Insulin

Insulin + amylin analogues

Pramlintide 30 micrograms was given to 16 patients using insulin pumps as an injection at meal times [ ]. Meal time insulin was reduced by 17%. Serum fructosamine improved. Nausea was the most common adverse reaction. There was no hypoglycemia.

Insulin + biguanides

In a multicenter, open, parallel-group study in the USA, 188 patients responded insufficiently to two oral hypoglycemic drugs (HbA _1c over 8.0%) and received either a third oral drug or metformin with insulin 70/30 mix twice a day [ ]. HbA _1c and fasting plasma glucose did not differ between the groups and neither did minor episodes of hypoglycemia. Weight gain was 3.5 and 2.9 kg respectively. Lack of efficacy or adverse effects prompted withdrawal in 13% of those taking triple therapy and 2% of those taking insulin + metformin. In those taking insulin + metformin cholesterol and triglycerides fell and the mean cost was $10.40 for the triple group and $3.20 for the insulin + metformin group.

Metformin was given as an adjunct to insulin in a double-blind, placebo-controlled study in 28 adolescents needing more than 1 unit/kg/day [ ]. The dose of metformin was 1000 mg/day when body weight was under 50 kg, 1500 mg/day when it was 50–75 kg, and 2000 mg/day when it was over 75 kg. Metformin lowered insulin requirements. The number of episodes of hypoglycemia increased compared with placebo. There was gastrointestinal discomfort in six patients taking metformin and five taking placebo.

A comparable placebo-controlled study was reported in 353 patients with type 2 diabetes for 48 weeks. All were taking insulin, and HbA _1c fell in those who also took metformin. Body weight was reduced by 0.4 kg by metformin and increased by 1.2 kg by placebo. Symptomatic episodes of hypoglycemia were more common with metformin. There were mild transient gastrointestinal complaints in 56% and 13% respectively [ ].

Insulin plus metformin (27 patients, 2000 mg/day) or troglitazone (30 patients, 600 mg/day) in patients with type 2 diabetes using at least 30 units/day was compared with insulin alone (30 patients) for 4 months [ ]. Body weight increased in the insulin and the insulin plus troglitazone groups. In the insulin plus metformin group there were significantly more gastrointestinal adverse reactions but less hypoglycemia than the other groups.

Insulin + meglitinides

In 80 patients taking metformin 850 or 1000 mg tds plus NPH insulin at bedtime, metformin was withdrawn and repaglinide 4 mg tds added in half of the patients for 16 weeks [ ]. In the repaglinide group the dose of insulin increased slightly and weight gain was 1.8 kg more. Mild hypoglycemia occurred more often in the metformin group; nightly episodes of hypoglycemia occurred only with repaglinide. One patient taking repaglinide had a myocardial infarction, and one had three separate hospitalizations for chest pain (myocardial infarction was excluded). No specific data were presented about gastrointestinal adverse effects or infections.

Insulin + meglitinides + sulfonylureas

When repaglinide 4 mg tds and glimepiride 160 mg bd in combination with bedtime NPH insulin were compared for 13 weeks in 80 patients, there were no differences in improvement of glycemic control, hypoglycemia, or weight gain, except in one patient taking gliclazide + NPH, who needed third party help for hypoglycemia [ ].

Insulin + thiazolidinediones

Insulin plus metformin (27 patients, 2000 mg/day) or troglitazone (30 patients, 600 mg/day) in patients with type 2 diabetes using at least 30 units/day was compared with insulin alone (30 patients) for 4 months [ ]. Body weight increased in the insulin and the insulin plus troglitazone groups. In the insulin + metformin group there were significantly more gastrointestinal adverse reactions but less hypoglycemia than the other groups.

The addition of rosiglitazone to insulin for 26 weeks in a double-blind study in 315 patients with inadequately controlled type 2 diabetes improved glycemic control and was well tolerated [ ]. There was a significant fall in hemoglobin, and some patients in both the rosiglitazone and placebo groups developed edema.

Hypoglycemia

Inadvertent hypoglycemia is the most frequent complication of insulin therapy [ ].

Frequency : In a review of severe hypoglycemia (requiring assistance) 11 studies of insulin-treated patients with type 2 diabetes were analysed; each included at least 50 patients followed for a minimum of 6 months [ ]. The reporting of severe hypoglycemia varied enormously, from none at all to 73 episodes per 100 patient years. Prospective and retrospective studies had different reporting rates, with lower rates in prospective studies. This may have been due to the study populations, with possible exclusions in prospective studies, including those with frequent hypoglycemia or hypoglycemia unawareness. Risk factors for severe hypoglycemia in type 2 diabetes were increasing age and duration of diabetes.

