Page 50 - Risk quantification and modification in older patients with colorectal cancer
P. 50

                                Chapter 2
Appendix B Criteria for scoring of risk of bias based on the CHARMS checklist
  Potential Bias Participant selection
Selective inclusion
Predictor assessment
Risk of bias
L Low risk if
M Moderate risk if H High risk if
Risk of bias
L Low risk if
Items to be considered for potential bias
- selection bias was unlikely,
- study avoided inappropriate inclusions or exclusions,
- in- and exclusion criteria were adequately described
- participants were enrolled at a similar presentation of their disease
- differences were accounted for by including appropriate predictors in the analysis
- not satisfying one of the above OR
- no adequate description of the recruitment of the study sample
- no adequate description of the sample for key predictors
if both items were not adequately described
- predictor definitions were the same for all participants - predictor measurement was blinded to outcome data - all predictors were available at the time the model is intended to be used
- predictors were measured with valid and reproducible methods such that misclassification was limited and if
- predictors were assessed in a similar way for all study participants
one of the criteria was not satisfied
if method for assessment of outcome was not adequately described
- the outcome was pre-specified and
- measured with sufficient validity and reproducibility and - measured in a similar way for all study participants and - if the outcome was assessed independently from the assessment of predictors
Note: for easy to obtain predictors such as gender, it is
not possible to assess outcome independent of predictor information
one of the criteria was not satisfied
the assessment of outcome was not adequately described
there was no loss-to-follow-up
- there were no important differences on key characteristics between included participants and those who were lost-to-follow-up or missing
- loss-to-follow-up was lower than 20% and - there were no important differences on key characteristics between included participants and those who were lost-to-follow- up or missing OR:
- loss-to-follow-up was higher than 20% but missing
data and loss-to-follow-up were imputed adequately or there were no important differences on key characteristics between included participants and those who were lost- to-follow-up or missing
  Treatment predictors; M do the modify outcome
and were they handled appropriately
  Outcome assessment
Attrition
Moderate risk if H High risk if
Risk of bias
L Low risk if
M Moderate risk if H High risk if Risk of bias
L Low risk if
M Moderate risk if
      48




















































   48   49   50   51   52