We anticipated that our consensus algorithm (1) would generate some controversy, but we are pleased with the general level of appreciation expressed in the letters, albeit with some disagreements. Dr. Cryer (2) specifically endorses the recommendation in our consensus algorithm to use insulin earlier in the treatment course of type 2 diabetes but takes issue with the relatively low frequency of severe hypoglycemia that we cited for insulin-treated type 2 diabetes, which was defined in accordance with the Diabetes Control and Complications Trial and compared with the rate in type 1 diabetes. As Dr. Cryer notes, our estimates were based on data from “clinical trials aimed at normoglycemia and achieving a mean A1C of ∼7%.” Dr. Cryer cited review articles (including some referenced by us) and other empiric studies (uncontrolled clinical trials) that suggested a much higher risk for severe hypoglycemia in insulin-treated type 2 diabetic patients than we described.
The reasons that we chose data from controlled clinical trials to establish the expected risk for severe hypoglycemia with insulin therapy, rather than refer to other clinical data referenced by Dr. Cryer, include their more careful and uniform assessment of adverse events, such as hypoglycemia; their use of consensus definitions established a priori; their ability to compare frequency of hypoglycemia among trials using intensive therapy in type 1 and type 2 diabetes; and, perhaps most importantly, their ability to examine the risk for hypoglycemia in the setting of studies specifically aiming for and achieving glycemic levels similar to those advocated by us. While we recognize that the selected cohorts and sometimes unrealistic setting of clinical trials may not accurately reflect what happens in clinical practice, we felt that the advantages noted above outweighed the potential problems with observational studies and, especially, retrospective studies, cited in Dr. Cryer’s letter.
Owing to space limitations, we did not include a comprehensive list of controlled clinical trials using insulin to achieve near normoglycemia, which support a relatively low rate of severe hypoglycemia. However, we would draw attention to several large trials (>1,400 participants), in addition to those cited in our original article, that both documented low rates of severe hypoglycemia (zero to five episodes per 100 patient-years) according to the Diabetes Control and Complications Trial definition and utilized intermediate-acting insulin (3–5), long-acting insulin (4,5), combinations of intermediate- and rapid-acting insulins (6), and insulin combined with oral agents (3,5,6). Moreover, large observational studies not cited by Dr. Cryer have noted rates considerably lower than those included in his table (7).
No one would argue that hypoglycemia of any degree can be both frustrating and frightening for patients, and severe hypoglycemia can be potentially life threatening. However, we think that the best data, derived from the careful collection of events using uniform definitions in clinical trials, suggest a relatively low risk of severe hypoglycemic events in insulin-treated type 2 diabetic patients. The ongoing ACCORD Study, which is aiming to achieve A1C <6.0%, should shed further light on the frequency of severe hypoglycemia in insulin-treated type 2 diabetic patients (8). Mindful of the potential risks posed by each of the therapies, we continue to endorse the algorithm proposed, including early use of insulin as indicated.
Drs. Porta and Trento (9) suggest that our algorithm, which recommended inititally treating patients with lifestyle plus metformin, dismissed the potential for lifestyle intervention in contributing to diabetes care. They further contend that we ignored a growing body of literature that supports lifestyle interventions, especially when delivered in a group setting, as a means of lowering A1C. However, the consensus document emphasized the importance of lifestyle intervention and recommended it as a first and continuing intervention in diabetes management. We also recommended treatment with metformin at the same time as initiation of lifestyle interventions as a realistic response to the common failure of lifestyle interventions in the clinical setting. This recommendation should not dissuade investigators or clinicians from developing and implementing innovative methods to enhance lifestyle interventions; it merely recognizes the absence of high-quality data to support the efficacy of solitary diet treatment of type 2 diabetes (a conclusion of the recent Cochrane Database Systems Review cited by Porta and Trento) (9,10). In fact, in the study by Porta and Trento, the vast majority of the patients were treated with oral hypoglycemic agents plus diet in order to maintain a stable A1C level of 7.0 to 7.4% over 4 years (11). While behavioral interventions, as implemented in the Diabetes Prevention Program (12) and the ongoing Look AHEAD (Action for Health in Diabetes) Study (13), may provide better glycemic results than that accomplished in the past, widespread implementation of such interventions in usual clinical care has not yet occurred.
Parkin and Davidson (14) suggest that the algorithm is incomplete as it “assumes that patients have only recently developed type 2 diabetes and that the A1C is only slightly elevated.” They further note that the algorithm is not aggressive enough in that evaluation and adjustment of therapy every 3 months may be inadequate for patients whose A1C is >10% and that the algorithm ignores the important contribution of postprandial glucose levels to A1C and risk of complications. Finally, Parkin and Davidson tout the putative cardiovascular benefits of acarbose and suggest that exenatide might “get more patients to target and avoid costly long-term complications.”
As indicated by the title, the consensus algorithm was designed to address the “initiation and adjustment of therapy.” Furthermore, the recommendations specifically took A1C into consideration, recommending early use of more effective glucose-lowering therapy (i.e., insulin) for patients with A1C >8.5%. Nothing in the algorithm precluded more frequent adjustments of therapy; in fact, we recommended more aggressive management in the presence of signs or symptoms of hyperglycemia or a catabolic state. However, for most patients, a 3-month period for adjustment of therapy, during which time A1C can equilibrate, should suffice. Finally, Parkin and Davidson’s faith in the cardiovascular disease effects of acarbose (as demonstrated in a prevention study) is not shared by the STOP-NIDDM investigators, who suggested that their unexpected finding of a beneficial effect of acarbose on cardiovascular disease was “hypothesis-generating and will need to be confirmed” (15). Although the role of postprandial glycemia in diabetes complications is far from established, the consensus guidelines included adjustment of therapy based on both pre- and postprandial glucose levels. No clinical trial in type 2 diabetes has demonstrated an added benefit of therapy that specifically lowers postprandial glycemia or complications. We do not at this time endorse newer and relatively expensive medications, such as exenatide, on the basis of their modest effectiveness in lowering glycemia, limited clinical data, and/or side effects.