A logistic regression model for the decision to perform access surgery

Area of Science:

• Periodontology
• Dental Surgery
• Biostatistics

Background:

• Advanced periodontal disease affects a significant portion of patients, often necessitating surgical intervention.
• Predicting the need for access surgery is crucial for effective treatment planning in periodontics.

Purpose of the Study:

• To identify clinical parameters significantly associated with the need for access surgery or extraction in advanced periodontal disease using multivariate logistic regression.
• To evaluate the accuracy, sensitivity, and specificity of a predictive model for surgical intervention.

Main Methods:

• A multivariate logistic regression analysis incorporating clinical parameters (tooth type, furcation involvement, bleeding on probing, attachment level, probing depth, mobility, BANA test score) was performed using generalized estimating equations (GEE).
• Receiver-operator characteristic (ROC) curves were generated to assess the model's performance at a probability cutpoint of 0.8.
• The model's predictions were compared against clinicians' decisions post-treatment (scaling and root planing plus antimicrobials).

Main Results:

• The GEE model identified significant predictors (p < 0.05) for access surgery or extraction.
• At a 0.8 probability cutpoint, the pretreatment model achieved 76.1% sensitivity and 75.3% specificity.
• Post-treatment, the model's agreement with clinicians' decisions was 80% accurate, with 90% specificity and 43% sensitivity.

Conclusions:

• Multivariate logistic regression effectively identifies clinical parameters predicting the need for periodontal surgery.
• The developed model demonstrates reasonable accuracy in predicting the need for access surgery or extraction.
• Clinical decision-making post-treatment showed high specificity but lower sensitivity when compared to the predictive model.