Impact of an Electronic Medical Record on Diabetes Quality of Care

O'Conner & Colleagues are to be commended for an excellent study of the impact of the electronic health record (EHR) on diabetes care. Lacking a prospective, randomized prospective trial, the authors did the next best thing: took advantage of a staggered roll-out of the intervention, using a parallel clinic site that - by System (HealthPartners) and patients - appeared to differ only by the presence or absence of the EHR. This 'real world' study found an association of higher frequency of A1c testing with the EHR but not overall glycemic control. In fact, there was evidence that the implementation of the EHR in the 1st year may have disrupted clinic operations enough to cause a transient increase in the A1c. Combine these findings with the relative dearth of any controlled studies that show the EHR results in greater health care quality, and physicians may begin to wonder if the capital investments required to support an EHR will truly result in any benefit to their patients.

Is all lost? Probably not:

Not all clinical operations are predicated on scientific evidence of improvement in clinical outcomes. For example, the EHR may result in certain marketing advantages, better documentation, more accurate coding and more efficient information retrieval by physicians (including consultants) and patients (who may be able to access their information on- line). In other words, there are potential advantages in other health care domains that may make the capital investment worth it. Operative word is potential: we have a long way to go in better understanding this.

Secondly, the implementation of this EHR in this setting was focused on the physicians. Our colleagues may find that a natural state of affairs, but that paradigm may be stifling. Basically, it makes little difference if the documentation, prompts/reminders and ordering is via paper or keyboard. But what if patients can enter their data themselves PRIOR to seeing the physician, if patients are prompted to ASK what their A1c is and if ordering is automated so that every person with a diagnosis code of 250 is scheduled for an A1c, and A1cs > 7 automatically lead to a suite of 'mass customization' options ranging from an educational email to a face to face appointment with a diabetes educator? The authors hint at this EHR ver 2.0 with discussion about decision support algorithms and patient activation.

Physicians have probably reconciled themselves to the fact that disruptive information technology is coming to the bedside and their clinic exam rooms. We need to better understand what it can do and can't do. Right now, based on the findings to date, it would appear the EHR by itself is not enough to solve the twin challenges of quality and cost. However, our task is to understand how to solve quality and cost with all tools at hand - including the EHR.

Jaan Sidorov, MD Medical Director, Care Coordination Geisinger Health Plan Danville PA

Competing interests:   None declared

Your study is a very valuable addition to support the increasing use of EMR, and demonstrating the obvious (process alone is not enough. Your work will help us avoid less than stellar initiatives in the future.

Have you had any experience with the Chronic Care Model, e.g. Wagner, with a more patient centered focus?

We are designing a system which will allow MD and patient access to an internet-based portal. And would like to know about the attempts of others.

Competing interests:   None declared

It has been shown by others already that simply knowing the guidelines and having an EMR are not enough to "translate research into practice." Academic detailing, in which a staff is brought to common ground of understanding (and accepting) the guidelines, regular reports of progress and, most important in our offices experience, involving and motivating an entire clinic staff results in better performance on clinical benchmarks such as diabetes control. THis generally is a more satisfying experience anyway and usually results in more effective use of the tools that an EMR provides.

Competing interests:   None declared

This excellent study by O’Connor et al yields at least three very important lessons for investigators engaged in research to translate clinical evidence into clinical practice (the so-called “second translational block”, with the first translational block from bench to bedside)(1):

First, the day of the “before-and –after” study design for quality improvement in diabetes is over. Studies using informatics applications or other strategies to improve diabetes management must be performed with simultaneous controls (ideally with the intervention randomly assigned). This study by O’Connor joins an increasing number of well-designed controlled trials that have found little impact of various informatics- based disease management strategies above and beyond temporal changes in care among control groups (2-4).

Second, interventions that are designed primarily to guide clinical decision-making at the time of a clinic visit (e.g. reminders and computerized decision support) may improve testing rates but rarely have any impact on risk factor levels. This may be a classic example of looking for one’s lost keys under the street lamps: electronic medical records are best suited for providing information, but what we really need are interventions that actually lead to changes in medication prescription. This may require venturing out towards system-level or behaviorally- oriented interventions to change physician practice patterns, particularly in the area of clinical inertia alluded to by Dr. Phillips.

Third, we need to think beyond the narrow confines of the clinic visit to take a population-based perspective. Simply put, it is very difficult to make medication changes to lower HbA1c for a patient who hasn’t been seen in clinic for 12 months, or who when last seen was primarily concerned with dealing with other more pressing issues. Face-to- face time is brief, so future interventions must also find ways to allow physicians to efficiently monitor and manage their panel of patients that go beyond the traditional clinic visit (5).

(1) Garfield, SA, Malozowski, S, Chin, MH, Venkat Narayan, KM, Glasgow, RE, Green, LW et al. Considerations for diabetes translational research in real-world settings.Diabetes Care. 2003 Sep;26(9):2670-4.

(2) Grant RW, Cagliero E, Nathan DM, Singer DE, Chueh H, Meigs JB.et al. A Controlled Trial of Population Management: Diabetes Mellitus - Putting Evidence into Practice. Diabetes Care, 2004 27:2299-2305.

(3) Murray MD, Harris LE, Overhage JM, Zhou XH, Eckert GJ, Smith FE, et al Failure of computerized treatment suggestions to improve health outcomes of outpatients with uncomplicated hypertension: results of a randomized controlled trial. Pharmacotherapy. 2004 Mar;24(3):324-37.

