Risk of new‐onset atrial fibrillation—does blood glucose matter?

Atrial fibrillation (AF) is the most common cardiac rhythm disorder and is considered a 21st-century cardiovascular disease epidemic. The lifetime risk varies between one in three and one in five [1, 2], and the Global Burden of Disease 2019 study found a global prevalence of more than 59 million individuals in 2019 [3]. The prevalence may rise to 15.9 million in 2050 in America and 17.9 million in 2060 in Europe [4, 5]. Patients with incident AF lose on average 2 years of expected lifetime compared with AF-free individuals from the general population [6]. Addressing modifiable risk factors is essential to facilitate primary prevention, which is crucial to reduce the global burden. Diabetes mellitus is a common chronic condition whose incidence and prevalence are rising. Multiple epidemiological studies suggest an association with the development of AF [7], and diabetes has been considered a modifiable risk factor through several years. In this issue of the Journal of Internal Medicine, Johansson et al. provided novel insights into the association between glycemic status in the general population and risk of AF [8]. The authors examined 88,889 participants from Sweden (mean age 49 years and 48.2% men) who accepted at least one invitation to a health examination between 1988 and 2014. Measures of glycemic status were categorized into normoglycemic, impaired fasting glucose (IFG), impaired glucose tolerance (IGT), previously unknown diabetes, and known diabetes. Participants completed a questionnaire about lifestyle factors, and body mass index (BMI) was calculated from height and weight. Each participant was followed for AF by linkage to the National Patient Register. During follow-up, 4958 participants were diagnosed with AF. The corresponding age- and sex-adjusted hazard ratios were (normoglycemia as reference) IFG: 1.17 (95% confidence interval 1.07–1.28), IGT: 1.05 (0.92–1.21), IFG + IGT: 1.23 (1.03–1.47), previously unknown diabetes: 1.41 (1.23–1.61), and known diabetes: 1.55 (1.31–1.82). Interestingly, when adding BMI to the age- and sex-adjusted model, the strength of the associations attenuated markedly: IFG: 1.10 (95% confidence interval 1.00–1.20), IGT: 0.98 (0.85–1.12), IFG + IGT: 1.05 (0.87–1.26), previously unknown diabetes: 1.17 (1.02–1.35), and known diabetes: 1.30 (1.10–1.54). Next, the authors adjusted additionally for systolic blood pressure, and all associations became non-statistically significant, except known diabetes 1.25 (1.05–1.47). In the final fully adjusted model, no statistical evidence of any association between blood glucose and AF was found. The associations for IFG, IGT, and previously unknown diabetes almost disappeared after adjusting for sex, age, and BMI. Obesity is an essential causal risk factor for the development of both type 2 diabetes mellitus and AF [9]. Obesity is, therefore, a significant confounder of the association between glycemic state and AF (Fig. 1). The results must be interpreted carefully given the observational nature of the study; however, such an attenuation suggests a substantial confounding effect of BMI. Based on their final model, the authors adjusted for several potential confounders and conclude that blood glucose is not causally associated with AF. Johansson et al. are to be commended for conducting a large study on the risk of AF and a continuum of glycemic states from normoglycemic to diabetes. The participants underwent a detailed examination program and the study period extended 26 years. Data on both IFG and OGTT provide insights into prediabetic states, which seem essential to consider when concentrating on primary prevention. Participant characteristics included individual-level data on lifestyle factors and social determinants of health. The study by Johansson et al. had some important limitations. As participation required acceptance of the invitation, self-selection bias is likely; hence, the representation of individuals may be associated with social determinants of health. Additionally, the generalizability of the findings to races and ethnicities other than White European is limited. Finally, the study included no data on glycated hemoglobin A1c, which is often used to diagnose diabetes in routine practice. This study questions the existing knowledge of diabetes as a causal risk factor. Is high blood glucose a noncausal biomarker rather than a causal risk factor for AF? Or could blood glucose mediate the effect of obesity or other cardiovascular risk factors? Nevertheless, the causal associations between chronic conditions, such as obesity and risk of AF, are complex. Many authors frequently use BMI to define obesity, but as the index is a ratio between two single risk factors, BMI may not be an optimal criterion for risk. Accordingly, attention has turned to investigating the effects of the various components of body mass. A high lean body mass (also referred to as fat-free mass) is closely associated with AF, whereas evidence is sparse for an independent influence of the mass of fat tissue on risk of AF [10]. Prevention of overweight and obesity is a key to reduce the burden of at least three major public health challenges—namely, diabetes, hypertension, and AF. Further studies are needed to improve our knowledge of the complex causal interplay among blood glucose, body mass, other risk factors, risk modifying drugs—such as SGLT2 inhibitors and GLP-1 receptor agonists—and AF to facilitate and optimize effective prevention. We encourage future research to concentrate on specific measures of body mass components, such as lean body mass and fat mass (Fig. 1). Furthermore, given that blood glucose mediates the effect of body mass, mediation analysis that quantifies the direct and indirect effects of body mass and its components is warranted. Finally, we will encourage all authors to use directed acyclic graphs to present their underlying assumptions of causal associations, as those will ease the interpretation of the results. Conceptualization; writing—original draft; writing—review and editing: Nicklas Vinter and Lars Frost. NV: Supported by grants from The Danish Cardiovascular Academy (PD2Y-2022002-DCA), The Health Research Fund of Central Denmark Region (R38-A1385-B1044), and The Danish Heart Foundation (16-R107-A3987). Advisory Board and Educational activities for AstraZeneca (no grant or fees were received personally). LF: Supported by the Health Research Fund of Central Denmark Region (Institutional grant). Educational activities for BMS and Pfizer. Advisory Board for AstraZeneca.