 To examine how cost sharing for prescription drugs affects compliance with antiplatelet therapy and subsequent health outcomes among patients with acute coronary syndrome (ACS).

 Retrospective outcomes study using administrative data from medical and pharmaceutical claims of patients enrolled at health plans offered by 26 large employers drawn from all regions of the country.

 A total of 14,325 patients were diagnosed as having ACS and underwent coronary stent implantation between 2002 and 2005. Each patient was followed up for a maximum of 2 years. Primary outcomes measures were adoption of outpatient antiplatelet therapy, adherence to outpatient therapy, hospital admissions, and healthcare expenditures.

 Patients with ACS who face higher coinsurance are less likely to adopt outpatient antiplatelet therapy within the first month after stent implantation and are more likely to discontinue treatment in the first year after stent implantation (P <.01). Higher coinsurance is also associated with an increased number of ACS rehospitalizations (P <.01). For patients in health plans with higher coinsurance rates, expected costs from ACS hospitalizations are  $2796 (38%) higher in the first year after stent implantation (P <.01).

 Higher copayments for prescription drugs are associated with lower utilization of antiplatelet therapy and with higher likelihood of rehospitalization among patients with ACS. As a consequence, total healthcare spending for patients with ACS increases by approximately $615 in the first year after stent implantation.

Articles

The American Journal of Managed Care

This article studies the association of pharmacy cost sharing with antiplatelet therapy utilization and health outcomes for patients with acute coronary syndrome.

Pharmacy Cost Sharing, Antiplatelet Therapy Utilization, and Health Outcomes for Patients With Acute Coronary Syndrome

 To investigate the effect of pharmacist management of poorly controlled diabetes mellitus in a community-based primary care group.

 Randomized controlled trial of pharmacist management of diabetes compared with usual medical care.

 Patients 18 years or older with glycosylated hemoglobin (A1C) levels of 9.0% or higher were enrolled. Patients were randomly assigned to an intervention group (n = 52) or a control group (n = 51). Management in the control group included the use of registries and targeted patient outreach. The intervention group participated in the same outreach program plus medication management, patient education, and disease control by a pharmacist.

 Nonparametric data showed median A1C decreases of 1.50% for the intervention group and 

0.40% for the control group (P = .06). Significantly more patients in the intervention group improved their A1C level by at least 1.0% relative to the control group (67.3% vs 41.2%, P = .02). Most of this benefit was seen for patients of nonwhite race/ethnicity compared with control subjects (56.3% vs 22.7%, P = .03). Male patients showed significantly greater benefit as well, with a median A1C decrease of 1.90% vs 0.15% for controls (P = .03).

 Patients with poorly controlled diabetes improved A1C levels significantly when pharmacist management was added to an aggressive organizational diabetes management program. Our results suggest that clinically trained pharmacists can help primary care providers improve diabetes management, especially among male patients and among patients of nonwhite race/ethnicity.

Pharmacist management resulted in improved glycosylated hemoglobin level among patients who had failed systematic outreach and regular use of an all-payer registry.

Improving management of patients with poorly controlled diabetes mellitus using pharmacist intervention is time intensive and resource intensive. 

Our data suggest that male patients and patients of nonwhite race/ethnicity are the greatest responders to pharmacist management. This finding needs to be confirmed by further research.

	Diabetes mellitus affects 23 million people in the United States.

 Adjusting for population age differences, survey data between 2004 and 2006 indicate that 6.6% of non-Hispanic whites, 7.5% of Asian Americans, 10.4% of Hispanics, and 11.8% of non-Hispanic blacks have diabetes mellitus.

 Totaling $174 billion in 2007, diabetes care accounts for 19% of total healthcare costs.1 Among patients with diabetes, 21% to 43% have glycosylated hemoglobin (A1C) levels exceeding 9.0% according to a 2004 study

; evidence suggests that attempting to achieve A1C levels below 7.0% will delay, ameliorate, or prevent the microvascular and neuropathic complications of diabetes.

 Data on trying to lower A1C levels below 7.0% are inconclusive.

 guidelines recommend a goal of less than 7.0% for A1C control.

	The medical literature demonstrates improvement in patient A1C levels with the use of clinical databases and case managers.

 Case managers coordinate all disease-related care and education, as well as support patient self-management. Early investigations showing the value of nurse case managers were conducted in an era of fewer medications and less aggressive management, usually involving patients enrolled in health plans, notably health maintenance organizations.

