Missing biases in your research design will lead you to see the wrong or limited results.

Missing biases in your research design will lead you to see the wrong or limited results.

Assessing Risk of Bias

  • Bias: systematic error leading to underestimation or overestimation of the true intervention effect. Not to be confused with imprecision as bias is often a systematic error in design, meaning multiple replications of the same study would reach the same ‘wrong’ conclusion.

 

  • Selection bias: systematic differences between baseline characteristics of the groups that are compared (prevent this with randomization and allocation concealment).

 

  • Performance bias: systematic differences between groups in care provided or exposure to factors other than interventions of interest (prevent this by blinding or masking participants and personnel).

 

  • Detection bias: differences between groups in how outcomes are determined (prevent this by blinding or masking of outcome assessors).

 

  • Attrition bias: differences between groups in withdrawals from the study (prevent this by minimizing lost to follow up, and analyzing all available data without exclusions).

 

  • Reporting bias: differences between reported and unreported findings (prevent by reporting all outcomes of interest even if not statistically significant or positive).