Author Affiliations: Center for Studying Health System Change, Washington, DC (Drs Pham and O’Malley); MacLean Center for Clinical Medical Ethics and the Division of General Internal Medicine, University of Chicago Hospitals, Chicago, Ill (Dr Alexander).
Patients face growing cost-sharing through higher deductibles and other out-of-pocket (OP) expenses, with uncertain effects on clinical decision making.
We analyzed data on 6628 respondents to the nationally representative 2004-2005 Community Tracking Study Physician Survey to examine how frequently physicians report considering their insured patients' OP expenses when prescribing drugs, selecting diagnostic tests, and choosing inpatient vs outpatient care settings. Responses were dichotomized as always/usually vs sometimes/rarely/never. In separate multivariate logistic regressions, we examined associations between physicians' reported frequency of considering OP costs for each type of decision and characteristics of individual physicians and their practices.
Seventy-eight percent of physicians reported routinely considering OP costs when prescribing drugs, while 51.2% reported doing so when selecting care settings, and 40.2% when selecting diagnostic tests. In adjusted analyses, primary care physicians were more likely than medical specialists to consider patients' OP costs in choosing prescription drugs (85.3% vs 74.5%) (P<.001), care settings (53.9% vs 43.1%) (P<.001), and diagnostic tests (46.3% vs 29.9%) (P<.001). Physicians working in large groups or health maintenance organizations were more likely to consider OP costs in prescribing generic drugs (P<.001 for comparisons with solo and 2-person practices), but those in solo or 2-person practices were more likely to do so in choosing tests and care settings (P<.05 for all comparisons with other practice types). Physicians providing at least 10 hours of charity care a month were more likely than those not providing any to consider OP costs in both diagnostic testing (40.7% vs 35.8%) (P<.001) and care setting decisions (51.4% vs 47.6%) (P<.005).
Cost-sharing arrangements targeting patients are likely to have limited effects in safely reducing health care spending because physicians do not routinely consider patients' OP costs when making decisions regarding more expensive medical services.
Patients face growing cost-sharing pressures through new insurance benefit structures that limit coverage for specific services.1- 3 Cost sharing is meant to offset payer expenditures by shifting responsibility for costs of care to patients and creating incentives for them to reduce health care demand. However, the complexity of clinical decision making and information asymmetries between patients and physicians often prevent patients from safely making decisions about health care use on their own.4 Rather, as physicians' decisions affect how 90% of every health care dollar is spent,5 whether increased cost sharing can effectively control health care spending depends on whether patients and physicians can together consider costs during clinical decision making.6,7
The RAND Health Insurance Experiment8 and subsequent studies9,10 have raised concerns that patients facing higher cost sharing tend to decrease use of both necessary and unnecessary services, potentially leading to worse health outcomes. In particular, these studies suggest that patients confronted with higher out-of-pocket (OP) costs more often choose to forego physician visits or care that has been recommended by a physician (eg, filling a prescription or obtaining an ordered mammogram).
Less evidence is available on how cost-sharing arrangements affect clinical decision making during patient-physician interactions. Patients and physicians often do not discuss issues around OP costs, which may be owing to embarrassment or fear that the quality of care may be compromised.11- 13 However, studies focused on patient-physician communication have examined cost sharing generally rather than comparing its effects on specific types of health care services (eg, diagnostic testing) and have explored only a limited number of physician and practice factors in relation to whether OP costs affect clinical decision making.12,14
Physicians serve as patients' agents, and increasingly are called on to deliver patient-centered care.15 Thus, physician sensitivity to patients' OP costs might be hypothesized to vary considerably based on several factors, including the amount of clinical discretion allowed within accepted practice standards for different types of decisions; the magnitude of economic burdens their patients would face; and how invested physicians are in patient satisfaction, which may vary with physician and practice characteristics.16,17 On the other hand, there are reasons to hypothesize little variation in physicians' consideration of OP costs, if most physicians lack detailed data on their patients' insurance benefits, and both patients and physicians generally avoid the awkwardness of discussing OP costs.
Although analyses of administrative claims may demonstrate the impact of greater cost sharing on health care utilization, such analyses have generally focused on pharmaceuticals alone18,19 and have not captured the mechanisms by which such changes take place, such as the degree to which physicians actively consider patients' OP costs. We analyzed data on how frequently physicians reported considering their patients' OP costs from copayments and deductibles when prescribing drugs, selecting diagnostic tests, and choosing care settings. We also assessed the physician and practice characteristics associated with this practice.
The Community Tracking Study Physician Survey20 is a periodic, nationally representative telephone survey of nonfederal US physicians conducted 4 times since 1996. The sample is clustered in 60 randomly selected metropolitan statistical areas and includes physicians who report spending at least 20 hours in direct patient care per week. Primary care physicians (PCPs) (including those engaged in the practice of general internal medicine, family practice, general practice, general pediatrics, geriatrics, and pediatrics/internal medicine) were oversampled, and specialists such as radiologists, anesthesiologists, and pathologists were excluded. Our analysis focuses on data from the 2004-2005 survey,20 which included 6628 respondents with a response rate of 53%. Other details about the survey and its design have been previously published.20
Our outcome measures came from 3 questions of the form “How often do you consider an insured patient's out-of-pocket costs for copayments and deductibles” in (1) “prescribing a generic over a brand-name drug if a generic option is available?” (2) “deciding the types of tests to recommend if there is uncertainty about a diagnosis?” and (3) “choosing between inpatient and outpatient care settings, when there is a choice?” Physicians responded “always,” “usually,” “sometimes,” “rarely,” or “never.” We dichotomized responses as always/usually vs all other responses.
We selected these types of decisions because physicians face different levels of discretion for each. That is, generic drugs are nearly always equivalent to brand-name drugs in efficacy, while different diagnostic tests or care settings may entail considerable trade-offs in safety, convenience, or other factors. For example, a physician trying to determine if a patient's chest pain is cardiac in nature may order a simple treadmill stress test; a stress test with an echocardiogram to visualize pump function; or a higher-risk but more definitive cardiac catheterization. Similarly, a patient with community-acquired pneumonia may be appropriately treated as an outpatient or admitted to a hospital, depending on the presence of comorbidities, severity of the infection, and presence of family or other support in the home.
We limited our inquiry to insured patients to isolate the effects of cost-sharing arrangements as distinct from those of potentially much larger OP expenses that uninsured patients confront. The 3 questions underwent cognitive testing with a focus group of physicians prior to survey fielding and were rotated during survey administration.
Our main independent variables of interest were self-reported physician and practice characteristics. Physicians' characteristics included years in practice; specialty (PCPs, medical specialists, and surgeons); medical school site (United States, Canada, or Puerto Rico, [hereinafter called “United States”] vs elsewhere); board certification; sex; race or ethnicity (non-Hispanic white, Hispanic, and all other); whether their compensation was dependent on measures of productivity or patient satisfaction (yes or no for each); what they perceived the overall effect of their financial incentives to be (expand services at all vs reduce services at all or no effect); and whether they agreed that they had adequate time to spend with patients during office visits (agree strongly or somewhat vs all other responses).
For the question on drug prescribing, we also included the reported proportion of the physicians' patients covered by drug formularies (<25%, 25%-50%, 51%-75%, and 76% or more). For the question on selecting diagnostic tests, we included physicians' reported ability to access referrals for high-quality diagnostic imaging and specialty care. And for the question on selecting care settings, we considered physicians' reported ability to access elective hospital admissions for their patients (all categorized as yes or no). Practice characteristics included type (solo or 2-person group; groups ≥3; staff or group model health maintenance organization [HMO]; hospital and/or university; or other); percentage of revenues derived from Medicare and managed care (each categorized by terciles); the number of managed care contracts; and whether the practice had information technology to access treatment guidelines. For the question on drug prescribing, we also assessed whether physicians had access to information technology in their practice to write electronic prescriptions.
We examined several other variables as proxy measures of the socioeconomic status of physicians' patient panels: the number of hours physicians reported providing charity care in the previous month (none, 1-4, 5-9, and ≥10 hours); the proportion of practice revenue derived from Medicaid (terciles); the median household income in the physician's practice ZIP code (quintiles); and the proportion of adults 25 years or older in that county who received 12 or more years of education (terciles).
We report herein the adjusted proportions of physicians who reported having considered patients' OP costs by their individual and practice characteristics. Adjustments were made using separate multivariate logistic regressions for each of the 3 outcome measures and included physician and practice characteristics that were significant in bivariate analyses. We also adjusted for the reported proportion of each physician's patients with a chronic condition as a measure of the illness burden among his or her patients.
All analyses were conducted using SUDAAN analytic software (release 7.0; Research Triangle Institute, Research Triangle Park, NC) and weighted with Community Tracking Study20 physician weights for national estimation to take into account the survey's complex sampling strategy and known differences between respondents and nonrespondents.
Most of the physicians (78.2%) reported regularly (always or usually) taking patients' OP costs into account when prescribing generic over brand-name drugs. The proportion of physicians routinely considering OP costs in prescribing drugs varied by both individual physician and practice characteristics, although these differences were small. In adjusted analyses, PCPs were more likely than medical specialists (85.3% vs 74.5% (P<.001) to regularly consider OP costs (Table 1).
Similarly, physicians in solo or 2-person practices were less likely to always or usually take OP costs into account than physicians in HMOs (79.8% vs 89.8%) (P<.001) (Table 2). Of note, neither having access to electronic prescribing nor the proportion of their patients covered by formularies was associated with physicians' reported frequency of considering OP costs in choosing generic drugs.
In contrast, more than half (51.2%) of physicians reported considering OP costs in choosing inpatient vs outpatient care settings, and two fifths (40.2%) did so in selecting diagnostic tests. As in decisions to prescribe generic drugs, medical specialists were significantly less likely than PCPs to routinely consider OP costs when selecting diagnostic tests (29.9% vs 46.3%) (P<.001) and care settings (43.1% vs 53.9%) (P<.001) (Table 1).
However, several physician and practice characteristics associated with considering OP costs for testing and care settings differed from those associated with considering costs for drug prescribing. Physicians in solo or 2-person practices were more likely to consider OP costs in choosing diagnostic tests and care settings than physicians in other types of practices. This association was independent of the socioeconomic mix in the practice area and the illness burden among a physician's patients (P<.05 for all comparisons). There was no relation between practice revenue sources and considerations of OP costs for decisions on diagnostic tests or care settings (Table 2).
Physicians' reported ability to access specific types of services was also associated with their considering OP costs. Physicians who were confident they could access high-quality diagnostic imaging, but not specialist referrals, were less likely than those who perceived worse access to consider patients' OP costs in selecting diagnostic tests (37.6% vs 42.1%) (P = .06). And physicians who were confident that they could access elective hospital admissions for their patients were less likely than those who perceived worse access to consider OP costs when selecting inpatient vs outpatient care settings (48.7% vs 54.8%) (P = .01) (Table 2).
Physicians treating patients of lower socioeconomic status tended to be more likely to consider OP costs. Physicians in the highest tercile of practice revenue derived from Medicaid were more likely than those in the lowest tercile (83.4% vs 77.4%) (P<.001) to regularly consider OP costs for drug prescribing (Table 1), while those who provided at least 10 hours of charity care a month were more likely than those providing no charity care to consider OP costs in both diagnostic testing (40.7% vs 35.8%) (P<.001) and care setting decisions (51.4% vs 47.6%) (P = .005) (Table 1).
We found no consistent association between physicians' reporting adequate time to spend with patients during office visits or their compensation structures and whether they routinely considered patients' OP costs in any of the 3 clinical situations.
As insurers and employers seek to control health care spending in the post–managed care era, patient cost sharing is likely to remain a prominent tool for influencing health care utilization. How well this works depends in part on physicians' sensitivity to their patients' OP costs, because physicians are responsible for decisions that affect how 90% of each health care dollar is spent.5 We found that while most physicians reported routinely considering patients' OP costs in clinically straightforward prescribing decisions, only half or fewer do so in more complex situations that allow greater clinical discretion.
Our survey was conducted in 2004-2005, several years after employers and insurers embarked on the current trend of “benefits buy-down” through expanding consumer copayments and deductibles.1,3 We found that, despite the attendant increases in cost-sharing burdens over that time period, most physicians did not take them into account when making common care recommendations surrounding diagnostic testing and choosing care settings.
To our knowledge, prior studies have not compared physicians' sensitivity to patients' OP costs for different services. The variation across different types of clinical decisions that we document is not unexpected. The wider clinical latitude involved in more complex decisions means that physicians may prioritize other considerations, such as accuracy of a test or ease of scheduling, above patients' OP costs as they approach these types of choices. And physicians and patients are less likely to have ready access to detailed cost data for diagnostic tests or care settings than for prescription drugs.
Nevertheless, that physicians report considering patients' OP costs less frequently in making decisions about more expensive services suggests that cost-sharing arrangements dependent on patients and physicians weighing such costs can have only limited effect on use of such services and, by extension, less impact on overall health care spending than payers might hope for.
We also found that specialists were modestly more resistant to OP costs than primary care physicians. This finding was consistent across all 3 types of care decisions and may be due to a number of factors. First, PCPs may be more sensitive than specialists to patients' economic burdens, perhaps because of more continuous care relationships and knowledge of patients' social circumstances that would facilitate discussing sensitive topics.21,22 Second, some specialist physicians are more likely to perform many types of diagnostic procedures including through self-referrals, which may create perverse disincentives for them to consider patients' cost burdens in selecting individual tests, although we did not find specific evidence that physicians' financial incentives affect consideration of patients' OP costs.23,24 Third, it is possible that specialists may perceive less discretion in decision making if they responded to the survey thinking in particular of patients within the purview of their individual specialties who may have more severe illness than similar patients treated by PCPs.25 Regardless of the mechanisms, these specialty differences compound concerns that cost sharing will have limited effects on the use of more expensive medical services.
