Objective. To examine the musculoskeletal manifestations in a large co
hort of patients (n = 410) diagnosed with either a well-established co
nnective tissue disease (CTD) (n = 197) or an early undifferentiated C
TD (n = 213) with a symptom duration of <1 year. This study was aimed
at determining the predictive value of demographic, clinical, and labo
ratory features on outcome in patients with unexplained polyarthritis
(UPA) (from the early undifferentiated CTD cohort; n 67) or rheumatoid
arthritis (RA) (from the well-established CTD cohort; n = 57), over a
5-year followup period. Methods. Patients from both cohorts were asse
ssed at years 1, 3, and 5. At the study visits, clinical data were col
lected in a standardized manner, and sera were obtained and stored. A
priori criteria were established for patient ascertainment and diagnos
is over the duration of the study, Standard statistics were used for c
omparisons of baseline characteristics in patients diagnosed as having
systemic lupus erythematosus, RA, undifferentiated CTD, and UPA at en
try into the cohorts. Baseline features in patients with UPA were exam
ined according to the different subsequent outcomes (RA, CTD, or undif
ferentiated CTD, remission [nonpersistent], or persistent or active UP
A). Baseline features in patients with RA whose disease remained activ
e versus those in whom remission was attained were also examined. Two
multivariable analyses, classification trees and polychotomous logisti
c regression, were performed to predict disease outcomes over time. Re
sults. The overall rate of ascertainment for the 410 patients ranged f
rom 90% at year 1 to 71% at year 5. Patients with established CTDs sho
wed a tendency for more stable diagnoses than those with early undiffe
rentiated CTDs (90-100% versus 45-70%). Consistent baseline predictors
of persistent active disease among patients with RA, in both univaria
te and multivariable analyses, were higher joint counts for pain and t
enderness and higher erythrocyte sedimentation rate (ESR). In similar
to 20% of patients who were classified as having RA when they original
ly entered the cohort, the disease was in remission at 5 years. Twenty
percent of the patients originally classified as having UPA developed
RA over the duration of the study. These patients tended to be older
and to have swelling of small joints at baseline. However, a consisten
t pattern of predictive variables could not be identified in the multi
variable analyses, other than at year 1 (higher small joint counts for
swelling and higher ESR). Conclusion. Baseline features (joint counts
and ESR) among RA patients were variously predictive of persistently
active disease at years 1-5. Consistent baseline predictors of outcome
among the patients with UPA only emerged at year 1. Remission occurre
d in similar to 20% of RA patients, whereas a similar percentage of pa
tients with UPA developed RA. These findings have implications with re
gard to treatment decisions in patients with early RA and/or UPA.