• :• ' N A L A R T I C L E A Multicenter Evaluation of Blood Glucose Awareness Training-ll DANIEL COX, PHD LINDA GONDER-FREDERICK, PHD WILLIAM POLONSKY, PHD DAVID SCHLUNDT, PHD DIANA JULIAN, MA WILLIAM CLARKE, MD OBJECTIVE — Blood glucose awareness training (BGAT) teaches individuals with insulin-dependent diabetes to more accurately estimate/detect their blood glucose (BG) fluctuations. It has not, however, consistently resulted in improved ability to detect low BG. To assess an enhanced version of BGAT (BGAT-II), with more focus on increasing sensitivity to low BG events, a multicenter study was undertaken. Following up on previous findings that BGAT is most effective with individuals who are least accurate in estimating BG, this study explicitly recruited subjects who did and did not report reduced awareness of hypoglycemia. RESEARCH DESIGN AND METHODS— Seventy-eight subjects from three research sites participated in a repeated baseline design. Subjects' BG estimation accu- racy and BG profiles were assessed 6 months before, immediately before, and imme- diately after BGAT-II. RESULTS — Post-treatment, BGAT-II led to better overall accuracy in detecting BG fluctuations and better detection of both low and high BG levels. This was achieved while the number of low readings of self-monitoring of blood glucose (SMBG) was reduced. Reduction in the number of low SMBG events was significant only for sub- jects reporting awareness of hypoglycemia. Detection of low BG was significant only for subjects reporting reduced awareness of hypoglycemia. Both groups demonstrated equivalent improvements in detection of high BG levels. CONCLUSIONS— BGAT may be an effective behavioral strategy for reversing hypoglycemic unawareness and an adjunct to intensive insulin therapy to reduce the occurrence of severe hypoglycemia. A series of studies (1-8) has found that blood glucose awareness training (BGAT) effectively teaches individuals with insulin-dependent dia- betes mellitus (IDDM) to more accurately recognize blood glucose (BG) fluctuations From the University of Virginia Health Sciences Center (D.C., L.G.-F., DJ, W.C.), Charlottes- ville, Virginia; Joslin Diabetes Center (W.P.), Boston, Massachusetts; and Vanderbilt University Medical School (D.S.), Nashville, Tennessee. Address correspondence and reprint requests to Daniel J. Cox, MD, Box 223, University of Virginia Health Sciences Center, Charlottesville, VA 22908. Received for publication 7 July 1994 and accepted in revised form 22 December 1994. AVOVA, analysis of variance; BG, blood glucose; BGAT, blood glucose awareness training; SMBG, self-monitoring of blood glucose. through symptom perception. The origi- nal BGAT involved both educational ma- terials on symptoms of BG fluctuations and homework exercises. Homework in- volved rating experienced symptoms, es- timating BG level on the basis of these symptoms, performing self-monitoring of blood glucose (SMBG), and then plotting estimated/actual BG readings on an error grid (9) to provide immediate feedback concerning estimation accuracy. While BGAT has led to consistent improvement in overall BG estimation accuracy, it has not specifically improved detection of ex- treme BG levels (<3.9 and >10 mmol/1). We recently followed up past BGAT sub- jects and, in comparison with control subjects, found that past BGAT subjects had fewer automobile crashes and had sustained improvement in metabolic con- trol. In addition, past BGAT subjects who were given booster training demonstrated better detection of low BG events com- pared with past BGAT subjects who did not receive booster training (10). These data have potentially significant implica- tions for patients undergoing intensive therapy, who are at increased risk of se- vere hypoglycemia. In an attempt to enhance BGAT and to more thoroughly evaluate its im- pact, we conducted a multicenter evalua- tion. BGAT-II involved seven 1.5-h classes that followed a standardized train- ing manual. This manual has an introduc- tory chapter, three chapters on internal cues (autonomic, neuroglycopenic, and affective symptoms), and three chapters on external cues (timing, amount and type of insulin injections, food consump- tion, and exercise performance). BGAT-II differed from our original BGAT in several respects. The original manual was 74 pages, while the BGAT-II manual was 132 pages. While the original manual dealt with external cues in a single chapter, the new manual devoted separate chapters to insulin, food, and exercise. The original manual was written in 1985, and the new manual was written in 1992 and updated with all new information. The new man- DIABETES CARE, VOLUME 18, NUMBER 4, APRIL 1995 523