POLICY
`Physicians’ Perceptions of Relevant
`Prescription Drug Costs: Do Costs to the
`Individual Patient or to the Population Matter Most?
`
`William H. Shrank, MD, MSHS; George J. Joseph, MS; Niteesh K. Choudhry, MD, PhD;
`Henry N. Young, PhD; Susan L. Ettner, PhD; Peter Glassman, MBBS, MSc;
`Steven M. Asch, MD, MPH; and Richard L. Kravitz, MD, MSPH
`
`Objectives: Physicians may be aware of at least 2 types of costs
`when prescribing: patient’s out-of-pocket costs and the actual costs
`of the medication. We evaluated physicians’ perceptions about rel-
`evant costs for prescription drugs and the importance of communi-
`cation about these costs.
`Study Design: Mailed survey to a random sample of 1200 physi-
`cian members of the California Medical Association, and a phone
`survey of a sample of nonresponders.
`Methods: Descriptive statistics of survey items, McNemar’s test
`to compare survey item responses, and logistic regression to eval-
`uate the relationship between physician, practice, and system vari-
`ables and physicians’ perceptions of relevant medication costs.
`Results: Of respondents with correct addresses, 49.6% respond-
`ed to the survey; 13% of nonresponders were contacted by phone.
`Approximately 91% and 80% of physicians reported that it is
`important to manage patients’ out-of-pocket costs and total med-
`ication costs, respectively. When comparing the relative impor-
`tance of managing the 2 types of costs, 59% of physicians agreed
`that managing patients’ out-of-pocket costs was more important
`than managing the total medication costs and only 16% disagreed.
`Physicians believed it was more important to discuss out-of-pocket
`costs than total costs with patients (P < .0001), but only 15% of
`physicians reported discussing out-of-pocket costs frequently and
`5% reported talking about total medication costs frequently.
`Physicians who managed more Medicare patients had a greater
`likelihood than physicians managing fewer Medicare patients of
`prioritizing out-of-pocket cost rather than total cost management
`(P = .038), and generalists had a greater likelihood than medical
`subspecialists (P = .046).
`Conclusions: Physicians prioritize managing out-of-pocket costs
`over total medication costs. Pharmacy benefit designs that use
`patient out-of-pocket cost incentives to influence utilization are
`addressing the costs to which physicians may be most responsive.
`When physicians face conflicts between managing patients’ out-
`of-pocket costs and total costs, they will likely try to protect the
`patients’ resources at the expense of the insurer or society. Efforts
`to align patients’, insurers’, and societies’ incentives will simplify
`prescribing decisions and result in better value in prescribing.
`(Am J Manag Care. 2006;12:545-551)
`
`Exposure to rapidly rising prescription drug costs
`
`has prompted insurers and policy makers to cre-
`ate various cost-control strategies to improve
`awareness of, and sensitivity to, these costs. In-
`terventions have been created to improve physicians’
`knowledge about the costs of medications in hopes that
`
`awareness of costs may lead to more value-based pre-
`scribing.1-4 Most notably, insurers have broadly imple-
`mented incentive-based pharmacy benefit plans to
`foster increased patient cost sensitivity, in hopes that
`consumer choice might steer prescribing toward more
`cost-effective medications.5 To serve as financial agents
`for patients and help them manage their out-of-pocket
`costs, physicians must also be aware of their patients’
`formularies and out-of-pocket cost requirements.
`These differing strategies highlight a critical problem
`for physicians. Physicians need to be aware of at least 2
`types of costs when prescribing: their patient’s out-of-
`pocket costs and the actual costs of the medication. For
`uninsured patients, managing a patient’s out-of-pocket
`costs is equivalent to managing the actual costs of med-
`ication. Most patients in the United States, however,
`have prescription drug coverage.6 Managing out-of-pock-
`et costs offers savings to individual patients, whereas
`managing total medication costs decreases overall pre-
`scription drug costs and may offer benefits to the popu-
`lation as a whole.
