• The economic impact of DUR interventions on the pharmacological management of asthma in children

      Raut, Monika Kumar; Stuart, Bruce C. (2001)
      The purpose of the dissertation was to assess the economic impact of a DUR intervention among children enrolled in the Pennsylvania Medicaid program who had overused SA beta2-agonist inhalation drugs. This research expanded on previous work, the base study, titled 'Pennsylvania Medicaid Retrospective Drug Utilization Review Program Improvement of the Pharmacological Management of Asthma in Children Enrolled in the Pennsylvania Medicaid Fee-for-Service Program' conducted by the Center on Drugs and Public Policy, University of Maryland and Pennsylvania Medical Society for the Commonwealth of Pennsylvania, Department of Public Welfare. This base study was descriptive in design and did not include a control group. The dissertation used scientifically controlled research methodologies to assess the base study results and evaluate whether a true DUR intervention effect existed. It used various multivariate techniques designed to control for non-intervention related influences. Four pre/post comparison series designs to address uncontrolled confounders such as regression to the mean, history, and maturation were used. These included (a) a self-controlled historical comparison group design, (b) a historical equivalent comparison group design, (c) a concurrent non-equivalent comparison group design, and (d) a regression discontinuity design. The self-controlled historical comparison group design was a single-group pre/post design that monitored trends in resource utilization among the intervention group in a year previous to the DUR claims review period. The historical equivalent comparison group design was a two-group design that compared the DUR intervention group to a comparison group identified by the same criteria as the intervention group in a year previous to the DUR claims review period. The concurrent non-equivalent comparison group design and the regression discontinuity design were two-group designs that utilized two different methods of analyses to compare the intervention group to a control group in the concurrent period. The end-points of interest were cost of asthma medications, asthma-related hospitalizations, and asthma-related emergency department visits. The study used Pennsylvania Medicaid administrative and prescription claims data for the period July 1997 to March 2000 residing at the CDPP. The robustness of each study design's findings to the base study results was analytically assessed. None of the four scientifically controlled study designs showed a significant positive DUR effect. The base study results were confounded by regression to the mean, history, and/or maturation.
    • Impact of Beneficiary Characteristics, Drug Plan Formulary Policies, and Environmental Factors on Medication Adherence among Low Income Beneficiaries Covered by Medicare Part D

      Shen, Xian; Stuart, Bruce C. (2016)
      Research suggests that low income individuals often fail to take medications as prescribed. Various individual characteristics have been studied in relation to adherence outcomes, however, little is known regarding how drug plan formulary policies and external environmental factors may affect individuals' medication adherence. This dissertation evaluated the independent effects of beneficiary characteristics, drug plan formulary policies, and external environmental factors on medication adherence for oral hypoglycemic agents (OHAs), statins, and renin angiotensin system (RAS) antagonists in a cohort of low-income subsidy (LIS) recipients enrolled in randomly assigned benchmark Part D plans. The data source included a random 5% sample of 2012 Medicare administrative claims and a customized dataset capturing beneficiaries' plan assignment history. Three hosts of beneficiary characteristics, including demographics, comorbidity burden, and health services utilization, were analyzed. The formulary policies of interest included non-coverage, prior authorization, and step therapy, while the environmental factors of interest were socioeconomic environment, availability of healthcare resources, health culture, evidence-based medicine practice, and quality of primary care. Results indicated that beneficiary characteristics, Part D plan, and external environmental factors all could significantly influence LIS recipients' medication adherence. Older age, male gender, use of multiple chronic medications were associated with higher medication adherence, whereas black race, Hispanic ethnicity, high comorbidity burden, and frequent hospitalizations and ER visits were inversely related to the adherence outcomes. Placing formulary restrictions on brand-name drugs could shift utilization toward generics and lower cost per prescription fill but had minimal impact on medication adherence among LIS recipients. Geographic variation in adherence rates was observed consistently across all three drug classes of interest. Those living in areas with low socioeconomic environment and poor quality of primary care were less likely to achieve acceptable levels of medication adherence than their counterparts. In conclusion, low income beneficiaries' medication adherence is influenced by multiple levels of factors. Policies aimed at improving low income population's adherence for chronic medications may consider plan- and environment-oriented programs in addition to interventions targeting at individuals' behaviors.
    • Osteoporosis Medication Use, Adherence, and Outcomes in Elderly Enrolled in Medicare Part D

