William B. Vogt, PhD

 Unwarranted geographic variation in spending has received intense scrutiny in the United States. However, few studies have compared variation in spending and surgical outcomes between the United States healthcare system and those of other nations. In this study, we compare the geographic variation in postsurgical outcomes and cost between the United States and Japan.

 This retrospective cohort study uses Medicare Part A data from the United States (2010-2011) and similar inpatient data from Japan (2012). Patients 65 years or older undergoing 1 of 5 surgeries (coronary artery bypass graft, abdominal aortic aneurysm repair, colectomy, pancreatectomy, or gastrectomy) were selected in the United States and Japan.

 Reliability- and case-mix—adjusted coefficient of variation (COV) values were calculated using hierarchical modeling and empirical Bayes techniques for the following 5 outcomes: postoperative mortality, the development of a complication, death after complication (failure to rescue), length of stay, and the cost of the hospitalization. Sensitivity analyses were also performed by calculating patient demographic-and case-mix–adjusted COV values for each outcome using weighted age- and sex-standardized values.

 The variability of the postsurgical outcomes was uniformly lower in the United States compared with Japan. Cost variation was consistently higher in the United States for all surgeries.

 Although the US healthcare system may be more inefficient regarding costs, the presence of higher geographic variation in postoperative care in Japan, relative to the United States, suggests that the observed geographic variation in the United States—both for health expenditures and outcomes&mdash;is not a unique manifestation of its structural shortcomings.

This investigation compares the amount of observed geographic variation in the United States and Japan for the surgical outcomes and costs of 5 common, higher-risk surgeries. Key findings include:

		The United States has less variation in surgical outcomes compared with Japan, but it has higher geographic cost variation.

		The surgical outcome variation in the United States is not unique; it exists even in a country that consistently ranks highly on public health quality measures. This variation is likely the result of unique structural inefficiencies in each healthcare system.

		There is an opportunity to reduce wasteful spending by standardizing the cost of care in the United States.

Countries around the world are attempting to contain increasing healthcare costs without sacrificing quality of care. Relative to other developed countries, the United States fares poorly on measures of both healthcare expenditure and health outcomes. Although the United States spent $7929 per person (16.4% of gross domestic product [GDP]) in 2010—the most of any Organisation for Economic Co-operation and Development (OECD) country&mdash;health outcomes, such as life expectancy, are among the worst in the developed world.

Much research has focused on small-area geographic variation in healthcare spending as evidence of inefficient healthcare in the United States.

 A recent report released by the Institute of Medicine (IOM) has affirmed that higher healthcare expenditures within a region are not related to better outcomes.

 In principle, the lack of relationship between spending and outcomes suggests that any unwarranted variation is potentially wasteful, and, if eliminated, might decrease costs while maintaining quality. Variations in healthcare expenditure have been identified as a potential area for up to a 30% reduction in waste (approximately $650 billion) in the US healthcare system.

Apart from issues of economic inefficiency, geographic variations in treatments and outcomes have important clinical and health policy implications. Outcome differences across regions are evidence for the presence of inequalities in care across regions, which represent a distinct and important problem in itself. Regional differences in quality of care may indicate opportunities to improve care for patients in places with worse outcomes. Identifying centers that have better outcomes than others and promoting the practice styles of those centers may be a practical way to improve patient outcomes.

Researchers have proposed several reasons for the observed geographic variation in the United States. Studies cite fragmentation of care,

 racial and socioeconomic barriers to care, and the diversity of financing and coverage options as reasons that the United States has unwarranted variation.

 Other studies have also investigated physician peer effects and the ways that social influences can affect medical decisions.

One premise underlying the interpretation of these findings is that the American healthcare system is unique in its inefficient stewardship of healthcare dollars. This naturally leads to a question: If the United States is so singularly ineffective at controlling costs and ensuring quality relative to other countries, then does it also experience singularly high geographic variation in costs and outcomes relative to other countries?

In this study, we compared the variability in postoperative outcomes and cost of hospitalization for 5 major surgeries in the United States and Japan. There are several key similarities and differences between the 2 countries. Japan is a modern OECD country that has access to technologies and treatments similar to those used in the United States; however, in 2010, Japan spent only $3204 per person (9.5% of GDP) on healthcare while attaining the highest average life expectancy at birth of any OECD country. Its system also provides universal access to care, and it is consistently well-graded from a public health standpoint.

 In other words, it does not have many of the shortcomings of the US healthcare system that potentially lead to unwarranted variation.