`Numerous studies have evaluated physicians’ knowl-
`edge about actual costs of prescription drugs, and have
`generally found that physicians estimate drug costs
`poorly.7-9 More recent research has shown that physi-
`cians are often unaware of patients’ formularies and out-
`of-pocket costs.10,11 Patients are often unaware of their
`costs for medications at the time of prescribing, and rely
`on physicians to be knowledgeable.12 These findings
`highlight the challenge of expecting physicians to be
`
`From the Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and
`Women’s Hospital, Boston, Mass (WHS, NKC); Harvard Medical School, Boston, Mass
`(WHS, NKC); VA Greater Los Angeles Healthcare System, Los Angeles, Calif (GJJ, SMA, PG);
`RAND Health (SMA, PG); the Department of Medicine, David Geffen School of Medicine
`at UCLA, Los Angeles, Calif (SLE, PG, SMA); the School of Pharmacy, University of
`Wisconsin-Madison, Madison, Wisc (HNY); and the Center for Health Services Research in
`Primary Care, University of California, Davis, Calif (RLK).
`This study was supported by The California Healthcare Foundation.
`Address correspondence to: William H. Shrank, MD, MSHS, Instructor, Division of
`Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital,
`Harvard Medical School, 1620 Tremont St, Ste 3030, Boston, MA 02120. E-mail:
`wshrank@partners.org.
`
`VOL. 12, NO. 9
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`Exhibit 1101
`IPR2017-00807
`ARGENTUM
`
`545
`
`© Ascend Media
`
`000001
`
`

`

`POLICY
`
`knowledgeable about both the patient’s out-of-pocket
`costs and the total costs of medications. Considering the
`low levels of knowledge of both types of costs, strategies
`to educate physicians about drug costs may be more
`effective if they focused on educating physicians about
`either, but not both, of these 2 kinds of costs, thus sim-
`plifying the issue for physicians.
`Little is known about what costs are most relevant to
`prescribing physicians; which costs they believe ought
`to be communicated to patients; and which costs they
`think are most valuable in their decision making. One
`qualitative study that evaluated this topic in Italy and
`the United Kingdom found that the physicians studied
`believed that costs to the system were as important as
`costs to individual patients.13 No US studies have evalu-
`ated whether physicians are more concerned with
`patients’ out-of-pocket costs or the total costs experi-
`enced by the population when prescribing. Findings in
`Europe may not be representative of perceptions in the
`United States, as cultural norms coupled with a more
`market-based healthcare delivery system may lead US
`physicians to emphasize management of the individual
`patient’s costs over societal costs.
`An evaluation of the costs that are relevant to physi-
`cians when prescribing medications could help explain
`the decision-making process when efforts to manage a
`patient’s out-of-pocket costs conflict with efforts to con-
`trol overall spending on medications. In addition, a bet-
`ter understanding of physicians’ perceptions about the
`relevant prescription drug costs and communication
`about those costs could assist in creating further inter-
`ventions that target the information that physicians
`would be most receptive to acting upon. This informa-
`tion could also assist in designing benefit programs that
`align physician, patient, and insurer incentives. Pro-
`viders have overwhelmingly adopted tiered pharmacy
`benefit designs to use patients’ out-of-pocket costs to
`steer prescription drug utilization. We surveyed physi-
`cians in California, a populous state with substantial
`managed care penetration and innovative pharmacy
`designs, to assess whether out-of-pocket cost-drivers are
`the most relevant costs to prescribing physicians.
`
`METHODS
`
`Study Sample
`Between June and December 2003, we surveyed a
`random sample of 1200 physician members of the Cali-
`fornia Medical Association (CMA). We selected our sam-
`ple size to estimate descriptive statistics, after
`conservatively predicting our response rate, with narrow
`confidence intervals (95% confidence intervals [CIs]
`with a width of less than 5 percentage points). Mem-
`
`bership in CMA includes approximately one third of
`California’s physicians.14 California was selected
`because it is a large and diverse state with substantial
`managed care penetration and reliance on formularies.
`We mailed each physician in our random sample an
`introductory letter followed by a survey including a $2
`gift certificate. The survey cover page identified the aca-
`demic affiliations of the investigators and stated that the
`survey was to be used for research purposes. Physicians
`who did not respond to the first survey were mailed up
`to 2 more surveys.
