Table 3

Quality assessment of the included studies
Selection bias Allocation bias Confounders Data collection methods Withdrawals and dropouts Comment on the analysis
Kohen, Brooks-Gun, Leventhal, & Hertzman, 2002 S M S S NA -No power calculation
-Results unambiguously reported
-Handling of missing data not reported
-Inappropriate statistical methods: multilevel model is required to answer research question
-Risk of clustering of children within the same families
Xue, Leventhal & Brooks-Gunn et al., 2005 M M S S NA -ICC calculated
-No power calculation
-Results are unambiguously reported
-Appropriate handling of missing data
-Appropriate statistical methods with remarks:
-Analysis of level 1 and 2 variances reported
Caughy & O’Campo, 2006 M M M S NA -ICC calculated
-No power calculation
-Results unambiguously reported
-Appropriate statistical methods with remarks
-No level 1 predicators entered in the model
-Analysis of level 2 variance not reported
-Small N
-Handling of missing data not reported
Drukker, Kaplan, Schneiders, Feron, & van Os, 2006 M M S S M -ICC calculated
-No power calculation
-Results are partially unambiguously reported
-Appropriate statistical methods with remarks
-Possible selective drop-out
-Analysis of level 1 and level 2 variance not reported
-Small N
-Handling of missing data not reported
Kohen, Leventhal, Dahinten, & McIntosh, 2008 S M S S NA -No power calculation
-Results are unambiguously reported
-Appropriate statistical methods with remarks
-Multilevel SEM would be more suited
-Not possible to assess level 1 and level 2 variance, calculate changes in r2, etc.
-Appropriate handling of missing data
Caughy, Nettles & O'Campo, 2008 W M S W NA -ICC not calculated
-No power calculation
-Results are unambiguously reported Appropriate statistical methods with remarks:
-Small N
-No analysis of level 1 and 2 variances reported
Karriker-Jaffe, Foshee, Ennett, & Suchindran, 2009 M M S S W -ICC calculated
-No power calculation
-Results are unambiguously reported
-Appropriate statistical methods with remarks
-No analysis of level 1/level 2 variance
-Appropriate handling of missing data
Odgers et al., 2009 M M S S NA -No power calculation
-Results are unambiguously reported
-Appropriate handling of missing data
-Appropriate statistical methods

W=weak; M=moderate; S=strong.

NA = not applicable; ICC = intraclass correlation coefficient; N = sample size; SEM = Structural Equation Modelling.

Vyncke et al.

Vyncke et al. BMC Public Health 2013 13:65   doi:10.1186/1471-2458-13-65

Open Data