Table 5.
Type of bias | Criteria | ||
---|---|---|---|
Low | High | Unclear | |
Selection bias: random sequence Generation |
Random-numbers table or a computer software program |
Tossing a coin, throwing dice or basing randomization on a date, name, etc. |
Methods not described |
Selection bias: allocation concealment |
Concealment of allocation means strict implementation of a randomized sequence without foreknowledge of the next assignment. Strategies that include central randomization by computer or phone and sequentially numbered envelopes that are both sealed and opaque are low risk. |
Strategies with a reasonable chance of being anticipated. |
Inadequate description of the envelope process, which is often the case, should be marked as “unclear risk of bias.” |
Performance bias |
If methods to mask participants and physicians were described and these methods had a reasonable chance of being effective. |
If treatment assignment known by patient/physician. |
If the study did not discuss methods of masking the participant or physician. |
Detection bias | Methods to mask the outcome assessor were described. | The outcome assessor was not-masked. |
Methods to mask the outcome assessor were not described. |
Incomplete outcome data/attrition bias |
Losses to follow-up are similar in number and in reason between intervention and control. An intention to treat analysis is performed. |
Losses to follow-up are more than 10% different between the intervention and control groups. |
If attrition is not reported or reasons for loss of attrition are not reported. |
Reporting bias | If the primary outcome(s) in the methods section is identical to those reported in the results section of the manuscript. If a protocol is available, we will also review for discrepancies between targeted outcomes and results. |
If the primary outcomes differ between the methods section or protocol and manuscript. |
If there is no reported primary outcome in a trial registry, the outcome will be considered “unclear.” |