TABLE 6

Risk of Bias of Included Studies: Observational Studies

First Author (Year)DesignaBiasConfounding
Armour (2009)26Retrospective cohortMediumLow
Berry (2011)23Retrospective cohortLowLow
Berry (2011)32Retrospective cohortMediumHigh
Cooley (2009)37Retrospective cohortMediumLow
Dosa (2001)33Prospective cohortMediumHigh
Feudtner (2009)35Retrospective cohortLowLow
Frei-Jones (2009)34Retrospective cohortLowLow
Gay (2011)31Retrospective cohortMediumMedium
Hain (2013)28Retrospective cohortHighHigh
Kun (2012)30Retrospective cohortLowLow
Lu (2012)24Cross-sectionalMediumMedium
Raphael (2011)39Prospective cohortLowLow
Todd (2006)36Cross-sectionalMediumMedium
  • a Instruments: Observational designs = Agency for Healthcare Research and Quality RTI Item Bank,17 Bias = Q1, 3, 7, 8, 9, 11; Confounding = Q6, 12, 13. Authors defined risk as follows: low risk if no affirmative answers, medium risk if 1 affirmative response, and high risk if >1 affirmative response.