Skip navigation

Using labs best

Bandolier 55 reported on a systematic review of laboratory test use which showed that a up to a third of tests were ordered for inappropriate purposes. The question left hanging was that of how to change behaviour to prevent silly requesting, and perhaps save rather significant sums, improve value for money, and reduce the huge loads placed on our hard-working laboratories.

Two more studies have swum into our ken, one a systematic review of interventions [1], the other a retrospective analysis of changes made in Ontario in the 1990s [2]. They give a powerful and positive message, that behaviour can be changed to the benefit of all.

What changes behaviour?

In their review, Solomon and colleagues [1] searched the English-language literature using several databases for studies to modify diagnostic test behaviour. Studies reviewed had to examine behaviour with an intervention and control and look at tests used for diagnostic procedures, and not just screening.

Because this is a difficult area where the normal rules of randomised trials may not always apply, they created their own set of methodological standards and applied them (blinded) to the 49 studies which met their inclusion criteria. The maximum score was 38 points.


There were only eight randomised studies, and the overall methodological quality was low, with a mean of 13 out of a possible 38 points. The majority (67%) of the studies targeted physicians in training, only 37% observed the results for 12 months or longer and only 45% looked at the impact on test ordering when the intervention had ended.

Reductions in test ordering by volume or cost was reported in 76% of the studies. The most impressive was that 86% of studies which looked at more than one behavioural change reported a reduction in test ordering. Audits used in conjunction with interventions designed to remove barriers to behaviour change were highly successful, with nine of 12 reporting reduced test ordering.

All of the interventions which used multiple lever strategies which addressing education, skills and barriers, and with feedback, were successful.

Making behaviour change

van Walraven and colleagues [2] examined the problem from a different perspective - looking back to see what the impact of a number of guidelines, test-ordering form and policy changes had been on the ordering of tests in Ontario. The background was an increase of 9.4 to 17.4 tests per person per year between 1976 and 1993.

Guidelines developed and introduced between 1991 and 1997 by expert panels were considered, and the effect on test ordering of erythrocyte sedimentation rate (ESR), renal dysfunction (microscopy, creatinine and urea), iron stores (iron and ferritin) and thyroid tests (TSH, T4-uptake) examined. Six tests (haemoglobin, glucose, sodium, uric acid, copper and aldolase) representing high, medium and low rates of use, and for which no guidelines were introduced, were used as controls.


Over the years 1991 to 1997 there was no change in age and sex standardised rates of use of any of the six control tests. By contrast, the effects of a variety of interventions made a big difference:

For ESR, removing a tick box from a request form plus guidelines discouraging ESR for asymptomatic patients led to a reduction from about 2000 to 500 tests per year per 100,000 population (Table).

Effect of policy, guideline and test form changes on test ordering
Clinical area Intervention Tests Effect of intervention
Haematology Change in form plus guidelines ESR 58% drop in requests (from 2000 to 500 per 100,000 persons)
Renal dysfunction Policy changes, guidelines and change in form Urinalysis ± microscopy Increase in urinalysis without microscopy and decrease in urinalysis with microscopy
    Urea Large reduction in urea tests, from about 1800 to 400 per 100,000 persons
    Creatinine No significant increase
Iron stores Policy and guidelines Iron binding 80% decrease in iron tests
    Ferritin Ferritin not significantly increased
Thyroid dysfunction Guidelines, policy changes and change in form Thyroxine and triiodithyronine resin uptake tests Over about three years total thyroxine tests fell from about 1300 per 100,000 to virtually zero
    TSH No significant increase in TSH tests

For renal dysfunction tests, guidelines and policy changes stressing the need for urine microscopy only when actually ordered led to changes (Table). Changes to request forms and guidelines stressing that urea was not needed in most situations led to a reduction by about 75% in urea requests with no increase in creatinine requesting.

Laboratories combined the restriction on iron and iron binding capacity tests with a guideline recommending ferritin testing alone to investigate iron deficiency led to an 80% iron tests with no significant increase in ferritin (Table).

Guidelines, policy changes and changes to requisition forms led to an almost complete elimination of thyroid uptake tests, without any concomitant increase in TSH tests when TSH alone was recommended.


Taken together these two papers show that diagnostic testing patterns of physicians can be changed, and that changes can have big consequences. Making an effective change usually means looking at not a single intervention, but several together. Problems are often multi-factorial, so addressing one aspect of the problem will usually fail.

In an accompanying editorial, the then editor of JAMA, George Lundberg, provides words of wisdom [3]. He gives a list of how to do things (truncated by Bandolier ):

  1. Know the literature and be certain you know the right things to do (have an evidence base, in other words).
  2. Get a cohort of influential physicians on your side to agree the proposed changes.
  3. Get on and do it, once you have agreement, and don't waste time on wider consultation.
  4. Education is important, so don't squirrel the knowledge, but let everyone know why you are doing what you are doing.
  5. Enjoy the ride, and be open about criticisms. React positively to valid complaints.
  6. Enjoy the success of providing a better, cheaper, faster and more effective diagnostic service.

Perhaps should be added another point - that of knowing that every effective change leaves some room for the next one. van Walraven's article makes the point that the changes in Ontario were calculated to save about 700,000 test request over about three or four years. Not much in money terms, because tests are cheap. But laboratories suffer enormous workloads, and just the thought of handling that many tests and samples makes one weak. Reducing the load means creating some small room for doing even better than we do now, and some space to think about the next step.
These may be 'just' management issues, but knowing what works and why opens the door to making laboratories more effective and more important.


  1. DH Solomon et al. Techniques to improve physicians' use of diagnostic tests. JAMA 1998 280: 2020-7.
  2. C van Walraven et al. Effect of population-based interventions on laboratory utilization. JAMA 1998 280: 2028-33.
  3. GD Lundberg. Changing physician behaviour in ordering diagnostic tests. JAMA 1998 280: 2036.

previous or next story in this issue