Shea AM, Curtis LH, Hammill BG, DiMartino LD, Abernethy AP, Schulman KA. Association Between the Medicare Modernization Act of 2003 and Patient Wait Times and Travel Distance for Chemotherapy. JAMA. 2008;300(2):189-196. doi:10.1001/jama.300.2.189
Author Affiliations: Center for Clinical and Genetic Economics, Duke Clinical Research Institute (Mss Shea and DiMartino, Drs Curtis and Schulman, and Mr Hammill), and Department of Medicine (Drs Curtis, Abernethy, and Schulman), Duke University School of Medicine, Durham, North Carolina.
Context The Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (MMA) altered reimbursements for outpatient chemotherapy drugs and drug administration services. Anecdotal reports suggest that these adjustments may have negatively affected access to chemotherapy for Medicare beneficiaries.
Objective To compare patient wait times and travel distances for chemotherapy before and after the enactment of the MMA.
Design, Setting, and Patients Analysis of a nationally representative 5% sample of claims from the Centers for Medicare & Medicaid Services for the period 2003 through 2006. Patients were Medicare beneficiaries with incident breast cancer, colorectal cancer, leukemia, lung cancer, or lymphoma who received chemotherapy in inpatient hospital, institutional outpatient, or physician office settings.
Main Outcome Measures Days from incident diagnosis to first chemotherapy visit and distance traveled for treatment, controlling for age, sex, race/ethnicity, cancer type, geographic region, comorbid conditions, and year of diagnosis and treatment.
Results There were 5082 incident cases of breast cancer, colorectal cancer, leukemia, lung cancer, or lymphoma in 2003; 5379 cases in 2004; 5116 cases in 2005; and 5288 cases in 2006. Approximately 70% of patients received treatment in physician office settings in each year. Although the distribution of treatment settings in 2004 and 2005 was not significantly different from 2003 (P = .24 and P = .72, respectively), there was a small but significant change from 2003 to 2006 (P = .02). The proportion of patients receiving chemotherapy in inpatient settings decreased from 10.2% in 2003 to 8.8% in 2006 (P = .03), and the proportion in institutional outpatient settings increased from 21.1% to 22.5% (P = .004). The proportion in physician offices remained at 68.7% (P = .29). The median time from diagnosis to initial chemotherapy visit was 28 days in 2003, 27 days in 2004, 29 days in 2005, and 28 days in 2006. In multivariate analyses, average wait times for chemotherapy were 1.96 days longer in 2005 than in 2003 (95% confidence interval [CI], 0.11-3.80 days; P = .04) but not significantly different in 2006 (0.88 days; 95% CI, –0.96 to 2.71 days; P = .35). Median travel distance was 7 miles (11.2 km) in 2003 and 8 miles (12.8 km) in 2004 through 2006. After adjustment, average travel distance remained slightly longer in 2004 (1.47 miles [2.35 km]; 95% CI, 0.87-2.07 miles [1.39-3.31 km]; P < .001), 2005 (1.19 miles [1.90 km]; 95% CI, 0.58-1.80 miles [0.93-2.88 km]; P < .001), and 2006 (1.30 miles [2.08 km]; 95% CI, 0.69-1.90 miles [1.10-3.04 km]; P < .001) compared with 2003.
Conclusion There have not been major changes in travel distance and patient wait times for chemotherapy in the Medicare population since 2003, the year before MMA-related changes in reimbursement.
