The dose-related hyper-and-hypokalaemic effects of salbutamol and its arrhythmogenic potential

1. The hypokalaemic effect of salbutamol after more than 30 min of administration has been well described. A hyper-and-hypokalaemic effect for adrenaline has been reported, but no such hyperkalaemic effect for salbutamol. 2. The possible hyper-and-hypokalaemic effects of salbutamol with the concomitant potential for pro-arrhythmia were assessed in the baboon (Papio ursinus). 3. Male and female baboons were anaesthetized with ketamine (15 mg kg-1) and maintained with 6% pentobarbitone as spontaneously breathing animals. Six baboons in each group received either 10, 100 or 500 micrograms kg-1 salbutamol i.v. Lead II of the ECG and femoral i.a. blood pressure were recorded continuously for 10 min. Arterial blood samples were collected at 0 min and then after 3 and 10 min of salbutamol administration. 4. All the animals developed sinus tachycardia (above 200 beats min-1) within 30 s of each dose of salbutamol administration and the high heart rate persisted throughout the experiment. All the animals were hyperkalaemic after 3 min and hypokalaemic after 10 min for each dose of salbutamol. Left ventricular conduction defects were seen in 3 animals during the hyperkalaemic phase. No arrhythmia was seen during the hypokalaemic phase. 5. Salbutamol has a transient hyperkalaemic and a more prolonged hypokalaemic effect in the baboon. The hypokalaemia could not be associated with arrhythmia although conduction defects were associated with the hyperkalaemia. 6. Since salbutamol is used as a bronchodilator in asthmatic patients and to treat acute hyperkalaemia, it is suggested that caution should be exercised when using salbutamol in high doses to treat acute asthma especially during the first few minutes of administration. The finding of hyperkalaemia with salbutamol questions its use in the treatment of hyperkalaemia.     

Effect of postexercise carbohydrate supplementation on muscle glycogen repletion in trained sled dogs

OBJECTIVE: To evaluate the effect of immediate postexercise carbohydrate supplementation on muscle glycogen (MG) repletion during the first 4 hours of recovery in sled dogs. ANIMALS: 24 Alaskan Huskies. PROCEDURE: Dogs were assigned to 1 of 3 treatment groups, and a muscle biopsy specimen was obtained 1 hour before and immediately (group A) or 4 hours (groups B and C) after a 30-km run. Immediately after exercise, dogs in group A and group C were given water; dogs in group B were given a glucose polymer solution (1.5 g/kg of body weight) in water. RESULTS: At 4 hours after exercise, MG concentration was significantly greater in group-B than in group-C dogs; the value in group-C dogs was not different from the value in group-A dogs immediately after exercise. Assuming similar rates of glycogen depletion between treatment groups, during the first 4 hours of recovery, group-B dogs replaced 49% of the glycogen used during exercise. Plasma glucose concentration was significantly greater in group-B than in group-A and group-C dogs at 100 minutes after exercise. CONCLUSIONS: Immediate postexercise carbohydrate supplementation in sled dogs leads to increased glucose concentration, which in turn promotes more rapid rate of MG repletion in the first 4 hours of recovery than is observed in dogs not given supplements. CLINICAL RELEVANCE: For dogs running in multiple heats on a single day or over several consecutive days, immediate postexercise carbohydrate supplementation may promote more rapid and complete recovery between bouts of exercise.     

The effect of plasma free fatty acids and long-chain triglycerides on glucose metabolism in uncomplicated falciparum malaria

To investigate the therapeutic potential of increased plasma free fatty acid (FFA) and triglyceride concentrations in hypoglycaemic patients receiving quinine, 32 untreated Thai adults with uncomplicated falciparum malaria were allocated at random to one of 4 regimens: 2 mg/kg/min dextrose infused over 60 min either alone (group A) or with a prior injection of 5000 units of heparin and simultaneous Intralipid infusion (group C), or 4 min/kg/min dextrose alone (group B) or with heparin and Intralipid (group D). Quinine (10 mg/kg) was also infused over 60 min in all cases. In patients of groups A and C, mean changes in plasma glucose concentrations from the beginning to the end of the infusion were 0.1 (SD 0.8) and 1.0 (SD 0.7) mmol/L respectively (P = 0.015). In groups B and D, plasma glucose increased by 1.8 (SD 1.2) and 2.2 (SD 0.4) mmol/L respectively (P < 0.5). Plasma FFA levels fell by approximately 50% during the infusion in groups A and B but increased by a similar percentage in groups C and D. Despite significant mean increases in plasma insulin during the infusion (from 12.2 milliunits (mu)/L in group A to 38.8 mu/L in group D), no rebound hypoglycaemia was observed in any patient during the ensuing 7 h. These data suggest that the glycaemic response to dextrose given at high rates, which match average glucose utilization in a severely ill patient with malaria, is not augmented by increased plasma FFA and long-chain triglycerides. However, this strategy increases the glycaemic efficacy of lower dextrose infusion rates and the combination could, therefore, reduce the volumes of hypertonic dextrose required to prevent hypoglycaemia in severely ill patients in whom optimal fluid balance is crucial.     

