Definition of study cohort of 85 088 men with clinically localized prostate cancer. AJCC indicates American Joint Committee on Cancer; HMO, health maintenance organization; SEER, Surveillance, Epidemiology, and End Results.
Specialists who were consulted prior to definitive treatment for 18 201 men who had a radical prostatectomy (A) and 35 925 men who had radiation therapy (B). Missing physician specialty codes on claims for radiotherapy accounted for the 7% of patients who received radiotherapy but for whom a visit with a radiation oncologist could not be identified.
Observed vs expected distribution of an individual urologist's patient to undergo a radiation oncologist evaluation.
Jang TL, Bekelman JE, Liu Y, Bach PB, Basch EM, Elkin EB, Zelefsky MJ, Scardino PT, Begg CB, Schrag D. Physician Visits Prior to Treatment for Clinically Localized Prostate Cancer. Arch Intern Med. 2010;170(5):440-450. doi:10.1001/archinternmed.2010.1
The 2 primary therapeutic interventions for localized prostate cancer are delivered by different types of physicians, urologists, and radiation oncologists. We evaluated how visits to specialists and primary care physicians (PCPs) by men with localized prostate cancer are related to treatment choice.
Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database, we identified 85 088 men with clinically localized prostate cancer diagnosed at age 65 years or older, between 1994 and 2002. Men were categorized by primary treatment received within 9 months of diagnosis: radical prostatectomy (n = 18 201 [21%]), radiotherapy (n = 35 925 [42%]), androgen deprivation (n = 14 021 [17%]), or expectant management (n = 16 941 [20%]). Visits to specialists and PCPs were analyzed by patient characteristics and primary therapies received and were identified using Medicare claims and the American Medical Association Physician Masterfile.
Overall, 42 309 men (50%) were seen exclusively by urologists, 37 540 (44%) by urologists and radiation oncologists, 2329 (3%) by urologists and medical oncologists, and 2910 (3%) by all 3 specialists. There was a strong association between the type of specialist seen and primary therapy received. Visits to PCPs were infrequent between diagnosis and receipt of therapy (22% of patients visited any PCP and 17% visited an established PCP) and were not associated with a greater likelihood of specialist visits. Irrespective of age, comorbidity status, or specialist visits, men seen by PCPs were more likely to be treated expectantly.
Specialist visits relate strongly to prostate cancer treatment choices. In light of these findings, prior evidence that specialists prefer the modality they themselves deliver and the lack of conclusive comparative studies demonstrating superiority of one modality over another, it is essential to ensure that men have access to balanced information before choosing a particular therapy for prostate cancer.
Most of the estimated 192 000 men diagnosed as having prostate cancer in the United States this year will have clinically localized disease at the time of diagnosis.1 Treatment strategies for these men include radical prostatectomy, radiotherapy, primary androgen deprivation therapy(PADT), or expectant management.
Selecting the appropriate treatment can be challenging, since no therapy has emerged as clearly superior.2,3 Patients rely on the clinical judgment, treatment philosophy, and recommendations of counseling physicians to help them make informed decisions. However, physician perceptions regarding the optimal management of localized prostate cancer vary widely according to specialty4,5 and geographic region.6 In a study in which US urologists and radiation oncologists were asked how they would want to be treated if they were found to have localized prostate cancer, 79% of urologists surveyed preferred radical prostatectomy, while 92% of radiation oncologists chose radiotherapy.4 In a later physician survey, investigators asked US urologists and radiation oncologists how they would treat patients with localized prostate cancer and found that specialists overwhelmingly would recommend the therapy that they themselves deliver.5 However, evidence is lacking on whether and how visits to primary care physicians (PCPs) and specialists who treat prostate cancer influence treatment patterns.
We conducted this study to characterize physician visit patterns prior to treatment for localized prostate cancer. Furthermore, we examined the association between patient characteristics and specialists consulted and how the types of physicians men visit relate to treatment choices.
