Table 3

Quality assessment based on evaluation of bias
Author, year Selection bias Exposure bias Outcome assessment bias Confounding factor bias Analytical bias Attrition bias Overall likelihood of bias based mainly on selection and confounding
O’Leary et al.2009 [30] Minimal (sample selected from a general population rather than a select group) Low (indirect assessment (postal survey, mailed question)) Minimal (direct question to mother) Minimal (assessed for common confounders) Minimal (analyses appropriate for type of sample (if matched)) Moderate (11 to 20% attrition but reasons for loss to follow-up not explained) Minimal
Faden and Graubard 2000 [29] Minimal (sample selected from general population, Eligibility Criteria explained) Low (indirect assessment (postal survey, mailed question)) Minimal (direct question to mother) High (not assessed for confounders) Low (analyses not accounting for common statistical adjustment and sample size calculation not performed but all eligible patients studied) Moderate (11 to 20% but reasons for loss to follow-up not explained) High
Greene et al. 1990 [31] Moderate (sample selection ambiguous but may be representative) Minimal (direct questioning (interview) or completion of survey by mother at the time of exposure or close to time of exposure) Minimal (direct question to mother) Minimal (assessed for common confounders) Low (analyses not accounting for common statistical adjustment, power calculation performed) High (>20% attrition but reasons for loss to follow-up not explained) Moderate

O’Keeffe et al.

O’Keeffe et al. Systematic Reviews 2014 3:1   doi:10.1186/2046-4053-3-1

Open Data