Demographics and Cognitive Screening
A total of 705 individuals were scheduled for screening appointments across the six 2017-2018 Memory Screening Day events (see Figure 1). One-hundred-thirty-one individuals (18.6%) failed to keep their scheduled appointment and 574 participants (81.4%) completed screening. The mean age of those who completed screening was 74.5 years (s.d.=7.4; range=45-96 years) and the mean level of education was 16.1 years (s.d.=2.9; range=0-20 years). There were 252 men (43.9%) and 322 women (56.1%). Although screening was targeted towards those age 65 or older (as indicated in the newspaper article), a small number of younger individuals (n=37) were allowed to participate due to a positive family history of dementia and significant anxiety related to their cognitive function. The age distribution of those screened is shown in Figure 2. There were 95 aged 65-69 years (16.6%), 309 aged 70-79 years (53.8%), 118 aged 80-89 years (20.6%), and 15 aged 90 years or older (2.6%). Almost half (46%) of those screened reported a positive family history of dementia (i.e., dementia in a first degree relative), and approximately half (46.4%) answered yes to “concern about memory by yourself or others” on the MCQ (see supplemental Table). The average score on the MCQ was 2.2 (s.d.=2.3, range=0-10).
Ethnicity/race was reported as non-Latino White for 492 participants (85.7%), Hispanic/Latino for 44 participants (7.7%), Asian-American for 18 participants (3.1%), African-American for 7 participants (1.2%), American Indian/Alaskan Native for 2 participants (0.3%), and Pacific Islander-American for 2 participants (0.3%). Eleven participants (1.9%) declined to state their ethnicity/race. Cognitive testing was administered in Spanish for 19 of the 44 Hispanic/Latino participants (43%), and in English for the remainder of the sample (n=555). The English-speaking participants averaged 23.8 (s.d.=4.0; range=6-30) on the MoCA, 11.4 (s.d.=4.4; range=0-22) on immediate Story recall, and 10.0 (s.d.=4.8; range=0-26) on delayed story recall. The 19 participants tested in Spanish averaged 24.4 (s.d.=5.7; range=6-30) on the MMSE, 15.8 (s.d.=5.8; range=0-22) on trials 1-3 total of the CERAD Memory Test, and 5.8 (s.d.=2.7; range=0-26) on delayed recall of the CERAD Memory Test.
Based on our cognitive test score algorithm, 297 participants (52.0%) were classified as “Normal, Age-Appropriate Performance”, 192 (33.6%) as “Possible Memory Decline”, and 82 (14.4%) as “Likely Memory Decline”. Three participants did not complete testing and were not classified. Demographic information for the three classification groups is presented in Table 2. The classification groups differed in age [F(2,568)=22.77; p<.001] with “Likely Decline” older than “Possible Decline” (p<.001) and “Normal” (p<.001), and “Possible Decline” older than “Normal” (p<.001). The groups also differed in education [F(2,565)=9.13; p<.001] with “Likely Decline” (p<.008) and “Possible Decline” (p<.001) less educated than “Normal”, but not different from each other. The percentage of men differed across the three groups [χ2=23.4; p<.001] with the “Likely Decline” group having a higher percentage than the other two groups [vs. “Normal”: χ2=23.4; p<.001; vs. “Possible Decline”: χ2=23.4; p<.001], and the “Possible Decline” having a higher percentage than the “Normal” group [χ2=4.8; p=.03]. There were fewer participants with a positive family history of dementia in the “Likely Decline” group than in the “Possible Decline” [χ2=9.1; p=.01] and “Normal” [χ2=10.6; p=.005] groups, which did not differ from each other. The percentage of Hispanic/Latino participants differed across groups [χ2=6.1; p<.05] with a higher percentage in the “Possible Decline” group than in the “Normal” group [χ2=6.0; p=.01] only. The classification groups differed in scores on the MCQ [F(2,475)=9.86; p<.001] with “Likely Decline” higher (i.e. more concerns) than “Possible Decline” (p<.01) and “Normal” (p<.001), and “Possible Decline” higher than “Normal” (p<.03). The percentage of each classification group that endorsed each MCQ item is shown in the supplemental Table.
The classification groups tested in English, as expected, differed in scores achieved on the MoCA [F(2,551)=282.72; p<.001], immediate story recall [F(2,549)=219.92; p<.001], and delayed story recall [F(2,549)=260.67; p<.001] (see Table 2). The three classification groups differed from each other on all three test measures (all p-values < .01). Groups tested in Spanish were too small for separate analysis (n=5 “Normal”, n=11 “Possible Decline”, n=3 “Likely Decline”).
