Abstract
Background/Aims The literature indicates Asians rate their health care experience lower than do non-Hispanic whites (NHWs). Cultural norms in rating similar experience may contribute to this, but before any adjustment, one should make sure differing ratings do not reflect true quality differences. We examined whether adjustment for the race/ethnicity (R/E) of survey respondents is warranted in assessing patient satisfaction across providers.
Methods We used 2011–12 patient satisfaction data from a large ambulatory care organization (116,516 surveys for 1,022 providers). Surveys were mailed to patients after randomly selected clinic visits. Using the scoring scheme used by survey organization, answers to each question (on a 5-point Likert scale) were converted to a 0–100 scale. The overall score for each visit was an average of 6 section scores: Access, Moving through the visit, Nurse/assistant, Care provider, Personal issues, and Overall assessment. The overall visit score was aggregated at provider, clinic and region levels. We focus on overall and section scores across 4 regions of the organization.
Results R/E differences were substantial, from 90.7 (NHW) to 85.2 (Asian). Regional scores ranged from 88.2 (region A) to 90.1 (region D). The ranking was not consistent, however, within R/E groups; Region A scored highest among Latino and 2nd among NHW patients, while region D scored lowest among Latino and 3rd among NHW patients. True differential treatment by R/E should be most apparent in “Care provider” and “Personal issues”, and less so in “Access” and “Moving through the visit”. In fact, the R/E difference (NHW vs. Asian) was smallest for “Care provider” (4.3) and largest for “Moving through the visit” (7.0) scores. R/E differences were much less pronounced if the top 2 Likert categories were combined.
Conclusions Patient satisfaction ratings are influenced by patient R/E in that similar experiences may not be rated equally. When R/E composition varies across providers or regions, scores need to be adjusted to account for this effect or scales less sensitive to differential ‘top box’ scoring may be appropriate.