 Patients in the nurse case management group attained significantly lower A1C levels by 0.6% to 1.0% compared with the control group.

 have used a model whereby pharmacists work in collaboration with primary care physicians. Given the ever-increasing options for treating diabetes, pharmacists should be well positioned to evaluate the merits of each therapeutic option and to provide patient education. Studies showing the benefit of pharmacists serving as case managers have been small and often nonrandomized. The few randomized controlled trials have been performed in large urban academic or diabetes referral centers. As an example, Choe et al10 showed improved A1C levels and process outcome measures in a single-university ambulatory internal medicine clinic for patients with diabetes. We add to the literature a randomized study that describes pharmacist management of patients with diabetes who failed to respond to systematic patient outreach using a regularly updated all-payer registry in the primary care setting.

	This was a 12-month prospective randomized controlled study. The primary outcome measure was the change in A1C level at the end of 1 year. Secondary outcomes were the percentage of patients with a 1.0% decrease in A1C level. This outcome was assessed in the population as a whole and asa function of minority racial/ethnic group status. The research committee and the institutional review board of Saint Mary’s Health Care, Grand Rapids, Michigan, approved this study.

	The study site was the Advantage Health Physician Network (AHPN), which uses an electronic registry that identifies all adults with diabetes mellitus. Thirteen AHPN offices participated, including 3 urban, 9 suburban, and 1 rural site. Clinical practice guidelines, diabetes indicators, and performance thresholds are tracked routinely and are shared regularly with providers and staff. Each office has specific personnel budgeted to support diabetes quality-related initiatives. The latest quality indicators for individual patients are available at each office visit. In addition, there is systematic telephone and mail outreach to patients who are due for diabetes-related care.

	Patients with diabetes 18 years or older having A1C levels of 9.0% or higher or no office visits within 12 months were contacted by a study nurse. The nurse saw the patient at his or her home and determined study eligibility. Patients were excluded from the study if an endocrinologist was managing their diabetes or if they were not expected to live for the duration of the study. Eligible patients were tested using a Bayer DCA 2000 (Bayer Corporation, West Point, CT) point-of-care A1C instrument and were offered enrollment if their A1C level was 9.0% or higher. The study nurse obtained institutional review board—approved informed consent at this point from patients who agreed to participate and to attend all study visits.

	Using sequential envelopes containing computer-generated group assignment, the research nurse randomized patients to the intervention group or to the control group. Envelopes were opened after patients were deemed eligible for the study. Both study groups received the aggressive outreach already described.

	One pharmacist (JPJ) provided all diabetes-related care for the intervention group. The pharmacist is a board-certified pharmacotherapy specialist, who participated in an American Society of Health-System Pharmacists diabetes management traineeship, an American Diabetes Association postgraduate course in diabetes management, and an American Association of Diabetes Educators training program.

	Patients in the intervention group also met with the pharmacist at their respective primary care site for an assessment of adherence, barriers to optimizing blood glucose levels, and current medication regimen.All intervention patients received individualized education regarding diabetes self-management, including diet, exercise, blood glucose level testing, medications, and insulin. The pharmacist followed guidelines of the Management of Hyperglycemia in Type 2 Diabetes.16 This included early switching to insulin therapy after failure of 2 oral medications. The patient’s primary care physician approved any changes in medication or therapy, although the pharmacist was given autonomy to adjust insulin doses as needed. The number of subsequent visits with the pharmacist was based on the need to further educate the patient about diabetes control or to monitor therapeutic changes. Follow-up visits were supplemented with telephone calls as needed for medication management. Patients were followed up for 12 months after study enrollment. We were unable to assess adverse events in the control group. There was 1 severe hypo-glycemic event in the intervention group. 

 were defined as those requiring assistance from another person.

	A power analysis indicated that at least 39 patients were needed in each study group to show a clinically significant decrease in A1C level of at least 1.0% (ß = 0.20 and α = 0.05). 

 Test and Fisher exact test were used where appropriate for nominal data. Mann-Whitney test was used for nonparametric quantitative data and unpaired 

 test for parametric quantitative data. Significance was set at 

 <.05. The statistical software used was NCSS 2004 (Kaysville, UT).

	Recruitment began in May 2006, and final 1-year follow-up for the last patient was in December 2007. Among all AHPN patients with diabetes, 902 had an A1C level of 9.0% or higher or had not been seen by their primary care provider within the previous 12 months. Of these, 491 patients were successfully contacted and were invited to participate in the study. Two hundred thirty-five patients were excluded; 191 (39.0%) declined participation, and 44 (9.0%) were managed by a specialist. An additional 152 (31.0%) had an A1C level of less than 9.0% at the time they saw the study nurse. The remaining 104 patients were randomized, 52 to the control group and 52 to the intervention group.

	One patient in the control group was excluded as an outlier based on A1C level exceeding 3 SDs from the mean A1C change. This patient was hospitalized for 3 weeks because of diabetes complications and subsequently attained an 8.6% decrease in A1C level. Therefore, the final analysis was based on 103 patients, 52 in the intervention group and 51 in the control group. Baseline demographic data are given in 

. The only significant difference between study groups at baseline was lower A1C level in the intervention group.

