Table 1

Grade assessment of certainty of estimate of effect size and summary of findings

GRADE assessment of certainty
Absolute risk
OutcomeRisk of biasInconsistencyIndirectnessImprecisionPublication biasNo of studiesNo of participantsRelative difference (95% CI)Medium riskExposed risk (95% CI)Certainty (reasons for downgrading)
All-cause mortalitySerious limitations (uncontrolled confounding)Serious inconsistency (heterogeneity)No serious indirectnessNo serious imprecisionUndetected3391 429OR=1.00 (0.94 to 1.06)323.5 age-standardised mortality rate per 100 000 person-years5323.5 (304.1 to 342.8) age-standardised mortality rate per 100 000 person-yearsVery low (risk of bias; inconsistency)
Self-rated healthSerious limitations (incomplete follow-up)Serious inconsistency (heterogeneity)No serious indirectness (borderline)No serious imprecisionUndetected545 904OR=1.26 (1.01 to 1.57)25.0%*29.6% (25.2% to 34.4%)Very low (risk of bias; inconsistency)
Mental health symptomsSerious limitations (uncontrolled confounding and incomplete follow-up)No serious inconsistency (borderline)No serious indirectnessNo serious imprecisionUndetected1155 121OR=1.44 (1.23 to 1.70)18.5†24.6% (21.8% to 27.8%)Very low (risk of bias)