Clinical trials in cognitively impaired older adults: home versus clinic assessments.

OBJECTIVE

To compare the reliability of instruments used in clinical trials involving cognitively impaired older adults when the instruments are administered in-home rather than in-clinic and to compare withdrawal rates is these two groups.

DESIGN

This study was part of a larger n-of-1 clinical trial to investigate the efficacy and safety of a MAO/A inhibitor (Brofaromine) in patients with Alzheimer's disease. Participants were initially assessed at the clinic (baseline) and then randomly allocated to in-home or in-clinic assessments for the remainder of the trial. The baseline and second assessment (performed before initiation of the treatment) were used for the reliability analysis. Withdrawal rates were examined over the course of the 6-month trial.

SETTING

Assessments took place at a geriatric clinic in an urban university teaching hospital and at residences of some of the patients.

PARTICIPANTS

Forty-six Alzheimer's disease patients participated in the study, of which, 22 were randomized to in-home assessments and 24 to in-clinic assessments.

MEASUREMENTS

Test-retest reliability was measured for all five instruments used in the study and was based on the first two assessments. Sample size requirements, based on within-group variance, were calculated. Withdrawal rates were obtained for the total duration of the trial.

RESULTS

Test-retest reliability of the instruments, as determined by intraclass correlations, was good in both groups but favored in-clinic for all but one instrument (range: 0.47-0.90 for in-home vs 0.57-0.92 for in-clinic). Sample size requirements based on reliability assessment data were found to be larger for some instruments when administered in-home. Only four in-home patients withdrew before completion of the study, compared with eight in-clinic patients.

CONCLUSION

The results suggest the in-home assessments in cognitively impaired older adults may result in lower withdrawal rates but may necessitate larger sample sizes to offset larger test-retest variability.