men and women, five reported outcomesseparately.
TheTOHPstudyincludedtwodiffer-ent cohorts (I and II)
and the study by He and cow-orkersprovidedseparatefindingsformenandwomenor, alternatively, for normal weight and overweight participants.
Overall, data on the relation betweensalt consumption and stroke were available from 14cohorts and on the relation between salt intake andcardiovascular disease from 14 cohorts.The overall study quality, evaluated by the DownsandBlackscore,averaged15.5(range12-18)onascaleof 19 (table 1).
Salt intake and risk of stroke
Table 2 provides data on the relation between salt intake and risk of stroke in each of the 14 cohortsincluded in our study. Figure 1 shows the results of the pooled analysis. In the pooled analysis, higher salt intake was associated with greater risk of stroke (rela-tive risk 1.23, 95% confidence interval 1.06 to 1.43;P
61%).Thefunnelplotdidnotshow asymmetry, thus excluding publication bias (Egger
0.26; see appendix on bmj.com). As shown infigure 1 for the individual cohorts included in the ana-lysis, we found a trend towards a direct associationbetween salt intake and risk of stroke in nine cohorts,which was significant in four. We observed a non-sig-nificant inverse trend in three cohorts.Sensitivity analysis showed that the pooled estimateof the effect of salt intake on risk of stroke did not varysubstantially with the exclusion of any one study; inparticular, the exclusion of the study by Umesawa et al,
which accounted for about 40% of all participantsin the meta-analysis and nearly 20% of all strokes,resulted in a pooled relative risk of 1.19 (1.03 to 1.39),P
Salt intake and risk of cardiovascular disease
Table 2 provides data on the association between salt intake and the risk of cardiovascular disease in 14cohorts. In the pooled analysis, there was an associa-tion between higher salt intake and risk of cardio-vascular disease (1.14, 0.99 to 1.32; P
0.07) (fig 2).The heterogeneity between studies was significant (P<0.01; I
80%), but the funnel plot did not show asymmetry, thus excluding publication bias (Egger
0.39; see appendix on bmj.com). The
Characteristics of prospective studies included in meta-analysis of studies on salt intake and stroke and cardiovascular disease (CVD)
Study Age (years) Sex No of people Outcome(s) Outcome assessmentSodium intakeassessmentStudy qualityscore
45-68 Men 7895 Total stroke Physicalexamination for residual ofstroke atbaseline,2 and 6 yearfollow-up; surveillance of hospitaldischargesand death certificates reviewedbyneurologist24h dietary recall 14Hu, 1992, Taiwan
36 Menand women 8562 Total stroke Casefindingthroughlocalhospitalreferralsandstudynurses. Certificationby computedtomographyHousehold surveyquestionnaire12Alderman, 1995, US,occupational
52; 54 Men; women 1900; 1037 Total CVD,totalstrokeReview ofhospitalchartsand death certificates (ICD-9): CVD I410,I430-I434, I436-I438; stroke asabovewithoutI41024h urinecollection12Tunstall-Pedoe,1997, Scotland
40-59 Men; women 5754; 5875 Total CVD Case notes requested for all hospital episodes of myocardialinfarctionandotheremergencyadmissionfor coronary heart disease,then extracted and codedaccordingto MONICA project criteria24h urinecollection15He, 1999, US,NHANESI
25-74 Men;women;non-overweight;overweight3686;5799;6797; 2688CVD death,totalstroke,stroke deathMortality basedondeath certificatereports.Incidentstroke based on death certificatereports inwhichunderlyingcause ofdeath was recorded withICD-9code(430-434.9,436or437.0-437.1)oroneormorehospitalstayswithdischarges withone ofthesecodes24h dietary recall 17Tuomilehto,2001,Finland
25-64 Men; women 1173; 1263 CVD death.totalstrokeNationalhospital discharge register ICD-8 and ICD-9,430-438, and 390-44824h urinecollection18Nagata, 2004, Japan
35 Men; women 13355; 15724 Stroke death Nationalvitalstatistics ICD-9 430-448 FFQ 18Cohen,2006, US,NHANESII
30-74 Menand women 7154 CVD death,stroke deathMortality basedondeath certificatereports ICD-9430-43824h dietary recall 18Geleijnse, 2007,Netherlands
55 Menand women 1448 CVD death,totalstrokeGPs registries (ICD-10): I20-I25,I46,I49,I50,I60-I67,I70-I74 and R96;I60-I67FFQ and overnighturine sodium15Cook, 2007, US,TOHPI,USA,TOHPII
30-54; 30-54 Menand women;men and women542;1873 Total CVD Notification ofnon-fataloutcomes inpost-trialsurveillance,reviewbyphysicianplusNationalDeathIndex24h urinecollection12Larsson, 2008,Finland
50-69 Men(smokers) 26,556 Total stroke Dischargediagnosesand deathcertificates (ICD-8,9,and 10)FFQ 15Umesawa, 2008, Japan
40-79 Men; women 23119; 35611 CVD death,stroke deathNationalVital Statistics ICD-9 FFQ 4
3 daydietary records18Cohen,2008, US,NHANESIII
30 Menand women 8699 Total CVD Vital statusand cause ofdeath (ICD-9 and ICD-10) 24h dietary recall 18
food frequency questionnaire.
ONLINE FIRST | bmj.com page 3 of 9