Data analysis
The proportion reporting pain and depression (3 and
more depressive symptoms out of the 8-item CES-D) at
follow-up was calculated according to baseline pain and
depression status. Chi-square analysis was conducted to
determine the odds ratios (ORs) of developing the two
outcomes according to baseline status. Logistic regression
analyses were then performed to evaluate independent
predictors of pain and depression at the follow-up
assessment, first predicting incident outcomes and then
predicting pain or depression status at 2 years with
baseline pain and depression as predictors. All predictive
variables were entered into the model, with an
inclusion criterion of pb0.05. Results were reported
using ORs and 95% confidence intervals (CIs).