Table 1 displays characteristics of 79,725 participants with missing data: the mean age was 63 years (SD: 55-72), 42.75% of participants were female, and 95.35% were White. Compared to non-stroke participants, the stroke participants had higher proportions being in the most deprived category (30.78% vs. 22.62%) and suffering from hypertension (56.14% vs. 51.7%), coronary heart disease (16.58% vs. 7.69%), diabetes (12.98% vs. 6.65%), Parkinson’s disease (2.28% vs. 1.41%), AF (28.06% vs. 11.6%), heart failure (14.02% vs. 5.39%), and hyperlipidemia (43.56% vs. 20.75%). The stroke participants were also more likely to have long-standing illnesses (56.35% vs. 34.9%), , a higher proportion were ever smokers (55.36% vs. 48.82%), and former drinkers (51.26% vs. 44.88%).
In this matched cohort study, which only matched for sex (female:42.8%vs 42.49%, P=0.50), year of birth, and APOE state (non-carrier: 71.12% vs 70.97%, P=0.92), we found that the proportion of blood type was very similar between the matched non-stroke cohort and the stroke cohort. For example, the percentage of type O blood was 43.77% in the non-stroke cohort and 42.74% in the stroke cohort (P=0.09).
In comparing the prevalent stroke and incident stroke cohorts, the prevalent cohort (mean age 55 years, IQR: 48-61) was younger than the incident cohort (mean age 71 years, IQR: 65-76). The prevalent stroke cohort also had higher rates of hypertension (35.13% vs. 27.39%), coronary heart disease (12.44% vs. 21.92%), long-standing illness (71.47% vs. 44.68%), and former drinker status (55.28% vs. 46.54%). Additionally, the prevalence of antihypertensive drugs (60.08% vs. 35.84%), cholesterol drugs (67.41% vs. 28.93%), and insulin use (3.97% vs. 2.84%) was greater in the prevalent stroke cohort.
Table 1. Participants characteristics at the study enrollment (2006-10).
Characteristic
|
Total
|
Matched Non-Stroke Cohort
|
Total Stroke
|
Incident Stroke Cohort
|
Prevalence Stroke Cohort
|
n = 79725
|
n = 66013
|
n = 13712
|
n = 7712
|
n = 6000
|
Age at index date, mean (SD), y
|
63(55-72)
|
63 (55-71)
|
64 (55-72) #
|
71 (65-76)
|
55 (48-61)
|
Sex, female, n (%)
|
34080(42.75%)
|
28254(42.8%)
|
5826(42.49%)
|
3329(43.17%)
|
2497(41.62%)
|
Ethnicity
|
|
|
|
|
|
1"white"
|
75630(95.35%)
|
62752(95.54%)
|
12878(94.43%) *
|
7244(94.43%)
|
5634(94.44%)
|
2"Asian or Asian British"
|
1478(1.86%)
|
1169(1.78%)
|
309(2.27%)
|
169(2.2%)
|
140(2.35%)
|
3"black or black British"
|
1111(1.40%)
|
854(1.3%)
|
257(1.88%)
|
147(1.92%)
|
110(1.84%)
|
4"Chinese"
|
169(0.21%)
|
147(0.22%)
|
22(0.16%)
|
17(0.22%)
|
5(0.08%)
|
5"Mixed & Other ethnic group"
|
933(1.18%)
|
762(1.16%)
|
171(1.25%)
|
94(1.23%)
|
77(1.29%)
|
Education attainment, Has a degree, n(%)
|
23205(29.55%)
|
19917(30.61%)
|
3288(24.43%) *
|
1983(26.17%)
|
1305(22.18%)
|
Townsend score
|
|
|
|
|
|
1 (least deprived)
|
20682(25.94%)
|
17799(26.96%)
|
2883(21.03%) *
|
