The Northwestern University Costs of Cancer Program consists ofa series of pilot studies that address the costs of cancer care. Theprogram is designed to serve as a template in preparation for undertakinga large-scale study of a nationally representative sample of cancerpatients-ie, in preparation for a cancer costs and services utilizationstudy in the future. In this article, we outline the theoretical frameworkassociated with a study of cancer costs and summarize findings fromour ongoing pilot studies in this area.
ABSTRACT: The Northwestern University Costs of Cancer Program consists of a series of pilot studies that address the costs of cancer care. The program is designed to serve as a template in preparation for undertaking a large-scale study of a nationally representative sample of cancer patients-ie, in preparation for a cancer costs and services utilization study in the future. In this article, we outline the theoretical framework associated with a study of cancer costs and summarize findings from our ongoing pilot studies in this area.
In the United States, cancer is the second leading cause of death after heart disease. One of every four deaths in this country is caused by cancer. According to the American Cancer Society, in the United States in 2002, there were an estimated 1,284,900 new cancer cases, not including 1 million new cases of skin cancer. In addition, an estimated 555,500 people died of cancer in 2002.
Cancer care is undoubtedly expensive, but the precise costs are unknown. Although the National Institutes of Health estimates that the overall annual costs of cancer care are $156.7 billion, no study has formally evaluated these costs at a national level. This is in contrast to the setting of human immunodeficiency virus (HIV), for which a recently completed $35 million study from the RAND Corporation reported on total medical costs for a nationally representative sample of 2,864 HIVinfected individuals in the United States. The HIV Costs and Services Utilization Study found that, in 1996, the total direct costs of illness were $6.8 billion. However, even this large study did not provide a totally comprehensive picture of the costs of an illness, as estimates of the indirect morbidity and mortality costs were not reported. Moreover, before embarking on this $35 million study, RAND researchers carried out a series of smaller pilot studies that evaluated the costs of HIV care in convenient populations. Using methodologies developed in these studies, the larger investigation was undertaken.
Understanding the costs of an illness is essential to policy makers. For cancer patients, policy makers are faced with many controversies such as the high costs of cancer drugs, access to clinical trials, palliative and end-stage cancer concerns, and the failure of insurance companies to reimburse for high-tech procedures. In order to make informed policies, empirical data on the costs of cancer care are needed. Again, HIV serves as a role model here. Armed with comprehensive data on the costs of HIV care, policy makers have developed comprehensive reimbursement programs such as the Ryan White Titles I, II, and III funds. These programs provide financial support for HIV-infected individuals in settings where the current health-care system might otherwise fail.
It is with this motivation that we have developed the Northwestern University Costs of Cancer Program (NUCCP). The NUCCP consists of a series of pilot studies that address the costs of cancer care. The program is designed to serve as a template in preparation for undertaking a largescale study of a nationally representative sample of cancer patients-ie, in preparation for a cancer costs and services utilization study in the future. In this article, we outline the theoretical framework associated with a study of cancer costs and summarize findings from our ongoing pilot studies in this area.
Reported costs of cancer represent the societal value of oncologyrelated resources consumed in providing cancer care, including the costs associated with drugs and devices, consultation with physicians, and hospitalization. However, most of the detailed information on cancer costs relates to the direct medical expenditures for cancer treatment, rather than for care related to treatment- related toxicity. Very little is known about the societal value of resources used as a result of toxicity, including both the direct medical costs for supportive care and the indirect costs of care to patients and providers. The indirect costs include the value of lost time from work as a result of the illness, decreased productivity while at work, time or money spent by people looking after the patient, premature retirement or death, and demotion in employment.
Additionally, out-of-pocket expenses for patients and the effect on future insurability all have significant costs associated with them, but depend on the perspective of the assessment. Health-care professionals, insurance companies, pharmaceuticals, the government, and patients all have interests in the economic impact of health care, but their respective goals differ greatly. Health-care professionals are concerned about the ability to perform their services within a budget (theirs and the patient's), whereas insurance companies must try to maximize cost efficiency to lower their own costs. Pharmaceutical manufacturers want to maintain their drug prices at higher levels to maximize profits.
The government must bear all of these goals in mind for each sector of the health-care economy in order to regulate the costs upon the consumer. In addition, they must evaluate cost from a societal point of view. Generally, costs can be considered from the perspective of the patient (travel costs, out-of-pocket expenses, lost work time), the employer (lost employee productivity, increased health insurance premiums), or society (overall costs to patients, employers, providers, payers, and the public as a whole).
