RSS Feeds
Glycemic Control in Type 2 Diabetes: Time for an Evidence-Based About-Face?
- Victor M. Montori, MD, MSc; and
- Mercè Fernández-Balsells, MD
- From the Mayo Clinic, Rochester, Minnesota, and Hospital Universitari de Girona Doctor J. Trueta, Girona, Spain.
Abstract
Some diabetes guidelines set low glycemic control goals for patients with type 2 diabetes mellitus (such as a hemoglobin A1c level as low as 6.5% to 7.0%) to avoid or delay complications. Our review and critique of recent large randomized trials in patients with type 2 diabetes suggest that tight glycemic control burdens patients with complex treatment programs, hypoglycemia, weight gain, and costs and offers uncertain benefits in return. We believe clinicians should prioritize supporting well-being and healthy lifestyles, preventive care, and cardiovascular risk reduction in these patients. Glycemic control efforts should individualize hemoglobin A1c targets so that those targets and the actions necessary to achieve them reflect patients' personal and clinical context and their informed values and preferences.
When the facts change, I change my mind. What do you do, sir?
—John Maynard Keynes
Routine treatment for type 2 diabetes has targeted tight glycemic control to reduce diabetes complications. A low hemoglobin A1c (HbA1c) value has been the objective of clinical care and a measure of its quality. In this article, we summarize trials that evaluated tight glycemic control in patients with type 2 diabetes and offer practical evidence-based suggestions for managing these patients.
Recent Evidence
We focus on large trials that compared clinical outcomes among patients with type 2 diabetes who were randomly assigned to tight versus less tight glycemic targets (1–5). We do not discuss trials that did not test contemporary treatment approaches (6, 7), trials of multifactorial risk reduction interventions (8, 9), or trials designed either to compare a single antihyperglycemic agent against placebo (10) or to assess glycemic durability (11).
The Table shows that trials that compared different glycemic targets involved different patient populations and had heterogeneous interventions, targets, and follow-up protocols. The UKPDS (United Kingdom Prospective Diabetes Study) trials (3, 4) are the oldest trials, involved patients with newly diagnosed diabetes, achieved less tight glycemic control, and had longer follow-up than the more recent trials. The results of all of the trials, except for the UKPDS metformin trial (3), suggest that tight glycemic control may not reduce the risk for all-cause or cardiovascular mortality, stroke, amputations, or even microvascular complications (Figure). These findings are inconclusive, however, because estimates of effects were imprecise as few patients developed complications and effects varied importantly across trials. Reported results of all trials did suggest that intensive glycemic control might reduce the risk for nonfatal myocardial infarction by about 16%. A clear adverse consequence of tight glycemic control was a 2- to 3-fold increased risk for severe hypoglycemia: Trials with the lowest HbA1c targets had the highest incidence of hypoglycemia. Intensive glycemic control also led to a 2% weight gain in all but the metformin versus conventional comparison in the UKPDS trial (3).
For all plots, no difference between intensive and less intensive glycemic control is denoted by the vertical continuous line. UKPDS(a) refers to the main UKPDS trial, which included 2729 patients in the intervention group and 1138 patients in the conventional group. We subdivided the main UKPDS trial in overweight patients with diabetes into 2 comparisons: UKPDS(b), which included 342 overweight patients in the intervention group with metformin and 411 in the conventional group, and UKPDS(c), which included all 1293 overweight patients in the intervention group (metformin, glibenclamide, and insulin) and 411 in the conventional group. When possible, we presented the results using data from UKPDS(c). We calculated the number of cardiovascular deaths in UKPDS(a) by subtracting all noncardiovascular deaths from all reported deaths. When trials used different outcome definitions, the asterisk and dagger identify which trial reported which outcome definition or combination of outcomes (when both symbols accompany the trial name). ACCORD = Action to Control Cardiovascular Risk in Diabetes; ADVANCE = Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation; CGC = conventional glycemic control; IGC = intensive glycemic control; RR = relative risk; UKPDS = United Kingdom Prospective Diabetes Study; VADT = Veterans Affairs Diabetes Trial.
