Page 93 - 南京医科大学自然版

第46卷第8期 陈芳怡，王 威，胡馨茹，等. 心理韧性与中国老年人全因死亡率关联强度的异质性：一项前瞻性队列
2026年8月 研究［J］. 南京医科大学学报（自然科学版），2026，46（8）：1197-1210 ·1199 ·

within this observation window. Follow ⁃ up continued No patients or members of the public were in⁃
from baseline until participant death，the last complet⁃ volved in the design，conduct，reporting or dissemina⁃
ed follow⁃up visit，or the end of the observation period， tion plans of this research. As a secondary analysis of
whichever occurred first. Of the 9 765 participants sur⁃ legally obtained public data，this study utilized anony⁃
veyed in 2011- 2012，we excluded those aged <65 mized information without the use of biological sam⁃
years（n=86），those with missing baseline psychologi⁃ ples. This study was approved by the Research Ethics
cal resilience data（n=1 479），those lost to follow⁃up in Committee of Peking University （IRB00001052 ⁃
2014（n=651）or 2017- 2018（n=40），and those who 13074）. Written informed consent was obtained from
died prior to the baseline assessment（n=28）. Conse⁃ all participants or their proxy respondents.
quently，7 481 participants were included in the final 1.2 Methods
analysis. A detailed overview of the study workflow is 1.2.1 Psychological resilience
［9］
presented in Figure 1. In accordance with previous research ，psycho⁃

Study design workflow
Step 1：Participants inclusion and exclusion Step 2：Investigate association

9 765 participants from Propensity score methods
CLHLS in 2011-2012

86 age<65 years
1 479 missed from the Data after IPW Data after PSM
baseline psychological resilience

8 200 potential eligible cohort
Kaplan⁃Merier curves
Log⁃rank test
651 loss to follow⁃up in 2014
40 loss to follow⁃up in 2017-2018
28 died prior to baseline
COX proportional hazards
regression model
7 481 participants available for analysis
Step 3：Explore heterogeneity
Raw data Obesity

Low resilience group ABSI
（scores<14，n=3 214） BRI
STEPP
WHtR
High resilience group
（scores≥14，n=4 267） BMI

图1 研究设计流程图
Figure 1 Study design workflow

logical resilience was measured using five items from psychological resilience scores ranged from 0 to 20.
［9］
the Wave 6 survey：① looking on the bright side of A previous study has employed the median resil⁃
things；② feeling fearful or anxious；③ feeling lonely ience score as a grouping threshold；this approach en⁃
and isolated；④making one’s own decisions；⑤feeling sures balanced group sizes and avoids assumptions re⁃
useless with age. Participants rated the frequency of garding score distribution，thereby enhancing statisti⁃
these experiences on a 5⁃point scale ranging from“al⁃ cal power in comparative analyses. Restricted cubic
ways”to“never”. Consistent with prior studies，items spline analysis of the association between psychologi⁃
1 and 4 were scored positively（always=4，often=3， cal resilience scores and all ⁃ cause mortality risk re⁃
sometimes=2，rarely=1，never=0），whereas items 2，3， vealed that the hazard ratio（HR）crossed the null value

and 5 were reverse ⁃ scored（always=0，often=1，some⁃ （HR=1）at a resilience score of 14，which coincided
times=2，rarely=3，never=4）. Consequently，total with the sample median. Consequently，we defined the