Can the prognosis of polymyalgia rheumatica be predicted at disease onset? Results from a 5-year prospective study.

To identify the features of PMR that may predict the duration of steroid therapy, the occurrence of relapses and the late development of GCA. Prospective cohort study of 176 patients with PMR, followed up for 5 years. Baseline factors associated with the duration of steroids therapy were identified using Cox regression. Predictors of relapse and the late development of GCA were identified using binary logistic regression. A total of 176 patients with PMR were included, of whom 124 stopped steroids within 5 years. The probability of stopping steroids within 5 years was independently reduced by an elevated plasma viscosity (PV) [hazard ratio (HR) = 0.49; 95% CI 0.29, 0.82 for a PV > or = 2.00 mPa s compared with a PV < or = 1.80 mPa s; overall P = 0.024] and by starting treatment at >15 mg prednisolone (HR = 0.63; 95% CI 0.41, 0.97; P = 0.036). Either of these independently reduced the chances of stopping steroids within a given time interval between 27 and 51%. No significant predictors of relapse were identified. Predictors of late GCA on univariable analysis were female sex [odds ratio (OR) = 8.16; 95% CI 1.06, 63.13; P = 0.044], HLA-DRB1*0101 or -*0401 alleles (OR = 4.95; 95% CI 1.05, 23.34; P = 0.043), PV > or = 2.00 mPa s compared with PV < or = 1.80 mPa s (OR = 10.64; 95% CI 1.28, 88.38; P = 0.029) and initial prednisolone dose >15 mg (OR = 4.53; 95% CI 1.61, 12.79; P = 0.004). A higher PV in PMR increases the risk of prolonged steroid therapy and late GCA. Female sex and particular HLA alleles may increase the risk of late GCA. Starting patients on >15 mg prednisolone is associated with a prolonged steroid duration.