Summary:
In patients with type 2 diabetes mellitus and periodontitis enrolled in a randomized clinical trial, periodontal therapy including scaling and root planing with machine learning analysis of baseline factors demonstrated significant improvement in clinical parameters (probing depth, clinical attachment level, plaque index) and HbA1c at 6 months compared to patients stratified by baseline characteristics to identify predictors of treatment success, with nine baseline factors identified as important for predicting treatment response including initial periodontal severity and glycemic control.
| PICO | Description |
|---|---|
| Population | Patients with type 2 diabetes mellitus and periodontitis enrolled in a randomized clinical trial of periodontal therapy. |
| Intervention | Periodontal therapy including scaling and root planing, with machine learning analysis of baseline factors. |
| Comparison | Patients stratified by baseline characteristics (probing depth, clinical attachment level, plaque index, HbA1c) to identify predictors. |
| Outcome | Significant improvement in clinical parameters and HbA1c at 6 months. Nine baseline factors identified as important predictors of treatment response. |
Clinical Context
Periodontitis and type 2 diabetes share a bidirectional relationship. Treatment of periodontitis can improve HbA1c by approximately 0.3-0.4%.
Clinical Pearls
1. Baseline Severity Predicts Response: Patients with worse baseline periodontal parameters tend to have more absolute improvement but may still have more residual disease.
2. Glycemic Control Matters: Higher baseline HbA1c predicted poorer periodontal treatment outcomes.
3. Plaque Control is Modifiable: Baseline plaque index predicts outcomes and represents a modifiable factor.
4. Machine Learning for Personalization: The identification of nine interacting factors suggests complex, multifactorial treatment response.
Practical Application
Assess factors that predict treatment response before initiating periodontal therapy. Coordinate with diabetes providers to optimize glycemic control.
Study Limitations
Post hoc analysis may identify associations that don’t replicate. Machine learning models can overfit. External validation needed.
Bottom Line
Nine factors including periodontal severity, glycemic control, and plaque levels predict periodontal treatment success in diabetic patients.
Source: Castro Dos Santos N, et al. “Factors Influencing the Response to Periodontal Therapy in Patients with Diabetes: Post Hoc Analysis of a Randomized Clinical Trial Using Machine Learning.” Read article