Generally, health-care costs are classified into three categories: direct medical costs, indirect morbidity costs, and indirect mortality costs. Direct medical costs include the aspects most closely associated with treatment, such as physician bills, laboratory work, drugs, and hospital stays. These costs are the factors most commonly used for analysis, and they often exclude the costs of building and providing health-care facilities such as hospitals and pharmacies.
Direct nonmedical costs are costs associated with treatment that do not involve the procedure itself, including transportation and dietary needs. Finally, indirect costs do not relate to the treatment itself, but they may occur as a consequence. Opportunity costs such as lost work time and reduced productivity fall into this category. It is important to note that to give a complete picture, indirect costs should be factored in as well. These will vary greatly between disparate socioeconomic groups, particularly in terms of cost-benefit ratios, a consideration to bear in mind when making claims for society as a whole.
Patients provide an important perspective because they are the most influenced by costs overall-particularly indirect costs-and these effects will determine the government's reaction to the state of health economics. In contrast to using only monetary costs, patient perspectives allow for a societal cost analysis, which is essential in determining overall costs.
The determination of these costs proves tricky. Even before comparing the direct and indirect costs, the prices and methods that health-care organizations use to convey them do not necessarily reveal information in a straightforward manner. For example, health care is often expressed in average costs of services and goods, but in order to make a strong comparison to indirect costs, marginal costs would be of greater utility.
The ratio of costs to charges (RCC) is a common way of expressing costs, although this method has received wide criticism for being based solely upon Medicare patients, limiting its applicability to the general population. A newer approach using relative- value units (RVUs), assigns value to medical procedures according to the needed resources. While this avoids problems associated with interinstitutional cost comparisons (another criticism of the RCC approach), its utility may be of limited value for individual institutions in terms of selfevaluations.[ 7]
Another difficulty underlying these analyses is the establishment of a comparison method for the data. For example, the dilemma of whether to recruit patients using diagnostic- based, treatment-based, or randomized criteria will bias the population to be studied, most likely skewing the collected data. While specific patient selections will have more utility for studies comparing various methods and management approaches, a randomized study will be more representative of the general population, and this may prove to be more useful for economic analyses.
It has also been recognized that patients in clinical trials often have better prognoses, and the data obtained through these studies may not accurately reflect the true costs involved.[ 8] Another factor that is often overlooked in cost analyses is the comparison between treatment and preventive measures. Many cancers and serious health conditions do not lend themselves to good preventive strategies. However, with certain conditions (eg, lung cancer), prevention plays an important role in epidemiology and should always be taken into consideration.
NUCCP Study 1: Out-of-Pocket Expenditures
Direct medical and nonmedical as well as indirect costs faced by cancer patients and their caregivers are currently being assessed through a longitudinal study. This study is jointly funded by a company that provides cancer insurance (AFLAC), a pharmaceutical company (Bayer), and the American Society of Clinical Oncology, a nonprofit organization. Breast, prostate, and colorectal cancer patients diagnosed within 2 years of study initiation, and their caregivers, are asked to fill out a comprehensive, multidimensional survey instrument.
The patient and caregiver surveys include questions about (1) direct medical costs (hospital stays, physician visits, pharmaceuticals, home health care, prosthetics, experimental medicines, special supplies); (2) direct nonmedical costs (transportation, parking, home health aides, psychological support, cost of meals); and (3) indirect costs (absence from work, lost wages, missed promotions). Caregivers are also surveyed regarding the physical and psychological demands of caregiving. A sample of patients also complete a 6-month log- book (filled out weekly) to prospectively assess out-of-pocket costs they incurred as they progressed through their cancer treatment. The 3-month retrospective survey and the 6-month prospective logbook provide data on the degree of financial burden that patients and their caregivers experience over the course of a 9-month treatment period.
Of the 93 patients currently enrolled in this study, 60 have completed the initial interview. The majority (88%) were female, the mean age of the participant was 50 years (range: 20-64 years), and 72% had at least one caregiver, of whom 84% were spouses. Breast cancer patients made up 87% of the study cohort, colorectal cancer, 9%, and prostate cancer, 3%.
