This monograph will review the burden of illness in oncology, suggest a framework for evaluating oncology costs and consequences, identify economic modeling formats in cancer care, and explore methods of cost control for cancer care.
(Am J Manag Care. 2006;12:S3-S16)
Cancer is the second leading cause of death in the United States, claiming approximately 570 280 Americans in 2005. The disease is responsible for 1 of every 4 deaths, and strikes approximately 1.37 million people per year.1 Although survival rates for the 15 most common cancers have been improving every year since 1993, the overall number of cancer deaths has continued to increase given the country's growing population and the aging of its citizens.2
Overall, the incidence of age-adjusted cancer rates in the United States has remained steady since 1992 for all cancer sites combined, although the incidence of some neoplasms–including melanoma and cancers of the kidney in men and women, prostate and esophageal cancer in men, and leukemia, non-Hodgkin's lymphoma, breast, thyroid, and bladder cancer in women–has increased.2 Broken down by sex, incidence rates have remained steady for men but increased 0.3% for women.1
This state of affairs will likely change, however. Indeed, the number of cancer cases in the United States is expected to double over the next 50 years to 2.6 million, the result of an aging and expanding population.1
Even without any significant increase in cancer incidence or mortality rates, costs have been increasing significantly. In 2004, the country spent $189.8 billion on overall costs for cancer, including $69.4 billion for direct medical costs, $16.9 billion for lost productivity caused by morbidity, and $103.5 billion for lost productivity due to mortality. Those costs were expected to increase in 2005 to $209.9 billion overall, including $74 billion in direct medical costs and $135.9 billion in indirect costs related to morbidity and mortality, about a 9.5% increase.1
Spending on cancer in the United States accounts for 4.7% of overall medical treatment expenditure and about 10% of Medicare spending.3 Moreover, 41.3% of Medicare drug expenditures were for oncology/hematology drugs, paid primarily through Part B reimbursement.4 Among cancer types, lung cancer tops the list of total expenditures, accounting for 13.3% of all cancer treatment dollars (~$9.6 billion), followed by breast cancer at 11.2% (~$8.1 billion).3
Although the cost of cancer treatment as a percentage of overall direct medical treatment expenditures has remained fairly consistent over the past 30 years, the National Cancer Institute (NCI) predicts these costs will begin rising at a faster rate, particularly as the population ages and the number of cancer cases increase. Additionally, the NCI notes that cancer costs will continue to rise as newer, more effective, and more expensive biologic treatments enter the pharmaceutical market.3
Relative to other pharmacotherapy, these new biologic treatments come with substantial costs and, in many instances, may only prolong life a few months longer than current treatments. For instance, adding the antivascular endothelial growth factor bevacizumab to a standard routine of irinotecan-5-FU/LV (IFL) for advanced colorectal cancer increased mean survival time by 4.7 months compared with those receiving IFL alone, a clinically significant difference, but a small one, nonetheless.5
The significant costs of these new drugs often put them out of the reach of any American without health insurance and pose an economic difficulty even for those with excellent insurance. In the case of colorectal cancer, for instance, one published cost comparison found differences of $21 000 and $30 000 between regimens containing bevacizumab or cetuximab and those using only fluorouracil and leucovorin.6 Overall, the cost of a standard regimen for treating advanced colorectal cancer has risen from about $500 to about $250 000 since the introduction of these new compounds.7
Meanwhile, the lung cancer drug gefitinib costs about $1800 per month and must be taken for months or even years; erlotinib, another biologic for metastatic non—small-cell lung cancer, costs about $2500 per month. Neither of these agents cure the disease; they only slow progression and extend survival time, enabling physicians to treat the cancer as a chronic condition.8 This is a paradigmatic shift in how cancer is treated. Rather than an acute disease with a high mortality rate, cancer is increasingly being viewed as a chronic disease, with some patients possibly receiving treatment with these and other equally expensive drugs for years, even decades.
As a result, health insurers are increasingly focusing on oncology costs. One survey found that 58% of responding health plans said they planned to focus on oncology in 2005 as part of a national movement towards specialty pharmacy programs to manage costs.9
The "oncologic time bomb" predicted by an American Cancer Society task force in 1999 has exploded.10
How health insurers, patients, healthcare professionals, and employers approach these rising costs over the next decades has tremendous implications for the entire healthcare system. One survey of 139 medical oncologists at 2 large academic hospitals found that although the majority (78%) said patients should have access to effective care regardless of cost, 71% said that rising costs would result in more rationing in the area of oncology treatment over the next 5 years.11
Burden of Illness in Oncology
Any discussion of oncologic costs must, of course, move beyond dollars and cents. Pertinent issues include not only direct medical costs, but indirect costs, survival, quality of life, health-related quality of life (HRQOL), and psychological issues.