Although the prevalence of various surgical procedures has been studied, to our knowledge, there have been only few and limited comparisons of the variability in surgical outcomes among countries.

 Surgical outcomes as a quality metric are ideal to examine, given their well-established definitions and close association between the surgical process and the postoperative outcome. If the observed variability in Japan is low, the implication would be that the United States is truly unique. If, however, the variability in Japan is high, that would suggest that variability could be an inherent feature of healthcare systems. Our main aim is to study whether the United States is truly uniquely inefficient or whether geographic variation is part of the natural fabric of modern healthcare delivery.

Our analysis uses 2 data sources: a large Japanese administrative database and the Medicare 5% sample standard analytical file. The Japanese administrative database contains anonymous patient and encounter-level information collected over the 9-month period between April 1, 2012, and December 31, 2012. The data were originally collected from hospital forms that were submitted to the Japanese government as part of an ongoing effort to reform the healthcare reimbursement system in Japan.

 Hospitals that participate in this program are designated as Diagnosis Procedure Combination (DPC) hospitals, and together, they account for more than half of all the acute care beds in Japan. The DPC hospitals represent approximately 20% of all hospitals in Japan; they include all major academic medical centers and most of the nation’s larger hospitals. Our sample encompasses 100% of the patients from roughly half of these DPC hospitals. Please see eAppendix Figure A (

) for a comparison of our sample with the full set of DPC hospitals. The data contain information on patient status at admission and discharge; cost of admission; and details on surgeries, procedures, and drugs administered during the patient’s stay at the hospital. Medical clerks and licensed information managers generally oversee the collection and reporting of DPC data at individual hospitals to optimize the accuracy of the data. Studies correlating data from electronic health records and DPC data have shown very high compliance.

 In total, our data encompass information for approximately 5 million patients in more than 600 hospitals located across every region of Japan.

Data on patients in the United States were obtained from the Medicare 5% sample from 2010 to 2011. This is a random sampling of 5% of Medicare beneficiaries across all hospitals in the United States and has been used extensively to analyze different outcome measures.

 In our study, we analyze all inpatient claims by hospitals for this 5% sample. The information about each hospitalization includes dates of hospital admission and discharge, discharge status, disease(s) diagnosed, and cost of hospitalization.

Patients undergoing any of 5 relatively common, higher-risk surgeries (abdominal aortic aneurysm repair, coronary artery bypass grafting, colectomy, gastrectomy, or pancreatectomy) were included in the analysis. We follow the definitions outlined in the US Agency for Healthcare Research Quality’s Inpatient Quality Indicator guidelines when available.

 These surgeries have been previously used in related investigations studying the variability of surgical quality and outcomes.

 Japanese patients were selected using a Japan-specific surgery classification scheme (K-codes). K-codes were individually selected and verified to correspond to their 

International Classification of Diseases, Ninth Revision, Clinical Modification

) counterparts for this study; they have similarly been used in other studies to create comparable cohorts.

 Comorbidities were defined using Elixhauser classification definitions for both the US and Japanese cohorts. Prior validation studies have shown that both 

 coding algorithms produce similar estimates of comorbidity prevalence in administrative data.

 Patients were restricted to those 65 years or older.

We investigated 5 outcomes: in-hospital mortality, postsurgical complications, failure to rescue, length of stay, and cost of hospitalization. Postsurgical complications are defined using 

 codes in the US Medicare dataset, and corresponding 

 codes for the Japanese dataset. Complications included pulmonary failure, pneumonia, myocardial infarction, deep vein thrombosis/pulmonary embolism, acute renal failure, postoperative hemorrhage, surgical site infection, and gastrointestinal bleeding. The specific 

 codes for these complications have been previously validated as having reasonable agreement when comparing administrative data with the medical record.

 Failure to rescue is defined as death after any of the previously listed complications. Variations in cost in the United States were calculated using the Medicare Part A reimbursement to hospitals.

We subtracted disproportionate share payments, indirect medical education payments, and payments classified under total prospective payment system capital from United States cost totals, because these represent intended systemic cost variation and would inflate our variation estimates for the United States. Regions of comparison were defined for each country so that they were similar; variability was compared among states in the United States and among prefectures in Japan, graphically depicted in eAppendix Figure B.

