Care of long-term cancer survivors†‡
Physicians seen by Medicare enrollees surviving longer than 5 years
This study was based on data from the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development, and Information, Centers for Medicare and Medicaid Services; Information Management Services; and the SEER Program tumor registries in the creation of the SEER-Medicare database.
These findings and conclusions are solely the responsibility of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention, the Department of Health and Human Services, or the United States government.
Abstract
BACKGROUND:
Studies have shown that follow-up care for cancer patients differs by physician specialty, and that coordination between specialists and generalists results in better care. Little is known, however, regarding which specialties of physicians provide care to long-term cancer survivors.
METHODS:
The authors used Surveillance, Epidemiology, and End Results data from 1992 through 1997 that were linked to 1997-2003 Medicare data to identify persons diagnosed >5 years earlier with bladder, female breast, colorectal, prostate, or uterine cancer. Physician specialties were assigned by combining Medicare data with the American Medical Association Masterfile and the Unique Physician Identification Number Registry. The percentage of long-term survivors who visited physicians of interest was determined by analyzing Medicare outpatient claims submitted 6 to 12 years after initial diagnosis.
RESULTS:
Over the entire study period, 46% of female breast cancer survivors, 26% of colorectal cancer survivors, and 14% of prostate cancer survivors saw hematologists/oncologists. Radiation oncologists were seen by 11%, 2%, and 14% of breast, colorectal, and prostate cancer survivors, respectively. Survivors also sought care from specialists related to their cancer: 19% of breast cancer survivors had a cancer-coded visit with a surgeon, 26% of colorectal cancer survivors visited a gastroenterologist, and 68% of prostate cancer survivors visited a urologist. The percentage of survivors who visited cancer and cancer-related physicians declined each year. In contrast, nearly 75% of female breast, colorectal, and prostate cancer survivors saw primary care providers, and these percentages did not decrease annually.
CONCLUSIONS:
The findings of the current study underscore the need to include both primary care providers and cancer-related specialists in education and guidelines regarding cancer survivorship. Cancer 2009. © 2009 American Cancer Society.
Cancer survivors are a large and growing population in the United States. In 2006, >11 million people in the United States were living with diagnosed cancer, and 65% of them had been alive >5 years since the cancer was diagnosed.1 Recognizing long-term consequences of cancer and its treatment, clinicians, researchers, and advocates have directed attention toward identifying and minimizing the chronic effects and health risks that cancer survivors may experience.2, 3 Four national reports concerning medical, psychosocial, and public health issues related to cancer survivorship have been released since 2004 and stated that survivors need long-term follow-up care from knowledgeable physicians.4-7 Many physicians providing this follow-up care may need additional training in how best to prevent complications and promote health among their patients who have survived cancer.
Previous studies have shown that follow-up care for cancer survivors treated by oncology specialists is different from care for survivors treated by generalists, and that coordination between specialists and generalists resulted in better patient care.8, 9 For example, among women diagnosed with uterine cancer at least 5 years earlier, those who visited obstetricians/gynecologists or gynecologic oncologists were more likely to receive preventive services, including a mammogram and colorectal cancer screening, than were those who saw an oncologist or a primary care physician.10 Mammography and colonoscopy use was found to be higher among long-term colorectal cancer survivors who visited oncologists than among those who visited primary care providers.11, 12 Among women with breast cancer, those who continued to see oncologists for 3 years after their diagnosis were found to be more likely to receive mammography than those who did not, but they were also found more likely to receive surveillance tests considered inappropriate according to current recommendations.9
The care that generalists provide to cancer survivors is important, because a recent projection estimated that by 2020 the annual demand for oncology visits will surpass the capacity of trained oncologists by 9.4 to 15.0 million visits.13 Thus, the future supply of oncologists might be insufficient to provide continued care to all cancer survivors. Physicians in specialties other than oncology will likely have increased responsibility in the care of long-term survivors. However, the roles of generalists and specialists in the coordinated care of cancer survivors remain unclear.12
Patterns of care for cancer survivors have been studied for time periods up to the first few years after treatment.9, 14-16 However, to our knowledge, little is known regarding what specialties of physicians provide care to long-term cancer survivors after they reach their milestone of 5-year survival. The purpose of the current study was to characterize the specialties of providers who care for Medicare beneficiaries who survive more than 5 years after their initial cancer diagnosis. We chose to focus on physician utilization during this later time period because of the paucity of literature that describes care among long-term survivors beyond 5 years from diagnosis. We determined the percentages of long-term survivors who visited cancer specialists, specialists in fields related to their cancer, and primary care physicians. We also examined the relation of patient sociodemographic factors and cancer site to the continuation of care from cancer specialists. Our rationale behind this work was that recognizing the physicians who care for long-term survivors is a first step toward improving the lifetime care of these men and women.
