TableĀ 1

Risk of bias in observational studies included in Roberts's meta-analysis according to ACROBAT checklist3

Mortality (Lipworth et al6)Mortality (De Vries et al7)CVD end point (Chan et al4)Myocardial Infarction (De Vries et al7)Stroke (De Vries et al7)Hypertension (Curhan et al8)Hypertension (Dedier et al9)Upper GI events (De Vries et al7)Acute kidney failure (De Vries et al7)Chronic kidney failure (Curhan et al8)Chronic kidney failure (Kurth et al10)Chronic kidney failure (Evans et al18)
Bias type
Bias due to confoundingCriticalCriticalCriticalCriticalCriticalLowLowCriticalCriticalSeriousLowSerious
Bias in selection of participants into the studyModerateLowLowLowLowModerateLowLowLowLowModerateSerious
Bias in measurement of interventionNAModerateLowModerateModerateLowLowModerateModerateLowLowModerate
Bias due to departure from intended interventionsNo informationSeriousSeriousSeriousSeriousModerateLowSeriousSeriousModerateSeriousModerate
Bias due to missing dataNo informationNo informationNo informationNo informationNo informationModerateLowNo informationNo informationLowNo informationLow
Bias in measurement of outcomesNo informationNo informationSeriousNo informationNo informationSeriousSeriousNo informationNo informationLowSeriousLow
Bias in selection of reported resultsLowModerateSeriousModerateModerateSeriousLowModerateModerateModerateLowModerate
OverallCriticalCriticalCriticalCriticalCriticalSeriousSeriousCriticalCriticalSeriousModerateSerious
  • ACROBAT, A Cochrane Risk Of Bias Assessment Tool; CVD, cardiovascular disease; GI, gastrointestinal.