Criteria for assessing the risk of bias (2023)

GENERATION OF RANDOM SEQUENCES

Selection bias (prejudiced assignment to interventions) due to insufficient generation of a random sequence.

Criteria for the assessment of a "low risk of bias".

The researchers describe a random component in the sequence generation process as:

• Refer to a table of random numbers;

• use of a computer random number generator;

• Coin toss;

• shuffle cards or envelopes;

• publication of data;

• Contest;

• minimization*.

*Minimization can be implemented without a random element and this is considered equivalent to random.

Criteria for judging “high risk” of bias.

The researchers describe a non-random component in the sequence generation process. Typically the description would imply a systematic approach rather than a random one, for example:

• Sequence generated by odd or even date of birth;

• String generated by a rule based on the entry's date (or day);

• A string generated by a rule based on the hospital or clinic's medical record number.

Other non-random approaches are much less common than the systematic approaches mentioned above and are more obvious. They usually involve an assessment or a method of non-randomly categorizing participants, for example:

• Allocation at the discretion of the doctor;

• Allocation based on participant preference;

• assignment based on the results of a laboratory test or series of tests;

• Allocation based on the availability of the intervention.

Criteria for assessing “unclear risk” of bias.

Insufficient information about the sequence generation process to allow for a 'low risk' or 'high risk' assessment.

HIDDEN TASK

Selection bias (allocation bias towards interventions) due to insufficient concealment of allocations prior to allocation.

Criteria for the assessment of a "low risk of bias".

Participants and investigators enrolling participants cannot predict attribution because one of the following methods, or an equivalent, was used to obfuscate attribution:

• Centralized assignment (including phone, web-based, and pharmacy-directed randomization);

• Consecutively numbered drug containers with identical appearance;

• Appropriately numbered, opaque and sealed envelopes.

Criteria for judging “high risk” of bias.

Participants or investigators enrolling participants can predict assignments and thus introduce selection bias, e.g. B. an assignment based on:

• Using an open schedule for random assignments (e.g. a list of random numbers);

• labeling envelopes were used without adequate safeguards (e.g., where the envelopes were unsealed, opaque, or sequentially numbered);

• toggle or rotation;

• Birth date;

• process registration number;

• Any other procedure that is not explicitly hidden.

Criteria for assessing “unclear risk” of bias.

Insufficient information to allow a “low risk” or “high risk” assessment. This is often the case when the method of obfuscation is not described, or is not described in enough detail to allow for a definitive assessment; B. When the use of association envelopes is described, but it is not yet clear whether the envelopes were consecutively numbered. , opaque and sealed.

BLINDING OF PARTICIPANTS AND STAFF

Performance bias due to knowledge of the interventions assigned by participants and staff during the study.

Criteria for the assessment of a "low risk of bias".

One of the following:

• No blinding or incomplete blinding, but the reviewers consider that the outcome is unlikely to be affected by lack of blinding;

• Blinding of study participants and key personnel was ensured and the blinding is unlikely to have been broken.

  (Video) Part 2: Assessing the risk of bias in measurement of the outcome: signalling questions 1-2

Criteria for judging “high risk” of bias.

One of the following:

• No blinding or incomplete blinding and the outcome is likely to be influenced by lack of blinding;

• An attempt was made to blind study participants and key personnel, but the blinding was likely broken and the outcome was likely influenced by lack of blinding.

Criteria for assessing “unclear risk” of bias.

One of the following:

• Insufficient information to enable a “low risk” or “high risk” assessment;

• The study did not address this finding.

BLIND ASSESSMENT OF RESULTS

Recognition bias due to outcome raters' knowledge of assigned interventions.

Criteria for the assessment of a "low risk of bias".

One of the following:

• No blinding of outcome assessment, but review authors consider outcome measurement unlikely to be affected by unblinding;

• Blinding of the outcome assessment was ensured and the blinding was unlikely to be broken.

Criteria for judging “high risk” of bias.

One of the following:

• Unblinded outcome assessment and outcome measurement is likely to be affected by unblinding;

• Blinding of outcome assessment, but the blinding has likely been broken and outcome measurement is likely to be affected by the lack of blinding.

Criteria for assessing “unclear risk” of bias.

One of the following:

• Insufficient information to enable a “low risk” or “high risk” assessment;

• The study did not address this finding.

INCOMPLETE RESULT DATA

Attrition bias due to the amount, nature, or handling of incomplete results data.

Criteria for the assessment of a "low risk of bias".

One of the following:

• There is no shortage of outcome data;

• Reasons for missing outcome data that are unlikely to be related to actual outcome (survival data are unlikely to be biased by censoring);

• Balanced missing outcome data in numbers between intervention groups, with similar reasons for missing data between groups;

• In the case of dichotomous endpoint data, the proportion of missing endpoints compared to the observed event risk is not sufficient to have a clinically relevant influence on the estimation of the intervention effect;

• With continuous outcome data, the plausible effect size (mean difference or standardized mean difference) between the missing endpoints is not sufficient to have a clinically relevant impact on the observed effect size;

• Missing data were imputed using appropriate methods.

Criteria for judging “high risk” of bias.

One of the following:

• Reason for missing outcome data is likely related to actual outcome, with numbers imbalanced, or reasons for missing data between intervention groups;

• In the case of dichotomous endpoint data, the proportion of missing endpoints compared to the observed event risk is sufficient to introduce a clinically relevant bias in the estimation of the intervention effect;

• For continuous endpoint data, plausible effect size (mean difference or standardized mean difference) between endpoints that are sufficiently absent to produce a clinically relevant bias in the observed effect size;

• “As treated” analysis performed with significant deviation of the intervention received from that assigned at randomization;

• Potentially inappropriate use of simple attribution.

Criteria for assessing “unclear risk” of bias.

One of the following:

• Insufficient reporting of attrition/exclusions to allow for a 'low risk' or 'high risk' assessment (e.g. random number not provided, reasons for missing data not provided);

• The study did not address this finding.

SELECTION REPORT

Reporting bias due to selective reporting of results.

Criteria for the assessment of a "low risk of bias".

One of the following:

• The study protocol is available and all pre-specified study results (primary and secondary) of interest to the review have been reported in the pre-specified manner;

  (Video) RoB 2.0: A revised tool to assess risk of bias in randomized trials pt1

• The study protocol is not available, but it is clear that the published reports contain all expected results, including those pre-specified (convincing texts of this kind may be rare).

Criteria for judging “high risk” of bias.

One of the following:

• Not all of the study's predefined primary endpoints were reported;

• One or more primary endpoints are reported using measures, analytical methods, or subsets of data (e.g., subscales) not previously specified;

• One or more reported primary endpoints were not pre-specified (unless a clear justification for reporting is provided, e.g. an unexpected adverse reaction);

• One or more relevant endpoints of the review are reported incompletely and therefore cannot be included in a meta-analysis;

• The study report does not include the results of a key finding that would be expected for such a study.

Criteria for assessing “unclear risk” of bias.

Insufficient information to allow a “low risk” or “high risk” assessment. Most studies probably fall into this category.

ANDERE BIAS

Bias due to issues not addressed elsewhere in the table.

Criteria for the assessment of a "low risk of bias".

The study appears to be free of other sources of bias.

Criteria for judging “high risk” of bias.

There is at least one significant risk of bias. For example the study:

• There was a potential source of bias related to the specific study design used; either

• It was claimed to be fraudulent; either

• He had another problem.

Criteria for assessing “unclear risk” of bias.

There may be a risk of bias, but there is:

• Insufficient information to assess whether there is a significant risk of bias; either

• Insufficient justification or evidence that an identified issue will lead to bias.