Routine use of low-dose intravenous insulin infusion in severe hyperglycaemia

A review if presented of the use of low-dose insulin infusion in the management of 58 episodes of severe diabetic hyperglycaemia. Neutral insulin in a dosage of 2-4 units per hour is infused via a paediatric giving set to achieve a sustained physiological elevation of insulin levels. This method is safe, simple and rapidly effective in lowering the blood glucose level, the mean rate of fall (62 mg/100 ml/hr, or 11% per hour) being unaffected by prior insulin therapy, acidosis or ketonuria. Classification of the hyperglycaemia as ketoacidotic or hyperosmolar is unnecessary before insulin therapy is instituted, as the relative decline in glucose level is the same in the hyperosmolar non-ketotic group as in the others. Proven infection significantly lowers the rate of fall of glucose level. Hypoglycaemia and hypokalaemia are rare during low-dose infusion. Early and adequate replacement with potassium phosphate is recommended, oral potassium supplements being continued for several days. Bicarbonate therapy is rarely indicated in the management of acidosis. No patient had cerebral oedema during treatment, and one elderly patient with extensive pneumonia and empyema died during the infusion. It is suggested that continuation of low-dose insulin infusion, together with 5% dextrose solution, after the plasma glucose level reaches 200 mg/100 ml, may hasten the clearance of ketones, preventing relapse.     

Changing from porcine to human insulin

The development of hypoglycaemia unawareness is associated with long duration of diabetes, improved glycaemic control, alcohol intake and recurrent hypoglycaemia. However, current evidence suggests that neither frequency of severe episodes nor mortality from hypoglycaemia are increased following a change from animal to human insulin. Nevertheless, a small number of patients continue to report an alteration in the nature of hypoglycaemic warning symptoms following a change in insulin species. This is possibly a consequence of a reduced catecholamine response to lowering blood glucose levels or to species differences in the effect of insulin on central nervous system function. In practical terms, it seems sensible to warn patients that the nature of the symptoms associated with hypoglycaemia might alter following conversion from porcine to human insulin. At the time of the changeover, patients should be encouraged to perform frequent blood glucose measurements. Also, the usual insulin dose should be reduced by 10% at the start of human insulin treatment. Other aspects of insulin treatment including injection technique, meal timing, exercise, etc. should be discussed. For patients who are convinced that loss of warning of hypoglycaemia occurred after conversion from porcine to human insulin, a change back to animal insulin would be preferred to relaxing glycaemic control in the first instance. Pressure should be brought to bear on the pharmaceutical industry to maintain the availability of animal insulins for the small number of patients who have experienced problems with human insulin.     

Muscle damage induced by experimental hypoglycemia

To determine organ damage due to hypoglycemia, we studied the effects of insulin dose and hypoglycemia duration on serum enzyme activity in rabbits. Thirty rabbits were randomly divided into five groups according to hypoglycemia duration and insulin dose: A2, hypoglycemia for 30 minutes with 2 U/kg insulin; A10, hypoglycemia for 30 minutes with 10 U/kg insulin; B2, hypoglycemia for 60 minutes with 2 U/kg insulin; B10, hypoglycemia for 60 minutes with 10 U/kg insulin; and C, no hypoglycemia with 10 U/kg insulin and 50% glucose. Insulin-induced hypoglycemia was reversed by intravenous injection of glucose. Alterations in serum enzyme activity and creatine kinase (CK) isoenzyme distribution were determined before and after insulin injection. Serum CK activity increased significantly in all hypoglycemic groups compared with preinjection values, and tended to remain high for 24 hours in both groups A10 and B10. Serum activity of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and lactate dehydrogenase (LDH) increased only in group B10. In addition, the level of band 4 of serum CK isoenzymes, which exists predominantly in skeletal muscle and myocardium, increased significantly in group B10. These results suggest that the increase in both serum enzyme and CK band 4 isoenzyme activities during hypoglycemia is primarily due to damage in muscle rather than liver, and that the hypoglycemia duration and insulin dosage may influence the extent of organ damage.     

