Search

Objective. DcSSc is associated with high morbidity related to widespread skin disease and poor prognosis due to earlier and more severe organ involvement. The objective of this study is to derive and validate a simple prediction rule for identifying patients at the time of initial diagnosis of SSc who are likely to progress to dcSSc. Methods. The Nijmegen cohort consists of 619 SSc patients. Logistic regression was used for predictive modelling. A prediction rule was created by rounding regression coefficients. Patients were stratified as being at low risk (<1) or high risk (≽1) of progression to dcSSc. Performance was analysed in 445 SSc patients from Madrid. Results. One hundred and seventy-four out of 535 patients were classified as dcSSc. The final model consisted of gender, time between RP and non-RP, sclerodactyly (first non-Raynaud symptom) and SSc-specific auto-antibodies. The model performed well in the derivation cohort [area under the curve = 0.78 (95% CI: 0.74, 0.82)] and validation cohort [area under the curve = 0.78 (95% CI: 0.74, 0.83)]. At the optimal cut point (1) for the prediction rule, sensitivity was 87% and specificity 61% in the derivation cohort, compared with 78% and 65% in the validation cohort. Upon application of the prediction rule to 392 lcSSc patients at initial diagnosis, 32 out of 34 patients were correctly classified as dcSSc. Conclusion. A simple prediction rule was designed to attribute a low/high risk category for development of dcSSc.This method is suited for assigning intensified screening at the time of initial diagnosis of SSc to patients most at risk for dcSSc. It provides the opportunity for early identification of potential dcSSc patients for enrolment into clinical trials. [ABSTRACT FROM AUTHOR]