Errors in systematic reviews: an example of computed tomography screening for lung cancer

Eur J Cancer Prev. 2014 Jan;23(1):43-8. doi: 10.1097/CEJ.0b013e3283616290.

Abstract

Systematic reviews are utilized in evidence-based medicine and are increasingly being used to help guide standards, guidelines, and clinical practice. The National Lung Screening Trial results prompted such a review of lung cancer screening literature. The review was endorsed by five major medical societies. We aimed at assessing its accuracy. Two independent groups of two reviewers reviewed the systematic review, including its source literature. Errors were placed into three major categories and tabulated: (i) selection of studies, (ii) misrepresentation of published reports, and (iii) errors in calculation and rounding. A total of 151 errors were found. There were 13 errors in selection of studies, 124 errors due to misrepresentation of published reports, and 14 errors in calculations and rounding. The extent of these errors raises concern about the credibility of the conclusions of the recent lung cancer screening systematic review. A process that allows for a thorough checking of data included in systematic reviews should be established.

MeSH terms

  • Data Collection / methods*
  • Diagnostic Errors*
  • Early Detection of Cancer*
  • Evidence-Based Medicine
  • Humans
  • Lung Neoplasms / diagnostic imaging*
  • Lung Neoplasms / prevention & control
  • Prognosis
  • Review Literature as Topic*
  • Tomography, X-Ray Computed*