Figure 4

Establishment of follow-up duration as a measure of patient’s dementia resistance and life expectancy by qualitative and quantitative analysis of trends. (A) Attrition rate between WAVE1 (2009–2011) and WAVE2 (2012–2013) of TILDA as a function of MMSE score (N = 6300) or age (N = 8512) available in WAVE1 data. (B) Distribution of all-cause mortality (black bars), dementia fraction detected at the beginning of follow up on first visit (hollow white bars with a black border), dementia fraction at the end of follow up (grey bars), comorbidities per person at the beginning of follow up and brought to scale by dividing by 10 (striped bars), measured in NACC version 3.0 (N = 15618) as a function of follow-up. The mortality was computed by counting all decedents accumulated during follow up and dividing the number by the initial number in the cohort. Similar method was applied to all-cause dementia, registering all cases that developed by the end of follow up. Long follow-up population (LFP, N = 1700) was compared with short follow-up population (SFP, N = 7300). The dashed line symbolizes similar average number of somatic comorbidities in each cohort at the beginning of follow-up. (C) Distribution of achieved lifespans (life expectancy, empty white bars), ages at the start (black bars) and ages at the end of follow up (grey bars) in NACC, version 3.0 (N = 15618) measured in each follow up cohort as a function of follow-up duration. Long follow-up population (LFP, N = 1700) is compared with short follow-up population (SFP, N = 7300). (D) Distribution of the numbers of patients in NACC, version 3.0 (N = 15612), identified in each follow-up duration cohort. The trendlines indicate proportion of each hypothetical component (LFP/SFP) in the cohorts formed based on follow up duration. (E) Semi-logarithmic plotting of all-cause mortality (AMRT) in the cohorts by age in the NACC sub-populations formed by variable durations of follow up. In the direction of arrow, the regression plots represent: (1) intermediate follow-up duration (N = 5861), (2) total NACC population (N = 14220), (3) SFP sub-population (N = 7078), (4) LFP sub-population (N = 1281). The regression equations are matched to the sub-populations.