Original article:
Nocturnal hypoglycemia in children and adolescents with insulin-dependent diabetes mellitus: prevalence and risk factors. Beregszŕszi M, Tubiana-Rufi N, Benali K et al.
J Pediatr 1997; 131: 27-33.

Summary
Blood glucose levels were measured hourly from 2200 h to 0800 h in 150 French children, 87% of whom were on twice-daily mixtures of soluble and isophane insulin. Values <3.3 mmol/l occurred on 47% of nights and were asympto-matic in half. Four factors were associated with nocturnal hypoglycaemia: (1) younger age, (2) two or more episodes of severe hypoglycaemia from the onset of diabetes, (3) more than 5% of self-measured blood glucose values <3.3 mmol/l in the previous month, and (4) daily insulin doses >0.85 U/kg. Two-thirds of nocturnal hypoglycaemia episodes began between 2200 h and midnight, lasting between 3 and 9 h in 30%. A pre-dinner blood glucose <5.2 mmol/l was highly predictive of subsequent nocturnal hypoglycaemia, as was a fasting level the next morning <6.7 mmol/l. Sixty percent of children whose blood sugar was >10.7 mmol/l at 2200 h had nocturnal hypoglycaemia.

Comment
The hours of sleep (and associated fasting) are the Achilles' heel of all insulin regimens because there is no satisfactory basal insulin; absorption of all long-acting insulins produces a parabolic curve rather than the constant basal level found in healthy people. This is compounded by the variability of absorption, which may lead to a difference in free insulin levels of 50% in the same subject on two different nights. Repeated studies in unselected adults with Type 1 diabetes have shown hypoglycaemia in 30-50% of all nights [1]. Nocturnal hypoglycaemia may last for 6 h or more and a striking finding in all studies is that three-quarters of all, even prolonged, episodes are asymptomatic (at least to the patient, but in half of these, the observer can detect an episode from abnormal restlessness, sweating and other signs). In children, symptomatic nocturnal hypoglycaemia is terrifying for the parents, and asymptomatic nocturnal hypoglycaemia, especially in very young children, worrying because of concern that it may impair normal brain development and neuropsychological functioning.
Three previous studies have suggested a somewhat lower frequency of nocturnal hypoglycaemia in children (18-34%) than adults, but this may be misleading since in one study, blood glucose was only measured three times and a substantial number of children in the other two were on once-daily morning insulin only. The study of Beregszŕszi et al. is one of the largest yet carried out in children and confirms that, as with adults, nocturnal hypoglycaemia is a frequent and troublesome problem. The risk factors are those one would expect from studies in adults: a previous history of severe hypoglycaemia, tight control (or at least demonstrably low blood sugars) in the previous month, and a high insulin dose. It is also worrying that the highest frequency was seen in the youngest children.
It would be reassuring and useful for the parents of a diabetic child to be able to predict whether hypoglycaemia was going to occur on a particular night so that they could do something to prevent it. In the present study a blood glucose value <5.5 mmol/l before dinner was highly predictive of subsequent nocturnal hypoglycaemia, although it needs to be emphasized that these French children ate their dinner at 1930 h and did not routinely have a snack before going to bed. The authors defend this strategy by pointing out that 60% of children whose blood sugar was >10.7 mmol/l at 2200 h still had nocturnal hypoglycaemia and that routinely giving a snack to all the children would have worsened nocturnal hyperglycaemia in the others. In adults, Pramming et al. [2] found that if blood glucose was <6.0 mmol/l at 2300 h, the risk of nocturnal hypoglycaemia was 80%, whereas if it was >6.0 mmol/l, the risk was only 12%. In a study in British children, a glucose value <7.0 mmol/l at 2200 h had a positive predictive value of 83% for nocturnal hypoglycaemia [3]. In adults on multiple insulin injections, Bendtson et al. [4] found that those with a blood glucose <6.0 mmol/l at 2300 h had a 100% chance of nocturnal hypoglycaemia. Looked at the other way, it has often been shown that a low fasting blood glucose indicates preceding nocturnal hypoglycaemia and this was also the case in the present French study.
If a 'normal' fasting blood glucose can only be achieved at the cost of symptomatic or asymptomatic nocturnal hypoglycaemia, what should paediatricians do? In a typically thoughtful editorial accompanying this paper [5], the late Julio Santiago suggests that maintaining 50% of pre-breakfast values between 4.4 and 8.8 mmol/l may be the most which can be accomplished safely. Indeed, for younger children 8-10 mmol/l may be more realistic. In fact, I have seen a number of adults in whom it is impossible to achieve normal fasting blood glucose concentrations although they still maintain an HbA 1c in the range of 7-8%.
In the future, solutions to the problem of nocturnal hypoglycaemia in children and adults will only come from better basal insulins. For the present, prevention must depend on setting realistic targets for fasting blood glucose and measuring blood sugar at bedtime.

References
1. Thow JC, Home PD. Nocturnal hypoglycaemia. In: Frier BM, Fisher BM, eds. Hypoglycaemia and diabetes: clinical and physiological aspects. London: Edward Arnold, 1993; 198-211.
2. Pramming S, Thorsteinsson B, Bendtson I et al. Nocturnal hypoglycaemia in patients receiving conventional treatment with insulin. Br Med J 1985; 291: 376-9.
3. Whincup G, Milner RDG. Prediction and management of nocturnal hypoglycaemia in diabetes. Arch Dis Child 1987; 62: 333-7.
4. Bendtson I, Kverneland A, Pramming S et al. Incidence of nocturnal hypoglycaemia in insulin-dependent dia-betic patients on intensive therapy. Acta Med Scand 1988; 223: 543-8.
5. Santiago JV. Nocturnal hypoglycemia in children with diabetes: an important problem revisited [editorial]. J Pediatr 1997; 131: 2-4.