In addition to establishing an outpatient prescription drug benefit for Medicare beneficiaries, the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (MMA) changed physician reimbursement for chemotherapy-related drugs and administration services. Before the enactment of the MMA, Medicare reimbursement to physicians for chemotherapy drugs often exceeded acquisition costs1 because many physicians obtained the drugs at substantially discounted prices. Some estimates placed Medicare payments at 3 times the cost of acquisition.2
In an effort to curtail this overpayment and align reimbursement more closely with market prices, the MMA reduced payments for chemotherapy drugs from 95% of average wholesale prices in 2003 to 85% in 2004. In 2005, payments were further reduced to 106% of manufacturer-reported average sales prices, which reflect the actual transaction prices of drug acquisition and are typically lower than the corresponding wholesale values.3 To offset the reductions, the MMA also increased reimbursement for drug administration services.4 The US Government Accountability Office estimated conservatively that Medicare payments for the 16 most commonly billed chemotherapy-related drugs would continue to exceed physicians' costs after the changes in reimbursement but by a smaller margin (22% in 2004 and 6% in 2005).1
Changes mandated by the MMA were directed at physician reimbursement and did not include substantial changes in reimbursement to institutions. Therefore, there was concern that the reduction in physician reimbursement would lead to closures of some private oncology practices, requiring the 80% of cancer patients who receive treatment in community settings to travel farther from their homes to local hospitals for treatment.5- 9 Moreover, without sufficient opportunity to plan and expand their services and without financial incentive to do so, hospital-based clinics might not have adequate resources to support the anticipated rapid influx of patients seeking chemotherapy, thereby further delaying provision of care.7 Opponents of the MMA also warned that quality of care might be negatively affected because financial constraints would necessitate the elimination of nursing and support staff4 and because cost shifting to patients in the form of co-payments might lead some patients to forgo care altogether.10
Despite these concerns, recent studies by the Medicare Payment Advisory Commission (MedPAC) and the National Patient Advocate Foundation found that patients were generally satisfied with their care and did not perceive changes in treatment following the enactment of the MMA.3,10,11 Still, there is limited empirical evidence about whether changes in reimbursement policy have influenced the location or timeliness of chemotherapy. Therefore, we examined patient wait times and travel distance for chemotherapy before and after the enactment of the MMA in a nationally representative sample of Medicare beneficiaries from 2003 through 2006.
We analyzed a 5% national sample of Medicare standard analytic files and corresponding denominator files for inpatient, outpatient, carrier, and durable medical equipment claims. The files are available from the Centers for Medicare & Medicaid Services (CMS) and represent a quasi-random sample of 5% of all Medicare beneficiaries. Beneficiaries are selected for the sample based on the last 2 digits of their Medicare beneficiary identification number.12 The inpatient files contain institutional claims for facility costs covered under Medicare Part A, and the outpatient files contain claims by institutional outpatient providers (eg, hospital outpatient departments, ambulatory surgery centers). The carrier files contain provider claims for services covered under Medicare Part B. The denominator files contain beneficiary identifiers, sex, race/ethnicity, birth dates, dates of death, zip codes, and information about program eligibility and enrollment.
We obtained all files for calendar years 2002 through 2006 from CMS. We eliminated invalid records and limited the analysis to persons living in the United States. For the carrier claims, we used the provider zip codes in the CMS files to determine locations of treatment. For inpatient and outpatient institutional claims, we linked Medicare provider identification numbers to Medicare cost report data to determine facility zip codes. The institutional review board of the Duke University Health System approved the study.
We included Medicare beneficiaries for whom a diagnosis of breast cancer (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis codes 174.0-174.9, 175.0-175.9, 233.0, and V10.3), colorectal cancer (codes 153.0-153.9, 154.0-154.8, 230.3-230.6, 159.0, V10.05, and V10.06), leukemia (codes 200.00-200.88, 202.00-202.28, 202.80-202.98, V10.71, and V10.79), lung cancer (codes 162.2-162.9, 231.2, and V10.11), or lymphoma (codes 202.40-202.48, 203.10, 204.00, 204.10, 204.20, 204.80, 204.90, 205.00, 205.10, 205.20, 205.80, 205.90, 206.00, 206.10, 206.20, 206.80, 206.90, 207.00, 207.10, 207.20, 207.80, 208.00, 208.10, 208.20, 208.80, 208.90, V10.60-V10.63, and V10.69) was reported on a single inpatient, outpatient, carrier, or durable medical equipment claim. We selected these cancer types because they are prevalent among elderly persons and chemotherapy is often indicated.
To specify the date of disease onset, we used the date of the earliest observed cancer claim. To be considered a new-onset or incident case, we required beneficiaries to have had no claims for any of the 5 cancer types in the previous calendar year. Note that this method does not prohibit the inclusion of patients with relapsed disease who may also have met this criterion. Therefore, the study sample represents a combination of incident and relapsed cases. We limited the sample to beneficiaries aged 67 years or older to minimize the risk of misclassifying prevalent cases as incident. In the event that a beneficiary was identified as having an incident case of more than 1 of the 5 cancer types in a given year, we retained the claim with the earliest incident date for the analysis.