In this study, we calculated outcome-specific coefficients of variation (COVs), defined as the sample standard deviation divided by the mean. The COV is a unitless measure that allows for a comparison of variation between a random variable with different means from 2 samples. For example, when used to measure the consistency of an elite athlete’s performance across multiple races, the COV will usually be between 1% and 2%.

Potential confounders, such as patient characteristics (age and gender for both countries’ cohorts, and additionally race for the US cohort), comorbidities, and urgent versus scheduled admissions, were included in all regression models. We additionally used hierarchical modeling techniques to adjust our COV estimates for reliability.

 Using random effects logistic regression models, we generated empirical Bayes predictions of the outcomes that were used to calculate our final COV values. This technique minimizes the impact of statistical outliers such as those from small-volume hospitals whose estimates can be considered less reliable than those of larger-volume hospitals. By variably shrinking the contribution of small-volume hospital estimates toward the average mortality rate across all hospitals, we are able to better estimate systematic, nonrandom variation as opposed to variation due to chance.

We additionally performed a sensitivity analysis by calculating age- and sex-standardized COV values for both countries to the 2000 US population, weighting all estimates by the number of procedures performed in each state and prefecture. These demographic- and case-mix—adjusted COVs were evaluated by performing multivariate regressions for each outcome of interest. We then calculated the adjusted COV (adjusted for age, gender, race, comorbidities, and urgency of admission) based on these regression estimates.

We assessed the geographic variation of in-hospital mortality, complication rate, failure to rescue rate, length of stay, and the cost of hospitalization for 5 major surgical procedures in the United States and Japan.

The Japanese surgical cohort had higher rates of colectomy and gastrectomy, together composing 74% of the analyzed surgeries versus 31% for the United States. The US surgical cohort was composed of a higher number of coronary artery bypass graft (CABG) procedures, representing 52% of the surgical cohort (

). Pancreatectomies were the least frequent surgery performed in both groups. Table 1 shows the raw descriptive statistics of each cohort, but they should not be used for comparison against each country as they do not take into account differing patient demographics and comorbidities between each country.

In comparing the United States and Japan’s reliability-adjusted COV values (

), the observed geographic variations among regions in Japan were higher than those of the United States for all 20 surgical outcome measures (4 outcomes for 5 surgeries). Our sensitivity analyses (

) reiterated these results, as the variation was higher in all 20 outcomes except for 1: there was lower variation in length of stay for colectomy in Japan. Of note, is that colectomy was the most frequently performed surgery in our analysis (41% of the total cohort) in Japan, but not in the United States (26% of the total cohort).

The variation in the cost of hospitalization was uniformly lower for Japan across all 5 surgeries for reliability-adjusted COVs. This was true in our sensitivity analysis as well. The level of cost variation between surgeries was similar across surgeries in Japan (adjusted COV values of 7.3 to 7.6), indicating that the specific type of surgery did not have much influence on the cost variability between regions. This is mostly true for the United States as well, as the adjusted COV values ranged from 21.7 to 22.8 with the exception of CABG with a COV of 25.3.

The level of variation for a given postsurgical outcome was similar across different surgeries in each country. For example, the approximate reliability-adjusted COVs for mortality across the 5 surgeries ranged from 15 to 17 in the United States, while ranging from 29 to 33 in Japan. The failure-to-rescue COV ranges from 16 to 19 in the United States and between 47 and 50 in Japan. One notable deviation from this pattern was length of stay in the United States. Four of 5 surgeries in the United States had a length-of-stay COV between 42 and 55, with the exception of CABG, which had a much lower COV at 29.

Our sensitivity analysis endorses the conclusions from our reliability-adjusted models, and it also demonstrates the importance of using the reliability-adjusted estimates. Though our conclusions remain the same, the estimates of variability are much higher and dispersed in our sensitivity analysis, especially for less frequently performed procedures such as pancreatectomy where small differences in outcomes could have large statistical manifestations (Table 3 and Figure 2).

In summary, the United States has lower variability for all postsurgical outcomes, but it has higher cost variability relative to Japan. The United States and Japan have low variability between surgeries for a given postsurgical outcome, suggesting that the differences in overall variation between the countries are not dependent on the specific surgery performed.

Much of the discussion around geographic variation in US healthcare expenditure and quality has focused singularly on the United States, with little work comparing the geographic variation in healthcare outcomes among other countries and the United States, especially for resource-intensive surgical procedures. Although previous studies have investigated whether the US healthcare system as a whole is uniquely inefficient compared with other countries,

 much of the literature around geographic variation does not address this topic, instead citing the inefficiency of the US healthcare system as a cause for geographic variation. This context is important when evaluating the causes and consequences of variation.

