| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence and request for reprints to Usha Sambamoorthi, PhD, Institute for Health, Health Care Policy, and Aging Research, Rutgers University, 30 College Avenue, New Brunswick, NJ 08901-1293. E-mail: sambamoo{at}rci.rutgers.edu
 |
Abstract |
|---|
 TOP Abstract MethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
Key Words: Out-of-pocket expenditures • Prescription drugs • Medicare • Elderly population
However, in many cases, the prior analyses have relied simply on descriptive data or used information from earlier periods. To inform policy choices, there is a considerable need for analyses that provide more insight on the burden of prescription drug costs and the way in which it is related to various characteristics of the elderly population. Because of rising drug prices and utilization patterns, changes in insurance coverage, the growth of Medicare HMOs, and other developments, earlier estimates are of limited current relevance. The more recent evidence (Davis et al., 1999; Gibson, Brangan, Gross, & Caplan, 1999; Long, 1994; Poisal et al., 1999) does not go beyond bivariate associations, which can provide a misleading view of the underlying determinants of burden.
The present study addresses this issue by estimating total and out-of-pocket expenditures for prescription drugs used by Medicare beneficiaries aged 65 years or older based on nationally representative data from the 1997 Medicare Current Beneficiary Survey (MCBS) Cost and Use data. We estimate the burden of out-of-pocket prescription drug (OOP-PD) expenditures in relation to income and compare OOP-PD spending and burden across major subgroups of the elderly population as defined by socioeconomic and health characteristics, distinguishing the impact of these factors by using multivariate models.
On the basis of previous work, we posit that initial disadvantages in economic and health circumstances lead to or constrain choices in such a way that by older ages the population is characterized by increasing diversity or inequality (Crystal, Johnson, Harman, Sambamoorthi, & Kumar, 2000; Crystal & Shea, 1990). For example, initial disadvantages in family income lead to differences in educational outcomes that create differences in work patterns, which lead in turn to differences in asset and pension income, and health. We argue that initial differences in health circumstances become magnified over time, so that at older ages those who find themselves with limited economic circumstances often find this compounded with greater health needs, poorer health care coverage, and higher, relatively nondiscretionary, out-of-pocket costs for health care. The two processes interact in many ways. Poor health during the working life leads to interruptions in employment, reducing pension and asset accumulation (both through lost income and health-related expenses). In turn, lower income and poorer job opportunities may lead to lower investment in health-producing activities or even be related to employment situations with greater health hazards.
There are, of course, mitigating circumstances and counteracting effects. For example, both education and health are, in an economic sense, human capital investments. People who make greater investments in education, all else equal, might make greater investments in health, including spending more of their income on health care. Despite these effects, however, we expect to find as a result of these processes of cumulative advantage and disadvantage that high out-of-pocket expenditures as a share of income are related to those personal characteristics that are known to be related to early disadvantage, including race, low levels of educational attainment, and gender. Examining the determinants of spending and burden in this manner can assist policy-makers in understanding the sources of need for further assistance with the costs of prescription drugs and evaluating potential solutions to this problem.
|
Methods |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
In 1997, a stratified random sample of 12,511 beneficiaries was selected to be representative of the entire population of aged and disabled beneficiaries enrolled in Medicare in 1997. The present study includes individuals aged 65 and older, living in the community, who were alive as of December 31, 1997. Data on individuals who enrolled in Medicare partway through the calendar year are not included because full-year survey data are not available on their health care utilization. (Although the files include out-of-pocket cost figures for such individuals, these are imputations from full-year beneficiaries and are not based on actual responses.) For the main analyses, we did not include beneficiaries who died during the calendar year, because full year data on their costs and income are not available. However, we did conduct sensitivity analyses including decedents, which did not significantly change our results. Thus, the final sample size for our main analyses was 8,814 persons.
|
Dependent Variables |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
Control Variables
Socioeconomic Characteristics
These variables included gender, race, age, education, and marital status. Information from the administrative files was used to classify subjects by gender. In multivariate analyses, males were used as the reference category. Race was characterized as African-American, White, and others. In multivariate analyses, White was used as the comparison group. Because the effect of age is likely to be nonlinear, we also categorized age into four groups: 65–69 years, 70–74 years, 75–79 years, and 80 and older. In multivariate analyses the youngest age group was used as the reference category. Marital status was classified as married, widowed, divorced or separated, and never married, with never married used as the comparison group in the multivariate analysis. Education was measured both in terms of years of education for multivariate models and as a categorical variable in bivariate analyses.
