Dementia and Hypoglycemic Episodes in Patients With Type 2 Diabetes Mellitus

Abstract

To the Editor: Dr Whitmer and colleagues reported a relationship between severe hypoglycemia and incident dementia in a large cohort of older people with type 2 diabetes. Although severe hypoglycemia may cause permanent cognitive impairment, the results of this study should be viewed with caution. It seems unlikely that a single episode of severe neuroglycopenia could increase the risk of dementia by 50%. Premorbid cognitive ability was assessed solely by educational attainment. Intellectually able people with diabetes may compensate for declining cognition and be less susceptible to the development of hypoglycemia. Vascular dementia is more common in people with lower cognitive ability and may confound a putative association with hypoglycemia. Cerebrovascular disease may both cause dementia and enhance the risk of hypoglycemia. Although the results were adjusted for the most overt diagnoses of stroke, which have a prevalence in the general population of 2% to 3%, no adjustment was made for subclinical cerebrovascular disease, which has a prevalence of around 18%. Other important confounding variables that were not excluded were a history of alcoholism, epilepsy, psychiatric illness, and head injury. Because the analysis did not adjust for many confounding factors, it was more likely to predict an association between hypoglycemia and dementia. Most severe hypoglycemia is treated in the community, and while 30% of people with insulin-treated type 2 diabetes require emergency assistance, very few are treated in the hospital. By contrast, hospital inpatient hypoglycemia has many causes, with renal insufficiency, malnutrition, sepsis, and liver disease being prominent factors often occurring in seriously ill people. In the study by Whitmer et al, episodes of severe hypoglycemia were identified retrospectively from coded entries in electronic hospital databases, which are frequently inaccurate. This may account for the surprising finding that 7.7% of those treated in the hospital for severe hypoglycemia had diet-controlled diabetes. Accurate measurement of the incidence of severe hypoglycemia in a diabetic population requires prospective recording; during this longitudinal study, most (unrecorded) hypoglycemic events in this cohort of patients would have occurred in thecommunity.Relianceonretrospective recordsofhospitaltreated hypoglycemia may have introduced selection bias, identifying people least able to self-manage diabetes and therefore most prone to severe hypoglycemia. Furthermore, the potential effects of changes in treatment modalities and of other comorbidities (such as hypertension) during the period of assessment have not been evaluated. Alex J. Graveling, MBChB, MRCP Brian M. Frier, BSc, MD, FRCP brian.frier@luht.scot.nhs.uk Department of Diabetes Royal Infirmary of Edinburgh Edinburgh, United Kingdom