British Journal of Dermatology | 2021
Using electronic health records to examine links between atopic dermatitis and obesity
Abstract
Atopic eczema is a common condition in adults and children, and is characterized by chronically occurring pruritis. Obesity is also common in both children and adults, and is characterized by excess inflammation in the form of adipose tissue. Given the inflammatory nature of both conditions, and recent evidence that adipokines may be a marker of atopic eczema, it stands to reason that risk may be increased for both conditions concurrently. Moreover, research has also found links between eczema and cardiovascular disease, a likely adverse outcome of obesity. Studies from the USA and Asia have shown associations between atopic eczema and obesity, but population-level links were not found in studies from Europe. In this issue of the BJD, Ascott et al. use electronic health record (EHR) data from patients who received care in UK general practices and hospitals to compare more than 441 000 people with diagnosed atopic eczema with more than 1 8 million people without diagnosed atopic eczema, to examine whether diagnosed atopic eczema was related to obesity. This study leverages clinical data to identify patients with atopic eczema using a validated algorithm that includes both diagnostic coding and presence of atopic eczema therapies recorded in the medical record, which is an improvement in methodology over patient-reported self-diagnosis of atopic eczema. While Ascott et al. found a small link between atopic eczema and obesity, severe eczema was not associated with obesity. The authors note that clinicians and patients can feel reassured that these conditions do not appear to be multiplicative risk factors for heart disease. This is an important finding; however, the next steps should consider how to use these data to guide clinical practice for patients who have both atopic eczema and obesity. Using EHR data to examine prevalence of possibly linked conditions is a good way to perform epidemiological studies because it uses data collected during routine medical practice to gain insight into clinical outcomes and practice without placing additional burden on the clinician or patients, who are the real stakeholders. Ascott et al. performed a number of sensitivity analyses, such as restricting cohort entry to include cases with the most complete data, and excluding certain weight classes (e.g. underweight) in separate models, which appeared robust and did not change the conclusions of the paper. Sensitivity analyses are important to check for differences that can bias the study when using EHR data, which can be messy as these data were not collected primarily for research. The authors acknowledge limitations common to EHR data, such as bias owing to missing body mass index measurements, but still provide a well-conducted study that is likely generalizable to patients in the UK, in addition to patients in other countries with similar healthcare systems. It is important to examine these associations at a population level in order to properly maintain guidelines for clinical practice, and epidemiological studies such as this one can contribute to the body of evidence needed to do so.