The study cohort comprised patients from the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, which links patient demographic and tumor-specific data collected by SEER cancer registries to health care claims for Medicare enrollees. Criteria used to create our analytical cohort are shown in Figure 1. Information on incident cancer cases was available from 16 cancer registries from 1994 through 2002, covering 26% of the US population. Greater California, Kentucky, Louisiana, and New Jersey case contributions began in 2000.7 SEER registries collect data on each patient's cancer site, extent of disease, histologic findings, date of diagnosis, and initial treatment. We staged patients according to the American Joint Committee on Cancer criteria.8
The Medicare program provides health care benefits to 97% of the US population 65 years or older. Hospitalization information for those eligible for Medicare Part A is available from the Medicare Provider Analysis and Review files. Outpatient and physician/supplier Medicare files for services rendered in physicians' offices and hospital outpatient departments are available for the 96% of Medicare beneficiaries who elect Part B coverage. Approximately 94% of SEER patients 65 years or older have been successfully linked with their Medicare claims.9
The date of diagnostic prostate biopsy was obtained from Medicare claims data and was available for all men in our study cohort. Primary therapy was defined as the most aggressive treatment first delivered within 9 months following diagnosis and was classified as radical prostatectomy, radiotherapy, PADT, or expectant management. A 9-month interval was selected to ensure sufficient time for multidisciplinary consultation with specialists and initiation of definitive treatment. We identified corresponding SEER variables and Medicare diagnosis and procedure billing codes for each therapy using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)10 and Current Procedural Terminology (CPT)11 codes and searched for these codes in SEER data and Medicare claims. To enhance specificity, we required stringent evidence of radiotherapy delivery in ICD-9-CM or CPT codes.
Specialists who treat prostate cancer and dates of specialist visits were identified by the Health Care Financing Administration specialty codes in Medicare claims and data from the American Medical Association (AMA) Physician Masterfile, which characterizes physician information such as demographics, specialty, and board certification.12 Specialist visits were defined for each patient as visits to a urologist, radiation oncologist, or medical oncologist from the time of diagnostic biopsy to either treatment date or 9 months after diagnosis, whichever occurred first. Linkage between Medicare claims and the AMA Masterfile was achieved using Unique Physician Identifier Numbers.
Visits to established PCPs were defined as visits to the same PCP during the 12-month window preceding cancer diagnosis and again during the window between diagnosis and either treatment or 9 months after diagnosis based on both inpatient and outpatient physician claims. In sensitivity analyses, we also considered mens' visits to PCPs between diagnosis and 9 months after, but for whom we could not identify a prediagnosis visit. Codes for classification of treatment and specialist and PCP visits are available from the corresponding author by request.
Patient characteristics were categorized from SEER-Medicare data and analyzed according to the variables given in Table 1. We calculated a modified Charlson comorbidity index13 from Part A and Part B Medicare claims during a prediagnosis interval from 13 months to 1 month. Patients were not stratified into high- and low-risk groups because prostate-specific antigen test results are not reliably reported in SEER-Medicare data.
We used χ2 tests when appropriate to examine whether physician visits and patient characteristics were associated with primary therapy rendered. We explored patient characteristics associated with having seen a particular specialist prior to definitive prostate cancer therapy or 9 months after diagnosis. All variables of interest were retained in multivariable logistic regression analysis.
To determine whether patients seen by particular urologists were more or less likely to be evaluated by radiation oncologists, we identified all urologists (n = 2038) who saw at least 10 cohort members (inclusion of those with smaller size panels would not generate stable estimates). We grouped urologists into panels, according to the proportion of their patients who saw a radiation oncologist prior to treatment (0%, >0% to ≤10%, >10% to ≤20%, . . . >90% to ≤100%). The expected proportion of a urologist's patients who visited a radiation oncologist was estimated using a multivariable logistic regression model to predict for each patient the probability of a radiation oncologist visit, after adjustment for patient age, race, ethnicity, marital status, tumor stage, Gleason grade, diagnosis year, SEER site, population of county, and census tract income and education. For each panel, these probabilities were summed across all urologists to obtain an expected estimate. The expected distribution of a urologist's patients to visit with a radiation oncologist was then compared with the observed distribution in our cohort to identify variation that could not be explained by patient factors adjusted for in the logistic regression model. This comparison allows for the distinction between urologists whose patients were less likely and those whose patients were more likely to see a radiation oncologist than would have been expected if the process of physician visits were random.14,15 All statistical analyses were performed using SAS version 9.1 (SAS Institute Inc, Cary, North Carolina) software.