1765(22.89%)
|
1118(18.63%)
|
2
|
20293(25.45%)
|
17054(25.83%)
|
3239(23.62%)
|
1917(24.86%)
|
1322(22.03%)
|
3
|
19688(24.69%)
|
16334(24.74%)
|
3354(24.46%)
|
1906(24.71%)
|
1448(24.13%)
|
4 (most deprived)
|
18962(23.81%)
|
14762(22.36%)
|
4220(30.78%)
|
2112(27.39%)
|
2108(35.13%)
|
Hypertension, n (%)
|
41700(52.30%)
|
34002(51.51%)
|
7698(56.14%) *
|
4611(59.79%)
|
3087(51.45%)
|
Coronary heart disease, n (%)
|
7140(8.96%)
|
4866(7.37%)
|
2274(16.58%) *
|
959(12.44%)
|
1315(21.92%)
|
Diabetes, n (%)
|
6015(7.54%)
|
4235(6.42%)
|
1780(12.98%) *
|
859(11.14%)
|
921(15.35%)
|
Parkinson's disease, n (%)
|
1158(1.45%)
|
845(1.28%)
|
313(2.28%) *
|
174(2.26%)
|
139(2.32%)
|
Atrial Fibrillation, n (%)
|
11278(14.15%)
|
7431(11.26%)
|
3847(28.06%) *
|
2428(31.48%)
|
1419(23.65%)
|
Heart Failure
|
5295(6.64%)
|
3373(5.11%)
|
1922(14.02%) *
|
1088(14.11%)
|
834(13.9%)
|
Hyperlipidemia
|
19541(24.51%)
|
13568(20.55%)
|
5973(43.56%) *
|
3351(43.45%)
|
2622(43.7%)
|
Long-standing illness, disability or infirmity, n (%)
|
29505(37.99%)
|
22028(34.21%)
|
7477(56.35%) *
|
3346(44.68%)
|
4131(71.47%)
|
Parents with CVD, n (%)
|
54656(75.03%)
|
44952(74.38%)
|
9704(78.2%) *
|
5421(77.33%)
|
4283(79.31%)
|
Parents with Dementia, n (%)
|
10047(14.55%)
|
8419(14.66%)
|
1628(14.01%)
|
932(14.14%)
|
696(13.83%)
|
Smoking status, ever smoker, n (%)
|
39308(49.63%)
|
31789(48.45%)
|
7519(55.36%) *
|
4029(52.66%)
|
3490(58.83%)
|
Drinking status, former drinker, n (%)
|
3076(46.12%)
|
2222(44.4%)
|
854(51.26%) *
|
357(46.54%)
|
497(55.28%)
|
Score of 4 curr dep symptoms
|
1 (0-2)
|
1 (0-2)
|
1 (0-3) #
|
1 (0-2)
|
1 (0-3)
|
Use of Antihypertensive Drugs
|
23502(29.78%)
|
17217(26.33%)
|
6285(46.44%) *
|
2729(35.84%)
|
3556(60.08%)
|
Use of cholesterol drugs
|
21176(26.83%)
|
14983(22.91%)
|
6193(45.76%) *
|
2203(28.93%)
|
3990(67.41%)
|
Use of insulin
|
1195(1.51%)
|
744(1.14%)
|
451(3.33%) *
|
216(2.84%)
|
235(3.97%)
|
BMI, Mean (SD)
|
27.04 (24.53-30.02)
|
26.92 (24.46-29.84)
|
27.77 (25.04-31.06) #
|
27.49 (24.81-30.67)
|
28.18 (25.36-31.55)
|
Blood-type haplotype
|
|
|
|
|
|
A
|
34405(43.22%)
|
28438(43.15%)
|
5967(43.58%)
|
3313(43.01%)
|
2654(44.31%)
|
AB
|
2886(3.63%)
|
2373(3.6%)
|
513(3.75%)
|
287(3.73%)
|
226(3.77%)
|
B
|
7608(9.56%)
|
6247(9.48%)
|
1361(9.94%)
|
771(10.01%)
|
590(9.85%)
|
O
|
34703(43.60%)
|
28851(43.77%)
|
5852(42.74%)
|
3332(43.26%)
|
2520(42.07%)
|
APOE ε4 carrier status
|
|
|
|
|
|
0
|
56679(71.09%)
|
46948(71.12%)
|
9731(70.97%)
|
5466(70.88%)
|
4265(71.08%)
|
1
|
21138(26.51%)
|
17490(26.49%)
|
3648(26.6%)
|
2065(26.78%)
|
1583(26.38%)
|
2
|
1908(2.39%)
|
1575(2.39%)
|
333(2.43%)
|
181(2.35%)
|
152(2.53%)
|
Abbreviations: CVD, Cardiovascular Disease; BMI, Body Mass Index, SD, Standard Deviation. Comparison between matched non-stroke cohort and total stroke cohort: * indicates P<0.05 using chi square test, # indicates P<0.05 using Kruskal-Wallis test.