Over 5% of deaths in diabetes can be attributed to hypoglycemia. The frequency increases with rigorous maintenance of normoglycemia [ , ]. In the Diabetes Control and Complications Trial (DCCT) [ ] the frequency of serious hypoglycemia was more than three times increased in the intensively treated group, and the frequency of the attacks was related to the concentration of HbA _1c [ ]. The UK Prospective Diabetes Study in patients with type 2 diabetes also showed an increased risk of hypoglycemia with more intensive treatment [ ].

In a 12-month open study of 64 patients with type 2 diabetes (mean age 58 years) who used either once-daily bedtime NPH insulin + tablets or twice-daily 30% soluble + 70% NPH, there was less hypoglycemia in the former (2.7 hypoglycemic events per person compared with 4.3); the improvement in HbA _1c was similar [ ]. Weight gain was also less in those who used once-daily NPH (1.3 kg compared with 4.2 kg).

The risk of hypoglycemia in intensive care has been studied retrospectively in 2272 patients. Hypoglycemia was associated with diabetes, sepsis, a Sequential Organ Failures Assessment score of 1 or more, the use of bicarbonate in renal replacement, female sex, use of vasopressors or insulin, and impaired nutrition without altered insulin dosage [ ]. Various protocols have been designed and studied to try to reduce the risk of hypoglycemia when infusing insulin to maintain normoglycemia in critically ill patients [ , ].

In a review of 102 consecutive drug-induced, hospital-related cases of coma, 23 were caused by insulin only, 14 by insulin + glibenclamide, and 3 by insulin + metformin [ ]. The likelihood of readmission for hypoglycemia after a previous admission was 2.9 times greater.

Of 546 Spanish diabetic children and teenagers, 14% had one period of hypoglycemia and 21% had more than one episode [ ]. The highest incidence was in the morning, possibly related to the frugal Spanish breakfast and abundant food intake in the evening.

Hypoglycemia was retrospectively monitored in 1055 patients who had had type 2 diabetes for more than 2 months, and who visited the clinic at least twice in 6 months [ ]. They all received aggressive treatment to reach near-normal blood glucose concentrations. Symptoms of hypoglycemia were mentioned by 12% of those treated with diet alone, 16% of those using oral agents, and 30% of those using any form of insulin. In only five patients, all using insulin, was there severe hypoglycemia. A low HbA _1c concentration at follow-up, symptoms of hypoglycemia at the initial visit, and younger age were independently associated with an increased incidence of symptoms.

During 12 months, 244 episodes of severe hypoglycemia in 166 patients were recorded in a district with a population of 367 051 people (8655 with diabetes); there were 69 (7.1%) episodes in people with type 1 diabetes, 66 (7.3%) in people with type 2 diabetes using insulin, and 23 (0.8%) in those taking a sulfonylurea. Age, duration of diabetes, and lower social class were risk factors. The total cost of emergency treatment was estimated to be no more than $92 078 per year [ ]. In a German study in a comparable group of 200 000 people, there were 92 cases in those with type 1 diabetes and 146 cases in those with type 2 diabetes during 3 years [ ]. The estimated costs were lower: $88 676 per year for type 2 diabetes and $16 258 for type 1 diabetes.

The frequency and nature of episodes of hypoglycemia have been quantified in a retrospective survey of 215 patients with type 2 diabetes treated with insulin [ ]. More than 90% used a mixture of short-acting or rapid-acting insulin plus an intermediate acting insulin twice daily, six used NPH insulin twice daily, and 13 used three or more injections per day. In the preceding year 32 of them had 60 periods of severe hypoglycemia (requiring assistance), 0.28 episodes/patient/year. One patient with 13 episodes had impaired awareness. The frequency increased with age, duration of insulin therapy, and duration of diabetes, but not with a lower concentration of HbA _1c or a higher dose of insulin. There was mild hypoglycemia in 64%. In those with mainly autonomic symptoms the occurrence of symptoms was inversely related to age. The neuroglycopenic symptoms were more frequent with increasing duration of insulin therapy.

In 1500 patients in intensive care, there was hypoglycemia (4 mmol/l and less) in 5.2% of the intensively treated group and 0.8% of those who received conventional therapy; 11% of drug errors were from insulin administration errors and it has been recommended that frequent checks be made of infusion systems [ ].

Mechanisms : The concentrations of the counter-regulatory hormones adrenaline and glucagon were higher before treatment of coma in insulin-treated persons admitted to hospital for hypoglycemia than in the fasting state. However, concentrations of adrenaline, glucagon, cortisol, and growth hormone are lower during a hypoglycemic episode in a diabetic than when these hormones are measured during induced hypoglycemia in non-diabetics [ ]. There was an inverse correlation between glucose and adrenaline in hypoglycemia, but no direct relation between the other hormones and blood glucose concentrations. The addition of intramuscular glucagon during intravenous glucose therapy did not result in different glucose concentrations at any time. In type 2 diabetes counter-regulatory hormones are reduced compared with controls, but the patients release the hormones at higher blood glucose concentrations than type 1 diabetics, and the glucagon response is not blunted [ , ].

