This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (13) Citing Articles via Google Scholar Google Scholar Articles by Thomas, C. P. Articles by Wallack, S. S. Search for Related Content PubMed PubMed Citation Articles by Thomas, C. P. Articles by Wallack, S. S. Related Collections Pharmaceuticals Health Spending

DataWatch

Growth In Prescription Drug Spending Among Insured Elders

Abstract

 
We examine growth in prescription drug use and spending in a well-insured elderly population in 1997 and 2000. We describe the high-cost segment of this population, identifying how it differs from the rest of the elderly regarding use and types of medications, and how stable this group is over time. Drug spending by the insured elderly rose more than 18 percent annually between 1997 and 2000. High-cost elders use more brand-name drugs, treat more conditions, and use more medications per condition. Once an insured elder becomes a high-cost user of prescription drugs, that person is likely to remain so. Our study suggests that a growing population of elderly is using many medications and may require considerable drug management.

In an insured population, prescription drug costs can be quite skewed, with a small proportion of the covered population incurring a large proportion of the costs. In 1996, for instance, only 4 percent of Medicare beneficiaries with prescription drug coverage incurred more than $2,500 in total drug costs, but they accounted for 28 percent of drug spending by all covered persons.¹

This high-cost segment of the elderly population is important for clinical reasons and for policy purposes as a prescription drug program for Medicare beneficiaries is being considered. First, as more elders use more medications, the issue of polypharmacy and inappropriate prescription drug use arises.² As people use more prescription drugs, some medications may be inappropriate, and combinations of drugs may have dangerous interactions. Additionally, for expenditure, benefit design, and drug management purposes, it is important to know how growth in the high-cost population might affect future Medicare spending for catastrophic drug coverage.

Our study uses prescription claims data from a well-insured population with varying benefit plans in a large pharmacy benefit management (PBM) company to examine several cost and utilization issues regarding the high-cost elderly. First, who among the insured elderly incurs the highest drug spending, and how do they differ from the rest of the elderly population? Second, how has this group changed in terms of size, use, and spending?

We examine the spending distributions of all persons age sixtyfive and older in 1997 and 2000 and focus on the highest spenders, or the top 12 percent of prescription drug users in 2000.We also select the cohort who moved from lower to higher expenditures and examine their changes in drug choice and intensity of use to determine the reasons for their spending increases.

Methods

Top Methods Results Discussion NOTES

 
Data source. Prescription drug claims data (the Brandeis Analytic File) were provided by AdvancePCS, a major PBM company that currently manages the prescription drug benefits of approximately eighty million persons in the United States.³ At the time of the study, PCS Health Systems (the portion of enrollees for which data were provided) covered approximately five million elderly, or 14 percent of the U.S. elderly population.

The data file used for the study was constructed in two steps. First, a set of 4,437 AdvancePCS carrier groups was selected, and then claims of enrollees in these carrier groups were gathered together to form the analytic file. In consideration of the linking of claims to enrollment information, data completeness, and data quality, the sample frame from which carrier groups were selected was restricted based on certain requirements: Carrier groups had to assign a unique identifier to each enrolled member (instead of one unique to the family), and they required only copayments at time of purchase rather than making the enrollee pay fully and then reimbursing a calculated share. In addition, the frame was restricted to carrier groups that were clients of PCS for the time period 1997–2000.

Claims sample. Enrollees in the claims sample had to be continuously enrolled with PCS for the entire fiscal year in question (1997 or 2000) so that their full year of use could be tabulated. The number of persons age sixty-five and older in the study was 99,655 in 1997 and 106,517 in 2000. Demographic information includes date of birth and sex. The age distribution in our sample is younger than the overall U.S. population over age sixty-five or the elderly sample for the Medicare Current Beneficiary Survey (MCBS).⁴ Also, our sample is 55 percent female, compared with 58 percent in the MCBS.