	The A1C changes were not normally distributed, so median values were used as the measure of central tendency. The overall median A1C reduction in the intervention group was 1.1% greater than that of the control group. This difference approached but did not achieve statistical significance. Post hoc subgroup analysis showed that male patients in the intervention group achieved a statistically significant improvement in their A1C level (median, −1.90%; interquartile range, −0.05% to −2.95%) versus the control group (median, −0.15%; interquartile range, 0.98% to −1.38%]). 

	From the perspective of individual patient improvement, statistically more patients in the intervention group achieved at least a 1.0% improvement in A1C level overall. Twice as many patients of nonwhite race/ethnicity and male patients (post hoc) in the intervention group exceeded the 1.0% improvement mark. No treatment effects using this measure were seen for patients of white race/ethnicity or for female patients (

	This intervention required considerable effort. The pharmacist averaged 6 office visits and 3 telephone calls per patient over the course of a year. Office visits lasted between 30 and 60 minutes. Telephone calls were 10 to 20 minutes in length. The pharmacist changed the drug regimen to basal-bolus insulin in 15 patients (28.8%) and discontinued oral medications completely in 15 patients (28.8%). In the control group, only 1 patient’s regimen was changed to basal-bolus, oral medications were discontinued in 1 patient, and insulin was added for 1 patient. As evidence of aggressive treatment in the control group, there were 6000 patients in the all-payer registry in 2007, and 55.1% had both an A1C level of less than 7.0% and low-density lipoprotein cholesterol (LDL-C) level of less than 100 mg/dL (to convert cholesterol level to millimoles per liter, multiply by 0.0259). This is substantially better than 2007 data in Medicaid beneficiaries, among whom the 90th percentile is 30.2% A1C control and 42.3% LDL-C control.

	This study was undertaken to ascertain the benefit of having a pharmacist work with primary care physicians to reduce A1C levels among patients with poorly controlled diabetes who failed to respond to systematic patient outreach in multiple community-based clinics. Greater health benefits accrue when moving from poor (A1C level >9.0%) to moderate glycemic control compared with moving from moderate to almost normal glycemic control. Vijan et al

 showed that 80% of reductions in blindness and in end-stage renal disease resulted from targeting less than 20% of patients with the worst A1C control. Our overall findings of significantly improved A1C outcomes among individuals in the intervention group are compatible with the findings of 2 recent comprehensive reviews

 of pharmacist management of diabetes. Our study was not originally powered to show subgroup differences. Despite this, statistically more patients of nonwhite race/ethnicity and male patients achieved at least a 1.0% improvement in A1C level. To our knowledge, this subgroup differentiation has not been previously identified and needs to be confirmed by future research. However, our finding of greater improvement with pharmacist intervention among patients of nonwhite race/ethnicity is congruent with results of studies that were limited to specific racial/ethnic groups. Studies

 among subjects of nonwhite race/ethnicity showed much greater improvement over control values than studies

 among subjects of white race/ethnicity. Minority racial/ethnic groups have not only poorer controlled diabetes but also more complications.

 Compared with non-Hispanic whites having diabetes, Mexican Americans and African Americans with diabetes have at least a 4-fold greater incidence of kidney disease, and Mexican Americans with diabetes have a 1.8-fold greater incidence of peripheral vascular disease.

 Given the disproportionate prevalence of diabetes and diabetic complications among patients of minority races/ethnicities, coupled with poorly controlled A1C level, case management with a pharmacist can be of benefit to these populations and is a tool that is likely to have clinically meaningful results.

	The intervention group showed more improvement, despite having lower baseline A1C levels than the control group. This could mean that diabetes in the control group was more difficult to manage. However, previous investigations have found greater improvement with higher baseline A1C levels, suggesting that the intervention group was actually at a disadvantage.10 Among male patients, approximately 2 patients had to be treated by a pharmacy case manager for 1 year to have 1 patient improve his A1C level by at least 1.0%. Among minority races/ethnicities, 3 patients had to be treated by a pharmacy case manager for 1 year to have 1 patient improve his or her A1C level by 1.0%. Among patients in the United Kingdom Prospective Diabetes Study

with type 2 diabetes mellitus, each 1.0% reduction in updated mean A1C level led to a 21% reduction in death, 14% reduction in myocardial infarction, and 37% reduction in microvascular complications. Therefore, male patients and patients of nonwhite race/ethnicity in our intervention group achieved A1C improvements that are likely to be clinically significant. We did not power the study to evaluate blood pressure, cholesterol, retinopathy, or nephropathy screening differences in the study groups. Results of other studies

 suggest that pharmacists can improve these measures as well.