• Low responses of counter-regulatory hormones during induced hypoglycemia in a 27-year-old woman with poorly controlled diabetes (HbA _1c 11%), with frequent hypoglycemic instances of which she was not aware, improved dramatically after 3 months of good regulation [ ]; only growth hormone showed no reaction.

Hypoglycemia increased beta-adrenoceptor sensitivity in healthy subjects but reduced it in type 1 diabetes [ ].

In 20 insulin-treated diabetic patients with episodes of severe hypoglycemia in 1982–84, re-evaluated in 1992–94, emergency visits were reduced from 1.05/year to 0.42/year between 1984 and 1994 [ ]. There were no cases of fatal hypoglycemia. There was no association with HbA _1c . Multiple daily insulin doses reduced the frequency of hypoglycemia to one-third. In 1984, unawareness was a predisposing factor and most of the patients had deficient counter-regulation (adrenaline). Most patients had a long history of insulin injections (mean 29 years). In 1994, six patients were partly and eight patients totally retested and compared with 10 matched control patients with type 1 diabetes. When hypoglycemia was induced by insulin the patients with frequent hypoglycemia reached values under 3.0 mmol much faster, and in six patients the test had to be stopped before the normal duration of 3 hours for hypoglycemia; this never happened in the control patients. Counter-regulation was deficient in both 1984 and 1994, indicating that reduced counter-regulation can be permanent and does not only depend on specific circumstances.

In 86 intensively treated patients with type 1 diabetes aged 7–18 years, the incidence of severe hypoglycemia correlated with the serum activity of acetylcholine esterase. Patients with acetylcholine esterase activity at the median or above reported 3.0 events/year and those with acetylcholine esterase activity below the median reported 0.5 events/year, suggesting that a genetic factor may play a role in the emergence of severe hypoglycemia [ ].

Autoimmune insulin hypoglycemia is rare, and is most often seen in East Asia. Patients have low concentrations of insulin and C-peptides, but insulin can be released from the insulin antibody complex or receptor antibodies can stimulate the message in the receptor.

• A 72-year-old woman with type 2 diabetes was treated with insulin but developed repeated attacks of hypoglycemia, with blood glucose concentrations below 2.2 mmol/l [ ]. She stopped taking insulin, but the attacks continued, causing disorientation, loss of consciousness, palpitation, and sweating. Her HbA _1c concentration was raised at 6.3%. There were insulin antibodies 58% and also insulin receptor antibodies. Treatment with prednisone and glucose tablets resolved the hypoglycemic episodes within 48 hours.

Presentation : Symptoms of hypoglycemia have to be expected when the blood glucose concentration is below 2.6 mmol/l. The effects of hypoglycemia vary from patient to patient and can vary in the same patient. The symptoms and signs are of two types: adrenergic effects, due to release of catecholamines, and neuroglycopenic effects, due to the effects of hypoglycemia on the nervous system.

• Adrenergic effects Hunger, restlessness, profuse sweating, palor, tachycardia, and palpitation.

• Neuroglycopenic effects Headache, confusion, drowsiness, fatigue, difficulties in finding words, frequent yawning, anxiety, blurred vision, diplopia, and numbness of the nose, lip, and fingers.

Some patients do not experience the noradrenergic symptoms of hypoglycemia. They are taken by surprise, may lose consciousness and have hypoglycemic blood glucose concentrations without any preceding symptoms. After frequent episodes of hypoglycemia, there is altered awareness of hypoglycemia [ ]. It is difficult to substantiate altered awareness. Neuroendocrine responses and symptoms of hypoglycemia, but not cognitive dysfunction, are shifted to lower plasma glucose concentrations after recent hypoglycemia [ ]. Repeated episodes of hypoglycemia reduce the awareness of symptoms of hypoglycemia. This is accompanied by a lower blood glucose concentration, to elicit the response of counter-regulatory hormones [ ]. Beta-blockers also suppress the adrenergic symptoms, apart from sweating, which is mediated by sympathetic cholinergic transmission. Training increases awareness [ ]. Intensively treated patients may have reduced preservation of higher brain functions [ ]. Some groups, mainly in the UK, have suggested that transfer from animal to human insulin increases unawareness [ ], but this was not substantiated in other studies [ ].

Every patient treated with insulin (or with hypoglycemic agents) who develops a neurological or psychiatric disorder has to be considered to be hypoglycemic until proven otherwise.

A rapid fall in blood glucose in a diabetic patient can cause symptoms of hypoglycemia, even when blood glucose concentrations are still normal or above normal. Experience with pumps has shown that many patients continue to feel hypoglycemic for a long time after normoglycemia has been restored. After an attack of hypoglycemia, patients often felt less well for a period of up to 48 hours. Headache, tiredness, and lack of initiative may disappear only gradually. These symptoms in the morning may indicate unnoticed hypoglycemic periods during sleep. Wet pyjamas or sheets may also indicate unnoticed hypoglycemia. During anesthesia, profuse sweating may indicate hypoglycemia.

