Table 9.

Comparison of approaches for assessing the quality of a body of evidence.

Approach	Study types assessed	Ratings for expected heterogeneity	Baseline quality of evidence	Downgrade domains	Downgrade ratings	Upgrade domains	Upgrade ratings	Ratings of quality of evidence (see also Table 8)
GRADE approach (Guyatt et al. 2011)	Studies of the effect of an intervention on a health outcome	–	- Randomized studies: High quality - Non-randomized studies: Low quality	Risk of bias Indirectness Inconsistency Imprecision Publication bias	−1 for a serious concern −2 for a very serious concern	Dose-response Strength of effect Residual confounding increases confidence in effect estimate	+1 for evidence of dose–response and for a large effect +2 for evidence for a very large effect	1.High quality 2.Moderate quality 3.Low quality 4.Very low quality
Navigation Guide approach (Lam et al., 2016a, Lam, 2016b)	Studies of the effects/harms and their severity/ probability of exposure to an environmental or occupational risk factor on a health outcome	–	Moderate quality for human observational studies	Risk of bias Indirectness Inconsistency Imprecision Publication bias	−1 for a serious concern −2 for a very serious concern	Dose-response Strength of effect Residual confounding increases confidence in effect estimate	+1 for evidence of dose–response and for a large effect +2 for evidence for a very large effect	1.High quality 2.Moderate quality 3.Low quality
OHAT approach (Office of Health Assessment and Translation, 2019)	Studies of the toxicity of exposure to environmental and occupational risk factors on health outcomes	–	- Four key featuresa of study design fulfilled: High confidence - Three features: Moderate confidence - Two features: Low confidence - One or no feature: Very low confidence	Risk of bias across studies Unexplained inconsistency Directness and applicability Imprecision Publication bias	−1 for a serious concern −2 for a very serious concern	1.Large magnitude of association or effect 2.Dose response 3.Residual confounding or other related factors that would increase confidence in the estimated effect 4.Cross-species/ population/ study consistency 5.Other		1.High confidence 2.Moderate confidence 3.Low confidence 4.Very low confidence
QoE-SPEO approach (presented in Appendix A in Supplementary data)	Studies of the prevalence of exposure to an occupational risk factor	2. No or only minor expected heterogeneity Low expected heterogeneity Medium expected heterogeneity High expected heterogeneity In Step 1 of QoE-SPEO	High quality Not a separate step in QoE-SPEO	Risk of bias Indirectness Inconsistency b Imprecision b Publication bias In Step 2 of QoE-SPEO	−1 for a serious concern −2 for a very serious concern In Step 2 of QoE-SPEO	- (no upgrading)	- (no upgrading)	1.High quality 2.Moderate quality 3.Low quality 4.Very low quality In Step 3 of QoE-SPEO

Footnotes: ^a The four features of study design assessed are: (i) controlled exposure; (ii) exposure prior to outcome; (iii) individual outcome data; and (iv) comparison group used. ^b Different sets of criteria depending on the judged level of expected heterogeneity of the prevalence of exposure to the occupational risk factor of interest.

GRADE – Grading of Recommendations Assessment, Development and Evaluation; OHAT – Office of Health Assessment and Translation; QoE-SPEO – Quality of Evidence in Studies estimating Prevalence of Exposure to Occupational risk factors.