Intended for healthcare professionals

Letters Your results may vary

Interpreting test results: don’t blame the tools

BMJ 2020; 368 doi: https://doi.org/10.1136/bmj.m1162 (Published 25 March 2020) Cite this as: BMJ 2020;368:m1162
  1. Santhanam Sundar, consultant oncologist
  1. Nottingham University Hospitals NHS Trust, Nottingham NG5 1PB, UK
  1. sundar{at}oncology.org

McCormack and Holmes discuss the imprecision of medical measurements and suggest a method for estimating variation in test results.1 Byron’s paraphrased proverb, “Good workmen never quarrel with their tools,” aptly summarises the art of interpreting test results.2

If there were a perfect laboratory test for a given condition, patients could cut out the intermediary and go straight to the laboratory for a direct-to-consumer diagnostic test. In practice, laboratory tests might be imperfect, but they complement other clinical information. Laboratory test results always need to be interpreted in the context of patient symptoms, medical history, clinical findings, and radiology investigations.3

Prostate specific antigen (PSA), for example, is used for prostate cancer screening, but PSA levels need to be interpreted with other clinical information.45 A low normal PSA level of 2.0 ng/ml does not rule out prostate cancer if digital rectal examination of the prostate is abnormal. An abnormally high PSA level of 100 ng/ml is not of concern if the patient has had bladder retention and prostate biopsy has shown inflammation only.

A PSA level of 0.5 ng/ml is not a good outcome for a patient who has had radical prostatectomy but is an excellent outcome for a patient who has had prostate radiation. Even a 50% error rate in the above clinical scenarios is of minimal clinical consequence compared with the medical history of patients.

Busy clinicians already have a plethora of laboratory tests to contend with, and adding complexity to existing test results should be done only if high quality evidence indicates improvement in clinical care.1

Footnotes

  • Competing interests: None declared.

References