MATERIALS AND METHODS
Data
To identify long-term cancer survivors and the physicians providing their care, we used linked data from the Surveillance, Epidemiology, and End Results (SEER) program and Medicare.17 This data set links clinical and demographic information including diagnosis date, cancer stage, patient's age at diagnosis, and patient's race/ethnicity from 11 population-based SEER cancer registries with hospital, physician, outpatient, home health, and hospice Medicare claims for people with cancer who are Medicare eligible. In a previous study, Medicare data were successfully linked to 94% of SEER registry patients aged ≥65 years.18 The Medicare data set we used was based on claims filed from January 1, 1997 through December 31, 2003, and was linked to SEER data for cancer cases diagnosed from January 1, 1992 through December 31, 1997. We also obtained data from the Unique Physician Identification Number (UPIN) Registry and the American Medical Association (AMA) Physician Masterfile to determine the specialties of physicians who were providing care.19, 20
Study Population
We used SEER data to identify 200,561 patients with a first diagnosis of invasive female breast, colorectal, prostate, uterine, or bladder cancer at age ≥60 years during the period 1992 through 1997 who had survived at least 5 years (hereafter referred to as long-term survivors). We excluded 36,809 long-term survivors with a history of multiple cancers or a known cancer recurrence, because they likely would have received more frequent follow-up care from oncology physicians than survivors without a new or recurrent cancer. The final cohort was comprised of 104,895 long-term survivors who were continuously enrolled in Medicare. We constructed the cohort by using Medicare eligibility information, to exclude those who 1) were enrolled in Medicare managed care plan at any time during 1997 through 2003 (n = 44,497); 2) were enrolled in Medicare before reaching age 65 years (n = 9372); 3) were not enrolled in both Medicare Part A and Part B (n = 5913); or 4) were not enrolled in Medicare when their 61st month of survival began (n = 2924). The final cohort was slightly larger than 200,561 minus the sum of all excluded long-term survivors, because some survivors met multiple exclusion criteria.
Patient and Tumor Characteristics
Sociodemographic variables were sex, age at diagnosis, race, and Hispanic ethnicity. We identified low-income survivors as those who had standard Medicaid coverage or were receiving state assistance for payment of Medicare premiums, deductibles, and coinsurance for at least 1 month during the study period of 1997 through 2003 (collectively known as “dual eligibles”). Rural-urban commuting area codes, which take into account population density, urbanization, and commuting flows, were used to indicate the survivors' residence at the time of diagnosis (metropolitan area, large town, small town, or rural area).21 Cancer stage at diagnosis was classified according to SEER summary staging.22 For prostate cancer, the localized and regional stages were combined in accordance with current reporting standards.21 We assessed comorbidity using Klabunde's modification of the Charlson comorbidity weights on outpatient diagnosis data on Medicare claims from the 6th survival year.23, 24
Physician Specialty and Visits
Because previous studies have shown that Medicare claims often do not identify cancer specialists,25, 26 we combined 3 data sources (Medicare claims, the UPIN Registry, and the AMA Masterfile) to determine the specialty of physicians who treated long-term survivors included in the study cohort. We considered physicians to be cancer specialists if they were so identified either in the AMA Masterfile or in both the Medicare claims and the UPIN Registry. If the identified specialties were discordant, the specialty classification in the AMA Masterfile took precedence, because the AMA does more active confirmation of specialty identification. If the cancer specialty was identified only in the Medicare claims or only in the UPIN Registry but not in the AMA Masterfile, the physician was considered misclassified, because the cancer specialization was identified in only 1 of the 3 available sources. We further restricted the identified cancer specialists to physicians most likely to have made clinical decisions regarding the care of cancer patients or who directly administered that care. Accordingly, we considered only physicians having Medicare claims with a radiation therapy Current Procedural Terminology (CPT) code (77000-77799) to be radiation oncologists. We also decided that physicians who were identified as specializing in pathology, blood banking/transfusion, dermatology, or anesthesiology likely had a supportive role in the oncological care of patients and did not consider them to be cancer specialists. Further discussion of the rules we used to determine whether to consider physicians to be cancer specialists and the effect of these rules on classifying physician specialties is detailed in a separate paper.27
We categorized cancer specialists as practicing gynecologic oncology, hematology/oncology (physicians identified as hematologists, oncologists, or both), musculoskeletal oncology, radiation oncology, surgical oncology, or multiple cancer specialties.