In vitro effects of diazepam on human ciliary function

This study demonstrated the in vitro effect of diazepam on human ciliary beat frequency (CBF) from fifteen normal subjects using brush cytology. Tube A contained culture medium, tube B the diluent that diazepam intravenous injectate was carried in and culture medium (controls). Tube C, D and E contained, 0.4 mg/l, 4.0 mg/l and 40.0 mg/l of diazepam in culture medium respectively. The mean effective diazepam concentration in plasma is 0.4 mg/l. CBF was measured photometrically. The most vigorous cilia were measured in 5 areas taking 10 readings on each sample, 30 minutes and 1 hour after mixing. Standard deviation (SD) and confidence limits were calculated along with significance testing (p < 0.05) using the paired t-test and ANOVA. The mean of the CBF of tubes A and B were 13.44 (SD 2.65) and 13.67 (SD 2.48). There was a reduction of the CBF with increasing concentrations of diazepam at 30 minutes, 11.32 (SD 2.14), 10.29 (SD 1.58) and 4.14 (SD 1.57) tubes C. D and E respectively. There was a significant lowering in CBF of 17% (p < 0.01) of diazepam at the mean effective plasma level (tube C) when compared against the controls. CBF decreased over time and at 1 hour was 10.57 (SD, 1.36), 9.02 (SD, 1.39) and 3.58 (SD, 1.31) tubes C, D and E respectively. A proposed mechanism of altered intracellular calcium flux via the action diazepam on GABA receptors is described.     

Effect of environmental temperature on glucose-induced insulin response in the newborn rat

Blood glucose and plasma insulin and glucagon concentrations were determined in full-term rats delivered by cesarean section and exposed to 37 degrees C. or 24 degrees C. environmental temperature during the first hours of extrauterine life. When newborn rats were maintained at thermal neutrality (37 degrees C.), a transient period of hypoglycemia of two hours occurred, associated with a rapid fall in plasma insulin and a rise in plasma glucagon concentrations. During cold exposure (24 degrees C.), the blood glucose level remained stable over the four hours studied; the decrease of plasma insulin was sluggish while the rise of plasma glucagon was unchanged. In newborn rats maintained at 37 degrees C., an intraperitoneal glucose load one hour after delivery produced a marked rise in blood glucose and plasma insulin concentrations one hour later. The distribution of experimental points suggested a sigmoidal dose-response curve. By contrast in newborn rats kept at room temperature (24 degrees C.) the same glucose load did not induce any increase in plasma insulin in spite of hyperglycemia. However, phentolamine resulted in pronounced plasma insulin rise in hypothermic newborns in response to glucose administration. From these observations it is concluded that the in-vivo unresponsiveness of the beta cells to glucose at birth, reported by others, is mainly due to the experimental conditions.     

Measuring hospital performance: are we asking the right questions?

The aim of this randomized cross-over trial was to evaluate the dose effect and systemic absorption of epinephrine nebulized at 2 and 5 mg in comparison with salbutamol (5 mg). Thirteen asthmatic patients (29 +/- 15 years, 4 men and 9 women) were randomly assigned to receive one nebulization of each of the three treatment regimens at 24 h interval. The evaluation concerned forced expiratory volume in 1 second (FEV1), heart rate, respiratory rate and arterial pressure. All measurements were done at baseline, every 15 minutes during the first hour, and hourly thereafter until return to baseline FEV1. Serum potassium was measured at baseline (T0) and sixty minutes after (T60). Plasma levels of epinephrine were measured at T0, T20, T60. Fifteen minutes after the beginning of nebulization FEV1 improved significantly over baseline in the three groups. These changes were similar in the three groups until T45, while FEV1 improvement was significantly greater in A5 and S groups than A2 group (+640 +/- 470 ml, +721 +/- 349 ml, +406 +/- 306 ml in A5, S and A2 groups respectively, p < 0.01). Bronchodilation lasted significantly longer with salbutamol than with epinephrine (p < 0.05). No side effects were recorded in spite of substantial and dose-dependent systemic absorption of epinephrine. CONCLUSION: Increasing epinephrine doses produces greater bronchodilation without additional side effects. However this bronchodilation lasts shorter than with salbutamol.     

