Accuracy of physicians' estimates of monthly drug costs.
Reichert S, Simon T, Halm EA. Physicians' Attitudes About Prescribing and Knowledge of the Costs of Common Medications. Arch Intern Med. 2000;160(18):2799-2803. doi:10.1001/archinte.160.18.2799
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2000
Compliance with medical therapy is often compromised because patients cannot afford to pay for medications. Inadequate physician knowledge of drug costs may unwittingly contribute to this problem.
To measure attitudes about prescribing and knowledge of medication costs and compare differences among attending physicians and residents.
Written survey of internal medicine house staff and general medicine attending physicians in an urban hospital-based primary care center.
One hundred thirty-four of 189 physicians responded (71% response rate). Seventy percent of respondents were house officers and 30% were attending physicians. Eighty-eight percent of physicians felt the cost of medicines was an important consideration in the prescribing decision, and 71% were willing to sacrifice some degree of efficacy to make drugs more affordable for their patients. However, 80% often felt unaware of the actual costs. Only 33% had easy access to drug cost data, and only 13% had been formally educated about drug costs. Regarding insurance coverage, 94% of physicians gave strong consideration to the cost of medications when patients were self-paying, 68% when patients had Medicare, and 30% when patients had Medicaid or were participants in a health maintenance organization with a prescription plan. Physicians' estimates of the cost of a month's supply of 33 commonly used medications were accurate in 45% of cases, too low for 40%, and too high for 15%. The costs of brand-name and expensive drugs were most likely to be underestimated. House officers were less cost-conscious than attending physicians.
Physicians were predisposed to being cost-conscious in their prescribing habits, but lacked accurate knowledge about actual costs and insurance coverage of drugs. Interventions are needed to educate physicians about drug costs and provide them with reliable, easily accessible cost information in real-world practice.
PRESCRIPTION drugs accounted for $93 billion in health care costs in 1998 and are the fastest growing component of health expenditures.1,2 Total drug expenditures are estimated to have reached $120 billion in 1999, making medication costs the second most expensive item in the national health care budget, surpassing the cost of hospital care.1- 3 These dramatically rising pharmaceutical costs are due in part to the growth in number of drugs dispensed, the high price of new agents, and direct advertising to consumers. While physicians are inundated with information about the availability and efficacy of drugs, they receive little information about actual drug costs in medical school, in residency training, or once in practice.
Studies conducted in the 1980s and early 1990s found poor knowledge of medication costs among family practitioners, neurologists, geriatricians, and pediatricians.4- 13 Over the last decade, the rise of managed care with its emphasis on containing costs has had an enormous influence on the practice of medicine. Whether or not increasing cost-consciousness has influenced physicians' attitudes about prescribing and knowledge of medication costs is unknown, and no studies have examined attitudes and knowledge about drug costs among general internists, who are the primary providers for a major proportion of adults.
These knowledge deficits, if they still exist, are important for 2 reasons. First, inattention to cost-effective prescribing contributes to the inefficient use of societal resources and rising pharmaceutical spending. Second, because many patients must pay for the full cost of their medications, expensive prescriptions may go unfilled or may be used less frequently than directed, resulting in compromised patient health.14
This topic has risen to national prominence on the health policy agenda regarding the absence of prescription drug coverage by Medicare and its potential consequences.15- 18 Eighty-five percent of Medicare beneficiaries receive at least one prescription each year, though nearly one third lack any supplemental insurance coverage of medications.18,19 Medicare beneficiaries with incomes near the poverty line, but who are not poor enough to qualify for Medicaid assistance, spend as much as one third of their incomes on out-of-pocket drug expenses.18,19 To make matters worse, patients in poor health have higher overall medication costs, but tend to have lower incomes.
The purpose of this survey was to measure attitudes about prescribing and knowledge of drug costs among general medicine attending physicians and internal medicine residents. We were interested in assessing the extent to which physicians are willing to consider the cost of medications and a patient's insurance coverage in the prescribing decision. We also sought to measure physicians' knowledge of actual drug costs. We were additionally interested in assessing differences in knowledge and attitudes between attending physicians and house officers.