We defined cancer-related specialties differently for each study cancer based on whether the care provided in that specialty directly related to the cancer of interest. Thus, we considered the following physicians to be cancer-related specialists: 1) urologists for bladder or prostate cancer patients; 2) colorectal surgeons and gastroenterologists for colorectal cancer patients; 3) obstetrician/gynecologists for uterine cancer patients; and 4) general surgeons for female breast, colorectal, and uterine cancer patients who had an evaluation and management visit for which a cancer-related diagnosis was coded on at least 1 of their Medicare outpatient claims (3-character International Classification of Diseases-9 diagnosis codes 140 through 239, V10, and V76).
Primary care providers included family practice, general medicine, internal medicine, geriatric medicine, and preventive medicine physicians.
Because physicians could have >1 specialty indicated in each of the data sources, we hierarchically assigned 1 specialty to each physician in the following order of precedence: cancer specialist, cancer-related specialist, other surgical specialist, other medical specialist, primary care provider, and all other. Except for the cancer and cancer-related specialties, the hierarchy of specialties was based on the number of years of training required.
Visits were defined as an office or other outpatient evaluation and management claim (CPT codes 99201-99205 and 99211-99215) for care that patients received at least 5 years after their initial diagnosis of cancer.
Statistical Analysis
We first described the sociodemographic characteristics, cancer stage at diagnosis, and number of comorbidities of the long-term cancer survivors, both overall and by type of cancer. We then calculated the percentage of long-term survivors who had at least 1 visit with a physician of interest at any time during the study period (6th to 12th survival year) for all study cancers, and then annually by years since diagnosis for breast, colorectal, and prostate cancer. The denominators included men and women who were alive at any point during the specified time since diagnosis. The denominators for survival years further from diagnosis included fewer survivors because of death, censoring, or transition to institutionalized care. We used chi-square tests to determine whether the percentage of survivors who had at least 1 office visit to various specialists differed significantly by patient characteristics. The differences among percentages by each characteristic were judged to be statistically significant if the P value was <.001. We also described the mean number of visits each year to physician specialty groupings for survivors in the 6th to 12th year from their diagnosis.
RESULTS
The mean age of the 104,895 long-term cancer survivors was 71.7 years. Approximately 86.9% were white and 3.6% were Hispanic, 11.8% were considered to be “low income” for at least 1 month during the study period, and 85.3% lived in a metropolitan area at the time of diagnosis (Table 1). The most common types of cancer among them were prostate cancer (n = 47,954), female breast cancer (n = 26,972), and colorectal cancer (n = 16,671). As would be expected in a population of long-term cancer survivors, most cancers were diagnosed at the localized and regional stage. Most survivors had either no (71.4%) or no more than 2 comorbidities (20.0%) during the 6th year after their cancer diagnosis.