For stroke survivors, over a median follow-up of 19.07 years (IQR: 15.87-23.97), the prevalent stroke cohort of 6,000 participants had 387 dementia cases: 66 AD, 132 VD, and 189 other types. The incident stroke cohort (7,712 participants) was followed for 4.76 years (IQR: 2.05-8.39) and had 442 dementia cases: 53 AD, 181 VD, and 208 other types.
Compared to non-stroke peers, stroke survivors had an increased risk of incident dementia, in the two stroke cohorts (all P<0.001). The risk of developing all-cause dementia was highest in those with intracerebral hemorrhage, followed by ischemic stroke and other stroke types (P<0.001). Details are shown in Figure 2 and Supplementary Figure 1-2.
Table 2. Age-specific hazard ratios for the association between stroke age diagnose category and subsequent dementia in stroke population.
|
No. of dementia cases/Person-years
|
Incidence rate per1000 person-years (95%CI)
|
HR (95% CI)
|
P value
|
Prevalence Stroke Cohort
|
|
|
|
|
Per 10-y increase
|
378/120672.52
|
3.13 (2.83, 3.46)
|
4.37(3.65, 5.23)
|
< 0.001
|
18-48 y (n = 1604)
|
64/43818.479
|
1.46 (1.14, 1.87)
|
1 [Reference]
|
|
49-55 y (n = 1520)
|
69/30594.408
|
2.26 (1.78, 2.86)
|
5.15(2.93, 9.04)
|
< 0.001
|
56-61 y (n = 1613)
|
126/28131.833
|
4.48 (3.76, 5.33)
|
18.85(10.57, 33.61)
|
< 0.001
|
>61 y (n = 1263)
|
119/18127.795
|
6.56 (5.49, 7.86)
|
42.60(22.93, 79.16)
|
< 0.001
|
Incident Stroke Cohort
|
|
|
|
|
Per 10-y increase
|
425/36717.742
|
11.57 (10.53 12.73)
|
3.41(2.84, 4.10)
|
< 0.001
|
18-65 y (n = 1973)
|
42/13097.636
|
3.21 (2.37, 4.34)
|
1 [Reference]
|
|
66-71 y (n = 2065)
|
152/12189.022
|
12.47 (10.64, 14.62)
|
3.90(2.52,6.01)
|
< 0.001
|
72-76 y (n = 2050)
|
147/8104.0795
|
18.14 (15.43, 21.32)
|
7.54(4.85, 11.74)
|
< 0.001
|
>77 y (n = 1624)
|
84/3327.0055
|
25.25 (20.39, 31.27)
|
12.16(7.45, 19.85)
|
< 0.001
|
Abbreviations: HR, Hazard ratios; CI, Confidential Intervals. Adjusted for sex, race, education level, the Townsend deprivation levels, hypertension, diabetes, hyperlipidemia, heart failure, atrial fibrillation, long-standing illness, disability or infirmity, depressive symptom, family history of dementia (whether parents has dementia), and family history of CVD (whether parents has CVD).
As shown in Table 2 and Figure 3, the age at stroke diagnosis was significantly associated with an increased risk of developing dementia among participants with stroke, which indicated that participants with older age at diagnosis of stroke had higher risks of developing all-cause dementia in the prevalence stroke cohort (adjusted HR per 10-year decrease, 4.37; 95% CI, 3.65-5.23, P < .001), and in the incident stroke cohort (adjusted HR per 10-year decrease, 3.41; 95% CI, 2.84-4.10, P < .001). Survivors with intracerebral hemorrhage has the highest HRs, followed by ischemic stroke, other stroke types, and non-stroke.
Table 3 illustrates the risks of all-cause dementia relative to the age at stroke diagnosis. In the prevalent stroke cohort, individuals who had a stroke diagnosed before age 48 showed the highest hazard ratio (HR) for developing all-cause dementia (adjusted HR, 2.88; 95% CI: 1.77-4.68, P < .001). This was followed by those diagnosed at ages 49 to 55 years (adjusted HR, 2.12; 95% CI: 1.42-3.16, P < .001), 56 to 61 years (adjusted HR, 2.09; 95% CI: 1.57-2.78, P < .001), and 62 years or older (adjusted HR, 1.43; 95% CI: 1.08-1.89, P = 0.013). In the incident stroke cohort, the highest HR was seen in individuals with stroke diagnosed before age 65 (adjusted HR, 3.71; 95% CI: 2.08-6.63, P < .001), followed by ages 66 to 71 years (adjusted HR, 3.69; 95% CI: 2.77-4.93, P < .001), 72 to 76 years (adjusted HR, 3.23; 95% CI: 2.47-4.21, P < .001), and 77 years or older (adjusted HR, 2.43; 95% CI: 2.74-3.41, P < .001).