Insurance
Insurance coverage measures included presence of any insurance supplemental to Medicare and presence of prescription drug coverage. Supplemental insurance plans included Medicaid (full and Qualified Medicare Beneficiaries/Specified Low-Income Medicare), self-purchased (Medigap), employer-provided coverage, private HMO, Medicare HMO, and other public plans. Individuals could have more than one source of supplemental coverage. Medicare HMOs are CMS-regulated Medicare + Choice plans offered directly by sponsors to Medicare participants, in which Medicare is the primary payer. Some Medicare beneficiaries are also covered by private HMOs, which may be employment-based policies for current employees or retirees or may involve coverage through an employed spouse.
For each individual, monthly prescription drug coverage variables were created. We then constructed a variable indicating whether the individual had prescription drug coverage for the full year, for a portion of the year, or no coverage at all. Individuals covered for all 12 months in 1997 were assigned to the full-year coverage category. Individuals with 1–11 months fell into the part-year coverage category; we also included in this category individuals who had no coverage according to our monthly coverage variables, but who had positive third-party annual pharmacy expenditures. Individuals with no pharmacy coverage and no positive third-party annual pharmacy expenditures were considered as having no coverage.
Health Variables
Respondents were asked to characterize their health in comparison with other people their age as excellent, very good, good, fair, or poor. Such self-reports have been found to be reliable predictors of morbidity and mortality (Greiner, Snowdon, & Greiner, 1996; Idler & Kasl, 1995, 1991; Idler, Kasl, & Lemke, 1990). A second health measure was based on the number of medical conditions reported by a respondent. We also used scales representing the number of limitations reported in activities of daily living (ADLs) and instrumental ADLs (IADLs). Our measure of ADL was a scale corresponding to the total number of ADL impairments reported, ranging from 0 (no ADL limitations) to 6 (complete limitation), from among any difficulty in bathing or showering, dressing, eating, getting in or out of chairs, walking, and using the toilet. Our measure of IADL was a scale corresponding to the total number of IADL impairments reported, ranging from 0 (no IADL limitations) to 6 (complete limitation), from among any difficulty in using a phone, doing light housework or heavy housework, making meals, shopping, and managing money. Respondents were asked whether they had ever been told by a doctor that they had medical problems related to heart, diabetes, cancer, stroke, arthritis, hypertension, emphysema, osteoporosis, Alzheimer's disease, or a mental or psychiatric disorder. A scale representing the number of medical problems reported by a respondent was created, with a value of 0 representing a respondent with no medical problems and a value of 10 (the highest possible value) representing a respondent who had all of the 10 above-mentioned medical problems.
Place of Residence
As of 1997, 14 states had created state pharmacy assistance programs (SPAPs) to provide access to prescription drugs, usually for persons aged 65 and older (Gross & Bee, 1999). Most of these programs provide assistance to low-income older people in paying for prescriptions, although the generosity of the programs varies widely (Fox, Trail, & Crystal, 2002). Because these can have an important impact on out-of-pocket spending burden, particularly for persons with low income, we created a variable to identify persons who live in a state where one of these programs exists. We also included metropolitan versus nonmetro area of residence as one of the control variables.
Level of Poverty
Our analysis also includes a measure of income to needs, with income measured as a percentage of the poverty level. The poverty levels used in the analysis are taken from the 1997 U.S. Department of Health and Human Services Poverty Guidelines (http://aspe.hhs.gov/poverty/97poverty.htm). In 1997, the poverty level was $7,890 for a single person and an additional $2,720 for each additional person. For married couples, we used the two-person level of $10,610 because the income variable in the MCBS is measured for the respondent, or the respondent and spouse if married for the calendar year. We operationalized this variable in the analyses as low income (≤200% of poverty) versus higher income.