Understanding the scope of these costs and benefits enables policymakers, researchers, insurers, and employers to evaluate the individual and societal benefits of oncology expenditures on the prevention and treatment sides, as well as across diverse racial, ethnic, socioeconomic, and geographic differences.
Framework for Evaluating Oncology Costs and Consequences
For decades, oncology outcomes have been measured in clinical terms, such as mortality or survival; progression-free survival (the time until the tumor begins to grow again); and clinical end points (tumor size, duration of response, hematological markers). With rising costs and greater treatment options, these relatively simple outcome markers are no longer sufficient.
Today, treatment decisions must also rely on economic and humanistic costs and outcomes associated with a given treatment, including direct and indirect medical costs, the treatment's effect on the patient's overall quality of life, and any benefits to the patient and society.12 In other words, how well do treatment options efficiently achieve desired outcomes?13
The importance of evaluating the cost of disease and value of treatments cannot be underestimated in today's healthcare system. With 40 million uninsured Americans, and healthcare costs making up 16% of the nation's economy (compared with 13.7% in 2000), the appropriate allocation of resources is a critical issue.14,15
To approach this issue in a systematic manner, researchers use outcomes research and pharmacoeconomics. Outcomes research involves the scientific and methodologically sound collection of data on the costs and consequences of various therapies. The data are then used to improve health status as part of a process known as outcomes management.16 Generally, 3 outcome dimensions may be considered: clinical, economic, and humanistic.12
Although traditional medical decision making has always included the latter 2 categories on a subjective basis dependent on the individual care provider, all 3 dimensions should be considered simultaneously when comparing treatment alternatives. Assessing the clinical, economic, and humanistic costs and benefits of a treatment regimen provides insight into the value of the respective treatments.12
Pharmacoeconomics is a component of outcomes research that brings a "systemic approach to the collection and analysis of data in decisions regarding the selection and use of pharmaceutical products and services."12
Pharmacoeconomics offers a set of tools to help decision makers better understand the value of treatment regimens. These methods provide a measure of the efficiency of treatments by systematically balancing the cost and consequences of treatment alternatives. There are 4 main pharmacoeconomic approaches to assess the value and efficiency of treatment interventions: cost-benefit analysis, cost-effectiveness analysis, cost-minimization analysis, and cost-utility analysis.
Cost-benefit analysis compares 2 or more interventions in terms of the total dollar cost of an intervention compared with the total monetary benefits. A cost-benefit ratio provides a measure of the "return on investment" for a given intervention or treatment. Interventions can then be compared to determine which provides the greatest dollar benefits for each dollar expended. For instance, suppose tamoxifen costs less than letrozole, but clinical studies find that letrozole is more effective as a postadjuvant breast cancer treatment in reducing the risk of recurrence (thus reducing direct and indirect costs from future medical costs and lost productivity). If the ratio of the dollar value of these benefits (decreased cost and improved productivity) to the dollar cost of treatment with letrozole is greater than the same ratio for tamoxifen, then letrozole provides the more efficient regimen.
Cost-effectiveness analysis compares the cost and consequences of 2 or more alternatives with a common therapeutic objective. In the case of tamoxifen and letrozole, the question might be not only which is the most cost effective with the greatest financially-based benefits, but which has the greater nonmonetary outcomes, such as years of life saved, hospital days avoided, tumor shrinkage, etc. The cost-effectiveness ratio would be calculated as the dollar cost per nonmonetary outcome (eg $500/years of life saved). The treatment with the lowest cost per outcome is the more efficient regimen. This is the approach used most often today, particularly in light of high-cost therapies that may or may not significantly extend life. In today's healthcare environment, a new treatment seldom displaces all other treatment options. Instead, we must decide if the added benefits of the new treatment are worth the additional costs, compared to existing treatments. This decision is aided by an "incremental analysis" where the incremental change in costs of 2 treatments is compared to the incremental change in benefits of the treatments. Incremental cost-effectiveness ratios quantify the cost per unit of benefit gained from using one treatment versus another. Generally, an incremental analysis should be included when 2 or more treatments are evaluated within a cost-effectiveness framework.
