The Association With Cardiovascular Disease and Type 2 Diabetes in Adults With Atopic Dermatitis

A Systematic Review and Meta-analysis

J.P. Thyssen; A.-S. Halling-Overgaard; Y.M.F. Andersen; G. Gislason; L. Skov; A. Egeberg


The British Journal of Dermatology. 2018;178(6):1272-1279. 

In This Article


In this systematic review and meta–analysis, we found no association between AD and hypertension, presumed type 2 diabetes, myocardial infarction and stroke, in the quantitative crude data analyses. Similar analyses using fully adjusted data from the articles gave essentially identical results. Overall, quantitative analysis indicated that AD was associated with angina pectoris based on only two publications and four cohorts.

Most studies included in the quantitative meta–analyses used questionnaire data to define AD. These studies are particularly vulnerable to misclassification as responders may confuse AD with other skin conditions such as psoriasis, lupus and nonatopic eczema (e.g. allergic and irritant contact dermatitis). Moreover, these studies used questions that were somewhat nonspecific for AD (e.g. 'has a doctor ever told you that you had eczema?' and 'during the past 12 months, have you had dermatitis, eczema or any other red inflamed skin rash?').[20] Importantly, the U.K. Working Party diagnostic criteria for AD may not reliably identify individuals with a history of AD when used in adults.[22] As AD often resolves in adolescence, most adults classified as having AD no longer have active disease. Therefore, systemic inflammation may no longer be present. Based on these observations, future studies that examine the association of AD with CVD should be conducted in patient populations with active AD, as seen when using registry data for current AD diagnoses.[4,9]

Systemic low–grade inflammation in AD has been put forward as an explanation for the increased risk of CVD in some studies,[23] although we could not confirm an association in this systematic review and meta–analysis. Indeed, increased serum levels of cytokines have been measured in patients with AD,[24] and elevated serum total IgE levels, which are observed in many patients with AD, have been linked to CVD.[25] Yet, the levels of specific IgE, particularly in relation to house dust mites, were inversely associated with CVD in a U.S. study,[26] whereas a recent Danish general population–based study refuted these observations.[27] Moreover, the T helper cell (Th)2 bias in AD and other atopic disorders, which is mutually suppressive to the Th1 bias observed in arteriosclerotic CVD, would further support the argument against a higher risk of CVD in adults with AD. There appears to be increased blood platelet activation in patients with AD compared with reference individuals, especially in patients with concomitant allergic rhinitis.[28] A very recent study showed an increased occurrence of cardiovascular risk proteins in serum.[29] Finally, one mouse study showed that persistent interleukin (IL)–1 secretion from the skin was associated with CVD,[30] and in line with this observation, an 'epidermal IL–1 march' was recently described as a potential link between the skin and vascular disease in AD.[31]

The higher risk estimates of CVD observed in selected North American and Asian populations could be related to obesity[32] and AD misclassification as described above. However, in Denmark, where a recent register–based study showed a decreased risk of CVD in patients with AD,[9] an inverse association with the occurrence of gallstones and erectile dysfunction, i.e. conditions traditionally associated with obesity and CVD, has also been observed in the same registers.[33,34] Limited data on physical activity are available, but one study showed that U.S. patients with AD had decreased levels of physical activity, whereas no difference in physical activity was found between Swedish adults with AD and those without AD.[35,36] In particular, the higher risk estimates of myocardial infarction observed in selected North American populations could therefore, at least in part, be due to differences in environmental, behavioural and cardiovascular risk factors. However, it should also be emphasized that the definition of myocardial infarction was based on relatively unspecific questionnaire data in the studies by Silverberg[8] but ICD–10 codes were used in the studies by Standl et al.[6] and Andersen et al.[9] (Table S1; see Supporting Information). In a recent meta–analysis, AD was significantly associated with smoking, thereby suggesting yet another causative factor for CVD.[37] Patients with AD often suffer from poor and interrupted sleep, mostly resulting from itch, but psychiatric disorders such as attention deficit hyperactivity disorder, anxiety and depression may also play a role. A cross–sectional questionnaire–based survey from the U.S.A. showed that adult patients with AD had higher odds of sleep disturbances, which included shorter sleep duration, trouble falling asleep and early morning awakenings.[38] Importantly, there is a positive relationship between sleep deprivation and hypertension, CVD and diabetes in adults, possibly owing to increased activity of the sympathetic nervous system.[39] Topical and systemic medication may also affect the risk of CVD and type 2 diabetes. For example, topical corticosteroid use was associated with type 2 diabetes in a large Dutch register–based study,[40] and a recent Danish study showed that the use of topical corticosteroids was a strong predictor of the occurrence of type 2 diabetes among patients with AD.[4] Moreover, systemic corticosteroids are often used to treat AD flares,[4] and well–established long–term effects include hypertension and type 2 diabetes.[41]