`Of the 1200 physicians who were mailed surveys,
`509 responded. An additional 173 addresses were iden-
`tified as incorrect because either the mailed surveys
`were returned to sender or phone calls to nonrespon-
`dents identified incorrect addresses. After removing
`incorrect addresses from the denominator, our response
`rate was 49.6%. We excluded an additional 34 physicians
`who reported that they do not prescribe in the outpa-
`tient setting either because of their scope of practice
`(radiologists, anesthesiologists, pathologists, administra-
`tors) or retirement.
`We did not have any baseline information about the
`physicians in our sample and could not comment on the
`differences in characteristics of survey respondents and
`nonrespondents. To better assess the generalizability of
`the respondents, we sequentially contacted by tele-
`phone a sample of physicians who did not respond to
`any of the 3 mailings and we performed an abbreviat-
`ed survey. We aimed to contact 10% of the nonrespon-
`ders by phone, but our resources allowed us to
`contact a total of 69 (13%) nonrespondents. We col-
`lected information about sociodemographics, several
`key predictor variables, and all dependent variables
`from nonrespondents.
`
`Survey Instrument
`The survey instrument was constructed through a
`collaborative and iterative process and was piloted
`extensively for validation purposes.10 The instrument
`included questions with multiple-choice, 5-point Likert
`scale, and discrete numerical responses.
`Dependent variables assessed physicians’ percep-
`tions about relevant costs at the time of prescribing. We
`discriminated between different types of costs by asking
`physicians to “refer to out-of-pocket costs (what the
`patient pays) or to the total costs of medications.” We
`asked physicians the extent to which they agree with
`the statement: “It is important to prescribe [a] drug that
`will minimize [their] patient’s out-of-pocket cost
`requirements when choosing among equally effective
`and safe medications.” We then asked physicians the
`extent to which they agree with a similar statement
`
`546
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`SEPTEMBER 2006
`
`000002
`
`

`

`Physicians’ Perceptions of Prescription Drug Costs
`
`referring to total medication costs. We also asked physi-
`cians whether they believe it is important to discuss out-
`of-pocket costs and total costs for prescription drugs,
`and asked physicians to estimate the frequency with
`which they engage in these discussions. Finally, we
`asked physicians the extent to which they agree or dis-
`agree with the following statement: “When choosing
`among equally effective and safe medications, I am more
`concerned with choosing the drug requiring the lowest
`out-of-pocket cost than choosing the drug with the low-
`est total cost.” All dependent variables used a 5-point
`Likert scale for responses (strongly agree, somewhat
`agree, neither agree nor disagree, somewhat agree,
`strongly agree).
`We collected a broad set of predictors to assess fac-
`tors that influence physicians’ perceptions about rele-
`vant costs when prescribing. In multivariate analyses,
`we controlled for physicians’ sociodemographic charac-
`teristics (age, sex) and specialty, and whether the prac-
`tice was hospital-based. We explored the relationship
`between patient insurance mix (ie, uninsured patients
`or patients in Medicare), practice characteristics (ie,
`practice size, location, number of prescriptions written,
`and the number of formularies prescribed from), and
`information technology systems use (computer order
`entry, Internet, handheld devices) with physicians’ per-
`ceptions about relevant costs.
`
`Data Analysis
`We used descriptive statistics to examine predictor
`variables and physicians’ perceptions about the impor-
`tance of managing patients’ out-of-pocket costs and total
`medication costs when prescribing. Physician responses
`were dichotomized for ease of interpretation. Physicians
`who somewhat or strongly agreed that out-of-pocket
`cost management is more important than total cost
`management were categorized together, while physi-
`cians who neither agreed nor disagreed, somewhat, or
`strongly disagreed with the statement were categorized
`together. We used logistic regression to evaluate the
`relationship between predictor variables and physicians’
`beliefs concerning the comparative importance of man-
`aging patients’ out-of-pocket costs versus the total med-
`ication costs. Missing
`independent variables
`in
`multivariate models were imputed using multiple impu-
`tations.15 Physicians with missing dependent variables
`were omitted from the analysis, and all independent
`variables had less than 20% missing values. Sensitivity
`analyses dropping respondents with missing values were
`qualitatively similar to models including those respon-
`dents. Data from the nonrespondent survey were not
`included in the final models reported. Rather, nonre-
`sponder data were used for sensitivity analyses. We cre-
`
`ated logistic models pooling data from responder and
`nonresponder surveys to see if inclusion of nonrespon-
`ders influenced the relationship between independent
`and dependent variables in our model. Statistical analy-
`ses were performed using SAS 9.1 (SAS Institute, Cary,
`NC; 2005) and Stata 8.1 software (Intercooled for
`Windows; STATA Corp, College Station, Tex; 2003).
