To the authors' knowledge, no analysis has examined the specific components of drug spending for overall cancer care. The authors' objective was to quantify and characterize trends in outpatient drug expenditures for cancer patients.
To the authors' knowledge, no analysis has examined the specific components of drug spending for overall cancer care. The authors' objective was to quantify and characterize trends in outpatient drug expenditures for cancer patients.
The authors retrospectively analyzed pharmacy and outpatient professional claims data from commercial and Medicare health maintenance organization enrollees with a solid tumor diagnosis in 1995 and 1998. Charges were subdivided by type of drug (antineoplastic drugs, chemotherapy adjuncts, supportive drugs, and drugs unrelated to cancer treatment).
In 1995, 14,663 cancer patients received outpatient drug treatment and 13,829 patients in 1998. Total charges increased from $17.9 million (mean charge of $1218 per patient) to $27.9 million (mean charge of $2003 per patient), an average annual increase of 16%. Antineoplastic therapy constituted the largest component of cancer-related drug costs (67%) and represented 76% of the increase from 1995 to 1998. Most charges were incurred in the professional setting for agents administered by injection. The primary explanation for the increases appeared to be a shift in treatment patterns toward newer, more expensive antineoplastic agents. Supportive therapy represented 17% of the increase in cancer drug costs, followed by chemotherapy adjuncts (7%). Charges for drugs unrelated to cancer therapy increased by 21% per year.
Antineoplastic therapy administered in an office or clinic was the single most important cost driver, with newer more expensive agents replacing older, less expensive drugs. Attempts to understand and control outpatient drug cost increases for cancer patients should focus primarily on antineoplastic therapy, especially the appropriate substitution of newer agents for older, less expensive alternatives. Some nonchemotherapy cancer drugs may offer an opportunity to improve quality of life with a relatively small effect on overall cancer drug costs. Cancer 2002;94:1142–50. © 2002 American Cancer Society.
Cancer is the second leading cause of death in the United States, representing one in four deaths. Costs associated with cancer care are substantial. The National Institutes of Health has estimated that in 1999 the total annual costs will exceed $107 billion, with $37 billion for direct medical costs alone. More than half of these direct costs are accounted for by breast, lung, and prostate carcinoma care.1 Furthermore, cancer costs are increasing steadily, rising an estimated 62% from 1985 to 1990.2
Understanding drug costs in cancer therapy should play an important role in the national debate on increasing costs for prescription drugs. Overall, spending on prescription drugs increased from 5.6% of national health expenditures in 1993 to 7.9% in 1998. During that time, growth in drug spending increased from 8.7% to 15.4% and reached 17% in 1999.3, 4 Double-digit growth rates for prescription drugs are projected through 2005.5 Moreover, drug spending trends are complex and are affected by many factors, including disease prevalence, demographics, changes in treatment patterns, new scientific advances, and inflation.6 Previous studies have examined the overall economic impact of cancer,7, 8 the direct costs of cancer care by type of tumor,9–11 stage at diagnosis,12 and phase of care,13 but there is surprisingly little quantitative information on drug costs for cancer as a whole. To our knowledge, no analysis has examined the specific components of drug spending for overall cancer care.
The purpose of this study was to examine trends in outpatient pharmaceutical expenditures for cancer treatment. Specifically, we explored three questions: 1) How much are costs increasing? 2) What drove those increases? 3) What does this suggest for the future? By including drugs administered by professionals in the outpatient setting as well as those dispensed through outpatient pharmacies, this report provides a comprehensive look at the outpatient arena of cancer-related drug therapy.
Retrospective medical and pharmaceutical administrative claims from Protocare Sciences' proprietary databases for the years 1995 and 1998 were used for this analysis. These data represent information on commercial and Medicare health maintenance organization (HMO) enrollees in 20 states in the South, Southeast and Midwest. To be eligible for inclusion in this analysis, a plan member had to have two separate ICD-9-CM (International Classification of Diseases, 9th revision, Clinical Modification) diagnostic codes for cancer diagnosis occurring at least 30 days apart and have continuous enrollment and coverage for both medical and pharmaceutical benefits for either of the two study years. Patient samples were drawn independently for 1995 and 1998. The same patients may not have been included in both years. The following ICD-9-CM codes were used to identify members carrying a diagnosis of cancer: breast, 174.0–174.9; prostate, 185 and 233.4; lung, 162.0–162.9; colorectal, 153.0–154.1 and 197.5; other tumors, 140–152, 154–161, 163–172, 175–184, 186–199.1, 230–234.9. Neoplasms of the lymphatic and hematopoietic tissues and nonmelanoma skin carcinomas were excluded from the analysis.