	The fact that basal-bolus insulin was started in 28.8% of patients in the intervention group versus 2.0% in the control group indicates a different approach to diabetes management by the pharmacist. In addition to adding insulin, the pharmacist simplified medication regimens by discontinuing oral medications in 28.8% of patients versus 2.0% in the control group. The average patient was in contact with the pharmacist for more than 300 minutes during the 1-year intervention. This significant time and cost highlight the commitment it takes to help patients overcome the difficulty of intensifying treatment. The time commitment is similar to that seen in the Diabetes Control and Complications Trial.

	This study extends findings of the Diabetes Ten City Challenge

 regarding the benefit of pharmacist intervention. We randomized all patients regardless of insurance status, whereas the Diabetes Ten City Challenge recruited volunteers and excluded one-quarter of them from their final outcomes because of missing economic data. In addition, our patients were selected as the most costly and challenging among patients with diabetes. As expected with higher baseline levels, A1C decrease was greater in our study than in the Diabetes Ten City Challenge. The Asheville Project studied a select insured population and evaluated the economic benefit of using pharmacists, certified diabetes educators, and enhanced benefit plans in a community pharmacy diabetes care program.

	Limitations of our study include the use of a single primary care group, although this is mitigated by the use of diverse sites (ie, urban, suburban, and rural). This study was not designed to prove whether a pharmacist (as opposed to other types of providers) is specifically needed to obtain the benefits of this case management approach. The subsidized nature of the pharmacist time and cost in this study threatens external validity. Finally, the participating pharmacist herein was clinically trained and experienced in patient care activities such as medication management and motivational interviewing. However, community pharmacists without specialized training in diabetes have been shown to improve outcomes in a health maintenance organization population.

 Therefore, the pool of available pharmacists may be greater than expected. Overall, the results of this study should be generalizable to similarly equipped primary care groups employing an experienced clinical pharmacist.

	The findings of this study should stimulate healthcare policymakers to evaluate the cost and benefit of including a pharmacist case manager in primary care networks as part of a systematic approach to diabetes management. Our study demonstrated improvement in disease management by a pharmacist for the most costly segment of the population with diabetes (those with A1C level >9.0%). This poorly controlled group still represents 10% of patients in our baseline population with aggressive interventions already in place (55.1% with A1C level <7.0% and LDL-C level <100 mg/dL). The benefit was seen among insured patients and among uninsured patients. For reasons yet to be explained, male patients and patients of nonwhite race/ethnicity showed disproportionately greater improvement. Our study provides a foundation for future research into pharmacist case management in a community-based primary care practice rather than in a health plan or university setting.

We thank Susan Emelander, RN, for her indispensable role in patient enrollment and follow-up. We thank Andrea Wendling, MD, Sister M. Rosita Schiller, PhD, Alan T. Davis, PhD, and Tracy Frieswyk, BA, for manuscript review and suggestions.

From the Grand Rapids Family Medicine Residency Program (JPJ), Ferris State University, Grand Rapids, MI; Depart-ment of Pharmacy Practice (JPJ), Michigan State University, East Lansing, MI; Saint Mary’s Health Care (JPJ, PJB), Grand Rapids, MI; and Advantage Health Physician Network (PJB), Grand Rapids, MI.

Financial support for this study was provided by Advantage Health Physician Network, Doran Foundation, Michigan Phar-macist Foundation, Priority Health, and Western Michigan Society of Health System Pharmacists.

Dr Jameson reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. Dr Baty is a practicing clinician in the Advantage Health Physician Network, the site where the study was conducted.

CLINICAL

Pharmacist management of poorly controlled diabetes mellitus in this randomized trial resulted in more patients decreasing their glycosylated hemoglobin level by at least 1.0%.

Pharmacist Collaborative Management of Poorly Controlled Diabetes Mellitus: A Randomized Controlled Trial

 To estimate the treatment effect of participation in an asthma intervention that was part of the National  Asthma Control Program.

 Cross-sectional; difference in outcomes between participants and comparable nonparticipants matched by  using propensity scores.

 Data on children who participated in asthma case management (n = 270) and eligible children who did not participate in case management (n = 2742) were extracted from a Medicaid claims database. We constructed measures of healthcare utilization, sociodemographics, and neighborhood characteristics. After creating a comparison group similar to the participants in terms of all characteristics before participation, we estimated the effect of the program on asthma outcomes.

 Participants were more likely to have vaccinations for pulmonary illness (95% confidence interval [CI] = 1.82, 4.81), to fill a prescription for controller medications (95% CI = 1.07, 2.19), and to have a refill for rescue medication (95% CI = 1.07, 2.07) after the program than comparable nonparticipants. There was no statistically significant difference in the number of nebulizer treatments or emergency department visits between the 2 groups.