`We calculated McNemar’s test statistics to compare
`responses to survey questions addressing beliefs about
`discussing out-of-pocket and total costs, and questions
`concerning the frequency with which physicians engage
`in such conversations. We performed t tests to compare
`characteristics of physicians in the nonresponder sur-
`vey with those of physicians who responded to the
`mailed survey.
`
`RESULTS
`
`Study Population
`Among respondents to the mailed survey, 475 physi-
`cians reported that they prescribe in the outpatient set-
`ting. Characteristics of respondents are in Table 1. On
`average, study physicians were 50 years old; 33% were
`generalists, 23% medical subspecialists, 18% surgeons,
`10% obstetrics and gynecology physicians, 8% emer-
`gency room physicians, and 3% psychiatrists. Physicians
`in this sample were similar to national averages in terms
`of age, sex, and specialty.16 Comparisons of mailed sur-
`vey respondents and nonrespondents who responded to
`our phone survey indicated no significant differences in
`age (mean 50.0 vs 52.9 years), sex (72% vs 78% male),
`number of prescriptions written (24.1 vs 22.2 prescrip-
`tions), or specialty.
`
`Physicians’ Perceptions of the Importance of
`Managing and Discussing Patients’ Out-of-pocket
`Costs and Total Medication Costs
`Physicians generally agreed with the statement,
`“When choosing between equally effective and safe med-
`ications, it is important to prescribe the drug that mini-
`mizes patients’ out-of-pocket costs.” Almost 91% of
`physicians surveyed somewhat or strongly agreed that it
`is “important” to try to minimize patients’ out-of-pock-
`et costs when prescribing, and only 5% disagreed
`(Figure). More than 80% of physicians surveyed some-
`what or strongly agreed that it is important to try to
`minimize the total costs of the medication when choos-
`ing between equally safe and effective medications, and
`only 8% disagreed. No significant differences were seen
`between survey respondents and nonrespondents.
`When asked if they agree that “it is important to dis-
`cuss out-of-pocket cost requirements with patients,”
`65% somewhat or strongly agreed and only 13% dis-
`
`VOL. 12, NO. 9
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`547
`
`000003
`
`

`

`POLICY
`
`Mean ± SD
`or Percent
`
`49.97 ± 9.8
`71.93%
`
`33.11%
`23.03%
`17.98%
`10.31%
`7.89%
`3.07%
`24.14 ± 23.7
`
`verse (data not shown; McNemar’s test statis-
`tic 41.3, P < .0001).
`When asked if they were more concerned
`with managing patients’ out-of-pocket costs
`than controlling the total cost of medication
`when choosing between equally safe and
`effective medications, almost 59% agreed
`(44% somewhat agreed and 15% strongly
`agreed), 25% neither agreed nor disagreed,
`and approximately 16% disagreed (11% some-
`what disagreed and 5% strongly disagreed
`[data not shown]).
