Patients' preferences for technical versus interpersonal quality when selecting a primary care physician

Health Services Research, August, 2005 by Constance H. Fung, Marc N. Elliott, Ron D. Hays, Katherine L. Kahn, David E. Kanouse, Elizabeth A. McGlynn, Mark D. Spranca, Paul G. Shekelle

We used [chi square] tests of independence to determine whether computer use was associated with the type of physician selected in the validity check and tradeoff pairs. We calculated the correlation coefficients between trust in different sources of information and our main outcome variable, the number of times the participant selected the physician who excelled in technical quality for the tradeoff pairs. We performed Wilcoxon's signed-rank tests to compare patients' trust in the two sources of information.

Finally, we estimated an ordered logit model to determine whether patient characteristics such as age, gender, race/ethnicity, chronic disease, number of visits to the primary care physician in the past year, caregiver status, subject understanding of the report cards (self-rated), computer use, income, and education were significant predictors of our main outcome variable, the number of times the participant selected the physician who excelled in technical quality for the tradeoff pairs (SAS, version 8).

RESULTS

Table 2 shows the distribution of participants according to gender, age, race/ ethnicity, educational level, health insurance, marital status, total combined family income in the past 12 months, computer use, employment status, caregiver status, number of PCP visits in past year, and chronic illness.

The two internal validity check pairs evaluated each participant's level of attention to the task. Ninety percent of participants chose the dominant provider both times, a figure consistent with the 90 percent of respondents who answered the dominant pair questions correctly in the conjoint analysis component of Phillips, Johnson, and Maddala's (2002) study on valuation. Infrequent computer users (computers less than once per month), who comprised 15 percent of the sample, were more likely to choose at least one nondominant physician in the validity check pairs (p< .0001). (2)

Figures 2A and B summarize the results of the tradeoffpairs. Two-thirds of our sample (95 percent CI: 62, 72 percent) chose the physician who excelled in technical care three or more times out of five, demonstrating an overall preference for technical quality of care. However, 33 percent (95 percent CI: 28, 38 percent) of the sample, chose the physician who excelled in interpersonal quality at least three times of five, suggesting that interpersonal quality was important for a substantial number of people in our sample. Figure 2A, which provides details about the choices participants made for each pair of physicians, also shows that the only pairing for which a majority of subjects did not choose the physician who excelled in technical quality was the one in which a physician average in both dimensions was paired with a physician with high technical and low interpersonal ratings (95 percent CI: 40, 51 percent).

Figure 3, which displays mean values for the responses to the questions in the paper questionnaire that assessed attitudes towards different dimensions of technical and interpersonal quality, indicates that dimensions of both technical and interpersonal quality are important to subjects. For example, participants rated communication as at least as important as preventative care.