Healthcare efficiency has previously been framed as a combination of productive and allocative efficiency. Productive efficiency refers to the impact of healthcare given a series of inputs such as physicians and hospitals, and allocative efficiency measures the benefits from the marginal cost of healthcare versus the cost of other goods. In this analysis, we characterize both the variation in healthcare outcomes (outputs) and the variation in procedure costs (inputs) across each country, therefore addressing both aspects of efficiency described above. Higher outcome variations reflect lower productive efficiency—that is, given the same level of healthcare inputs, there are differences in achieving desirable outcomes. In the same manner, higher cost variations reflect lower allocative efficiency&mdash;that is, in order to achieve the desirable outcome, the amount and therefore the value of each healthcare dollar spent varies greatly.

In this study, we demonstrate that although cost varies more in the United States compared with Japan, all postsurgical outcome measures show less variability in the United States. This suggests that geographic variation in outcomes is not a manifestation of US-specific inefficiencies, but is more likely an inherent part of healthcare delivery and a shared characteristic among healthcare systems.

Our conclusions are in line with the results of previous studies comparing the United States, United Kingdom, and Germany that show that among a group of medical conditions, no single country consistently spent less money and achieved better health outcomes across different indications.

 Our finding suggests that every country struggles in its own way to achieve productive efficiency in healthcare. In other words, building a healthcare system where the least expenditures result in the best outcomes is a universal problem.

The fact that the United States had higher observed cost variation than Japan is also consistent with previous analyses. In 2008, a report by the Congressional Budget Office concluded that the geographic variation in healthcare spending was higher in the United States than in Canada and the United Kingdom, citing the fact that the financing of healthcare is more centralized in those countries.

 In Japan, the government centrally controls healthcare costs by budgeting and setting standard reimbursement rates for procedures. Although our results show that there is more cost variation in the United States compared with Japan, this is unlikely to represent US-specific unwarranted variation in healthcare resource use. A recent report by the IOM found that if geographic variation were eliminated for postacute care services, variation in Medicare spending would fall by 73%.

 Much of the geographic variation in expenditure in the United States is actually attributable to postacute care and not to excess use of care during hospitalizations.

The most likely reason for Japan’s increased outcomes variability compared with that of the United States is the fact that Japan has many more hospitals per capita than the United States. In 2010, Japan had 68 hospitals per million individuals compared with 19 hospitals per million in the United States—a more than 3-fold difference.

 This translates into surgeons performing high-risk procedures, such as those in this study, relatively infrequently. Similarly, hospital staff may have less experience caring for these postoperative patients and are likely to be less prepared for both preventing and resolving postoperative complications. There is an extensive literature, both in the United States and in Japan, on the benefits of centralized surgical centers, where increased surgical and postoperative care experience, as well as differences in staffing, can translate to better outcomes for patients.

Other reasons that explain the high variability in Japan include the less-centralized regulation of hospitals and medical care, and the lower level of regulation over physicians’ training and development.

 In 1 study, adherence to clinical practice guidelines for cholangitis were measured in hospitals across Japan. Results showed markedly varied compliance that correlated directly with in-hospital mortality rates.

 Patients in Japan have universal access to any facility, and in general they do not require referrals to visit a specialist. Furthermore, physicians and nurses are currently licensed for life; they do not have requirements for license renewal or continuing medical education.

 Both of these factors could contribute to higher variability in quality of care that patients could receive, especially for higher-risk, lower-volume procedures.

Previous studies demonstrate that clinical decision making is affected by the supply of specialists, local training frameworks, and regulatory factors, all of which are more fragmented in Japan than in the United States. In addition, geographic variation in rates of surgery reflect physician beliefs about the level of appropriateness of the surgery and the extent to which patients have a say in their treatment.

 The decentralized nature of Japan’s healthcare system, with its many hospitals and unstructured referral system, presumably exacerbates variation by creating more physician “micro-communities” that lead to variable care.

Potential drivers for more homogeneous outcome performance in the United States may arise from the fact that the United States has greater standardization of the surgical training curriculum relative to Japan, as there is no standardized national surgical curriculum in Japan or even the equivalent of the Accreditation Council for Graduate Medical Education.

 The fact that the United States has a multiplicity of accrediting organizations that all work toward achieving a higher quality of care also undoubtedly plays a role.

