Vicki Fung, PhD

 The introduction of generic second-generation antipsychotics (SGAs), starting with risperidone in July 2008, could reduce antipsychotic spending and cost-related use barriers. This study examines associations between generic risperidone use and spending and adherence after introduction among Medicare Advantage (MA) beneficiaries.

 The study included MA beneficiaries receiving SGA treatment prior to July 2008. We examined antipsychotic spending using linear models, adherence (proportion of days covered ≥80%) using logistic models, and nonpersistence (time to first gap in antipsychotic use >30 days) in 2009 using Cox proportional hazard models, comparing beneficiaries with versus without generic use, adjusting for individual and plan characteristics.

Between July 2008 and December 2009, 22.8% of beneficiaries had ≥1 fill of generic risperidone: 73% of those previously using branded risperidone and 6.7% of those previously using other SGAs. Beneficiaries in private fee-for-service (FFS) versus health maintenance organization (HMO) plans had lower rates of generic use (hazard ratio [HR], 0.73 [95% CI, 0.56-0.96]); however, cost-sharing levels were not associated with generic use. Compared with beneficiaries who continued using other SGAs, those who switched from other SGAs to generic risperidone in 2008 had lower out-of-pocket spending (—$214; 95% CI, –$314 to –$115), higher adherence (odds ratio, 2.34; 95% CI, 1.62-3.40) and lower rates of nonpersistence (HR, 0.56; 95% CI, 0.46-0.69) in 2009.

Generic use was concentrated among patients previously using branded risperidone. HMO plans appeared to be more effective at encouraging generic use than unmanaged private FFS plans; however, patient financial incentives had limited influence on switching. Additional opportunity remains to encourage greater generic SGA use, reduce spending, and potentially improve treatment adherence and outcomes.

	Risperidone was the first widely used second-generation antipsychotic (SGA) available as a generic. We examined patterns of use in the first 18-months of introduction and found the following: 

		Use of generic risperidone was concentrated among those previously using branded risperidone (ie, bioequivalent substitution); there was limited therapeutic (non-bioequivalent) substitution from other brand SGAs. 

		Rates of generic risperidone take-up were greater in Medicare Advantage HMO plans than in private fee-for-service plans; however, patient financial incentives had limited influence on generic use. 

		Beneficiaries who switched to generics had lower out-of-pocket spending and higher levels of drug adherence compared with beneficiaries who remained on brand SGA therapy.

Antipsychotics are among the top selling classes of drugs in the United States, largely driven by spending on second-generation antipsychotics (SGA). Antipsychotics are one of 6 protected drug classes within the Medicare Part D prescription drug program, meaning that plans are required to include all or substantially all drugs within the class on plan formularies. Medicare spending on antipsychotics was $5.9 billion in 2009, second only to antihyperlipidemics.

Several commonly used SGAs have recently lost patent protection, starting with Risperdal (risperidone) in July 2008, and then Zyprexa (olanzapine), Geodon (ziprasidone) and Seroquel (quetiapine) in 2011 and 2012. The entry of new generic SGAs could result in substantial cost savings for Medicare and other payers. The Congressional Budget Office estimated that the overall Part D savings attributable to generic substitution in 2007 were approximately $33 billion.

However, realizing savings associated with the availability of generic antipsychotics could be more challenging than in other therapeutic classes. Physicians and patients are often hesitant to change psychotropic drug regimens for stable patients due to concerns about lack of tolerability or effectiveness of new regimens.

 Market reports suggest that the entry of generic risperidone did not decrease the market share of other SGAs, suggesting limited generic substitution across molecules (ie, nonbioequivalent substitution).

 Even for bioequivalent substitution of the same molecule, some studies note concerns about potential reductions in adherence due to patients’ anxieties regarding changes in the name, packaging, or appearance of the drug.

 Conversely, greater generic substitution could improve adherence by reducing the out-of-pocket costs associated with treatment

; however, little is known about the effects of switching to generic SGAs on spending or adherence.

We examined changes in SGA treatment choices among Medicare Advantage (MA) beneficiaries in the first 18-months after generic risperidone introduction (July 2008—December 2009) and Part D plan characteristics associated with generic use. We also examined the associations between generic risperidone use and antipsychotic spending and adherence.

This study included noninstitutionalized beneficiaries enrolled in MA prescription drug plans offered by a national carrier, including health maintenance organizations (HMOs), preferred provider organizations (PPOs), and private fee-for-service (FFS) plans. Identifying information on the plans included in the study has been removed, including markets served and exact details on cost sharing and plan structures.

MA-HMO plans have the most closed physician networks, whereas PPOs have more open networks. Until 2011, private FFS plans were not required to have formal provider networks and included the same physicians as in traditional Medicare (96% of eligible physicians, nationally).

 Because plans with tighter network structures have a greater range of tools with which to influence physician behavior, we hypothesized that generic use would be greater in HMO versus private FFS plans. Examples of commonly used tools include providing feedback to physicians on their relative rates of generic use and drug spending, and the use of physician financial incentives to encourage generic drug use.

Antipsychotics are a protected class under Part D; thus, plans have to include all drugs within the class in their formularies, but could vary tier placement and/or cost-sharing amounts for drugs within the class. We hypothesized that greater cost sharing for brand relative to generic drugs would be associated with higher rates of generic risperidone use.

All study plans had a similar formulary for antipsychotics in terms of tier placement, but co-payment levels for these tiers varied across plans. In 2007, most SGAs were preferred brand drugs (tier 2). Following the introduction of generic risperidone in 2008, risperidone was on tier 1 and Risperdal (risperidone) became nonformulary, or a nonpreferred brand, with a try/fail requirement for generic risperidone. Most of the other commonly used SGAs remained preferred brands without utilization management requirements, while some SGAs became nonpreferred brands.

We focused on beneficiaries with at least 1 SGA fill between January and June 2008, and examined SGA use patterns after the introduction of generic risperidone in July 2008. Beneficiaries were censored upon disenrollment or death. Because we focused, in part, on the relationship between cost-sharing levels and drug choices, we excluded beneficiaries receiving Part D low income subsidies (LIS) who had minimal cost sharing. In sensitivity analyses among LIS beneficiaries, findings were consistent.

We linked Part D Event files, medical claims, plan information, and beneficiary characteristics for all subjects. To identify the type of SGA use, we used the Medi-Span Electronic Drug File (version 2)

Generic Risperidone Use and Part D Plan Characteristics

We examined the cumulative proportion of beneficiaries with any generic risperidone use between July 2008 and December 2009. We used Cox proportional hazard models to examine the plan characteristics associated with time to first generic risperidone use. We stratified both analyses by whether beneficiaries were previously using brand risperidone or other SGAs before generic introduction.

We classified beneficiaries’ plan types as HMO, PPO or private FFS, and beneficiaries’ benefit designs as enhanced or basic Part D—enhanced plans included coverage for generics during the standard coverage gap (eg, after $2510 in total drug spending in 2008) and basic plans had no gap coverage. We also examined the relative cost of brand versus generic drugs during the initial coverage period prior to the gap, and focused on the preferred brand (tier 2) versus generic (tier 1) co-payment differential because the most commonly used SGAs were tier 2 drugs throughout the study period.

These models adjusted for beneficiary age, gender, race/ethnicity, Part D comorbidity risk score (RxHCC), mental health diagnoses, reason for Medicare entitlement (disabled vs aged), and prior drug spending. As a measure of beneficiaries’ stability on their SGA regimens, we adjusted for the length of time beneficiaries were receiving the last SGA used prior to July 2008. Among non-risperidone users, we also adjusted for the generic name of the last SGA used prior to July 2008.

