Skip navigation


One of the objectives of Bandolier is to analyse and summarise articles to ease the huge task of the clinician, purchaser, provider - and sometimes, we hope, patient - faced with the huge and increasing mass of medical research publications.

Sometimes it is necessary to recommend actually reading an article yourself because they merit individual and special attention. These "mindstretchers" will not be easy articles; they will not be softened by the process of predigestion that usually accompanies editing. Sometimes articles will be recommended because the contents are very important and on other occasions because the process and methods offer an excellent educational opportunity. Not every "mindstretcher" will be appropriate for every reader - but one will come your way soon enough.

Mindstretcher 1: Guidelines for Meta-analyses Evaluating Diagnostic Tests

Diagnostic tests are much more difficult to evaluate than treatments. For treatment the outcome is relatively straightforward, but for a diagnostic test the outcomes are much less straightforward and much more difficult to assess. Different techniques are required to measure the added value that a new diagnostic test offers.

The major review of meta-analyses evaluating diagnostic tests in Annals of Internal Medicine not only describes the meta-analyses included in the reviews but also has an excellent discussion of the criteria used within the reviews and the criteria used to appraise and classify the reviews themselves.

Eleven meta-analyses were found, covering the following tests:

  • PET assessment of myocardial perfusion.
  • CT to stage lung cancer.
  • Fluoroscopy for the diagnosis of coronary artery disease.
  • Parathyroid imaging.
  • The accuracy and usefulness of thermography for lumbar radiculopathy.
  • Assessment of dipstick tests for urinary tract infection.
  • A comparison of three different radiological procedures for the detection of lumbar disc herniation.
  • Diagnostic significance of carcinoembryonic antigen in the differential diagnosis of malignant mesothelioma.
  • Tests for human immunodeficiency virus antibodies.
  • Tecnetium venography in the diagnosis of venous thrombosis of the lower limb.
  • The reliability of non-invasive carotid studies.

Some of these tests are common and some are rare, but the paper describing their meta-analysis is excellent. It comes from some of the superstars in this field, including Thomas Chalmers and Frederick Mosteller and involves Dr Irwig from the University of Sydney, who has particular expertise in this area. This paper will take you an hour or two, but it offers an excellent opportunity to stretch the mind and extend your education.


L Irwig, ANA Tosteson, C Gatsonis, J Lau, G Colditz, TC Chalmers, F Mosteller. Guidelines for meta-analyses evaluating diagnostic tests. Annals of Internal Medicine 1994 120: 667-76.

previous or next story in this issue