To gain insight into the characteristics and representativeness of
the respondents, we tested (1) for differences in age between responders
and nonresponders at follow-up using a t test and (2) the
relation between having or not having MSDs at baseline and
response at follow-up using univariate logistic regression analysis.
Statistical significance was set at a ¼ 0.05. The prevalence and 95%
confidence interval of the MSDs were calculated using the followup
data. Confidence intervals were calculated using the Wald
method when the sample consisted of more than 150 persons and
the adjusted Wald method when the sample consisted of < 150
persons