Variability in Outcomes Across the Trials
Results for some outcomes, such as all-cause and cardiovascular mortality, varied across trials (Figure). For instance, the UKPDS metformin trial (3) reported that tight glycemic control reduced mortality risks, whereas the ACCORD (Action to Control Cardiovascular Risk in Diabetes) trial (5) reported that tight control increased these risks. The favorable effect of metformin observed in the UKPDS is not evident in the latest comparative effectiveness systematic review (12) or in ADOPT (A Diabetes Outcome Progression Trial) (11), and the mechanism of this finding remains unknown. Possible explanations for the increased mortality risk with tight control in the ACCORD trial include adverse effects from hypoglycemia or rosiglitazone and chance. The data monitoring committee and National Institutes of Health stopped the trial early because of mortality risks, and early stopping can lead to overestimates of the effect of interventions on the monitored outcome (13).
The choice of participants and follow-up duration could explain differences across trial results. The most recent trials enrolled patients with long-standing diabetes who had cardiovascular disease and had relatively brief follow-up. A hypothesis-generating subgroup analysis from the ACCORD trial suggested that patients without a previous cardiovascular event may have had a lower risk for cardiovascular events with tight glycemic control than patients with a previous event (5). This subgroup analysis and the UKPDS metformin trial findings suggest that patients with earlier and milder disease could benefit from aggressive treatment. This is somewhat contradicted by the comparable outcomes of tight glycemic control reported in the main UKPDS trial in patients with newly diagnosed diabetes (4) and in the ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation) trial in patients with long-standing diabetes (1). Thus, this possibility should be evaluated in long-term clinical trials, because premature assumption of its truth exposes a very large population to a resource-intensive intervention with still-uncertain benefits and certain harms.
Perhaps differences in medications used to achieve glycemic control could explain the observed differences across the trials. Some research, for example, has linked rosiglitazone to adverse cardiovascular outcomes (14) and metformin to cardiovascular benefits (12). The tight glycemic control trials, however, were not designed to link outcomes to the medications used. Thus, perhaps glycemic control is effective, but not with the treatment strategies tested, or we may have failed to fully understand the mechanisms by which diabetes causes complications and may have chosen incorrect therapeutic targets and goals.
The nature of the outcomes measured, reported, and emphasized in the trials may confuse or mislead interpretation of results and cause “apparent” inconsistency of findings. In the ADVANCE trial, focus on the composite end point suggests that tight control was associated with “prevention of all diabetes related complications by 10%” (1), whereas this effect is driven mainly by a reduction in the incidence of albuminuria. In UKPDS, decreases in the number of patients requiring photocoagulation made up most of the effect of tight glycemic control captured by the composite end point “any [of 14] diabetes complications” (4). To aid interpretation and avoid being misled by composite end points, we presented (Figure) and recommend focus on the effect of treatment on the components of composite end points that matter to patients.
Focus on a surrogate end point, such as HbA1c, could also mislead if it did not capture all of the effect of the intervention on the outcomes that matter. The appropriateness and adequacy of the surrogate end point may be specific to patients, interventions, and outcomes. For instance, HbA1c seems to appropriately capture the effect of intensive insulin therapy on microvascular complications in patients with type 1 diabetes (15). Yet, the trials we reviewed in type 2 diabetes show that reductions in HbA1c level achieved with contemporary therapies do not uniformly predict major benefits for either micro- or macrovascular outcomes in patients with type 2 diabetes. Furthermore, analyses of UKPDS data showed that intensive therapy with metformin was associated with improved macrovascular outcomes compared with intensive therapy with other agents despite both groups achieving similar HbA1c levels (3).