Inclusion in the incident cohort was conditional on survival to chemotherapy, and the initial chemotherapy visit was required to be in the same calendar year as the incident diagnosis (ICD-9-CM diagnosis code V58.1 [V58.11 in 2006]; ICD-9-CM procedure code 99.25; and Current Procedural Terminology, Fourth Edition/Healthcare Common Procedure Coding System, Level II, National Codes 96400-96549, G0345-G0362, Q0083-Q0085, and J8510-J9999 [C8953-C8955 in 2006 only]). To identify the date of the first chemotherapy visit, we used the date of the earliest observed chemotherapy claim following an incident cancer diagnosis. We determined treatment setting from the file type in which the claim was observed. Claims in inpatient files represent services provided in inpatient facilities. Claims in outpatient files represent services in institutional outpatient facilities, such as hospital outpatient departments and rural health clinics. Claims in carrier files are submitted by noninstitutional providers and, thus, represent services provided in 1 of several locations, including community-based physician offices, hospitals, independent laboratories, patient homes, or other sites.13 For claims in carrier and durable medical equipment files, we used the “line place of service” variable to determine treatment setting.
We excluded carrier and durable medical equipment claims from settings other than inpatient facilities, outpatient facilities, and physician offices (n = 358). In the event that an institutional claim and an office-based carrier claim were observed on the same day, we retained the office claim for the analysis. Because of relatively high levels of seasonal migration among elderly persons in the United States, colloquially known as “snowbirds” or “sunbirds,”14 we excluded 1311 beneficiaries who traveled more than 100 miles for treatment (median, 377 miles; interquartile range, 152-945 miles). Finally, we excluded 329 beneficiaries with a prior diagnosis of metastatic cancer of the same cancer type (ICD-9-CM codes 198.81, 197.4, 197.5, 197.6, 197.0, 197.1, and 197.2).
For characteristics of patients in the incident cohort, we present categorical variables as frequencies and continuous variables as means with SDs. We identified comorbid conditions using the coding algorithms described by Birman-Deych et al15 and Quan et al.16 Specifically, we searched all inpatient, outpatient, and carrier claims for 365 days preceding the date of the incident diagnosis for evidence of cerebrovascular disease (ICD-9-CM codes 362.34 and 430.x-438.x), chronic obstructive pulmonary disease (codes 416.8, 416.9, 490.x-505.x, 506.4, 508.1, and 508.8), coronary heart disease (codes 410.x-414.x, 429.2, and V45.81), dementia (codes 290.x, 294.1, and 331.2), diabetes mellitus (code 250.x), hypertension (codes 401.x-405.x and 437.2), metastatic cancer of a different type (codes 196.x-199.x), peripheral vascular disease (codes 093.0 437.3, 440.x, 441.x, 443.1-443.9, 47.1, 557.1, 557.9, and V43.4), and renal disease (codes 403.01, 403.11, 403.91, 404.02, 404.036, 404.12, 404.13, 404.92, 404.93, 582.x, 583.0-583.7, 585.x, 586.x, 588.0, V42.0, V45.1, and V56.x). To test the significance of changes from 2003 through 2006, we used χ2 tests for categorical variables and Kruskal-Wallis tests for continuous variables. Medicare beneficiaries report race/ethnicity at the time of enrollment. In this analysis, we used the reported categories “black” and “white” and combined all others and missing values as “other/unknown.”17
To classify patient zip codes as rural or urban, we used rural-urban commuting area codes and the associated zip code approximations.18 We used the 2005 Area Resource File from the Health Resources and Services Administration to classify county of residence by poverty level. For each county, we obtained the proportion of persons living below the poverty level. We grouped the values by quartiles and defined beneficiaries in the highest quartile as living in counties with the highest poverty level and beneficiaries in all other quartiles as living in counties with the lowest poverty level.
We measured days from the incident diagnosis to the date of the first chemotherapy visit. We also calculated the distance traveled for treatment using the beneficiary's zip code and the zip code of the facility where chemotherapy was administered. Distances were measured in miles. We summarized days and distance over time and by treatment setting, cancer type, rural-urban status, and age (<75 vs ≥75 years). Because the initial, transitional MMA-mandated changes in Medicare payment for chemotherapy drugs and administration began in 2004, with additional adjustments implemented in 2005, we selected 2003 as the reference year.