Data Analysis
In the bivariate analysis, we tested for subgroup differences in the average amount of prescription drug expenditures with t statistics. We also present the differences in median total and OOP-PD expenditures. For each of the subgroups, 95% confidence intervals around these estimates were constructed, and nonoverlapping confidence intervals indicate statistically significant differences. Bivariate group differences in OOP-PD burden were tested for significance with the chi-square test statistic.
Ordinary least square (OLS) regressions were used to isolate subgroup differences in prescription drug expenditures and level of OOP-PD expenditures. For these analyses, expenditures were transformed to a logarithmic scale to reduce skewness. Effect estimates for continuous independent variables on the log of monthly expenditures can be interpreted as percentage change for each unit of change in the independent variable. The effect of dummy variables in terms of percentage of expenditures can be estimated by exponentiating the regression coefficients of dummy variables and subtracting 1 (i.e., percent change = eß-1) (Halvorsen & Palmquist, 1980). To further reduce the influence of extreme values, to test the robustness of the analyses, and to make interpretation of the results more straightforward, we also used median regressions to jointly estimate the effect of predictors on total and OOP-PD expenditures. Because median regressions do not account for design effects of the MCBS data, we report results from both the OLS and median regression specifications.
Logistic regressions were estimated to predict the probability of spending greater than 10% of income on prescription drugs and to isolate the effects on this outcome of characteristics such as gender, age, race, education, geographical location, insurance coverage, and health status. We examined OOP-PD burden by using nested logistic regression models. In Stage 1, we entered only the socioeconomic, geographic, and health characteristics. In Stage 2, in addition to variables entered in Stage 1, level of poverty was entered. Parameter estimates from logistic regressions were transformed into odds ratios (with 95% confidence intervals) associated with each independent variable, which represent the relative risk ratios for a 1-unit change in the variable in question. Odds ratios exceeding 1 indicate an increased likelihood of utilization relative to the comparison group, whereas odds ratios less than 1 indicate a decreased likelihood of use. In the statistical analysis except for median regressions, we estimated standard errors by using linearization methods that account for intracluster correlation in the multilevel sampling design utilized by MCBS, using the procedures of the SAS-Callable SUDAAN statistical software system (Shah, Barnwell, & Bieler, 1997).
|
Results |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
Prescription Drug Use
Table 1 provides bivariate comparisons of the proportion of subgroups with any prescription drug use. Eighty-eight percent of older Medicare beneficiaries in our sample used prescription drugs at some point during the calendar year 1997. We found significant differences in the use of and costs of prescription drugs among subpopulations. Overall, a significantly higher proportion of women, older individuals, married persons, individuals with prescription coverage, QMB/SLMB or Medicaid coverage, and individuals in poor health status had prescription drug use. Odds ratios and 95% confidence intervals from logistic regression on any prescription drug use when the effects of demographic, socioeconomic characteristics, insurance coverage, and health status of respondents were controlled for are also presented in Table 1. As in the bivariate analysis, we observed significant differences by gender, age, education, prescription drug coverage, health status, and level of poverty. These results indicate that prescription drugs were more likely to be used by women, older individuals, persons with more education, pharmacy coverage, a greater number of chronic medical conditions, and persons with higher income.
|
|
Annual Out-of-Pocket Expenditures on Prescription Drugs
Table 2 also provides level of OOP-PD expenditures (columns 7 and 8). On the average, respondents spent $347 on prescription drugs. Nearly half of annual expenditures (48%) on prescription drugs were borne by survey respondents. In dollar terms, racial minorities spent less out of pocket on prescription drugs than Whites. Those with prescription drug coverage spent less out of pocket than persons without any prescription drug coverage. Those with Medicaid, as expected, spent less out of pocket on prescription drugs. Similarly, having certain types of supplemental policy coverage was associated with lower expenditures on prescription drugs. For example, OOP-PD costs for those in Medicare HMOs averaged $214, whereas expenditures for respondents with self-purchased supplemental insurance policies averaged $503. The median expenditures were $120 and $299, respectively. Those with employer-sponsored coverage, who typically pay only part of the cost of the policy, spent $277 (median = $154) on prescription drugs. As with bivariate comparisons, functional impairment, self-reported health status, and number of medical conditions contributed significantly to higher out-of-pocket costs, even after other covariates were controlled for (Table 2, column 9). However, even with controls for other characteristics, self-purchase of supplemental policies was associated with more out-of-pocket costs than others who did not have self-purchased policies. Those living below 200% of federal poverty guidelines spent less out of pocket on prescription drugs than those with higher incomes. We found similar patterns when the data were analyzed with median regressions. There were also a few differences in the predictors in the median regression specification compared with the OLS specification. Whereas age was a significant predictor in the OLS specification, it was not in the median regression. In the median regression, partial drug coverage decreased annual out-of-pocket expenditures by $58 compared with the elderly people without drug coverage. Elderly people with Medicare HMO coverage spent $33 less out of pocket than those without Medicare HMO, suggesting that Medicare HMO participation reduces expenditures at the 50th percentile of the distribution rather than at the average.