Even though we looked at carriers that were clients of PCS throughout 1997–2000, we did not require study subjects to be enrolled in both 1997 and 2000, except where specified. This means that people who died or lost benefits after 1997 were included in the base year (1997) calculations. Conversely, people who obtained coverage or reached age sixty-five after 1997 were included in the 2000 data. While persons were selected separately for each study year, there is considerable overlap in the sample, making it useful for comparisons across years; 74 percent of individuals in 1997 are also included in the 2000 sample. Essentially, the sample differs across years as covered persons age into the sixty-five-and-older group and as some retirees lose coverage or die. For the specific analyses looking at utilization changes in the same persons over time, we created a subsample of drug claims of the 72,115 enrollees over age sixty-five who were included in PCS enrollment files for both full years, 1997 and 2000.

Drug claims criteria. Claims in this study include all outpatient medications traditionally covered by prescription drug insurance. ⁵ "Total expenditures" are defined as the transaction costs paid for each claim and do not include dispensing fees. Expenditures are reported in actual 1997 and 2000 dollars, unadjusted for inflation, to best approximate actual costs and to keep our results consistent with those of other studies. However, where appropriate, general inflation rates are noted in the text as comparisons. Total cost is broken down into the portions paid by AdvancePCS and by the insured person (out-of-pocket cost per claim). As in any claims data set, use and spending are recorded only for drugs purchased through AdvancePCS plans. Therefore, out-of-pocket spending is limited to copayments, coinsurance, and deductibles, and no out-of-plan purchases or premium costs are recorded. In AdvancePCS plans with maximum benefits, enrollees receive a discount on medications after they have exceeded their maximum allowable costs. These discounted purchases are also included in the claims data, and enrollees’ expenditures beyond the cap are thus included as out-of-pocket costs.

Rebate information. Rebate information is not included in the transaction price of a medication in this study, since it depends on overall sales volume and share, rather than on individual claims. Also, it varies across classes of drugs and specific drugs within those classes, and it is not included in the claims database. According to AdvancePCS, the overall level of rebates in the aggregate did not change greatly during the study period. However, for several classes of medications, using price information without rebates may overestimate the true cost of brand-name prescriptions.

Results

Top Methods Results Discussion NOTES

 
Use and spending. Mean total drug spending for the elderly rose from $827 in 1997 to $1,378 in 2000, an 18.5 percent annual increase, compared with 2.3 percent average annual general inflation during this period (Exhibit 1).⁶ The annual increase for the younger population was 15.6 percent. The major component explaining this difference was in number of prescriptions per person, which grew at 7.1 percent for the elderly compared with 4.7 percent for the nonelderly. As expected, the elderly started with utilization rates that were almost three times higher than those for younger enrollees in 1997. Even with this higher baseline, utilization increased at higher rates among the elderly than among the young.

Distribution of expenditures. The proportion of the elderly population that used any prescription drugs rose only slightly, from 81.9 percent in 1997 to 83.6 percent in 2000, but the spending distribution changed far more (Exhibit 2). The 1997 median expenditure of $488 per person rose to $925 in 2000. The top 10 percent of prescription drug spenders incurred $2,322 or more in 1997 and $3,579 or more in 2000.

Higher-cost prescription drug users do not differ from lower-cost users by average age or sex, but fewer high-cost enrollees are age eighty-five or older. However, high-cost elders in our samples not only used more medications than others did, but the medications they used were more expensive and less often generic (Exhibit 4). High and very high cost elders filled an average of more than fifty prescriptions per year, and the cost per prescription for this group was twice that of the low-cost group ($61–$78 versus $33). The mean number of days’ supply per prescription for all of the cost groups was similar, indicating that the higher cost per medication for high-cost elders was not due to the longer duration of prescriptions. In 2000 the mean cost per prescription of elders in the very high cost group was 28 percent higher than for those in the $3,000–$4,000 range, and their total drug cost was 30 percent higher per day.

While elders overall used a proportion of generic drugs similar to that of the younger population, the proportion of generics decreased from 44 percent to 30 percent with higher cost groupings. Consistent with this, the percentage of total spending devoted to generics decreased by about half, from 16.6 percent to 8.4 percent.