(4) Tierney WM, Overhage JM, Murray MD, Harris LE, Zhou XH, Eckert GJ et al. Can computer-generated evidence-based care suggestions enhance evidence-based management of asthma and chronic obstructive pulmonary disease? A randomized, controlled trial. Health Serv Res. 2005 Apr;40(2):477-97.

(5) Lester WT, Grant RW, Barnett GO, Chueh HC. Facilitated Lipid Management Using Interactive E-mail: Preliminary Results of a Randomized Controlled Trial. Medinfo. 2004;2004:232-6.

Competing interests:   None declared

Diabetes is already a major public health problem in the United States and the future U.S. diabetes burden is expected to increase dramatically. (1) Fortunately, however, several effective treatments to reduce or prevent diabetes-related complications are available (2), and much effort has gone into the development of evidence-base guidelines for delivery of these treatments. However, the existence of a guideline does not ensure delivery of quality care, and in spite of much effort, the quality of diabetes care remains suboptimal in this country (2), regardless of a particular health care system, or population, or geographic location.

In this issue, O’Connor et al reports on the use an electronic medical record (EMR) intervention to improve the quality of care for persons with diabetes. (3). While the intervention found improvement in the process of accessing risk factor control care, this improvement did not result in better risk factor control (i.e., in lower A1C and lipid levels). There currently is much debate about whether act of measuring a risk factor leads to better risk factor control. Further study is needed.

The measurement of quality indicators is important, not only because it can help health care systems focus their efforts more efficiently, but also because the process of measuring these indicators and reporting results can draw much needed attention to the quality improvement efforts. The old adage, “what gets measured, gets done” may apply in this case. Quality indicators can also be very useful outcome measures in as the case for O’Connor et al’s study. We have a long way to go toward getting optimal care for persons with diabetes - but interventions that improve the quality of diabetes care will help get us there.

Michael M. Engelgau, MD, MS, Associate Director, Division of Diabetes Translation, CDC

Disclaimer: The findings and conclusions in this report are those of the author(s) and do not necessarily represent the views of CDC.

1. Engelgau MM, Geiss LS, Saaddine JB, Boyle JP, Benjamin SM, Gregg EW, et al. The evolving diabetes burden in the United States. Ann Intern Med 2004;140:945-50.

2. Narayan KMV, Benjamin E, Gregg EW, Norris SL, Engelgau MM. Diabetes translation research: where are we and where do we want to be? Ann Intern Med 2004;140:958-63.

3. O’Connor PJ, Crain AL, Rush WA, Sperl-Hillen JM, Gutenkauf JJ, Duncan JE. Impact of electronic medical record on diabetes quality of care. Ann Fam Med 2005;3:300-6.

Competing interests:   None declared

We were pleased to see that an electronic health record could act effectively as a system for reminding clinicians regarding proper testing as well as proper intervals for testing. As your study so aptly demonstrates, this can work very effectively with an electronic system. We would submit that actually affecting the parameter, in this case hemoglobin A1C is certainly more difficult to achieve and is dependent on many more factors than just testing frequency. We are optimistic rather than disappointed regarding this result and believe that computerized healthcare information technology solutions still have a long way to go regarding functionality and sophistication. In addition as with any new method of taking care of patients, providers and their offices will gradually come up with very effective ways to improve the quality of patient care.

With very kind regards,

Physician Staff at the Center for Health Information Technology David C. Kibbe, MD MBA Louis Spikol, MD Steven Waldren, MD

Competing interests:   None declared

O’Connor and colleagues address an important and timely topic—do EMR reminders cause clinicians to treat chronic illness more appropriately? Opportunities to observe controlled comparisons of EMR implementation are rare and valuable. O’Connor et al. deserve our thanks for making the most of one such opportunity.

Despite the technical limitations of O’Connor et al.’s report—very small patient sample size, only quasi-experimental control, just one clinic per treatment arm, and an ADA treatment-threshold recommendation (8.0% HbA1c) that, until 2004, was well above the mean HbA1c in the study sites--I strongly agree with the authors’ conclusion that we need more sophisticated clinical decision support and more potent patient activation applications than crude alerts provide. At Kaiser Permanente Northwest, which has had the Epic system that O’Connor evaluated since 1995, and predecessor systems since the mid-1980s, we are now preparing to pilot a computer simulation-based evidence integrator to calculate the most valuable clinical actions for each individual patient with diabetes. This system also will give clinicians graphic displays of the impacts of treatments that can be shown to patients, plus the capacity to order all associated prescriptions, tests, referrals and messages with a single mouse-click. In a clinical setting in which so many alerts and reminders are now built into the EMR that many clinicians claim to ignore all of them, front-line primary care clinicians view this new approach with genuine enthusiasm. Consistent with O’Connor et al.’s conclusions, what they especially like is its potential to (1) end-run the logjam of alerts, single-purpose guidelines, and hard-to-personalize trial data and (2) get patients to agree to treatment.

Kaiser Permanente’s inter-regional experience also supports O’Connor’s final observation that, “in the absence of more advanced EMR capabilities, less expensive and less disruptive care-improvement strategies may improve chronic disease care as effectively as EMRs.” At various times, KP regions have made abrupt improvements in HbA1c, lipid levels, retinal screening rates and even “hard” outcomes like cardiovascular event rates using paper records, person-to-person contact, and other non-electronic means. However, data from computer systems were hugely useful in supporting all of these approaches. The limitations of crude electronic alerts and reminders should not blind us to the enormous total value of EMRs, or delay us in achieving what I see as the next big stage of evidence-based medicine, personalized (and ultimately genetically informed) treatment recommendations calculated from data archived by EMRs.

Competing interests:   None declared