Fig. 2

Odds Ratio (95% CI) estimated from the logistic regression analysis of the significant clinical predictors associated to MO resolution after 6 months of anti-CGRP MAbs treatment. Odds ratio (95% CI) estimated from the 10-fold CV of the stepwise logistic regression with selection criteria of minimizing the AIC for a variable to be eliminated from the selected model. The final logistic regression model, in the test set, presented an accuracy of 0.744 [0.579–0.870] and a Kappa value of 0.430 (p = 0.047). Only headache intensity remained as a statistically significantly independent predictor in the model (OR [95%]: 0.236[0.054–0.975]; p = 0.049). Unilateral Pain, BZD Use, BAI and AM burden were not statistically significant in the logistic regression analysis (If the 95% CI includes/crosses the value of 1, then there is insufficient evidence to conclude that these covariables are statistically significantly and independently associated to the outcome variable). MO: medication overuse, OR: odds-ratio; CI: confidence interval; BZD: benzodiazepines; AM: acute medication; BAI: Beck anxiety inventory; AUC: area under the curve