OBJECTIVE
To test the previously developed prediction model on another sample of patients and obtain the refined coefficients of formulas and graphs on pooled data in development of methodology for predicting the period till psoriatic arthritis onset in patients with moderate and severe psoriasis.
MATERIAL AND METHODS
We analyzed the data of psoriasis patient registry of the Russian Society of Dermatovenereologists and Cosmetologists. The main sample consisted of patients included in the registry between September 24, 2018 and November 12, 2021 (n=760). The sample for model applicability testing consisted of patients included in the registry between November 13, 2021 and December 19, 2022 (n=398). The quantile lines were obtained on the main sample using polynomial quantile regression analysis. We analyzed applicability of these results to external data via comparing 95% confidence intervals of quantiles of periods till psoriatic arthritis onset, calculated on the new data, and quantile regression lines obtained on the main sample. The formulas of quantile lines for the pooled data were derived by polynomial quantile regression analysis.
RESULTS
Quantile regression lines obtained on the main sample were in good agreement with 95% confidence intervals of quantiles based on the new data. This model was applicable to external data. After combining main and new data, we obtained the adjusted coefficients for the formulas describing quantile lines.
CONCLUSION
Applicability of prediction model for calculating the 95%, 75%, 50% and 25% quantiles of period till psoriatic arthritis onset to external data was confirmed. Refined predictive formulas can be used in clinical practice to assess the likely period till psoriatic arthritis onset in patients with moderate and severe psoriasis.