Conclusion and Recommendations
Current best evidence requires a change in emphasis in our care for patients with type 2 diabetes. Clinicians should prioritize supporting well-being and healthy lifestyles, preventive care, and cardiovascular risk reduction for these patients. The randomized trial evidence that we reviewed does not strongly support tight glycemic control as more beneficial than harmful in reducing the risk for diabetes complications. Although we should not dismiss potentially effective approaches (for example, early tight glycemic control for patients with newly diagnosed diabetes), we require additional research to confirm or refute such approaches before we impose them on patients, particularly given the unequivocal patient burden, cost, and harm of serious hypoglycemia associated with tight control.
Given that patients with diabetes often have comorbid conditions, clinicians should avoid glycemic control interventions that overwhelm the patients' capacity to cope clinically, psychologically, and financially. Tight disease-centered goals that require highly complex and burdensome treatment programs may promote frustration, nonadherence, and financial stress in some patients. For instance, many patients will not benefit from and could reduce or eliminate glucose self-monitoring (16–18).
Patients may opt to control their glycemia to a level that best balances the burden of medication, including the risk for hypoglycemia, with the benefit in reducing symptoms, which may appear with glycemia greater than 10 mmol/L (>180 mg/dL). Keeping the HbA1c level between 7% and 7.5% (estimated average glucose level, 8.5 to 9.5 mmol/L [150 to 160 mg/dL]) seems reasonable and feasible for many patients. For others—particularly those with severe insulin deficiency—achieving this range requires substantial effort, including physiologic insulin dosing (that is, basal-bolus regimens) and intense monitoring.
Glycemic targets can be adjusted up or down according to the burden of treatment; side effects; and the patient's context, values, and preferences. Given the possibility that tighter control may be beneficial, some patients who are less concerned about downsides, and are ready to do whatever may possibly help, may choose tighter control. The need to set individual glycemic targets suggests that HbA1c targets for clinical use cannot be the same when used to measure quality of care. Policymakers who want to use HbA1c as a performance measure should use an upper limit, such as an HbA1c level greater than 9%, to indicate possibly inadequate care, rather than one that would invite clinicians to ignore patient burden, context, and goals (for example, HbA1c level <7%).
Once clinicians and patients have set an HbA1c target, which is often an iterative process, they need to decide how to achieve the target. Because we cannot confidently distinguish the relative effectiveness of different diabetes medications in reducing complications (12, 19), we recommend basing their selection on such factors as burden of administration and side effects. We have developed and are studying tools to promote patient involvement in choosing diabetes medications (20). We hope that tools and tactics that encourage patient involvement in treatment decisions prove to be effective and lead to treatment programs that are both evidence-based and consistent with patients' context, values, and preferences.
Article and Author Information
-
Grant Support: The American Diabetes Association awarded Dr. Montori a Clinician Investigator grant in 2004. Novo Nordisk, a maker of insulin, subvented the American Diabetes Association granting program but did not have direct contact with the investigators and did not play any role in the awarding of the grant to the research team. Dr. Fernández-Balsells has received grant support from the Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo (BA08/90035), Government of Spain.
-
Potential Financial Conflicts of Interest: Grants received: V.M. Montori (American Diabetes Association), M. Fernández-Balsells (Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Government of Spain). Dr. Montori's research group does not take funding from pharmaceutical or medical device corporations.
-
Requests for Single Reprints: Victor M. Montori, MD, MSc, Knowledge and Encounter Research Unit, Mayo Clinic, Plummer 3-35, 200 First Street SW, Rochester, MN 55905; e-mail, kerunit{at}mayo.edu.
-
Current Author Addresses: Dr. Montori: Knowledge and Encounter Research Unit, Mayo CTSA, Plummer 3-35, 200 First Street SW, Rochester, MN 55905.
-
Dr. Fernández-Balsells: Servei d'Endocrinologia, Hospital Universitari de Girona Doctor J. Trueta, Carretera de França s/n, 17007 Girona, Spain.
