It has been suggested that much effort expended in teaching diabetic diets is ineffective and wasteful. We have tested a different system by randomly allocating 75 newly diagnosed obese Type 2 diabetic patients to usual 'unstructured' clinic care or to group education by diabetes specialist nurses and a dietitian. Patients allocated to group education attended five 90-min group sessions during the first 6 months. Six months after diagnosis they had lost more weight (median (95% Cl), 7 (5.5-9) vs 2(1-5)kg, p less than 0.002) and were better controlled (HbA1:7.5 (7.0-8.1) vs 9.5 (8.7-10.4)%, p less than 0.001) than those randomized to the usual clinic system. At 1 year (after no further visits) the difference in weight loss was less (5.5 (4-6.5) vs 3 (2-4) kg, p less than 0.05) and diabetic control was similar (HbA1:9.0(8.2-9.8) vs 9.9(8.9-10.9)%. At 1 year only 14(39%) of the education group and 9(23%) of those attending the clinic had a fasting blood glucose less than 7.0 mmol l-1.

A knowledge of the renal threshold for glucose is important in the interpretation of urine tests in diabetes. We describe a simple method for determining renal threshold from blood and urine tests performed by the patients themselves. In a group of 65 insulin-dependent diabetic patients there was a wide variation in renal threshold, with a mean of 130 mg/dl (range 54-180 mg/dl). Threshold tended to rise with age, and it is suggested that the higher the renal threshold, the higher is the mean blood glucose achieved by the patient (r = 0.50, P = < 0.001). The change in blood glucose required to convert urine tests from 0% to 2% is very variable and ranged from 36 to 288 mg/dl (mean 110 mg/dl). The simple method that we describe may improve understanding of the significance of urine test results without the necessity for hospital admission.

Summary and conclusionsAdmission to hospital is usually recommended to achieve the best possible diabetic control during pregnancy. We have used blood glucose monitoring at home to find out if patients can achieve equally good control outside hospital. Twenty-five consecutive diabetic patients were studied, of whom 20 had taken insulin before pregnancy. Six of their 14 previous pregnancies had ended in perinatal death. The 25 women performed 4247 blood glucose measurements during their pregnancies. Overall the mean blood glucose concentration was 71 mmol/l (128 mg/100 ml); before meals the mean was 6 5 mmol/l (117 mg/100 ml). Mean concentrations were lower in the third trimester, but at no stage was control in hospital significantly better than at home. The mean hospital stay

The characteristics and outcome of 68 newly diagnosed Type 2 diabetic patients who presented with clinically evident peripheral neuropathy were compared with matched controls who had no neuropathy at diagnosis. All subjects (34 male) whose median age was 68 (range 47-89) yr were identified from a computerized diabetes register and presented in 1982-1990. The groups were compared at diagnosis for haemoglobin A1, body mass index, blood pressure, smoking, and alcohol consumption, and for co-existent coronary and peripheral vascular disease. Mortality and morbidity were recorded to March 1991. Significantly more patients with neuropathy had co-existent peripheral vascular disease: 24(35%) compared to 6(9%) controls (p = 0.0021). Twenty (30%) of those with neuropathy and no controls had retinopathy at diagnosis, which was sight-threatening in 10. Seven (10%) with neuropathy but no controls presented with foot ulcers, one requiring limited amputation. Two more patients with neuropathy and one control subsequently developed foot ulcers resulting in one or more amputation in each group. Twenty-one (31%) of those with neuropathy and 14 (21%) controls died (p = 0.2109). In conclusion more diabetic patients with clinically evident peripheral neuropathy at diagnosis have peripheral vascular disease than matched patients without neuropathy. It is likely that macrovascular disease either exacerbates or causes the neuropathy in this group of patients. They are at high risk of developing foot ulceration and high priority should be given to foot care in planning their management.

Blood lipid estimations were performed on 118 diabetics receiving different forms of treatment and with varying degrees of control of their diabetes. Elevated blood lipids were found predominantly in elderly patients being treated with hypoglycaemic drugs and could not be accounted for by poor diabetic control. Patients with elevated plasma triglycerides were significantly heavier tha~ those with normal levels.