Effect of galactose and sugar substitutes on blood insulin levels in normal and obese individuals

Eighteen male patients between the ages of 25 and 50 were given on a double blind randomized basis (A) 40 gms. galactose (B) 50 gms. arabinogalactan and 0.11 gm. sodium saccharin (C) 2 gm. methyl cellulose and 0.083 gm. sodium saccharin and (D) 4 gm. galactose, all in 200 ml water. Blood glucose, galactose and insulin levels were determined during a six hour period before and after ingestion. The three first mentioned solutions tasted equally sweet, the fourth was essentially tasteless. None of these feedings altered plasma insulin or glucose levels. It appears that in contrast to other conclusions reached by earlier investigators sweet taste is unable to induce insulin secretion through neurogenic pathways.     

Errors in the diagnosis of benign prostatic hyperplasia

Monitor blood glucose level before, during and for up to 24 hours after exercise. Ensure refined carbohydrate snack is taken prior to exercise. Reduce insulin dosage if possible. Inject insulin away from any exercising muscle. Remember that glycogen stores are replenished in two phases: immediately after the exercise and two to three hours later. These are the key risk times for hypoglycaemia. If blood glucose control is poor (14 mmol/litre or higher) prior to exercise, the 'stress' effect of the exercise may cause further increases in the blood sugar level unless control is achieved.     

Effects of somatotropin and salbutamol in three genotypes of finishing barrows: blood hormones and metabolites and muscle characteristics

We evaluated the effects of recombinant porcine somatotropin (pST) and the beta-adrenergic agonist salbutamol on plasam endocrine and metabolite profiles and muscle chemistry in three genotypes of growing barrows (n = 96, 139 d old). Treatments were in a 2 x 2 x 3 factorial arrangement, and main effects were pST (0 or 4 mg/d) and salbutamol (0 or 2.75 ppm); the three genotypes including purebred Meishan (M), 1/4 Duroc, 3/4 White composite (D x Wc), and 1/4 Meishan, 3/4 White composite (M x Wc). Individual pigs were injected daily with buffer or pST at 0700 and allowed ad libitum access to a corn-soybean meal diet (1.2% lysine) and water for 33 d. Plasma was obtained 4 h after injection and 3 h postprandially on d 0, 14, and 28 for determination of growth hormone (GH), insulin, IGF-I, glucose, urea N (PUN), NEFA, triglycerides, and cholesterol. Longissimus and semitendinosus samples were obtained, and protein, RNA, and DNA were quantified and loin chop shear force was measured. In general, plasma hormones and metabolites on d 14 and 28 were not affected by salbutamol in the absence of pST, although salbutamol tended to increase d 14 and 28 GH concentrations. Salbutamol lowered plasma IGF-I (d 14 and 28, P < .05), insulin (d 14, P < .01), and NEFA (d 28, P = .07) when pST was administered, although concentrations still exceeded those for control pigs. Salbutamol reduced (P < .05) IGF-I in M and M x Wc pigs, and GH was not changed in M pigs. Meishan pigs had a greater increase in glucose with pST than M x Wc or D x Wc pigs, although the effect was not consistent over time. Individual treatment with pST caused GH, insulin, IGF-I, glucose, NEFA, and triglycerides to be increased and PUN to be decreased on d 14 and 28. Cholesterol on d 14 and 28 was decreased by pST in M pigs, whereas no effects were found in the other genotypes. Muscle protein and RNA were increased by salbutamol and were consistently lowest for M pigs. Furthermore, pST did not affect muscle protein, but it increased RNA more in M pigs than in the others. Overall, pST and salbutamol seemed to act separately and by different mechanisms to alter muscle composition, but blood criteria generally representing fat metabolism (insulin, glucose, NEFA, triglycerides) were interactively affected. Meishan pigs tended to have greater changes in muscle and plasma composition with pST treatment than did M x Wc or D x Wc pigs.     