We used multivariable linear regression models to examine the adjusted relationship between year of diagnosis and mean distance traveled for treatment and days from diagnosis to treatment. The models included age, sex, race/ethnicity, cancer type, geographic region, and comorbid conditions. In a similar fashion, we used quantile regression to examine the adjusted relationship between year of diagnosis and median wait times and travel distance.
We used SAS software, version 9.1.3 (SAS Institute Inc, Cary, North Carolina) for all analyses, and we considered P < .05 to be statistically significant.
Table 1 shows the baseline characteristics of the study population. In this nationally representative sample of Medicare beneficiaries, there were 5082 incident cases of breast cancer, colorectal cancer, leukemia, lung cancer, or lymphoma in 2003; 5379 cases in 2004; 5116 cases in 2005; and 5288 cases in 2006. These values reflect the number of patients who received chemotherapy in the same calendar year as their diagnosis, and they do not include patients who received other types of care. Thus, the values are not a true measure of cancer incidence in the Medicare population. Demographic characteristics and comorbid conditions were relatively consistent across the years of the study. With the exception of cancers other than those studied herein, all comorbid conditions were observed more frequently in 2006 than in 2003. The prevalence of hypertension, diabetes mellitus, peripheral vascular disease, and renal disease increased annually.
In each year, approximately 70% of patients had their first chemotherapy visit in a physician office, and no more than 10% received chemotherapy in an inpatient hospital setting. The distribution of treatment settings in 2003 was not significantly different from 2004 (P = .24) or 2005 (P = .72); however, we observed a small but significant difference between 2003 and 2006 (P = .02). The proportion of patients receiving chemotherapy in inpatient settings decreased from 10.2% in 2003 to 8.8% in 2006 (P = .03), and the proportion of patients in institutional outpatient settings increased from 21.1% to 22.5% (P = .004). The proportion of patients in physician offices remained at 68.7% (P = .29).
The median wait time for treatment was about 28 days (Table 2). Wait times were shortest for patients undergoing chemotherapy in inpatient settings and longest for patients in institutional outpatient settings. The median wait time for treatment in physician offices increased significantly from 29 days in 2003 to 31 days in 2005 (P < .001). However, the median wait time in 2006 was not significantly different from 2003 for any of the treatment settings. Median travel distance was approximately 8 miles (12.8 km) and varied little by treatment setting. Between 2003 and 2006, median travel distance increased from 7 miles (11.2 km) to 9 miles (14.4 km) for inpatient settings (P = .01) and from 7 miles to 8 miles for physician offices (P < .001) but remained at 7 miles for institutional outpatient settings (P = .04).
Table 3 shows median wait times by cancer type in 2003 and 2006. In both years, wait times were shorter for patients with leukemia, lung cancer, or lymphoma than for patients with breast or colorectal cancer. Wait times increased by 5 days for patients with colorectal cancer (P = .07) and by 3 days for patients with breast cancer (P = .15) or lung cancer (P < .001). Wait times decreased by 2.5 days for patients with leukemia (P = .15) and by 3 days for patients with lymphoma (P = .006). Wait times for patients in rural locations were 3 days longer in 2006 than in 2003 (P = .04), whereas wait times for patients in urban locations were 1 day shorter (P = .05) (Table 3). At the 75th percentile, wait times for patients in rural locations increased by 5 days. Median wait times for patients aged 75 years or older were unchanged (P = .70), whereas wait times for patients younger than 75 years increased by 1 day (P = .80). Similarly, wait times increased by 1 day for patients in counties with the least poverty (P = .03) but decreased by 4 days for patients in counties with the most poverty (P = .006).
Table 3 shows median travel distance by cancer type in 2003 and 2006. Distance traveled by patients with lymphoma was unchanged (P = .37). Travel distance increased by 1.2 miles (1.9 km) for breast cancer (P = .007), 1.5 miles (2.4 km) for colorectal cancer (P = .003), 0.9 miles (1.4 km) for lung cancer (P = .08), and 1.4 miles (2.2 km) for leukemia (P = .10). Although patients in rural areas traveled longer distances than patients in urban areas, median travel distance increased by only 1.2 miles in rural areas (P = .04) and by 0.7 miles (1.1 km) in urban areas (P < .001) (Table 3). Travel distance increased by 0.7 miles for patients aged 75 years or older (P = .001) and by 1.2 miles for patients younger than 75 years (P = .001). Travel distance increased by about 1 mile (1.6 km) in both high-poverty (P = .05) and low-poverty (P < .001) counties.