A rise in C reactive protein has been observed during spontaneous attacks of hypoglycemia in diabetics and after experimental hypoglycemia in healthy controls [ ].

Attacks of hypoglycemia are often preceded by less marked attacks, which are unnoticed or not reported to the family or physician. Attacks can be caused by reduced resistance to insulin or switch to a type of insulin with a different duration of action. The reasons for hypoglycemia can be inaccurate or excessive insulin injections, heavy physical exercise, or omission of meals. The action of highly purified insulins and some new analogues, even when in long-acting form, is somewhat faster and shorter compared with less pure formulations. Errors in injection techniques, such as superficial subcutaneous injection, forming nodules or causing bleeding, can introduce variation in the absorption of insulin, resulting in an increase in the mean administered dose and in inadvertent hypoglycemia. Twice-daily isophane insulin for 6 months has been compared with once-daily ultralente insulin in 60 patients [ ]. Isophane was associated with fewer attacks of hypoglycemia, lower HbA _1c , lower evening glucose concentrations, and greater patient satisfaction.

The Somogyi effect, unnoticed hypoglycemia during sleep, causes a rebound increase in morning blood glucose with accompanying glycosuria. When, in response to this, the evening dose of long-acting insulin is increased and the risk of nocturnal hypoglycemia also increases, creating a vicious cycle. Blood glucose monitoring late at night helps to establish the diagnosis. Of 39 poorly controlled patients with insulin-treated diabetes aged 9–66 years, 22 had recurrent nocturnal hypoglycemia, the best clinical clue to which was intermittent symptoms, however mild and infrequent they appeared to be [ ].

Hypoglycemia can also be induced by concomitant diseases, for example renal disease, hepatic disease (cirrhosis), hypopituitarism, hypoadrenocorticalism, hypoglucagonism, hypothyroidism, malnutrition, anorexia nervosa, pregnancy, termination of pregnancy, recovery from infections, operations, or stress.

Brittle diabetes Hypoglycemic periods are often seen in patients with so-called brittle diabetes, many of whom are overtreated with insulin. Changes in the insulin regimen (reduced use of long-acting insulin, frequent small injections of short-acting insulin) or the use of continuous infusion pumps can often lead to better results, but not everyone with brittle diabetes responds in that way [ ]. Apparently brittle diabetes may in fact be due to factitious hypoglycemia, the hypoglycemic periods of which are caused by surreptitious self-injection of insulin, Munchausen syndrome by proxy [ ], or manipulation of the prescribed doses. In factitious hypoglycemia, low blood glucose concentrations are accompanied by high insulin concentrations but low concentrations of C peptide [ ]. Suicide attempts with insulin may be less uncommon than is often thought. A Medline search between 1966 and 1999 identified 46 papers containing 69 cases of factitious hypoglycemia; 46 were women, 52 were not diabetic, 29 had no close links with diabetes in their environment [ ]. In 47 the hypoglycemia was induced by insulin. Two patients died and one had severe impairment of intellectual function and short-term memory. In 32 cases unnecessary surgical procedures were performed.

Cardiovascular effects of hypoglycemia The cardiovascular effects of hypoglycemia include angina pectoris, dysrhythmias, electrocardiographic changes, and coronary thrombosis. Raised concentrations of catecholamines and reduced concentrations of potassium contribute to cardiac damage during hypoglycemia.

Respiratory effects of hypoglycemia A 19-year-old woman with diabetes developed hypoglycemia with pulmonary edema [ ]. This has previously been seen as a complication of insulin shock therapy for psychiatric illnesses.

Nervous system effects of hypoglycemia When hypoglycemia does not resolve spontaneously or is not terminated, cerebral dysfunction becomes manifest as confusion or reduced consciousness. Lethargy and depression or obstructive behavior develop and are accompanied by loss of consciousness, snoring, deep respiration, and facial paralysis. Neurological involvement can appear as cramps, paralysis, hemiplegia, or paraplegia. Epileptic seizures can accompany attacks of hypoglycemia. In deep coma the pupils are dilated, but they may react to light. Coma can develop very rapidly.

In a teaching hospital in Edinburgh, 56 admissions of 51 patients for hypoglycemia were registered during 12 months; 41 patients had diabetes mellitus and 33 were using insulin [ ]. There was a high incidence of neurological effects. Psychiatric illness or alcoholism was common. Four patients died but only one as a direct consequence of hypoglycemia. A further six patients died within 15 months, not related to hypoglycemia.

In 37 drivers with type 1 diabetes, fewer corrective actions were taken when the blood glucose was below 2.8 mmol/l [ ]. This was related to increased neuroglycopenic symptoms and increased electroencephalographic theta-wave activity. The authors suggested that diabetics should not begin to drive when the blood glucose concentration is in the 4.0–4.5 mmol/l range. In two editorials, the possibility of further restricting driving licenses in people with diabetes has been discussed [ , ]. However, there is no evidence of higher accident rates in drivers with diabetes.