Generic Substitution, Spending, and Adherence

We examined the association between SGA drug use patterns and antipsychotic spending (total and out of pocket) and treatment adherence and persistence. Our primary analyses examine variations in these outcomes in 2009 based on 2008 switching patterns. We focused on 4 SGA use patterns, as defined by the last SGA used prior to July 2008 and SGA use between July and December 2008: 1) bioequivalent generic substitution (brand risperidone to generic risperidone), 2) nonbioequivalent generic substitution (non-risperidone SGA to generic risperidone), 3) nonbioequivalent brand switching (non-risperidone SGA to other SGA brand), and 4) no change (same non-risperidone SGA brand throughout 2008). Beneficiaries with no SGA fills between July and December 2008 were excluded from these analyses (19.9% and 23% of brand risperidone and non-risperidone users, respectively). Because Part D Event files do not capture data on uncovered drugs, we reliably cannot assess brand risperidone use after July 2008.

Antipsychotic Spending, Adherence, and Nonpersistence

We examined total and out-of-pocket antipsychotic drug spending in 2009. We examined adherence monthly (January 2008-December 2009) and annually (2009) using the proportion of days covered (PDC). The PDC was calculated as the sum of available days’ supply of all antipsychotics divided by the total days in each time period. We defined nonpersistence as having a gap in antipsychotic supply >30 days.

To examine the associations between patterns of SGA use (eg, generic substitution) in 2008 and annual 2009 drug spending and adherence (ie, PDC >80%), we use linear and logistic regression models, respectively, adjusting for plan characteristics and covariates described above, as well as antipsychotic PDC levels in 2008. To examine the association between patterns of SGA use and nonpersistence in 2009, we used Cox proportional hazard models, adjusting for the same covariates.

To examine changes in monthly adherence levels by SGA use patterns, we used linear fixed effects (within-person) regression models to estimate changes in monthly PDC relative to the month prior to generic introduction (June 2008). These models are robust to potential measured and unmeasured time-stable confounders; we adjusted for time-varying covariates, including calendar month, beneficiaries’ Part D risk scores, and tier 2 versus tier 1 co-payment differentials.

Among all subjects, 24.3% used brand risperidone prior to generic introduction in July 2008. Overall, 41.3% had schizophrenia or bipolar disorder, 46.1% were disabled, and about 35.4% and 57.1% were in HMO and private FFS plans, respectively (

). Mean cost-sharing levels were approximately $3 (tier 1) and $25 (tier 2).

 presents the cumulative percent of beneficiaries who used generic risperidone between July 2008 and December 2009. Among those using brand risperidone prior to July 2008, 67.7% used generic risperidone by December 2008 and 73% by December 2009. Among those using other brand SGAs prior to July 2008, 6.7% used generic risperidone by December 2009.

In multivariate analyses, greater cost-sharing differentials for generic versus brand drugs and enhanced Part D coverage were not significantly associated with generic use rates (

). In contrast, enrollment in private FFS versus HMO plans was associated with lower rates of nonbioequivalent generic substitution (hazard ratio [HR], 0.73; 95% CI, 0.56-0.96); there were no significant differences in bioequivalent generic substitution rates.

Compared with beneficiaries who continued using the same non-risperidone SGA before and after generic introduction, those who switched to risperidone in 2008 (nonbioequivalent generic substitution) had similar 2009 total spending, but lower out-of-pocket antipsychotic spending (difference = —$214; 95% CI, –$314 to –$115). In contrast, those who switched to a different brand SGA in 2008 (nonbioequivalent brand switching) had higher total and out-of-pocket spending in 2009. Beneficiaries who switched from brand to generic risperidone (bioequivalent substitution) had lower total antipsychotic spending (difference = –$943; 95% CI, –$1,042 to –$843) and out-of-pocket spending (difference = –$445; 95% CI, –$485 to –$404).

Antipsychotic Adherence and Nonpersistence

Beneficiaries who switched antipsychotics, either to generic risperidone or other brand SGAs, had higher odds of adherence in 2009 versus those who remained on the same brand SGA: odds ratio (OR), 1.72 (95% CI, 1.48-2.00) for bioequivalent generic substitution; OR, 2.34 (95% CI, 1.62-3.40) for nonbioequivalent generic substitution; OR, 1.43 (95% CI, 1.04-1.96) for nonbioequivalent brand switching (

). These beneficiaries also had lower rates of nonpersistence in 2009, defined as the time to first gap in SGA use >30 days, compared with those who continued using the same brand SGA in 2008.

 presents within-person changes in the PDC by antipsychotics in each month in 2008 and 2009 relative to June 2008—the month prior to generic introduction. Declines in use toward the end of each calendar year are consistent with the pattern of an increasing number of beneficiaries reaching the coverage gap and losing drug coverage. Those with nonbioequivalent generic substitution had the smallest relative reductions in SGA use by the end of 2009 compared with other beneficiaries (—2.6 percentage points in PDC; 95% CI, –8.3 to 3.0), including those who switched to a different brand SGA (–15.7 percentage points; 95% CI, –20.4 to –10.9); those who stayed on the same brand SGA throughout 2008 had the largest reductions in use by the end of 2009 (–27.5 percentage points; 95% CI, –28.9 to –26.1).

This study focused on the effects of the introduction of the first widely used SGA, risperidone, in Medicare Part D. We examined the levels of generic risperidone use in the first 18 months of introduction among beneficiaries receiving ongoing SGA drug therapy. Fewer than 1 in 4 beneficiaries used generic risperidone by the end of the study period, and the vast majority of use was among those previously using brand risperidone. Generic risperidone use was associated with lower total and out-of-pocket antipsychotic spending, as well as higher levels of adherence and persistence to SGA therapy.

Articles

This study examines the plan characteristics associated with generic risperidone use and spending and adherence outcomes associated with use among Medicare Advantage Part D beneficiaries.

The Introduction of Generic Risperidone in Medicare Part D

To understand when patients use secure e-mail messaging with healthcare providers across several types of questions or concerns, associations between out-of-pocket costs for in-person visits and use of secure messaging, and to examine patient-reported impacts on care-seeking behavior and overall health.

 Cross-sectional survey of patients in an integrated healthcare delivery system, with access to a patient portal to send secure e-mail messages to providers at no out-of-pocket cost.

 The study included patients with a chronic condition (N = 1041). We described patient-reported preferences for contacting providers and patient-reported impact of e-mail use on phone calls, in-person visits, and overall health. We used multivariate analyses to examine patient characteristics associated with using e-mail as a first contact method, and effects on care-seeking and health.

Overall, 56% of patients sent their provider an e-mail within 1 year, and 46% reported e-mail as their first method of contact for 1 or more types of medical concerns. After adjustment, higher out-of-pocket costs for in-person visits were significantly associated with choosing e-mail as a first method of contact (

 <.05). Among patients who had e-mailed their provider, 42% reported that it reduced their phone contacts, 36% reduced in-person office visits and 32% reported e-mailing improved their overall health.

 Patients reported using e-mail broadly to initiate conversations with their providers, and patients with higher out-of-pocket costs for in-person visits were more likely to choose e-mail as a first contact method. Use of secure e-mails reduced patients’ use of other types of healthcare and resulted in improved overall health.

	In a survey of patients with a chronic condition and with access to send secure e-mail messages to providers, nearly half had used e-mail as their first method of contacting providers for various types of health concerns. 

		Patients with higher out-of-pocket cost-sharing for visits were significantly more likely to report e-mail as their first method of contact with a health concern. 

		More than 1 in 3 patients who sent an e-mail to providers reported that it reduced their phone contacts or office visits. 

		Nearly one-third of patients who sent an e-mail to providers reported that it had improved their overall health.