`
`15.35%
`1.10%
`6.14%
`42.11%
`
`43.42%
`26.54%
`13.38%
`15.79%
`
`37.28%
`55.26%
`6.58%
`30.92%
`
`39.69%
`14.47%
`19.30%
`25.66%
`29.33 ± 21.5
`9.73 ± 16.7
`
`Factors Associated With Belief That
`Managing Out-of-pocket Costs Are More
`Important Than Managing Total
`Medication Costs
`Few physician characteristics were associ-
`ated with physicians’ perceptions about
`which costs are relevant at the time of pre-
`scribing (Table 3). Physician age and sex
`were unrelated to the outcome; of physician
`specialties, only medical subspecialists were
`less likely to report that out-of-pocket costs
`were more important than generalists (odds
`ratio 0.56, P = .046). Physicians who man-
`aged more Medicare patients were more like-
`ly to report that out-of-pocket costs are more
`important than total costs. As an example of
`the magnitude of the effect, every 10-percent-
`age-point increase in Medicare patient load
`was associated with approximately a 10%
`increase in the odds of reporting that man-
`agement of out-of-pocket costs was more
`important (P = .038). No clear or consistent
`relationship patterns were seen between the
`number of formularies prescribed from, number of pre-
`scriptions written, practice location, or practice size. No
`significant relationship was found between physicians
`who used certain information technology tools (eg, com-
`puter order entry or handheld devices) and physician
`preferences about relevant costs.
`
`Table 1. Physician Sample Characteristics
`
`Physician Characteristic (N)
`
`Age (450), y
`Male (456)
`Specialty (456)
`Generalist
`Medical subspecialist
`Surgeon
`Ob-Gyn
`Emergency room
`Psychiatrist
`Average number of prescriptions/day (453)
`Type of facility (456)
`Academic
`Veterans Administration
`County facility
`Hospital-based
`Practice size (456)
`Solo
`Medium
`Large
`Very large
`Practice setting (456)
`Urban
`Suburban
`Rural
`Computer order entry (456)
`Number of formularies prescribed from (456)
`0 to 1 formulary
`2 to 5 formularies
`6 or more formularies
`Don’t know
`Average percent of patients enrolled in Medicare (449)
`Average percent of uninsured patients (450)
`
`agreed (Table 2). Approximately 47% of physicians
`agreed that “it is important to discuss total costs of med-
`ications with patients,” and 23% disagreed. Almost 23%
`of physicians reported that they agreed that it is impor-
`tant to talk about out-of-pocket costs but did not agree
`that it is important to discuss total costs, while only 4%
`of physicians reported the converse (data not shown;
`McNemar’s test statistic 56.9, P < .0001). Yet when
`physicians were asked if they engage in these conversa-
`tions with patients, only 15% of physicians surveyed
`reported that they discuss out-of-pocket costs with
`patients most or all of the time and only 5% reported
`that they talk about the total costs of medication most
`or all of the time (Table 2). More than 12% of physi-
`cians reported that they discuss out-of-pocket costs
`most or all of the time and discuss total costs less fre-
`quently, while only 1% of physicians reported the con-
`
`DISCUSSION
`
`Our survey of California physicians offers insight
`into the cost information that physicians consider
`most important when prescribing. Although physicians
`attempt to minimize both out-of-pocket costs and total
`costs of medication when choosing between equally
`effective and safe options, most believe that manage-
`ment of patients’ out-of-pocket costs was more impor-
`
`548
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`SEPTEMBER 2006
`
`000004
`
`

`

`Physicians’ Perceptions of Prescription Drug Costs
`
`Figure. Percentage of Physicians Who Agreed That It Is Important to
`Manage Out-of-pocket and Total Costs of Medications
`
`tant. These findings are instructive in
`the setting of the widespread imple-
`mentation of tiered, incentive-based
`pharmacy benefit designs, which rely
`on varying out-of-pocket costs as a
`mechanism to influence physician
`prescribing and manage drug costs.
`When using incentive-based pharma-
`cy benefit designs to align the patient’s
`financial incentives with those of the
`insurer, efforts to educate physicians
`about patients’ out-of-pocket costs will
`likely be a more effective mechanism
`to control prescription drug costs than
`efforts to educate physicians about the
`total costs of medication.
`Physicians who managed more pa-
`tients enrolled in Medicare expressed
`greater concern about out-of-pocket cost management.
`At the time of the survey, approximately one third of
`Medicare enrollees had no prescription drug coverage.17
`Although the passage of Medicare Part D provided addi-
`tional coverage to many elderly people, the broad use
`of tiered formularies and lack of coverage in the “donut
`hole” will leave the elderly who enroll exposed to
`substantial out-of-pocket costs.18 In addition, medical
`subspecialists were less concerned with patients’ out-of-
`pocket costs as compared with total medication costs,
`suggesting that incentive-based formularies may be least
`effective in influencing prescribing among this physi-
`cian population. Yet the most notable findings in our
`multivariate analysis were that very few physician or
`practice variables significantly influenced beliefs about
`relevant costs for prescription drugs, and California
`physicians believed that out-of-pocket costs were more
`important to manage than total medication costs.
