Introduction

Bias is a systematic error in the way different groups are handled in a study leading to inaccuracy.

A trial evaluating a treatment is less valid if both the investigator and the patient know what treatment they’re getting. However, it is not always possible to blind a patient to the treatment they get or the physician to the treatment they give.

Causes of bias:

Selection

Is selection systematically different between one group and the other?

Measurement/Observation

Were observations or measurements recorded differently between the groups?

Follow-up

Were more patients lost to follow up in one arm of the study?

Cross-over

Did patients get moved across from the intervention to control arm or vice-versa?

Blinding

The risk of bias can be reduced through blinding. Blinding is a quality of study design whereby study participants, those involved with their management and those collecting and analysing clinical data may be prevented from knowing which treatment or intervention has been provided. Blinding may be particularly important where the outcome of interest is subjective (eg. pain) rather than objective (eg. death). However, blinding is not always possible to achieve in a study.