Use across therapeutic categories. In further examining the high-cost elderly, we looked at how high-cost users differed from low-cost users within the therapeutic categories used most often. More than half of elders even in the low-cost category used medications for cardiovascular disease, with a relatively low mean cost per prescription (Exhibit 5). No other disease was as well represented in the low-cost category. Fewer than 10 percent of this group were on diabetes medications, and few took gastrointestinal medications. Although nearly half of the low-cost group had claims for an antibiotic, because of the short-term nature of these prescriptions and the lower average cost per prescription, this did not move people into the higher cost categories. Thus, high-cost elders had on average several more chronic conditions than did their lower-cost peers. For example, among persons in the high and very high cost categories, nearly 90 percent used cardiovascular medications, but in addition, more than half of this group took a gastrointestinal medication, half took a lipid-lowering medication, and nearly half were on antidepressants or anti-arthritis medication. The relatively high mean cost per prescription for the anti-ulcer and lipid-lowering medications poses a particularly large cost burden. This confirms a clinical picture of high-cost elders as medically complex persons requiring treatment for many conditions, both physical and mental. This group clearly presents a challenge for clinical and costmanagement.⁹

Changes in expenditures over time. To determine the extent to which elders move into higher cost categories over time, we looked at changes in spending groups between 1997 and 2000 for the population that was enrolled for both full years.¹⁰ Exhibit 6 indicates a stability in drug spending that has been seen in earlier research.¹¹ More than half of the overall population remained in the same cost category over the years studied. The one exception to this was the high cost category, in which there was greater movement upward into the very high cost group, perhaps because of the smaller spending range in this category. For all cost levels, transitions tended to be toward the higher cost categories, not the lower. More than half of those who were high cost in 1997 were very high cost by 2000. Of the 41 percent in the middle in 1997 who had shifted by 2000, 29 percent moved up, and only 12 percent moved down. Finally, the 6 percent of persons who were very high cost in 1997 but showed no claims in 2000 most likely are institutionalized or deceased and had not yet been removed from the enrollment files.

For beneficiaries who moved from lower to higher cost groups, per capita expenditures for several medications increased dramatically between 1997 and 2000 (Exhibit 7). The top ten medications accounted for 40 percent of the overall three-year spending growth for this group. This is both the result of more people taking these drugs and each person having more prescriptions. Several of the medications with the greatest growth in per capita spending are too new to show up in 1997 claims (Celebrex, Vioxx, Plavix, and Enbrel). While Exhibit 7 shows only one dimension of what occurs over time to place people in higher cost categories, it illustrates how the addition of new medications drives their spending growth.

Top Methods Results Discussion NOTES

Comparisons to other studies. The overall distributions we show for the elderly’s prescription drug spending and out-of-pocket costs are similar to those presented in a study of Merck-Medco pharmaceutical claims in 1997–1998.¹² We found that 83.6 percent of beneficiaries purchased a prescription drug in 2000, compared with 82 percent in 1998 in the Merck-Medco claims. Also, mean expenditures for enrollees age sixty-five and older went from $827 in 1997 to $1,361 in 2000 in our study, compared with $1,099 in 1998 in the Merck-Medco claims. The 1998 number reported in the Merck-Medco study is somewhat consistent with ours, in spite of the fact that we used actual costs per claim and they used average wholesale price (AWP) plus a dispensing fee.

The Merck-Medco mapping of disease prevalence by drug claims showed similar findings to ours in terms of the proportion of persons taking medications in particular disease categories (Exhibit 5). However, while they examined spending by enrollees with several disease combinations, little detail is presented on how persons who incur the highest costs differ from those with lower costs, and what moves people from lower to higher costs within diseases.

In the 1996 MCBS the distribution of population spending was similar to ours at the high-cost end.¹³ High spenders in the 1996 MCBS were defined as the top 20 percent of beneficiaries who had a claim, or those who spent more than $1,066 in 1996.¹⁴ Although we take a somewhat finer cut on high-cost elders (12 percent rather than 20 percent) and only include well-insured elders, in our study age was also not a predictor of high spending. Our results showing that higher expenditures are associated with more disease categories (rather than just more expensive medications) are also consistent with the MCBS in that high spenders were more likely to be in poor health and have three or more chronic conditions.

