Optimizing the use of laboratory tests – A study on the applicability and efficacy of different interventions to improve the appropriateness of laboratory test requests in primary care

Laboratory tests are important for screening, diagnosis, and monitoring. Millions of tests are requested each year for these purposes. However, in recent decades, their use has become increasingly disproportionate to the clinical need. In the Region of Southern Denmark (RSD), primary care is responsible for approximately half of the tests requested at biochemistry laboratories, and there are indications that they are inappropriately overused or underused. Both situations lead to unnecessary costs in the healthcare system and can potentially harm the patients.

The aim of the present Ph.D. thesis was to:  

 Provide a comprehensive and systematic review of studies on interventions to improve the use oflaboratory tests in primary care and assess their risk of bias.
 Evaluate the effectiveness and appropriateness of four different interventions, either individually orin combination, in a primary care setting using vitamin D as the model test.
 Evaluate the long-term effect of these interventions.
 Explore the GPs’ evaluation of the interventions. 

We performed a systematic review covering the period 1947–2023, including studies that usedinterventions based on three different categories. The first was based on an educational component, whichincluded studies that consistently reduced the number of tests in the range of 5–35%, with a mainly lowrisk of bias, as evaluated by the Cochrane Effective Practice and Organization of Care (EPOC). The secondwas based on administrative changes, such as studies that updated test profiles or designed new requestforms; these studies achieved greater reductions (up to 94%) but were of medium or high risk of bias. Thethird was studies that used IT-based strategies, like soft blocking and hard blocking. The use of soft blockingstrategies in primary care has increased in recent years; the overall quality of the studies has generally beengood, with interventions that have achieved reductions ranging from 13–46%. Recent studies have focusedon hard-blocking interventions in primary care; however, these studies have primarily been classified ashaving a medium or high risk of bias, indicating the need for further evaluation. 

In 2020, we implemented a large, randomized study. GP clinics in the RSD were randomly assigned to oneof seven groups, one of which was a control group. The remaining six groups applied the followinginterventions, alone or in various combinations, for one year: distribution of guidelines, feedback reports,non-interruptive alerts, and/or interruptive alerts. As previously mentioned, vitamin D was used as a testmodel, owing to its extensive use in primary care and because an official guideline of its use was available. 

We found that pop-up alerts resulted in the greatest reduction of ordered vitamin D (up to 46%),irrespective of the combination with feedback reports; the use of guidelines and feedback reports did notcause any significant change in the number of ordered tests compared to the control group. Furthermore,we evaluated the variance in the number of tests between the clinics in each group and the changes in thenumber of results that reported sufficient (> 50 nmol/L) and insufficient (≤ 50 nmol/L) levels of vitamin D.We found that most clinics, independent of their pre-intervention level of ordered vitamin D tests, tendedto reduce the number of tests used in the interventions. None of the interventions increased the absoluteor relative number of low vitamin D results, which would have been expected if the test had been usedmore appropriately because of the interventions.

In the first and second follow-up years, we observed a significant increase in vitamin D requests across allthe groups, including the control group. However, the effect of the interventions remained visible in threeout of the four groups that used pop-up alerts, as they were still significantly lower compared to the controlgroup in the second follow-up year. Regardless of the intervention received, the relative numbers ofvitamin D results (<50 nmol/L), increased in the follow-up years. Furthermore, the odds of the doctors tofind patients with a low level of vitamin D increased significantly in all the groups including the controlgroup when the second year of follow-up was compared to the background period (ranging from 20–42%).  

During the final week of the intervention, an electronic survey was sent to the 638 GPs who were includedin the intervention groups. Despite reminders, the responsive rate was only 21% but we considered itrepresentative, as we found no significant differences between the respondents and the non-respondentsin their gender, age, and type of clinic. 

The survey demonstrated that approximately half of the GPs found the guidelines helpful, and a similarproportion “always” or “often” read the feedback report. However, as mentioned above, theseinterventions did not result in either a reduction in the number of tests or more appropriate use of thetests, as judged by the fraction of low results. The pop-up alerts, which significantly reduced the number ofordered tests, were accepted by the GPs if they were displayed for a short period for frequently used tests,and long periods or forever for rarely used tests. Nevertheless, the findings suggested that pop-up alertsshould be used with caution to avoid alert fatigue. 

The findings of this thesis indicated that pop-up alerts can significantly decrease the number of laboratorytests requested by GPs and are well-accepted by GPs. However, there is a risk of alert fatigue if they areused too much. Therefore, it may be appropriate to use short display periods for commonly used tests andlong to permanent display periods for rarely used tests. Guidelines and feedback reports in the format weused did not have a significant impact on test numbers, but half of the GPs still found them helpful and regularly read them. Unfortunately, none of the interventions used in the study improved theappropriateness of test requests, during the intervention period, as the absolute number of low levels ofvitamin D also decreased.

None of the interventions was successful in reducing the number of vitamin D requests in the follow-upyears; however, all groups, including the control group, demonstrated increased appropriateness. Thisobservation cannot be attributed to the interventions but was probably due to an unspecific effect of thegreater focus on the use of vitamin D tests caused by this large and long-lasting study. Further researchshould focus on interventions that are effective and considered helpful by GPs. We suggest that GPs shouldbe provided with tools that improve patient selection, such as improving requesting profiles, sendingfeedback reports that include relevant clinical information, and displaying pop-up alerts for a short time forcommonly used tests.