Controlling for age, sex, race/ethnicity, cancer type, geographic region, rural-urban status, proportion of persons living below the poverty level, and comorbid conditions, average wait times were not significantly different from 2003 to 2004 (difference, 0.28 days; 95% confidence interval [CI], –1.55 to 2.10 days; P = .76) or from 2003 to 2006 (difference, 0.88 days; 95% CI, –0.96 to 2.71 days; P = .35), but there was an increase of approximately 2 days from 2003 to 2005 (difference, 1.96 days; 95% CI, 0.11-3.80 days; P = .04). Patients with a history of metastatic cancer of a different type received chemotherapy 18.1 days sooner than other patients (95% CI, –19.60 to –16.57 days; P ≤ .001). Compared with white patients, black patients waited an average of 3.7 days longer for initial chemotherapy (95% CI, 1.21-6.27 days; P = .004).
Patients traveled slightly farther on average for chemotherapy in 2004 (1.47 miles [2.35 km]; 95% CI, 0.87-2.07 miles [1.39-3.31 km]; P < .001), 2005 (1.19 miles [1.90 km]; 95% CI, 0.58-1.80 miles [0.93-2.88 km]; P = .001), and 2006 (1.30 miles [2.08 km]; 95% CI, 0.69-1.90 miles [1.10-3.04 km]; P < .001) relative to 2003. Controlling for other factors, black patients traveled 3 miles (4.8 km) less than white patients (95% CI, –3.88 to –2.21 miles [–6.21 to –3.54 km]; P < .001).
Multivariable quantile regression models on median wait times and travel distance yielded results consistent with the linear regression analysis. Relative to 2003, there was no increase in median days in 2004 (difference, 0.38 days; 95% CI, –0.77 to 1.54 days; P = .52), an increase of 2.20 days in 2005 (95% CI, 1.05-3.34 days; P < .001), and no increase in 2006 (difference, 0.93 days; 95% CI, –0.33 to 2.19 days; P = .15). Median travel distance increased by 0.70 miles (1.12 km) in 2004 (95% CI, 0.34-1.06 miles [0.54-1.70 km]; P < .001), 0.80 miles (1.28 km) in 2005 (95% CI, 0.43-1.18 miles [0.69-1.89 km]; P < .001), and 0.85 (1.36 km) miles in 2006 (95% CI, 0.49-1.20 miles [0.78-1.92 km]; P < .001).
Opponents of the MMA have predicted negative consequences of the legislation since before its enactment.2,5- 10,19,20 In anticipation of reduced revenues, oncology practices reported closing satellite facilities and reducing office staff.8,9 However, current available evidence does not suggest that Medicare beneficiaries or cancer care providers have been adversely affected. In this analysis of a nationally representative sample of Medicare beneficiaries, we did not find a significant change in the distribution of patients by treatment setting from 2003 to 2004 or from 2003 to 2005, although we observed a small shift in the provision of initial chemotherapy from inpatient facilities to institutional outpatient settings between 2003 and 2006. This finding contrasts with predictions that changes in Medicare reimbursement for outpatient chemotherapy would result in a rapid shift toward provision of chemotherapy in inpatient settings.
Consistent with predictions of a migration of patients from community-based practices to hospital settings, concerns were expressed that increased travel requirements and longer wait times at overburdened facilities would delay the initiation of chemotherapy. However, in this analysis, median wait times in 2006 were not significantly different from those in 2003 for any of the treatment settings. Across all treatment settings and after adjustment for other factors, there was no significant change in time to treatment from 2003 to 2004. Patients waited approximately 2 days longer for chemotherapy in 2005 compared with 2003, but there was no difference between 2003 and 2006. Median wait times observed herein are consistent with or lower than those reported elsewhere.21,22 Clinical effects of delays in treatment remain largely unknown, but analyses of time to initiation of adjuvant chemotherapy for breast cancer have shown no difference in disease-specific or overall mortality for patients who received treatment within 1 to 3 months of definitive surgery.23,24 In this study, we evaluated days from diagnosis—not surgery—to the initial chemotherapy visit.
