Limites et Biais de l’Evaluation Retrospective d’une Technique de Diagnostic

A propos d’une étude de la scintigraphie cérébrale
  • A. Alpérovitch
  • F. Cavaillolés
  • C. Soussana
  • B. Thébault
  • B. Bok
Conference paper
Part of the Lecture Notes in Medical Informatics book series (LNMED, volume 11)

Abstract

In order to assess retrospectively the diagnostic value of radionuclide brain scans, 399 cases which fulfilled the required diagnosis criteria were reviewed. These 399 cases constituted only 22% of the scans performed during one year in the hospital. In this sample, the test overall accuracy was equal to 79.5. False positive errors (4%) were less frequent than false-negative ones (16.5 %). By comparing scintigraphic diagnoses with preliminar clinical diagnoses, the main result appeared to be the following. When these two diagnoses were identical, the final diagnosis agreed with the preliminar clinical hypothesis in 97 % of the cases; but when they were different, final diagnosis and preliminar clinical hypothesis agreed in only 15% of the cases.

But limits and biases of such a retrospective study make suspectáble this apparently positive conclusion. Indeed, they lead to i) overestimate the proportion of abnormal scans; ii) under estimate the proportion of false-negative errors; iii) over estimate, on the overall, the diagnosis value of this test. But it seems, unfortunately, that neither a prospective study nor a change in the efficacy criteria which are taken into account, can permit to eliminate important errors in assessment of diagnostic tests.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    P. O. Alderson, M. Mikhael, R. E. Coleman, M. Gado. Optimal utilization of computerized tomography and radionuclide brain imaging. Neurology, 1976, 26: 803–807Google Scholar
  2. 2.
    E. F. Bernstein. The current place of non invasive diagnostic techniques in evaluating cerebrovascular disease. J. Roy. Soc. Med., 1978, 71: 709–710Google Scholar
  3. 3.
    F. Cavailloles, A. Alperovitch, C. Soussana, B. Thebault, B. Bock. Clinical usefulness of brain scintigrams. Int. J. Applied Radiat., (sous presse)Google Scholar
  4. 4.
    G. H. Boulay, J. Marshall. Comparison of EMI and radioisotope imaging in neurological disease. Lancet, 1975, ii: 1294–1297Google Scholar
  5. 5.
    B. J. Mcneil, S. J. Adelstein. Determining the value of diagnostic and screening tests. J. Nuclear Med., 1976, 17: 439–448Google Scholar
  6. 6.
    T. Poynard, A. Alperovitch. Evaluation des moyens de diagnostic - Etude critique des principaux indices mesurant la valeur diagnostique. Journal de Radiologie (sous presse)Google Scholar
  7. 7.
    D. Ransohoff, A. R. Feinstein. Problems of spectrum and biais in evaluating the efficacy of diagnostic tests. New Engl. J. Med., 1978, 299: 926–930Google Scholar
  8. 8.
    A study of the efficacy of diagnostic radiologic procedures. Report from the American College of Radiology, 1977.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1981

Authors and Affiliations

  • A. Alpérovitch
    • 1
  • F. Cavaillolés
    • 2
  • C. Soussana
    • 2
  • B. Thébault
    • 2
  • B. Bok
    • 2
  1. 1.INSERMVillejuifFrance
  2. 2.Hopital BeaujonClichyFrance

Personalised recommendations