Estimated mean age at diagnosis of Alzheimer’s dementia according to levels of depressive symptoms, neuroticism and cognitive activity (n=2860)
| Status at baseline* | Incident Alzheimer’s dementia (n) | Estimated mean age at diagnosis of Alzheimer’s dementia† (years) | P value‡ |
| Depressive symptoms | |||
| No depressive symptoms (n=1293) | 328 | 92.1 | Reference |
| Mild depressive symptoms (n=654) | 185 | 89.5 | 0.1 |
| Significant depressive symptoms (n=914) | 272 | 86.9 | 0.001 |
| Neuroticism | |||
| Lowest tertile/least neuroticism (n=923) | 193 | 93.1 | Reference |
| Second tertile (n=958) | 288 | 90.5 | 0.03 |
| Highest tertile/most neuroticism (n=979) | 304 | 88.8 | <0.001 |
| Cognitive activity | |||
| Lowest tertile/least activity (n=1198) | 347 | 89.2 | Reference |
| Second tertile (n=803) | 216 | 90.8 | 0.02 |
| Highest tertile/most activity (n=859) | 222 | 92.6 | <0.001 |
*Depressive symptoms measured using the 10-item Center for Epidemiological Studies Depression Scale. No symptoms were defined as a score of 0; mild symptoms as a score of 1–2; and significant symptoms as a score of 3–10. Neuroticism measured using the NEO Five-Factor Inventory (range, 0–48 points). The bottom tertile included scores <12; the second tertile 13–17; top tertile ≥18. Cognitive activity included self-reported frequency over the past year of four activities: reading the newspaper, reading magazines, reading books, playing games. Responses for each activity were averaged to create a score from 1 (once a year or less) to 5 (every day/almost every day). The tertiles were defined by scores of ≤3.5; 3.6–4.0; >4.0.
†Age at diagnosis was estimated using the mean parameters from an extended accelerated failure time model, with a covariate for years of education; education was set as median years of education (16) in the population. This simplified model with education and no other covariates yielded results within approximately 10% of the estimates in the full model with all covariates.
‡P value is from the coefficient comparing each risk factor group to its reference group within a single extended accelerated failure time model controlled for sex, education, cohort, physical activity and number of comorbidities. Separate models were created for depressive symptoms, neuroticism and cognitive activity.