| Characteristic | Total (n=104,895), % | Type of Cancer | ||||
|---|---|---|---|---|---|---|
| Female Breast (n=26,972), % | Colorectal (n=16,671), % | Bladder (n=7429), % | Prostate (n=47,954), % | Uterine (n=5869), % | ||
| Female | 42.4 | 100 | 56.4 | 29.6 | — | 100 |
| Age at diagnosis, y | ||||||
| 60-64 | 17.2 | 19.2 | 14.0 | 15.4 | 16.9 | 21.7 |
| 65-69 | 24.8 | 23.0 | 20.5 | 23.1 | 27.5 | 24.9 |
| 70-74 | 25.5 | 23.1 | 22.9 | 24.9 | 28.0 | 24.2 |
| 75-79 | 18.4 | 18.2 | 20.1 | 19.5 | 17.8 | 17.6 |
| ≥80 | 14.2 | 16.5 | 22.5 | 17.2 | 9.9 | 11.7 |
| Race | ||||||
| White | 86.9 | 89.1 | 86.1 | 92.5 | 84.3 | 93.1 |
| Black | 6.7 | 5.0 | 5.8 | 2.6 | 9.0 | 3.1 |
| Other | 6.2 | 5.7 | 7.9 | 4.6 | 6.5 | 3.6 |
| Unknown | 0.2 | 0.2 | 0.3 | 0.3 | 0.2 | 0.2 |
| Ethnicity | ||||||
| Non-Hispanic | 95.4 | 96.1 | 96.0 | 96.7 | 94.4 | 96.3 |
| Hispanic | 3.6 | 3.4 | 3.6 | 2.5 | 3.9 | 3.3 |
| Unknown | 1.1 | 0.6 | 0.4 | 0.8 | 1.7 | 0.5 |
| Low income | 11.8 | 14.6 | 16.1 | 11.5 | 8.7 | 12.2 |
| Residence at time of diagnosis | ||||||
| Metropolitan area | 85.3 | 85.4 | 83.7 | 84.5 | 85.9 | 85.2 |
| Large town | 5.8 | 5.7 | 6.4 | 7.1 | 5.5 | 5.5 |
| Small town | 7.3 | 7.2 | 8.2 | 7.0 | 7.2 | 7.4 |
| Rural area | 1.6 | 1.7 | 1.7 | 1.4 | 1.5 | 1.9 |
| SEER summary stage at diagnosis | ||||||
| Localized | 39.0 | 75.2 | 55.4 | 86.1 | — | 85.8 |
| Regional | 13.2 | 21.8 | 39.5 | 10.5 | — | 10.0 |
| Distant | 1.8 | 1.2 | 2.6 | 0.3 | 2.1 | 2.1 |
| Localized/regional (prostate only) | 40.5 | — | — | — | 88.6 | — |
| Unstaged | 5.5 | 1.8 | 2.6 | 3.2 | 9.4 | 2.1 |
| Charlson Index at 6th survival y | ||||||
| 0 | 71.4 | 72.7 | 69.3 | 65.9 | 71.9 | 73.4 |
| 1-2 | 20.0 | 19.7 | 20.9 | 22.7 | 19.6 | 19.4 |
| ≥3 | 8.6 | 7.6 | 9.8 | 11.4 | 8.5 | 7.2 |
- SEER indicates Surveillance, Epidemiology, and End Results program.
During the 6th to 12th year since their cancer diagnosis, of the 94.2% of the survivors who had an office evaluation and management visit with a physician of interest, 32.9% visited a cancer specialist, mainly a hematologist-oncologist (23.6%) or a radiation oncologist (9.9%) (Table 2). Visits to hematologists/oncologists were more common among survivors with breast (45.6%) and colorectal (26.3%) cancer than among those with prostate, bladder, or uterine cancer (13.9%, 11.1%, and 10.1%, respectively). A radiation oncologist was seen at least once by 13.6% of prostate cancer survivors and 11.3% of breast cancer survivors. Surgical and gynecologic oncologists were seen by 3.5% or fewer survivors of all categories, except for uterine cancer survivors, of whom 8.0% had at least 1 visit to a gynecologic oncologist. Overall, 48.4% of long-term survivors had a claim for visiting a cancer-related specialist. The type of specialist seen most frequently varied by cancer type; 19.3% of breast cancer survivors saw a general surgeon who coded for a cancer-related visit, 25.9% of colorectal cancer survivors saw a gastroenterologist, 62.1% of bladder and 67.8% of prostate cancer survivors saw a urologist, and 34.9% of uterine cancer survivors saw an obstetrician/gynecologist. Overall, the percentages of long-term survivors who saw a primary care physician or a noncancer-related medical specialist were 74.2% and 72.8%, respectively; whereas the percentages who saw cancer or cancer-related specialists were 32.9% and 48.4%.
