Skip to main content

Table 4 Risk of bias assessments

From: The effectiveness of smoking cessation interventions for socio-economically disadvantaged women: a systematic review and meta-analysis

Studies Bias arising from the randomisation process Bias due to deviations from intended interventions Bias due to missing outcome data Bias in measurement of the outcome Bias in selection of the reported result Overall bias
Andrews et al. 2016 [48] Low risk Low risk Low risk Some concerns: no information on blinding of outcome assessors Some concerns: no information on whether statistical analysis was pre-planned Some concerns
Bernstein et al. 2015 [47] Some concerns: Some differences in baseline characteristics between groups Low risk Low risk Low risk Low risk Some concerns
Collins et al. 2019 [49] Low risk Low risk Retention rate relatively low   High risk: authors changed statistical analysis plan due to low retention ratea High risk
Curry et al. 2003 [50] Low risk Low risk Low risk High risk: no biochemical validation in primary outcome Some concerns: no information on whether statistical analysis was pre-planned High risk
Etter et al. 2016 [51] Some concerns: Some differences in characteristics between groups, but accounted for in analysis Low risk Low risk Low risk Some concerns: No information on whether statistical analysis was pre-planned Some concerns
Gilbert et al. 2017 [46] Low risk Low risk Low risk Low risk Low risk Low risk
Glasgow et al. 2000 [52] Low risk Low risk Low risk Some concerns: no information on blinding of outcome assessors Some concerns: no information on whether statistical analysis was pre-planned Some concerns
Haas et al. 2015 [53] Low risk Low risk Some concerns: Retention rate relatively low and higher assessment rate in control group High risk: no biochemical validation in primary outcome Some concerns: no information on whether statistical analysis was pre-planned High risk
Manfredi et al. 2004 [54] Some concerns: differences in racial characteristics between groups Low risk Some concerns: Retention rates low, but equally low in both groups High risk: no biochemical validation in primary outcome Some concerns: no information on whether statistical analysis was pre-planned High risk
Solomon et al. 2000 [55] Low risk Low risk Low risk Some concerns: no information on blinding of outcome assessors Some concerns: no information on whether statistical analysis was pre-planned Some concerns
Solomon et al. 2005 [56] Low risk Low risk Low risk High risk: no biochemical validation in primary outcome Some concerns: no information on whether statistical analysis was pre-planned High risk
  1. aThe authors explain the change in analyses as “We planned to test the interaction between treatment and other smokers in home at p<0.05 when initial models demonstrated main effects of both variables. Although each of the predictors and outcome variables contained small numbers of missing values, an analysis of complete data only would have reduced our sample by about one third. To retain our sample and avoid bias arising from missing data, we used multiple imputation methods, which also estimate SEs that incorporate the uncertainty due to imputation