Abstract

BACKGROUND: Patients increasingly rely on readily available physician reviews to inform their provider choices. Sentiment analysis and machine learning techniques quantitatively analyze written prose to understand patient desires from physician encounters. Patient perspectives on their medical care have been understudied in neurosurgery. OBJECTIVE: To analyze patient reviews of neurosurgeons to uncover trends between patient ratings of their encounters and content in their reviews. METHODS: Identification of neurosurgeons and demographic data were collected from 115 Accreditation Council for Graduate Medical Education–accredited programs using public data. Healthgrades.com was used to obtain online written and star rating reviews which were analyzed using a machine learning sentiment analysis package to generate a sentiment score. Student t tests compared differences between demographics and outcomes from the sentiment analysis. Multivariate regression was performed to examine associations between sentiment rating and word/word pair frequency. RESULTS: One thousand two hundred eighty-four neurosurgeons were found to have review profiles which consisted of 6815 reviews. Analysis revealed a direct correlation between sentiment score and star rating (r2 = 0.554, P < .0001). There were no differences in the sentiment score based on neurosurgeons' sex; however, younger surgeons had more positive reviews (P = .022). Word frequency analysis showed that reviews were less likely to be positive if they included “pain” (odds ratio [OR]: 0.28, CI: 0.24-0.32, P < .0001) or “rude” (OR: 0.03, CI: 0.01-0.06, P < .0001). Reviews were more likely to be positive when they included “kind” (OR: 3.7, CI: 2.6-5.3, P < .0001) or “pain-free” (OR: 3.1, CI: 2.1-4.7, P < .0001). CONCLUSION: Top-rated reviews demonstrate the importance of compassion in patient satisfaction. The word “pain” arose for both negative and positive reviews. Pain management seems to be a salient component of patients' evaluation of their neurosurgical care, thereby underscoring the importance of guiding patient pain expectations.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call