Premium
Novel insights into risk of dementia after traumatic brain injury: A systematic review, meta‐analysis, and heterogeneity analysis
Author(s) -
Gardner Raquel C.,
Bahorik Amber L,
Mangal Paul,
Allen Isabel Elaine,
Yaffe Kristine
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.038676
Subject(s) - meta analysis , publication bias , funnel plot , dementia , medicine , confidence interval , study heterogeneity , cohort study , relative risk , traumatic brain injury , demography , psychiatry , disease , sociology
Background Traumatic brain injury (TBI) is an established risk factor for dementia. Risk estimates have varied substantially across studies with recent studies reporting higher risk in veterans, men, or at the extremes of age. We performed the largest and most comprehensive systematic review of risk of post‐TBI dementia with the aim of specifically investigating contributors to heterogeneity including age, sex, and veteran status. Method We performed a systematic review and meta‐analysis of all‐cause dementia after all‐severity TBI (search window 1/1990‐1/2019). We identified observational studies reporting age‐adjusted risk for all‐cause dementia after TBI among individuals with average age ≥40 years. Data were pooled using random‐effects models. Between study variability was assessed using the I 2 index. We further evaluated sources of heterogeneity according to covariates including mean age, sex distribution (majority male vs. majority female), veteran status, design (case‐control vs. cohort), outcome/exposure ascertainment (ICD‐9 vs. other) and publication year using sub‐groups and meta‐regression analysis. Funnel plot inspection and Egger and Begg statistics were used to evaluate publication bias. Result Data from 41 studies representing N=7,736,173 individuals were included in the analysis. Overall pooled risk ratio (RR) for dementia after TBI was 1.71 (95% confidence interval [CI] 1.47,1.98) with no evidence of publication bias. There was substantial heterogeneity (I 2 = 98.6%; Q test p<0.001). Removal of each study in turn did not reduce heterogeneity. No significant differences were observed in risk estimates for dementia after TBI according to veteran status, study design, outcome/exposure ascertainment, or publication year (all p’s >0.09). However, risk estimates were significantly higher in studies with lower mean age (mean age <50y RR 2.52 [1.80,3.53]; 50‐70y RR 1.73 [1.34,2.24]; >70y RR 1.41 [1.17,1.71]; p=0.01; see Figure) and in studies that were majority male (majority male RR 2.16 [1.74‐2.69]; majority female RR 1.44 [1.22,1.70]; p=0.01). Mean age and sex distribution accounted for 27.5% and 27.9%, respectively, of the observed heterogeneity. Conclusion TBI is associated with a 71% increased risk for dementia. While risk was not modified by veteran status, risk was significantly higher in studies with younger mean age and majority male populations. These findings may inform targeted TBI prevention efforts.