Showing Test Prices Cuts Costly Lab Orders, Five-Year Swedish Study Finds

When a county in southern Sweden began displaying the price of every clinical chemistry test next to its name in the electronic ordering system, and later made primary care centers pay the full laboratory bill, the ordering habits of physicians changed in ways that persisted for years. A new retrospective study from Kronoberg County, published in Discover Health Systems, followed monthly laboratory test requests from January 2010 through December 2018 and found that the two financial interventions reduced test ordering over the long term, but the effect was far from uniform. It depended strongly on how expensive a test was and on which tier of the healthcare system the physician worked in.

The research team, Kim Ekblom and Annika Petersson, had previously shown in a shorter study, published in 2018, that displaying test costs to all staff with access to the computerized patient record decreased laboratory ordering from secondary healthcare centers but not from primary healthcare centers. When a full cost charge was then introduced for primary care, privately operated centers reduced their ordering while publicly operated centers did not. Because both interventions were permanent, the investigators set out to determine whether those short-term changes would hold up over a five-year follow-up, and whether expensive tests behaved differently from cheap ones.

The study used a retrospective interrupted time-series design, a quasi-experimental approach well suited to policy changes that sweep across an entire population at once. The county had 33 primary healthcare centers, 22 run by the regional council and 11 privately operated, plus two secondary healthcare centers. All tests were analyzed by central laboratories or sent to a tertiary laboratory, with no point-of-care testing included, and patient self-referral was not allowed. In Sweden, public healthcare is tax funded with no cost reimbursement system, and before the interventions the Kronoberg laboratory was primarily budget financed, with each healthcare unit holding its own separate budget.

The first intervention, cost display, arrived in September 2013 for both primary and secondary care. The price in Swedish crowns appeared next to the laboratory test name in the Cambio Cosmic ordering system, visible to the referring physician both when ordering and when the result was presented. The displayed cost covered everything: reagents, controls, instruments, salaries and overhead. The second intervention, cost charge, followed in January 2014, when all primary healthcare centers became responsible for paying the full laboratory test costs out of budgets defined by the number of listed patients. Secondary care centers never faced the charge, which made the two settings a natural comparison of information alone versus information plus financial responsibility.

For the analysis, the researchers selected two sets of ten analytes each. Set 1 contained the ten most frequently ordered tests costing more than 100 Swedish crowns according to the 2016 price list, chosen to capture the largest possible sample among the most expensive analytes. Set 2 contained the most frequently ordered tests representing a range of common medical issues, from glucose and cholesterol to C-reactive protein and sedimentation rate. The main outcome was the number of tests ordered per physician visit per month, obtained from the Department of Analytical Support at Region Kronoberg. Data from publicly operated primary care centers were excluded because the earlier study had shown no significant decrease there.

The statistical machinery was deliberately careful. The team used Poisson regression with interaction terms between time and the interventions, allowing them to separate an immediate level shift from a change in the trend slope, that is, an instant effect from a long-term effect, each compared against the projected pre-intervention trajectory. Four models were constructed for each analyte, all adjusted for seasonal variability: a basic model with a constant linear trend, one allowing a change after cost display, one allowing a change after cost charge, and one allowing changes after both. The best-fitting model for each analyte was selected using the Bayesian information criterion. The pre-intervention baseline ran from January 2010 to August 2013, yielding 44 monthly measuring points, which the authors note reduced short-term overdispersion and autocorrelation.

The results were strikingly heterogeneous. In secondary healthcare, the cost display model best explained the ordering pattern for 19 of 20 analytes, consistent with those centers never being subjected to the charge. The instant effect of cost display was scattered: it immediately reduced orders for 3 of the 10 expensive tests in set 1, left 4 unchanged, and was actually associated with increases in carbohydrate-deficient transferrin and methylmalonate. In the cheaper, common-issue set, only plasma glucose dropped immediately, while five analytes, including alanine aminotransferase, cholesterol, creatinine, potassium and sedimentation rate, increased. Over the long term, though, the picture clarified: the intervention effect on the change rate was reduced for 5 of 10 expensive analytes, and only three tests, S-CDT, cortisol and homocysteine, saw their increase rates escalate further.

In privately operated primary care, where both display and charge applied, the combined model best explained ordering for nearly all analytes in the common-issue set. Cost charge instantly reduced orders for 3 of 10 expensive tests, and in the common-issue set it reduced every analyte except potassium. The long-term effects diverged across both sets, but the interventions reduced the change rate for a majority of analytes, with the effect more pronounced among the expensive tests. Notably, the increase rate of vitamin D orders, a test whose overuse has been documented before, was reduced by the interventions, saving what the authors describe as considerable laboratory and healthcare resources. There were no incident reports related to the interventions anywhere in Region Kronoberg during the entire period.

Why would some tests briefly increase in the instant aftermath of seeing a price? The authors speculate that cost display might have created an anticipatory ordering effect: if prices turned out lower than physicians expected, orders could rise, and the displayed price might even have inadvertently signaled value or importance to some clinicians. No concurrent change in regional screening protocols or testing availability occurred, so the explanation remains open. The more durable pattern, they argue, has a plausible logic. Expensive tests tend to be ordered to diagnose specific medical issues, and when clinical suspicion is weak the test is easier to forgo. Cheap tests, by contrast, are often used as broad screening and are frequently bundled into panels; the county had roughly 250 panels during the study period, most built around low-cost analyses. Cost awareness among healthcare practitioners is generally limited, so seeing a figure may be enough to curb the pricier requests.

The authors are candid about the limits of what their intervention achieved and what their study can show. No comprehensive data on clinical outcomes related to the intervention were obtainable, because the healthcare system is too complex with too many confounders. The design lacked an external control group, and an adjacent county could not serve as a comparator because every Swedish county is autonomous with widely discrepant economic systems, which would introduce numerous confounders. Medical practices change constantly, although no major structural or organizational changes or guideline revisions occurred during the study that would solely explain the results. There are also clinical questions worth asking: homocysteine testing has been recommended to increase in favor of vitamin B12 and folate testing, a trend visible in secondary care but not primary care, while ferritin, a key test for diagnosing iron deficiency, the most common nutrient deficiency worldwide, decreased in primary care. The absence of incident reports, the authors suggest, may indicate the interventions adjusted test numbers toward an appropriate level rather than eliminating clinically relevant testing. Their conclusion is measured: cost display and cost charge can be implemented to reduce laboratory testing in both the short and long term, but the effect depends on the setting and the cost of the test, and further studies are needed to establish the full extent of the savings.

Subject of Research: Long-term effects of cost display and cost charge on clinical chemistry laboratory test ordering in Swedish healthcare

Article Title: The effect of cost display and cost charge on laboratory test requests

Article References: Ekblom, K., & Petersson, A. (2026). The effect of cost display and cost charge on laboratory test requests. Discover Health Systems, 5(1), Article 88. https://doi.org/10.1007/s44250-026-00432-w

Image Credits: AI Generated

DOI: 10.1007/s44250-026-00432-w

Keywords: laboratory testing, cost display, cost charge, interrupted time series, primary healthcare, secondary healthcare, Kronoberg County, clinical chemistry, test ordering behavior, health economics, Poisson regression, Sweden