Introduction: In order to assess patient experiences of telemedicine, researchers and administrators use the net promoter score (NPS), based on a likelihood to recommend (LTR) question. However, there is reason to doubt validity of this metric for this purpose. We assessed the degree to which the LTR question reflects actual patient preferences about telemedicine. Methods: Using data from a patient experience survey collected in Spring 2020, we compared LTR responses to open comments. Through content analysis, we transformed comments into categorical variables and used those variables in a multiple logistic regression model to predict LTR responses. We also thematically analyzed comments to further elucidate our results. Results: Only about half the comments mentioned telemedicine at all. Around 6% of comments were wholly incongruent with LTR responses. In many comments, ideas about telemedicine were semantically entangled with ideas about providers. Our logistic regression found strong associations between sentiments expressed in comments and LTR responses. However, comments about telemedicine were relatively poor predictors for LTR compared to comments about the provider. Discussion: NPS, which is included on many patient experience surveys used by health systems across the United States, has limitations for use as a measure of the acceptability of telemedicine for patients. Patients have more than telemedicine in mind when responding to the LTR question, and ratings conflate attitudes about providers, office policies, and staff with the telemedicine modality. More direct measures are necessary for meaningful research on the acceptability and usability of telemedicine for patients.
Keywords: acceptability of telemedicine; likelihood to recommend; net promoter score; patient experience metric; survey validation.