- © 2008 by American Society of Clinical Oncology
Current and Future Utilization of Services From Medical Oncologists
- From the Applied Research Program, and the Surveillance Research Program, National Cancer Institute, Bethesda; and Information Management Services, Silver Spring, MD
- Corresponding author: Joan L. Warren, PhD, Health Services and Economics Branch/Applied Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute, Executive Plaza North, Rm 4005, 6130 Executive Blvd, MSC 7344, Bethesda, MD 20892-7344; e-mail: joan_warren{at}nih.gov
Abstract
Purpose There is limited information on the current use of oncologists and projections of future need. This analysis assesses current utilization patterns and projects the number of people with cancer and their use of oncologists’ services through 2020.
Methods Data from the Surveillance, Epidemiology, and End Results cancer registries and Medicare physician claims were used to estimate oncologists’ services from 1998 to 2003. We estimated the portion of patients with cancer who saw an oncologist, the mean number of visits, and the clinical setting where care was provided. Care was divided into initial, continuing, and last-year-of-life phases. Projections for future number of patients with cancer and visits were calculated by applying incidence and prevalence rates derived from Surveillance, Epidemiology, and End Results data to census population projections through 2020.
Results The percentage of patients who saw an oncologist was 47% during the initial-care phase, 36% during the continuing-care phase, and 70% in the last year of life. The number of visits varied by age, sex, cancer site, and phase. The total number of cancer patients in the United States is projected to increase 55%, from 11.8 million in 2005 to 18.2 million in 2020. Total oncology visits are projected to increase from 38 million in 2005 to 57 million in 2020.
Conclusion Utilization of oncologists’ services will increase appreciably between 2005 and 2020; this will be driven predominantly by an increase in survivors of cancer and by the aging of the population. The United States may face an acute shortage of medical oncologists if efforts are not taken to meet this growing need.
INTRODUCTION
Cancer incidence rates in the United States have declined significantly from 1995 to 2004.1 Although the incidence rate is decreasing, the number of people newly diagnosed with cancer is expected to increase because of population growth and the aging of the US population.1 Survivors of cancer are expected to increase in number as survival time for patients with cancer has lengthened through improved treatment and earlier detection. A consequence of these changes in demographics and survivorship is that the total number of people living with cancer in the United States is expected to grow significantly in the future.
Medical oncologists will play a major role in providing care for many of these patients. A recent study by the Association of American Medical Colleges2 found that the current supply of oncologists is adequate; however, the authors project an acute shortage by 2020. A primary component that will drive the need for medical oncologists is the number of people with cancer who are in need of care. To date, little data have been published on the current use of medical oncologists or on forecasts regarding future utilization. In this analysis, we provide data on the current utilization patterns of medical oncologists’ services and provide projections of the number of people with cancer and their use of oncologists’ services through 2020.
METHODS
Data Sources
The analysis combined data from three sources: cancer registry data from the Surveillance, Epidemiology and End Results (SEER) program maintained by the National Cancer Institute (NCI), Medicare data, and the American Medical Association (AMA) Physician Masterfile Data. The SEER registries collect population-based clinical information about incident cancer patients for people in defined geographic areas. The SEER areas have changed with time and range from nine registries in 1975 to 1991, which included 10% of the US population, to an expansion of 13 registries in 2000, which included 25% of the US population. The SEER data for each patient contain every occurrence of a primary incident cancer; the month and year of diagnosis; the cancer site; the stage of disease at diagnosis; and follow-up vital status, including the month and year of death and cause of death for people who died. We excluded patients ascertained by death certificate or autopsy or those with an unknown month of diagnosis.
People reported to SEER have been matched against Medicare's data, as described elsewhere.3 For people reported to SEER who were age 65 years or older, 94% have been linked to Medicare's master enrollment file. The SEER-Medicare data used for this analysis included people diagnosed with cancer from 1973 to 2002 and their Medicare claims through 2003.
The Medicare data contain claims for all Medicare covered services for beneficiaries with fee-for-service coverage. Claims are not available for HMO enrollees, who accounted for 22.6% of all patients with cancer in our analysis. The physician claims included a unique number for each physician (UPIN) and a self-reported provider specialty.
Current Utilization of Oncologists’ Services
We used physician claims in the SEER-Medicare data for patients age 65 years and older to estimate current utilization of oncologists’ services, defined as services during 1998 through 2003. Oncologists were identified by linking the UPIN from the physician claim to the AMA Physician Masterfile, which has information about each physician's primary and secondary specialty. If the physician specialty could not be determined from the AMA data, the self-reported specialty from the Medicare data was used.