The median distance traveled for initial chemotherapy was 7 miles in 2003 and 8 miles in subsequent years. Across all treatment settings and after adjustment for age, sex, race/ethnicity, cancer type, geographic region, rural-urban status, poverty at the county level, and comorbid conditions, patients traveled an average of 1.5 miles farther for treatment in 2004 than in 2003, 1.2 miles farther in 2005 than in 2003, and 1.3 miles farther in 2006 than in 2003. Previous studies have reported an inverse relationship between travel distance and likelihood of radiotherapy25- 28; however, we are unaware of any study that has examined the relationship between distance traveled and outcomes. Nonetheless, it seems unlikely that increases in travel distance of less than 2 miles would be clinically significant.
In general, our findings are consistent with a recent MedPAC evaluation of the effects of MMA-related payment changes using Medicare claims data, commercial drug information, provider site visits, and patient focus groups. MedPAC found that despite changes in reimbursement, more patients received chemotherapy in physician offices and more total chemotherapy services were provided in 2004 and 2005 than in 2003. In addition, these increases were found to be consistent across geographic regions. Of the patients and providers who participated in the MedPAC assessment, none reported a decrease in the quality of cancer care following implementation of the MMA.3 Echoing these findings, a recent survey of patients who received chemotherapy either before or after the enactment of the MMA found no difference in wait times and equal rates of satisfaction with oncology care.11 In a comparison of 2005 reimbursement amounts to actual purchase prices for 39 drugs representing more than 94% of all Medicare hematology/oncology drug spending, the Office of Inspector General of the Department of Health and Human Services found that physician practices were able to purchase 35 of the 39 drugs at prices lower than the reimbursed amounts.29
There are several possible explanations for why we did not observe greater changes in patient wait times or travel distance after the enactment of the MMA. First, it might be too early to assess the full impact of MMA-related changes in reimbursement. In the short term, oncology practices may have been able to absorb financial losses or compensate for those losses by providing other services. The cumulative, long-term effects of lower reimbursement may still lead to office closures and reductions in community-based oncology services for Medicare beneficiaries. Second, nurses and other support staff—not patients—may bear the brunt of the impact of these changes. If practices have indeed reduced the number of staff that they employ, remaining employees may be working longer and harder to provide care for the same number of patients. Although there have been numerous anecdotal reports of staff reductions, a quantitative analysis of this issue has not been performed. Third, it is possible that the financial impact of the reimbursement changes has not been as substantial as was initially expected, so the incentive to change the delivery of care was minimal.
Moreover, examining the aggregate impact of MMA-related changes may obscure important changes in access to care for certain subgroups. It is plausible that patients in rural or underserved areas or those with limited resources might have been affected more dramatically. When we stratified our analysis by rural-urban status, we found that the median wait time for patients in rural areas increased by 3 days from 2003 to 2006 and that wait times for the top quartile of patients in rural areas increased by 5 days. The clinical significance of this difference is unclear, but the magnitude of the difference is greater than that observed in other subgroups. Careful analysis of the impact of MMA-related changes on patients in rural areas using a larger sample of Medicare beneficiaries may be warranted. Moreover, analyses stratified by cancer type suggest that wait times and travel distance increased for some cancer types and decreased for others. These changes may reflect temporal changes in treatment regimens, but we cannot explore this possibility without detailed clinical data.
Our analysis has some limitations. First, the coding of diagnoses and procedures in claims data may not always be accurate or complete and the quality of care received is unknown. However, previous studies have shown that diagnosis information in Medicare claims can be used to identify incident cancer with high specificity30 and that claims are a valid source of information for the identification of chemotherapy services.31- 33
Second, Medicare data do not include claims for beneficiaries during periods of enrollment in managed care. Similarly, ascertainment bias results when a person does not have contact with the health care system. A cancer diagnosis can only be recorded if there was a visit; therefore, the effect of ascertainment bias is a bias toward accepting the null hypothesis.
Third, the effects of reimbursement changes may have been mitigated by other factors. For example, payments made to physicians for concurrent CMS demonstration projects may have offset reductions in reimbursement, and physicians may not have fully responded to the implications of the new reimbursement system by 2006. In addition, where fixed costs are high, physicians may have limited ability to make sudden changes to their practice.
