Table 2.

Risk of bias analysis: neonics and human health.

Reference	Comparison groups appropriate	Confounding/modifying (design/analysis)	Identical experimental conditions	Blinding subjects and researchers	Outcome data complete	Exposure characterization confidence	Outcome assessment confidence	All measured outcomes reported	Other validity/statistical issues
Source: acute studies
Elfman et al. 2009	++	–	+	++	++	–	–	++	–
Forrester 2014	NA	––	NA	NA	NA	––	––	+	–
Mohamed et al. 2009	NA	+	NA	NA	–	–	+	++	–
Phua et al. 2009	NA	++	NA	NA	NA	–	+	++	–
Source: chronic studies
Carmichael et al. 2014	++	–	NA	NA	–	–	–	+	–
Keil et al. 2014	+	–	NA	NA	–	+	–	–	–
Marfo et al. 2015	++	–	NA	NA	–	+	+	+	–
Yang et al. 2014	++	–	NA	NA	–	–	–	+	–
Note: ++, definitely low risk of bias; +, probably low risk of bias; ––, definitely high risk of bias; –, probably high risk of bias; NA, not applicable. Overall rating: Tier 3, probably high risk of bias, low to moderate confidence. Under OHAT, all chronic studies would be dropped as too weak for inclusion, as would the most recent (2014) acute study. All were retained to enable this review.