| Physician Category* | Total (n=98,803), % | Type of Cancer | ||||
|---|---|---|---|---|---|---|
| Female Breast (n=25,543), % | Colorectal (n=15,343), % | Bladder (n=6954), % | Prostate (n=45,382), % | Uterine (n=5581), % | ||
| Cancer specialist† | 32.9 | 54.0 | 28.4 | 13.4 | 26.6 | 23.9 |
| Hematologist/oncologist | 23.6 | 45.6 | 26.3 | 11.1 | 13.9 | 10.1 |
| Radiation oncologist | 9.9 | 11.3 | 2.2 | 1.8 | 13.6 | 5.5 |
| Surgical oncologist | 1.9 | 3.5 | 1.0 | 1.2 | 1.6 | 0.8 |
| Gynecologic oncologist | 0.8 | 0.9 | 0.2 | 0.3 | 0.0 | 8.0 |
| Cancer-related specialist | 48.4 | 19.9 | 36.6 | 62.2 | 67.8 | 36.8 |
| General surgeon (cancer-coded claim) | 6.4 | 19.3 | 7.9 | — | — | 2.6 |
| Urologist | 35.5 | — | — | 62.1 | 67.8 | — |
| Colorectal surgeon | 0.9 | — | 6.1 | — | — | — |
| Gastroenterologist | 4.0 | — | 25.9 | — | — | — |
| Obstetrician-gynecologist | 2.0 | — | — | — | — | 34.9 |
| Primary care provider | 74.2 | 76.0 | 72.7 | 73.8 | 73.5 | 76.8 |
| Medical specialist | 72.8 | 70.7 | 65.9 | 75.1 | 76.0 | 72.7 |
- * Hematologist/oncologist refers to physicians identified as a hematologist, oncologist, or both. Primary care provider refers to family practice, internal medicine, or generalist physicians. Medical specialists include all subspecialties of medicine with the exception of gastroenterologists, who are considered to be cancer-related specialists for colorectal cancer survivors.
- † Musculoskeletal oncologists and physicians with multiple oncology specialties were not shown, because they were seen by <1% of cancer survivors.
Figure 1 shows, by physician specialty and years since diagnosis, the percentages of long-term breast, prostate, and colorectal cancer survivors who had a physician evaluation and management visit during Years 6 to 12 after diagnosis. Overall, 56.8% to 64.9% of these survivors saw a primary care physician, and 49.2% to 61.8% saw a non–cancer-related medical specialist at least once in any given year; these percentages did not decline until Year 12. Among women with breast cancer, 41.3% visited a hematologist-oncologist during their 6th year of survival; however, this annual percentage had fallen to 20.2% by the 12th year. The percentages of long-term colorectal cancer survivors who saw a hematologist-oncologist were 23.8% in Year 6, declining to 10.7% in Year 12; for prostate survivors, the percentages in all years from diagnosis were <9.0%. The percentage of prostate cancer survivors who visited a urologist fell from 59.1% during the 6th year to 39.2% during the 12th year since diagnosis. The percentage of survivors who visited radiation oncologists ranged from 10.7% at Year 6 to 3.4% at Year 12 among those with prostate cancer, from 9.2% to 2.1% among those with breast cancer, and from 1.4% to 0.4% among those with colorectal cancer. The percentage of colorectal cancer patients who visited a gastroenterologist fell from 15.1% in Year 6 to 9.6% in Year 12.
Percentage of long-term cancer survivors who had at least 1 visit annually with a physician of interest is shown by year since diagnosis in (a) breast cancer long-term survivors, (b) colorectal cancer long-term survivors, and (c) prostate cancer long-term survivors.