Once the diagnosis of cancer had been made, out-of-pocket expenditures were considerable for both patients and caregivers, including 35% who used sick days (mean: 53 h/mo). In addition, 15% lost work-related salary (mean: 47 h/mo), 15% felt that they had been bypassed for a job promotion due to illness, 5% felt that they had received a demotion due to illness, and 20% reported that they had considered voluntary unemployment (Table 1). Fifteen percent of caregivers missed work to assist with the patient's care, for an average of 23 h/mo.
For 30% of the cancer patients, insurance carriers refused to pay for some aspect of care-mainly wigs, referrals, out-of-state consultations, antidepressive drugs, private rooms, basic supplies, and part of the cost of radiation, colonoscopy, chemotherapy, and anesthesia. Mean out-ofpocket expenditures were $546/mo for patients and $247/mo for caregivers. In addition, loss of income due to absence from work amounted to $390/mo for patients and caregivers.
A second pilot study evaluated the impact of the financial burden associated with cancer on the choice of the institution providing care. Prostate cancer patients whose medical care had recently been transferred from non-Veterans Administration (VA) settings to the VA system were interviewed to determine reasons for transferring care. Explanatory choices included cost of physician services, cost of private health insurance, nonfinancial considerations (including quality and coordination of the VA), cost of radiation therapy, cost of surgery, cost of hormonal agents, and cost of follow-up care. Patients were also given the opportunity to provide reasons not listed among the alternatives.
A 3-minute survey questionnaire was completed for 104 prostate cancer patients at the Chicago VA Health Care System, representing an 80% response rate. Almost half of the patients were African-American (46.7%), and the mean monthly income was $1,197. More than half of the VA patients had private health insurance before transferring care (64.2%), 26.4% maintained their private health insurance after the transfer, 46.2% had their cancer diagnosed at a non-VA medical center, and 70.8% received their initial care at a non-VA site.
NUCCP Study 2: Transferal to Veterans Administration Care
Out-of-pocket costs associated with hormonal therapies for prostate cancer, primarily oral nonsteroidal antiandrogens, were the most common reason for transferring care (34.9%), followed by physician co- payments (30.2%), quality considerations (26.2%), and copayment for luteinizing-hormone-releasing hormone (LHRH) analogs (8.5%, see Table 2). When limited to only VA patients with metastatic disease (n = 59), the cost of nonsteroidal antiandrogen therapy was by far the most important reason for transfer to the VA system (44.1% of the transfers), followed by physician copayments (23.7%), quality considerations (18.6%), and LHRH costs (13.6%).
A third pilot study addressed the financial impact of cancer on the patient following diagnosis and treatment of a curable malignancy, hairy cell leukemia. Without empirical data, it is likely that policy makers often assume that the financial impact of cancer is minimal for individuals who are cured from the illness.
In this study, funded by the Hairy Cell Leukemia Survivors Group, persons who were treated and cured from hairy cell leukemia were surveyed to determine any changes in health insurance status (such as premium, deductible, copayments for physician visits, and prescriptions); life insurance history (policy holdings prior to and after their diagnosis, difficulties obtaining a policy after diagnosis); and changes in lifestyle habits (exercise, eating habits, use of vitamins or nutritional supplements, use of alternative medical practitioners, and use of psychological counseling, as well as overall impact of the cancer experience on lifestyle habits).
NUCCP Study 3: Working Statistics for Hairy Cell Leukemia Patients
Completed surveys were received from 31 cancer survivors with a mean age of 56.5 years (range: 37-82 years). Patients reported a mean age of diagnosis of 48.7 years (range: 24-73), and 78.6% reported an annual income of $50,000 or more. Half were employed, 36.7% were retired, and 13.3% were unemployed. Onefifth (22.6%) had taken a leave of absence from work for an extended period of time because of their cancer (Table 3).
After treatment, 8 patients had private health insurance, 6 belonged to an HMO, 10 were covered by a PPO, 7 used Medicare, and 1 patient was not insured. Health insurance benefits had been discontinued or coverage was refused for four study participants (12.9%) due to their history of cancer. Additionally, the majority of patients (63.3%) reported that they had incurred significant out-of- pocket medical costs for their medical care following their diagnosis. Of the 13 respondents who attempted to obtain a life insurance policy following treatment for hairy cell leukemia, 76.9% faced difficulty or were turned down.
