Less Testing: Better for Patients and Health Care Systems?

Lab tests are common in modern medicine. Patients expect them, doctors base decisions on them. But do we need so many tests?  Probably not, in many cases. Researchers at Harvard University recently reviewed studies of lab tests over the last 15 years and estimated that one in five are unnecessary.

Over-testing inconveniences patients, drains resources from health care systems and can lead some patients through the "Ulysses syndrome" — a nerve-racking and ultimately futile series of tests and interventions based on an initial false-positive or uncertain result.

But reducing unneeded tests has proven difficult. On one hand, doctors may hesitate to skip tests for fear of missing early signs of illness in their patients. On the other hand, a major barrier is lack of agreement on what exactly constitutes an unneeded test, and how to identify such tests in practice.

Eric Morgen is a clinical and research fellow in the Department of Laboratory Medicine and Pathobiology  at the University of Toronto. He recently published a paper on repeat tests — the same test on the same patient, multiple times — with University of Calgary professor Christopher Naugler.

The pair looked at six common tests of patients in and around Calgary over one year. Their results, published in the American Journal of Clinical Pathology, showed that 16 per cent of these repeats were inappropriate, with estimated lab costs of up to $2.2 million. Scaled for population, excess costs were around $160 million in Canada and $2.4 billion in the U.S.

Morgen spoke with Faculty of Medicine writer Jim Oldfield about his study, unnecessary tests and how technology could help limit them.

Why did you look at repeat tests?

Well, if you want to reduce unnecessary testing, the first thing to do is figure out exactly which tests are unnecessary. This is surprisingly difficult, because usually it requires a doctor to review the clinical situation and decide if a particular test was unnecessary — and even then there are different opinions of what “unnecessary” means. We realized that repeat tests present an opportunity to simplify and automate this process for certain tests. Also, they comprise a large proportion of lab tests: for many common tests, repeats within a year of the initial test account for well over half of test volumes.

And why did you choose the six tests you looked at?

We chose tests where we could provide very simple, concrete criteria for when a repeat was inappropriate. Part of the difficulty in studying appropriateness in lab tests is that there aren't clear clinical guidelines on when to test patients in every situation, so researchers often apply different criteria of "appropriate" from study to study. With the six tests we looked at there are either clear guidelines or reasonably good evidence on when to repeat them, so we can be confident about whether they were appropriate or not. They're also relatively common tests — such as cholesterol and vitamin D levels.

How many tests did you include in your study?

We had data on everyone tested in the Calgary area  — about 1.4 million patients. But we focused the analysis on a random sample of nearly 400,000 tests in 100,000 patients. That data is representative of all instances of our six chosen tests in patients in and around Calgary over one year, because all tests for people in that area go through one lab. The size of the sample population was definitely a strength in this study, and we think the results could be useful for system-level changes that reduce costs or improve care.

So what would system-level change based on this data look like?

It can be very hard to change physician behaviour. Doctors build their expertise through years of training and experience, so education or awareness campaigns alone often aren't enough to alter their practice. It's probably more effective to intervene with an electronic reminder or question — some kind of flag that enters a doctor's consciousness in the act of ordering a test. With the six tests we looked at, you could create those automated flags, which is one of the big advantages of our approach. As well, it would be possible to track the number of inappropriate repeats over time, to monitor the success of the intervention.

Could we roll out such systems on a wide scale?

I think so. It wouldn't be hard to coordinate a top-down solution in Calgary and areas where most results go through a single lab, such as Manitoba and parts of other provinces. Toronto would be more difficult, but it could be done institution by institution.

How do you calculate the cost savings from such an intervention?

For our study, we provided a range of potential direct cost savings for the health care system, based on various price estimates from Canada and the U.S. On the lower end of the range, we looked at the internal costs of these tests — how much extra it costs a hospital laboratory to run that test, and on the higher end we looked at how much private laboratories would charge a hospital for the same test.

But it's very hard to quantify all the other indirect costs of an inappropriate test. Some studies have tried to measure downstream costs of over-testing, which can include unnecessary use of clinical services, follow-up tests, hospitalizations and procedures. But most of that work hasn’t been done, as the editorial  that accompanies our paper points out.

It’s also worth remembering that cost savings aren’t nearly the whole story. The inconvenience, anxiety and unpleasantness of some of these situations are really important for patients, and everyone wants to avoid them. However, on the flip side we need to be careful about putting up barriers to testing. It definitely shouldn't be hard for doctors to order tests, but right now, it's probably too easy to order too many.

Do you have any advice for patients facing a repeat test?

Well, most of the time it will probably be entirely appropriate. Remember, we showed that 16 per cent of the repeat tests we studied are inappropriate, and while we think the real number is somewhat higher, the large majority of repeats are likely to be fine. So I don’t think patients need to worry about every repeated test. That said, it also depends on your preferences as a patient — which doctors are sensitive to — and it can definitely have an impact if you say you prefer to avoid tests when not absolutely necessary.