We surveyed internal medicine house staff (n = 145) and full-time and part-time attending physicians (n = 44) in the Division of General Internal Medicine at the Mount Sinai campus of the Mount Sinai–New York University Medical Center, New York, NY. These physicians staff and write nearly all the prescriptions for the main hospital-based primary care clinic as well as several small, private outpatient practices. The written survey was 3 pages, anonymous, and self-administered. It was distributed in February 1998. We sent a reminder notice and replacement questionnaires to all nonrespondents in March 1998.
Demographic variables were level of training, year of graduation from medical school, and sex. Physicians were asked to agree or disagree with 8 medication attitude statements on a 5-point Likert scale (1, strongly disagree; 2, somewhat disagree; 3, no opinion; 4, somewhat agree; and 5, strongly agree). We also asked them to agree or disagree that "The cost of medications is more of a concern to me when my patient's insurance status is: Medicare, Medicaid, self-pay, or HMO [health maintenance organization] with prescription plan." Another question asked about the sources respondents use to obtain information about drug costs. The second half of the survey had physicians estimate the average wholesale price (AWP) of a 30-day supply of 33 commonly used outpatient medications (Table 1). Each medication was listed by generic and brand name (if appropriate), dose, and frequency. Respondents were given a choice of 5 different price categories: $1 to $10, $11 to $30, $31 to $50, $51 to $80, and >$80. These 33 medications were selected from longer lists of the 100 most frequently prescribed drugs in the United States, adapted to reflect pharmaceuticals commonly used in our local primary care practices. The list represents a variety of brand-name and generic medications of varying costs across several therapeutic classes including asthma and allergy, diabetes mellitus, analgesics, antihypertensives, antibiotics, and antisecretory agents, among others.
We report agreement with the drug attitude statements (4, somewhat agree, and 5, strongly agree). Physician cost estimates were compared with the standard AWP listed in the 1998 Drug Topics Red Book.20 When the actual AWP fell into the same price category as the respondent estimate, this was counted as an accurate answer. When the AWP was greater than the physician estimate, this was an underestimate, and vice versa for overestimates. Drug prices that were estimated correctly by half or more of respondents were considered to be correctly estimated. We used χ2, Fisher exact, and Wilcoxon tests to evaluate differences between responders and nonresponders and between attending physicians and house officers. In order to compare the accuracy of cost estimates between attending physicians and house officers, we created a total cost score in which price category estimates were converted into standardized z scores (ranging from + 5 [overestimate by 5 categories], to –5 [underestimate by 5 categories], with 0 indicating accurate). The total cost score was normally distributed. Negative total cost scores were interpreted as overall underestimation of drug costs. Two-sided values of P≤.05 were considered statistically significant. All analyses were performed with PC SAS 6.12 statistical software (Statistical Analysis Systems Inc, Cary, NC).
We received completed surveys from 134 of 189 physicians (71% response rate; 66% of house officers and 86% of attending physicians). The characteristics of respondents are summarized in Table 2. Seventy percent of those responding were house officers. The remaining 30% were attending physicians, who had a median of 9 years in clinical practice (range, 6-20 years). Response rates were similar across training levels.
Overall, 88% of respondents felt that cost was an important consideration when making medication choices, and 71% were willing to sacrifice some degree of efficacy in order to make drugs more affordable for their patients. However, 80% often felt unaware of the actual costs of medications, and only 13% reported ever having any formal education about the cost of medications. Sixteen percent of physicians reported asking patients about the costs of their medications. Only 8% of physicians preferred brand-name medications over generic drugs regardless of cost.
One third of respondents (33%) reported easy access to drug cost information. The resources they reported using were The Medical Letter on Drugs and Therapeutics (65%), other physicians (52%), patients (45%), pharmacists (41%), drug company representatives (25%), advertisements (17%), and journals (10%).
