One hundred sixty-two of the 213 participants (76%) completedthe 10-week studyF81 in each arm. To complete thestudy, participants had to attend at least two-thirds of theexercise sessions. Attrition by 10 weeks affected both interventionarms equally: 25 in CBSTand 26 in TC. Reasons forattrition in the CBST arm were did not meet the minimumtwo-thirds total session attendance requirement in the firstfew weeks of the study and were asked to discontinue(n55), dropped out because of other time commitments(n511), could not continue because of unrelated medicalillness (n57), and experienced program-induced exacerbationof chronic musculoskeletal conditions that could notbe alleviated through program alterations (n52). TC correspondingnumbers were similar: poor attendance (n58),time commitments (n512), and unrelated medical illness(n56). There was no TC group attrition due to musculoskeletalcomplaints. Of those who dropped out before studycompletion, there was a difference between CBST and TCaccording to age, such that the early dropouts (beforeWeek5) tended to be younger in TC than CBST (Po.03 accordingto logistic regression analysis). Specifically, by Week 5,approximately 19% of those aged 85 and older in CBST(compared with 4% in TC) dropped out, whereas 34% ofthose aged 65 to 74 in TC dropped out (compared with11% in CBST). Group differences in additional attrition at10 weeks were no longer significant. There were no significantdifferences in any of the baseline balance, stepping, ormobility performance measures presented below betweendropouts and those who completed the study. Of those whocompleted the study, mean sessions attended was 25.7 2(range 20–30) for CBSTand 25.0 2 (range 20–30) for TC,for a mean percentage training attendance rate of 86% forCBST and 83% for TC. The flow of participants throughthe study is summarized in Figure 1.