Critical appraisal:Mayre-Chilton KM, Talwar BP, Goff LM 2011

From Cancer Guidelines Wiki

Risk of bias assessment: case control – risk factors

Bias in selection of participants into study
Selection of cases and controls
Drawn from different populations but unlikely to introduce bias (moderate risk)
Bias due to error in outcome measurement
Definition of cases (outcome)
Outcome imprecisely specified OR outcome self-reported and cases not blind to hypotheses related to outcome OR insufficient information to tell (high risk)
Definition of controls (Not relevant when outcome of interest is recurrence and there is expected to be a short latent interval from exposure to the hypothesised risk factor to a symptomatic recurrence)
Objective evidence of no past history of outcome of interest (low risk)
Bias due to error in exposure measurement
Measurement of exposure
Structured interview not blind to case or control status OR self-administered questionnaire OR insufficient information to tell (high risk)
Temporality of exposure
Exposure precedes onset of disease in cases but exposure in controls is not from a calendar time corresponding to exposure in cases (moderate 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 (response) rate for cases
< 50% participation rate (< 60% response rate) OR insufficient information to tell (high risk)
Participation (response) rate for controls
≥ 60% participation rate ( ≥70% response rate) (low risk)
Difference in participation rate (response rate) between cases and controls
Participation or response rate in cases >10 to ≤15 percentage points different from controls (moderate risk)
Bias due to missing data
Difference in missing data for exposure between cases and controls
Difference in missing data for exposure ≥10 to <20 percentage points (moderate risk)
Bias due to confounding
Comparability of cases and controls with respect to potentially important confounding variables (Requires prior specification of potentially important confounders)
No potentially important confounders or only age controlled by design or in analysis OR insufficient information to tell (high risk)
Analysis bias
Analysis appropriate to design
None of the above OR insufficient information to tell (high 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)

Jutta's tick icon.png This appraisal has been completed.

Mayre-Chilton KM, Talwar BP, Goff LM. Different experiences and perspectives between head and neck cancer patients and their care-givers on their daily impact of a gastrostomy tube. J Hum Nutr Diet 2011 Jun 8 Abstract available at
Assigned to
Topic area
Guidelines:Some topic
Clinical question
Form:Quality appraisal cc risk factors

Section below only relevant for Cancer Council Project Officer

Edit appraisal assignment