 The program did increase the use of preventive healthcare by participants. Over the time period we studied, these behaviors did not decrease healthcare utilization for asthma exacerbations. We were unable to discern  whether the lack of effect was because of the nature of the program, heterogeneity of the effects, or barriers outside the program’s control.

An asthma case management program increased preventive care but did not significantly decrease markers of exacerbations.

Estimated Effect of Asthma Case Management Using Propensity Score Methods

 To assess knowledge and understanding of cost-sharing responsibilities and whether that knowledge and understanding influence actual and perceived use of healthcare services.

 A 3000-person random sample was drawn from a state employee database in Massachusetts.

 Survey responses and claims analyses were used to assess knowledge of cost sharing and healthcare utilization over a 3-year study period. Trend models and logistic regression were used.

 Nearly two-thirds of respondents (62%) accurately recalled the percentage of premium that they paid; 67% recalled the correct copayment for a doctor’s visit. Younger, less educated, and lower-income employees recalled their copayment more accurately than older, more educated, higher-income colleagues. Half of the respondents accurately reported the copayment amount for an emergency department visit. Greater knowledge of overall healthcare costs was positively associated with higher utilization of office visits (P <.01). Knowledge of specific office visit and emergency department copayments had no significant relationship with utilization. Self-reported delays and reductions in utilization were much more pronounced than the actual claims data indicated.

 Employees were reasonably well informed about their cost-sharing responsibilities. Knowledge of costs was associated with higher office visit utilization during the study period. Respondents who were more knowledgeable about their specific copayments for office visits and emergency department visits also were more likely to behave in what appeared to be a more cost-efficient manner, with more office visits and fewer emergency department visits.

Educating employees about their cost-sharing responsibilities could lead to more efficient use of the healthcare system.

Knowledge of Cost Sharing and Decisions to Seek Care

 To test the hypotheses that older patients with colorectal cancer (CRC) and rural patients are less likely to undergo surgery, radiation, and chemotherapy.

 A total of 6561 patients with CRC between January 1998 and December 2003 were identified by incident International Classification of Diseases for Oncology codes from the Nebraska Cancer Registry. In multivariate logistic regression analyses, we studied the association of age and residence county (rural vs urban and micropolitan) with each of 3 CRC treatments by anatomic site.

 After adjusting for patient demographics, insurance payer, ratio of providers to population, and cancer stage, patients with colon cancer living in micropolitan counties were more likely to receive chemotherapy than those living in rural counties (P <.001). Compared with patients aged 19 to 64 years, patients with colon cancer 85 years and older (P <.001) and patients with rectal cancer 75 years and older (P <.05) were less likely to undergo surgery. Patients with CRC 75 years and older were less likely to receive radiation, and patients with colon cancer 65 years and older and patients with rectal cancer 75 years and older were less likely to receive chemotherapy (P <.001 for both).

 In Nebraska, older patients with CRC were less likely to undergo surgery, radiation, and chemotherapy. Patients with colon cancer in rural counties were less likely to undergo chemotherapy than those in micropolitan counties. Decision makers need to consider issues of age and rural residence in patient access to CRC treatments.

POLICY

Older patients with colorectal cancer were less likely to undergo surgery, radiation, and chemotherapy; rural patients with colon cancer were less likely to undergo chemotherapy.

Age and Rural Residence Effects on Accessing Colorectal Cancer Treatments: A Registry Study

An extensive literature has documented the impact of high copayments on medication use. These studies suggest that for each 10% increase in cost sharing, prescription drug use decreases by 2% to 6%, depending on the class of drug and indication for its use.

 In the case of “essential” medications, these reductions in medication consumption should be associated with adverse clinical consequences.

 After all, if randomized trials and other forms of high-quality evidence have established the relationship between the (greater) use of a drug and improvements in health outcomes, then the opposite must logically be true. Numerous studies demonstrate this apparent association.

 For example, patients who discontinue evidence-based medications after an acute myocardial infraction

 are 3 times more likely to die than patients who remain adherent.

However, the validity of these observations is threatened by an important bias: patients who stop therapy are often sicker than patients who do not, either because of true clinical differences or their likelihood of engaging in healthy activities (see 

). This “sick stopper bias” is a direct analog of the “healthy user effect,” whereby patients who adhere to preventive therapies are more likely than nonadherent patients to engage in a broad spectrum of healthy  behaviors such as use of preventive services.

 As a result, adherent patients are less likely to experience adverse health outcomes (eg, motor vehicle accidents) that are related and unrelated to the therapeutic effects of the drugs being evaluated.