Secure electronic messages offer patients and physicians an additional channel for communication and may serve as a unique mechanism for healthcare delivery. Although recent estimates show that most physicians in the United States are not yet regularly communicating with their patients electronically, most patients indicate an interest in communicating directly with their healthcare providers online.

 In order to receive the second stage of federal incentive payments for “meaningful use” of electronic health records and to avoid eventual financial penalties, clinicians will need to offer patients access to their health information through Web-based tools and to exchange secure electronic messages.

 However, the effects of these policy-driven shifts toward more electronic information and communication on patient care-seeking decisions and healthcare utilization are unclear.

There is limited information on what types of health concerns patients discuss with their providers using secure messaging, subsequently conflicting with evidence on the ways in which having Web-portal access affects in-person care-seeking behavior. Additionally, there is little attention paid to how patients’ cost-sharing for in-person visits affects their choice to contact their providers by e-mail.

 To understand how patients report their use of secure messaging tools to discuss concerns or questions about their health with their healthcare providers, we surveyed patients in a large health system that offers patient—provider secure electronic messaging tools, focusing specifically on patients with a chronic condition. We examined patient preferences for contacting healthcare providers across a variety of types of questions and concerns, whether cost-sharing for in-person visits was associated with this decision, and the patient-reported impact of secure message use on their in-person visits and overall health. We hypothesized that preferences for first contact method would vary by type of health concern, and that they would be affected by out-of-pocket costs.

Kaiser Permanente Northern California is an integrated delivery system with more than 3 million members. Health system members who register to use the password-protected patient portal website are able to exchange secure electronic messages directly with a provider in their healthcare team. Patients and providers are each notified when they have received a new message. There is no charge to the patient for using the patient portal, which also offers patients the ability to view lab results, request medication refills, and to view portions of their health records and visit summaries. Patients in this health system can also schedule nonemergency office visits through the portal website or by telephone.

The Kaiser Foundation Research Institute Institutional Review Board reviewed and approved the study protocol, waiving the requirement for informed consent.

Our source population for this study included all adult patients (18 years or older) who were in at least 1 of the health plan’s clinical chronic disease registries for asthma, coronary artery disease, congestive heart failure, diabetes, or hypertension during 2010. We studied patients with chronic conditions in order to focus on patients with at least 1 condition to discuss with a provider, but our survey asked generally about patient—provider messages for any type of condition or concern. Because the survey included questions about the impact of patient out-of-pocket costs (eg, co-payments and deductibles) on care-seeking behavior, we limited our sample to patients who were continuously enrolled through an employer-sponsored health insurance plan with no cost-sharing changes in their health plan benefits during 2010.

To ensure a sufficient number of participants who had recently sent a secure message and who had higher out-of-pocket costs for in-person care, we used a stratified random survey sample. Using automated records from the 12-month period before our study began, we randomly selected 25% of our sample from those who had not sent any secure messages and 75% from those that had sent at least 1 secure message. We also stratified the sample to include 25% from patients with a high-deductible health plan and 75% from patients without a high-deductible health plan. The high-deductible plans in our study met the IRS requirements for health savings account eligibility, including a deductible of $1200 or higher for an individual ($2400 for family coverage) that applied to most healthcare services, including nonpreventive office visits.

Beginning in June 2011, we mailed each potential participant a study introduction letter, a reply postcard, and a questionnaire with a postage-paid return envelope. We offered the option of completing the survey questionnaire by mail, by Web-based electronic survey, or by telephone interview in order to include of as many respondents as possible regardless of technology access or preferences, and in order to maximize response rates. Between June and December of 2011, trained interviewers contacted those who had not responded to our initial mailing in order to conduct telephone interviews, attempting to reach potential participants during different times of the day on weekdays and weekends. As needed, interviewers also called respondents who had mailed the written survey to complete and clarify any missing items. At the end of our data collection period, we again mailed a copy of the survey with a prepaid return envelope to all potential participants we had not been able to reach. All study participants received a $5 coffee gift card.

Of the total 1314 potential respondents contacted: 117 could not be reached after 15 phone call attempts, 183 were ineligible for study participation (a language barrier or health problem prevented them from completing an English-language interview or survey or they could not be reached due to incorrect contact information), and 1041 patients completed the study questionnaire (the response rate among eligible participants was 79%). Among all respondents, 51% completed the survey by telephone, 34% returned the survey by mail, and 15% completed the Internet-based survey. Comparing respondents with nonrespondents, women and respondents older than 65 years were more likely to complete the survey (

The survey questionnaire asked respondents how often they use the Internet for any reason and what devices they use to access the Internet. We also asked participants to report if, in the previous 12 months, they had any of 5 types of health concerns: 1) questions about a medical test result, 2) questions about a new health condition, 3) questions about an ongoing or chronic health condition, 4) questions about a medication, or 5) a request for a referral.  For each type of concern reported, we then asked respondents how they first contacted their provider or the healthcare system (ie, telephone call for advice or to schedule a visit, e-mailing their provider, visiting the emergency department, or no contact at all). We grouped phone calls for advice and to schedule a visit together since the patients’ first method of contacting the health system was by phone.

If e-mail was the preferred method of provider contact for a given type of concern, we then asked respondents what other method they would have used to contact their provider if the option to send a secure e-mail message had not been available. Respondents who reported past use of secure messaging also answered questions about which provider they had e-mailed and how long it took to receive a response from the provider. We also asked whether using secure messaging affected the number of times they contacted their provider by phone or the number of in-person office visits, and if using the secure messaging tool to e-mail their provider had an impact on their overall health status.

We asked all respondents to report their health plan’s cost-sharing requirements for doctor’s office visits and several demographic characteristics, including education, annual household income, race/ethnicity, marital status, and self-reported health status. Using health plan administrative data, we identified participants’ age, gender and DxCG comorbidity score (diagnosis-based risk score).

Because our study used a stratified sample, all analyses and results presented were weighted using study sampling proportions to represent the overall source population of patients with chronic conditions. We describe participants’ access to the Internet, with participants who reported never or rarely using the Internet categorized as not accessing the Internet. We also calculated the percentages of respondents who reported the 5 specific types of concerns in the last 12 months, their preferred method of contact for each type of concern, and the percentage who had registered to use the patient portal website. Among those who were Internet users and had registered to use the portal, we calculated the percentage who reported using secure messaging with any healthcare provider in the last 12 months, the type of provider e-mailed, and the average time to provider response. Among those who e-mailed their provider, we calculated the percentages who reported that e-mails with their provider changed the number of calls or visits with providers, or had an impact on their overall health.

Among patients who reported having any of the 5 specific types of health concerns, we used multivariable logistic regression to examine the association between reporting a preference for using e-mail to contact providers for any type of concern and visit cost-sharing levels, adjusting for patient characteristics, including gender, race-ethnicity, age, marital status, education, self-reported health, income, and comorbidity score. All multivariate analyses account for the survey sampling strategy in the point estimates and variance estimation (Stata version 9.0 [

 LP, College Station, Texas] using the svlogit command). We also used multivariable logistic regression to examine the association between reporting a preference for using e-mail as the first method to contact providers and reported impacts on phone contact and office visit rates, with adjustment for patient characteristics. We categorized respondents as having high visit cost-sharing if they reported having a deductible that applied to office visits or out-of-pocket cost (eg, co-payments or deductibles) of $60 or higher. We then computed the adjusted percentage of respondents who reported using e-mail as a first method of contact by fitting results from the logistic regression model by each patient characteristic and reported cost-sharing levels for office visits.

 shows the characteristics of the 1041 participants in our study after weighting. Among the 71% of participants who access the Internet at least monthly, 79% reported sending at least 1 secure message to their provider in the past year (24% sent 1-2, 29% sent 3-5, and 26% sent >5 e-mails). 

 shows the types of providers they e-mailed and the reported provider response times.