A retrospective questionnaire was sent to 195 consecutive patients addressing questions of severe hypoglycemia, coma, awareness of hypoglycemia, and fear of hypoglycemia [ ]. The mean duration of diabetes was 20 years and 82% had received intensive therapy. Coma was reported in 19% and severe hypoglycemia in 41%. Coma was independently related to neuropathy, beta-blockers, and alcohol.

In reaction to a report of pulmonary edema and hypoglycemia [ ] it has been noted that in many cases, one or more seizures precede pulmonary edema (acute respiratory distress syndrome), suggesting a neurogenic mechanism [ ].

In 304 insulin-treated patients, 8.2% of those with type 1 diabetes and 2.2% of those with type 2 diabetes had blood glucose concentrations below 4 mmol/l on arrival in the clinic [ ]. None had complained of symptoms of hypoglycemia or had taken glucose, but when questioned, 59% had autonomic symptoms and 12% had neuroglycopenic symptoms; 29% were asymptomatic.

Autonomic failure can occur as a result of hypoglycemia, since antecedent hypoglycemia causes both defective glucose counter-regulation and lack of awareness of hypoglycemia. The role of the brain in lack of awareness of hypoglycemia and the question of whether the brain is the primary site for sensing hypoglycemia has been discussed [ ]. In older people, markedly fewer autonomic symptoms are reported and there is greater slowing of psychomotor performance. In young children, hypoglycemia can lead to more mood and behavioral disturbances than in adults, although they also occur in the latter. It may be that the cortical responses to recurrent hypoglycemia are less plastic and reversible than the hypothalamic and glucose-sensing functions [ ].

Hypoglycemic neuropathy has been described in association with an insulinoma [ ].

Hypoglycemia can cause dysesthesia.

• A 26-year-old woman had numbness and tingling in her hands and feet on awakening [ ]. They were most pronounced in the hands and resolved within minutes. Her insulin regimen had recently been changed from NPH 50 U in the morning to 35 NPH and 5 regular in the morning and 8 NPH and 5 regular before dinner. The morning glucose concentrations averaged 3.3 mmol/l. One month after reducing the dose of insulin the symptoms of hypoglycemic neuropathy had disappeared and low morning glucose concentrations were rare.

Psychological and psychiatric effects of hypoglycemia Using evidence from auditory-evoked brain potentials and hypoglycemic clamps, it has been argued that antecedent hypoglycemia not only reduces awareness, but also that several aspects of cognitive function are attenuated during subsequent hypoglycemia 18–24 hours later [ ]. However, there were no effects of repeated hypoglycemia on cognitive function in patients included in the DCCT, a large American study that included more than 1400 patients, which showed that normalization of blood glucose prevents or delays the development of secondary (microvascular) complications in type 1 diabetes [ ].

Nevertheless, long periods of hypoglycemia can cause permanent brain damage. There is concern that frequent attacks of hypoglycemia impair brain function but there are few hard data.

Hypoglycemic coma due to insulin with extensive mental changes has been reported, including a review of six comparable cases in patients aged 37–56 years, whose coma lasted from 36 hours to 31 days [ ].

• A 37-year-old man could not be wakened in the morning. He had injected insulin without eating. His blood glucose was 1.5 mmol/l and he did not improve with intravenous glucose 16 g. In hospital he remained unconscious with a blood glucose of 12.2 mmol/l. There was no alcohol in the blood, his pH was 7.35, and he had a normal anion gap (18 mmol/l). His serum creatinine concentration was 288 μmol/l and his creatine kinase activity was high, suggesting rhabdomyolysis. A brain CT scan was normal and repeated electroencephalography showed slow waves with reduced voltages but no focal changes or irritation. He gradually recovered and was discharged after 6 days. Because of the dissociation between physical and mental improvement he was checked after 6 months and still had antegrade memory loss and problems with memory, complaining that he needed reminders on paper, and had less vitality and reduced emotionality.

Of the six reviewed patients, two died in coma; the other four had neuropsychological problems that did not improve after 6 months and up to 2 years. They had comparable electroencephalographic changes. During coma there was hypokalemia and hypocalcemia combined with increased lipolysis; this may have accounted for the permanent cerebral changes.

Of 20 patients with severe hypoglycemic coma and 20 with no or light coma, those with hypoglycemia had chronic depression and anxiety and performed persistently more poorly in several cognitive tests [ ].

In 42 patients with at least two episodes of severe hypoglycemia in the previous 2 years and 51 patients with no episodes, low blood glucose, hypoglycemia-impaired ability to do mental subtractions, and awareness of neuroglycopenia and hypoglycemia predicted future severe attacks of hypoglycemia [ ]. In another study, blood glucose awareness training increased adrenaline responses to hypoglycemia [ ]. However, in a reanalysis of data from the Diabetes Control and Complications Study, a large study relating the development of secondary complications to less strict control of blood glucose [ ], there was no effect of repeated hypoglycemia [ ].

