Critical Thinking & Technology Assesment > Common Biases > Case Mix Bias

Common Biases: Case Mix Bias

  • Case mix bias

    Case mix, or selection, bias occurs when cases are selected that inaccurately reflect the range of cases that occur in the general population. This type of bias is common in studies of new technologies where much is known about the cases chosen for the study but much less known about the broader population to which the technology could be applied.

    Selection bias powerfully affects case studies and case series (see Research Hierarchy) because the most difficult cases are frequently excluded from the sample set. It has been suggested that a study plagued with selection bias becomes an evaluation of the "wellest of the well and sickest of the sick." Thus, an interpreter using these extreme cases is more likely to have a favorable view of the technology than the clinician that uses the technology in general practice where he or she will encounter a broader spectrum of disease cases.

    • Example

      Selection bias is prominent in cancer screening studies (e.g. screening mammography) that do not utilize randomized controls. The reasoning lies in that subjects who receive screening are likely to have varying cancer incidence rates than those who do not receive screening.

    © Copyright Rector and Visitors of the University of Virginia 2013