ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
2
Navigation:
Risk Of Bias In Non-randomized Studies - of Exposure (ROBINS-E), Launch Version, 1 June 2022
3
Go to Planning Stage
CONFIDENTIAL - PLEASE DO NOT SHARE
4
Go to Preliminary Considerations A-D
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
5
Go to Preliminary Considerations E
6
Go to Domain 1 vA
ROBINS E includes seven domains of bias:
7
Go to Domain 1 vB
• Domain 1: Risk of bias due to confounding
8
Go to Domain 2 vA
• Domain 2: Risk of bias arising from measurement of the exposure
9
Go to Domain 2 vB
• Domain 3: Risk of bias in selection of participants into the study (or into the analysis)
10
Go to Domain 2 vC
• Domain 4: Risk of bias due to post-exposure interventions
11
Go to Domain 3
• Domain 5: Risk of bias due to missing data
12
Go to Domain 4
• Domain 6: Risk of bias arising from measurement of the outcome
13
Go to Domain 5
• Domain 7: Risk of bias in selection of the reported result
14
Go to Domain 6
15
Go to Domain 7Each bias domain in ROBINS-E is addressed using a series of signalling questions that aim to gather important information about the study and the analysis being assessed. Many signalling questions have response options ‘Yes’, ‘Probably yes’, ‘Probably no’, ‘No’ and ‘No information’. For these, ‘Yes’ and ‘Probably yes’ have the same implications for risk of bias, and ‘No’ and ‘Probably no’ have the same implications for risk of bias; the distinction allows the user to distinguish between situations where definitive information is available from those where a judgement is made. Other signalling questions have different response options, specific to the question, which may be used to distinguish between different risks of bias.

16
Go to Addressing Appropriateness
17
Go to Summary
18
After the relevant signalling questions have been completed, three judgements are made as follows.
19
1. The risk of bias in the result that arises from this domain. This should be interpreted as risk of material bias that has the potential to impact on the estimated effect of exposure on outcome. A suggested judgement is generated using an inbuilt algorithm, based on the answers to the signalling questions. The judgements and their broad interpretations are as follows.
20
Please enter below reference information for the study under evaluation:JudgementInterpretation
21
Reference:Low risk of bias*there is little or no concern about bias with regard to this domain
22
Some concernsthere is some concern about bias with regard to this domain, although it is not clear that there is an important risk of bias
23
High risk of biasthe study has some important problems in this domain: characteristics of the study give rise to a high risk of bias
24
Very high risk of biasthe study is very problematic in this domain: characteristics of the study give rise to a very high risk of bias
25
*For Domain 1 (Risk of bias due to confounding),this is referred to as “Low risk of bias (except for concerns about uncontrolled confounding)”, in which confounding is very well addressed but cannot be eliminated as a possibility. This is because a risk of bias due to uncontrolled confounding cannot be excluded in an observational study.
26
ROBINS-E is intended to provide a framework for making informed and reasonable judgements about risk of material bias in studies of the effects of exposure on outcome. On occasion, answers to the signalling questions may not yield an appropriate risk of bias judgement based on the algorithm. Therefore, suggested risk of bias judgements produced by the algorithms can be overridden, in which case justification should be provided. We aim for transparency and reasonableness rather than mechanistic adherence to every word of the tool’s contents.
27
2. The predicted direction of bias, balancing the various issues addressed within the domain. Response options for this depend on the type of bias being addressed.
28
3. Whether the risk of bias (arising from this domain) is sufficiently high, in the context of its likely direction and the magnitude of the estimated exposure effect, to threaten conclusions about whether the exposure has an important effect on the outcome. This last judgement should take into account both the finding of the study (including its magnitude and the strength of evidence around it) and a broad assessment of bias (through the likelihood of it being present, its likely direction and its likely magnitude. This is a challenging judgement to make, and detailed guidance has not been developed for this. Response options are Yes / No / Can’t tell.
29
30
After completing all seven bias domains, an overall judgement made for each of the three considerations above. Judgements for the first and third are derived from the domain-level judgements using an algorithm. As for bias domain-level judgements, justification should be provided when the overall judgement suggested by the algorithm is overridden.
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100