Cost-minimization analysis is used when all relevant and available treatment choices have clinically equivalent outcomes but different costs. In this instance, the approach chosen is based entirely on the lowest treatment cost. With a cost-minimization analysis, the assumption of equivalent treatment outcomes should be clearly supported by clinical evidence.
Finally, cost-utility analysis compares 2 or more treatments in which costs are measured in dollars and outcomes are measured in quality-adjusted life-years (QALYs), which adjust treatment outcomes for patient preferences. The approach that yields the lowest cost per QALY is the more efficient treatment.17 Cost-utility analysis is particularly relevant when regimens differentially affect quality of life during and after treatment. Consider, for example, 2 chemotherapy regimens that are equivalent in terms of their survival benefit. The 2 regimens differ, however, in terms of their emetogenic potential. It is reasonable to assume that the least emetogenic treatment would be preferred. Cost-utility analysis adjusts the outcome to account for patient preferences for the regimen with the least amount of nausea.
Interestingly, an analysis by Tengs comparing cost-effectiveness estimates with cost-utility estimates found little difference between the 2 unless the intervention had significant side effects that greatly reduced quality of life or the intervention itself greatly improved quality of life over the long term.18
Choosing the perspective.
Before implementing an economic analysis of treatment interventions, the perspective from which the costs and benefits will be calculated must be determined. Will it be from the perspective of the individual patient and his or her family? The payer? The provider? Or society as a whole? Specification of study perspective is crucial in identifying relevant cost and outcomes.
For instance, from the patient perspective the most important outcome may be additional years of life and overall quality of life. From the provider perspective, the most important outcome may be overall reimbursement and clinical end points of treatment. However, from society's perspective, the more important outcomes may be reducing costs associated with absenteeism and decreased productivity as well as the loss of an individual's contribution to society.19
Once the perspective is determined, the actual measurement of costs commences. Cost can be categorized as direct medical costs, direct nonmedical costs, indirect costs (lost productivity), and humanistic costs (quality of life, patient satisfaction).
Direct medical costs.
Direct medical costs include not just pharmaceutical costs, but all costs relating to treatment, including laboratory testing, physician visits, surgical costs, inpatient care, and imaging costs. Identifying direct medical costs for treatment of a specific disease can help policymakers, payers, and clinicians identify areas of focus. For instance, one study modeling the lifetime direct costs of treating metastatic breast cancer in the United States found that the combined cost of all therapies (ie, chemotherapy, hormonal therapy, and radiation therapy) accounted for just 12% of total medical costs. This suggests that focusing on other direct medical costs, specifically hospital and terminal care, would be an appropriate method for controlling or reducing costs associated with the final stages of the disease.20
Direct nonmedical costs.
These costs typically fall on the patient and his/her family, and include items such as the cost of transportation to and from healthcare provider offices and hospitals; lodging for long-term treatment courses, such as autologous bone marrow transplants; and costs for lifestyle items like wigs and prostheses as treatment progresses. Nonmedical costs for the physician may include unreimbursed care.
These costs are composed primarily of the patient's lost productivity in the workplace, home, and society as a whole, both from the illness and premature death, as well as time and productivity lost by the patient's family, friends, and others involved in the patient's care.19
Two approaches commonly used in estimating costs associated with lost or diminished productivity are the human capital approach and the willingness-to-pay approach. The human capital approach is based primarily on the value of wages lost from decreased productivity for those in the workforce.
For example, according to Chang et al, higher absenteeism rates for employees with cancer translated into higher costs ($373 vs $101; <.05). Cancer patients also had, on average, more monthly short-term disability days than controls (5.2 vs 0.2), resulting in mean monthly costs of $698 vs $25 (<.05). Additionally, cancer patients' caregivers had higher mean monthly costs of deductibles and copayments and also had more absenteeism per month than caregivers of the control group.21
The willingness-to-pay approach measures productivity costs based not only on lost wages, but also on the value the patient places on their own health and well-being.22 Willingness-to-pay basically estimates what people would be willing to pay to avoid the illness episode and its sequelae.