 The existence of managed care in the United States is unique, and the role that national payers have in emphasizing quality of care also encourages more uniform quality of care across the country. Public disclosure of surgical outcomes in the United States may also influence quality, as public reporting has been extensively studied as an effective quality improvement strategy by encouraging changes in the behavior of providers.

 However, with respect to surgical outcomes the evidence is mixed.

 The higher variation in United States costs could be due to regional adjustments and variation in diagnoses billed for the same procedure.

Articles

Compared with Japan, the United States has substantially less geographic variation in surgical outcomes, but it has higher variation in cost.

Geographic Variation in Surgical Outcomes and Cost Between the United States and Japan

To assess the broad impacts of Medicare Part D and the extent to which prior concerns have been realized.

We used administrative data to summarize beneficiary enrollment and plan participation in Part D, and compared pharmaceutical use and out-of-pocket spending before and after the introduction of Part D. We characterized the benefit designs of the 10 largest Part D plans in 2006 and compared them with the benefit designs of 7 non—Part D plans often cited as examples of low-cost or comprehensive drug benefits. 

By 2008, nearly 90% of seniors had drug coverage at least as generous as the standard Part D benefit. Excluding premiums, annual out-ofpocket spending in the 10 largest Part D plans was comparable to that of other private and public drug benefits, with the most prominent differences attributable to out-of-pocket spending on drugs

not covered in the plan. Poorer beneficiaries have gained the most from Part D in terms of increased access to medications and reduced out-of-pocket spending.

Coverage under Part D is comparable to that under non—Part D plans with respect to key features that are likely to be important to Medicare beneficiaries—access to medications and out-of-pocket costs. Nonetheless, concerns remain over drug pricing and gaps in coverage. The government should continue to monitor

the competitiveness of the Part D market to ensure it meets the diverse needs of Medicare beneficiaries.

After more than 2 years of experience, data are starting to accumulate on the performance of Medicare Part D.

Nearly 90% of Medicare beneficiaries have prescription drug coverage as generous as the standard Part D benefit.

Part D was associated with a 16% annual decrease in out-of-pocket spending and a 7% increase in the number of prescriptions.

Coverage under Part D is comparable to that of other private and public drug benefits with respect to access to medications and out-of-pocket costs.

Poorer beneficiaries have gained the most from Part D.

Prior to the introduction of Medicare Part D in 2006, there were concerns that the program would not appeal to all types of beneficiaries, nor entice a sufficient number of health plans to offer a benefit. Failing to attract a broad cross-section of beneficiaries could jeopardize the long-term viability of the program if, for example, only the sickest patients enrolled. Similarly, failing to attract a sufficient number of plan sponsors would diminish competition, which was widely viewed as the cornerstone of a well-functioning program. Other concerns focused on low-income beneficiaries, many of whom were previously covered by Medicaid’s prescription drug benefit. Would they be negatively affected by potentially more expensive private plans with more restrictive formularies? Would seniors be able to navigate the myriad of plan choices and benefit options, including the gap in coverage (“doughnut hole”) in Part D? Finally, many analysts feared that the cost of the program would exceed original Congressional Budget Office estimates if private plans could not obtain sufficient price discounts or constrain growth in per capita utilization.

Nearly 3 years into the Medicare Part D program, enough data have accumulated on the program’s performance for a preliminary picture to emerge. We examined beneficiary and plan participation in the program, and how Part D has affected beneficiaries’ access to medications, utilization, and financial risk. We compared the 10 largest Part D plans in 2006 with 7 non—Part D plans often cited as examples of low-cost or generous pharmacy benefits. We found that although Part D has exceeded expectations in many ways, several questions remain that warrant careful monitoring in the future.

Beneficiary Enrollment and Plan Participation

We used data from the Centers for Medicare & Medicaid Services to characterize beneficiaries’ drug coverage and plan participation in Medicare Part D. Estimates of Medicare beneficiaries with prescription drug coverage were from enrollment files. Data on plan participation were derived from the Prescription Drug Plan Landscape File.

To characterize the Part D program, we examined the 10 largest prescription drug plans (PDPs) based on 2006 enrollment. These 10 plans accounted for 46% of total Part D enrollment in 2006. We compared these 10 plans with 7 public and private plans often cited as examples of low-cost or comprehensive drug benefits with respect to formulary coverage, cost-sharing features, and out-of-pocket costs. The 7 non—Part D plans included the Federal Employees Health Benefits Program (FEP Basic), the California Public Employees’ Retirement System (CALPERS Choice/CALPERS Care), the California State Medicaid program (Medi-Cal), the Veterans Affairs plan (VA National Formulary), the Department of Defense plan (TRICARE), Anthem Blue Cross, and Kaiser Permanente.