Among patients who reported having each type of health concern, 

 shows the reported first contact method. After multivariate adjustment, patients who were female, who used the Internet more often, or who reported higher out-of-pocket cost for in-person visits (≥$60) were statistically significantly more likely to choose to e-mail their provider as a first method of contact (

). After adjusting for patient characteristics, 85% of patients with higher cost-sharing for in-person visits reported choosing e-mail as their first contact method compared with 63% of patents with lower cost-sharing.

Across the types of health concerns, 85% to 91% of those who reported e-mail as their first method of provider contact indicated that if e-mail was not available, they would instead have contacted their provider by phone for advice or to schedule a visit; 4% to 11% reported that they would have scheduled a visit through the health plan website. Less than 3% reported that they would have avoided contact with their provider altogether if e-mail had not been available, and less than 1% reported that they would have instead sought care at the emergency department.

Among all participants who exchanged a secure message with their provider, 

 shows the reported impact on office visits and phone calls with the health system. Overall, 42% reported that the number of times they contacted the health system by phone decreased because of e-mail exchanges with their provider, 36% reported that their number of office visits decreased because of e-mail exchanges with their provider, and 32% reported that being able to communicate via e-mail improved their overall health. In multivariate analyses adjusting for patient characteristics, participants who reported e-mail as their first method of contact for any type of concern were significantly more likely to report that using e-mail had decreased their number of phone contacts with health system by phone and the number of office visits they had (

 <.001). After adjustment, 50% of those who reported using e-mail as their first method of contact reported that it decreased the number of calls, 42% reported that it decreased the number of office visits, and 35% that it improved their overall health compared with 14%, 11%, and 22%, respectively, of those who did not have any preference for e-mail as a first method of contact.

In a survey of patients with a chronic condition within a healthcare setting that offers patients the ability to send secure electronic messages to their healthcare provider, we found that a large proportion of patients used e-mail as their first method of contacting providers across multiple types of health-related concerns. Patients with higher out-of-pocket cost-sharing for in-person visits were statistically significantly more likely to report e-mail as their first method of contact with a health question. While the majority of patients who e-mailed their provider reported that this did not affect their number of phone contacts or office visits, more than 1 in 3 reported that their phone contacts or office visits decreased because of e-mail exchanges with providers, and nearly one-third reported that exchanging e-mails with their provider had improved their overall health

Secure patient portals have been proposed by the Institute of Medicine as a promising method of decreasing medical errors and increasing healthcare quality.

 Although the majority of US adults value Web-based access to their medical records as important, in 2013, only 28% of Americans were offered access to their online medical record.

  However, Stage 2 of the Meaningful Use federal programs that offer payment through the Medicare and Medicaid programs for demonstrating “meaningful use” of electronic health records will require clinicians to access and exchange secure e-mails with patients. Still, although our study was conducted in an integrated delivery system, many other reimbursement arrangements may limit the use of patient—physician e-mail, and Medicare does not generally provide separate payment for secure message exchanges.

It is unclear how these broad changes in electronic data and communication access will change patient care-seeking behavior and outcomes. Our study finding of high rates of use of electronic messaging as the first method of contact for a range of health-related questions will likely become increasingly widespread as more patients and clinicians gain access to, and begin using, these tools regularly. Because patients in our study largely reported that they would call the health system for advice or to schedule a visit if e-mail had not been available, this initial contact shift likely has consequences for telephone communication and in-person utilization patterns as well. We found that one-third of those who e-mailed their provider reported that this communication decreased their use of in-person visits.

With the rapid growth of healthcare costs, health plan benefits that include high levels of patient cost-sharing (such as high deductibles), in order to curb demand for services, have been growing in popularity with policy makers and employers. High-deductible health plan enrollment will likely increase further with implementation of health reform in the Affordable Care Act.

 Several studies have demonstrated that cost-sharing, in various forms (eg, co-payments, deductibles), reduces utilization of medical services,

  but may lead to decreases in necessary care, including for chronic conditions.

 In addition to the cost-sharing associated with a doctor’s office visit, patients often face additional barriers and costs to seeking in-person care, such as difficulty getting time off from work or arranging transportation or childcare. The convenience of e-mail may reduce these barriers to accessing care. Also, our finding of higher rates of patients using secure messaging as their first method of contact when facing higher out-of-pocket costs for in-person care may signal changes in patients’ care-seeking patterns that could become increasingly common as patient cost-sharing levels continue to grow and the ability to communicate with providers electronically becomes more widespread.

Existing evidence on the impact of patient use of patient portal websites on healthcare use and health outcomes has been limited and the results are conflicting.

 Previous studies do not capture the patient-reported choice to use electronic tools relative to more traditional in-person types of healthcare services; it is challenging to identify switches between types of services or delays in healthcare service use using administrative data sources. We uniquely captured patients’ own reports of initial methods for contacting their healthcare providers according to type of health concern and patient-reported impacts of e-mail on visits and health. Some patients explained that issues such as lab orders, medication changes, and patient questions could be addressed by e-mail and could reduce the need for an in-person visit. In addition, patients noted that greater information availability and engagement in their own care improved the management of their conditions and overall health. Overall, we found that patients regularly used secure messages to initiate conversations with their providers across a wide range of types of health questions and concerns and many report that these exchanges improved their health.

The direct generalizability of our findings is limited to the populations and delivery system setting of our study; however, since we studied patients’ self-initiated e-mail exchanges with their providers, our findings may reflect more general patient care-seeking behavior. Also, while our study is based on patient-reported survey data, automated clinical data sources are unlikely to capture patients’ initial health concerns and care-seeking decisions that we present. These patient-reported findings do complement previous studies, in the same integrated delivery system, in which providers report that electronic tools and messaging facilitated care coordination, by improving communication and providing informational continuity.

 Additionally, although our study focused only on out-of-pocket charges for healthcare services such as co-payments and deductibles, patients may have also had other unmeasured personal expenses for transportation, lost wages, childcare, etc, associated with any healthcare encounter. Overall, the study data are cross-sectional; thus, these findings are not designed to establish causality. Future studies should continue to examine the impact of patient—provider e-mail use on healthcare-seeking behavior, clinical care delivery work flow, and patient outcomes.

We found that patients with chronic conditions sent secure e-mail messages to their providers as a first method of contacting the healthcare delivery system for a wide variety of health concerns, and that higher out-of-pocket costs for in-person visits were associated with use of e-mail. Nearly one-third of respondents who used this tool reported that e-mailing their healthcare providers had improved their overall health, and more than 1 in 3 reported that e-mailing their provider decreased their phone calls to their providers or their in-person doctor’s office visits. As more patients gain access to patient portal tools associated with an electronic health record, patient-provider e-mails may shift the way that healthcare is delivered and have the potential to impact efficiency, quality, and health outcomes.

 Kaiser Permanente Division of Research (MR, IG, NG), Oakland CA; University of Tennessee Health Science Center (IG), Memphis TN; Mongan Institute for Health Policy, Massachusetts General Hospital (VF), Boston MA.

 Kaiser Permanente Community Benefit Fund.

 The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (MR, NG, IG, VF); acquisition of data (MR, IG); analysis and interpretation of data (MR, NG, IG, VF); drafting of the manuscript (MR, NG, IG); critical revision of the manuscript for important intellectual content (MR, IG, VF); statistical analysis (MR, IG, VF); provision of patients or study materials (MR, IG); obtaining funding (MR, IG); administrative, technical, or logistic support (MR, IG). 

 Mary Reed, DrPH, Kaiser Permanente Division of Research, 2000 Broadway, Oakland, CA 94612. E-mail: Mary.e.reed@kp.org.

