Impact of Electronic Health Record Clinical Decision Support on Diabetes Care: A Randomized Trial

I read with interest this study as it provides useful insight into the effectiveness of clinical decision support for physicians.

It is not clear if this is a study with a positive or neutral result. The authors state towards the end of the Results section: “The intervention had no significant positive or negative impact on diastolic blood pressure and LDL cholesterol values or proportion remaining in control for hemoglobin A1c, diastolic blood pressure, or LDL cholesterol values.”

The improvement in HbA1c is difficult to celebrate as the confidence interval comes within a hairline of no effect (0.06%).

The large and comparable improvements in the no-intervention arm underline the need for control groups in studying these new and in vogue technologies to fix the healthcare system.

Testing multiple endpoints in clinical trials has been discouraged by statisticians and research methodologists. Furthermore, it is difficult to unravel the effect of individual intervention components (the CDSS, workflow re-engineering and financial incentives) in a study with concurrent multiple interventions. Statistical analysis of such a complex study design with cluster randomization is formidable.

Overall, the authors’ effort is commendable and supports the need for caution and more research in this area of emerging health information technology.

Competing interests:   None declared

The paper by O’Connor et al. presents some welcome positive results that provide evidence that computerized clinical decision support can assist clinicians in their efforts to improve diabetes care. As the authors point out, prior studies have focused on process outcomes, such as the rate of LDL cholesterol testing, so this study of intermediate clinical outcomes is especially promising and important. It is admirable that the authors recognized that it would take more than the technological intervention itself to improve outcomes. The four implementation strategies—patient specific clinical decision support (CDS), changes in staff responsibilities, timing of the reminders, and financial incentives—in addition to training nurses and physicians, were well planned and likely contributed to the success of the project. The degree to which attention is paid to these organizational issues has been shown to be critical for clinical systems implementation in general. For example, the study by Han et al. (1) indicating increased mortality after implementation of computerized provider order entry (CPOE) on one hospital unit was conducted at a site that clearly did not pay sufficient attention to these issues. (2) Two subsequent studies in similar hospitals using the same technology showed no increased mortality (3) and decreased mortality (4) when careful planning, system configuration, and organizational change management were accomplished.

We should celebrate the positive results here, but at the same time question why they were somewhat modest and how the implementation could have been done even better. As qualitative researchers in medical informatics, we believe that asking the “why” questions is as important as studying outcomes if we are to inform future implementations. For example, the subjects, if interviewed, could have told the researchers the value of the training, why they only used the application in 62% of the visits, to what extent compensation influenced both physicians and nurses, and how the entire system could have been improved. In those cases in which the system was used, the level of satisfaction was remarkably high and again, it would be helpful to know why.

Looking to the future, the potential for CDS interventions for diabetes care will continue to grow as more primary care provider offices adopt and use electronic health record (EHR) systems. The Diabetes Wizard application described here is sophisticated in that it includes detailed clinical algorithms for providing patient-specific decision support, but the physician receives a printout of the recommendations and, after the visit, completes a form, a process that should be unnecessary in this setting, which has an EHR. If the Diabetes Wizard were used in the exam room, as is often done these days, it could serve as a way of engaging the patient in a three way doctor/patient/computer conversation. In addition, if the Wizard had provided “offered choices” that allowed the clinician to simply click on the Wizard’s suggested action to complete the order, clinicians might have viewed the application as an efficiency aid and used it much more frequently. Again, with careful attention to implementation planning, provider training, and user-centered information system design, such conversations can improve the quality of the encounter. (5,6) The potential, especially for diabetes patients, is exciting, and warrants a great deal more study.

1. Han YY, Carcillo JA, Venkataraman ST, et al. Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics 2005;116(6):1506-1512. 2. Sittig DF, Ash JS, Zhang J, et al. Lessons from unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics 2006;118(2):797-801. 3. Del Beccaro MA, Jeffries HE, Eisenberg MA, Harry ED. Computerized provider order entry: no association with increased mortality rates in an intensive care unit. Pediatrics 2006;118(1):290-295. 4. Longhurst CA, Parast L, Sandborg CI, et al. Decrease in hospital-wide mortality rate after implementation of a commercially sold computerized physician order entry system. Pediatrics 2010;126(1):14-21. 5. Shield RR, Goldman RE, Anthony DA, et al. Gradual electronic health record implementation: new insights on physician and patient adaptation. Ann Fam Med 2010;8(4):316-326. 6. Johnson KB, Serwint JR, Fagan LA, et al. Computer-based documentation: effects on parent-provider communication during pediatric health maintenance encounters. Pediatrics 2008;122(3):590-598.

Competing interests:   None declared