Factitious insulinoma

A 32-year-old woman was admitted with signs of recurrent hypoglycaemia. Within 72 hours hypoglycaemia was successfully provoked by prolonged fasting. Also, blood samples demonstrated high levels of serum insulin and C-peptide and the insulin-glucose ratio was abnormally high. An insulinoma was strongly suspected. However, extensive imaging displayed no tumour in the pancreas. The patient also had extensive psychological and social problems. The psychiatrist suggested a factitious disorder. High serum concentrations of insulin and C-peptide in combination with the psychiatric disorder led to the suspicion of abuse of sulfonylurea derivatives by the patient. This was confirmed by toxicological screening. A patient with unexplained hypoglycaemia, especially if an insulinoma cannot be detected, should be suspected of abusing sulfonylurea derivatives.     

The three-dimensional bone interface of an osseointegrated implant. I: A morphometric evaluation in initial healing

This study was designed to determine the effect of a potent cholecystokinin antagonist, L-364,718, on canine pancreatic endocrine function following partial pancreatectomy. Plasma glucose, insulin, and glucagon were determined over a 2-hr interval following an intravenous bolus of 0.5 g/kg glucose in a 50% solution. The following groups were established: normal animals (group A, n = 5), normal animals pretreated with 20 nmole/kg L-364,178 (group B, n = 5), partially pancreatectomized animals (group C, n = 5), and partially pancreatectomized animals pretreated with 20 nmole/kg L-364,178 (group D, n = 5). In contrast to animals with an intact pancreas, pretreatment with L-364,718 following partial pancreatectomy resulted in a significant decrease in peak insulin (group C = 132.8 +/- 13.0 microU/ml vs Group D = 90.4 +/- 16.1 microU/ml, P < 0.05) and the basal-to-peak insulin difference (group C = 111.9 +/- 11.5 microU/ml vs group D = 77.5 +/- 16.6 microU/ml, P < 0.05). Despite this, the rate of glucose utilization (K value) was significantly increased in the partially pancreatectomized animals given the antagonist (group C = -1.22 +/- 0.22%/min vs group D = -2.79 +/- 0.427%/min) and there were no significant differences in basal or peak glucose when comparing the groups given L-364,718 with the groups given placebo (group A vs B and group C vs D). Thus, the CCK antagonist L-364,718 significantly decreases peak insulin in partially pancreatectomized animals but not in nonoperative control animals. There is a paradoxical increase in the rate of glucose utilization but no effect on glucose homeostasis. The effect of this antagonist in other models of reduced islet cell reserve (i.e., pancreas transplantation) remains to be determined.     

Changes in plasma and urine magnesium following subcutaneous insulin

In 7 of 13 patients given subcutaneous insulin as part of the assessment of their endocrinological status, serum magnesium levels fell below the lower limit of the normal range for the laboratory 1 to 2 hours after insulin administration. This relative hypomagnesaemia was not accompanied by any change in the red cell magnesium concentration, and did not bear a close relationship to the degree of hypoglycaemia developed. During the 2 1/2 hour period following insulin therapy there was renal conservation of magnesium ion.     

The action profile of lispro is not blunted by mixing in the syringe with NPH insulin

OBJECTIVE: To assess the effect of mixing the insulin analog lispro (Humalog) with NPH (Humulin I) before injection on lispro's fast, short action profile. RESEARCH DESIGN AND METHODS: A total of 12 healthy volunteers received subcutaneous abdominal injections of 0.1 U/kg regular insulin and 0.2 U/kg NPH insulin as follows: lispro and NPH injected separately (treatment group A), lispro and NPH mixed in the syringe up to 2 min before single injection (treatment group B), and human regular insulin and NPH mixed and injected as in group B (treatment group C), on separate occasions, in random order. Plasma glucose was maintained for 12 h by intravenous 20% glucose. Pharmacokinetic and pharmacodynamic parameters were compared by analysis of variance for repeated measures. RESULTS: Peak plasma insulin levels (2.6 +/- 0.8 vs. 2.2 +/- 0.6 vs. 1.9 +/- 0.6 ng/ml, P = 0.075), total glucose infused (121.5 +/- 32.8 vs. 135.0 +/- 49.0 vs. 117.3 +/- 39.9 mg.kg-1.min-1, P = 0.53), and maximum glucose infusion rate (GIRmax) (8.3 +/- 0.9 vs. 8.0 +/- 1.7 vs. 7.1 +/- 2.4 mg.kg-1.min-1, P = 0.65) were not significantly different between treatments. The times until peak insulin concentrations were similar in treatment groups A and B, but significantly shorter than in treatment group C (0.9 +/- 0.3 and 1.2 +/- 0.2 vs. 2.0 +/- 0.4 h, respectively, P = 0.042). The times until GIRmax were also not different (113.9 +/- 41 and 122.0 +/- 45 vs. 209.0 +/- 51.3 min, respectively, P = 0.002). The glucose infusion rate (GIR) then fell to 50% GIRmax more quickly in treatment groups A and B than in treatment group C (239.9 +/- 40.5 vs. 292.4 +/- 133.3 vs. 399.5 +/- 78.3, respectively, P = 0.005). CONCLUSIONS: The action profile of lispro is not attenuated by mixing lispro with NPH in the syringe immediately before injection. The advantages are available to those individuals who need to combine types of insulin before injection to achieve optimal diabetes control.     

