Skip navigation

Testing times

Bandolier has long been concerned about diagnostic testing. We have argued that new diagnostic tests are introduced with lower levels of evidence than new treatments, and that this plays an important part in rising health care costs.

The evidence-base for diagnostic tests, and their use as motors for change through 'technology creep' have been slippery topics, hard to grip. New information will help those who do the tests and those who use them.

Systematic review methods

A Cochrane methods working group under the direction of Les Irwig and Paul Glasziou in Australia have, within the last few weeks, posted on the Internet some recommended methods for systematic reviews of screening and diagnostic tests.
their Internet address is:
This is an important document, not because it has all the answers, but it informs us as how to seek answers about diagnostic tests. It draws together a very useful bibliography on diagnostic test evaluation and overview, and has excellent guidelines on how to work. It will be an invaluable tool to move knowledge forward in this difficult subject, and, though it will take some time, it will encourage groups and individuals to begin to put the bricks into this particular wall.

Test more, find more, treat more

The importance of knowledge about diagnostic tests is demonstrated by two articles and an editorial in JAMA. Their thrust is that increase use of diagnostic tests is directly related to increases in treatments, without any rational link to need. The studies [1,2] are good examples of how observational studies from administrative data can be used to pose important questions about how we run our health service.

Invasive cardiac procedures

The first study [1] was a survey of utilisation rates for cardiac stress tests, coronary angiography and revascularisation procedures in 12 coronary angiography service areas in New England. In each area data were collected from 1992 and 1993, and then the rates of diagnostic testing and procedures correlated using the correlation of determination (R 2 , where a value of 1 means that all of the change in one parameter is determined by change in the other, and where a value of 0 means that none of it is).

Strong positive relationships were found both between testing and subsequent coronary angiography (R 2 = 0.61) and between coronary angiography and subsequent revascularisation (R 2 = 0.82).

The differences in procedures between locationswere not explained by differences in prevalence, or the provision of local surgical services. The authors concluded that the relationships between testing and subsequent procedures reflect underlying uncertainty about when to test for and treat ischaemic heart disease in a population where the potential reservoir of coronary disease is huge. So strategies chosen to evaluate those with symptoms can have a dramatic effect on how much disease is discovered. Test more, find more, treat more.

Impact of diagnostic testing over time

The other study [2] was a survey of the relationship between frequency of testing in a number of different clinical situations - cardiac catheterisation, spinal imaging, mammography, swallowing studies and prostatic biopsy - in 30 million elderly Americans over a 7 year period. The bulk of the variance in therapeutic intervention rates was accounted for by diagnostic testing rates (R 2 values between 0.82 and 0.93).

The authors concluded that " a substantial increase in diagnostic testing closely tracked the increase of clinically relevant downstream procedures". They suggest that managing diagnostic tests could be an important strategy for controlling the increased use of therapeutic interventions.


An accompanying editorial [3] says "studies based on administrative data are often better at raising questions than answering them". Intervention rates and judgements of appropriateness of care vary. We now know that the implications of increasing testing leading to increasing volume of care are immense.

A cardinal principle of public health is "never do a test unless you know what you are going to do with the result". This applies equally well to the introducion of new tests. As soon as they are used, the dominoes start to fall.

The problem for all of us is to know exactly where the line is that we are trying to hold. More evidence about diagnostic tests would help us find it.


  1. 1 DE Wennberg, MA Kellett, J Dickens et al. The association between local diagnostic testing intensity and invasive cardiac procedures. Journal of the American Medical Association 1996 275: 1161-4.
  2. D Verrilli, G Welch. The impact of diagnostic testing on therapeutic interventions. Journal of the American Medical Association 1996 275: 1189-91.
  3. AM Epstein. Use of diagnostic tests and therapeutic procedures in a changing health care environment. Journal of the American Medical Association 1996 275: 1197-8.

next story in this issue