Use of an open-ended question to supplement a patient satisfaction questionnaire in a medical residents' clinic

Objectives: To determine (1) the proportion of responses to an open-ended q uestion related to patient satisfaction that could be categorized into 1 or more of 9 previously developed domains of outpatient care and (2) whether any other important aspects of care could be identified by adding the open- ended question to a satisfaction questionnaire. Study Design: A 3-month observational study was done at the internal medici ne clinic of an urban teaching hospital. Patients and Methods: As part of a patient satisfaction study, 511 visitors were asked after their visit, "What are the 1 or 2 things that are most im portant to you when you see a doctor?" The responses were categorized indep endently by 2 raters into 1 or more of the 9 domains. When these 2 raters d isagreed, the responses were read to a third rater. When either all 3 rater s disagreed, or at least 1 rater thought a new domain was mentioned, those responses were categorized by consensus. Interobserver reliability between raters 1 and 2 was calculated by using Cohen's kappa statistic. Results: The 355 responses were categorized as follows: 303 (85.4%) identif ied one or more domains that were part of the previously developed taxonomy , 9 (2.5%) identified a new domain, 11 (3.1%) identified both old and new d omains, and 32 (9.0%) could not be categorized. Cohen's kappa was 0.57 (P<. 001). Cultural sensitivity and physician honesty were the additional domain s identified, by 1.1% and 4.5% of respondents, respectively. Conclusions: The previously developed taxonomy of domains can be used in th is setting to categorize the targe majority of open-ended responses. Such;r esponses can identify important aspects of care that were either previously unidentified or were already identified but given low ratings. This inform ation then can help improve quality of care.