The effect of hypoglycemia on cognitive function has been investigated in 142 children aged 6–15 years with type 1 diabetes intensively treated for 18 months; 58 had 111 periods of treatment. There were no effects on cognitive functions [ ]. In 29 prepubertal children, with diabetes for at least 12 months and using twice-daily mixed insulin, observed for two nights, asymptomatic hypoglycemia occurred in 13 children on the first night and in 11 children on the second night; cognitive performance was not altered, but mood was reduced [ ].

In healthy volunteers hypoglycemia caused significant deterioration in short-term attention, whereas sustained attention and intelligence scores did not deteriorate [ ].

Sensory effects of hypoglycemia Temporary blindness after severe coma [ , ] and retinal damage by gazing in the sun during hypoglycemia [ ] have been reported.

Differential diagnosis : When a patient does not react rapidly to sufficient therapy, other diagnoses have to be considered during suspected attacks of hypoglycemia.

Alcohol can confuse the diagnosis of hypoglycemia [ ]. Alcohol inhibits gluconeogenesis. It makes the patient more susceptible to hypoglycemia and can even cause hypoglycemia in healthy individuals. The symptoms of alcohol abuse and hypoglycemia are almost identical. If hypoglycemia is predominant, glucose administration will help. In attacks of hypoglycemia, the symptoms disappear rapidly after glucose intake.

Vascular episodes in older diabetics can mimic attacks of hypoglycemia. True epilepsy or strokes can cause comparable symptoms or accompany hypoglycemia.

Differentiation of hypoglycemia from hyperglycemic coma is usually not difficult. The development of hyperglycemic coma takes a longer time and the blood glucose concentration is high. However, urine testing may show positive glycosuria, if urine produced before the hypoglycemic period is still in the bladder. Even ketonuria may be present if a patient has been fasting for a long period.

Timing : The emergence of attacks of hypoglycemia depends on the times and amounts of food eaten and the duration of action of the insulin used (see Table 2 ). When only one type of insulin is used, the symptoms of hypoglycemia mostly occur at the end of the period of maximal insulin activity. Modern insulin therapy involves using a combination of long-acting and short-acting insulins. Long-acting insulins are given once or twice a day in combination with and/or in addition to short-acting insulins, which are given 2–4 times a day. Hypoglycemia can then develop at times when the combined effects are most prominent. Hypoglycemia in the mid-morning can be a consequence of the action of the long-acting insulin of the previous day and the short-acting insulin given earlier in the same morning. Hypoglycemia during the night or in the early morning can be caused by too much long-acting insulin or by short-acting insulin late at night without sufficient food. Repeated symptoms of hypoglycemia at the same time of the day indicate that the timing of the insulin injection or the relative proportions of long-acting and short-acting insulins have to be changed. If the interval between insulin injection and the subsequent meal is very short, the effective insulin concentration in the blood will still be low when glucose is absorbed from the gut. This will cause very high postprandial glucose concentrations. An increase in the dose of insulin will then cause hypoglycemia at a later time. It is therefore advisable to try first to increase the interval between the injection and start of the meal. New synthetic insulins, like lispro insulin, give a more rapid increase and fall of insulin concentrations than regular insulin; in that case, postprandial hypoglycemia can occur 1–3 hours after the injection [ ].

Risk factors : In a review of studies of tight blood glucose control in intensive care there were increased risks of hypoglycemia in patients who received intensive insulin therapy [ ]. The risk was greatest in those with sepsis. In those with conventional insulin therapy and sepsis the risk of hypoglycemia was 2.9% compared with 1.2% in those without sepsis. When septic patients received intensive insulin therapy the risk of hypoglycemia was 20% compared with 6.8% in those without sepsis. The risk of hypoglycemia coincided with an increased death rate in all patients, regardless of whether they received intensive or conventional insulin therapy (OR = 3.2 in the surgical ICU study and 2.9 in the medical ICU study). However, hypoglycemia was not thought to be directly responsible for the increased death rate, more a marker of those at increased risk. Larger studies appropriately designed to achieve normoglycemia are required to answer questions about the adverse effects and benefits of using intensive insulin therapy in this setting.

Insight into a possible mechanism of this increased mortality has come from a study of patients with subarachnoid hemorrhage who received continuous intravenous insulin if they had hyperglycemia. Although blood glucose concentrations remained within the reference range at above 4.4 mmol/l, 19 of 24 patients receiving insulin had a reduction in cerebral glucose concentration to less than 0.6 mmol/l, assessed using microdialysis samples. However three of seven patients who were not receiving insulin also had reduced cerebral glucose concentrations. This reduction occurred about at 4 days after the insulin infusion was started. This suggests that very low concentrations of cerebral glucose, which are harmful, can occur regardless of the blood glucose concentration. However, as patients did not become hypoglycemic it is not known whether hypoglycemia would have lowered the cerebral glucose concentrations further [ ].

