Background: The modified Telephone Interview for Cognitive Status (TICS-m) assesses cognitive status via the telephone and has been used to recruit for clinical trials by screening for amnestic mild cognitive impairment (aMCI). The utility of screening for aMCI has not been validated, and it is unknown which questions best predict aMCI.
Methods: The Alzheimer's Disease Cooperative Study (ADCS) used the TICS-m to recruit for an aMCI clinical trial. If telephone respondents screened positive for aMCI on the TICS-m, they were referred to a clinical site where they were assessed for the operational criteria of aMCI. Univariate analyses identified the TICS-m questions that best predicted aMCI, creating a final model using forward stepwise logistic regression.
Results: Of 52,722 calls, 16,312 were screened for trial entry. Of these, 4,883 received the TICS-m. The 2,431 that screened positive for aMCI were referred to a clinic, and 527 arrived for clinical assessment. Of these, 266 met operational criteria for aMCI. The positive predictive value of the TICS-m in this population is 50.9% (95% confidence interval [CI], 46.2% to 54.8%). The final model included 5 variables: (1) 10-word list delayed recall (p < 0.001), (2) day of month (p = 0.002), (3) season (p = 0.009), (4) last name of current president (p = 0.020), and (5) month (p = 0.036). This model has a predictive accuracy of 0.61 (95% CI, 0.51 to 0.71) which is similar to the entire TICS-m. The predictive accuracy of the 10-word list delayed recall alone was 0.63 (95% CI, 0.54 to 0.72).
Conclusion: The TICS-m is not optimally efficient for recruitment of subjects for aMCI clinical trials. The 10-word delayed recall has a predictive accuracy similar to that of the entire TICS-m.