Mapping online patient reviews onto a data-informed service-experience framework: a content analysis of tertiary...

Online patient reviews may provide patient-generated textual feedback on hospital service experience and may complement formal patient-reported experience measures (PREMs), but evidence from ordinary Chinese prefecture-level cities remains limited. This study examined whether online patient reviews could be systematically mapped onto a data-informed service-experience framework in tertiary hospitals in Wuxi, China.

We conducted a cross-sectional content analysis of 2,381 online patient reviews from Dianping for tertiary hospitals in urban Wuxi. The extracted rating variable referred to each review's overall platform rating. Free-text comments were used to identify service-experience content behind ratings. After preliminary reading, pilot coding, clinician discussion, and framework refinement, the final five-domain coding framework was fixed before annotation. Each review was classified by main experience domain, secondary experience domain, and text sentiment. We described domain distributions and examined associations with rating levels, hospital type, time period, and campus characteristics. Factors associated with low ratings were analyzed using logistic regression with campus-clustered robust standard errors, with sensitivity analyses using mixed-effects, GEE, ordinal, and sentiment-excluded models.

Online patient reviews showed interpretable domain patterns within the fixed, data-informed framework. Efficiency/Process was the most common main domain, followed by Perceived Quality and Convenience/Environment. Low-rating reviews were more often centered on Attitude/Communication and accompanied by Negative or Mixed text sentiment. Compared with reviews primarily focused on Perceived Quality, reviews centered on Attitude/Communication showed stronger associations with low ratings, whereas reviews centered on Efficiency/Process or Convenience/Environment showed weaker associations. Negative and Mixed text sentiment were both strongly associated with low ratings. Contextual variables such as hospital type and parking availability were less consistent across modeling frameworks and should be interpreted cautiously.

Online patient reviews of tertiary hospitals in Wuxi could be systematically mapped onto a fixed, data-informed five-domain framework and may provide supplementary signals for identifying recurrent perceived service frictions, particularly those related to communication and process improvement. These findings may offer complementary information for patient-centered service monitoring in similar urban tertiary-hospital contexts, but should be interpreted alongside formal PREMs, patient-experience surveys, complaint records, and hospital quality indicators.