OOP-PD Expenditure Burden
Because cost burden is of key importance in terms of ability to pay for health care costs not covered by third-party payers, we also examined OOP-PD cost burden. On average, 3% of income was spent on prescription drugs. To avoid the disproportionate influence of a few cases with very high income and expenditure ratio, we classified OOP-PD burden into four categories: (a) 0%, (b) >0%–5%, (c) >5%–10%, and (d) >10%. Table 3 shows variations in OOP-PD cost burden categories by beneficiary characteristics. Nearly 8% of the sample respondents spent more than 10% of their income on prescription drugs. This represented 2,339,000 older individuals. As shown in the table, the burden was distributed disproportionately. Expenditures as a proportion of income were greater for women, African-Americans, the widowed, respondents with no high school education, residents living in nonmetro areas, and individuals living in non-SPAP states; they were lower among individuals with pharmacy coverage, Medicaid, or supplemental coverage including Medicare HMOs. However, individuals who had purchased supplemental policies such as Medigap had a higher burden than those who did not. As expected, health status was strongly associated with high OOP-PD burden, defined as spending greater than 10% of per capita income. For example, 16% of respondents with six or more chronic conditions had high burden. Similarly, burden is greater for lower-income individuals, with over 13% of the beneficiaries with incomes below 200% of poverty spending more than 10% of their income on prescription drugs compared with only 2% of beneficiaries with incomes above 200% of the poverty line.
|
|
Among the health status variables, self-reported health status and number of chronic medical conditions were associated with high burden. Excellent or very good health status was associated with a reduction in the odds ratio for high burden of 38% compared with those reporting poor health. Each chronic condition increased the odds ratio of having high burden by nearly 45%. Contrary to the bivariate comparisons, ADL and IADL impairments were not strong predictors of high OOP-PD burden. Inclusion of level of poverty (Model 2) did not affect the results, except that when level of poverty was in the model, education was no longer a significant predictor of high burden.
Sensitivity Analyses
OOP-PD Expenditures Burden: Alternative Specification
Because the MCBS asks only about a couple's income, in the original specification we divided income by two for those who are married, which may introduce measurement error into the analysis. To test the robustness of our findings and minimize the measurement error, we then computed the burden of OOP-PD expenditures as a percent of total household income and included household size as one of the predictors. Results from these analyses (data not shown) indicated that under this specification only 5% of the sample respondents spent more than 10% of their income on prescription drugs. However, the subgroup differences in the burden and predictors of the burden remained essentially unchanged under this alternative specification.
Inclusion of Decedents
Previous research using the 1996 MCBS data shows that pharmaceutical costs in the last year of life were somewhat higher ($653) than in nonterminal years ($593; Hoover, Crystal, Kumar, & Sambamoorthi, 2002). Because inclusion of costs associated with the last months of life may upwardly bias the estimates, we conducted sensitivity analyses by including respondents who died during the calendar year. These results indicated (data not shown) that the average and median costs were similar to the results without decedents. Average expenditures for prescription drugs were $710 and out-of-pocket expenditures were $343. Median expenditures on prescription drugs were $451 and median out-of-pocket expenditures were $166. Including decedents in the sample did not substantially alter the findings in terms of subgroup differences and predictors of expenditures and burden.