A market value may be used even for those not in the workforce. This market value considers what the patient would have been paid for nonworkforce activities, such as housekeeping and child rearing. Additionally, an opportunity-cost approach may be used, which considers those wages that might have been lost as a result of the patient's illness if the patient had been employed outside the home.17
The direct effect of cancer-related treatment on patients' daily activities may also be considered. A study of ovarian cancer patients who experienced chemotherapy-related hematologic or neurologic side effects found that indirect costs for patient and caregiver work loss and caregiver support payments were greater, in most instances, than the direct medical costs for thrombocytopenia and neurotoxicity.23
Humanistic costs and outcomes consider the effect of treatment on a patient's HRQOL and satisfaction with that treatment. These measures have gained acceptability in assessing the value of various cancer treatments beyond traditional clinical outcomes. Given the growing understanding that traditional mortality end points may not reflect the actual effect of cancer and its treatment on patients and that many cancers have no cure, these measures are being used as end points for clinical trials of new cancer therapies.19,24 The humanistic value of any treatment regimen should be balanced against the clinical and economic consequences of the treatment.
Although similar, HRQOL and patient satisfaction represent different constructs. Health-related quality of life is a multidimensional construct that captures the effects of an illness or its treatment on aspects that are relevant and important to a patient's well-being. These dimensions typically include physical, emotional, and social functioning as well as symptoms.25
All treatments affect patients' HRQOL in a positive or negative way. For instance, any adverse events related to pharmaceutical treatment have an effect, as does the time and energy required to receive those treatments.
In addition to its use as a primary or secondary end point in clinical cancer trials, HRQOL is also used to compare the effects of interventions, to assess funding priorities at the National Institutes of Health against the overall burden of the disease, to index the relative burden of cancer for individual nations or groups of nations, and to evaluate the performance of healthcare providers and managed care plans.13,26 Considering HRQOL issues also helps identify the effects of treatment changes on patient well-being and predict survival.27
Satisfaction with care measures how well a particular treatment meets the expectation of the patient. These expectations may involve aspects of quality of life or the ability to function at a particular level.28
Including patient preferences for a given treatment and its consequences in an outcome analysis can have significant effects. For example, a cost-utility analysis of paclitaxel in combination with cisplatin for patients with advanced ovarian cancer published in 1997 found that although first-line treatment costs with the drugs were about 4 times higher per cycle than the then-standard treatment of cyclophosphamide and cisplatin, incorporating patient treatment preferences into the model revealed a much smaller cost difference per quality-adjusted, progression-free year.29
Numerous tools are available to assess HRQOL from a general (generic) or disease-specific perspective across a range of illnesses. One of the most widely used generic instruments is the MOS SF-36. If one is interested in the effects of a particular aspect of a therapy or its treatment, a disease- specific HRQOL instrument is more appropriate. Numerous cancer-specific instruments, such as the FACT-Anemia and the Prostate Cancer Treatment Outcomes Questionnaire, exist. The choice of instrument depends on the patient, the disease, and the treatment under investigation.28
Incorporating HRQOL and Patient Preferences
Clegg et al incorporated HRQOL and patient preferences into a cost-effectiveness and cost-utility analysis of 4 drugs for treating non—small-cell lung cancer: vinorelbine, gemcitabine, paclitaxel, and docetaxel.30 The review found that although survival gains of a few months were modest with the new compounds, they were clinically significant relative to survival in the untreated group. The researchers also found the treatment resulted in gains in quality of life compared with best supportive care, because some of these newer forms of chemotherapy had fewer negative effects on HRQOL.
The analysis concluded that the newer drugs extended life at a cost for year of life gained (YLG) that was much lower than for other treatments.30 This study illustrates the importance of considering the impact of treatment choice on patient quality of life and satisfaction with care. Therapies that minimize adverse events or meet patient expectations are more likely to be used and accepted by the patient. Patient acceptance translates into improved adherence, which should have a positive effect on treatment outcomes.
Economic Modeling in Cancer Care
Economic models are used to describe the essential elements, consequences, and complications of a treatment decision under conditions of uncertainty. As such, models provide estimates when data are either not available or are insufficient to make a decision. They can also be used to extrapolate existing data to predict other events.
Models are also used when the relevant clinical trials have not been conducted or the requisite data are not collected as part of the trial and to extrapolate long-term outcomes, such as survival, when only short-term end points, such as tumor size or neutrophil count, were measured.
Decision analysis is a systematic approach to evaluating a problem or decision that incorporates both clinical and economic costs and consequences of the decision under conditions of uncertainty. Decision analysis enables the evaluation of complex clinical choices after weighing the risks and benefits of alternatives and the likelihood (probability) that any of these alternatives will occur. Without this approach, data for such decisions might take years to collect through clinical trials. Decision analysis is often used when competing treatment interventions, each with its own tradeoffs, are available and a specific treatment decision must be reached.
