Blood Pressure And The Risk Of Dementia: A Systematic Review And Dose-response Meta-analysis Of Prospective Studies

Background:As a common modifiable vascular risk factor,hypertension is considered to play an important role in the development of dementia.Although the understanding of the relationship between cardiocirculatory dysfunction and brain health has improved significantly over the last several decades,it is still unclear whether hypertension constitutes a potentially treatable risk factor for dementia.A number of epidemiological studies have revealed a controversial correlation between blood pressure（BP）and dementia’s risk,hindering the clinical translation for preventative practice.Objective:The present study is to meta-analyze the dose-response relationship between dementia and its subtypes and BP and shed more light on the optimal BP management strategy for dementia prevention.Methods:PubMed,EMBASE,and Ovid（from inception to May 1,2017）were systematically searched in accordance with the recommendations in PRISMA 2009guidelines and Meta-analysis Of Observational Studies in Epidemiology（MOOSE）Group.Keywords include“hypertension”,“blood pressure”,“pressure”,“dementia”,“Alzheimer”,“cohort”,and“prospective”.No restrictions were imposed.In order to avoid the loss of relevant studies,bibliographies of the relevant meta-analyses or reviews were hand-searched.Strict criteria were formulated before literature screening and data extraction.We only included one with a larger sample size or longer follow-up when several articles are based on the same cohort.The DerSimonian and Laird random effects model was selected to calculate a summary estimate and its 95%confidence interval（CI）.First,we examined the association between high BP（high systolic blood pressure（SBP）and high blood pressure（DBP））and dementia（all dementia and Alzheimer’s disease（AD））risk on the basis of the effect estimates and their 95%CIs in each included study.The I² statistic and Q test were utilized to assess the heterogeneity.We select p<0.05 as the statistically significant level and set I² value ranges to indicate different heterogeneity（<30%indicates low heterogeneity;30-60%indicates possibly moderate heterogeneity;>60%indicates possibly high heterogeneity）.Moreover,we performed a sensitivity analysis by removing each individual study from the pooled analysis to evaluate the influence of each included study.The Newcastle-Ottawa Quality Assessment Scale（NOS）,which has been partially validated for evaluating the quality of observational studies in a meta-analysis,was used to rate the quality of studies included.Univariate meta-regression analyses by the age at BP measurement,population gender,and NOS score were conducted to investigate the potential sources of heterogeneity in qualified group（the number of studies included>10）.In the primary analyses,stratified analyses based on specific BP levels（per 10 mmHg increment）and population characteristics（age at BP measurement,gender,hypertensive treatment status and APOE-4 genotype）were conducted for all dementia,AD and vascular dementia（VD）.Subsequently,a dose-response analysis based on SBP and DBP was conducted using generalized least squares regression（the two-stage GLST in StataSE software）as proposed by Greenland and Longnecker（called GL method）.In the primary stage,we examined potential associations between dementia and BP using study-specific restricted cubic spline models with four knots at fixed percentiles（5,35,65,and 95%）of the exposure distribution.In the next stage,the variance/covariance matrix and the study-specific estimates（standard error）were estimated.By testing the joint effect of the spline transformations,we investigated the holistic significance of the curve.The P value for nonlinearity was calculated by testing the null hypothesis that the coefficient of the second spline is equal to zero.Results:A total of 38473 papers were yielded after deduplication.A total of 38427 papers were excluded after reviewing the titles and abstracts with 46 potentially eligible papers left.As necessary supplements,three potential articles were identified in applicable meta-analyses and reviews.After full-text screening,27 were excluded.One article was further excluded for ineligible data.Finally,21 articles were included in the primary analysis and five articles in the dose-response analysis（4 for all dementia and 3 for AD）.Twenty-three studies from 17 articles with 830631 participants and 4384 cases for all dementia,17 studies from 12 articles with 1707445 participants and 3481 cases for AD and 11 studies from 8 articles with 1693690 participants and 1067 cases for VD were included.We identified that the associations between BP and dementia varied with population characteristics.In the dose-response analysis,five studies（three in the age group of 62 to 82 and three in the age group of 70 to 86.5）were used to reveal the relationships between SBP and all dementia risk.And four studies（population aged from65 to 86.5）were used to reveal the relationships between DBP and AD risk.A nonlinear relationship was found between SBP and all dementia risk in population aged≥65 years(p_nonlinearity<0.05).SBP between 110 and 120mmHg played a protective role in population aged 62 to 82（p for heterogeneity=0.0717;p for model=0.0168;p for nonlinearity=0.0068）,while SBP above 162 mmHg would significantly increase the risk in those aged70 to 86.5（p for heterogeneity=0.2153;p for model=0.0216;p for nonlinearity=0.0352）.It is also noteworthy that there is a linear association between DBP and AD risk in the population aged≥65 years（p for heterogeneity=0.0704;p for model=0.0189;p for nonlinearity=0.2549）,such that the risk decreased by 3%for per 10 mmHg increase in a specific range of DBP（RR:0.97;95%CI 0.94-0.99）.Conclusions:These findings indicate that BP management strategy for dementia prevention might be tailored according to specific population characteristics.