Table 3. Association between age at stroke diagnosis and incidence dementia.
|
No. of dementia cases/
Person-years
|
Incidence rate per 1000
person-years (95%CI)
|
HR (95% CI)
|
P
|
SHR (95% CI)
|
P
|
Prevalence Cohort
|
|
|
|
|
|
|
18-48 y (n = 9624)
|
|
|
|
|
|
|
Matched Non-Stroke
|
76/224943.93
|
0.34 (0.27, 0.42)
|
1 [Reference]
|
|
|
|
Stroke
|
64/43818.479
|
1.46 (1.43, 1.87)
|
2.88(1.77, 4.68)
|
< 0.001
|
2.60(1.63, 4.14)
|
< 0.001
|
49-55 y (n = 9120)
|
|
|
|
|
|
|
Matched Non-Stroke
|
130/159579.63
|
0.81 (0.69, 0.97)
|
1 [Reference]
|
|
|
|
Stroke
|
69/30594.408
|
2.26 (1.78, 2.86)
|
2.12(1.42, 3.16)
|
< 0.001
|
1.84(1.37, 2.46)
|
< 0.001
|
56-61 y (n = 9678)
|
|
|
|
|
|
|
Matched Non-Stroke
|
273/149671.41
|
1.82 (1.62, 2.05)
|
1 [Reference]
|
|
|
|
Stroke
|
126/28131.833
|
4.48 (3.76, 5.33)
|
2.09(1.57, 2.78)
|
< 0.001
|
1.71(1.15, 2.66)
|
< 0.001
|
>62 y (n = 7576)
|
|
|
|
|
|
|
Matched Non-Stroke
|
332/97177.444
|
3.42 (3.07, 3.80)
|
1 [Reference]
|
|
|
|
Stroke
|
119/18127.795
|
6.56 (5.48, 7.86)
|
1.43(1.08, 1.89)
|
0.013
|
1.31(0.99, 1.74)
|
0.057
|
Incident Cohort
|
|
|
|
|
|
18-65 y (n = 11838)
|
|
|
|
|
|
|
Matched Non-Stroke
|
41/70853.644
|
0.58 (0.43, 0.79)
|
1 [Reference]
|
|
|
|
Stroke
|
42/13097.636
|
3.21 (2.37, 4.34)
|
3.71(2.08,6.63)
|
< 0.001
|
3.31(1.91, 5.74)
|
< 0.001
|
66-71 y (n = 12390)
|
|
|
|
|
|
|
Matched Non-Stroke
|
193/68829.315
|
2.80 (2.44, 3.23)
|
1 [Reference]
|
|
|
|
Stroke
|
152/12189.022
|
12.47 (10.64, 14.62)
|
3.69(2.77, 4.93)
|
< 0.001
|
3.08(2.30, 4.16)
|
< 0.001
|
72-76 y (n = 12300)
|
|
|
|
|
|
|
Matched Non-Stroke
|
252/44210.945
|
5.70 (5.04, 6.45)
|
1 [Reference]
|
|
|
|
Stroke
|
147/8104.0795
|
18.14 (15.43, 21.32)
|
3.23(2.47, 4.21)
|
< 0.001
|
2.80(2.14, 3.67)
|
< 0.001
|
>77 y (n = 9743)
|
|
|
|
|
|
|
Matched Non-Stroke
|
144/15984.129
|
9.01 (7.65, 10.61)
|
1 [Reference]
|
|
|
|
Stroke
|
84/3327.0055
|
25.25 (20.39, 31.27)
|
2.43(2.74, 3.41)
|
< 0.001
|
2.27(1.64, 3.13)
|
< 0.001
|
Abbreviations: HR, Hazard ratios; SHR, sub-distribution hazard ratio; CI, Confidential Intervals. Adjusted for sex, race, education level, the Townsend deprivation levels, hypertension, diabetes, hyperlipidemia, heart failure, atrial fibrillation, Long-standing illness, disability or infirmity, depressive symptom, family history of dementia (whether parents has dementia), and family history of CVD (whether parents has CVD).