A few remaining causative factors for CVD in AD warrant mentioning. Autoimmune comorbidities co–occurring in patients with AD, e.g. rheumatoid arthritis and inflammatory bowel disease, have been associated with CVD and could add to the overall burden, although the prevalence of these conditions in patients with AD remains low.[42,43] Furthermore, vitamin D insufficiency in patients with AD could affect the risk of hypertension, myocardial infarction and diabetes, for example.[44–46] A large German study showed no evidence for shared genetic risk variants of AD and CVD.[6] Increased expression of IL–17 has been shown in predominately Asian subtypes of AD and in ichthyoses, which, at least in theory, could increase the risk of CVD in Asian patients who might be affected by a more psoriatic form of AD.[3,47–49]

Certain limitations apply to the interpretation of the present results. A limited number of studies were available for meta–analysis and none were from outside Germany, Denmark, U.S.A., Canada or Taiwan. Information about AD severity and smoking were often lacking. Importantly, a causal link cannot be determined based on observational data, and furthermore, most studies were cross–sectional. Misclassification possibly affected the association estimates between AD and presumed type 2 diabetes, as not all studies provided sufficient information about the type of diabetes. While most cases of diabetes are expected to be type 2 diabetes, we cannot exclude the possibility that cases of type 1 diabetes affected the estimates. However, the number of type 1 cases is assumed to be low and would therefore not affect the estimates to a large extent.[50] As AD is an umbrella term that covers different endophenotypes, stratification by specific IgE levels or filaggrin gene mutations might have detected specific risk groups; however, this was not possible.[51] We only searched for articles using the term 'atopic' to increase specificity. However, studies using other common terms such as 'children' or 'flexural eczema' to study AD were potentially missed, which obviously could have biased the outcome. Publication bias may have occurred, as the initial studies found more positive associations than more recent studies. We emphasize that meta–analyses, such as the present one, are not designed to assess whether an association is independent of traditional risk factors. Lastly, although we presented estimates based on both crude and adjusted data, we emphasize that data based on adjusted estimates should be interpreted with caution, as studies may differ considerably with regards to covariates included in the adjusted models. Nonetheless, no significant difference was observed between analyses using crude and adjusted data in any of the analyses.

AD was not associated with myocardial infarction, hypertension, diabetes or stroke. There was a modest association with angina pectoris based on four studies. The current sum of evidence renders it unlikely that AD itself is an independent risk factor for cardiometabolic disease. Differences in estimates are likely explained by study design differences and AD misclassification, variations in body weight, physical activity, use of corticosteroids and other lifestyle features, although the relative contribution of these factors remains unknown. Importantly, as demonstrated in multiple cohorts,[14,17] the magnitude of CVD and associated risk factors is markedly lower in AD compared with psoriasis populations. Future well–designed prospective clinical studies may help to elucidate the relationship between lifestyle and environmental factors, AD disease severity and associated comorbidities.