COVID-19 Altered content but not rate of negative dental reviews

Research output: Contribution to conferenceAbstractpeer-review



Since March 2020, COVID-19 has restricted access to National Health Service (NHS) dental care. This has led to reports of growing dissatisfaction. This study uses text mining, sentiment analysis and topic modelling of patient reviews left on the website to identify the rate and content of dissatisfied reviews before and after COVID-19 restrictions.

A Python script scraped reviews from the website from June 2019-June 2021. The DetectSentiment Amazon Web Services algorithm was used to categorise reviews by sentiment. Negative sentiment reviews were assigned to pre- and post- COVID-19 restriction timepoints. Review texts were prepared for topic modelling with stopword removal, stemming, lemmatisation, and tokenisation. Latent Dirichlet Allocation (LDA) topic modelling was performed using the Gensim LDAmodel Python package, using 10 topics per timeframe. The pyLDAvis Gensim package was used to visualise and interpret salient themes and terms.

Of 15458 reviews, 1849 reviews displayed negative sentiment. There was no significant difference in proportion of negative reviews before (n=943, 13.79%) and after (n=906, 13.25%) the introduction of COVID-19 restrictions (chi-square = 0.9883, p = .320). Topic modelling coherence scores were similar in both pre (0.45) and post (0.48) groups. The three predominant themes prior to COVID-19 restrictions focused on interpersonal relationships (terms: ‘Receptionist’, ‘Rude’, ‘Staff’); treatment problems (terms: ‘Treatment’, ‘Emergency’, ‘Pain’); and cancellation of appointments (terms: ‘Appointment, Time, Cancel’). The three predominant themes after COVID-19 restrictions related to treatment charges (terms: ‘Pay, Practice, Treatment’); difficulty attaining appointments (terms ‘’Appointment’, ‘Call’, ‘Wait’); and difficult attaining emergency care (terms: ‘Emergency’, ‘Phone’, ‘Reception’). ‘Pain’ was a more prevalent term in the post-restriction group.

COVID-19 restrictions changed content of reviews. The salient themes identified reflect reports of patients being unable to attain NHS dental appointments and seeking private care at increased cost. LDA can be used to aggregate patient complaints for targeted service improvement.
Original languageEnglish
Publication statusPublished - 25 Jun 2022
Event2022 IADR/APR General Session & Exhibition - VIRTUAL EXPERIENCE
Duration: 20 Jun 202225 Jun 2022


Conference2022 IADR/APR General Session & Exhibition


Dive into the research topics of 'COVID-19 Altered content but not rate of negative dental reviews'. Together they form a unique fingerprint.

Cite this