Health Care Financing Administration Common Procedure Coding System codes were used to identify cancer-related drugs administered in an outpatient facility by a physician. Uniform System Classification codes were used to identify cancer-related pharmacy prescriptions. For this analysis, all medications administered in the outpatient setting were included. Patients received these agents either through a standard outpatient pharmacy prescription (pharmacy) or through administration by a physician or other health professional at an outpatient medical facility or office (professional), typically injected or parenteral agents. Drugs administered as part of an inpatient hospitalization were not included in this analysis.
We use the term “claim” to represent a request for payment for the administration of an injectable drug or the filling of a prescription. The term “charge” represents the dollar amount associated with a claim. Charges were used in lieu of actual amounts paid, which were not available.
To assess drug costs as specifically as possible, no ancillary claims (i.e., office visits, intravenous setup, etc.) were included. For each study year, the number of claims and total charges were tallied for individual drugs in both settings. Numbers of claims and charges then were aggregated across settings, yielding totals for each drug. Average charges per claim were calculated by dividing the total charges by the total number of claims. The average number of claims per treated patient was calculated by dividing the total number of claims by the total number of patients within the group being considered.
Cancer-related drug therapy was grouped into three major categories: 1) antineoplastic drugs, 2) chemotherapy adjuncts, and 3) supportive drugs. Antineoplastic medications were those directed at controlling the neoplastic process. Chemotherapy adjuncts represented those agents used to counteract significant adverse effects secondary to chemotherapy—agents that reasonably could be expected to enhance a patient's ability to continue or complete a current course of chemotherapy. These agents included growth factors, antiemetics, steroids, and organ/cytoprotective agents. Supportive drugs included those used to treat complications or symptoms of conditions frequently associated with cancer, but which would be unlikely to have a direct effect on disease free survival time or tumor progression. These agents included psychotherapeutic drugs, analgesics, antibiotics and other antiinfectives, nutritional supplements, and gastrointestinal medications other than antiemetics.
All prescription medications not included in the above categories, i.e., drugs for routine patient conditions, were aggregated into a single item (unrelated pharmacy) and included for completeness.
Distribution of patient characteristics was compared between years using chi-square tests. Average annual rates of increase or decrease were calculated assuming a constant rate of change, using the formula for annualized rate of return14:
The number of plan members continuously enrolled for both medical and pharmacy benefits was 1.56 million in 1995 and 1.24 million in 1998. Of these, the number of eligible patients (i.e., those receiving a cancer diagnosis) was 20,481 in 1995 and 15,798 in 1998, representing approximately 1.3% of the member population in both years. Selected characteristics of these study populations are presented in Table 1. The most common tumors for both years were prostate and breast carcinomas, followed by colorectal and lung carcinomas. The population in 1998 was somewhat older than that of 1995 (P < 0.0001), and the distribution of cancer diagnoses differed between the study years (a small percentage increase was observed in breast, prostate, and colorectal carcinoma cases). Because these differences did not appear to be clinically important for the purpose of comparing drug utilization and costs between years, crude (unadjusted) comparisons were used. Although the proportion of diagnosed patients receiving therapy increased, the actual number of patients receiving drug treatment decreased from 14,663 (72% of those diagnosed) in 1995 to 13,829 (88% of diagnosed) in 1998.