As shown in Table 3, the percentage of survivors with claims for cancer specialist visits was higher among women, and the percentage with claims for cancer-related specialist visits was higher among men. This finding is because women with a history of breast cancer sought care from oncologists who were classified as cancer specialists; whereas men with prostate cancer sought care from urologists who were classified as cancer-related specialists. Overall, the percentages of long-term cancer survivors who saw cancer specialists and cancer-related specialists declined by age group. The percentages with visits to primary care providers and other medical specialists did not differ as much by age. The percentage of survivors who saw cancer-related specialists was higher among blacks than among whites; however, no differences were found in the percentages who sought cancer specialists by race. Among Hispanic survivors, the percentages who saw primary care and other medical specialists 5 years past a cancer diagnosis were lower than for non-Hispanic survivors. Low income survivors were less likely than those not needing state assistance with Medicare coverage to have visited physicians in every category. Residents of metropolitan areas at the time of diagnosis were more likely than large town, small town, and rural residents to have visited cancer- and non–cancer-related specialists. Survivors whose cancer was diagnosed at a distant stage were more likely to have visited cancer specialists than were those whose cancer was diagnosed at the localized stage. Survivors with ≥2 comorbid conditions were less likely to have visited cancer physicians, but more likely to have visited noncancer specialists than were those with fewer comorbid conditions.
| Characteristic | No. of Long-Term Cancer Survivors Who Visited a Physician | Physician Specialty* | |||
|---|---|---|---|---|---|
| Cancer Specialist, % | Cancer-Related Specialist, % | Primary Care Physician, % | Noncancer Medical Specialist, % | ||
| All | 98,799 | 32.9 | 48.4 | 74.2 | 72.8 |
| Sex† | |||||
| Men | 56,968 | 25.9 | 64.1 | 73.2 | 75.2 |
| Women | 41,831 | 42.4 | 27.1 | 75.6 | 69.7 |
| Age at diagnosis, y† | |||||
| 60-64 | 17,277 | 38.2 | 49.8 | 76.0 | 72.9 |
| 65-69 | 24,994 | 36.5 | 52.9 | 74.9 | 75.3 |
| 70-74 | 25,532 | 34.4 | 51.2 | 74.8 | 75.3 |
| 75-79 | 18,093 | 30.4 | 46.7 | 73.7 | 72.7 |
| ≥80 | 12,903 | 19.4 | 34.9 | 70.1 | 63.2 |
| Race‡ | |||||
| White | 86,296 | 32.8 | 47.9 | 74.5 | 73.8 |
| Black | 6228 | 33.7 | 52.6 | 72.3 | 63.0 |
| Other | 6095 | 33.3 | 51.9 | 72.7 | 69.5 |
| Unknown | 180 | 25.6 | 41.7 | 75.0 | 68.9 |
| Hispanic origin† | |||||
| Non-Hispanic | 94,311 | 33.1 | 48.4 | 74.6 | 73.0 |
| Hispanic | 3442 | 31.2 | 46.8 | 67.7 | 67.0 |
| Unknown | 1046 | 23.2 | 57.1 | 64.6 | 78.6 |
| Low income† | |||||
| No | 88,198 | 33.5 | 49.9 | 74.5 | 73.7 |
| Yes | 10,601 | 27.4 | 36.7 | 71.7 | 65.2 |
| Residence at time of cancer diagnosis† | |||||
| Metropolitan area | 84,188 | 33.8 | 49.2 | 73.6 | 74.4 |
| Large town | 5771 | 28.1 | 48.0 | 76.9 | 65.3 |
| Small town | 7267 | 26.7 | 41.5 | 78.2 | 63.2 |
| Rural area | 1573 | 28.6 | 39.5 | 78.8 | 61.9 |
| SEER summary stage at diagnosis† | |||||
| Localized | 38,636 | 32.6 | 32.8 | 76.0 | 71.5 |
| Regional | 12,852 | 53.5 | 30.2 | 72.2 | 66.9 |
| Distant | 1685 | 54.5 | 44.5 | 66.2 | 62.2 |
| Localized/regional (prostate) | 40,292 | 26.3 | 67.7 | 73.6 | 76.3 |
| Charlson comorbidity index† | |||||
| 0 comorbidities | 70,548 | 33.9 | 49.0 | 74.4 | 71.5 |
| 1 comorbidity | 19,919 | 31.0 | 48.0 | 75.4 | 75.2 |
| ≥2 comorbidities | 8332 | 29.0 | 45.2 | 70.4 | 78.7 |
- SEER indicates Surveillance, Epidemiology, and End Results program.
