Background/Purpose:   The ability to risk stratify patients for short term mortality is important in SSc patient care and clinical trial design; but there is no externally validated short term mortality model in early diffuse cutaneous systemic sclerosis (dcSSc). We have previously used a prospectively enrolled large American single center SSc databank to develop and internally validate a 4-factor risk stratification model for two year mortality.  Our objective was to externally validate this two year mortality model in a European population of early dcSSc Caucasian patients. 

Methods:   A large European single-center prospectively enrolled SSc databank was used to identify an inception cohort of adult early dcSSc seen for an initial visit between 2000 and 2006 to serve as the external validation cohort.  Patients were considered to have early dcSSc if their first visit occurred< 2 years after the first symptom attributable to SSc with skin thickening proximal to the elbows or knees.  Vital status was determined using the UK National Care Records Service.

We used a 4-factor model (age at first visit, gastrointestinal severity, skin thickness progression rate and presence of anemia) to calculate a total sum score (range of -1 to 8) and risk stratify patients into low, moderate and high risk for two year mortality in the original derivation and external validation cohort.  Multiple data imputations were used for missing data in the validation cohort.  Stratum-specific chi-squares were compared for the low, moderate and high risk groups.

Results:   Of 160 European early dcSSc patients, 110 were Caucasian, and formed the external validation cohort.   The validation cohort had a mean age of 51.2 ± 12.4 years, was 74% female and a mean modified Rodnan skin score (mRss) of 26.6 ± 8.1.  Median disease duration from first SSc symptom was1.02 years (IQR 0.78, 1.49), or 0.78 years (0.52, 1.17) from first non-Raynaud symptom.  There was no significant difference in age, gender, disease duration or mRss between the validation and derivation cohort.  At two years, 7.3% had died in the validation cohort, compared to 22.6% in the original derivation cohort (p=0.0005).   The stratum-specific comparison of mortality rates using this risk model are shown in Table 1.  There was no difference in prediction of low or moderate risk groups between the models. 

Conclusion:   We have now validated a two year mortality risk stratification model for early dcSSc patients in American and European cohorts.  In this external validation, the model accurately predicted those at low and moderate risk of death at two years from the first visit.  There was a significantly lower rate of death in the validation cohort, possibly related to the inclusion of only recent patients or differences in the underlying populations, which may have affected performance in the high risk group. This model may be used to risk stratify patients for short-term mortality.