Page 57 - Contributed Paper Session (CPS) - Volume 2

CPS1419 Jinheum K. et al.
4. Illustrative real example
PAQUID data were collected to investigate the effects of dementia
on mortality. Samples were taken from community residents of two
southwestern regions (Gironde and Dordogne) of France (Helmer et al.,
2001). The population consists of residents aged65 or above between
1988 and 1990, whose socio-demographic characteristics and mental
health status were recorded every two to three years. Among a total of
3,675 persons selected to participate in the study, 832 (22.6%) were
diagnosed with dementia, 639 of whom died. The remaining 2,843
participants (77.4%) did not experience dementia but 2,298 of them
died.
In this article, we performed an analysis based on ‘paq1000’ data,
which included 1,000 randomly selected observations from the PAQUID
data (Touraine et al., 2015). The paq1000 data consist of several pieces
of information, such as the mental health status (diagnosed with
dementia or dementia-free), dead or alive status, ages (including a
participant’s age at the start of study ( ), their age at the last dementia-

free visit ( ), their age when they were diagnosed with dementia ( ),

their age at their time of death ( ), their age at censoring ( ), gender,

and educational background (educated or non-educated in terms of
graduation from elementary school, say certificate). For an illustration
of the proposed method, we assume a subject’s age at death to be
censored in an interval around the observed value. To this end,
assuming that is age-at-death of subject , the endpoints of the

interval ( , ] for subject are defined as = − and = +

1

, where and are random variates generated from (0,3).
1
2
2
When a subject’s age at death is censored at time , we set = and

= ∞, i.e., right-censored at .

For the 0 → 1 transition, men showed intensity that was 1.899 times
higher than that of women, yielding a decently significant result with a
-value of 0.0016. The non-educated group showed an intensity that
was 1.277 times higher than the educated group, although this was not
significant (a -value of 0.3293). For the 1 → 2 transition, the intensity
for women is 1.222 times higher than it is for men, but this is not
significant (a -value of 0.711). The intensity for the educated group is
1.248 times higher than the non-educated group, but this is not
significant (a -value of 0.6808). Finally, for the 0 → 2 transition, the
intensity for women is 6.816 times higher than it is for men with a very
46 | I S I W S C 2 0 1 9