      Loh, Feng-Hua; Stuart, Bruce C. (2016)
      Background: Osteoporosis affects an enormous number of people of both sexes, and osteoporosis-related fractures are costly to treat. Yet osteoporosis is poorly managed and managed differently by sex and residential setting. Therefore, this study aimed to assess the difference in medication use, adherence, and outcomes between men and women and among women, between long-term care (LTC) facility and community residents. Methods: Using the 2006-2008 Chronic Condition Data Warehouse 5% national random sample of Medicare beneficiaries, this retrospective study identified elderly 70 years and older with osteoporosis enrolled in Medicare Part A, B, and D stand-alone prescription drug plans from January 1, 2006 through December 31, 2008, or death. Use of bisphosphonates, calcitonin, parathyroid hormone and selective estrogen receptor modulator was tracked over the 3-year period. Treatment effectiveness was measured as hazard of fracture after treatment initiation. Modified Poisson regression was used for analyzing the effect of sex and residential status on osteoporosis medication use. Cox proportional hazard model was used for analyzing the effect of medication use and adherence on fracture risk. Results: The samples included 96,408 females, 8,465 males and 90,956 females, and 2,083 males and 10,262 females enrolled in Medicare Part D for aims 1, 2, 3, respectively. Utilization was lower among LTC residents (RR 0.89, 95% CI [0.87, 0.91]). Bisphosphonates were prescribed less often to LTC residents (RR 0.79, 95% CI [0.75, 0.83]) compared to among community residents. Prevalence of osteoporosis medication use in men was substantially lower than that in women (25.2% vs. 44.3% in 2006). Good adherence decreased the hazard of fracture in both sexes (HR 0.86, 95% CI [0.75, 0.99]). No difference in either treatment or adherence effect on fracture between men and women existed. Conclusion: Prevalence of osteoporosis medication use is low in elderly women enrolled in Part D whether community dwelling or LTC residents. Elderly men are undertreated for osteoporosis compared to elderly women. There is strong confounding by indication in the effect of osteoporosis medications on the risk of fracture; however, good adherence reduces the risk of fracture. There is no evidence for heterogeneity in treatment response among men and women.
    • Prevalence, predictors, and consequences of propoxyphene use in the aged

      Kamal-Bahl, Sachin J.; Stuart, Bruce C. (2003)
      Propoxyphene, an opiate-like analgesic, is considered inappropriate among patients 65 years or older. Despite this, propoxyphene is commonly prescribed among elderly patients. The objectives of this study were (1) to estimate the prevalence and persistence of propoxyphene use, (2) to determine patient and physician characteristics and the main clinical indications associated with propoxyphene use, and (3) to establish whether propoxyphene use is associated with an increased risk of fractures in elderly persons. The Medicare Current Beneficiary Survey (MCBS), a nationally representative dataset of the Medicare population, and the Area Resource File, which provided county-level physician characteristics, were used to estimate the prevalence and predictors of propoxyphene use. Since drug events are not dated in the MCBS, the MarketScan, an employer-based claims dataset was used to examine persistence of propoxyphene use, indications for its use and its association with fractures. Descriptive and multivariate (logistic and time-dependent Cox regression) analyses were conducted. The results indicated a high prevalence of propoxyphene use in the community-dwelling elderly Medicare population, with over 2 million beneficiaries receiving it in each year from 1993(6.8%) to 1999(6.5%). More than half the propoxyphene users had only one prescription in the year and the mean prescription was written for 11 days supply. Beneficiaries most likely to fill propoxyphene included: females, Medicaid beneficiaries, residents of the South and Midwest, and those with history of osteoporosis and hip fractures. Beneficiaries residing in counties where the proportion of medical specialists was higher were less likely to receive propoxyphene. The most common indications for propoxyphene were cancer, back disorder, osteoarthritis, and joint pain. Cox regression results indicated a 2.5 times increase in the hazard of fractures for persons who used propoxyphene in the last 14 days as compared to those who did not (HR 2.5; 95%CI 2.4--2.7). A positive dose-response relationship was also observed between propoxyphene use and risk of fractures. These results should help design and target interventions to reduce propoxyphene use in elders. Establishing an increased risk of fractures should strengthen the case for these interventions.
    • Supplemental Insurance and Health Care Utilization and Spending among Elderly Cancer Survivors