Decision analysis begins with a question, such as, "Which treatment option maximizes the outcome of interest?" The question should be framed by factors such as target patient population, diagnosis and treatments, reason for the analysis, and study perspective. For instance, is the decision analysis being conducted to determine treatment options for an individual patient or to guide policy decisions?
To use decision analysis, there must be a criterion for the optimal decision, whether that is maximizing survivability, life expectancy, quality of life, or cost effectiveness.31 Decision analysis is often used as a tool to estimate average or expected cost in a pharmacoeconomic evaluation.
Data for decision analysis models are obtained from a variety of sources, including clinical trials, meta-analyses, administrative or medical databases, and expert opinions. In addition to specifying the sources of economic and clinical data, decision analysis models should also make all assumptions transparent to the reader.
Assumptions for a decision analysis may include estimates for length of life, medical costs, levels of disability, and effect on quality of life.32
Decision analysis models are often presented in the form of a "decision tree" with symbols such as squares, circles, or triangles indicating key decision points. Square nodes are "decision nodes," because the branches represent a potential option or choice. Circular nodes are "chance or probability nodes," because each branch represents future outcomes that are beyond the control of the decision maker and are uncertain.33 Chance nodes have probabilities of occurrence associated with each possible outcome. Finally, the triangle node represents the end point or outcome cost.
A simple decision tree is depicted in the Figure. This decision tree depicts a choice between 2 chemotherapy agents, both of which are effective in treating the cancer. In this example, Drug X has a lower incidence of nausea than Drug Y (30% vs 40%). Drug X is also slightly more efficacious than Drug Y, but also more expensive. Dollar amounts displayed at the terminal nodes (triangles) represent the total cost of treatment, including drug cost, for the particular arm of the decision tree. When the average or expected cost is calculated, Drug X is shown to have a lower average cost per patient treatment ($3080 vs $3124).
In the oncology arena, decision analysis has been used to model a variety of treatment decisions, such as the use of bone marrow transplants, colony stimulating factors, prostatic-specific antigen testing, and screening mammography. When complete, these decision analysis models provide not only an "optimum" solution but also a clear delineation of the underlying structure and assumptions used. Decision analysis models can be modified as new or additional information about treatment options becomes available.
Decision analysis may use static or dynamic models. A static model is best used when the event in question has a relatively well-defined start and end points that occur within a finite period of time, such as the development of community-acquired pneumonia. The onset, progression, and treatment of this acute condition is well defined (the condition is typically "cured" within a few weeks) and thus well-suited to a static model.
A dynamic model, on the other hand, is more useful for chronic conditions, such as heart failure or even cancer, in which the patient improves but then has a recurrence. In this case, the patient is in various health states over a long period of time. Conditions such as these are more amenable to dynamic models, like Markov models, where the model is more fluid and the patient is allowed to move among various health states as the illness improves or progresses.
Examples of Economic Models
In Lee et al, researchers used a Markov decision analysis to determine the cost effectiveness of postmastectomy radiation therapy (PMRT) in high-risk premenopausal breast cancer patients (those with positive lymph nodes). They derived clinical data for the model from a large meta-analysis of adjuvant systemic therapy trials for breast cancer with and without PMRT and estimated cost data from other literature.34
The model estimated the number of recurrences, relapse-free and overall survival, and costs over a 15-year period using a discount rate of 3%. Cost-effectiveness ratios were calculated per incremental QALY and YLG. One- and two-way sensitivity analyses were performed to determine the sensitivity of results to clinical and economic assumptions.
The analysis found that PMRT reduced the risk of relapse by 31%, projecting a 15- year relapse-free survival rate of 52% with PMRT and 43% without. PMRT increased overall survival from 48% to 55%, resulting in an incremental 0.29 YLG per patient. Fifteen-year costs also increased from $40 800 to $48 100, which translated into a cost per additional YLG of $24 900. Based on their analysis, the authors concluded that PMRT offered "substantial clinical benefits" in a cost-effective manner.
Another example comes from Sanders et al, who used decision analysis to assess the cost effectiveness of a potential vaccine for human papillomavirus (HPV) as a way to prevent cervical neoplasms. Cervical cancer is one of the most common malignancies in women, diagnosed in about 13 000 women per year and responsible for the death of about 4000 women per year.35
Researchers built a decision tree to evaluate length of life and cost for vaccinating adolescent girls against high-risk types of HPV. The target population was all adolescent girls in the United States, with a base case of 12-year-old girls (sexual activity before age 12 is considered rare). This assumption was tested in a sensitivity analysis.
