In a prospective study to analyse stroke topography and outcome in diabetics and to determine the prognostic value ofblood glucose and glycosylated haemoglobin estimation, we evaluated 176 patients with acute stroke. The patients were classified into four groups on the basis of history, fasting glucose, and glycosylated haemoglobin: euglycaemic patients with no history of diabetes, stress hyperglycaemia, newly diagnosed diabetics, and known diabetics. A high prevalence of undiagnosed diabetes was shown. No difference was found in the type or site of stroke between the four groups. No difference was found in the site of symptomatic or incidental lesions on computerised axial tomography. Patients with stress hyperglycaemia and known diabetics had more severe strokes. Mortality was higher in patients with stress hyperglycaemia, newly diagnosed diabetics, and the combined diabetes groups. This increased mortality was evident in the hyperglycaemic and diabetic groups, even after excluding patients with cerebral haemorrhage. Stroke severity and mortality also increased independently with blood glucose in the euglycaemic group. We conclude that there is a correlation between admission glucose concentration, diabetes, and poor stroke outcome, which may not be attributed to stroke type or location.Epidemiological and necropsy studies show that diabetic patients have a higher incidence of ischaemic stroke than non-diabetic patients.' 2 In the Framingham study the incidence of thrombotic stroke was 2-5 times higher in diabetic men and 3-6 times higher in diabetic women than in those without diabetes.3 Wolf and Kannell reported that even when other risk factors such as hypertension and ischaemic heart disease are taken into account diabetes remains an independent risk factor for stroke.4 Previous studies have found a range of prevalence of undiagnosed diabetes in acute stroke populations from 6%' to 42%6.The type and topography of diabetes-related cerebral infarction may differ from brain infarcts in non-diabetics. In a necropsy survey Kane and Aronson7 found that diabetics had more lacunar lesions when compared with non-diabetics, especially in the distribution of the parasagittal perforating arteries. Peress et al8 also reported a higher occurrence of lacunar infarcts in diabetics compared with non-diabetic patients. In the Harvard cooperative stroke registry,9 hypertension and diabetes were present respectively in 75% and 29% of lacunar cases and 71% and 43% of cases in the South Alabama population study.'o Several animal studies of experimental cerebral ischaemia have shown that hyperglycaemia increases the severity of ischaemic brain dam-

A large number of experimental studies in animals and retrospective or non-randomised prospective studies in humans provide support for the concept that the microvascular complications of diabetes mellitus are dependent on hyperglycaemia. This review focuses on four potential biochemical pathways linking hyperglycaemia to changes within the kidney which can plausibly be linked to the functional and structural changes characterising diabetic nephropathy. These four pathways are the polyol pathway, non-enzymatic glycation, glucose autoxidation and de novo synthesis of diacylglycerol leading to protein kinase C and phospholipase A2 activation. Rather than being independent, there are several potential interactions between these four pathways which may explain confusing and overlapping effects observed in studies examining inhibitors of individual pathways. As many of the steps which follow on glucose metabolism are subject to modification by dietary and pharmacological means, the further delineation of the pathogenetic sequence leading to tissue damage in diabetes should allow a logical and effective approach to the prevention or treatment of the complications of diabetes.

The effect of calcium supplementation on bone mineral density (BMD) was evaluated in female twin pairs aged 10-17 years with a mean age of 14 years. Forty-two twin pairs (22 monozygotic, 20 dizygotic; (including one monozygotic pair from a set of triplets) completed at least 6 months of the intervention: 37 pairs to 12 months and 28 pairs to 18 months. BMD was measured by dual-energy X-ray absorptiometry (DXA). In a double-blind manner, one twin in each pair was randomly assigned to receive daily a 1000 mg effervescent calcium tablet (Sandocal 1000), and the other a placebo tablet similar in taste and appearance to the calcium supplement but containing no calcium. Compliance (at least 80% tablets consumed), as measured by tablet count, was 85% in the placebo group and 83% in the calcium group over the 18 months of the study, on average increasing dietary calcium to over 1600 mg/day. There was no within-pair difference in the change in height or weight. When the effect of calcium supplementation on BMD was compared with placebo at approximately 6, 12 and 18 months, it was found that there was a 0.015 +/- 0.007 g/ cm2 greater increase in BMD (1.62 +/- 0.84%) at the spine in those on calcium after 18 months. At the end of the first 6 months there was a significant within-pair difference of 1.53 +/- 0.56% at the spine and 1.27 +/- 0.50% at the hip. However, there were no significant differences in the changes in BMD after the initial effect over the first 6 months. Therefore, we found an increase in BMD at the spine with calcium supplementation in females with a mean age of 14 years. The greatest effect was seen in the first 6 months; thereafter the difference was maintained, but there was no accelerated increase in BMD associated with calcium supplementation. The continuance of the intervention until the attainment of peak bone mass and follow-up after cessation of calcium supplementation will be important in clarifying the optimal timing for increased dietary calcium and the sustained, long-term effects of this intervention.