      Ke, Xuehua; Davidoff, Amy J., 1957-; Stuart, Bruce C. (2014)
      Little is known about the impact of complex interactions of financial resources, health status, and attitudes towards health care seeking upon access to and utilization of health care beyond the acute cancer diagnosis and treatment phases. The study examined the relationship between supplemental medical and prescription insurance and use and spending on selected health care services for elderly Medicare cancer survivors compared to non-cancer controls. Data were pooled from the 1997-2007 Medicare Current Beneficiary Survey (MCBS) including linked historical and concurrent claims. A cancer survivorship year was randomly selected from one of the MCBS years that met the following criteria: (1) being post cancer-diagnosis year 2 or later, (2) no active treatment, and (3) no enrollment in hospice, or death, and with complete Cost and Use survey data available. The observation year for non-cancer controls was randomly selected from one of the MCBS years meeting the selection criteria (3). Spending was adjusted to 2007 dollars. Univariate and bivariate analyses were used to describe sample baseline characteristics. Naive generalized linear models and two-stage residual inclusion methods were conducted to examine whether there is a differential effect of supplemental insurance on health care utilization and spending for cancer survivors vs. non-cancer controls. The study included 3,958 cancer survivors and 7,056 non-cancer controls. Breast (21%), prostate (21%), and colorectal (16%) were the three most common cancer sites among cancer survivors. The study showed that cancer survivors had generally higher socioeconomic status, were more likely to have supplemental insurance, had more comorbidities, and had stronger preferences for medical care than non-cancer controls. Cancer survivors tended to use more selected preventive services and had higher total healthcare and prescription drugs spending, while use less recommended medications for diabetes than non-cancer controls. In the main multivariate analyses, supplemental insurance had an effect on use and spending on selected health care services, but had no differential effect for cancer survivors vs. non-cancer controls. The results help fill a gap in understanding the relationship of supplemental insurance to health care utilization and spending for elderly Medicare cancer survivors.
    • Topics in Medicare Prescription Drug Enrollment in the Low-Income Subsidy Population

      Hendrick, Franklin B.; Stuart, Bruce C. (2016)
      In Medicare Part D, random assignment and potential yearly reassignment to premium-free stand-alone prescription drug plans (PDPs) is the default plan enrollment option for low-income subsidy (LIS) recipients. Randomization can cause medication access issues, but the impacts on medication use remain unclear because those who choose plans have not been separately examined. Medicare Advantage prescription drug plans (MAPDs) and tailored dual eligible Special Needs Plans (D-SNPs) have financial incentives to improve the medication adherence of Medicare-Medicaid dual eligibles compared to Medicare fee-for-service (FFS), but this relationship has not been assessed. The study used 2006-2009 Medicare administrative data and a customized dataset that differentiated plan election types. In aim 1, 29,784 LIS recipients assigned in 2007 were followed for three years and only 26% became choosers, with half selecting MA plans. PDP choosers appeared sicker and had higher Part D costs than non-choosers. In contrast, MA enrollees had fewer chronic conditions and lower costs than non-choosers. Choosers' plans covered more drugs than non-choosers' plans. In aim 2, medication use and costs were compared among 28,610 statin users who either accepted or opted out of reassignment. Cross-sectional and difference-in-differences (DID) regression models examined changes in statin use and costs. Compared to reassignees, the 7.6% who opted out were less likely to discontinue (-0.8%) and switch statins (-7.0%) and exhibited relative increases over time in brand name use (6.8%) and 30-day fill costs to Medicare ($2.82) and plans ($3.23). In aim 3, drug adherence with statins and angiotensin-converting enzyme inhibitors/angiotensin receptor blockers (ACEIs/ARBs) was compared for samples of dual eligibles enrolled in FFS, MAPDs, and D-SNPs over 24 months, including subsamples who switched from FFS to managed care. Analyses included cross-sectional and DID regressions. Drug adherence was slightly higher among managed care dual eligibles when compared to FFS dual eligibles, but a percentage of those switching from FFS experienced disruptions in use. In conclusion, few randomized LIS recipients choose their own plans. Opting out of reassignment has minimal impact on statin adherence and costs. Additional safeguard policies are needed for dual eligibles switching from FFS to Medicare managed care.
    • Underutilization of cardiovascular medications by aged Medicare beneficiaries with heart disease