To account for the varying intensity of services throughout the course of cancer, we classified health care services into three mutually exclusive phases: the incident phase that captured care for 12 months from the month of diagnosis; the last year of life for people who died as a result of cancer; and continuing care, which included all care between the initial year and in the last year of life. The incident phase had the most current patients, which included people in the SEER-Medicare data who were diagnosed with cancer in 2002. We removed people who had more than one primary cancer, as it would not be possible to determine if the care was for a new or previously diagnosed cancer; 23% of patients diagnosed in 2002 had multiple cancers. The last-year-of-life phase included people diagnosed in 1998 who died as a result of cancer between 1998 and 2003. Their claims were reviewed from the time of death retrospectively for up to 12 months, although hospice care was excluded, as it was not possible to determine which doctor provided care. The continuing-care phase included services for up to 5 years for people diagnosed in 1998 if they were not included in phases of initial care or the last year of life. Patients were required to have Medicare Part A and B and fee-for service coverage for each phase, so that claims were available for the services provided during the phase.
We determined the number and percentage of patients who had an oncologist visit by identifying at least one Medicare bill submitted by a medical oncologist in each phase. Rates of use of oncology services are presented by sex and age and for selected cancer sites (ie, breast, colorectal, lung, and prostate). These four sites account for 60% of all cancers diagnosed in elderly patients in the United States.1 We calculated the mean number of visits for those patients who saw an oncologist. We determined, for specific cancers, the impact of chemotherapy use on the frequency of oncology visits by calculating the mean number of oncology visits for those patients who did not receive chemotherapy separately from those who did. Chemotherapy included those services covered by Medicare; most oral agents were not included. We estimated the mean number of visits for chemotherapy administration for the chemotherapy group separately from visits for other reasons. Finally, we evaluated the place of service where oncologists were providing care by using the place-of-service code on the physician claim. The place of service was divided into three, mutually exclusive categories: office/clinic visits; inpatient contacts; and other visits. A list of codes used in this analysis and their definitions are in Appendix Table A1 (online only).
Estimation of the Future Number of Incident and Prevalent Patients With Cancer
To project the future number of incident and prevalent patients with cancer, we needed longitudinal data to improve our estimates of cancer prevalence. Therefore, we used a subset of the registries included in the current utilization estimates, which were those nine registries that have participated in the SEER program since 1975. We identified patients reported by these registries between 1975 and 2002 to obtain age- and sex-specific cancer rates of tumors.4 We used information in the first cancers for people with multiple cancers. Projections of future incidence rates were derived by applying a 3-year average of the most recent age- and sex-specific incidence rates from the SEER data (2000 to 2002) to US Census Bureau population projections to estimate the annual number of new patients with cancer in the United States.5 To determine the number of prevalent patients in the United States through 2020, projections of the new patients with cancer and of cancer survival rates were used to estimate the number of prevalent patients. Prevalence was estimated for 5-year age groups through age 84 years by time from diagnosis. In people age 85 years and older, prevalence was estimated by applying the proportions for those age between 80 and 84 years to the US population age ≥ 85 years.
The time since diagnosis was used to classify patients into the different phases of care. The number of people in the initial year was calculated by subtracting people who died in a given year from the number of incident patients. The number of person years for the continuing phase was calculated by subtracting people in the initial and last-year-of-life phases from the total number of prevalent patients. The number of people in the last-year phase was determined by estimating the number of new patients and the number of prevalent patients that died as a result of cancer in a given year. Patients who survived less than 24 months after diagnosis contributed up to 12 months of utilization data to the last-year phase; additional months of data then were included in the initial phase. A detailed description of the methodology is given elsewhere.6 We combined the population projections with estimates of the current use of oncologists’ services to produce phase-specific projections in 5-year increments for the total annual visits.
RESULTS
The assessment of the current use of oncologists’ services included 117,776 people in all phases (Table 1). The portion of patients who saw an oncologist varied by phase: 47% of people in the initial-care cohort, 36% of people in the continuing-care phase, and 70% in the last year of life. In both the initial- and continuing-care phases, women were approximately twice as likely to see an oncologist as were men, although the use of oncology services during the last year of life was comparable between men and women (ie, 69% and 71%, respectively). We used the χ2 test to assess the difference in the percentage of patients who saw an oncologist by age group. There was a significant difference between age groups (P < .0001), which reflected the lower use of oncologists for people age 85 years and older.