We used data from the Fingertip Formulary and Internet sources to characterize the benefit designs and cost-sharing arrangements of each of these 17 plans. The Fingertip Formulary is an online database of formulary designs for nearly all commercial, Medicare, Medicaid, and public insurance plans (http://www.fingertipformulary.com). The data in our analysis contain formulary designs as of July 16, 2007. In cases where there were discrepancies between sources, we contacted the plans directly to resolve any ambiguities.

We compared coverage in each plan for the 300 drugs most widely prescribed to seniors between January and October 2006. The list of medications was based on Verispan’s Vector One data, a national-level prescription and patient tracking service that collects nearly half of the retail prescriptions dispensed each month in the United States. The list of 300 drugs was reduced to 252 in some analyses because of 33 generic/brand duplicates and 15 drugs not covered under Part D. We also examined how broadly plans cover medications for chronic diseases of the elderly to assess the potential clinical consequences of formulary exclusions.

Public programs with statutory or regulatory limits on cost-sharing often use formulary restrictions or other types of administrative requirements to limit access to specific medications. For example, plans may require prior authorization (requiring permission before certain drugs can be dispensed), step therapy (requiring use of lower-cost medications before providing coverage for more expensive alternatives), or quantity limits (which restrict the number of pills or prescriptions dispensed per month or per patient) to control utilization and improve patient safety. We examined the extent of these restrictions in each of the 17 plans.

It often is difficult to translate the stated pharmacy benefit into actual prices that consumers face. Multitier formularies are the standard for most private plans, and they also have gaps in coverage, out-of-pocket limits, and discounts for purchases through mail-order or in-network pharmacies. These added complexities mean that the price a consumer will pay for a given drug depends in which tier it is placed, where it is dispensed, and at what time of year. To address this issue, we estimated average beneficiary out-of-pocket costs in each plan for a standardized set of pharmacy claims. The set of drug claims was generated by drawing a random sample of seniors from 14 large employers providing retiree drug coverage in 2004 (non—Part D plans, national in scope). The sample included beneficiaries age 65 years and older who were continuously enrolled in a plan for the entire year and who had at least 1 pharmacy claim for the 300 most common drugs. We randomly selected 10% of enrollees from each plan, up to a maximum of 500 per plan. We then created a “market basket” of drug claims used by this random sample of retirees, restricted to the 300 most common drugs. This fixed set of pharmacy claims then was processed through the benefit designs of each plan to calculate the average out-of-pocket costs in each of the 17 plans. This approach incorporates the plan’s cost-sharing arrangements and other factors such as where the drug was dispensed (eg, network or mail-order pharmacy) and cumulative spending to date.

For drugs that were not covered by the plan—either because they were excluded from the formulary or because the beneficiary’s cumulative spending was below the deductible or in the doughnut hole&mdash;we assigned an out-of-pocket payment equal to the full price of the drug (excluding rebates). We used the full price of the drug for each Part D plan as reported on the Medicare Web site in July 2007. Some Part D plans did not report prices for all of the drugs. In those cases, we imputed the average price of a drug across all other Part D plans with pricing data. Similarly, because we did not have price data for the employer plans, we assigned the average price of each drug in the Part D plans.

Our approach to estimating out-of-pocket costs for a fixed basket of pharmacy claims had 2 principal limitations. Most importantly, it assumed no demand response (ie, that patients’ drug use or “bundle of claims” was fixed across plans and thus invariant to the benefit design). In reality, patients will alter their choice of drugs in response to changes in cost-sharing.

 Second, the price we used for uncovered drugs was measured with some error, given the limitations on quantity and dosage and our use of average prices in the 10 largest Part D plans. Given these limitations, our out-of-pocket estimates provide a useful measure of relative plan generosity, but should not be viewed as an accurate measure of the actual cost burden to beneficiaries.

We estimated the effects of Part D on pharmacy utilization and spending using a “before and after” design. For the baseline period, we analyzed data from the 2004 Medicare Current Beneficiary Survey (MCBS). For the follow-up period, we analyzed 2006 enrollment and claims data from a large Part D plan. These datasets are not perfectly comparable. First, the MCBS relies on self-reports and has been shown to undercount drug spending.