Among 85 088 men comprising our cohort, 18 201 (21%) underwent radical prostatectomy, 35 925 (42%) received radiotherapy, 14 021 (17%) received PADT, and 16 941 (20%) received expectant management. For those who received radiotherapy, 61% had external beam radiation, 25% had brachytherapy, and 14% had both. Table 1 presents primary therapy of cohort members, according to patient characteristics. Men aged 65 to 74 years were more likely to receive radical prostatectomy or radiotherapy. Older patients were more likely to receive expectant management or PADT.
Patients were evaluated by one of 2832 index urologists who performed the diagnostic biopsy. Including the initial diagnostic procedure and all physician visits in the subsequent 9 months, 4298, 1545, and 1552 unique urologists, radiation oncologists, and medical oncologists, respectively, saw men in our study cohort. Overall, 42 309 men (50%) were seen exclusively by urologists, 37 540 men (44%) by both urologists and radiation oncologists, 2329 men (3%) by both urologists and medical oncologists, and 2910 men (3%) by all 3 specialists.
Factors associated with a radiation oncologist visit were patient age, race, tumor stage and grade, year of diagnosis, and population of residence (Table 2). Specifically, patients aged 70 to 74 years, who were white, had tumors with a higher tumor stage (T2) and higher tumor grade, were diagnosed in more contemporary eras (2000-2002), and lived in a more populous region were at higher odds of having visited a radiation oncologist prior to receiving treatment.
Table 3 gives primary treatment according to patient age and specialist visit patterns. Overall, 34% of men who were evaluated exclusively by urologists had radical prostatectomy. Among men aged 65 to 74 years who were evaluated solely by urologists, radical prostatectomy was the most frequent form of therapy. In contrast, among all men, irrespective of age, who saw both radiation oncologists and urologists, radiotherapy was the predominant treatment modality received, having been delivered in 83% of these men.
Those seen by urologists with or without medical oncologists were more likely than those evaluated by both urologists and radiation oncologists to receive PADT or expectant management. This association was pronounced for men 75 years or older. Notably, radiation oncologists saw an increasingly smaller proportion of patients as men aged. For instance, only 16% of cohort members 75 years or older were evaluated by a radiation oncologist.
Figure 2 illustrates visit patterns according to those treated with radical prostatectomy and radiotherapy. Among 10 687 patients aged 65 to 69 years who underwent a radical prostatectomy, 80% consulted only a urologist (Figure 2A). Among those receiving radiotherapy, 93% of men not surprisingly saw both a urologist and radiation oncologist (Figure 2B). Missing physician specialty codes on claims for radiotherapy accounted for the remaining 7% of patients who received radiotherapy but for whom a visit with a radiation oncologist could not be identified.
Figure 3 displays histograms of the observed (gray bars) and expected (black bars) proportion of each urologist's panel of patients to have a radiation oncologist visit. There were some urologists whose patients were less likely and some whose patients were more likely to see a radiation oncologist than would have been expected if the process were random. The difference between the observed and expected distributions was significant after adjusting for patients characteristics (P < .001).
Of the 85 088 cohort members, 22% saw a PCP between diagnosis and definitive treatment or 9 months after diagnosis. For 17%, these visits were with established PCPs, namely those whom they had also visited in the 12 months preceding diagnosis. Overall, 79% of cohort members were seen by a PCP in the 12 months before diagnosis but not again between diagnosis and definitive therapy or 9 months after diagnosis.
Older patients and those with more comorbid conditions were more likely to see PCPs (Table 4). However, visits to an established PCP were not associated with a greater likelihood of visits to specialists. For example, among men who saw a PCP after cancer diagnosis, 82% saw only a urologist, whereas 13% and 4% also saw a radiation oncologist and medical oncologist, respectively. For those who did not see a PCP, 50% saw both a urologist and radiation oncologist.
Independent of patient age and comorbid conditions, men who saw a PCP after diagnosis were more likely to have expectant management compared with those who did not see a PCP (Table 5). When this analysis was further stratified to include prostate cancer specialists, results were similar (data not shown).