Attitudes Towards Research and Specific Research Procedures
Three-hundred-seventy-eight of the 574 (66%) screened participants agreed to be included in the ADRC Research Registry and to be contacted about future research participation. Those who agreed and those who declined did not differ in average age [t(572)=1.75,p=.08], but those who agreed had more education [t(566)=-3.02,p=.003] than those who declined (see Table 3). The groups did not differ in the percentage of females [χ2=1.52, p=.22], percentage of those with positive family history [χ2=4.15, p=.13] or percentage of Hispanic/Latino individuals [χ2=1.73, p=.19]. Those who agreed to additional research scored worse on the MCQ than those who declined [t(476)=-2.71, p=.007]. The two groups did not differ on MoCA scores [t(550)=-1.76, p=.08], but those who agreed had worse scores on immediate [t(550)=-2.72, p=.007] and delayed [t(550)=-2.01, p=.045] Story recall than those who declined. Different percentages of individuals from the “Normal” (65.0%), “Possible Decline” (72.4%) and “Likely Decline” (52.4%) screening classifications agreed to further research [χ2=10.3, p=.006] with “Likely Decline” less likely to agree than “Possible Decline” [χ2=10.2, p=.001] or “Normal” [χ2=4.3, p=.04].
Those that agreed to be included in the Research Registry (n=378) were asked about their willingness to participate in various AD research procedures and the availability of a study partner. The number of individuals who responded “Yes,” “Maybe” or “No” to various potential research procedures is presented in Table 4. Overall, 79.1% responded “Yes” or “Maybe” to a research lumbar puncture (LP), 82.0% responded “Yes” or “Maybe” to a research MRI, and 82.0% responded “Yes” or “Maybe” to autopsy. In addition, 77.0% responded “Yes” or “Maybe” when asked if a study partner would be available. More than half of the participants (192/378; 50.8 %) agreed to all 3 procedures and reported an available study partner. When we separately examined responses from the 25 Hispanic/Latino participants who completed the questionnaire, 68.0% (17/25) responded “Yes” or “Maybe” to a research LP, 84.0% (21/25) responded “Yes” or “Maybe” to a research MRI, 80.0% (20/25) responded “Yes” or “Maybe” to autopsy, and 72% (18/25) responded “Yes” or “Maybe” when asked if a study partner would be available.
Recruitment and Enrollment into Additional Research
The 378 individuals entered into the Research Registry were reviewed to determine if they satisfied inclusion and exclusion criteria for the ADRC longitudinal observational study. One-hundred-seven (28.3%) did not satisfy these criteria and were not eligible for recruitment into the study. The primary reason for exclusion was related to some aspect of the participant’s health or medical history (e.g., history of stroke, significant psychiatric disorder) that was inconsistent with study criteria (n=63). Other reasons included lack of an available study partner (n=15), young age (n=13), severe cognitive impairment (n=9), infeasible travel distance (n=5), or non-compliance with study procedures (n=2).
Efforts were made to contact the remaining 271 individuals by telephone. Forty-three individuals could not be contacted (11.4%). The remaining 228 were contacted and asked about willingness to enroll in the ADRC longitudinal study after study procedures had been fully described. One-hundred-nineteen declined to participate and cited unwillingness to comply with one or more required study procedures (n=73), lack of interest (n=35), or a variety of personal reasons (e.g., traveling, family emergency) (n=11). Eighty-eight individuals were successfully enrolled in the ADRC longitudinal study and agreed to all required study procedures (i.e., yearly cognitive testing, baseline LP and MRI, autopsy). Twenty-one were referred directly to affiliated research studies, including clinical trials, and were successfully enrolled. These individuals either made a specific request to join a clinical trial or did not meet all criteria for enrollment in the ADRC longitudinal study. Overall, 19% (109/574) of all screened individuals were successfully recruited into AD research as part of the ADRC longitudinal observational study (88/574; 15.3%) or an affiliated AD-related study or clinical trial (21/574; 3.7%).
Validation of Memory Screening Classifications
As part of their participation in the ADRC longitudinal study, 86 individuals who had been screened received comprehensive clinical, neurological and physical evaluations, and a consensus diagnosis (or classification) of dementia (and etiology), Mild Cognitive Impairment (MCI), or Cognitively Normal (ADRC procedures have been described in detail previously; [24]). The ADRC evaluations generally occurred within 6 months after screening. Two of those screened enrolled in the ADRC but did not complete the full set of evaluations. The concordance between screening classification and ADRC “gold standard” diagnosis/classification is shown in Table 5. The overall accuracy of the screening classification was approximately 72%. Accuracy was 88.9% for the “Normal” classification, 36% for the “Possible Decline” classification (if an ADRC diagnosis of either MCI or dementia is considered correct), and 71.4% for the “Likely Decline” classification (if an ADRC diagnosis of either MCI or dementia is considered correct).
Fourteen of 20 individuals who received an ADRC diagnosis of dementia or MCI had been classified as having “Possible Decline” or “Likely Decline” by screening (i.e., 70% sensitivity). Most of the accurately classified (by screening) truly-impaired (by ADRC diagnosis) individuals had been classified as “Possible Decline” by screening (n=9 of 14). Forty-eight of 66 individuals who were considered “normal” based on the ADRC evaluation were classified as “Normal Cognition” by screening (73% specificity). Most misclassified truly-normal individuals (by ADRC diagnosis) had been misclassified as “Possible Decline” by screening (n=16 of 18). The negative predictive value of the memory screening procedure (i.e., no true decline by ADRC evaluation when memory screening classified as normal) was 89%, while the positive predictive value (i.e., true decline by ADRC evaluation when memory screening classified as having decline) was only 44%.