Patients with higher out-of-pocket visit costs, such as co-payments and deductibles, initiated provider contact by e-mail more often. Patients report that e-mailing changed their care-seeking behaviors and improved overall health.

Patient-Initiated E-mails to Providers: Associations With Out-of-Pocket Visit Costs, and Impact on Care-Seeking and Health

 With the introduction of Part D drug benefits, Medicare began to collect information on diagnoses, treatments, and clinical events for millions of beneficiaries. These data are a promising resource for comparative effectiveness research (CER) on treatments, benefit designs, and delivery systems.

 To explore the data available for researchers and approaches that could be used to enhance the value of Medicare data for CER.

 Using currently available Medicare data for CER is challenging; as with all administrative data, it is not possible to capture every factor that contributes to prescribing decisions and patients are not randomly assigned to treatments. In addition, Part D plan selection and switching may influence treatment decisions and contribute to selection bias. Exploiting certain program aspects could address these limitations. For example, ongoing changes in Medicare or plan policies and the random assignment of beneficiaries with Part D low-income subsidies into plans with different formularies could yield natural experiments.

Refining policies for time to data release, provision of additional data elements, and linkage with more beneficiary level information would improve the value and usability of these data. Improving the transparency and reproducibility of findings, and potential open access for qualified stakeholders are also important policy considerations. Data needs must be reconciled with current policies and goals.

 Medicare data provide a rich resource for CER. Leveraging existing program elements, combined with some administrative changes in data availability, could create large data sets for evaluating treatment patterns, spending, and coverage decisions.

	Medicare now collects information on diagnoses, treatments, prescription drug use, and clinical events for millions of beneficiaries. These data are a promising resource for comparative effectiveness research (CER) on treatments, benefit designs, and delivery systems; however, there are a number of challenges to using these data for CER.

		 We explore the data available for researchers and approaches that could be used to enhance the value of Medicare data for CER.

		 Leveraging existing program elements, combined with some administrative changes in data availability, could create large data sets for evaluating treatment patterns, spending, and coverage decisions.

	Information on the safety, effectiveness, and value of medical care requires detailed clinical data from large numbers of patients receiving care in real-world settings. With the introduction of Medicare Part D prescription drug benefits in 2006, Medicare began to collect information on the use of prescription drugs for more than 27 million beneficiaries.

 Previously, this information was not widely available for Medicare beneficiaries, although drug use data were available for dual-eligible Medicare beneficiaries via Medicaid claims. The more comprehensive collection of drug use data allows for linkage with previously available information from Parts A and B including

	inpatient and outpatient diagnoses, and major clinical events for millions of beneficiaries, including many persons over the age of 65 years. These data provide a promising resource for assessing the comparative effectiveness of many types of care across a range of settings and geographic areas in the United States.

	Because Medicare collects data for payment and administrative purposes and not for research, there are several limitations to using these observational data. The care beneficiaries receive will vary depending on where they live, the types of Medicare plans they choose, and their physicians and hospitals. In other words, there may be factors associated with both the care beneficiaries receive and the outcomes of care; these factors confound assessments of care effectiveness and limit the validity of simple comparisons. However, exploiting certain program aspects and ongoing natural experiments within the Medicare program can mitigate some biases associated with purely observational data.

	In this review article, we discuss these strengths and limitations of using Medicare data for comparative effectiveness research (CER) and propose policy recommendations for improving the usefulness of these data for patients, providers, and policy makers.

	There is a profound need for more evidence to guide clinical and policy decisions on drug treatments, devices, interventions, care delivery, payment models, and delivery systems. For example, while there is arguably substantial trial evidence supporting the use of many prescription drugs, this evidence often provides limited guidance for actual clinical decisions.

	There are 2 main approaches for developing comparative effectiveness evidence: (1) clinical trials, including randomized clinical trials (RCTs) and pragmatic trials, and (2) studies using observational data from actual practice. While doubleblinded RCTs represent the gold standard for generating clinical evidence, they have a number of practical limitations. Specifically, trials have historically compared single drugs with placebo rather than with existing alternative drugs, or examined them in combination with commonly used drug regimens. Trials are done under rigorous experimental conditions (efficacy) rather than real-world situations (effectiveness), and are not designed to evaluate costs or rare adverse events. Randomized clinical trials also tend to be expensive and examine relatively short-term effects. Moreover, trials may have limited generalizability to specific subgroups of patients (eg, the elderly, racial/ethnic minorities, those with severe diseases) because they tend to target relatively homogeneous patient groups rather than the broader mix seen in actual practice.

Pragmatic trials attempt to overcome some of these limitations by focusing on more heterogeneous groups of patients and by evaluating effectiveness under routine care; however, existing evidence from pragmatic trials is in short supply and funding for these types of studies is limited.

	Studies using observational, longitudinal data sets could provide complementary information that addresses many of the limitations associated with RCTs. For example, Medicare collects information on millions of beneficiaries and allows for linkages across a range of claims data, including inpatient, outpatient, and prescription drug data. Having a large sample of individuals is critical for ensuring adequate statistical power for studying rare conditions or specific patient subgroups. Use of observational data (including Medicare data), however, requires consideration of numerous factors such as the types of Medicare plans for inpatient, outpatient, and drug services; coverage/cost sharing for treatments; availability of physicians and hospitals; and clustering of patients by physician. In addition, there can be variations in practice patterns across geographic areas. Examinations of drug use within Part D should consider these various levels of analysis and account for factors

	that could affect drug use, adherence, and ultimately outcomes, including a range of patient, provider, and plan-level characteristics. In short, the strengths and limitations of both clinical trials and observational data analyses should be considered carefully when evaluating the value of these approaches for addressing specific questions.

	The following sections describe relevant structural aspects of the Medicare Part D program, the Medicare data available for researchers, and potential approaches that could be used to create quasi-experiments and enhance the value of historical Medicare data for CER.

	Medicare currently collects diagnostic and treatment information through 4 programs: Part A (inpatient), Part B (outpatient), Part C (Medicare Advantage, which includes medical information for beneficiaries enrolled in managed care organizations), and Part D (prescription drugs). Part D is administered by private plans either as stand-alone Prescription Drug Plans (PDPs) that supplement traditional Medicare or Medicare Advantage Prescription Drug (MAPD) plans that bundle Part A, B, and D benefits. Part D is a voluntary benefit; in 2009 about 27 of 45 million Medicare beneficiaries were enrolled in a Part D plan, including 9.6 million lowincome beneficiaries who received additional premium and cost-sharing subsidies from Medicare. The Centers for Medicare & Medicaid Services (CMS) randomly assigns low-income subsidy beneficiaries who have not chosen a Part D plan to qualified stand-alone drug plans.

	Beneficiaries choose their Part D plans; these plans have some autonomy in determining their benefit structures and formulary drug lists provided they meet basic Medicare requirements. For example, all plans must offer benefits at least as generous as the defined standard. Medicare also requires that plans cover at least 2 drugs within a therapeutic class. However, plans can determine coverage for specific drugs within a class, as well as tier placement and utilization management requirements.

	The use of utilization management tools such as prior authorization has grown since Part D’s introduction; these tools are most often used for drugs that are newer, more expensive, or more risky, with greater potential for adverse effects or with less available evidence on the possible benefits or harms.

MEDICARE DATA CURRENTLY AVAILABLE FOR RESEARCH

	Since the introduction of the drug benefits, Part D plans have submitted detailed information on prescription drug events for all Part D beneficiaries to CMS. Researchers can apply for access to Research Identifiable Files that include beneficiarylevel information on Part A, B, and D claims. Proposals are reviewed by the CMS privacy board; if approved, researchers must sign a Data Use Agreement specifying the terms of use, including the destruction or return of data at the study’s end.