In the incidence stroke cohort, we assessed the stability of results across different sexes, APOE ε4 carrier status, and blood types, shown in Figure 4. For females, the highest risk of developing all-cause dementia post-stroke was in those diagnosed before age 65 (adjusted HR, 7.14; 95% CI: 3.32-15.38, P < .001), followed by ages 66 to 71 years (adjusted HR, 4.98; 95% CI: 3.35-7.40, P < .001), 72 to 76 years (adjusted HR, 2.77; 95% CI: 1.96-3.92, P < .001), and 77 years or older (adjusted HR, 2.43; 95% CI: 1.58-3.73, P < .001). For males, although all HRs were above 1, no decreasing trend (i.e., similar to that for females) was observed (Supplemental Figure 3 and Table 1).
For APOE non-carrier participants, the highest risk of developing all-cause dementia post-stroke was observed in those diagnosed before the age of 65 (adjusted HR, 6.27; 95% CI: 3.49-11.27, P <. 001), followed by those aged 66 to 71 years (adjusted HR, 4.97; 95% CI: 3.62-6.84, P < .001), 72 to 76 years (adjusted HR, 4.46; 95% CI: 3.27-6.10, P < .001), and 77 years or older (adjusted HR, 2.37; 95% CI: 4.59-3.53, P < .001). For APOE carriers in the 18-65 age group, there was no statistically significant difference (adjusted HR, 2.21; 95% CI: 0.98-4.99, P =0.055). And the risk of developing dementia decreased with increasing stroke onset age. This trend is consistent with that of the APOE non-carrier group (Supplemental Table 2).
65 had the highest dementia risk (adjusted HR, 3.71; 95% CI: 2.08-6.63, P < .001), followed by diagnosis at ages 66 to 71 (adjusted HR, 3.69; 95% CI: 2.77-4.93, P < .001), 72 to 76 (adjusted HR, 3.23; 95% CI: 2.47-4.21, P < .001), and 77 or older (adjusted HR, 2.43; 95% CI: 2.74-3.41, P < .001). No such trend was evident for individuals with type O blood (Supplemental Figure 4 and Table 3).
In the prevalence cohort, stroke survivors who were elder than 70 and APOE carrier had the highest HRs, and in the incident cohort, stroke survivors who were elder than 55, APOE carrier and male had the highest HRs. Negative INTMs were observed for age at stroke diagnosis and APOE state in both prevalence and incident cohort. For participants aged >70 years with a history of stroke, the RERI was 7.37(4.91, 9.83), with an SI of 1.90(1.59, 2.26), indicating that the integration of stroke and its diagnose age had a greater dementia risk than the sum of the independent effect of the two factors. By calculating AP, we could know that 44% (95%CI: 36–53%) of the combined risk were due to an additive interaction. For survivors who suffered a stroke and aged >55 years, the RERI was 11.42(7.6, 15.23), with an SI and AP of 1.95(1.66, 2.28) and 47% (95%CI: 39–55%), respectively. In the prevalence cohort, APOE carrier with a stroke the RERI was 2.11(0.64, 3.57), with an SI and AP of 1.36(1.11, 1.67) and 24% (95%CI: 10–37%), respectively. While in the incident cohort the additive effect of APOE state and stroke was not statistically significant. Similar results could be drawn when considering the sex and stroke interaction.
Table 4. The effect of age at stroke diagnosis, sex, APOE state and their additive interaction on dementia.