|
Discussion |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
The findings on the characteristics of the heavily burdened are relevant to the approach taken to creating a Medicare drug benefit, as well as to the need for such a benefit. Major competing approaches in Congress have been divided between those that would rely on federal subsidies for competing, individually marketed, private pharmacy-only insurance policies with varying benefit structures, and those that would add a standard set of pharmaceutical benefits to existing Medicare coverage (typically with income-related premiums, cost sharing, or both). Negotiating the former approach might be difficult for Medicare beneficiaries who currently experience high pharmacy costs, who tend to have multiple chronic health conditions and functional impairment, poor overall health, limited formal education, and very advanced age (Crystal, 2001). On one hand, to the extent that insurers were permitted to vary their benefit structures, they would have considerable incentive to seek to shape them so as to attract fewer of these heavily burdened individuals, who tend to have predictably high pharmaceutical utilization. On the other hand, unless federal subsidies covered a higher share of the costs than anticipated in Congressional private-insurance-based proposals to date—with a correspondingly higher price tag—insurers would also be exposed to considerable adverse selection, as high utilizers would be disproportionately likely to enroll (Crystal, 2001).
This tendency might be exacerbated by the skewed distribution of pharmaceutical utilization, which often involves long-term treatment of chronic conditions and is therefore predictable. Some Congressional proposals have sought to avoid this problem by offering pharmacy coverage only as a one-time election at the time of initial Medicare enrollment. As the heavily burdened typically have incomes below 200% of the poverty line, however, many of them might lack the discretionary income to make this election in advance of need.
All things considered, the concentration of high pharmacy cost burden in a definable subgroup of older people suggests the need to pool their costs of coverage into a broad insurance pool that includes individuals with lower anticipated expenditures, if an insurance-based approach is taken. At the same time, it suggests the desirability of public-sector strategies that explicitly target those who experience high burden, either because of low income or unusually high pharmacy expenditures. A number of states have developed programs aimed directly at one or the other of these subgroups (Crystal, Fox, Silberberg, Trail, Reinhard, & Cantor, 2002; Fox et al., 2002). As our findings indicate, these programs are successful in reducing burden in that the presence of a SPAP in a beneficiary's state is associated with less incidence of high burden. The fact that a significant incidence of high burden remains probably reflects the wide variation in generosity among these programs (Fox et al., 2002).
Our multivariate findings indicate that levels of prescription drug expenditures, out-of-pocket costs, and burden are associated with a number of factors, including gender, insurance coverage, health status, and income. Elderly women were more likely to use prescription drugs and appeared to maintain greater burden than elderly men, even after insurance, health, and, even more important, income were controlled for. Consistent with earlier research, our findings suggest that paying for prescription drugs is becoming a significant and growing problem for older women (Blustein 2000; Gibson et al., 1999; Lassila et al., 1996; Neuman et al., 1999; OWL, 2000; Poisal & Chulis, 2000; Rogowski et al., 1997; Stuart et al., 1991). Although it may not be feasible to subsidize prescription drug expenditures for women as a subgroup, prescription drug policies targeted to those who are most heavily burdened will also be of particular benefit to women, who tend to have low incomes and older age on average. In addition, women may have higher expenditures as a result of a higher rate of morbidity (Lassila et al., 1996; OWL, 2000). For example, a Medicare prescription drug benefit that would pay all of the drug costs for individuals below 100% of the poverty line and 80% of the costs for individuals living up to 150% of the poverty level, as suggested in Moon and Storeygard (2002), would go a long way in protecting women against high burden.
We find that Blacks spent less than Whites on prescription drugs, largely as a result of differences in the level of use, rather than the probability of any use (there is no significant difference in probability of any use by race). The biviariate analysis of burden, however, suggests that the burden of prescription drug costs is higher among persons. After other covariates are controlled for, Blacks are not found to have significantly lower burden than Whites, suggesting that differences in insurance, health, and other factors may explain the bivariate relationship. These findings do suggest that policies that addressed these differences would appear to remove racial differences in the burden of prescription drug costs.
An interesting finding in our study was that self-purchased supplemental policies were not associated with reduced probability of high OOP-PD costs. Beneficiaries with self-purchased insurance policies paid nearly 70% of their drug expenditures out of pocket. After all other characteristics were controlled for, they were twice as likely as those without self-purchased insurance policies to spend greater than 10% of their income on prescription drugs, perhaps suggesting the inadequacy of coverage offered by these plans. It must be noted that our analysis does not include premium costs, which are substantial, and therefore may underestimate the burden. These results may also reflect patterns of self-selection into self-purchased coverage.