 The final rule permitting release of the newly available Part D data to researchers was issued in May 2008. In addition to usual protections for beneficiary privacy, this rule includes additional protections for commercially sensitive plan information. Since the initial release of Part D data, CMS has rolled out an increasing number of data elements and linkages, and is continuing to assemble supplemental data files.

 outlines the available Part D research data files.

 The primary data source for Part D drug utilization is the Part D Event (PDE) files, which are currently available for

	2006-2008. These files contain detailed information on each drug event for PDP and MAPD plan beneficiaries, and encrypted beneficiary, pharmacy, prescriber, and plan identifiers that allow linkage with other files such as inpatient and outpatient claims data, and Part D plan characteristics files. The PDE contains information on each drug dispensed including the National Drug Code, the quantity dispensed, and days of supply, allowing for the examination of therapy adherence and persistence based on dispensing data, which have been previously validated.

	The PDE data also capture cost information such as total drug costs and patient payments. These data allow for examination of variation in spending patterns and cost-of-care analyses. The PDE data also specify the benefit phase during which each prescription was filled (eg, deductible or initial coverage phase) based on the benefit structure implemented by each beneficiary’s plan, which affects costs for both patients and payers. This file also includes plan-specific information on the formulary coverage for each drug dispensed, including the tier and utilization management requirements. Plan-level information on formulary and benefit structures can be valuable for identifying quasi-experiments or instrumental variables for statistical analyses.

	The PDE data can be linked with information on individual beneficiary characteristics. The Beneficiary Summary File contains beneficiary-level information on basic demographics, including age, sex, race, and geographic location. This file also describes beneficiaries’ months of enrollment in Parts A, B, C, and D, including the type of coverage (eg, retiree, Part D stand-alone plan, Part D Medicare Advantage plan), dual-eligibility status, and whether they are receiving the Part D low-income subsidy. The Beneficiary Annual Summary File contains additional information on patients’ inpatient diagnosis-related groups, as well as 2 sets of chronic condition flags.

	Since the Part D data rule was originally issued, CMS has rolled out an increasing number of data elements that supplement the PDE and beneficiary data (Table 1). The Drug Characteristics File includes drug names (generic and brand) and strength and dosage form information by the National Drug Code. Plan characteristics files can be linked to the PDE data using the encrypted plan identifier to examine detailed information on plan type, cost-sharing levels, premium information, and service area. The Pharmacy Characteristics File includes information on the type of pharmacy where beneficiaries filled their prescriptions. Lastly, the Prescriber Characteristics File contains information on the prescriber’s specialty, credentials, and geographic location.

	Part D data can also be linked to traditional Medicare data on beneficiaries’ other medical claims (eg, hospital, skilled nursing facility, hospice, physician). For example, the Inpatient Standard Analytic File includes claims for inpatient stays, including diagnosis and procedure codes (

International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM]

), Diagnosis Related Groups (a classification system used for prospective Medicare payments), date, facility, and cost information. The Carrier File contains claims for noninstitutional providers, largely physicians. It also contains diagnosis and procedure codes (

 and CMS Healthcare Common Procedure Coding System codes), reimbursement amounts, and provider identifiers.

	Information at the beneficiary level on inpatient and outpatient service use and diagnoses within the Medicare Advantage program is not currently available to researchers.

	Monthly enrollment information is available in the Beneficiary Summary File so beneficiaries’ transitions between traditional fee-for-service Medicare and Medicare Advantage can be tracked. Prior to the introduction of Part D, beneficiaries could switch between traditional Medicare and Medicare Advantage plans monthly. In 2006, beneficiaries were restricted to switching only during the first 6 months of the year, and from 2007 onward, the first 3 months of the year.

	Medicare claims data also can be linked with Medicaid data for dual-eligible beneficiaries, as well as data sets focused on specific subpopulations or collected via surveys, such as the Surveillance Epidemiology and End Results Cancer Registry, the Long Term Care Minimum Data Set, the Home Health Outcome and Assessment Information Set, the Medicare Current Beneficiary Survey, and the Health and Retirement Survey. While the available sample size is more limited when linking to these data sources, they can provide a richer set of sociodemographic, health, and clinical characteristics than is available in the claims data alone.

	A substantial body of literature describes the limitations of using observational data for CER, as well as potential strategies for addressing these limitations.

 Data and measurement quality, the formation and stability of comparison groups, and methods to deal with crossover effects or switching between therapies are particularly relevant for observational data studies and have been described elsewhere.

	Despite the range of information available on Medicare beneficiaries’ drug treatments and diagnoses, simple analyses using currently available Medicare data are likely to yield biased results (

). Because patients are not randomly assigned to treatments, there may be important differences between the comparison groups with respect to the factors that contributed to patients receiving different treatments or therapies, such as disease severity, risk factors, and comorbidities (eg, confounding by indication).

 The lack of randomization creates analytic challenges because it is not possible using administrative data to capture all factors that contribute to physicians’ prescribing decisions.

	Plan selection and switching within the Part D program beneficiaries also create challenges. Beneficiaries choose to enroll in plans with varying characteristics that can influence treatment decisions, such as different delivery systems (eg, Medicare Advantage vs PDP), benefit designs, and formularies. Beneficiaries are encouraged to choose plans basedon these characteristics to minimize treatment disruptions and lower out-of-pocket drug costs, so there is likely to be subst antial self-selection into Part D plans.

 For example, there was evidence of substantial adverse selection into plans that offered full coverage during the standard coverage gap in 2006, prompting these plans to leave the market the following year.

 In addition, the availability of employer-sponsored drug coverage for current or retired employees also influences the characteristics of beneficiaries choosing to enroll in PartD plans; an estimated 8.3 million beneficiaries received retiree drug coverage in 2010. Beneficiaries can also switch to and from plans during the open enrollment phase of each year (eg, based on changes in clinical need). Beneficiaries’ plan choice and geographic location may have important implications for their access to different treatment alternatives.

 While a number of patient, provider, and plan characteristics are available in Medicare data sets, simply adjusting for these factors in a multivariate model is unlikely to sufficiently address selection differences between patient groups that contribute to treatment decisions, plan choice, or health outcomes.

Structural Aspects of the Medicare Program

This review article explores the Medicare data available for researchers and approaches that could be used to enhance the data%u2019s value for comparative effectiveness research.

Using Medicare Data for Comparative Effectiveness Research: Opportunities and Challenges

	Medicare bases its risk adjustment method for Medicare Advantage plan payment on the relative costs of treating various diagnoses in traditional Medicare. However, there are many reasons to doubt that the relative cost of treating different diagnoses is similar between Medicare Advantage plans and traditional Medicare, including the varying applicability of care management methods to different diagnoses and the varying degrees of market power among suppliers of services to plans. We use internal cost data from a large health plan to compare its cost of treating various diagnoses with Medicare’s reimbursement. We find substantial variability across diagnoses, implying that the current risk adjustment system creates incentives for Medicare Advantage plans to favor beneficiaries with certain diagnoses, but find no consistent relationship between the costliness of the diagnosis and the difference between reimbursement and cost.

	The Centers for Medicare and Medicaid Services—Hierarchical Condition Categories (CMSHCC) system that Medicare uses to reimburse health plans establishes relative prices for different diagnoses based on fee-for-service system data. This makes the implicit assumption that health plans reduce costs equiproportionately across diagnoses. This article tests that assumption.

		We overwhelmingly reject that the relative cost of diagnoses in the health plans in our sample is the same as that in the CMS-HCC.

		The magnitude of the errors is large, exceeding 100% for some HCCs.

		This is particularly true for uncommon conditions for which certain providers have substantial market power but Medicare uses lower administered prices.