|
|
HR (95% CI)
|
P value
|
INTM
|
P value
|
RERI (95% CI)
|
AP (95% CI)
|
SI (95% CI)
|
Prevalence Cohort
|
|
|
|
|
|
|
|
|
Age at date index
|
|
|
|
|
|
|
|
|
<=70
|
Matched Non-Stroke
|
1 [Reference]
|
|
|
|
|
|
|
Stroke
|
5.06(4.1, 6.25)
|
<.0001
|
|
|
|
|
|
>70
|
Matched Non-Stroke
|
5.17(4.32, 6.17)
|
<.0001
|
|
|
|
|
|
Stroke
|
16.6(13.65, 20.18)
|
<.0001
|
0.64(0.49, 0.82)
|
0.0006
|
7.37(4.91, 9.83)
|
0.44(0.36, 0.53)
|
1.90(1.59, 2.26)
|
APOE state
|
|
|
|
|
|
|
|
|
Non-carrier
|
Matched Non-Stroke
|
1 [Reference]
|
|
|
|
|
|
|
Stroke
|
4.72(3.98, 5.61)
|
<.0001
|
|
|
|
|
|
Carrier
|
Matched Non-Stroke
|
3.08(2.63, 3.6)
|
<.0001
|
|
|
|
|
|
Stroke
|
8.91(7.39, 10.73)
|
<.0001
|
0.61(0.48, 0.79)
|
0.0001
|
2.11(0.64, 3.57)
|
0.24(0.1, 0.37)
|
1.36(1.11, 1.67)
|
Sex
|
|
|
|
|
|
|
|
|
Female
|
Matched Non-Stroke
|
|
|
|
|
|
|
|
Stroke
|
3.69(3.03, 4.49)
|
<.0001
|
|
|
|
|
|
Male
|
Matched Non-Stroke
|
1.15(0.98, 1.35)
|
0.0924
|
|
|
|
|
|
Stroke
|
4.28(3.59, 5.09)
|
<.0001
|
1.01(0.79, 1.30)
|
0.9336
|
0.44(-0.34, 1.22)
|
0.1(-0.07, 0.28)
|
1.16(0.89, 1.5)
|
Incident Cohort
|
|
|
|
|
|
|
|
|
Age at date index
|
|
|
|
|
|
|
|
|
<=55
|
Matched Non-Stroke
|
1 [Reference]
|
|
|
|
|
|
|
Stroke
|
3.43(2.75, 4.26)
|
<.0001
|
|
|
|
|
|
>55
|
Matched Non-Stroke
|
10.6(8.79, 12.79)
|
<.0001
|
|
|
|
|
|
Stroke
|
24.44(19.77, 30.22)
|
<.0001
|
0.67(0.52, 0.88)
|
0.0032
|
11.42(7.6, 15.23)
|
0.47(0.39, 0.55)
|
1.95(1.66, 2.28)
|
APOE state
|
|
|
|
|
|
|
|
|
Non-carrier
|
Matched Non-Stroke
|
1 [Reference]
|
|
|
|
|
|
|
Stroke
|
3.25(2.74, 3.86)
|
<.0001
|
|
|
|
|
|
Carrier
|
Matched Non-Stroke
|
3.3(2.88, 3.8)
|
<.0001
|
|
|
|
|
|
Stroke
|
6.34(5.27, 7.62)
|
<.0001
|
0.59(0.46, 0.75)
|
<.0001
|
0.78(-0.32, 1.88)
|
0.12(-0.04, 0.28)
|
1.17(0.94, 1.45)
|
Sex
|
|
|
|
|
|
|
|
|
Female
|
Matched Non-Stroke
|
1 [Reference]
|
|
|
|
|
|
|
Stroke
|
2.47(2.02, 3)
|
<.0001
|
|
|
|
|
|
Male
|
Matched Non-Stroke
|
1.31(1.14, 1.51)
|
0.0002
|
|
|
|
|
|
Stroke
|
3.31(2.79, 3.92)
|
<.0001
|
1.02(0.80, 1.32)
|
0.8518
|
0.53(-0.07, 1.13)
|
0.16(-0.01, 0.33)
|
1.3(0.96, 1.77)
|
Abbreviations: HR, Hazard ratios; SHR, sub-distribution hazard ratio; CI, Confidential Intervals; INTM, Multiplicative Interaction. The multivariate regression models were adjusted for race, education level, the Townsend deprivation levels, hypertension, diabetes, hyperlipidemia, heart failure, atrial fibrillation, Long-standing illness, disability or infirmity, depressive symptom, family history of dementia (whether parents has dementia), and family history of CVD (whether parents has CVD). RERI and AP larger than 0 and SI large than 1 indicate a positive additive interaction effect.
The inflammation indicators and NFLA scores played significant mediating roles in the association between stroke event and incident dementia, with mediation proportions of 5.8 % and 5.4 %, respectively. The proportion of mediation was higher in the <=55 diagnose age group (6.9%:2.7%-19.9%) than the >55 age group (5.8%:2.1%-15.1%). Figure 5 summarizes the potential mediating effect of inflammation between stroke event and incident dementia.
Our sensitivity analyses produced results that were consistent with the main analyses, supporting stability in the findings. These results are detailed in the supplementary material (Supplemental Table 4-6).