Our findings differ to some extent with other elements of the MCBS. For instance, we show a smaller proportion of our elderly group in 1997 using any medications than is reported for persons with insurance coverage for 1996 in the MCBS elderly and disabled populations (81 percent versus 89 percent).¹⁵ The MCBS reports mean total annual expenditures for beneficiaries of all ages with insurance for the full year in 1996 to be $828, with out-of-pocket spending of $219.¹⁶ We found much lower out-of-pocket spending of $118 per enrollee in 1997. Our lower finding is perhaps not surprising, since most plans in our database were employer-based, generally known to have lower rates of out-of-pocket to total spending than other private Medicare supplemental insurers have.¹⁷ We do not have detailed data on MCBS plan type for those enrolled the entire year in employer coverage and over age sixty-five. However, 1998 MCBS spending by beneficiaries over age sixty-five with drug coverage at least a portion of the year was $812 to $998 per elder, and 20–25 prescriptions per person, similar to our numbers.¹⁸

Movement into higher cost groups. We can see patterns in movement from low or middle to higher cost groups. Medication therapy can become very expensive when more than one disease is treated. Taking one lipid-lowering agent (no generic statins are available yet) immediately places a person in a higher cost group. What moves persons into high cost groups, therefore, is not only the addition of disease groups, but also the way in which prescription drugs are used to treat diseases. As we have shown, a major difference in the medication profile of persons who moved from under $3,000 per year to over $3,000 per year is spending increases for expensive brand-name medications (Exhibit 7). It is clear, also, that once people move into the high cost group, they are likely to stay there for at least several years, confirming other studies’ findings.

Drug and disease management. Using claims data alone, we cannot make inferences about the clinical appropriateness of the drug use patterns seen here, or about the relationship to health outcomes or use of other medical services. Also, because this group includes a wide range of benefit program designs, we cannot determine the degree to which the drug profile of this group is managed. All drugs provided through PCS Health Systems, and now AdvancePCS, are subject to basic screens at the point of sale to identify inappropriate drugs and combinations, and further analyses of these data will be directed at how these work. Certain clients are more heavily managed than others, through various utilization management strategies. Highcost users are made up primarily of persons taking multiple medications for several diseases. Our findings suggest that disease management directed at a single disease may be difficult for this group, even in tightly managed systems. Management of complex sets of conditions will be increasingly important as this group grows in size.

A Medicare drug benefit. Finally, our study stopped short of analyzing the impact of various plan design strategies. To truly predict how this high-cost population would affect the cost of a Medicare prescription drug benefit, it is important to learn the impact of cost sharing and other controls on them. Adjusted for plan design and utilization management components, these data could be used to model the cost and impact of various Medicare prescription drug plans under consideration.

The rapid growth in the population with high and very high drug spending, and the likelihood that their costs will remain high, has critical importance in the administration of any Medicare drug benefit plan. This is particularly so if plans are administered by at-risk competing PBMs or other entities. Experience has shown that a large number of elders exhaust managed care drug benefit caps and disenroll disproportionately as they do so.¹⁹ The selection issue and the risk-sharing arrangements for high-cost elders could in large part determine the viability of any program. As discussions regarding a drug benefit for seniors continue, this portion of the elderly must be given increased attention.

Editor's Notes

 
Cindy Parks Thomas is a senior research associate at the Brandeis University Schneider Institute for Health Policy. Grant Ritter is senior scientist at the institute. Stanley Wallack is a professor at the Heller Graduate School at Brandeis University and the institute’s founder and director.

This research was funded by the U.S. Department of Health and Human Services Office of the Assistant Secretary for Planning and Evaluation (ASPE). The authors thank Dominic Hodgkin and Elizabeth Dichter, Schneider Insititute; Andrew Garling, Christine Syrotiuk, and Duane McKinley, AdvancePCS; and members of the RxHealthValue Coalition, for their help in providing data and for helpful comments. They also thank Joan Sokolovsky and Jack Hoadley at ASPE for guidance in developing this research effort and comments on earlier drafts; and several anonymous reviewers.

