Skip navigation

False Positive Explosion

Any diagnostic test has to be assessed with respect to its sensitivity - the proportion of cases it picks up - and its specificity - the proportion of people without the condition who are confirmed as normal. With the continuing progress of technology it is only to be expected that new diagnostic tests are becoming more sensitive, and most diagnostic tests, be they laboratory or imaging, emphasise sensitivity as an important selling point.

There are, however, few diagnostic tests which can discriminate absolutely between healthy and sick populations (that is have a specificity of 100%). For the individual tested, and for the Health Service, the specificity of any diagnostic test is probably more important.

As the number of tests inexorably increases, we are seeing an explosion of false positive tests. Bandolier will be carrying articles on false positives in future, but in this edition we focus on two types of false positives with imaging.

Perils of false positives

For the individual, being falsely diagnosed as having a disease can cause tremendous suffering which is not always easily set aside. For the Health Service, the generation of many false positives places increasing burdens on overstretched resources, resources which, in the end, are wasted.

False positives cause problems for patients, doctors and the Health Service because the GP who elicits a false positive biochemical test will almost certainly have to take some further action.

Most test results do not fall conveniently into simple yes/no answers with obvious and huge differences. Normal and abnormal values overlap, particularly with effects of age and medicines. Definition of a "positive" is often set by using the upper 95th centile of a population as a cut off. This defines 5% of the population as being "abnormal" or "positive" - but it ain't necessarily so!

Just recently the Association of Clinical Biochemists and BUPA have produced a short handbook [1] on expected values for standard laboratory tests. Looking at tens of thousands of people, the results are given by age and sex, and colour coded so that the meaning of any result can quickly be obtained.

Thus a "high" alkaline phosphatase in a 30-year old man may have much more importance than the same result in a 75-year old woman.

This booklet is worth having handy when interpreting laboratory results.

The incidentaloma

A fascinating study from the Annals of Internal Medicine demonstrated just how commonly a pituitary adenoma could be diagnosed among normal people having an MRI scan. Scans were performed on 100 normal volunteers (70 women, 30 men, aged 18 to 60 years) before and after administration of paramagnetic contrast agent.

MRI scans from volunteers were mixed randomly with those of 57 patients with Cushing's disease, and all the scans were interpreted independently by three blinded and experienced reviewers.

Ten percent of normal volunteers were judged to have pituitary adenomas by two of the three reviewers (they had no pituitary endocrine disease). It is important to know that 10% of the normal adult population has pituitary abnormalities on MRI scanning, and that must be remembered when diagnosing pituitary disease - the positive predictive value of identification of a pituitary adenoma in patients with Cushing's disease was 86%. Between one and two patients with Cushing's syndrome and a pituitary adenoma identified by MRI scanning may not have Cushing's disease.

False positive slipped discs

Eight out of ten individuals will have low back pain at some time in their lives, and at any given time there are probably hundreds of thousands of patients suffering back pain in the UK. The costs of this, both to the Health Service and to society in general is huge.

How back pain is related to abnormalities in the lumbar spine is an area of controversy, but it is difficult to ignore a radiological abnormality in such a patient.

A paper in the New England Journal of Medicine suggests that ignoring such a finding may be in order. MRI examinations were performed on 98 asymptomatic people (50 men and 48 women aged 20 to 80 years); those with a history of back pain lasting more than 48 hours, or lower back problems were excluded.

These 'normal' MRI scans were mixed with 27 abnormal scans from patients with back pain, and all the scans were then evaluated blind by two experienced neuroradiologists. The readers independently evaluated the status of the five intervertebral disks in the lumbosacral spine in all 125 subjects.

The results showed that only 36% of the 98 asymptomatic subjects had normal disks at all levels. With the results of the two readings averaged, 52% had a bulge at least one level, 27% had a protrusion and 1% an extrusion. More than one disk was involved in 38% of people.

Thus MRI scans showed that "abnormal" pathology was the norm. An accompanying editorial [4] states laconically that "the recent increase in the rates of lumbar spine surgery may be related in part to the availability of new imaging techniques".

False positive explosion

Increasing power to test may not only serve to increase sensitivity, but may also decrease specificity, create false positives and thus anxiety and unnecessary cost. Bandolier will be happy to share any experiences readers have of the false positive explosion.

References:

  1. TP Whitehead, D Robinson, AC Hale, AR Bailey. Clinical Chemistry & Haematology - adult reference values. BUPA Medical Research, London,1994.
  2. WA Hall, MG Luciano, JL Doppman, NJ Patronas, EH Oldfield. Pituitary magnetic resonance imaging in normal human volunteers: occult adenomas in the general population. Annals of Internal Medicine 1994 120: 817-20.
  3. MC Jensen, MN Brant-Zawadki, N Obuchowski, MT Modic, D Malakasian, JS Ross. Magnetic resonance imaging of the lumbar spine in people without back pain. New England Journal of Medicine 1994 331: 69-73.
  4. Editorial: Magnetic resonance imaging of the lumbar spine. Terrific test or tar baby? New England Journal of Medicine 1994 331: 115-6.



previous or next story in this issue