Utilization of Oncologists’ Services Among Medicare Beneficiaries Diagnosed With Cancer in the SEER Areas by Phase of Care
Phase-specific utilization of oncologists’ services varied by cancer site. During the initial-care phase, 75% of those with breast cancer saw an oncologist, contrasted with 13% those with prostate cancer (Table 2). During the initial phase, use of oncologists varied by stage; people most likely to require chemotherapy (ie, those with advanced stage) were most likely to see an oncologist (Table 3). Site-specific differences persisted through the continuing-care phase, which ranged from 16% of men with prostate cancer who saw an oncologist to 55% of women with breast cancer. In the last year of life, variation in the percentage of patients who saw an oncologist ranged from 53% for patients with prostate cancer to 75% for patients with breast cancer.
Utilization of Oncologists’ Services Among Medicare Beneficiaries Diagnosed With Cancer in the SEER Areas by Phase of Care
Percent of Patients Who Saw an Oncologist During the Initial Phase
Among the people who saw an oncologist, the mean number of visits to an oncologist varied by cancer site. When people who received chemotherapy were excluded, though, the mean number of oncology visits was similar between cancer sites. During all phases, patients who received chemotherapy had a much higher mean number of visits than did patients who did not receive chemotherapy. In patients who received chemotherapy, many oncologist visits were for drug administration. However, visits for reasons other than chemotherapy were frequent as well. During the last year of life, the mean number of visits increased for all cancers and all patients, although the mean number of visits for patients who received chemotherapy (33.8) was markedly higher than for those patients who did not (10.5). We reviewed the diagnoses from oncology visits during the continuing-care period and during the last year of life and found that, for each phase, greater than 80% of the diagnoses were for the cancer, the treatment, conditions related to cancer progression, or complications from treatment (data not shown).
The proportion of oncology care provided in different settings varied by phase (Fig 1). During the initial- and continuing-care phases, the overwhelming majority of services occurred in the office/clinic (82% and 87%, respectively). In the last year of life, office/clinic visits declined to 55% of all services. Visits to patients in the hospital accounted for 18% of all bills during the initial-care phase, for 12% during the continuing-care phase, and for 41% during the last year of life.
Percent of oncologists’ services for Medicare beneficiaries by place of service and phase of care, 2002.
The number of people of all ages who live in the United States and who have cancer is projected to grow from 11.8 million in 2005 to 18.2 million in 2020 (Table 4), which is a 55% increase. The estimated increase in the number of patients during this 15-year interval will vary by phase: 35% in the initial phase, 57% in the continuing-care phase, and 48% the last year of life. The total number of oncology visits is projected to increase from 38 million in 2005 to 58 million in 2020. Figure 2 shows the increase in the total annual oncology visits by phase of care. The use of oncology services will be driven predominantly by people in the continuing-care phase, as the total visits will increase from 29 million in 2005 to 45 million in 2020.
Total annual oncology visits by phase of care, 2005 to 2020.
Projections for the Number of Patients With Cancer and Oncologist Visit Data for 2005-2020
DISCUSSION
The findings from this study demonstrate that there will be a significant increase in the utilization of medical oncologists’ services between 2005 and 2020. Our projection of a 50% increase in the number of annual visits is slightly lower than a 55.8% increase in demand reported in an analysis by Cooper7 that was sponsored by the American Society of Clinical Oncology. Cooper's approach differs from ours, as he tied utilization of physician services to economic trends, included different trends in the utilization of primary care physicians and specialists during the past 50 years, and used a slightly higher estimate of population growth.
Estimates of future demand are based on a number of assumptions. Projections through 2020 assume constant incidence and are based on the most current 3 years of available incidence data. If there are changes in current modes of detection or risk factors, these could cause incidence to change markedly. These impacts could occur in the near term (eg, the role of mammography in the substantial increase in the incidence of ductal carcinoma in situ). Increased detection could impact survivorship by increasing lead time and by extending life expectancy. Changes in treatment also could increase the number of cancer survivors.
In our analysis of current use of services, less than half of patients with newly diagnosed cancer in 2002 saw a medical oncologist, likely because there was no indication for chemotherapy. However, during the period of this study, the portion of newly diagnosed elderly patients who consulted with oncologists grew from 40% in 1998 to 47% in 2002. This trend is consistent with other reports of the growing use of chemotherapy for patients with cancer,8,9 especially in the elderly population.10-12 We observed that the intensity of oncology services was significantly higher among patients who received chemotherapy. If the percentage of patients who receive chemotherapy continues to increase, there may be greater demand than is reflected in our projections. The introduction of new chemotherapeutic agents may increase the number of patients who are being treated; development of oral chemotherapies may reduce the number of visits for chemotherapy administration.
