Diagnostic accuracy of IL1β in saliva: The development of predictive models for estimating the probability of the occurrence of periodontitis in non-smokers and smokers.

Aim: To obtain salivary interleukin (IL) 1β-based models to predict the probability of the occurrence of periodontitis, differentiating by smoking habit.
Materials/methods: A total of 141 participants were recruited, 62 periodontally healthy controls and 79 subjects affected by periodontitis. Fifty of the diseased patients were given non-surgical periodontal treatment and showed significant clinical improvement in 2 months. IL1β was measured in the salivary samples using the Luminex instrument. Binary logistic regression models were obtained to differentiate untreated periodontitis from periodontal health (first modelling) and untreated periodontitis from treated periodontitis (second modelling), distinguishing between non-smokers and smokers. The area under the curve (AUC) and classification measures were calculated.
Results: In the first modelling, IL1β presented AUC values of 0.830 for non-smokers and 0.689 for smokers (accuracy = 77.6% and 70.7%, respectively). In the second, the predictive models revealed AUC values of 0.671 for non-smokers and 0.708 for smokers (accuracy = 70.0% and 75.0%, respectively).
Conclusion: Salivary IL1β has an excellent diagnostic capability when it comes to distinguishing systemically healthy patients with untreated periodontitis from those who are periodontally healthy, although this discriminatory potential is reduced in smokers. The diagnostic capacity of salivary IL1β remains acceptable for differentiating between untreated and treated periodontitis.
(© 2020 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)