- * Cancer specialists were defined as gynecologic oncologists, hematologists, oncologists, musculoskeletal oncologists, radiation oncologists, or surgical oncologists. Cancer-related specialists were defined as urologists for patients with bladder or prostate cancer; colorectal surgeons and gastroenterologists for patients with colorectal cancer; obstetrician/gynecologists for patients with uterine cancer; and general surgeons for patients with cancers of the female breast, colorectum, and uterus, as long as the survivors had at least 1 cancer-coded claim. Primary care providers were defined as family practice, internal medicine, or generalist physicians. Medical specialists were defined as practitioners of all other subspecialties of medicine except gastroenterology, which was considered to be a cancer-related specialty for colorectal cancer survivors.
- † Differences among percentages by characteristic had a significance level of P < .001 for each physician specialty grouping.
- ‡ Differences in percentages by race were significant (P < .001) for visits to cancer-related specialists, primary care physicians, and noncancer medical specialists, but not for cancer specialists (P > .07).
As shown in Figure 2, the average annual number of all evaluation and management visits to physicians for all study cancers combined declined from 9.1 in the 6th year to 7.9 in the 11th year of survival, whereas the average number of visits per year to primary care specialists and noncancer specialists remained fairly stable until the last study year. The decrease in the number of visits to cancer and cancer-related specialists impacted the overall decrease in evaluation and management visits by year. Because men with prostate cancer comprised 46% of our study cancers, we also did a separate analysis of their visits (data not shown). The mean number of evaluation and management visits among prostate cancer survivors was higher each year compared with the overall results. This finding was because of a higher mean number of visits to cancer-related specialists, mainly urologists (1.6 visits in Year 6, 1.4 visits in Years 7-9, and 0.7 visits in Year 12). The mean number of visits to primary care providers was the same each year for all survivors and men with prostate cancer. Visits to cancer specialists for prostate cancer survivors were fewer than the average number of visits among all survivors in the earlier years since diagnosis (0.8 visits in Years 6, 7, and 8), but similar in the remaining years.
The mean number of outpatient visits per year among long-term cancer survivors is shown by total and by type of physician. *Cancer specialists were defined as gynecologic oncologists, hematologists, oncologists, musculoskeletal oncologists, radiation oncologists, or surgical oncologists.
DISCUSSION
Overall, we found that only approximately one‒third of cancer survivors continued to seek care from physicians whose specialties were related to their original cancer at any point after reaching the milestone of 5-year survival. Breast cancer survivors continued to visit with hematologists/oncologists and other cancer specialists more than any other group of long-term cancer survivors, and >67% of long-term prostate cancer survivors continued visiting urologists. Much higher percentages of survivors received care from primary care providers and non–cancer-related medical specialists than from cancer and cancer-related specialists. Furthermore, whereas the percentage of long-term cancer survivors who received care each year from cancer and cancer-related specialists decreased each year out from the 6th survival year, the percentage who visited primary care providers and non–cancer-related medical specialists did not.
We also found that age, Hispanic ethnicity, rural residence, low income, and multiple comorbidities were associated with a decreased likelihood of cancer survivors receiving continued care from cancer specialists. Possible explanations for this finding include assumption of care by primary care providers, less access to specialized care, and competing health demands among survivors in these categories.
In the future, we would like to further explore how the type of physician seen by long-term cancer survivors relates to outcomes, including the use of preventive care, the incidence of late effects, the incidence of secondary and recurrent cancers, and ultimately, survival. Results from a previous study indicated no differences in the rate of cancer recurrence between survivors treated by family physicians and those treated by oncologists.28 Similar results from additional studies would help reassure physicians and survivors that noncancer specialists can provide follow-up care to cancer survivors without compromising the quality of care.
The findings of the current study highlight that primary care providers have a central role in providing care for cancer survivors, especially in years further out from diagnosis. This trend was observed by Snyder et al, who found that, during the first year after completion of active cancer treatment, cancer patients were being seen by both oncologists and primary care providers.14 Then, between the 1st and 5th year after cancer diagnosis, the percentage of survivors who visited oncologists decreased, whereas the percentage who saw only primary care providers increased.15 We showed that this trend continued from the 6th to the 11th year of cancer survival.