Enrollment in Medicare HMOs, and prescription drug coverage, significantly decreased the financial burden on elderly households associated with prescription drug use. Although these findings are encouraging, newer data suggest that dropout of HMOs from Medicare+Choice and the trend to less generous benefits of HMOs may leave elderly people vulnerable for high financial burden. For example, in 2001, only 70% of Medicare+Choice HMOs offered prescription drug coverage and more than 300 HMOs had dropped out of the Medicare+Choice program (Briesacher et al., 2002; Gold, 2001). These results suggest the limited ability of the Medicare+Choice program to address the need for protection from high pharmaceutical costs by Medicare beneficiaries.
Our study underscores the particularly high cost burden that pharmaceutical expenditures create for elderly people with poor health status and chronic diseases, even after insurance coverage and income are controlled for (Gibson et al., 1999; Lillard et al., 1999; Steinberg et al., 2000); in order to provide adequate protection for this group, there is a need for consistent, defined-benefit pharmacy coverage with low consumer cost sharing and without coverage caps—characteristics that are absent from much of the available private coverage as well as some of the Congressional proposals for prescription drug coverage. In considering alternative strategies to address prescription drug needs for elderly people, our findings suggest the need for particular focus on the needs of the subgroup who are currently heavily burdened, and that disproportionately represents the oldest-old, those with low incomes, and those with multiple chronic health conditions and high functional impairment.
|
Limitations |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
Ideally, to examine these relationships, an analysis would focus on a long-term panel study of individuals. However, data available for such analysis are extremely limited. Our effort to examine with cross-sectional data the burden that develops as a result of a lifetime process can only be regarded as suggestive of the relationships that exist. Nonetheless, the approach is valuable in identifying more fruitful areas of research. Further research can focus on identifying the paths that lead to the differences in burden of prescription drug costs that we identify here.
|
Footnotes |
|---|
1 Institute for Health, Health Care Policy, and Aging Research, Rutgers University, New Brunswick, NJ.
2 Department of Health Policy and Administration, Penn State University, University Park, PA.
Decision Editor: Laurence G. Branch, PhD
Received for publication April 5, 2002. Accepted for publication September 11, 2002.
|
References |
|---|
|
TOPAbstractMethodsDependent VariablesResultsDiscussionLimitationsReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  N. E. Schoenberg, H. Kim, W. Edwards, and S. T. Fleming Burden of Common Multiple-Morbidity Constellations on Out-of-Pocket Medical Expenditures Among Older Adults Gerontologist, August 1, 2007; 47(4): 423 - 437. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. B. Soumerai, M. Pierre-Jacques, F. Zhang, D. Ross-Degnan, A. S. Adams, J. Gurwitz, G. Adler, and D. G. Safran Cost-related medication nonadherence among elderly and disabled medicare beneficiaries: a national survey 1 year before the medicare drug benefit. Arch Intern Med, September 25, 2006; 166(17): 1829 - 1835. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 U. D. Patel and M. M. Davis Falling into the Doughnut Hole: Drug Spending among Beneficiaries with End-Stage Renal Disease under Medicare Part D Plans J. Am. Soc. Nephrol., September 1, 2006; 17(9): 2546 - 2553. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Wei, A. Akincigil, S. Crystal, and U. Sambamoorthi Gender Differences in Out-of-Pocket Prescription Drug Expenditures Among the Elderly Research on Aging, July 1, 2006; 28(4): 427 - 453. [Abstract] [PDF]
|
|
|
|
 |
|
 I. M. Kronish, A. D. Federman, R. S. Morrison, and J. Boal Medication utilization in an urban homebound population. J. Gerontol. A Biol. Sci. Med. Sci., April 1, 2006; 61(4): 411 - 415. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Klein, C. Turvey, and R. Wallace Elders Who Delay Medication Because of Cost: Health Insurance, Demographic, Health, and Financial Correlates Gerontologist, December 1, 2004; 44(6): 779 - 787. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