|No. of patients with cancer diagnosis||20,481||100||15,798||100|
|Tumor type (ICD-9-CM codes)bc|
|Prostate (185, 233.4)||5896||28.8||5038||31.9|
|Colorectal (153.0–154.1, 197.5)||2389||11.7||1950||12.3|
Charges, number of claims, and patients receiving outpatient drug therapy are presented for the study population as a whole in Table 2. Total charges increased from $17.9 million ($1218 per treated patient) in 1995 to $27.9 million in 1998 ($2003 per patient), an average annual increase of 16% (18% per patient). The total number of claims increased by 7.8% per year during the same period, whereas the number of patients being treated decreased by 1.7% per year. Increases in professionally administered drugs ($5.1 million) and in pharmacy claims ($4.9 million) contributed equally to the overall increase in drug charges. However, whereas the number of pharmacy claims increased by 8.7% annually during this period, the number of professional claims decreased by 4.9% annually. Thus, charges per claim increased considerably faster for professional (20%) than for pharmacy (8.3%) claims. In both years, charges for professionally administered drugs were roughly 9 times higher per treated patient than charges for drugs obtained through a pharmacy ($8570 vs. $936 in 1998). The number of claims per treated patient decreased by 1.7% annually for professionally administered drugs but increased by 9.5% for drugs dispensed through a pharmacy.
|Associated factors||1995 ($)||1998 ($)||Change 1995–1998|
|Increase (decrease) ($)||Rate of change (%)|
|Total no. of claims||343,215||430,270||87,055||7.8|
|Charge per claim||52.02||64.84||12.83||7.6|
|Charge per treated patient||1,217.58||2003.03||785.45||18.0|
|Claims per treated patient||23.41||30.89||7.48||9.7|
The impact of various types of therapy on cancer drug charges is presented in Table 3. Antineoplastic therapy made the largest contribution to drug costs, representing 67% of cancer drug costs in 1998 and 76% of the $6.9 million increase in charges from 1995 to 1998. Most antineoplastic charges were incurred in the professional setting. Chemotherapy adjuncts represented 18% of charges, but 7% of the 1995–98 increase. Supportive therapy made up 15% of 1998 charges and accounted for 17% of the cost increase. Charges for outpatient pharmacy drugs unrelated to cancer therapy increased by 21% per year.
|Therapy type||1995 charges ($)||1998 charges ($)||Percentage of 1998 cancer drug charges (%)||Charges 1995–1998|
|Increase (decrease) ($)||Rate of changea %||Percentage of cancer drug increase|
|Subtotal: cancer drug therapy||13,786,630||20,706,799||100||6,920,169||14.5||100|
|Subtotal: unrelated pharmacy||4,066,787||7,193,464||3,126,677||20.9|
Antineoplastic drugs with the greatest impact on charge differences between 1995 and 1998 are presented in Table 4. Paclitaxel ($1.4 million increase) was the most significant contributor to the increase, followed by irinotecan ($0.93 million), carboplatin ($0.57 million), goserelin ($0.53 million), and docetaxel ($0.50 million). Paclitaxel alone accounted for 20% of the total increase in cancer drug costs, an increase driven by a doubling of claims during the study period. Four of the top 10 contributors to cost increases (irinotecan, docetaxel, gemcitabine, and topotecan) were new drugs, becoming available after 1995. Several older drugs showed decreases in overall charges during the study period, including etoposide, cisplatin, cyclophosphamide, flutamide, and mitomycin. These decreases were driven by decreases in utilization (i.e., numbers of claims) that were, in some cases, quite substantial.
|1995||1998||1995–98 difference||Rate of changea|
|No. of claims||Charges ($)||No. of claims||Charges ($)||Charges ($)||Charge per claim ($)||Claims (%)||Charge per claim (%)|
|Greatest positive impact|
|Greatest negative impact|
|Mean charge/claim ($)||99.97||125.62|
|Total antineoplastic (all drugs) ($)||47,770||8,669,847||40,950||13,933,092||5,263,245||158.75||(5.0)||23.3|
|Mean charge/claim ($)||181.49||340.25|
The contribution of other agents to cancer drug charges is presented in Table 5. Among chemotherapy adjuncts, antiemetics were the largest contributors to charges, followed by hematopoietic growth factors (mainly colony-stimulating factors and erythropoietins), organ and cytoprotective adjuncts, and steroids. Although antiemetics accounted for the largest proportion of total charges within this group, charges for antiemetics actually decreased by $76,698 between 1995 and 1998. As a group, chemotherapy adjuncts saw an increased charge per claim of only 3.1% per year. Among supportive therapies, gastrointestinal agents (primarily proton-pump inhibitors), made up the largest cost component, followed by psychotherapeutic agents (primarily antidepressants) antibiotics and other antiinfectives, and analgesics (primarily narcotics). In this group, claims volume increased by 6.1% per year, with a 9.9% per year increase in charges per claim. Psychotherapeutic agents showed the largest cost increase among nonchemotherapy drugs.