`
`Unfortunately, the financial incentives of patients
`and insurers are frequently discordant in tiered formu-
`laries. A recent article by Neumann and colleagues
`evaluating the cost effectiveness of preferred formulary
`medications in Florida Medicaid and a large managed
`care organization found that few drugs had any cost-util-
`ity data available, and of the drugs that did, most pre-
`ferred drugs were “more costly and less effective than
`alternatives.”19 These inconsistencies highlight the
`challenges that physicians face when attempting to
`prescribe, as efforts to minimize out-of-pocket costs
`may conflict with efforts to manage total medication
`costs. When faced with conflicts in which prescriptions
`that minimize a patient’s out-of-pocket cost will lead to
`increased total medication costs (or vice versa), our
`study suggests that physicians are more likely to choose
`medications that minimize patients’ out-of-pocket costs,
`contributing to the rising overall costs of prescription
`
`Table 2. Physicians’ Responses (%) to Questions Addressing Patients’ Out-of-pocket Costs and Total Costs
`of Medications
`
`Important
`to Discuss
`Out-of-pocket
`Costs With
`Patients
`
`Important
`to Discuss
`Total Costs
`With Patients
`
`Frequency of
`Discussing
`Out-of-pocket
`Costs With Patients
`
`Frequency of
`Discussing
`Total Costs
`With Patients
`
`Strongly disagree
`Somewhat disagree
`Neither agree nor disagree
`Somewhat agree
`Strongly agree
`
`5
`8
`20
`44
`21
`
`8
`15
`30
`33
`14
`
`18
`27
`39
`13
`2
`
`30
`35
`30
`5
`0
`
`Never
`Seldom
`Some of the time
`Most of the time
`Always
`
`Some columns do not add to 100% due to rounding to the nearest integer.
`
`VOL. 12, NO. 9
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`549
`
`Total Costs
`
`Out-of-pocket Costs
`
`Strongly
`Disagree
`
`Somewhat
`Disagree
`
`Somewhat
`Agree
`
`Strongly
`Agree
`
`Neither
`Agree
`Nor
`Disagree
`
`70
`
`60
`
`50
`
`40
`
`30
`
`20
`
`10
`
`0
`
`Physicians,%
`
`000005
`
`

`

`POLICY
`
`Table 3. Factors Influencing Whether Physicians Agree That It Is More Important to Manage Patients’ Out-of-
`Pocket Costs Than Total Medication Costs
`
`Odds Ratio
`
`95% Lower
`Confidence Interval
`
`95% Upper
`Confidence Interval
`
`Intercept
`Physician age
`Male physician
`Number of prescriptions
`Systems use when prescribing
`Use handheld device
`Use Internet
`Use handbooks
`Computer order entry when prescribing
`Academic physician (yes)
`Practice in county facility (yes)
`Hospital-based practice
`Practice size
`Medium-sized practice*
`Large-sized practice*
`Very large-sized practice*
`Practice location
`Urban practice†
`Suburban practice†
`Percentage of patients insured by
`Medicare
`Uninsured
`Number of formularies prescribed from
`2 to 5‡
`6+‡
`“Don’t know”‡
`Physician specialty
`Medical subspecialist§
`Surgeon§
`Ob-Gyn§
`Emergency room§
`Psychiatrist§
`
`0.95
`0.99
`1.34
`1.00
`
`0.99
`0.87
`0.91
`1.21
`1.28
`0.76
`1.26
`
`1.70
`0.98
`0.75
`
`1.80
`2.00
`
`1.01
`1.00
`
`2.96
`1.38
`1.26
`
`0.56
`0.64
`1.68
`0.44
`1.20
`
`0.18
`0.97
`0.82
`0.99
`
`0.83
`0.68
`0.74
`0.67
`0.71
`0.31
`0.81
`
`1.01
`0.48
`0.34
`
`0.77
`0.89
`
`1.01
`0.99
`
`1.44
`0.73
`0.70
`
`0.32
`0.34
`0.80
`0.19
`0.35
`
`5.12
`1.02
`2.21
`1.00
`
`1.18
`1.11
`1.12
`2.19
`2.31
`1.84
`1.96
`
`2.88
`2.00
`1.68
`
`4.20
`4.46
`
`1.02
`1.01
`
`6.09
`2.60
`2.25
`
`0.99
`1.20
`3.56
`1.05
`4.09
`
`P
`
`.95
`.57
`.25
`.26
`
`.92
`.26
`.35
`.53
`.42
`.54
`.30
`
`.048||
`.97
`.48
`
`.17
`.09
`
`.038||
`.92
`
`.003||
`.32
`.44
`
`.046||
`.16
`.17
`.07
`.78
`
`Results of multivariate logistic regression after multiple imputation of missing variables.
