Hypoglycemia at Night: A Cause for Alarm?

S leep is important to us. We don't fully understand why, but we spend about a third of our lives doing it. All animals sleep, variably from 2 h (giraffes) to 18 h (pythons) in a 24-h day. Humans sleep very variably, but with a mean of around 8 h. If we are sleep deprived, we have problems of executive function, memory, and appetite control. The importance of sleep may be underlined by the tendency to sleep more when opportunity affords, making up the so-called “sleep debt.”

It has long been known that hypoglycemia is common during sleep in insulin-treated diabetes—and that much of it occurs without the patient's knowledge. Gale and Tattersall, ¹ conducting in-hospital hourly venous blood glucose measurements in a small and diverse group of people with poorly controlled insulin-treated diabetes in 1979, found that over half were hypoglycemic in the night and that on half of those occasions the hypoglycemia occurred without the patient's knowledge. A similarly sized study, conducted 30 years later, using continuous glucose monitoring (CGM) on 2 nights in people specifically with type 1 diabetes and on very different insulin regimens, likewise discovered that about half the group experienced nocturnal hypoglycemia, most of which was again asymptomatic. ² Depressingly, small and large studies using CGM show high rates of nocturnal hypoglycemia, in patients using both multiple daily injection regimens and on insulin infusion (pump) therapy. ^3,4 Although, in the longer-term study, only 8.5% of 36,467 nights included hypoglycemia (about two episodes per person per month), approximately one-quarter of those episodes lasted 2 h or more and only 3% of the 176 patients included had no nocturnal hypoglycemia at all. ⁴

The consequences of nocturnal hypoglycemia are not fully understood. It can be sufficiently severe to trigger a seizure in susceptible individuals ⁵ and has been associated with sudden death, ⁶ perhaps by triggering a fatal cardiac arrhythmia, ⁷ but, as suggested by the numbers just quoted, such dreadful outcomes are, happily, exceedingly rare. Posthypoglycemic transient paralysis is well described, ⁸ but formal attempts to discover more subtle neurological consequences have had varied outcomes: King et al. ⁹ found reduced measures of well-being but no evidence of cognitive dysfunction the morning after nocturnal hypoglycemia; Jauch-Chara et al. ¹⁰ described impairment of memory consolidation. Acute hypoglycemia has also been associated with endothelial dysfunction and changes in coagulation state and markers of inflammation, which have been hypothesized to contribute to vascular consequences of diabetes (reviewed by Desouza et al. ¹¹ ). This hypothesis remains unproven, as the higher rates of hypoglycemia with intensive control in such trials as the Diabetes Control and Complications Trial were also associated with protection from vascular complications over time—presumably primarily because of the reduced exposure to hyperglycemia. Nevertheless, it remains plausible. As with hypoglycemia at other times, nocturnal hypoglycemia can result in diminished protection from hypoglycemia next day. ¹² Whether this relates to the lower metabolic demand of the body during deep sleep, and is therefore not a problem, is doubtful, at least in iatrogenic hypoglycemia, as the metabolic rate falls by less than 10% during sleep. The lack of a stress response to hypoglycemia in deep sleep may explain why so many episodes of nocturnal hypoglycemia are unappreciated by the patients undergoing them. ¹³ Fear of nocturnal hypoglycemia is very prevalent among people with diabetes and their families and may lead to overeating, or insulin underdosing, with subsequent impact on the ability to achieve the degree of glycemic control that best avoids long-term diabetes complications. Better methods of avoiding hypoglycemia during sleep are urgently required.

One of the expectations of the development of CGM was that it would provide patients with an effective hypoglycemia alarm. Practical experience, however, shows that audible alarms are often unperceived by the wearer, who sleeps through them. ¹⁴ In a novel article in the present issue, Ly et al. ¹⁵ describe that it is the sleep rather than the hypoglycemia that underlies this. The study is inevitably small but fails to find any difference in the acoustic signal that detectably arouses sleeping adolescents with type 1 diabetes when applied during deep sleep, whether at euglycemia or hypoglycemia. Their data suggest—and parents of adolescent children will not be surprised to discover—that an auditory alarm would need more closely to resemble a jet engine on take off than a human voice to awaken a sleeper as blood glucose falls. Current alarms on CGM systems are just too subtle to be heard.

How do we resolve these problems? Further research into the consequences of unperceived nocturnal (or daytime) hypoglycemia is to be welcomed, but whether new data in this area will address the concerns of patients and their families is doubtful, even were the results to be reassuring. Newer treatments for diabetes are often marketed with the claim of less hypoglycemia risk, but once a patient is dependent upon exogenous insulin, such claims are relative, and the evidence tends to relate to mild rather than severe episodes (see, for example, Riddle et al. ¹⁶ ). Nocturnal hypoglycemia is one more reason to pursue actively the ultimate target of restoring glucose-responsive insulin delivery to insulin-deficient patients. Closed-loop insulin delivery has had some early successes, ^17,18 and cell replacement therapy likewise. ¹⁹ Both need more development to be generally applicable and research investment to develop cost-effective ways of restoring truly normal glucose control.