 Second, the MCBS is a stratified national sample of Medicare enrollees, whereas the Part D data are a convenience sample, albeit a large one.

To address the first problem, MCBS spending was adjusted upward from 2004 to 2006 using the consumer price index prescription drug series, and both spending and number of fills were adjusted upward by 15% to account for the undercount. To address the second problem, we constructed demographic weights based on age, sex, and state of residence so the Part D enrollees were demographically representative of the Medicare population. We re-weighted the Part D data to match the MCBS for each age-sex-state-subsidy cell and used these weights in all reported analyses.

We also approximated low-income subsidy (LIS) eligibility in the MCBS prior to Part D introduction based on a reported household income less than 150% of the federal poverty line. No asset information is available in the MCBS. Prior work suggests that nearly 14 million Medicare beneficiaries would qualify for the LIS based on this income test alone, of which 2.4 million (17%) would be ineligible based on assets.

For each of the 3 eligibility groups (dual-eligible, LIS, general), we examined the distribution of prescriptions and out-of-pocket spending before and after the introduction of Part D. In particular, we compared the means, variances, and various quartiles of the distributions of these variables.

 shows the breakdown of prescription drug coverage among Medicare beneficiaries as of January 2008. Nearly 90% of seniors had drug coverage at least as generous as the standard Part D benefit. More than 57% were enrolled in a Part D plan (15% dual-eligibles), with the remainder covered by employer plans (15%) or some other creditable plan (17%).

Widespread enrollment in Part D is partly attributable to robust plan participation across states. In January 2006 alone, there were on average 43 PDPs in each state, with states at the 25th percentile having 41 plans. In 2007 and 2008, this number rose to 54 plans, with states at the 25th percentile having 53 plans on average. Despite the large number of plan offerings, beneficiaries enrolled in Part D were highly concentrated in a few plan sponsors. Nearly 44% of enrollees in 2007 were in plans offered by UnitedHealth and Humana, while almost a quarter were covered by Universal American Financial, WellPoint, WellCare, Kaiser Permanente, and Coventry. A similar degree of concentration occurred in the Medicare Advantage market prior to Part D.

characterizes the formulary coverage and costsharing arrangements of the 10 largest Part D plans, as well as the 7 comparison plans. For each of the plans, we show the number of covered drugs in each tier (or a single tier in plans where only 1 tier exists), as well as the number of drugs not covered. Among the 10 largest Part D plans, between 147 and 157 of the 300 most common medications were covered in tier 1. Further, the most (least) expansive plans excluded 4 to 6 (24 to 41) medications. There was little variation across plans in the number of drugs assigned to the first 2 tiers. Thus, variation in the number of covered drugs across Part D plans was driven by how many products were assigned to the third tier versus not covered at all. Coverage under the largest Part D plans was comparable to that under many of the non—Part D plans such as Anthem Blue Cross, FEP Basic, CALPERS, and TRICARE, which excluded 0 to 6 medications. By contrast, the Kaiser Permanente and VA plans excluded 75 and 84 medications, respectively.

Because certain drugs are more widely used than others, excluded medications may account for a relatively smaller or larger fraction of total prescriptions. The last column of Table 2 shows the proportion of total national prescriptions accounted for by the list of excluded drugs in each plan. For example, the 84 drugs excluded from the VA National Formulary accounted for nearly one-quarter (24.7%) of all prescriptions dispensed to seniors in 2006 among the 300 most common medications. By contrast, drugs excluded from even the most restrictive Part D plan accounted for only 12.6% of total prescriptions, and drugs excluded from the least restrictive Part D plan accounted for just 4.4% of total prescriptions.

To assess the potential clinical consequences of excluding drugs from the formulary, we also examined how broadly plans cover medications for chronic diseases of the elderly, specifically diabetes, heart disease, hypertension, and high cholesterol. 

 shows the number of excluded drugs in each plan. The 10 largest Part D plans covered the vast majority of the drugs in these therapeutic classes, with the exception of angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers (ACEIs/ARBs). The most restrictive Part D plan excluded just 4 of 11 antihyperlipidemic agents, 1 of 13 antidiabetic agents, and 1 of 10 beta-blockers. Although coverage was less generous for ACEIs/ARBs, even the most restrictive Part D plans covered 14 to 18 of the 22 drugs in the class. In contrast, the VA National Formulary excluded 7 of 13 antidiabetic agents, 12 of 22 ACEIs/ARBs, and 5 of 11 antihyperlipidemic agents.

