovations in the future to lower drug costs for consumers remains to be seen.
This paper was presented at the AcademyHealth Annual Research Meeting in Seattle, WA, June 13, 2011. This study was supported in part by a VA HSR&D Career Development Award (CDA 09-207) to Dr. Gellad, as well as by the RAND University of Pittsburgh Health Institute (RUPHI) and the Clinical and Translational Science Institute (CTSI) Translating Research into Practice (TRIP) Program. The CTSI is made possible by Grant Number UL1 RR024153 to the University of Pittsburgh from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. Dr. Donohue was supported by the following grants (AHRQ 1R01HS017695; 1R34 MH082682; P30 MH090333). Dr. Zhang was supported by National Institute of Mental Health (NIMH) RC1 MH088510 and AHRQ R01 HS018657. This study was conducted while Dr. Banthin was employed by the Agency for Healthcare Research and Quality. This work represents the opinions of the authors alone and does not necessarily represent the views of the Department of Veterans Affairs, the Agency for Healthcare Research and Quality, the Congressional Budget Office, or the United States Government.