Fourth, our analysis of distance relies on measurements between zip code centroids. Previous research has shown a high correlation between travel times and straight-line distances between zip code centroids.34 Nevertheless, some zip codes encompass large geographic areas, and our analysis does not include any measurement of travel times.
Fifth, the sample size allowed us to detect differences of approximately 1 mile and 2.8 days with 80% power. Although statistically significant, these differences are unlikely to be clinically meaningful. Finally, in this analysis, we implicitly assumed that temporal changes were related to the MMA when, in fact, random variation exists over time.
As measured by travel distance and time to chemotherapy, our findings do not support anecdotal reports that the enactment of the MMA has changed access to chemotherapy in a meaningful way. Given the slow transition to full implementation of the reimbursement changes mandated by the MMA and the limited amount of follow-up data available at present, it may be premature to observe a relationship between these changes and delivery of care. With the aging of the US population, the number of elderly individuals with cancer is expected to increase proportionally, with incidence doubling in less than 30 years.35 As the burden increases, researchers should continue to monitor the effects of major policy changes on Medicare beneficiaries' access to care.
Corresponding Author: Lesley H. Curtis, PhD, Center for Clinical and Genetic Economics, Duke Clinical Research Institute, PO Box 17969, Durham, NC 27715 (email@example.com).
Author Contributions: Ms Shea and Dr Curtis had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Shea, Curtis, Schulman.
Acquisition of data: Shea.
Analysis and interpretation of data: Shea, Curtis, Hammill, DiMartino, Abernethy, Schulman.
Drafting of the manuscript: Shea.
Critical revision of the manuscript for important intellectual content: Shea, Curtis, Hammill, DiMartino, Abernethy, Schulman.
Statistical analysis: Shea, Curtis, Hammill.
Obtained funding: Schulman.
Administrative, technical, or material support: Shea, DiMartino, Schulman.
Study supervision: Curtis, Schulman.
Financial Disclosures: Dr Curtis reports receiving research and salary support from Allergan Pharmaceuticals, GlaxoSmithKline, Lilly, Medtronic, Novartis, Ortho Biotech, OSI Eyetech, Pfizer, and Sanofi-Aventis. Dr Curtis has made available online a detailed listing of financial disclosures (http://www.dcri.duke.edu/research/coi.jsp). Dr Schulman reports receiving research and/or salary support from Actelion, Allergan, Amgen, Arthritis Foundation, Astellas Pharma, Bristol-Myers Squibb, The Duke Endowment, Genentech, Inspire Pharmaceuticals, Johnson & Johnson, Kureha Corporation, LifeMasters Supported SelfCare, Medtronic, Merck, Nabi Biopharmaceuticals, National Patient Advocate Foundation, North Carolina Biotechnology Center, Novartis, OSI Eyetech, Pfizer, Roche, Sanofi-Aventis, Schering-Plough, Scios, Tengion, Theravance, Thomson Healthcare, Vertex Pharmaceuticals, Wyeth, and Yamanouchi USA Foundation; receiving personal income for consulting from Avalere Health, LifeMasters Supported SelfCare, McKinsey & Co, and the National Pharmaceutical Council; having equity in and serving on the board of directors of Cancer Consultants; having equity in and serving on the executive board of Faculty Connection LLC; and having equity in Alnylam Pharmaceuticals. Dr Schulman has made available online a detailed listing of financial disclosures (http://www.dcri.duke.edu/research/coi.jsp). No other disclosures were reported.
Funding/Support: This study was funded by a grant to Duke University from the National Patient Advocate Foundation as manager of the Global Access Project, Washington, DC.
Role of the Sponsor: The National Patient Advocate Foundation had no role in the design and conduct of the study or the collection, analysis, and interpretation of the data. Representatives of the sponsor reviewed a draft of the manuscript. The authors had full control over the preparation of the manuscript and the decision to submit the manuscript for publication.
Previous Presentations: This study was presented in part at a meeting of the Global Access Project, September 20, 2007, Washington, DC; and at the AcademyHealth Annual Research Meeting, June 10, 2008; Washington, DC.
Additional Contributions: We thank Damon M. Seils, MA, Duke University, for editorial assistance and manuscript preparation. Mr Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted.