Patient-Physician Communication About Out-of-Pocket Costs FREE

About 20% of US health care costs are paid by patients out-of-pocket, usually at the point of health care utilization through deductibles, co-payments, and payment of uncovered services.¹ During the past 2 decades these costs have risen considerably in absolute terms.^2- 3

Despite this, little is known about patient-physician communication regarding out-of-pocket costs. Such information is important for several reasons. First, high patient expenses are a persistent concern among the general public in the United States.^4- 5 Physicians' fiduciary obligation to patients may be difficult to fulfill if they are not aware of patients' cost preferences in their decisions about treatment. Second, high out-of-pocket costs are associated with medication nonadherence and may impede physicians' efforts to provide high-quality care.^6- 9 Third, cost sharing has been proposed as a means to reduce health care utilization, but the RAND Health Insurance Experiment¹⁰ and other studies^11- 12 suggest that higher out-of-pocket costs may lead to decreased demand for both nonessential and essential health care. One explanation for this nonselective effect is that patients and physicians do not discuss out-of-pocket costs and thus patient demand for health care services is modified without physician input.

We undertook a paired study of general internists and their patients to compare their beliefs and practices about discussions of out-of-pocket costs. By using a paired study design, we were able to compare directly the reports of each group. Although we examined out-of-pocket costs for all types of health care, we included several items focusing specifically on out-of-pocket prescription costs. These costs account for almost half of overall expenditures for prescription medicines, affect a large sector of the population, and are the focus of ongoing debates regarding a Medicare prescription drug benefit.^1,13

Our recruitment strategy is illustrated in Figure 1. We invited the participation of the 3 largest academic general internal medicine practices in the Chicago metropolitan area. We also invited a convenience sample of 9 community-based affiliates of these practices. With the assistance of the Medical Group Management Association (MGMA), we contacted each of 21 additional physician groups that were members of MGMA and located within the metropolitan area.

Potential participants were told the general goal of the study, "to learn about factors doctors and patients consider as they make medical decisions," but they were not prospectively told the specific study aims. Trained interviewers attended a morning or afternoon session of each participating physician's practice and invited consecutive patients to complete a short orally administered questionnaire after their visit. Interviewers attempted to recruit at least 3 patients per physician. In some cases, based on interviewer availability, we consecutively enrolled additional patients for a given physician. At the end of the half-day, each participating physician completed a written questionnaire about at most 3 of his or her enrolled patients. Therefore, our total patient sample includes some patients (n = 110) for which paired physician data are not available. We included all patients 18 years or older and excluded patients who did not speak English or who were cognitively impaired. To measure patient participation rates during the latter two thirds of the study period, we assessed the number of ineligible patients due to language or cognitive barriers, who declined to participate, or who agreed to participate but left prior to being surveyed (Figure 1) . We used a script to obtain oral informed consent from patients and written informed consent from physicians. The institutional review boards at each participating academic site approved the study protocol and strict measures were undertaken to preserve participant confidentiality.

Based on pilot interviews with physicians, we developed patient and physician questionnaires (available on request) to examine their beliefs and practices regarding discussions of patient out-of-pocket costs. The patient survey examined whether they wanted to discuss out-of-pocket costs with their physician, the burden of their out-of-pocket costs, and their experiences of discussing these costs with the physician they were seeing. The physician survey included some items that inquired about the study patients burden from out-of-pocket costs, whether a discussion of these costs had ever taken place, and other items that asked more generally about whether they believe patients in their practice want to discuss out-of-pocket costs. We defined burden as the presence of at least 1 of 3 measures: problems paying medical bills, subjective burden from out-of-pocket costs, and cost-related medication nonadherence within the previous 12 months. In addition, the patient survey assessed patients' age, race, sex, income, insurance status, self-reported health, and number of chronic conditions. The physician survey included items examining length of time in practice, sex, knowledge of specific study aims prior to participation, and the use of patient medical records to complete the questionnaire. Both surveys included approximately 50 items and required about 10 minutes to complete. Surveys were extensively piloted and revised before administration from March through November 2002.

We defined our primary outcome variable as patients' report of having discussed out-of-pocket costs with their physician. We reasoned that this report was less likely than physicians' report to be subject to recall and socially desirable response bias.¹⁴ We defined our secondary outcome variable as physicians' recognition of patients burdened by their out-of-pocket costs. We then used the χ² test and logistic regression to examine the bivariate and multivariate relationships between each patient, physician, and practice characteristic and these outcome variables. In the final multivariate model our predictor variables included basic patient characteristics, variables approaching significance on bivariate analysis (P<.20), and those in which we had a substantive interest. Finally, because our primary and secondary outcomes were common (>10%), we derived prevalence (risk) ratios (PRs) from the odds ratios to better estimate the true associations between our predictor variables and the outcome variables of interest.¹⁵ For continuous predictor variables, we approximated the incidence of the outcome among the unexposed group to be that of the lowest quartile of the sample. To account for the clustering of patients within physicians and physicians within practices we repeated analyses using hierarchical logistic modeling and standard logistic regression with robust SEs. These results yielded virtually identical findings and are not presented herein. Analyses were performed using JMP version 4.04 (Cary, NC), Stata version 7 (College Station, Tex), and HLM2 (Lincolnwood, Ill).

A total of 484 patients and 133 physicians participated (Figure 1). The patient participation rate during the latter two thirds of the study was approximately 80%. Patients enrolled during the first third of the study were almost all from academic practices. Due to differences in the case mix of academic practices sampled, patients enrolled from these practices during the first third of the study were more likely (P<.05) to be older (mean age 55 vs 51 years), nonwhite (73% vs 50%), without a college education (35% vs 56%), and with more comorbidities (mean 2.19 vs 1.74); however, there were no differences in the frequencies of discussions or burden of out-of-pocket costs based on the period of study participation (P<.05).

Of the 484 patients, the mean age was 54 years, 64% were women, 42% were black, 68% had private or employer-purchased insurance, and 90% had at least partial prescription drug coverage (Table 1). Only 1 patient was uninsured. Of the 133 participating physicians, the median length in practice was 8 years (range, 1-55 years), 55% were men, and 55% worked within the 3 academic practices.

The proportion of patients reporting a burden from their out-of-pocket costs varied from 14% reporting a problem paying medical bills to 25% of respondents reporting a subjective burden from their out-of-pocket costs. Seventy-seven (16%) of the 484 patients reported cost-related medication nonadherence during the previous year.

Sixty-three percent of patients reported a preference to talk with their physician about their out-of-pocket costs before receiving a test or treatment. Similarly, 79% of physicians believed that patients in general want to discuss their out-of-pocket health care costs prior to receiving a test or treatment, and 90% of physicians said that they should consider patients' out-of-pocket costs as they make clinical decisions.

Four hundred eight patients (85%) reported they had never discussed their out-of-pocket costs with their physician. This group of 408 patients included 56 of the 77 patients (73%) who reported cost-related medication nonadherence during the previous 12 months. Of the 374 patients for whom paired physician data were available, physicians reported that they had never discussed out-of-pocket costs with 65% of these patients.

On bivariate analysis, several patient, physician, and practice characteristics were associated with patients' reports of having discussed their out-of-pocket costs with their physician (Table 2 and Table 3). On multivariate analysis, patients who were seen in a community practice and those who felt burdened by their out-of-pocket costs were each significantly more likely to report having discussed their out-of-pocket costs with their physician than were their counterparts (Table 2 and Table 3). For example, the adjusted PR of reporting a discussion of out-of-pocket costs was 5.19 (95% confidence interval [CI], 1.86-8.93) among patients seen in a community practice as compared with those seen in an academic practice. Despite these multivariate associations, 75% of patients seen in a community practice or burdened by their out-of-pocket costs did not report discussing these costs with their physician. Although practice type and practice size were highly correlated, on multivariate analysis there was a trend toward greater frequency of discussions reported by patients seen in smaller practices. These results did not change significantly when we limited our analysis to the 366 patients (76%) who reported that the physician they saw was both their primary care physician and someone they had seen previously.

Only 16% of patients believed that their physician was aware of the magnitude of their out-of-pocket costs. Twenty-one percent of physicians reported that in general they know how much their patients are spending out-of-pocket. Physicians were more likely to recognize that the study patients felt burdened by out-of-pocket costs if they had had a prior discussion of these costs with the patient. For example, among the 118 patients burdened by their out-of-pocket costs for which paired physician data were available, physicians correctly recognized this burden among 24 (80%) of 30 patients who reported a prior discussion of their out-of-pocket costs compared with 45 (51%) of 88 patients who reported that no such discussion had taken place. The unadjusted PR of a physician recognizing a patient's burden was modestly greater among patients reporting that a prior discussion of out-of-pocket costs had taken place (PR, 1.57; 95% CI, 1.20-1.81). After adjustment for potentially confounding patient, physician, and practice characteristics that were associated with burden recognition, a similar but statistically nonsignificant association persisted between patients' reports of prior discussions of out-of-pocket costs and physicians' recognition of patients' burden (PR, 1.44; 95% CI, 0.87-1.80).

To the best of our knowledge this is the first study focusing on the extent to which physicians and patients discuss patients' out-of-pocket costs. We found that both patients and physicians believed that discussions of out-of-pocket costs were important, yet these discussions were uncommon. This finding is especially noteworthy given the substantial minority of patients reporting a burden from their out-of-pocket costs and the finding that physicians appeared more likely to recognize this burden when a prior discussion of out-of-pocket costs had occurred.

Our findings have important implications for patients, clinicians, and policymakers. Patients should be encouraged to raise concerns about their out-of-pocket costs with their physicians. A recent study by the American Association of Retired Persons^16- 17 found that the general public had poor knowledge of generic medicines, a finding that led to a widespread media campaign to encourage seniors to discuss with their physicians the cost-savings associated with generic medicines. Our results support these efforts and suggest that although physicians may not frequently initiate discussions of out-of-pocket costs, they generally believe the topic is an important one. The difference between what physicians reported they should do and what they actually do is striking. This discrepancy is important because greater discussion of unvoiced concerns about out-of-pocket costs among patients who have them may facilitate their participation in the patient-physician relationship and in clinical decision making.¹⁸

Although nearly all of our patients had at least partial prescription drug coverage, our study provides similar estimates as previous work demonstrating that a substantial minority of patients forgo prescribed medications due to cost.^4,6 Our findings may understate the burden that out-of-pocket prescription costs pose on the public, especially among uninsured and elderly patients. Clinicians may wish to be alert for the risk factors associated with burden from out-of-pocket costs for prescription medicines (eg, advanced age, multiple comorbidities, low income), to be attentive to patient clues that suggest a burden from these costs,¹⁹ and to invite patients to raise their concerns. Physicians may also work to minimize these expenses by using generic and lower-cost brand name prescriptions. Enhanced communication between patients burdened by out-of-pocket prescription costs and their clinicians may heighten physicians' awareness of this burden and thereby help physicians to ensure that patients use the most cost-effective medicines possible.^20- 21

Finally, our findings may provide insights for policymakers as to how greater cost-sharing may fail to discriminate between appropriate and inappropriate health care utilization as well as pose a particular burden on vulnerable populations with the greatest health care expenditures.^10- 12,22

Our study had several limitations. First, we examined a group of general internists and their patients in 1 metropolitan area of the country. Our patient population was more likely to be black and more likely to be insured than the general population. Further work is needed to assess the generalizability of our results. Second, our findings are based on self-report. Questions about actual behavior are subject to several biases including recall bias and socially desirable response bias. Third, because of the modest number of subjects who reported discussions of out-of-pocket costs, our data do not allow us to fully distinguish the effect of practice type (academic vs community practice) from the effect of practice size or other practice characteristics (eg, payer mix, patient turnover, proximity of billing to clinical services) that may be associated with the frequency of discussions about out-of-pocket costs. However, our data suggest that both practice type as well as practice size may be independently associated with discussions of out-of-pocket costs. Fourth, because of the modest number of subjects who reported discussions of out-of-pocket costs, our power to discern patient, physician, and practice characteristics associated with these discussions was limited. Fifth, we do not have information on participation rates for patients enrolled during the first third of the study. Finally, our cross-sectional study design did not examine the content of discussions about out-of-pocket costs or assess how often physicians explicitly consider patients' out-of-pocket costs without discussing these costs.

Among our respondents, both patients and physicians believed that discussions of out-of-pocket costs were important, yet these discussions seldom occurred. Possible barriers to discussions that were reported by patients and physicians included discomfort discussing financial issues, insufficient time, and a belief that there were not viable solutions to patients' concerns. Although most patients reported a desire to discuss their out-of-pocket costs with their physician, the fact that patients' awareness of their out-of-pocket costs may not be prompted at the point-of-service may pose an additional hurdle to patient-physician communication about these costs. Further research is needed to better understand and address these and other barriers that may prevent patients and physicians from discussing out-of-pocket costs more often.

Research is also needed to assess the impact of increased patient-physician communication about out-of-pocket costs on patient satisfaction, utilization of care, and outcomes. Continued increases in out-of-pocket costs for many US residents make these issues particularly salient,^5,23 and although little research has examined communication about out-of-pocket costs, effective patient-physician communication in other settings has been associated with improved patient satisfaction and health outcomes.^24- 25