`Reference categories: *Solo/small practice, †rural practice, ‡0-1 formulary, §generalist/primary care physician.
`||P <.05.
`
`drug care. Nonetheless, in most cases, less expensive
`drugs are selected as preferred on patients’ formularies,
`and physicians who are aware of patients’ preferred for-
`mulary options may prescribe the medication that con-
`serves both patient and insurer resources.
`Our findings offer critical insight into physician pre-
`scribing behavior that may be instrumental to insurers
`who attempt to design efficient pharmacy benefit struc-
`tures to manage costs while maximizing quality.
`Previous studies have demonstrated that when branded
`drugs become available over-the-counter (OTC), costs
`for the insurer decrease.20,21 However, more recent stud-
`ies have found that insurers’ coverage decisions influ-
`
`ence costs; insurers who decreased coverage of OTC
`medications experience increased overall costs, where-
`as insurers who provide coverage for the OTC medica-
`tion experience reduced overall costs.22,23 These
`findings may be due, in part, to physicians’ willingness
`to prescribe more expensive medications for patients
`whose insurance coverage allows them to access the
`medication for lower out-of-pocket costs. These find-
`ings suggest that as medications become available
`OTC, coverage decisions should consider both the total
`costs of the medication and the out-of-pocket costs so
`that the incentives of the insurer, patient, and society
`are aligned.
`
`550
`
`THE AMERICAN JOURNAL OF MANAGED CARE
`
`SEPTEMBER 2006
`
`000006
`
`

`

`Physicians’ Perceptions of Prescription Drug Costs
`
`Our study was limited by the fact that we surveyed
`only physicians from California who are members of the
`CMA. Physicians from California may have different
`perceptions about drug costs than physicians elsewhere.
`Staff-model HMOs comprise substantial market share in
`California, and a large percentage of physicians in our
`sample prescribed from either 0 or 1 formulary. In
`addition, only approximately one third of the physi-
`cians in California are members of the CMA,24 and the
`possibility exists that CMA members differ in some
`way from nonmembers. Further study in a national
`sample would be informative. Another study limitation
`was that just under half of the physicians surveyed
`responded to the survey, which may have introduced
`selection bias. Yet our nonresponder survey suggested
`that the respondents were similar to the nonrespon-
`dents, and our results were robust to the inclusion of
`the nonresponders.
`These findings offer additional guidance concerning
`formulary design. Tiered formularies seem to address the
`costs that are most likely to influence physician and
`patient behavior, and present a logical approach to man-
`aging prescription drug costs. However, efforts are neces-
`sary to develop formularies in which decisions to select
`specific medications as preferred are more transparent
`and easily accessible. Although previous studies have
`found that doctors and patients discuss patients’ out-of-
`pocket costs infrequently,12,25 no previous study has dif-
`ferentiated communication about out-of-pocket and total
`costs for medications. We found that communication
`about total costs occurs even less frequently than discus-
`sions about out-of-pocket costs. If we aspire to create a
`collaborative healthcare system in which patients and
`physicians make decisions about care together, any con-
`flicts should be eliminated between therapies that mini-
`mize out-of-pocket costs while increasing total costs (or
`vice versa) so that doctors and patients can engage in
`discussions about costs that are more easily comprehen-
`sible. Such simplification may require formulary deci-
`sions to be based on cost-effectiveness evidence about
`medications. There has been a growing interest in link-
`ing out-of-pocket costs to the expected benefits of a
`drug.26 Such a benefit-based or value-based approach to
`formulary design may offer the most transparent mecha-
`nism for helping doctors and patients to communicate
`about costs and benefits of medications and lead to more
`cost-effective decision making.