|No. of claims||Charges ($)||No. of claims||Charges ($)|
|Hematopoietic growth factors||2698||1,206,800||3361||1,554,173|
|Subtotal: chemotherapy adjuncts||18,691||3,124,823||19,721||3,614,443|
|Subtotal: supportive therapy||91,257||1,991,960||109,013||3,159,263|
|Subtotal: nonchemotherapy cancer drugs||109,948||5,116,783||128,734||6,773,706|
Drug therapy in cancer patients is an important clinical and financial issue and is likely to become more critical in coming years. Important questions that need to be addressed include: How much are costs increasing and why? and What does this suggest for the future? In the current study, overall outpatient drug-related spending increased by 56% between 1995 and 1998, an average annual rate of 18% on a per-patient basis. During the same period, the consumer price index increased 7.0%, an average annual increase of 2.3%.15 The data presented here focus on the substantial majority of utilization, because as much as 90% of cancer patient care occurs in the outpatient setting, including most chemotherapy.16
Antineoplastic therapy was the single largest cost component, accounting for 67% of all cancer drug expenditures in 1998 and 76% of the increase in spending from 1995 to 1998. Among these agents, the primary change appeared to be a shift in treatment patterns toward newer, more expensive agents (averaging more than $800 per claim) and away from older, less expensive drugs (with per claim charges in the range of $100). During the study period, treatment of metastatic breast carcinoma shifted toward agents such as the taxanes and vinorelbine, replacing less expensive agents like doxorubicin, cyclophosphamide, and methotrexate.17, 18 Much of this change likely resulted from increasing use of the newer drugs in adjuvant programs.19
In colorectal carcinoma, irinotecan ($2080 per claim in our study) was approved for second-line therapy in 1996,20 after firstline therapy with fluorouracil ($22 per claim in 1998) and leucovorin ($315 per claim). In prostate carcinoma combined androgen blockade, a therapy combining a luteinizing hormone-releasing hormone analog with an antiandrogen agent, became more widespread.21 In non-small cell lung carcinoma, the taxanes, vinorelbine, gemcitabine, and irinotecan all came into prominence.22, 23 Thus, newer agents with considerably higher acquisition costs replaced (or were added to) older agents in each of the most common cancers.
The important question of whether any additional benefits provided by newer agents justify their additional cost is controversial.24 Ideally, more expensive agents should provide better health outcomes (e.g., improved disease free survival, improved overall survival, or better quality of life). Death rates for the four major cancers (prostate, breast, colorectal, and lung) have decreased significantly from 1990 to 1995.25 Although some of these decreases may be related to decreased incidence, evidence supports the idea that better adjuvant therapy has contributed to improved survival in the cases of breast26 and colon carcinoma.27 However, because it can take a decade after a drug's release to evaluate its true effect in a population, decision makers often are faced with a dilemma: either give wide access to newer more expensive agents that have improved response rates but whose long-term benefits are incompletely understood, or restrict these newer agents and potentially condemn some patients to shorter lives or greater suffering.
The second most costly group of drugs in our study was chemotherapy adjuncts—drugs that counteract negative effects of neoplastic therapy (e.g., nausea, vomiting, bone marrow suppression). The chemotherapy adjuncts accounted for 18% of outpatient cancer drug spending in 1998 and 7% of the increase from 1995 to 1998. Some of these agents have attracted considerable attention because of their high acquisition costs. For example, colony-stimulating factors ($439 per claim in 1998) have been the subject of guidelines,28 editorials,29 and pharmacoeconomic reviews.30 The serotonin antagonist antiemetics also have attracted attention. In our study, these drugs had a significantly higher unit cost than other antiemetics: $276 per claim in 1998, compared with an average $21 per claim for other antiemetics. Although antiemetics in aggregate accounted for the largest portion of total charges among the adjunctive agents, total charges for antiemetics actually decreased during the study period. Despite the frequent focus on these individually expensive agents, they contributed relatively little to the overall increase in spending.