Physicians' concerns for costs of medications varied widely based upon their patients' insurance status. When patients had no medical insurance, 94% of respondents agreed that the cost of medicines was a strong concern. For a patient with Medicare coverage, two thirds of physicians (68%) agreed that cost was a concern. Fewer than 1 in 3 physicians (30%) felt that cost was a concern for patients with Medicaid insurance or HMO prescription plans.
In order to measure physicians' knowledge of medication costs, we asked them to estimate the AWP of a month's supply using 5 cost categories. Of the 33 commonly used drugs that we asked about, the majority of physicians were accurate in their estimates for 45% of the drugs (n = 15) and inaccurate in their cost estimates for 55% (n = 18); 40% (n = 13) were underestimated and 15% (n = 5) were overestimated. The mean ± SD total cost accuracy score was –0.74 ± 8.09, indicating an overall tendency to underestimate costs (range, 19 to +24). The prices of 90% of the generic drugs were correctly estimated, while the prices of 52% of brand-name drugs were underestimated (Figure 1). The prices of 91% of the expensive drugs (cost >$80/month) and the prices of 86% of the high-priced drugs ($51-$80/month) were underestimated, while the prices of 80% of the most inexpensive and moderately priced drugs (<$30/month) were correctly estimated.
Attending physicians were more cost-conscious than house officers (Table 3). Comparing the proportion of overall underestimation of drug costs using the total cost accuracy score, house officers were 50% more likely to underestimate drug costs than attending physicians (relative risk [RR], 1.5; 95% confidence interval [CI], 1.01-2.22). We found similar results when we compared the mean ± SD total cost scores for the 2 groups (house officers, –1.7 ± 8.1; attending physicians, 1.5 ± 7.6; P = .04). In addition, house officers were more likely to report feeling unaware of medication costs (RR, 1.36; 95% CI, 1.10-1.68), but they were less likely to agree that cost was an important concern for patients with Medicare insurance (RR, 0.74; 95% CI, 0.58-0.94). The sources of drug cost information also differed by training level. Residents were twice as likely to ask other physicians for cost information and much less predisposed to query The Medical Letter on Drugs and Therapeutics or pharmacists. Comparing responses among house officers, there was a trend toward more senior residents having more knowledgeable and cost-conscious responses. There were no differences between the 2 groups in any of the other attitudes or practices we examined, including preference for brand-name drugs, willingness to sacrifice efficacy for affordability, or prescribing attitudes for self-paying, Medicaid, or HMO patients.
Too often, patients may fail to fill prescriptions or may ration their medications because they are too costly.14 The ability to pay for expensive medications is most strained for those without a drug coverage benefit (no insurance or Medicare alone) and of limited financial means. For physicians who care for the poor and elderly, such occurrences are all too common and contribute to suboptimal control of chronic diseases. Physicians may unwittingly exacerbate problems in many of these cases when they prescribe drugs without knowing the cost of medicines for their patients.
In our survey of general medicine attending physicians and internal medicine residents, we found reasons for optimism and concern. The good news was that, in general, physicians appeared predisposed to practice cost-effective medicine. Nearly all respondents felt that the cost of medications was important, and most were willing to sacrifice some degree of clinical efficacy to make therapy more affordable. However, the bad news was that the providers we surveyed were not well equipped with the requisite knowledge or cost information resources needed to facilitate cost-effective prescribing in everyday practice. Four out of 5 physicians indicated that they were often unaware of actual drug costs, and most underestimated the cost of common brand-name medications we inquired about. Poor knowledge about medication costs was most acute among house staff, who were 50% more likely than attending physicians to underestimate prices.
We were also struck by the fact that nearly one third of physicians did not appear to understand that Medicare does not pay for medications. This misunderstanding was much more common among house staff. The substantial economic burden borne by Medicare beneficiaries without supplemental insurance to pay for their medications has recently emerged as a major national health policy issue.15- 19 In our own practice, the high price of medications for Medicare patients of modest financial means is a frequent contributor to poor compliance and suboptimal control of chronic medical problems such as hypertension, diabetes, and asthma. From the perspective of an individual patient, it makes sense that physicians are much less concerned about the out-of-pocket cost of drugs for persons with Medicaid or HMO drug plan coverage. However, from a societal perspective, the prospect of practicing cost-effectively for only a subset of patients may prove to be detrimental.