`
`REFERENCES
`
`1. Korn LM, Reichert S, Simon T, Halm EA. Improving physicians’ knowledge of the
`costs of common medications and willingness to consider costs when prescribing.
`J Gen Intern Med. 2003;18:31-37.
`2. Frazier LM, Brown JT, Divine GW, et al. Can physician education lower the cost
`of prescription drugs? A prospective, controlled trial. Ann Intern Med.
`1991;115:116-121.
`3. Beilby JJ, Silagy CA. Trials of providing costing information to general practition-
`ers: a systematic review. Med J Aust. 1997;167:89-92.
`4. Hart J, Salman H, Bergman M, et al. Do drug costs affect physicians’ prescrip-
`tion decisions? J Intern Med. 1997;241:415-420.
`5. Kaiser Family Foundation and Health Research and Educational Trust. Employer
`Health Benefits: 2004 Summary of Findings. September 9, 2004. Available at:
`http://www.kff.org/insurance/7148/index.cfm. Accessed July 24, 2006.
`6. Kaiser Family Foundation. Prescription Drug Trends [fact sheet]. November
`2005. Available at: http://www.kff.org/insurance/upload/3057-04.pdf. Accessed May
`23, 2006.
`7. Glickman L, Bruce EA, Caro FG, Avorn J. Physicians’ knowledge of drug costs
`for the elderly. J Am Geriatr Soc. 1994;42:992-996.
`8. Reichert S, Simon T, Halm EA. Physicians’ attitudes about prescribing and
`knowledge of the costs of common medications. Arch Intern Med. 2000;160:
`2799-2803.
`9. Hoffman J, Barefield FA, Ramamurthy S. A survey of physician knowledge of
`drug costs. J Pain Symptom Manage. 1995;10:432-435.
`10. Shrank WH, Young HN, Ettner SL, Glassman P, Asch SM, Kravitz RL. Do the
`incentives in 3-tier pharmaceutical benefit plans operate as intended? Results from
`a physician leadership survey. Am J Manag Care. 2005;11:16-22.
`11. Alexander GC, Casalino LP, Meltzer DO. Physician strategies to reduce
`patients’ out-of-pocket prescription costs. Arch Intern Med. 2005;165:633-636.
`12. Shrank WH, Fox SA, Kirk A, et al. The effect of pharmacy benefit design on
`patient-physician communication about costs. J Gen Intern Med. 2006;21:334-339.
`13. Hassell K, Atella V, Schafheutle EI, Weiss MC, Noyce PR. Cost to the patient or
`cost to the healthcare system? Which one matters the most for GP prescribing deci-
`sions? A UK-Italy comparison. Eur J Public Health. 2003;13:18-23.
`14 Medical Board of California. 2004-2005 Annual Report. Available at:
`http://www.medbd.ca.gov/Pubs_Annualrept_2004-2005.pdf. Accessed May 9, 2006.
`15. Rubin DB, Schenker N. Multiple imputation in health-care databases: an
`overview and some applications. Stat Med. 1991;10:585-598.
`16. American Medical Association. Physician Characteristics and Distribution in
`the US, 2003-2004 Edition. Chicago, Ill: American Medical Association; 2003.
`17. Kaiser Family Foundation. How Do Patterns of Prescription Drug Coverage and
`Use Differ for White, African American, and Latino Medicare Beneficiaries Under
`65 and 65+ [chartpack]. July 30, 2003. Available at: http://www.kff.org/medicare/
`upload/How-Do-Patterns-of-Prescription-Drug-Coverage-and-Use-Differ-for-White-
`African-American-and-Latino-Medicare-Beneficiaries-Under