For some patients, a lack of health insurance can influence use of oncologists’ services. A recent study reported that, in people age 19 to 64 years, 18% were uninsured in 2006.13 This study demonstrated that people without insurance were less likely to undergo breast and colorectal cancer screening, may be diagnosed at a later stage of cancer, and have poorer survival rates compared with insured individuals.13 Even for people with Medicare coverage, the out-of-pocket expenses related to cancer care are significant.14 The future utilization of oncologists may be impacted if the portion of people in the United States without adequate health insurance continues to increase. Utilization of oncologists may also decline if third-party payers restrict coverage of new, costly, targeted chemotherapeutic agents.
Some patients may use their established relationships with their oncologists to obtain primary care services. We evaluated care for the 30,698 patients in the continuing-care phase and found that, of those who received influenza vaccinations or cholesterol screenings, less than 4% of these services were billed by oncologists (data not shown). A study of patients with colorectal cancer15 found that patients treated by primary care physicians were significantly more likely to receive preventive care than those treated only by an oncologist. These findings provide little evidence that oncologists are providing large amounts of primary care to their patients.
Our analysis identified that the vast majority of oncology services occurred in the physician's office, where physician extenders can assist with many routine patient needs. Use of oncologists’ services was greatest during the last year of life, and more than 40% of oncologists’ services occurred in the hospital. Increased use of hospitalists could reduce the need for oncologists. Greater and earlier use of hospice, in which services are often provided by nononcologists, could also ease the need for oncologists during the last year of life. In our analysis of the last-year-of-life phase, 50% of patients had a hospice claim and a mean and median length of hospice stay of 41.1 and 19 days, respectively. These data suggest that a greater number of patients could be referred to hospice and at an earlier time antecedent to death.
There are limitations to our analysis. Our estimates of current utilization may be low, as they were derived from people age 65 years and older in the SEER-Medicare data. Some older patients with cancer may be less likely than younger patients to be referred to an oncologist, as chemotherapy would not be indicated because of the shorter life expectancy and poorer health status. The analysis depended on the AMA data to identify oncologists; for physicians who were not found in the AMA data, we used the self-reported specialty on the Medicare claim. AMA data have been found to be a reliable source of physician specialty.16 An earlier study found that, for physicians who appeared in both the AMA and Medicare data, there was 81% agreement between the two data sources that the doctor was an oncologist.17 Our phase-specific analysis demonstrates that much of the growing use of oncologists’ services will be the result of the increasing number of cancer survivors. Two of the most common cancer types—breast and prostate—have long natural histories. Therefore, it is difficult to determine if the 5 years included in the continuing-care phase capture total utilization for cancer survivors. A final limitation is that the SEER areas used to project future incidence and prevalence of cancer do not represent the totality of the United States. The SEER areas had lower incidence rates than most other states and have been found to have higher socioeconomic status, greater urban population, and more specialty care than the rest of the U.S. population.18,19 These limitations could result in a low estimate of cancer prevalence but higher rates of treatment by oncologists.
These projections are estimates only. They do, however, reveal that the utilization of oncologists is projected to increase markedly between now and 2020. These projections should serve as a catalyst to initiate efforts to increase the supply of oncologists. In the absence of such efforts, the United States may face an acute shortage of medical oncologists to care for patients with cancer in the next 15 years.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The authors indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Joan L. Warren, Angela B. Mariotto, Martin L. Brown
Administrative support: Joan L. Warren
Collection and assembly of data: Joan L. Warren, Angela Meekins, Marie Topor
Data analysis and interpretation: Joan L. Warren, Angela B. Mariotto, Angela Meekins, Marie Topor, Martin L. Brown
Manuscript writing: Joan L. Warren, Angela B. Mariotto, Angela Meekins, Martin L. Brown
Final approval of manuscript: Joan L. Warren, Angela B. Mariotto, Angela Meekins, Marie Topor, Martin L. Brown
Appendix
Codes Used to Identify Medical Oncologists and the POS
Acknowledgments
We thank Edward Salsberg and Clese Erikson from the Association of American Medical Colleges for their suggestions about the design of this analysis and for their comments on the manuscript.
Footnotes
-
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received September 27, 2007.
- Accepted March 7, 2008.