We evaluated how variations in the types of specialists seen after prostate cancer diagnosis relate to men's ultimate treatment choice. Among a population-based cohort of Medicare beneficiaries with clinically localized disease, we found that almost half of men received consultation from both a urologist and radiation oncologist prior to definitive therapy, which was typically radiation. There were also a proportion of men who never saw a radiation oncologist or other nonurologic specialist and received radical prostatectomy.
Age influenced treatment as well, with younger patients more frequently receiving radical prostatectomy if they saw a urologist exclusively. For example, among men aged 65 to 69 years and 70 to 74 years who were evaluated only by urologists, 70% and 45%, respectively, had radical prostatectomy. In contrast, 78% and 85% of men in these respective age groups evaluated by both radiation oncologists and urologists received radiotherapy. Rates of PADT and expectant management increased with age for men evaluated by urologists alone and for those evaluated by both urologists and medical oncologists. While only 16% of men 75 years or older with prostate cancer visited a radiation oncologist, few of these men were treated expectantly.
When we examined the practice patterns of urologists who cared for at least 10 cohort members, we found that urologists varied widely in the proportion of their patients who were also evaluated by radiation oncologists. If patient factors were the sole determinant of patterns of specialty evaluations, then we would have expected a random distribution of the tendency of an individual urologist's patients to have a radiation oncologist evaluation, after accounting for important patient factors. Instead, we identified some urologists whose patients were consistently evaluated by radiation oncologists and others whose patients were seen more seldom, suggesting that referral patterns may be influenced by factors other than those specific to the patient's condition, such as geographic access to radiation oncologists or radiotherapy facilities or personal viewpoints of counseling physicians. Further research is needed to explore and characterize these factors.
Our findings provide new insight into the relationship between physician visit patterns and receipt of therapy for localized prostate cancer. Prior physician surveys suggest that urologists and radiation oncologists might recommend their own treatment modality based on their stated preferences in response to hypothetical survey questions.4,5 The pattern of specialist visits and treatment that we observed suggests that these preferences may be affecting treatment decisions of Medicare patients.
The survey by Fowler et al5 suggests that for men with a life expectancy of less than 10 years, 68% of urologists and 81% of radiation oncologists believe that either radical prostatectomy or radiotherapy confers survival advantage. Neither group of specialists was supportive of expectant management for any but a limited subset of men.5 On the contrary, we observed increasing frequency of conservative management with age among men seen by urologists alone. Furthermore, men older than 80 years seldom underwent radical prostatectomy, suggesting that in clinical practice urologists are sensitive to evidence-based guidelines18 for the care of elderly men and exercise discretion in the patients to whom they offer curative surgery.
We also observed high rates of radiotherapy delivery among patients, particularly older men, who were seen by radiation oncologists and urologists compared with those seen only by urologists. This finding is not easily interpretable. Urologists may recommend, based on a risk assessment of age and tumor features, that their patients seek a radiation oncologist evaluation if they are appropriate candidates for and are interested in radiation treatment and have access to radiotherapy facilities. However, radiation oncologists, once consulted, may strongly advocate radiotherapy.4,5
The variation in physician visit patterns seen in our study raises 2 key questions. Are all men with localized prostate cancer provided with balanced information on all the available options to treat their cancer? Should all men with localized prostate cancer seek opinions from different specialists or advice from their PCPs? Our study does not answer these questions, nor does it imply that evaluation by a single specialist prior to treatment leads to recommendations that are inappropriate or that the quality of care for these men is deficient. However, high rates of radiotherapy observed in men who saw both radiation oncologists and urologists vs those who only saw urologists and variation in the tendency of urologists' patients to visit radiation oncologists raise concerns that treatment decisions may be guided not only by objective evidence and patient preferences but also by specialty bias.
Multidisciplinary consultation with physicians who have particular expertise in prostate cancer, be they urologists, radiation oncologists, or medical oncologists, should ensure that men receive a balanced perspective on the risks and benefits of all available therapeutic options. However, the practical implications of such an initiative could be substantial in terms of the effect on health care expenditures and the operational efficiency of busy physician practices. Moreover, a radiation oncologist evaluation is not necessary for all men with localized prostate cancer, such as those who are expected to die of competing causes in fewer than 5 to 10 years, for whom the adverse consequences of local therapy outweigh the potential survival benefits obtained from intervention.19 Also, for men who have received a well-balanced consultation from a urologist and elect to have radical prostatectomy or expectant management, forgoing evaluation with a radiation oncologist seems reasonable.