 In this way, the apparent association between medication discontinuation and adverse clinical outcomes (

) may actually be due to confounding by a patient’s healthy lifestyle and health choices (

), even in analyses that use multivariable models to carefully adjust for observable clinical covariates (

, Philipson and colleagues evaluate the association between cost sharing for clopidogrel, medication discontinuation, and rates of acute coronary syndrome (ACS) rehospitalization.

 Clopidogrel is an essential drug in the management of patients who have experienced an ACS and undergone percutaneous coronary intervention.

 Thus, an evaluation of the implications of high cost sharing for this drug is an important contribution to the literature. The authors demonstrate that patients with 10% higher coinsurance for clopidogrel have a 0.74 percentage point greater risk per quarter of medication discontinuation and a 0.76 percentage point greater risk per quarter of ACS rehospitalization.

The authors recognize the pitfalls of directly evaluating the association between the observed reductions in clopidogrel use related to higher cost sharing and the subsequent changes in the rate of hospitalization, and thus do not do so. Nevertheless, their results clearly imply this relationship, and the 2 associations they do evaluate may be confounded by the same factors they seek to avoid (see the Figure). The relationship between higher cost sharing and reduced medication use (

) characteristics. Similar concerns could exist for the relationship between cost sharing and utilization (

If, for example, sicker patients chose or ended up in plans with higher levels of cost sharing, the apparent association between higher cost sharing and lower utilization may be the result (in part or in whole) of these patient characteristics. Philipson et al evaluated selection effects by measuring the average number of patient comorbidities in the year after the index ACS discharge. Other factors inherent to the variations in plan design that are the focus  of the authors’ investigation are more difficult to address. For example, patients with certain preferences or medical histories (ie, before their index ACS) may seek employers or jobs with less or more generous benefits (even if these factors don’t influence plan choice after the ACS has occurred). More directly, the generosity of drug coverage for clopidogrel is likely related to cost sharing for other evidence-based medications which, if stopped, also confer near-term risks.

 Thus, it may not be accurate to attribute all of the observed effects of high cost sharing to clopidogrel alone, especially since the primary independent variable in Philipson et al’s analysis is an overall measure of plan generosity (ie, for all drugs and not specifically for clopidogrel).

Perhaps most telling is the fact that the increased rate of clopidogrel discontinuation associated with higher cost sharing is almost identical to the increased rate of cardiac rehospitalization. This implies that all patients who discontinued clopidogrel were rehospitalized because of it or that a smaller

number of patients had multiple subsequent hospitalizations all as a result of cost-related medication discontinuation. A rebound phenomenon has been associated with stopping clopidogrel, but the expected event rate is far lower than that observed.

 Further, the actual cost for an ACS hospitalization

for patients with high drug cost sharing was greater than that for patients with low cost sharing. This too may reflect the characteristics of patients who discontinued clopidogrel (ie, sick patients stop therapy and sick patients also have longer and more complicated hospitalizations), rather than the consequences of the discontinuation itself.

How might such confounding be addressed? In the absence of randomized cost-sharing evidence,

 a variety of design and analytic strategies might help. Markers of health-seeking behavior, such as receipt of immunizations

regression analyses. High-dimensional propensity scores

 may provide for better covariate balance and more valid inferences of treatment effects. When a natural cost-sharing change occurs, interrupted timeseries methods limit confounding to co-occurring events, especially when a control group is used.

It is a clear policy priority to understand the clinical and economic trade-offs associated with high patient cost sharing for essential medications. This is especially true as the level of cost sharing faced by commercially insured patients continues to increase,

 and the economy stagnates. While Philipson et al’s results are consistent with a large literature on this topic, their analysis highlights the challenges in quantifying a relationship that seems to be a truism.

From the Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA.

Dr Choudhry receives research funds from the Commonwealth Fund, the Robert Wood Johnson Foundation, and Aetna for work related to setting copayments for cardiovascular medications. He is a consultant to Mercer Health and Benefits, Inc.

Concept and design; drafting of the manuscript; and critical revision of the manuscript for important intellectual content.

 Niteesh K. Choudhry, MD, PhD, Brigham and Women’s Hospital, 1620 Tremont St, Ste 3030, Boston, MA 02120. E-mail: 

 Prescription drug cost sharing: associations with medication and medical utilization and spending and health. 

 Medication adherence after myocardial infarction: a long way left to go. 

 Copayment levels and medication adherence: less is more. 

4. Ho PM, Rumsfeld JS, Masoudi FA, et al.

 Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. 

5. Colivicchi F, Bassi A, Santini M, Caltagirone C.

 Discontinuation of statin therapy and clinical outcome after ischemic stroke. 

6. Brookhart MA, Patrick AR, Dormuth C, et al.

 Adherence to lipid-lowering therapy and the use of preventive health services: an investigation of the healthy  user effect. 

