Tighter glucose control while minimizing the risk of severe hypoglycemia is associated with lower mortality among critically ill cardiac patents, new research suggests.
Researchers reported inon the outcomes of a multicenter retrospective cohort study in 1,809 adults in cardiac ICUs. Patients were treated either to a blood glucose target of 80-110 mg/dL or 90-140 mg/dL, based on the clinician’s preference, but using a computerized ICU insulin infusion protocol that the authors said had resulted in low rates of severe hypoglycemia.
The study found patients treated to the 80-110 mg/dL blood glucose target had a significantly lower unadjusted 30-day mortality compared to patients treated to the 90-140 mg/dL target (4.3% vs. 9.2%; P less than .001). The lower mortality in the lower target group was evident among both diabetic (4.7% vs. 12.9%; P less than .001) and nondiabetic patients (4.1% vs. 7.4%; P = .02).
Researchers also saw that unadjusted 30-day mortality increased with increasing median glucose levels; 5.5% in patients with a blood glucose of 70-110 mg/dL, 8.3% mortality in those with blood glucose levels of 141-180 mg/dL, and 25% in those with a blood glucose level higher than 180 mg/dL.
Patients treated to the 80-110 mg/dL blood glucose target were more likely to experience an episode of moderate hypoglycemia, compared with those in the higher target group (18.6% vs. 8.3%; P less than .001). However, the rates of severe hypoglycemia were low in both groups, and the difference between the low and high target groups did not reach statistical significance (1.16% vs. 0.35%; P = .051).
The authors did note that patients whose blood glucose dropped below 60 mg/dL showed increased mortality, regardless of what target they was set for them. The 30-day unadjusted mortality in these patients was 15%, compared with 5.2% for patients in either group who did not experience a blood glucose level below 60 mg/dL.
“Our results further the discussion about the appropriate BG [blood glucose] target in the critically ill because they suggest that the BG target and severe hypoglycemia effects can be separated,” wrote Andrew M. Hersh, MD, of the division of pulmonary and critical care at San Antonio Military Medical Center, and his coauthors.
But they said the large differences in mortality seen between the two treatment targets should be interpreted with caution, as it was difficult to attribute that difference solely to an 18 mg/dL difference in blood glucose treatment targets.
“While we attempted to capture factors that influenced clinician choice, and while our model successfully achieved balance, suggesting that residual confounding was minimized, we suspect that some of the mortality signal may be attributable to residual confounding,” they wrote.
Another explanation could be that hypoglycemia was an ‘epiphenomenon’ of multiorgan failure, as some studies have found that both spontaneous and iatrogenic hypoglycemia were independently associated with mortality. “However, given the very-low rates of severe hypoglycemia found in both groups it is unlikely that this was a main driver of the mortality difference found,” the investigators wrote.
The majority of patients in the study had been admitted to the hospital for chest pain or acute coronary syndrome (43.3%), while 31.9% were admitted for cardiothoracic surgery, 6.8% for heart failure including cardiogenic shock, and 6% for vascular surgery.
The authors commented that a safe and reliable protocol for intensive insulin therapy, with high clinician compliance, could be the key to realizing its benefits, and could be aided by recent advances such as closed-loop insulin delivery systems.
They also stressed that their results did not support a rejection of current guidelines and instead called for large randomized, clinical trials to find a balance between benefits and harms of intensive insulin therapy.
“Instead our analysis suggests that trials such as, and the conclusion they drew, may have been accurate only in the setting of technologies, which led to high rates of severe hypoglycemia.”
No conflicts of interest were declared.