Type of insulin The type of insulin used may influence the frequency of hypoglycemia.

The authors of a systematic review of whether there is a difference in the frequency and awareness of hypoglycemia induced by human or animal insulins identified 52 randomized, controlled trials; 37 were double-blind [ ]. They found no support for the supposition that human insulin per se affects the frequency, severity, or symptoms of hypoglycemia. In a few studies, mainly of less rigorous design, there was an effect when people were transferred from animal to human insulin, indicating increased frequency or reduced awareness of hypoglycemia.

Insulin detemir and glargine may reduce the incidence of hypoglycemia compared with NPH insulin, although results are variable [ ]. Studies in patients with type 2 diabetes tend not to be as significant. Studies in patients with type 1 diabetes show most benefit in reducing nocturnal hypoglycemia [ , ]. The difference between patients with type 1 and type 2 diabetes may relate to the lower frequency with which hypoglycemia is experienced in patients with type 2 diabetes.

Insulin glargine (n = 76) has been compared with NPH insulin + lente (n = 81) in a multidose regimen in adolescents (mean age 13 years, 45% girls) with type 1 diabetes. There were more episodes of blood glucose concentrations below 4 mmol/l in those who received glargine (116 events per patient year compared with 94), but there were no differences in severe hypoglycemic events [ ]. In a review in which insulin glargine was compared with NPH insulin in type 2 diabetes, although overall symptomatic hypoglycemia was not reduced in all studies, nocturnal hypoglycemia did appear to be reduced [ ].

In 28 children aged 3.1–8.3 years using twice- or thrice-daily insulin, blood glucose was measured with a subcutaneous continuous glucose monitoring system on 3 consecutive days and nights [ ]. Hypoglycemia was defined as a blood glucose concentration below 3.3 mmol/l for longer than 15 minutes. The prevalence of hypoglycemia was 10% and it was more common at night than during the day (19 versus 4.4%). Hypoglycemia at night had a longer duration (median 3.3 hours) and was asymptomatic in 91% of the episodes. The highest prevalence occurred at between 04.00 and 07.30. On a thrice-daily insulin injection regimen, nightly hypoglycemia was less frequent, but the frequency was higher on the following morning. With increasing age there was less hypoglycemia.

Susceptibility factors : A questionnaire about the prevalence of severe hypoglycemia in relation to susceptibility factors was answered by 387 patients in 1984 and by 641 patients in 1998; 178 patients answered both questionnaires [ ]. The following changed significantly from 1984 to 1998: multiple injection therapy increased from 71 to 98%, daily self-monitoring from 17 to 48%, episodes of nocturnal hypoglycemia from 76 to 83%, and lack of awareness from 40 to 55%; HbA _1c fell from 7.6 to 7.4%.

Children Hypoglycemia is an important problem in children [ ]. Children do not always establish the connection between the symptoms of threatening hypoglycemia and the danger involved. Overdosage of insulin is relatively common [ , ].

In a systematic review of short-acting analogues, 42 randomized controlled trials in 7933 patients with type 1, type 2, and gestational diabetes showed only minor differences in overall hypoglycemia [ ]. The standardized mean differences of episodes per patient per month were − 0.05 and − 0.04 in adults with type 1 and type 2 diabetes respectively, comparing short-acting analogues with regular insulin. There were no differences between children and pregnant women with type 1 diabetes and women with gestational diabetes. The change in HbA _1c was small. Hypoglycemia remains a clinical problem whether analogue or regular insulin is used.

Impaired hypoglycemic awareness was associated with an increased rate of severe hypoglycemia in 130 children and adolescents (aged 3–17 years) [ ]. One-third of the severe episodes developed without warning symptoms. Impaired awareness, young age, and recent attacks of hypoglycemia were independent susceptibility factors.

Elderly people Older patients are particularly susceptible to hypoglycemia [ ]. Factors such as cerebral blood flow, P _a O ₂ and P _a CO ₂ , permeability of the blood–brain barrier, and the presence of underlying neurological defects influence the hypoglycemic effects.

Critically ill patients The use of insulin during intensive therapy for critically ill patients has been reviewed [ ]. In a safety study from the GIST (post-stroke hyperglycemic management) trial in 25 patients using a GKI (glucose, potassium, insulin) infusion one patient required therapy for symptomatic hypoglycemia [ ]. Of 452 patients, mean age 75 years, 20 had blood glucose concentrations below 4 mmol/l within 30 minutes of stopping the GKI infusion and required intravenous dextrose. The patients had been randomized to GKI infusion or saline to maintain blood glucose concentrations at 4–7 mmol/l, and only 69 of the 452 had type 2 diabetes.