NUCCP Study 3: Lifestyle Changes
Following their experience with hairy cell leukemia, many respondents reported positive lifestyle changes (Table 4), including increased exercise (61.3%) and healthier diets (51.6%). Patients also reported an increase in the use of vitamins or herbal nutritional supplements, alternative medicine practitioners, and psychological counseling. Nearly half (48.4%) of the participants indicated that their experience with cancer had a large impact on their lives, while 32.3% reported a little impact, and 19.4% reported no impact. Participants reported having a greater appreciation for life, loved ones, and physical health, and many reported having made significant employment and/or lifestyle adjustments as a result of cancer. Clearly, even when patients are cured of their cancer, they still carry a significant financial burden associated with their previous diagnosis.
A fourth pilot study addressed an often overlooked aspect of cancer care-the costs of toxicity related to treatment. In this study, funded by the ALZA pharmaceutical company (which is now part of Johnson & Johnson), all women with ovarian cancer and chemotherapy-associated neurotoxicity, neutropenia, and/ or thrombocytopenia who received care at the Northwestern University gynecology-oncology clinic were eligible for inclusion. Consenting patients were queried about resources used as a result of toxicity, including outpatient visits from traditional and alternative health-care providers, labs, phone calls, hospitalizations, home health visits, medications (including over-the-counter drugs), medical devices, time missed from work, and time spent by caregivers to assist with activities of daily living during episodes of toxicity.
Resource utilization was tracked from study entry, defined as coincident with the onset of toxicity, and continued through the end of the toxicity "episode" (up to 9 months for neurotoxicity and 3 months for hematologic toxicity). Economic variables were collected at baseline and every 3 months during the follow-up period. All patients were recruited within 1 month of experiencing the toxicity. Economic data included a detailed listing of the medical care services received by subjects, attribution of the cause of each service, and loss of work time because of toxicities.
Primary analyses included only resources that are directly attributed to toxicity or its treatment. A cost value was assigned to each resource. Unit costs came from hospital bills for inpatient stays, the Medicare Physician Fee Schedule for outpatient services, the Red Book for pharmaceuticals,[ 11] an average price from three national drug stores for devices, and clinician estimates for phone calls to medical providers (15 minutes to a physician or a nurse).
Standard unit costs were multiplied by the average number of visits, labs, hospitalizations, medications, and medical devices to calculate an estimate of the average cost per person in each of the three study groups. Indirect costs (ie, productivity loss) were based on modified Labor Force, Employment, and Earnings data. A total cost of toxicity was derived for each participant. The proportion that each component cost contributed to the corresponding aggregated cost (indirect, direct or total) was assessed for statistical significance across the three toxicity groups, using the chi-square test.
NUCCP Study 4: Costs for Comorbidities With Ovarian Cancer
• Direct Costs-The study included economic information regarding neurotoxicity (n = 42), neutropenia (n = 26), and thrombocytopenia (n = 15) following ovarian cancer chemotherapy. Direct medical costs (Table 5) were highest for neutropenia (mean: $7,546 per episode), intermediate for thrombocytopenia (mean: $3,268 per episode), and lowest for neurotoxicity (mean: $688 per episode). The proportion of the total direct costs for each of the seven component costs was found to be significantly different among the three groups (c2 = 228.3, P < .001).
For the hematologic toxicities, inpatient medical care accounted for the largest part of the direct medical costs-$6,204 for neutropenia (82% of the total direct costs) and $2,654 for thrombocytopenia (81% of the total direct costs)-followed by outpatient pharmaceuticals ($721 for neutropenia, 10% of the total direct costs). In contrast, neurotoxicity, although disabling, was not associated with any inpatient care and accounted for $432 in outpatient pharmaceuticals and devices (63% of the total direct costs).
• Indirect Costs-Indirect costs relating to patient and caregiver work loss and payments for caregiver support were substantial, accounting for $4,200, $3,834, and $4,282 for ovarian cancer patients who developed neurotoxicity, neutropenia, and thrombocytopenia, respectively (Table 5). Differences in proportions of the total indirect costs for each of the three component costs were found to be statistically significant among the three groups (c2 = 113.1, P < .001). Hematologic toxicity indirect costs were primarily associated with patient work loss-$2,422 for neutropenia (63% of the total indirect costs) and $2,122 for thrombocytopenia (50% of the total indirect costs)-whereas neurotoxicity was associated with $2,837 in caregiver work loss (67% of the total indirect costs).
• Total Costs-Proportions of the total costs (direct and indirect) for each of the three component costs were found to be significantly different among the three groups (c2 = 40.8, P < .001). In total, the costs of chemotherapy- related neurotoxicity, neutropenia, and thrombocytopenia were $4,908, $11,830, and $7,550, with indirect costs accounting for 86%, 43%, and 57% of the total costs of toxicity, respectively (Table 5).