Finally, although PCPs are well positioned to provide insight regarding life expectancy that may influence treatment choice,20 we found PCP visits to be infrequent between diagnosis and receipt of prostate cancer therapy and not associated with a greater likelihood of visits to other specialists. In addition, independent of patient age, comorbid conditions, or the type of prostate cancer specialist seen, men who saw PCPs were more likely to be followed expectantly. The reasons for higher rates of expectant management in men who saw PCPs compared with those who did not are not entirely clear from our data. It is possible that PCPs are more willing than specialists to consider expectant management or that those patients who are risk stratified as inappropriate candidates for aggressive intervention are referred back to their PCPs.
Several limitations of our study warrant mention. Our findings may not be applicable to men younger than 65 years. As an observational study, we cannot explain the frequency of specialist visits. Indeed, there is no normative standard against which to benchmark our results. It is not possible to identify the proportion of men who should be seen by radiation or medical oncologists or to determine whether the observed number of visits with specialists other than urologists is too low or too high. In addition, the potential for inaccurate coding exists for any analysis based on administrative claims.21,22 However, the unequivocal nature of a prostate cancer diagnosis and the financial incentive to code accurately for consultations and therapeutic interventions in this disease minimizes the potential for coding errors. Intangible factors that comprise decision making on the patient's part cannot be evaluated with SEER-Medicare data. Without beneficiaries' care preferences and refusal rates to see additional consultants, we cannot determine whether patients decided against radiation oncologist consultations despite urologist encouragement. Finally, although specialist visits could be determined, referrals cannot be ascertained reliably using claims.
Despite these limitations, our study demonstrates a strong association between the types of physicians men visit after a prostate cancer diagnosis and the eventual treatment strategy chosen. This finding and the known preferences of prostate cancer specialists for the treatment they themselves deliver4,5 underscores the need to ensure that all men are well informed and have access to balanced information prior to making this important treatment decision.
Correspondence: Thomas L. Jang, MD, MPH, Urologic Oncology Program, Division of Urology, The Cancer Institute of New Jersey, 195 Little Albany St, New Brunswick, NJ 08901 (firstname.lastname@example.org).
Accepted for Publication: September 24, 2009.
Author Contributions: Drs Jang and Bekelman contributed equally to this manuscript. Drs Jang, Bekelman, and Schrag had full access to all 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: Jang, Bekelman, Bach, Zelefsky, Scardino, and Schrag. Acquisition of data: Jang, Bekelman, Liu, and Schrag. Analysis and interpretation of data: Jang, Bekelman, Liu, Bach, Basch, Elkin, Zelefsky, Scardino, Begg, and Schrag. Drafting of the manuscript: Jang, Bekelman, Liu, Bach, Basch, Elkin, Zelefsky, Scardino, Begg, and Schrag. Critical revision of the manuscript for important intellectual content: Jang, Bekelman, Bach, Basch, Elkin, Zelefsky, Scardino, Begg, and Schrag. Statistical analysis: Jang, Bekelman, Liu, Bach, Begg, and Schrag. Obtained funding: Bekelman, Scardino, and Schrag. Administrative, technical, and material support: Bach, Scardino, and Schrag. Study supervision: Bach, Zelefsky, Scardino, and Schrag.
Financial Disclosure: None reported.
Funding/Support: This study was funded by a National Institutes of Health Ruth Kirchstein National Research Service Award (T32 CA 82088-07) (Dr Jang) and grants 1F32 CA 123964-01 (Dr Bekelman) and R21 CA98353 (Dr Schrag) from the National Cancer Institute.
Role of the Sponsor: The funding bodies had no role in the design, data extraction, and analyses; in the interpretation of the results and preparation of the manuscript; or in the decision to submit the manuscript for publication.
Additional Contributions: This study used the linked SEER-Medicare database. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute, the Office of Research, Development, and Information, Centers for Medicare & Medicaid Services, Information Management Services, Inc, and the SEER Program tumor registries in the creation of the SEER-Medicare database.