	For a quarter of a century, Medicare has risk adjusted its payments to private plans that accept at-risk contracts and participate in Part C of the Medicare program, now known as Medicare Advantage (MA). Risk adjustment means Medicare pays plans more for enrollees who are expected to use more services and less for enrollees who are expected to use fewer services, thereby better matching enrollee reimbursement to expected use and, most important, minimizing plan incentives to select against the sick. For example, Medicare pays more for a beneficiary with cancer than for a beneficiary with no chronic disease.

	Medicare’s initial risk adjustment system, introduced in 1985, accounted for only the enrollee’s age, sex, county of residence, institutional status, Medicaid eligibility (for the noninstitutionalized), and whether a working beneficiary had employment-based insurance that was primary. Although Medicare paid plans more for enrolling older beneficiaries than younger beneficiaries, reflecting their greater medical spending, the risk adjustment system at that time included no direct measures of health status.

	Without adequate risk adjustment, a plan could make money by attracting low-cost healthy beneficiaries with a tightly managed low-premium product, whereas a plan enrolling high-cost beneficiaries would not be paid commensurately more and could lose money.

 In particular, if expected profitability varies with the health status of the beneficiary, plans have incentives to manage care for different services more or less aggressively. Services that are predictable by the beneficiary (and can be used as a device for selection), that are predictive of total medical costs, and that are less profitable are the services a plan would cut back on or manage aggressively to deter enrollment by enrollees who want those services, and conversely.

 provide evidence that service-level selection is operative in Medicare.

	Because plan reimbursement was set by Congress in 1985 at 95% (later 100%) of predicted spending in traditional Medicare (TM) in the beneficiary’s county, the enrollment of those whose utilization was less than this amount resulted in increased overall spending by Medicare.

 For example, the Congressional Budget Office

 estimated in 1994 that Medicare was paying 8% more to private plans than it would have paid had the same beneficiaries instead been enrolled in TM. In 2006, Medicare changed from paying plans a take-it-or-leave-it price to a bidding system, but the bids were for an average risk mix, and the actual monies paid to the plans continued to be adjusted for the risk mix of the plan’s actual enrollees. Therefore, to the degree the risk adjustment system remains inadequate, the MA program continues to be vulnerable to selection problems.

	In response to recommendations of the Prospective Payment Assessment Commission and the Physician Payment Review Commission (the predecessors of the Medicare Payment Advisory Commission), the Balanced Budget Act

 mandated that Medicare introduce a risk adjustment method that would account for the beneficiary’s health status in addition to demographic variables, such as age and sex, that it already used. In response, the Centers for Medicare and Medicaid Services (CMS [née Health Care Financing Administration]) contracted for the development of a risk adjustment method that became known as CMS—Hierarchical Condition Categories (CMS-HCCs). In addition to most of the demographic factors that the prior risk adjustment system had used, this method also adjusted plan reimbursement on the basis of diagnoses recorded on claims from the prior year.

	In 2000, an initial version of this method, based solely on diagnoses from inpatient claims, was implemented.

 This initial version was used through 2003, but to avoid an incentive to hospitalize a beneficiary simply to record a diagnosis and obtain higher reimbursement, the method initially applied only to 10% of the reimbursement to plans; the remaining 90% was based on the prior system that did not use diagnostic data. Starting in 2004, diagnostic data from outpatient claims were considered reliable enough to use to reimburse, and a transition began such that the new system of CMS-HCCs, which used diagnoses recorded in both the inpatient and outpatient settings, by 2007 applied to 100% of the Part C payment.

 Although the developers initially distinguished 189 categories of conditions, the final set of CMS-HCCs that CMS implemented had only 70 categories, which balanced concerns about coding feasibility, adequate sample size, and predictive accuracy.

	The principal task of the new risk adjustment method was to distinguish how much more or less expensive beneficiaries with a given diagnosis or diagnoses were to treat (in terms of total Medicare-covered services) relative to beneficiaries with other diagnoses. A key problem that CMS and its contractor faced in answering this question was that the data to determine the relative costliness of those with various diagnoses were available only from TM. Using data on relative costs in TM to pay for treatment in MA presumes that the relative cost of diagnoses in the managed system has the same structure as TM (ie, any reduction in cost from care management is equiproportionate across all diagnoses or CMS-HCCs).

	The equiproportionality assumption is unlikely to hold for several reasons. We have known for many years that between-area variation in rates of procedures is related to the degree of physician discretion in treating the condition.

 For diagnoses with less discretion in choice of treatment, there is less variation across areas. This finding seems likely to carry over into managed care as well, meaning that diagnoses with little discretion will be treated similarly to TM and those with more discretion may be treated in a less costly fashion. For example  plans could selectively contract with those physicians and hospitals that are more conservative in their management of conditions with more physician discretion or could place them in a favorable tier such that patients would pay less to use them.

	In addition, health plan interventions in the care process, such as integration of care, utilization management, and prior authorization, almost certainly affect different conditions differently. For example, in the case of chronic problems, long-term compliance with a treatment regimen is often important, and health plans can influence compliance with prescribed medication for a chronic disease, such as diabetes mellitus, through benefit design and disease management.

 Such interventions are unlikely to be appropriate or used for emergency treatment, such as that of an aortic dissection. In general, acute problems tend to be less amenable to care management than chronic problems because rapid action by the physician on the scene may be important in treating an acute problem and no intervention by a third party may be feasible.

	Furthermore, the number and effectiveness of potential treatment options vary across medical conditions, limiting the ability of any plan to achieve gains through more efficient care or better contracting. At one extreme, if there is only a single treatment option available nationally from a single manufacturer, the cost of this option may not vary substantially. At the other extreme, when there are multiple treatment options with comparable effectiveness (eg, diuretic prescription drugs), plans could obtain favorable prices through negotiation and encourage greater use of the highest-value option through care management programs. Even if there is only 1 standard treatment option, its cost may vary among local providers (eg, procedures performed at teaching vs nonteaching hospitals). Furthermore, if the degree of improvement in compliance that managed care can achieve and the resulting reduction in cost varies across chronic conditions, as is surely the case, the equiproportionate assumption would also fail.

	One dramatic, but dated, demonstration of differences in treatment patterns between managed and “unmanaged” care comes from the RAND Health Insurance Experiment,

 which randomized families of beneficiaries younger than 65 years to a staff model health maintenance organization (HMO) in the 1970s and others to an indemnity insurance plan like TM that had no utilization management features (but had cost sharing similar to the staff model HMO). Those assigned to the HMO had similar ambulatory visit rates but 40% lower hospitalization rates and, using the same set of price weights for specific services to impute spending, 28% less spending.

 Therefore, for diagnoses that do not customarily result in hospitalization, the CMS-HCC weights that one would have estimated using data on the cost incurred by the staff model HMO probably would have differed little from the weights estimated using data from those with indemnity insurance, whereas for many of the diagnoses with nontrivial proportions hospitalized, the weights would have differed substantially simply from the lower frequency of hospitalization.

	Finally, health plans contract with providers who have varying degrees of market power. Particularly for rare or unusual diagnoses, there may be only 1 local or even only 1 regional provider. That provider can probably obtain higher reimbursement from managed care plans relative to Medicare’s administratively set reimbursement compared with other diagnoses for which there are more numerous local providers, thus giving plans a stronger negotiating position.

	In short, the relative price structure that Medicare uses to reimburse at-risk MA plans almost certainly contains errors, because the cost structure of health plans across persons with various diagnoses likely differs substantially from that of TM. The aim of this study was to determine how different the structure of the risk adjustment scheme would be if it were based on health plan costs rather than data from TM and whether the scheme creates incentives to select the healthy and avoid the sick. We do not examine actual selection by plans or beneficiaries.