Management Patients have to be instructed to have a rapidly absorbed form of carbohydrate (for example dextrose tablets) available at all times and to use it when the first symptoms of hypoglycemia are felt. Often they fail to do so [ ]. If patients are used to self-monitoring, it is advisable that they monitor blood glucose first, although they should be told that a rapid drop in blood glucose concentration without reaching hypoglycemic values can cause a hypoglycemic reaction. Hypoglycemic reactions are sometimes difficult to discriminate from other feelings of malaise. Carbohydrate will always give rapid relief if the diagnosis is hypoglycemia and it is always safe to try taking it.

The conscious patient should take oral glucose at once. The treatment of choice in hypoglycemic coma is immediate intravenous injection of 20–50 g of dextrose. The patient may try to resist the injection, and help with the immobilization of the arm may therefore be needed. Injection of concentrated glucose solution outside a blood vessel leads to inflammatory and necrotic reactions. If intravenous injection of glucose is impossible, 1 mg of glucagon can be injected subcutaneously. However, in patients with residual insulin secretion (type 2 diabetes) glucagon may elicit extra insulin secretion and perpetuate hypoglycemia [ ]. The Epipen (a pen filled with a solution of adrenaline) is not a good substitute for the glucagon pen (in which glucagon must be dissolved before it can be used) in the treatment of hypoglycemia [ ].

When the patient has taken high doses of long-acting insulin, hypoglycemia may relapse after a single dose of glucose has provided temporary relief, and monitoring should continue for a longer period. After hypoglycemic reactions elicited by long-acting insulins or oral hypoglycemic drugs the patient should be observed for possible recurrence during the next few days [ ]. The longer the duration of coma, the poorer the prognosis. Persistent posthypoglycemic coma can be due to cerebral edema. Fever can accompany this severe form of coma, which requires treatment with intravenous mannitol and glucocorticoids. Severe coma can last for several days and require intensive management. Encephalopathy with neurological symptoms can be the consequence.

Infusion of too much glucose over an extended period in the treatment of hypoglycemia is dangerous [ ]. Glucose utilization in the postabsorptive state is 2 mg/kg/minute and can increase, if insulin concentrations are high, to 6 mg/kg/minute (about 600 g/day in a 70 kg individual). It is better to infuse glucose in concentrations of 10 or 20% rather than 40%.

The opinion of experts about when and how to treat asymptomatic hypoglycemia in children varies greatly [ ]. Hypoglycemia in children is often undetected. Using a subcutaneous continuous glucose monitoring system [ ] or the non-invasive Glucowatch biographer [ ], hypoglycemic periods were more frequent and prolonged than when only fingerprick testing was available. For treating hypoglycemia in children, small doses of glucagon are suggested. The contents of a 1 mg/ml ampoule can be drawn into a 1 ml U100 syringe. For children under 2 years, 20 micrograms should be given initially; 10 micrograms is added for every year, up to 150 micrograms at age 15. When the effect is insufficient, the dose can be repeated once or twice [ ].

Awareness of hypoglycemia Avoidance of hypoglycemia for 3 months can improve hypoglycemic awareness for a period of over 3 years [ ]. In contrast, supervised induction of brief hypoglycemia twice weekly reduced clinical awareness of hypoglycemia by 33% and reduced the important adrenaline response, so reducing the behavioral and physiological defences against hypoglycemia [ ].

Fat metabolism

• A 14-year-old adolescent girl, who had only used regular insulin developed lipoatrophy [ ]. Histology showed lipoblastoma-like cells, a possible sign of dedifferentiation. Skin tests showed no signs of allergy. The concentrations of immunoglobulins and TNF-alfa were normal.

Body weight

The DCCT Research Group has reported that patients in the intensive treatment group had substantial excess weight gain [ ]. In the first 9 months a group of patients who received intensive treatment gained 3.3 kg, compared with 1.2 kg in the control group; the percentage of people who gained more than 5 kg/m ² was consistently higher with intensive therapy. This weight gain was related to both lean body mass and fat.

In comparisons of insulin detemir with NPH insulin, insulin detemir appears to be associated with less weight gain [ ].

Of 12 981 patients with type 2 diabetes who took part in the PREDICTIVE study, an open observational study of patients taking insulin detemir, 2377 required basal insulin [ ]. Those with a BMI of more than 25 kg/m ² had a slight reduction in weight proportional to their baseline BMI, and those with a BMI of more than 31 had the greatest reduction in weight (1.5 kg). Patients with a BMI of less than 25 kg/m ² had a slight weight gain (0.55 kg). There are many possible explanations for this, including the fact that the study was not randomized. However, others have reported similar findings. Pooled data from two randomized studies of 900 people with type 2 diabetes, lasting 22 and 24 weeks, showed that patients with a BMI of more than 35 kg/m ² who used NPH insulin gained a mean of 2.4 kg and those who used insulin detemir lost a mean of 0.5 kg [ ]. Patients with a BMI of less than 35 kg/m ² gained weight, but those who took detemir gained less than those who took NPH insulin.