Application of sentiment and word frequency analysis of physician review sites to evaluate refractive surgery care

Victoria Vought; Rita Vought; Andrew S Lee; Irene Zhou; Mansi Garneni; Steven A Greenstein

doi:10.1016/j.aopr.2024.03.002

Application of sentiment and word frequency analysis of physician review sites to evaluate refractive surgery care

Adv Ophthalmol Pract Res. 2024 Mar 7;4(2):78-83. doi: 10.1016/j.aopr.2024.03.002. eCollection 2024 May-Jun.

Authors

Victoria Vought¹, Rita Vought¹, Andrew S Lee¹, Irene Zhou¹, Mansi Garneni², Steven A Greenstein^{1

3}

Affiliations

¹ Institute of Ophthalmology and Visual Science, Rutgers New Jersey Medical School, Newark, NJ, USA.
² Washington University School of Medicine, St. Louis, MO, USA.
³ Cornea and Laser Eye Institute, Teaneck, NJ, USA.

Abstract

Background: Online physician reviews increase transparency in health care, helping patients make informed decisions about their provider. Language processing techniques can quantify this data and allow providers to better understand patients' experiences, perspectives, and priorities. The objective of this study was to assess patient satisfaction and understand the aspects of care that are valued by patients seeking refractive care using sentiment and word frequency analysis.

Methods: Written reviews and Star ratings for members of the Refractive Surgery Alliance Society practicing in the United States were collected from Healthgrades, a popular physician rating website. Surgeons with at least one written review were included in the study. Reviews were scored from -1 (most negative) to +1 (most positive) using Valence Aware Dictionary sEntiment Reasoner (VADER). Reviews were stratified by demographic characteristics, namely gender, region, and years in practice. Word frequency analysis was applied to find the most common words and phrases.

Results: A total of 254 specialists and 3104 reviews were analyzed, with an average of 4.4/5 stars and mean 48 ratings each. Most physicians had positive reviews (96%, average VADER = 0.69). Younger physicians (<20 years since residency) had significantly higher Stars rating than senior peers (>20 years) (P < 0.001). A similar trend was observed in VADER score (0.71 vs 0.69), although not statistically significant (P = 0.06). No statistical differences were observed between Stars rating and VADER score by gender (P = 0.66, P = 0.83) or by geographical region (P = 0.74, P = 0.07). "Staff" (n = 1269), "professional" (n = 631), "office" (n = 523), "questions" (n = 424), and "friendly" (n = 386) were frequently used in reviews, along with phrases such as "the staff" (n = 273) and "my questions" (n = 174). "Surgery" (n = 719), "staff" (n = 576), "procedure" (n = 251), "experience" (n = 243), and "professional" (n = 240) were the most common words in positive reviews, while "surgery" (n = 147), "office" (n = 86), "staff" (n = 54), "time" (n = 47), and "insurance" (n = 28) were the most commonly used in negative reviews.

Conclusions: Both the average Stars and VADER sentiment score suggest a high satisfaction among refractive patients. Word frequency analysis revealed that patients value non-clinical aspects of care, including interactions with staff, insurance coverage, and wait-times, suggesting that improving non-clinical factors could enhance patient satisfaction with refractive surgery.

Keywords: Patient sentiment; Physician reviews; Quality improvement.