There is a substantial body of evidence documenting performance gaps related to appropriate monitoring of laboratory tests in the presence of medication use, as well as the poor outcomes associated with such shortcomings. Our study found only a relatively small proportion (approximately 5%) of patients on targeted medications were overdue for appropriate laboratory monitoring at the time of office visits with primary care physicians, and that nearly half of these patients had appropriate testing within two weeks of the office visit. Electronic reminders to physicians were not effective in increasing the rates of appropriate laboratory monitoring in this setting.
The prevalence of overdue routine laboratory monitoring among patients in this study was low compared to previous work in this area, where other investigators have documented much higher rates of overdue creatinine testing with non-steroidal anti-inflammatory medication use (30%) and metformin use (52%). 15,16
These low rates of overdue testing may relate to the fact that our health system employs a unified clinical data repository across all outpatient and inpatient settings, and thus captures lab values from episodes of clinical care that may not be available in settings that lack such a comprehensive system. Alternatively, the low overdue rate might reflect the benefits of an electronic medical record system that increases access to patient information such as up to date medication lists. While we did not collect additional data on redundant testing, it is important to note that these high rates of testing likely come at the expense of over-utilization, as prior studies have documented rates of redundant inpatient laboratory testing within our system of nearly 10%. 37
While a number of studies have shown that computerized reminders are efficacious in the inpatient setting, their use in the outpatient setting is less well described. A few randomized trials have shown improvements in ordering appropriate laboratory tests upon medication initiation with the use of outpatient computerized reminders. 38–40
However, only one previous randomized trial has evaluated the effect of outpatient computerized reminders on recommended routine monitoring. In an Israeli healthcare provider network, approximately 20% of patients taking diuretics were overdue for an annual potassium check, and reminders were effective in increasing monitoring compliance. 41
Barriers to provider utilization of clinical reminders are limitations of our study that might have contributed to the negative findings. For example, our previous work indicates that only one-third of providers in our system notice reminders during clinical encounters. 25
The reminders are visible from the patient summary screen, which is a common portal for physicians. However, since our reminders are passive, it is possible for a clinician to not use this screen and thereby circumvent the reminder system. Physicians may also be less likely to utilize reminder systems as workload increases. 42
While inaccuracy of recommendations is often cited as a barrier to effective clinical decision support tools, 43
this was unlikely in our system as we were able to capture both laboratory results and medication prescriptions in a relatively complete manner. Other studies have noted decreases in the use of reminders over the course of a study, attributed to reminder fatigue. 44,45
Clinical content could also have limited the utility of the reminders if physicians felt that the recommendations were unnecessary, or if the interval was too frequent. In order to address these issues in our evaluation, we chose a conservative monitoring interval of one year and drug-lab recommendations in which there were clear FDA concerns or significant adverse effects from a lack of monitoring. Laboratory test values obtained outside our health system were not electronically available, but could have been known by providers. This could falsely increase overdue rates and contribute to apparent reminder non-compliance by displaying reminders for labs appropriately performed elsewhere. However, it is important to note that the vast majority of our patients' labs are performed within the Partners system, and would therefore show up in our Central Data Repository. In addition, patient non-compliance is included in reminder non-compliance, since only ordering a laboratory test is insufficient to fulfill guideline compliance. Finally, clinicians who viewed the reminders still needed to order the lab tests in a separate step using a paper lab ordering form. This additional step could have been an obstacle to taking action based on the reminder.
A final explanatory mechanism for our failure to demonstrate an effect with the reminder system is the ceiling effect associated with the high baseline rates of adherence. However, a post-hoc power calculation using higher rates of compliance (90% baseline rate, detecting 5% absolute difference) required 582 clinic visits per arm to achieve 90% power, suggesting adequate power among the reminders in this study to detect such a treatment effect.
In summary, we found relatively high rates of appropriate laboratory monitoring in our study population, and an electronic reminder system did not significantly improve rates of such testing. Future studies should focus on laboratory monitoring for different classes of medications, laboratory monitoring at medication initiation, and settings with lower rates of baseline adherence to monitoring recommendations.