The analysis assumed universal vaccination, although during the sensitivity analyses the researchers did evaluate the cost effectiveness of targeting only high-risk girls. The model included data on incidence of HPV infection; low-versus high-risk HPV types; rates of HPV progression; cancer surveillance, treatment, and progression; benign hysterectomy (assuming such women were protected from cervical cancer); HPV vaccine characteristics; and the effects of HPV infection and cervical cancer on quality of life and costs.
The researchers found that although a prophylactic vaccine against high-risk HPV types was more expensive than current practice, it resulted in greater quality-adjusted life expectancy. They predicted that such vaccination would avoid more than 224 255 cases of HPV; 112 710 cases of squamous intraepithelial lesion, a precancerous condition; 3317 cases of cervical cancer; and 1340 deaths related to those cancers. Their sensitivity analyses found the vaccine would be cost effective, even assuming booster shots every 3 years and an efficacy as low as 40%.35
Methods of Cost Control in Cancer Care
During the late 1990s and early part of this century, the growth of managed care organizations began to decline. Simultaneously, health insurance premiums began to skyrocket for both employers and employees, rising by 13.9% in 2003 (the highest increase since 1990), 11.2% in 2004, and 9.2% in 2005. Since 2000, premiums for family coverage have increased 73% compared with inflation growth of 14%.36
Tools such as pharmacoeconomics, decision analysis, clinical guidelines, and outcomes research help health plans and employers determine which therapies are most effective and under which conditions they are most efficient. The challenge of health pl
Address correspondence to: Claiborne E. Reeder, RPh, PhD, University of South Carolina College of Pharmacy, 715 Sumter Street, Room 109, Coker Life Sciences Bldg, Columbia, SC 29208; firstname.lastname@example.org.
1. American Cancer Society. United States cancer statistics, 2002. Incidence and mortality. Available at: http://www.cdc.gov/cancer/npcr/uscs/index.htm. Accessed January 11, 2006.
J Natl Cancer Inst.
2. Edwards BK, Brown ML, Wingo PA, et al. Annual report to the nation on the status of cancer, 1975-2002, featuring population-based trends in cancer treatment. 2005;97:1407-1427.
3. Cancer Trends Progress Reportâ€”2005 Update. Available at: http://progressreport.cancer.go. Accessed January 5, 2006.
4. Medicare Payment Advisory Commission. Report to the Congress: Medicare payment policy. Washington, DC: MedPAC; March 2001.
N Engl J
5. Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. 2004;350:2335-2342.
N Engl J Med.
6. Schrag D. The price tag on progressâ€”chemotherapy for colorectal cancer. 2004;351:317-319.
Wall Street Journal.
7. Marcus AD. Price becomes factor in cancer treatment. September 7, 2004:D1, D7.
8. Appleby C. Payers' costly battle against cancer. 2005;19-25.
Specialty Pharmacy News
9. MCOs' oncology strategies focus on provider issues for 2005. . 2005;2:1-3.
10. Congialose CB, Blair, AE, Borchardt JS, et al. Purchasing oncology services. 2000;88:2876-2886.
11. Nadler ES, Eckert B, Neumann PJ. Do oncologists believe new cancer drugs offer good value? Abstract presented at: American Society of Clinical Oncology; May 13-17, 2005; Orlando, Fla. Abstract 6011.
12. Kozma CM, Reeder CE, Schulz RM. Economic, clinical, and humanistic outcomes: a planning model for pharmacoeconomic research. 1993;15:1121-1132.
and Practices of Managed Care Pharmacy.
13. Reeder CE, Kozma CM, McCollam AR. Overview of pharmaceutical outcomes, pharmacoeconomics, and quality of life. In: Ito SM, Blackburn SA, eds. Alexandria, Va: Academy of Managed Care Pharmacy; 1995:103-112.
14. Smith C, Cowan C, Heffler S, Catlin A. National health spending in 2004: recent slowdown led by prescription drug spending. 2006;25:186-196.
15. Centers for Medicare & Medicaid Services, Office of the Actuary, National Health Statistics Group. National Health Expenditures. http://www.cms.hhs.gov/NationalHealthExpendData/