NOTES

Top Methods Results Discussion NOTES

 

1. Office of the Assistant Secretary for Planning and Evaluation,U.S. Department of Health and Human Services, Report to the President: Prescription Drug Spending, Utilization, Coverage, and Prices (Washington: ASPE,May 2000).
2. S.M. Willcox, D.U. Himmelstein, and S. Woolhandler, "Inappropriate Drug Prescribing for the Community-Dwelling Elderly," Journal of the American Medical Association 272, no. 4 (1994): 292–296[Abstract/Free Full Text]; and U.S. General Accounting Office, Prescription Drugs and the Elderly: Many Still Receive Potentially Harmful Drugs Despite Recent Improvements, Pub. no. GAO/HEHS-95-152 (Washington: GAO, 1996).
3. The data were provided by PCS Health Systems, prior to its purchase by Advance Paradigm Health Systems, forming Advance PCS. At the time the claims were incurred, PCS Health Systems covered fifty million lives.
4. The proportion of elderly in our database ages 65–74 ranged between 62 percent and 64 percent compared with the U.S. elderly in 2000 (and the 1997 MCBS) of 52 percent. There also are fewer elderly age eighty-five and older: 67 percent of our elderly sample, compared with 12 percent nationally.
5. We excluded any claims that were for prescription-related supplies (for example, special shampoos, batteries); medications in strengths available over the counter (for example, Motrin, 200 mg, or ibuprofen, 200 mg); or any injectable medications routinely administered only in the hospital.
6. Bureau of Labor Statistics, "Consumer Price Indexes," Inflation Calculator, <www.bls.gov/cpihome.htm> (28 May 2001).
7. There are twenty-two therapeutic categories as measured by this index.
8. It is likely that the very high cost users also include a large number of persons who are using high-cost medications to treat one or two expensive conditions such as cancer or neurological disease.
9. One limitation in interpreting the data is the large "miscellaneous" category. This category likely includes anticancer medications and several expensive new medications. The number of people who incur very high costs from a limited number of very expensive medications is likely to be small but clearly warrants further analysis.
10. This portion of the analysis excludes persons who disappeared from enrollment files by 2000. Such persons have either died, switched from AdvancePCS to another PBM, or lost prescription drug coverage.
11. N.E. Coulson and B. Stuart, "Persistence in the Use of Pharmaceuticals by the Elderly," Journal of Health Economics 11, no. 3 (1992): 315–328.[Medline]
12. E.P. Steinberg et al., "Beyond Survey Data: A Claims-Based Analysis of Drug Use and Spending by the Elderly," Health Affairs (Mar/Apr 2000): 198–211.
13. ASPE, Report to the President, Table 2-14.
14. Ibid., Tables 2-15–2-17.
15. Ibid., Table 2-1.
16. Ibid., Table 2-3.
17. Ibid., Table 2-23; and J.A. Poisal and L. Murray, "Growing Differences between Medicare Beneficiaries With and Without Drug Coverage," Health Affairs (Mar/Apr 2001): 74–85.
18. Poisal and Murray, "Growing Differences."
19. T.S. Rector, "Exhaustion of Drug Benefits and Disenrollment of Medicare Beneficiaries from Managed Care Organizations," Journal of the American Medical Association 283, no. 16 (2000): 2163–2167[Abstract/Free Full Text]; and E.R. Cox, B.R.Motherall, and K. Fairman, "Exhaustion of Prescription Drug Benefits and Medicare Beneficiaries Disenrollment from Managed Care," Journal of the American Medical Association 284, no. 2 (2000): 2596–2597.[Free Full Text]

                      What's this?

This article has been cited by other articles:

				S. S. Wallack, D. B. Weinberg, and C. P. Thomas Health Plans' Strategies To Control Prescription Drug Spending Health Aff., November 1, 2004; 23(6): 141 - 148. [Abstract] [Full Text] [PDF]

				J. F. Moeller, G. E. Miller, and J. S. Banthin Looking Inside The Nation's Medicine Cabinet: Trends In Outpatient Drug Spending By Medicare Beneficiaries, 1997 And 2001 Health Aff., September 1, 2004; 23(5): 217 - 225. [Abstract] [Full Text] [PDF]

				M. W. Smith and G. J. Joseph Pharmacy Data in the VA Health Care System Med Care Res Rev, September 1, 2003; 60(3_suppl): 92S - 123S. [Abstract] [PDF]

				S. H. Altman and C. Parks-Thomas Controlling Spending for Prescription Drugs N. Engl. J. Med., March 14, 2002; 346(11): 855 - 856. [Full Text] [PDF]