Outside of comparisons of academic medical centers vs community-based practices, few studies have examined practice type differences in decisions regarding specific services.12 We found that physicians in closed-model HMOs are most likely to consider OP costs in drug prescribing while those in solo or 2-person practices are most likely to do so in recommending diagnostic tests and care settings. One potential explanation may be that physicians in HMOs are more likely than those in other practice settings to have ready access to information on copayments for formulary vs nonformulary drugs to which they can refer during patient encounters.26 In contrast, similar decision aids for choosing diagnostic tests or care settings are unlikely to be as available or straightforward to use, given the greater clinical latitude involved in these types of decisions. In this context, the strength of individual patient-physician relationships that support communication about sensitive economic issues may become more important determinants of whether physicians consider cost burdens. Although in prior research both patients and physicians report that inadequate time is a barrier preventing greater communication about OP costs,12 in the present study we did not find an association between physician's reports of having adequate time to spend with patients and physicians' routinely considering OP costs.
Our finding that physicians who reported worse access to high-quality imaging services and elective hospital admissions were more likely to consider OP costs in selecting tests and care settings is intuitive. It suggests that cost-sharing arrangements will have their greatest effects where the supply of the relevant services is already constricted, providing further motivation for physicians to weigh their costs against their benefits and the time patients would spend waiting to access them.
The tendency toward greater consideration of OP costs among physicians treating patients with lower socioeconomic status is also expected. Lower-income insured Americans are more likely than the wealthier insured to have high OP health care spending burdens relative to their family incomes.3 Physicians may either intuit this or make assumptions about the relationship between patients' cost burdens and their socioeconomic status. But as more patients face cost-sharing arrangements, such preconceived notions of which patients are most vulnerable could grow less reliable.
Our results should be viewed in light of several limitations. We examined only 3 types of clinical decisions. Physicians may consider OP costs more or less frequently in other clinical situations, especially those involving even higher-cost or higher-risk services such as surgical procedures. Social desirability bias in a telephone survey may also result in underestimation or overestimation of physicians' likelihood of considering patients' cost burdens, although we carefully worded questions on diagnostic testing and care settings to avoid implying a normative response. (We expect physicians to associate little stigma with prescribing a generic over a brand-name drug regardless of their motivation.)
It is difficult to judge the accuracy of self-reported measures of physicians' decision-making processes, although physicians accurately interpreted the questions during cognitive testing, and we observed a reasonable distribution of responses. Similarly, because survey items asked about physicians' overall experience, we cannot address the influence of OP cost considerations on decision making in individual care encounters. Further research may clarify whether physicians may be more sensitive to cost burdens experienced by particular patient subgroups such as those exposed to “doughnut holes” and varying copayments vs deductibles.19,27 Finally, our results do not address the effects of cost sharing on care decisions that patients may make outside of their interactions with physicians, for example, by foregoing recommended therapy.9,10,28
More aggressive interventions such as accessible, detailed, and user-friendly information on costs of specific services might broadly change physicians' clinical decision making,29 but only if they result in a significant improvement in how often and how well patients and physicians communicate with one another about cost burdens. To evaluate the impact of such efforts, policymakers will need to track physicians' sensitivity to patients' OP costs over time, particularly for costly, discretionary services. How such clinical decisions are made at the point of care is critical to the viability of cost-sharing arrangements as a cost-control tool and, as importantly, to the overall health and economic burden they pose to patients.
Correspondence: Hoangmai H. Pham, MD, MPH, Center for Studying Health System Change, 600 Maryland Ave, SW, Suite 550, Washington, DC 20024 (firstname.lastname@example.org).
Accepted for Publication: December 12, 2006.
Author Contributions: Dr Pham had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pham and Alexander. Acquisition of data: Pham. Analysis and interpretation of data: Pham, Alexander, and O’Malley. Drafting of the manuscript: Pham and Alexander. Critical revision of the manuscript for important intellectual content: Pham, Alexander, and O’Malley. Statistical analysis: Pham and O’Malley. Obtained funding: Pham. Study supervision: Pham.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Robert Wood Johnson Foundation.
Role of the Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Internal Medicine editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Register and get free email Table of Contents alerts, saved searches, PowerPoint downloads, CME quizzes, and more
Subscribe for full-text access to content from 1998 forward and a host of useful features
Activate your current subscription (AMA members and current subscribers)
Purchase Online Access to this article for 24 hours
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 19
Customize your page view by dragging & repositioning the boxes below.
and access these and other features:
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.