Plans also can place administrative restrictions on specific drugs or classes of medications that can greatly limit access. 

shows the number of drugs in each plan subject to prior authorization requirements, quantity limits, step therapy, and other restrictions. The most common formulary restriction in Part D plans is a quantity limit, where the plan will only cover a drug up to a designated quantity. If prescribing physicians feel it is medically necessary to exceed the set limit, they must get prior approval from the plan. Among the 10 largest Part D plans, the median quantity limit applies to 54 of the top 300 drugs.

Prior authorization is commonly used in state Medicaid programs as a cost containment tool, but is far less common in Part D plans. Half of the 10 largest plans do not impose prior authorization requirements on any drug, while the most restrictive Part D plan requires it for just 7 medications. Step therapy is used in some Part D plans, but is typically applied to a small number of drugs. Although VA does not use any of these formulary restrictions, it imposes other types of restrictions to control utilization. For example, 34 of the 180 formulary drugs in the VA National Formulary (among the top 300 drugs) are only covered for specific indications, dosages, or intake formulations.

presents predicted annual out-of-pocket costs (excluding premiums) under each plan for our market basket of claims and decomposes these costs into out-of-pocket expenditures for covered and noncovered drugs. Predicted out-of-pocket spending for the 5 least expensive Part D plans was roughly $1000 annually, which was modestly higher than spending in Anthem Blue Cross, FEP Basic, and CALPERS. With the exception of WellCare Signature, out-of-pocket spending in the remaining Part D plans was between $1300 and $1400 annually, making these plans more comparable in spending to the VA plan.

A critical question is how Part D affected pharmaceutical use and spending, both overall and for specific groups of beneficiaries. Our analyses comparing MCBS data in 2004 with Part D claims in 2006 suggest that Part D was associated with a 16% annual decrease in out-of-pocket spending and a 7% increase in the number of prescriptions (results not shown). These estimates are consistent with other findings using 2006 data from a large national pharmacy chain.

 Our analysis also suggests that these changes were concentrated among the poor. Average out-of-pocket spending among the dual-eligibles and LIS population declined markedly, but was largely unchanged for the general Part D population. Equally important, Part D was associated with reduced financial risk for low-income populations, as measured by the variance of out-of-pocket spending (

). Prior to Part D, the probability of having out-of-pocket drug spending higher than $1000 in a year was 24% in the LIS population and 3.6% in the dualeligible population. After Part D, those probabilities were less than 1% for each group.

Medicare Part D generated much confusion at the time of its introduction in January 2006. Some beneficiaries did not understand the benefit designs and were not sure how Part D interacted with existing drug coverage or whether they might gain or suffer financially from enrolling.

 However, after more than 2 years of experience, the assessment has changed. Nearly 90% of Medicare beneficiaries have prescription drug coverage at least as generous as the standard Part D benefit. Policies to protect plans from excessive losses in the first few years (through reinsurance and risk corridors) as well as efforts to educate beneficiaries about plan choices led to a large number of sponsors and a wide array of options for beneficiaries to choose from. Despite the large number of plans, it appears that most beneficiaries who enroll in the program are making plan choices that reflect both their health status and the market circumstances.

Enrollment in Part D was widespread for several reasons. First, the large federal subsidy for Part D plans—74.5% of the premium is paid by Medicare&mdash;dramatically reduced the cost of coverage for most beneficiaries. Second, individuals previously covered under Medicaid were automatically enrolled in a private Part D plan. Third, the widespread availability of low premium plans (less than $20 per month), combined with penalties for late enrollment, made coverage more appealing to healthy seniors who might otherwise have been dissuaded from enrolling. However, there remains a substantial core of seniors who need to be educated that enrolling in Part D is in their own interest, and reaching them should continue to be a health policy priority.

Despite the variation across Part D plans, annual out-of-pocket spending in the 10 largest plans was only modestly higher than that in the other private and public drug plans, with the most prominent differences attributable to out-of-pocket spending on drugs not covered in the plan. Poorer beneficiaries, specifically dual-eligibles and those eligible for additional LISs, have gained the most from Part D in terms of increased access to medications and reduced financial risk.

Coverage under Part D is comparable to that under non–Part D plans with respect to key features likely to be important to Medicare beneficiaries.

William B. Vogt, PhD

Articles

Geographic Variation in Surgical Outcomes and Cost Between the United States and Japan

Medicare Part D After 2 Years