	We studied more than 300,000 Medicare beneficiaries enrolled in risk contracts in a large MA-HMO plan during 2006 and 2007. This MA-HMO insurer offered multiple types of benefit arrangements to its beneficiaries. The MA risk adjustment system does not adjust for differences in benefit arrangements between MA and TM, and neither do we. Data from 2005 provided prior year diagnoses for 2006. We included beneficiaries who had aged into Medicare eligibility by the beginning of the study year; we excluded the institutionalized because of small numbers. The 2006 cohort consisted of 322,237 persons, and the 2007 cohort comprised 336,507 persons, as the plan gained enrollment over this period.

	Per our protocol, we have masked the identity of the plan. As is generally the case, the hospitals and physicians in the MA-HMO we studied treat both Medicare and commercially insured enrollees; therefore, the mix of providers and capital equipment are configured to treat both types of enrollees. In other words, Medicare reimbursement is not the only influence on the choices of inputs and in turn on the cost structure by disease.

	We included all Medicare beneficiaries in the MA-HMO who were enrolled throughout the year before the study year and throughout the study year (2006 or 2007), as well as those who died during a study year. We allowed up to a 2-month gap in membership in a calendar year because we conjecture that such gaps represent data processing errors rather than true changes in membership status. However, we believe that any medical claims from this gap are included in our utilization data. For the decedents, we replicated the methods by Pope et al

; we annualized the decedent’s spending by multiplying by 12, divided by the number of months eligible, and then weighted the observation by the reciprocal of that ratio. We excluded the small number (3%-4%) who did not have coverage for both Parts A and B or who left the plan in the middle of the year. Excluding the nondecedents with less than a full year of eligibility avoided the issue of how best to annualize spending for beneficiaries who disenrolled because they left the service area or changed plans in the middle of a calendar year and had missing spending data.

	We obtained internal cost accounting data for the services the plan provided to beneficiaries; these data approximate (or are proportional to) total allowed charges. We then replicated the methods that Pope et al

used to derive the CMS-HCC weights, substituting the plan’s cost accounting data at the beneficiary level for the TM spending data that Pope et al used. We compared the resulting coefficients of the HCCs with the coefficients of the CMS-HCC from the model by Pope et al that CMS used to reimburse plans. To simplifycomparison of the distribution of MA-HMO values with the values in the study by Pope et al, we rescaled the distribution of MA-HMO values to have the same mean as the distribution of the values estimated from TM. Rescaling (ie, normalizing the weights) corrects for any factor that affects all HCCs proportionately.

	Because the MA-HMOs’ geographic distribution of beneficiaries differs from the national TM distribution that Pope et al used in their calculations, absolute spending levels (and hence absolute values of weights) could differ for reasons that are incidental to our purposes herein, such as variation in nominal wage levels. Moreover, it is well known that there are geographic differences in the treatment of various conditions, 

 an additional reason why there could be differences in relative costs by geography. We make no effort to reweight our data to match the TM geographic distribution but note that, from the point of view of Medicare policy, any geographic differences in relative costs simply add to any distortions in the current structure because the structure of the CMS-HCC is the same across all regions.

	We proceeded by following the specification by Pope et al.

 We regressed the annual accounting cost for each beneficiary in the MA-HMO sample on the following dummy variables: age and sex (24 categories), Medicaid-×-sex-×-disability (4 categories), HCC (70 categories), HCC-×-disability (5 categories), and HCC interactions (6 categories). This gave a total of 81 HCC-related coefficients that we compared with those in the study by Pope et al. After initial estimation, we constrained several of the MA-HMO coefficients such that categories with a higher ranking in the disease hierarchy would have at least as high predicted costs. Specifically, we constrained the coefficients of the following HCCs to be equal: hcc008 = hcc009, hcc067 = hcc068, hcc081 = hcc082, hcc107 = hcc108, hcc075 = hcc154, and hcc161 = hcc177. These coefficients should be monotonically ordered but were not in our sample, presumably because of small numbers. Our specification differed from that by Pope et al in a minor aspect: because we were unable to determine which of the older beneficiaries may have been eligible for Medicare before age 65 years because of disability, we did not estimate intercept terms corresponding to older beneficiaries who were disabled before age 65 years and those who were not. Specifically, Pope et al included interaction terms for originally disabled and sex (originally_disabled—×-female and originally_disabled–×male). Instead, our estimates effectively are the average intercept over all older persons. Because most older beneficiaries were not eligible before age 65 years for reasons of disability (about 15% of all Medicare beneficiaries are younger than 65 years at any point in time) and because those who were eligible because of disability rather than by becoming 65 years old are distributed throughout the HCCs, any bias from this difference should not materially affect our results.

	We compared our results for 2006 with the relative risk scores from the 2004 version of the CMS-HCC software, which CMS used for payment in 2006 and are based on the values by Pope et al.

 The values that CMS used in 2005 and 2006 were based on a 5% sample of 1999 and 2000 data from TM. We similarly compared our 2007 results with the results from the 2007 CMS-HCC software. As a descriptive comparison of the risk adjustment structure derived from the MA-HMO data with that derived from the TM data, we computed the percentage differences between the values derived from the MA-HMO data and from the TM data for those HCCs whose values in the MA-HMO data were estimated with a sufficient degree of precision (specifically, whose standard error was <20% of the absolute difference between the weights from TM and from the MA plans, divided by the mean of these 2 values). For this purpose, werescaled the MA-HMO values to have the same mean as the TM values over all 81 coefficients.

	We adopted the one-fifth cutoff for the descriptive statistic to represent a compromise between having enough HCCs to obtain a meaningful distribution of differences between MAHMO and TM values but also to be sufficiently precise that we did not show large deviations simply from sampling error, especially in the MA-HMO estimates; 46 HCCs in 2006 and 45 HCCs in 2007 satisfied this criterion.

	If the equiproportionate assumption held, these differences would bunch around zero. Some differences could arise from differences in relative input prices that led to a different mix of inputs in MA-HMO treatment; other differences could arise from varying distributions of the demographic variables or illness severity within HCCs between TM and the MA-HMO sample. There is also a negligible discrepancy because the means of the TM and MA-HMO distributions were adjusted to be equal across the 81 coefficients rather than across the 46 (in 2006) and 45 (in 2007) used in the descriptive analysis. Nonetheless, it seems unlikely that differences from these causes would be large. However, differences in treatment patterns owing to utilization management could be large. Some of those differences could arise from input price or demographic differences, but most of them probably arise from differences between the incentives that full capitation offers and those of an unmanaged fee-for-service reimbursement environment, as well as differences in market power in contracting with providers.

	As already noted, some differences between the 2 sets of weights would be expected from sampling error; even with more than 300,000 observations, some of the HCCs have only a few hundred observations in the MA-HMO data, and the values by Pope et al

 have a sample only about 6 times as large (Pope et al used a 5% sample of 2 years of data), so that sampling error is also relevant to the values by Pope et al. We describe herein a formal test across all the coefficient values of whether the relative price structures are similar.

 show the results of the percentage differences between the weights derived from TM and from the MAHMO sample for those HCCs in which the standard error of the MA-HMO values is less than 20% of the difference between the TM and MA-HMO coefficients (46 HCCs in 2006 and 45 HCCs in 2007). 

Health plan accounting data are used to test how well the CMSHCC risk adjustment system tracks relative costs of treating various diagnoses: not very well.

Vicki Fung, PhD

Articles

The Introduction of Generic Risperidone in Medicare Part D

Patient-Initiated E-mails to Providers: Associations With Out-of-Pocket Visit Costs, and Impact on Care-Seeking and Health

Using Medicare Data for Comparative Effectiveness Research: Opportunities and Challenges

The Structure of Risk Adjustment for Private Plans in Medicare

Medicare Part D Coverage Gap and Diabetes Beneficiaries