Our results confirm the findings of studies in the 1980s and early 1990s that physicians believe that the costs of drugs are important,6,10,13 while their actual knowledge of costs was poor.4- 7,9,11,12 Thus, primary care/internal medicine physicians in 1998 appear to be no more knowledgeable about costs than their family practice, geriatrics, pediatrics, and neurology colleagues were a decade ago.4,6,7,12 While we confirmed previous reports that physicians underestimate the cost of expensive medicines and overestimate the cost of inexpensive ones,4,6,8,12 we found that their accuracy was more closely related to whether a drug was a brand name or generic. In contrast to previous studies of pediatricians and family practitioners,6,7 we found differences in attitudes between attending physicians and house officers. We suspect that variations in study design and statistical power may explain these discrepancies.
Two factors likely contribute to continued inadequate knowledge of medication costs. First, medical schools and residency training programs provide little or no formal education about medication costs and insurance coverage of pharmaceuticals. Second, there are few ways to obtain reliable drug price information in a timely fashion. Cost information is rarely if ever included in medical journals, textbooks, or drug-prescribing guides (including the Physicians' Desk Reference).
While the most commonly cited cost reference was The Medical Letter on Drugs and Therapeutics, a peer-reviewed publication, we were somewhat troubled by reports of relying on drug company representatives and advertisements for price data. The accuracy of information provided by pharmaceutical sales representatives has been questioned.21,22 In addition, greater exposure to drug representatives has been associated with higher prescribing costs among primary care physicians.23 The greater use of pharmaceutical representatives by attending physicians compared with house staff may reflect the fact that policies governing interactions between attending physicians and industry personnel at our institution are less restrictive than those for residents. However, since house staff were more likely to ask other physicians, presumably attending physicians, for drug cost information, we are concerned about the propagation of biased or inaccurate information. Few physicians have had sufficient training during medical school and residency regarding professional interaction with sales representatives.24
Several limitations of our study are worth acknowledging. We had 2 methodological challenges in trying to measure physician knowledge of drug costs: how cost estimates should be elicited, and what the criterion is for costs. Because of concerns about false precision and lower completion rates for surveys that require free-text responses, we presented 5 price categories. While we tried to space out the price categories across the range of actual costs using round number ranges (eg, $1-$10, $11-$30), the scale is inherently arbitrary. For some decisions about cost-effectiveness, competing therapies may be priced within a single price category. We decided to use the industry standard 1998 Drug Topics Red Book AWP for the "true" costs, a common convention.20 However, we are cognizant that there is often considerable variability in actual prices for the same medicine at different pharmacies, in different areas, and among different insurance plans. Because we surveyed internal medicine physicians affiliated with a single academic medical center, our results may not be generalizable to other settings or disciplines. Finally, the study was conducted in New York, where managed care penetration is intermediate and physicians' exposure to aggressive prescription cost-reduction programs (pharmacy benefit management firms and outpatient formularies) may be less than that experienced by providers in more mature managed care marketplaces.
Given that residents had the most significant knowledge deficits, training programs should include sessions on cost-effective prescribing and types of health plan drug coverage. In addition, trustworthy sources of drug cost information should be made widely available in residents' ambulatory clinics so that data can be found in real time when needed. At our own institution, we have developed a drug cost guide for 100 commonly used outpatient drugs that we have widely distributed as a pocket-sized booklet. We have also made it available on the institution's Intranet website, accessible from any physician workstation. We hope interventions like these will help empower physicians with the tools needed to practice cost-effective prescribing, foster better medication compliance, and ultimately improve health outcomes.
Accepted for publication March 2, 2000.
The authors thank all of the physicians who participated in this study.
Reprints: Ethan A. Halm, MD, MPH, Department of Health Policy, Box 1077, Mount Sinai Medical Center, One Gustave L Levy Place, New York, NY 10029 (e-mail: email@example.com).