The last category, supportive therapies, represented 15% of spending and 17% of the cost increases measured in this study. The supportive therapies included those drugs that treat the patient's general symptoms such as pain, depression, or heartburn. Since Cleeland et al's landmark article in 1994,31 it has been widely agreed that increased treatment of the suffering associated with cancer—specifically, physical and psychological pain—are appropriate. Furthermore, at least in the case of narcotic analgesics and antidepressants, effectiveness has been established, and these agents have only a modest impact on total spending. For other supportive drugs, the issues are less clear. Using the gastrointestinal agents as an example, the primary pattern driving costs was a shift away from H2 blockers toward the more expensive proton-pump inhibitors.
Drugs unrelated to cancer care, e.g., drugs used to treat common illnesses such as hypertension and hyperlipidemia, accounted for 26% of all outpatient drug spending in 1998 and 31% of the increase from 1995 to 1998. These patients have noncancer comorbidities that consume a significant amount of pharmaceutical resources.
Our study population was drawn from a relatively large managed care data set, which is likely to be representative of similar managed care plans but not necessarily of the general U.S. population. In using diagnostic codes associated with professional claims to identify persons with cancer, we have obtained a subset of prevalent cases that is biased toward persons undergoing active treatment for cancer.
The current study examined changes in costs to a health care system rather than changes in the costs of a course of treatment, which may be shorter or longer than a calendar year. Because we selected claims by calendar year rather than 1-year costs from the time of diagnosis, our results may overestimate the number of new cases and underestimate the 1-year per-patient costs. These results do not measure cancer incidence or prevalence, nor do they describe the costs of an “average” course of cancer treatment.
Although we may make conclusions regarding patterns of treatment in these cancer patients, we cannot make specific statements regarding appropriateness of care. First, individual drugs were not linked to specific diagnoses. Second, patient specific information such as the stage and type of cancers, the patterns of care (i.e., drug combinations, duration of therapy) and the use of other treatment modalities were not analyzed. Third, because the analysis was conducted at the claims level, we cannot be certain of the number of patients receiving a specific therapeutic regimen. Differences in treatment intensity (including longer duration of therapy) rather than in the number of patients treated may have accounted for a portion of the differences in claims between the study years.
Drugs administered as part of an inpatient hospitalization were not included in this analysis. Thus, our results underestimate total drug expenditures, and some portion of the observed increase may have been because of shifts from the inpatient to the outpatient setting.
Finally, as mentioned in “Methods,” charges were not adjusted for differences in the patient population between the study years. Thus, differences in disease severity or case mix may have influenced our findings.
On the basis of the results of this study, what does the future hold? First, in defining the role of antineoplastic agents in overall costs, it becomes clear that decisions about drug spending in cancer are difficult and are likely to remain so. Despite a decrease in age-adjusted cancer incidence and death rates,32 the absolute number of cancer patients will grow as the population ages. To the degree that survival is enhanced by newer treatments (in the absence of complete cure), more patients will require late-stage and second-line therapies represented by the antineoplastic cost drivers observed in this study. As newer, more expensive agents become available, physicians will want to use them. To the degree that these newer agents may provide improved benefits (i.e., better survival, fewer side effects, better quality of life, etc.), restricting them will provide cost savings only at the expense of patient care. Decision makers, then, must face the problem squarely: there are no easy answers because there are no complete data.
Second, outpatient formulary mechanisms generally affect only those drugs dispensed through a pharmacy. Even the most aggressive formulary manipulations are unlikely to affect trends that are driven primarily by antineoplastic therapy, which is generally administered by a professional and generally not under the purview of the standard controls exercised by pharmacy benefit managers. Finally, nonchemotherapy drugs that substantially improve quality of life33 at acceptable cost should be encouraged, especially in areas where underuse is thought to occur.34 In many cases, these drugs would appear to offer a large potential benefit, with a relatively small effect on overall cancer drug costs.
The authors thank Andy Mosso and Ruby Vendiola for their analytic help, Merle Haberman for technical input, George Goldberg for clinical input, and Gary Persinger and Seonyoung Ryu for comments on the article.