Critical appraisal:Baars JE, Looman CW, Steyerberg EW, Beukers R, Tan AC, Weusten BL, et al 2011 2 1

Risk of bias assessment: nested case control – risk factors

Bias in selection of participants into cohort study

Sources of cases and controls

<img alt="File:Jutta's info icon.png" src="/australiawiki_test/images/d/d9/Jutta%27s_info_icon.png" width="16" height="16"> Drawn from the same population (This will usually be the case when a case-control study is nested in a single cohort containing exposed and unexposed people and cases accrue during follow-up of the whole cohort) (low risk)

Selection of cases and controls

Cases and controls are randomly selected from all available cases and controls, controls matched to cases by <img alt="File:Jutta's info icon.png" src="/australiawiki_test/images/d/d9/Jutta%27s_info_icon.png" width="16" height="16"> risk set (Defined by sex, age group, date of entry into cohort and date of case-defining event) (either at selection or during analysis) (low risk)

Bias due to error in outcome measurement

Definition of cases (outcome)

Outcome precisely specified and with pathological or other objective confirmation (low risk)

Definition of controls

Objective evidence of no past history of outcome of interest (low risk)

Was outcome of interest likely to have been absent at the time to which the exposure refers?

Yes (low risk)

Was follow-up long enough for outcome to occur as a consequence of the measured exposure? (Requires prior specification of a sufficient follow-up period)

Yes (low risk)

Bias due to error in exposure measurement

Measurement of exposure

Objective measurements from pre-existing records or <img alt="File:Jutta's info icon.png" src="/australiawiki_test/images/d/d9/Jutta%27s_info_icon.png" width="16" height="16"> baseline (Existing at or before baseline, where baseline is the time at which a participant is recorded to have entered the cohort or, if obtained after baseline, a time before onset of symptoms of the outcome or any likely effect of the developing outcome on the exposure) physical or biological assessment, each blind to case or control status (low risk).

Was the same method used to measure exposure in cases and controls?

No OR insufficient information to tell (high risk)

Bias due to non-participation

Participation rate in cohort

Participation rate in exposed cohort ≤10 percentage points different from non-exposed cohort OR exposed and non-exposed are from the same cohort (low risk)

Participation (response) rate for cases

Not applicable – new data not being collected from participants

Participation (response) rate for controls

Not applicable – new data not being collected from participants

Difference in participation rate (response rate) between cases and controls

Not applicable – new data not being collected from participants

Bias due to missing data

Completeness of follow-up of cohort

Active or passive follow-up of participants with methods for ascertainment of outcome and death clearly described AND with methods for ascertainment of emigration from population-at-risk clearly described or censoring at date of last follow-up OR there is a plausible estimate of >90% follow-up (low risk)

Accuracy of dates of outcome or censoring

One or more of dates of outcome or censoring not ascertained to within one year OR insufficient information to tell (moderate risk)

Difference in follow-up between exposed and non-exposed members of cohort

Completeness of follow-up in exposed and unexposed participants is very unlikely to be the same and difference between the two is, or is likely to be, large (≥10%) OR insufficient information to tell (high risk)

Difference in missing data for exposure between cases and controls

Difference in missing data for exposure < 10 percentage points (low risk)

Bias due to confounding

Comparability of cases and controls with respect to potentially important confounding variables (Requires prior specification of potentially important confounders)

Age and other potentially important confounders measured and controlled by design or in analysis (low risk)

Analysis bias

Analysis appropriate to design

When controls are frequency-matched to cases, matching variables are controlled in the analysis OR when controls are individually matched to cases, a conditional analysis is used or matching variables are controlled in the analysis (low risk)

Covariates are appropriately included in statistical analysis models

Variables measuring the same underlying concept or lying in the same causal pathway ARE NOT included together as covariates in statistical analysis models (low risk)

Overall risk of bias

High risk of bias

Additional comments: Please replace this text and include any additional comments in regards to your risk of bias rating

This appraisal has been completed.

Article: Baars JE, Looman CW, Steyerberg EW, Beukers R, Tan AC, Weusten BL, et al. The risk of inflammatory bowel disease-related colorectal carcinoma is limited: results from a nationwide nested case-control study. Am J Gastroenterol 2011 Feb;106(2):319-28 Available from: http://www.ncbi.nlm.nih.gov/pubmed/21045815.

Assigned to

User:Cecilia.taing

Topic area

Guidelines:Colorectal cancer/Colonoscopy surveillance

Clinical question

What is the most appropriate time interval for surveillance in IBD patients?

Form: Form:Quality appraisal nested cc risk factors
Outcomes: Risk of colorectal cancer based on duration of IBD; Risk of colorectal cancer based on duration of PSC (verse no PSC)

Section below only relevant for Cancer Council Project Officer

Edit appraisal assignment