Changes in attention with hypo- and hyperglycaemia in children with insulin dependent diabetes mellitus

We compared the results of a computerized attention test (TOVA) in 38 children with insulin dependent diabetes mellitus in relation to various spontaneously occurring blood glucose levels. Testing was performed at the following blood glucose levels: <3.3 mmol/1 (hypoglycaemia), 3.3-8.3 mmol/1 (normoglycaemia) and >8.3 mmol/1 (hyperglycaemia). The attention (sum of errors and response time) varied significantly with the blood glucose level (P = 0.002). The highest number of errors of omission and the longest response time was observed during the test run with hypoglycaemia. Age, sex, age at manifestation of the disease, metabolic control and the results of the intelligence test had no significant influence on these results. We found that attention in children with diabetes was significantly reduced compared to TOVA norms especially during mild hypoglycaemia (P < 0.001). Irrespective of the blood glucose levels, reaction time and the variability of the reaction time differed significantly between TOVA norms and diabetic children (P < 0.01). CONCLUSION: In children with diabetes mellitus a significant reduction in attention was found at mild hypoglycaemia but as well at low normal blood glucose levels. Attention deficits due to transient lowering of blood glucose may therefore occur in diabetic children even before they are aware of hypoglycaemic symptoms.     

Double blind clinical and laboratory study of hypoglycaemia with human and porcine insulin in diabetic patients reporting hypoglycaemia unawareness after transferring to human insulin

OBJECTIVES: To compare awareness of hypoglycaemia and physiological responses to hypoglycaemia with human and porcine insulin in diabetic patients who reported loss of hypoglycaemia awareness after transferring to human insulin. DESIGN: Double blind randomised crossover study of clinical experience and physiological responses during slow fall hypoglycaemic clamping with porcine and human insulin. SETTING: Clinical investigation unit of teaching hospital recruiting from diabetes clinics of five teaching hospitals and one district general hospital. SUBJECTS: 17 patients with insulin dependent diabetes mellitus of more than five years' duration who had reported altered hypoglycaemia awareness within three months of transferring to human insulin. MAIN OUTCOME MEASURES: Glycaemic control and frequency of hypoglycaemic episodes during two months' treatment with each insulin. Glucose thresholds for physiological and symptomatic responses during clamping. RESULTS: Glycaemic control did not change with either insulin. 136 hypoglycaemic episodes (eight severe) were reported with human insulin and 149 (nine severe) with porcine insulin (95% confidence interval -4 to 2.5, p = 0.63). 20 episodes of biochemical hypoglycaemia occurred with human insulin versus 18 with porcine insulin (-0.8 to 1, p = 0.78). During controlled hypoglycaemia the mean adrenaline response was 138 nmol/l/240 min for both insulins; neurohormonal responses were triggered at 3.0 (SE 0.2) versus 3.1 (0.2) mmol/l of glucose for adrenaline and 2.5 (0.1) versus 2.5 (0.1) mmol/l for subjective awareness. CONCLUSIONS: These data suggest that human insulin per se does not affect the presentation of hypoglycaemia or the neurohumoral, symptomatic, and cognitive function responses to hypoglycaemia in insulin dependent diabetic patients with a history of hypoglycaemia unawareness.