A fifth pilot study evaluated the costs of phase II clinical trials for cancer patients. The pilot effort included assessment of total direct medical charges for 6 months of care for 35 patients who received care in phase II clinical trials and for 35 matched controls (based on age, sex, disease, stage, and treatment period) at five cancer centers. Cost data were obtained for hospital and ancillary services from automated claims files at individual study institutions. The analyses were based on the perspective of a third-party payer.
The mean age of the phase II clinical trial patients was 58.3 years, compared with 57.3 years for control patients. The study population included persons with cancer of the breast (n = 24), lung (n = 18), colon (n = 16), prostate (n = 4), and lymphoma (n = 8). The ratio of male-to-female patients was 3:4, with greater than 75% of patients having stage III to IV disease.
NUCCP Study 5: Total Mean Charges for Phase II Cancer Clinical Trials
As shown in Table 6, total mean charges for treatment from the time of study enrollment through 6 months were similar: $57,542 for clinical trial patients and $63,721 for control patients (1998 US$; P = .4).
In the pilot projects conducted by the NUCCP, we have identified various aspects of cancer care for which information on costs has been lacking. These areas represent the "hidden" costs of cancer, which, although not included in insurers' estimates for the illness, are nonetheless real. A similar framework was outlined by HIV investigators as part of the nationally representative HIV Costs and Services Utilization Study, conducted by researchers at the RAND Corporation.
Estimates of provider burden that result from the proliferation of managed care plans and administrative responsibilities that accompany medical care today are being developed in a recently initiated pilot study by the NUCCP. It is our goal to conduct a large-scale Cancer Costs and Services Utilization Study-a comprehensive economic assessment of cancer care that will build on these pilot efforts.
As with the RAND study of HIV costs and services, for which several small pilot studies both assisted policy makers faced with uncertainties in their legislative endeavors and provided preliminary data prior to the comprehensive national study, it is hoped that a similar process will be carried out for oncology. Armed with this information, federal, state, and local policy makers will be able to develop comprehensive programs to ensure that cancer patients receive high quality but affordable cancer care.
Financial Disclosure: The authors have no significant financial interest or other relationship with the manufacturers of any products or providers of any service mentioned in this article.
American Cancer Society Facts & Figures.Atlanta, American Cancer Society, 2002.
Bozzette SA, Hays RD, Berry SH, et al:A Perceived Health Index for use in personswith advanced HIV disease: Derivation, reliability,and validity. Med Care 32:716-731,1994.
Pascal A, Cvitanic MA, Bennett CL, et al:State policies and the financing of AIDS care.RAND Notes. N-2942-HCFA, 1989.
The Ryan White Comprehensive AIDSResources Emergency (Care) Act. U.S. Departmentof Human and Health Services web site.Available at http://www.hhs.gov/news/press/1998pres/981218d.html. Accessed November11, 2002.
Boyer M: The economics of lung cancer.Lung Cancer. 14:13-17, 1996.
Statistical Abstract of the United States1999, 119th ed. US Department of CommerceEconomics and Statistics Administration, USCensus Bureau.
Waters TM, Weinberg PD, Bennett CL:Health economics in the treatment of colorectalcancer. Cancer Invest 16:582-587, 1998.
Phan LT, Lee JU, Calhoun EA, et al: Anupdated look at the societal cost of toxicitymanagement and out-of-pocket costs for patientsand their caregivers (abstract). Proc AmSoc Clin Oncol 21:254a, 2002.
Hounshell J, Tomori C, Newlin R, et al:Changes in finances, insurance, employment,and lifestyle among persons diagnosed withhairy cell leukemia. Oncologist 6:435-440,2001.
Calhoun EA, Chang CH, Welshman EE,et al: Evaluating the total costs of chemotherapy-induced toxicity: Results from a pilot studywith ovarian cancer patients. Oncologist 6:441-445, 2001.
Red Book. Medical Economics Co, Inc,Montvale, NJ, 1999.
Bennett CL, Stinson TJ, Vogel V, et al:Evaluating the financial impact of clinical trialsin oncology: Results from a pilot studyfrom the Association of American Cancer Institutes/Northwestern University Clinical TrialsCosts and Charges Project. J Clin Oncol18:2805-2810, 2000.