OBJECTIVE

To identify clinical factors that affect hypoglycemia perception in type 2 diabetes.

METHODS

Prospective observational study of 344 insulin-treated subjects randomly selected from pharmacy records at three large medical centers. At entry, subjects underwent an extensive psychological evaluation and then monitored their blood glucoses in their usual fashion for up to 52 weeks using a glucose meter capable of storing 1000 readings. For blood glucoses <or=3.33 mmol/l (60 mg/dl), subjects recorded the severity of symptoms in a log book. Symptoms were scored '0' if they were asymptomatic, '1' for symptoms that were mild-to-moderate, and '2' if the subject had a diminished level of consciousness or required the assistance of others. A mean blood glucose and mean symptom score were calculated for all hypoglycemic episodes detected by each patient.

RESULTS

One hundred seventy-six subjects (51.2%) had a median of 4.5 hypoglycemic events during 42.5+/-8.0 weeks of follow-up. The mean hypoglycemic blood glucoses and mean symptom scores were modestly correlated (r=0.153; P=0.043). After adjusting for blood glucose, symptom scores were lower in the elderly and higher in subjects with higher diabetes knowledge scores, microvascular complications, and higher entry hemoglobin A1c. Multiple linear regression showed that the latter three factors were independently predictive of higher symptoms.

CONCLUSIONS

Conventional clinical data may be useful in identifying patients with poor hypoglycemia perception. Patients with better long- and short-term glycemic control are at higher risk and may benefit from instruction on all aspects of hypoglycemia.