      Doshi, Jalpa A.; Stuart, Bruce C. (2003)
      Despite publication of evidence-based guidelines and numerous quality-improvement efforts, studies have documented under use of effective cardiovascular medications in a large proportion of elderly patients with heart disease. This study has two objectives. The first is to examine utilization of effective cardiovascular medications for four common conditions among a nationally representative sample of institutionalized and community-dwelling aged Medicare beneficiaries. These include: warfarin in atrial fibrillation, beta-blockers in myocardial infarction, statins in hyperlipidemia and coronary artery disease, and ACE-inhibitors in diabetes with concomitant cardiovascular risk factors. The second aim is to assess policy-relevant risk factors associated with under use of these cardiovascular medications; specifically, to examine whether the risk of under treatment varies with the presence, type, and generosity of drug coverage; beneficiary age; physician specialty; geographic location; and type of residential setting. The study also examined whether the risk of under treatment of elderly beneficiaries has decreased over time. The data source was the Medicare Current Beneficiary Survey (1997--2000) supplemented with data from Medicare claims. Descriptive as well as multivariate analyses were conducted. Propensity score techniques were used to compare underuse between community and nursing home residents. Substantial underuse of cardiovascular medications was found across the four diseases. While the use of these medications in the outpatient setting has increased between 1997 and 2000, there still remains ample opportunity for improvement. Older age was negatively associated with cardiovascular medication use across the disease samples. Type and generosity of prescription coverage was correlated with use of the expensive statins and ACE-inhibitors/ARBs, and uncorrelated with use of the cheaply available warfarin and beta-blockers. Significant geographic variation was observed in all four cardiovascular medications. No significant differences were observed in cardiovascular medication use among patients treated by generalists versus cardiac specialists. Similarly, the risk of under treatment was the same in nursing homes and in the community. These findings point towards the need for increased educational efforts directed toward the physicians as well as patients. The findings relating to prescription coverage have direct implications for the Medicare drug benefit design. Overall, the results of this study can provide guidance to researchers, policy makers, and clinicians concerned with improving the quality of cardiovascular medication use among aged Medicare beneficiaries with heart disease.
    • Use of Antipsychotics in Medicare Part D: The Role of Coverage Restrictions and Impact of Medication Adherence on Hospitalization and Spending For Beneficiaries with Serious Mental Illness

      Roberto, Pamela; Stuart, Bruce C. (2016)
      Background: Coverage restrictions for antipsychotics are associated with access problems in Medicaid but their impact in Medicare Part D is unknown. The relationship between antipsychotic adherence, hospitalization risk, and treatment costs among Part D enrollees with serious mental illness also has not been systematically examined. Evaluating the association between adherence and outcomes is complicated by a recent policy change requiring the redaction of substance abuse claims from the Medicare research files, as hospitalizations and spending for beneficiaries with serious mental illness-many of whom have comorbid substance abuse disorders-may be systematically underreported. Methods: Enrollment, formulary, utilization, and spending data were obtained from the CMS Chronic Conditions Warehouse for 2011-2012. Effects of coverage restrictions on antipsychotic utilization were assessed by exploiting a unique natural experiment in which low-income beneficiaries were randomly assigned to prescription drug plans with varying levels of formulary generosity. The scope of the substance abuse claims redaction was determined by comparing unredacted beneficiary-level expenditure data to expenditures calculated from the redacted claims. Antipsychotic adherence was measured by the proportion of days covered (PDC) and stratified into four categories: PDC<0.70; 0.70≤PDC<0.80; 0.80≤PDC<0.90; and PDC≥0.90. Probit regressions and two-part generalized linear models were used to examine relationships between adherence in year one and hospitalizations and expenditures, respectively, in year two. Results: Despite considerable variation in the use of coverage restrictions across Part D plans, there was no evidence that restrictions significantly altered utilization patterns or reduced overall utilization of antipsychotics. Nearly one in five Part D enrollees with serious mental illness had claims affected by the redaction and average Part A spending among those with redacted claims was underreported by 57%. Based on analysis of the unredacted data, adherence to antipsychotic therapy was associated with significantly lower probability of psychiatric hospitalization and lower hospital expenditures. Conclusions: Coverage restrictions do not limit access to antipsychotics among Part D enrollees. Benefits of sustained adherence to antipsychotics for beneficiaries with serious mental illness include lower probability of psychiatric hospitalization and lower hospital spending. Researchers should exercise caution when using redacted Medicare claims to analyze utilization and spending outcomes for this population.