7. Dormuth CR, Patrick AR, Shrank WH, et al.

 Statin adherence and risk of accidents: a cautionary tale. 

8. Philipson TJ, Mozaffari E, Maclean JR.

 Pharmacy cost sharing, antiplatelet therapy utilization, and health outcomes for patients with acute coronary syndrome. 

 Meta-analysis of the efficacy and safety of clopidogrel plus aspirin as compared to antiplatelet monotherapy for the prevention of vascular events. 

 The economic consequences of non-evidence-based clopidogrel use. 

11. Tricoci P, Roe MT, Mulgund J, et al. 

Clopidogrel to treat patients with non-ST-segment elevation acute coronary syndromes after hospital discharge. 

12. Cannon CP, Braunwald E, McCabe CH, et al; Pravastatin or Atorvastatin Evaluation and Infection Therapy-Thrombolysis in Myocardial Infarction 22 Investigators.

 Intensive versus moderate lipid lowering with statins after acute coronary syndromes [published correction appears in  

 Incidence of death and acute myocardial infarction associated with stopping clopidogrel after acute coronary syndrome [published correction appears in 

14. Choudhry NK, Brennan T, Toscano M, et al.

 Rationale and design of the Post-MI FREEE trial: a randomized evaluation of first-dollar drug coverage for post-myocardial infarction secondary preventive therapies.

15. Kulik A, Brookhart MA, Levin R, Ruel M, Solomon DH, Choudhry NK.

 Impact of statin use on outcomes after coronary artery bypass graft surgery. 

16. Ho PM, Magid DJ, Shetterly SM, et al.

 Medication nonadherence is associated with a broad range of adverse outcomes in patients with coronary artery disease. 

17. Schneeweiss S, Rassen JA, Glynn RJ, Avorn J, Mogun H, Brookhart MA.

 High-dimensional propensity score adjustment in studies of treatment effects using health care claims data. 

18. Rassen JA, Brookhart MA, Glynn RJ, Mittleman MA, Schneeweiss S.

 Instrumental variables I: instrumental variables exploit natural variation in nonexperimental data to estimate causal relationships. 

19. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D.

 Segmented regression analysis of interrupted time series studies in medication use research. 

20. Kaiser Family Foundation and Health Research and Educational Trust.

 Employer Health Benefits 2007 Annual Survey. 

http://www.kff.org/insurance/7672/index.cfm

The relationship between higher cost sharing and reduced medication use may be confounded by patients' healthy lifestyles or their health choices.

Relationship Between High Cost Sharing and Adverse Outcomes: A Truism That's Tough to Prove

 To identify potentially low-value services for inclusion in value-based insurance design (VBID) programs and to discuss challenges involved in incorporating such information.

 We searched the Tufts Medical Center Cost-Effectiveness Analysis Registry (

) to identify examples of low-value services, defined as interventions that make health worse without saving money or those that cost at least $100,000 per quality-adjusted lifeyear gained. We restricted our attention to papers published since 2000. We supplemented this literature review with a list of services recently rejected by the United Kingdom’s National Institute for Health and Clinical Excellence for coverage by the UK’s National Health Service.

 The list of potentially low-value services includes several drugs to treat cancer, as well as other therapies such as left ventricular assist devices and lung volume reduction surgery. Building negative incentives into VBID programs to discourage use of low-value care will involve a number of challenges, including identification of appropriate candidates; the scope of services to be covered (ie, whether VBID should be expanded beyond drugs to address medical devices, procedures, and diagnostics); and whether VBID programs should target specific subgroups.

 Identifying noncontroversial low-value services and designing VBID programs to discourage their use will not be easy. However, to fulfill their promise of improving value and moderating cost growth, VBID programs should target low-value as well as high-value care.

Building negative incentives into value-based insurance design programs to discourage use of low-value care will involve a number of challenges.

Low-Value Services in Value-Based Insurance Design

Quality care has been defined as providing the right services to the right patient at the right time and place. Health insurance can support the provision of quality care by offering meaningful coverage of the right services for  the right patient—perhaps even at the right time and place. Defining what’s “right,” however, is a critical step in the process.

Health insurance traditionally has offered coverage for services that are generally accepted to be associated with improved outcomes. The advent of managed care brought about a greater scrutiny of the association between medical services and health outcomes, and evidence-based medicine has started to provide a knowledge base from which to examine this association.

 We have made substantial progress in offering coverage for services that are generally associated with improved outcomes.

