Most quality indicators and guidelines are aimed at ensuring that patients receive an appropriate minimal level of testing and treatment, rather than curbing potentially unnecessary testing. Recently, however, as part of a quality assessment initiative, Veterans Administration (VA) researchers explored repeat hemoglobin A1c testing across the VA Healthcare System. Their findings are the subject of this issue of Strategies.
The American Diabetes Association (ADA) and other professional organizations for some time have recommended using hemoglobin A1c (HbA1c) measurements to monitor and manage the treatment of diabetes. Given that HbA1c values reflect average glycemia over several months, determining HbA1c values more frequently than that is potentially wasteful in patients with stable disease. Indeed, ADA in its 2011 Standards of Medical Care in Diabetes recommends only twice-yearly HbA1c testing in patients who are meeting treatment goals and have stable disease, and quarterly testing for those whose therapy has changed or who are not meeting glycemic goals (Diabetes Care 2010;34:S 11-S61).
In this context, VA researchers sought to examine several aspects of outpatient HbA1c testing across the entire VA Healthcare System, which has focused extensively on the quality of ambulatory care and has issued clinical practice guidelines on diabetes (Am J Med 2011;124:34-9).
The investigators looked at the distribution and frequency of HbA1c tests among newly diagnosed diabetics, and how this testing correlated with ADA and VA clinical practice guidelines. “Hemoglobin A1c has received a lot of attention at VA as a clinical quality indicator for diabetes, so it was only natural to choose it as one of the lab tests to monitor,” explained first author Archana Laxmisan, MD, MA, instructor of general internal medicine at Baylor College of Medicine in Houston. “The guidelines on hemoglobin A1c testing have been concerned with whether diabetics are not getting the care they need. They haven’t looked at the flipside of whether there is an overuse of recommended tests. But in the current climate of healthcare, that’s the kind of question we have to ask— are we ordering more tests than necessary?” Laxmisan conducted the study during the time she was a VA quality scholar at the Center for Research in the Implementation of Innovative Strategies for Practice at the Iowa City VA Medical Center.
Laxmisan and her colleagues performed a retrospective analysis of more than 1 million patients with a primary or secondary diagnosis of diabetes who were seen in VA clinics during fiscal year 2006. The data came from the VA’s nationwide electronic health record, which includes a laboratory results dataset. Because they were only interested in newly diagnosed diabetics, the researchers excluded patients with a diabetes-related visit during the preceding 2 years, as well as those who had no visits over the same period. In all, 130,538 patients met the study’s inclusion criteria. The primary outcome measures were the proportion of patients receiving repeat HbA1c testing within 30 and 90 days of an initial test. Secondary outcomes included the distribution and frequency of repeated HbA1c tests in 365 days, and the proportion of patients who received more than four repeat tests within 365 days of their initial HbA1c results.
The researchers found that overall, approximately 62% of patients had an initial HbA1c value <7%, 26% had results between 7% and 9%, and 13% had initial concentrations >9%. While 28.8% of patients had no repeat HbA1c tests at all within 365 days of their initial test, 4.2% had more than four repeat tests. In addition, 8.3% of patients had at least one repeat HbA1c test within 30 days of their first test, while 30.8% had a minimum of one repeat test within 90 days of the index test.
As one might expect, patients with higher initial HbA1c results ≥9% had repeat testing—15% within 30 days and 48% within 90 days. However, repeat testing was common even among patients with well-controlled diabetes with HbA1c levels <7%. For example, 26% of patients with well-controlled disease had repeat testing within 90 days and 7% had repeat testing within 30 days. The researchers also found that older patients and African Americans were less likely to receive repeat HbA1c tests. In the case of elderly patients, after multivariate adjustment, the odds ratio of receiving at least four tests in 1 year rose, suggesting they have less aggressive testing due to increased morbidity, according to the authors. In contrast, the odds ratio for African Americans declined after multivariate adjustment. “That finding wasn’t a primary outcome of our study; it was a finding that supported other studies suggesting there is an under-utilization of diabetes care in this population,” observed Laxmisan.
The study provided important baseline information and points out the need for additional guidance and controls around test ordering, according to Laxmisan. “The VA has an electronic health record, but it doesn’t have certain clinical decision support features that might have prevented this. An example would be a simple alert that informs the physician that a hemoglobin A1c test was ordered 30 days ago and asks whether he or she still wants to order it,” she explained. “That’s a fairly simple rule to implement and it probably would have stopped the ordering of those tests that were unnecessary.”
David Bruns, MD, agreed that this type of alert strategy can be successful in curbing unnecessary test ordering, and is one he spearheaded at the University of Virginia School of Medicine in Charlottesville, where he is professor of pathology, director of clinical chemistry, and associate director of the molecular diagnostics laboratory. “We cut down ordering of complete blood count with differential by about two-thirds simply by inserting a note in the electronic health record that pops up when someone tries to order the test within the 24 hours following a previous test. It informs them that the test was run within 24 hours and is not needed,” he explained. “The physician order entry system also had an option to order a CBC with differential test daily for X number of days, which we removed. It took several months for us to plan this intervention, but once we implemented it, our workload for this test plummeted and has remained at that level for years without further intervention.” Bruns was not involved in the study.
He also pointed out that the findings from a separate study by different researchers published in 2007 probably are generalizable in the U.S. outside the specific health system in which it was carried out. This analysis of ordering patterns in a non-VA health system revealed an almost identical rate of 8% of tests repeated within 30 days and a similar rate of repeat testing within 60 days (Clin Chim Acta 2007;378:201-5).
Bruns cautioned, however, that any alert strategy would need to be carefully planned and implemented. “A pop-up alert can be effective just so long as it doesn’t aggravate physicians too much. So you have to decide if it’s a high enough priority to be worth doing. No intervention like this is free,” he observed.
While Bruns said he hoped the study would spur laboratorians to look at ordering patterns in their own institutions, he emphasized that a certain amount of repeat HbA1c testing might be unavoidable given concerns about biological and analytical variations involved in glucose testing. “Some of these tests probably were repeated because the physician didn’t trust the results. A patient could have come in with blood glucose measurements from a meter, which the doctor felt didn’t match the first hemoglobin A1c result, so he or she ordered a repeat hemoglobin A1c test close in time to the first test,” he said.
Laxmisan explained that the nature of the study didn’t enable the research team to determine precisely why repeat testing occurred, but she speculated that if such a pattern existed in the VA system it was likely to be found elsewhere. “We can only guess the clinical reasons why so many of these tests happened as frequently as they do,” she said. “However, the VA has put a fair amount of scrutiny into its own practices. So while our findings may underwhelm some people who’ll say, ‘well, you only had so much repeat testing’, the fact that we found so much with the VA system makes us strongly suspect that there’s probably more overutilization of testing in the private setting.”