The implication of the findings of this study is that physicians who care for long-term survivors need up-to-date guidance concerning appropriate follow-up and preventive measures, especially as the evidence base for what constitutes appropriate long-term surveillance of cancer survivors expands. The 2006 Institute of Medicine report, From Cancer Patient to Cancer Survivor: Lost in Transition, detailed many physical, psychosocial, and practical issues confronting long-term survivors and suggested approaches to improve care, such as cancer treatment summaries to facilitate better coordination of care for survivors.5 Related to this Institute of Medicine report, the American Society of Clinical Oncologists has developed resources, such as treatment summary templates, to help with the transition to survivorship once active treatment is completed.29 It is unknown whether and to what extent physicians providing care to long-term survivors, particularly primary care providers and urologists, are aware of and use these resources. Our findings could be useful to direct efforts to improve survivorship care toward the most relevant physician specialty groups.
A major strength of the current study was that we combined data from 3 sources to more accurately classify the specialties of physicians treating cancer survivors. To our knowledge, no previous study has used this amount of combined data to limit potential misclassification. The SEER-Medicare data, which capture 17% of all cancer patients in the United States, are population-based, valid, and representative of different areas across the county. In addition, we characterize physicians seen by cancer patients whose cancer had been diagnosed 6 to 12 years previously, a longer cancer survival period than most studies among cancer survivors. In addition, because we focused on cancers with high 5-year survival rates, our results reflect the physicians seen by of a majority of cancer survivors. Among all US cancer survivors, 66% had a type of cancer included in our study.1
The current study also had several limitations. One limitation was that despite our best attempt to characterize the specialty of physicians treating cancer survivors by combining multiple data sources, there is no way to ascertain the actual specialty of individual physicians. Many subspecialists practice general medicine and, conversely, generalists may focus on a particular clinical area without formal fellowship training.25, 30 In addition, distinguishing between general and oncologic surgeons or between gynecologists and gynecologic oncologists has been shown to be especially difficult, because any of these specialists can choose to focus exclusively on cancer patients or not.31 Furthermore, specialty data from Medicare claims and the UPIN Registry rely exclusively on self-reports, and although AMA Masterfile data on physicians' specialties are derived in part from specialty boards, they are also based in part on physician surveys. Thus, the percentages reported may not accurately reflect the particular specialists seen, and we cannot assess whether there was potential under- or overestimation based on the available data. A second study limitation is that 46% of the study cohort had prostate cancer. Therefore, the overall results presented are heavily influenced by the findings for men diagnosed with prostate cancer. The role of urologists, who are classified as cancer-related specialists in caring for prostate cancer survivors, accounts for the high percentages of overall, male, and nonwhite survivors who visited with cancer-related specialists. A third limitation is that although we excluded survivors with a known secondary cancer from our study because they would be more likely to seek care from a cancer specialist, SEER-Medicare data do not identify all survivors in whom recurrence or metastasis has occurred.32 Therefore, some of the long-term survivors in our study who visited cancer specialists and cancer-related specialists may have been receiving care related to a recurrence. In addition, we included visits near the end of life, a period known for increased healthcare utilization and not reflective of routine care.33, 34 A fourth limitation is that, in Year 12 of this study, the number and percentages of long-term survivors who had visits were less than expected. Our results may have been affected in this latter year by attrition because of death, data censoring, and transitioning to institutional or hospice care, which does not generate “allowed” Medicare evaluation and management visits. Finally, Medicare claims-based data represent only Americans aged ≥65 years who are not enrolled in Medicare managed care organizations and only capture services billed, which may not reflect actual care.
In conclusion, long-term cancer survivors have more contact with primary care physicians and other non–cancer-related providers than with cancer specialists or cancer-related specialists. As a result, non–cancer-related providers need to understand their patients' cancer history and treatment as well as the potential late effects of cancer and how to optimize the health and well-being of cancer survivors. Findings from this study provide necessary background information for subsequent studies of the outcomes of long-term cancer survivors by the specialty of their physicians. Efforts to improve the health and well-being of long-term cancer survivors such as the development and use of personalized survivorship care plans should be fully supported and disseminated through professional education to physicians of all specialties, especially primary care providers and urologists. We hope that our results will encourage partnerships with appropriate medical specialty organizations to develop and disseminate education and interventions designed to improve the care of long-term cancer survivors.
Conflict of Interest Disclosures
Supported by contract 200-2002-00,575 from the Centers for Disease Control and Prevention to RTI International.