The value-based insurance design (VBID) movement seeks to push traditional coverage on 2 fronts: (1) beyond yes/no decisions on coverage to gradients based on value and (2) beyond generalities to specific expectations of outcomes for the right patient. If a treatment is impressively effective in improving health  outcomes, but is not provided or adhered to because of health insurance reasons, VBID suggests that enhanced coverage should be provided to encourage patients to do the right thing. Recent evidence shows that this enhanced coverage is possible in specific instances.

 Conversely, if a treatment has little effect on outcomes, limited coverage should be provided to discourage patients from doing the wrong thing. Beyond yes/no decisions (and within the constraints of avoiding unreasonably complex plan designs), coverage should match the value offered by the service to a patient with a particular condition.

A good examination of the available information doesn’t bring us much closer to the goal of providing a complete matrix of coverage for a given treatment based on value and patient conditions.

 Comparative effectiveness research (CER) is much more likely than prior clinical research to include comparisons across treatments that are meaningful to payers. Ensuring that comparisons consider patient conditions and stratify outcomes will require vigilance.

Comparative effectiveness research recognizes the importance of costs, although ensuring that analyses include costs also will require vigilance. The recent Institute of Medicine report on priorities for CER included cost comparisons for 19 of the 100 conditions.

 A further requirement of research for the VBID application of CER will be identification of treatments that are sensitive to patient cost sharing.

At a high level, VBID may be applied to a yes/no coverage decision. Applying gradients of coverage based on value requires that patients respond to cost  sharing at meaningful levels. A clearly high-value or low-value treatment that already has 95% or 5%, respectively, selection and adherence would not benefit  from a VBID approach. Although development of a clinical and economic information base will take time and resources, it is feasible. Tuft’s Center for the Evaluation of Value and Risk in Health is a public service aimed at making such information available (

https://research.tufts-nemc.org/cear/Default.aspx

Successful implementation of VBID requires both information and courage. I suggest that the larger challenge is finding the courage to conduct the research and to implement coverage based on the findings. Among the elephants in the room is the relationship between the profitability of treatments and their outcomes. Administrators would be loath to talk about it publicly, but certain services are more profitable than others. Even before the seminal work of Siu et al,

 we knew that a considerable percentage of hospitalizations are unnecessary. Strict adherence to the evidence-based medicine that is encouraged through VBID may require rethinking our delivery systems. Our delivery systems may not be prepared for VBID.

Nearly a decade ago managed care was making yes/no coverage decisions and moving beyond generalities to assigning expectations of outcomes for specific patients. Granted, the information base for those decisions may not have been ideal and the tools may not have been in the right hands. The backlash against managed care highlighted the limitations of our information. More notably, that backlash highlighted our discomfort with making difficult decisions.

 A new era of VBID may require that medical directors contact patients about coverage, as opposed to speaking only with physicians.

Making coverage decisions on the basis of value for specific patients involves many challenges. Yet we must go down that road. With increased expenses for health insurance, pressures for reforming not only the availability but also the nature of coverage will rise. I hope that we find the courage to provide quality healthcare that is covered in a quality way.

 From the Department of Health Management and Policy, University of Michigan, Ann Arbor, MI.

 Dr Smith reports serving as a consultant to Benefit Based Decisions, LLC.

 Concept and design; drafting of the manuscript; and supervision.

 Dean G. Smith, PhD, Department of Health Management and Policy, University of Michigan, 109 Observatory, Ann Arbor, MI 48109-2029. E-mail: 

 Evidence that value-based insurance can be effective. 

3. Neumann PJ, Auerbach HR, Cohen JT, Greenberg D.

 Low-value services in value-based insurance design. 

 ‘Marginal medicine’: targeting comparative effectiveness research to reduce waste. 

5. McMahon LF Jr, Hayward R, Saint S, Chernew ME, Fendrick AM.

 Univariate solutions in a multivariate world: can we afford to practice as in the “good old days”? 

6. Committee on Comparative Effectiveness Research Prioritization, Institute of Medicine.

 Initial National Priorities for Comparative Effectiveness Research.  Washington, DC: National Academies Press; 2009. 

http://www.iom.edu/Reports/2009/ComparativeEffectivenessResearchPriorities.aspx

7. Siu AL, Leibowitz A, Brook RH, Goldman NS, Lurie N, Newhouse JP.

 Use of the hospital in a randomized trial of prepaid care. 

 The managed care backlash: perceptions and rhetoric in health care policy and the potential for health care reform. 

Provision of quality healthcare covered in a quality way requires both information and courage.

Getting the Right Services Covered by Health Insurance

To examine whether physicians' use of information technology (IT) was associated with better knowledge of drug costs.

A 2007 statewide survey of 247 primary care physicians in Hawaii regarding IT use and self-reported knowledge of formularies, copayments, and retail prices.

High rates of health information technology use by physicians were only modestly associated with better knowledge of drug costs.


April 2010