There are moments during every physician’s day when she or he gives medical advice based on well-established evidence– “The data show that starting medication A for this disease will reduce the risk of death by 20%.” There are also moments when she or he may give advice just because it seems like the right thing to do, though evidence may be lacking– “It makes sense that using medication B might help in the treatment of this disease.” Sometimes advice based on common sense or medical tradition turns out to be misguided (e.g., bed rest for back pain, niacin for atherosclerotic vascular disease). And sometimes advice that makes sense is spot-on correct, as shown by a paper published in this week’s issue of NEJM, “Glycemic Control and Excess Mortality in Type 1 Diabetes Mellitus, “ by Lind and colleagues.
According to a recent Centers for Disease Control and Prevention report, each year over 18,000 people in the United States are diagnosed with type 1 diabetes (T1D). People with T1D are at increased risk for both microvascular complications (e.g., neuropathy, nephropathy) and macrovascular complications (e.g., coronary disease, stroke), as well as morbidity associated with these conditions. As above, it might make sense that better glycemic control, as measured by a lower hemoglobin A1c level, would be associated with improved outcomes for these disease states. However, unlike prior research that has demonstrated a clear association between lower HbA1c levels and improved outcomes from microvascular complications, the relation between mortality and glycemic control has remained less well defined.
Lind and colleagues describe a prospective cohort study of patients with T1D who were enrolled in Sweden’s National Diabetes Registry. For each person with diabetes enrolled in the study, the study also included 5 matched controls from Sweden’s general population. Participants were followed from enrollment until death or study completion. Outcome data collected for all participants included date of death (if it occurred) and relevant associated diagnoses. Data collected specifically for participants with T1D included albuminuria status, kidney function, and updated mean HbA1c level. Cox regression models were used to compare outcomes between persons with T1D and the matched controls.
Between 1996 and the end of 2011, 33,915 patients with T1D and 169,249 controls were enrolled in the study. In their subsequent analyses of these populations, Lind et al. found that persons with poor glycemic control (HbA1c ≥9.7%) had 8-10 times the risk of mortality compared to the control population, and significantly higher mortality than persons with T1D who had appropriate glycemic control.
At first glance, the results of Lind et al. are no surprise—better glycemic control can improve outcomes in T1D. However, Lind also provides us with humbling data. First, while the risk of mortality in T1D appears to be modifiable and dependent on adequate glycemic control, our progress in improving T1D outcomes may have stalled over the last two decades. In comparing study time periods (1998-2004 cf. 2005-2011), there was no significant improvement in excess mortality risk for the T1D population. Additionally, the authors found that even when the glycemic control of persons with T1D was appropriate (updated mean HbA1c ≤ 6.9%), they still had twice the risk of mortality as compared to the control population.
As clinicians these results may leave us wondering: what can we do to help improve outcomes in T1D? We know there has been an historical gap between guidelines and the actual quality of care patients receive–only 13-15% of persons with T1D reach their HbA1c goal. Similar data exist for other diabetes quality metrics. Innovation in patient engagement, quality improvement projects around guideline adherence and identification of additional outcomes metrics may be appropriate starting places for our collective efforts.
On top of this we should ask, if appropriately controlled T1D still carries an increased risk of mortality, what are we missing? Continued research on insulin replacement strategies (e.g. the bionic pancreas) and mitigation of the end-organ effects of diabetes is needed.
After reading Lind et al., we may feel inclined to congratulate ourselves–our clinical intuition was correct after all. However, by strengthening the known association between glycemic control and mortality, Lind and colleagues have sounded an important warning: clinicians and researchers still have much progress to make in improving our understanding of T1D and the quality of care we provide each day.