More information worse decisions? Evaluating potential overuse of laboratory diagnostics in diagnosing anaemia

M. Kip, M.L.J. Oonk, M.D.L. Levin, G.C.M. Kusters, E. Koffijberg

Research output: Contribution to journalMeeting AbstractAcademic


Objectives: More information is often thought to improve medical decision making, which may lead to overuse of relatively cheap tests. For diagnosing the underlying cause of aneamia in general practice, up to 14 different laboratory tests may be ordered by general practitioners (GPs), according to the Dutch guideline. This study assessed which of these 14 tests actually contribute to diagnosing an underlying cause of anaemia, and which subset of tests is optimal for establishing a correct diagnosis.

Methods: Data from a previous questionnaire presenting GPs (n=139) with three varying real-world anaemia cases, and all 14 corresponding test results, was used. GPs selected the expected underlying cause (anaemia of chronic disease, iron deficiency anaemia, renal anaemia, 'unknown or other'). The correct underlying cause was established by an expert panel (i.e. GP, internist, clinical chemist). In this study, multinomial regression was used to determine the value of each test for establishing a particular underlying cause. Logistic regression was used to determine the value of each test for establishing the correct underlying cause. Stepwise backward selection using the Akaike Information Criterion was applied to determine the optimal subset of tests.

Results: Only 9 (60%) of the laboratory tests, and patient age, contributed to diagnosing an underlying cause of anaemia (CRP, ESR, ferritin, folic acid, haemoglobin, leukocytes, MDRD, reticulocytes and serum iron). Diagnosing the correct underlying cause required just five (33%) tests (CRP, ferritin, folic acid, MCV and transferrin), and patient age.

Conclusions: Of the full set of 14 laboratory tests recommended by the Dutch guideline, only a subset enhances the ability of the GP to diagnose an underlying cause of anaemia, from a statistical perspective. A subset of five tests has most added value. Extending this set with more tests may lead to lead to a decrease rather than increase in correct diagnoses.
Original languageEnglish
Pages (from-to)S908-S908
JournalValue in health
Issue numberSuppl. 3
Publication statusPublished - Nov 2019
EventISPOR 22nd annual European Congress 2019 - Copenhagen, Denmark
Duration: 2 Nov 20196 Nov 2019
Conference number: 22

Cite this