Clinical Parameters Cannot Accurately Predict Periodontal Disease Progression

Objectives: We have previously applied linear mixed models (LMM) to longitudinal measurements of clinical attachment loss (CAL) to identify progressing periodontal sites. Using this methodology, we classified subjects as progressing or stable after 12 months of monitoring for changes in CAL. We then explored if baseline clinical parameters could be used to predict which subjects would present periodontal disease progression.
The objective of the study was to determine if mean baseline clinical parameters could predict periodontal disease progression at the subject level.
Methods: We monitored 302 periodontitis subjects for up to 12 months without therapy. Periodontal clinical parameters were measured bimonthly and LMM predictions of CAL measurements were used to classify sites as progressing or stable. Subjects were then categorized as stable (no progressing site) or progressing (≥3 progressing sites). We applied forward model selection with logistic regression on the mean clinical parameters at baseline: probing depth, CAL, bleeding on probing, plaque index (PI), gingival index, and suppuration, adjusting for age.
Results: Out of the 302 participants, 143 were classified as stable, while 64 subjects were deemed progressing. The remaining 95 subjects, with only 1-2 progressing sites were excluded from the present analysis. Mean baseline PI was the only predictor of periodontal disease progression selected by the logistic regression model, with an odds ratio of 9.2 (95% confidence interval: 1.8-45.9). The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was 0.62 (Fig.1).
Conclusions: The final model included only PI as a predictor and had a low 62% accuracy for discriminating progressing and stable subjects. Our findings confirm previous reports indicating that periodontal clinical parameters are poor risk predictors for periodontal disease progression.