Commentary

In the current issue of EPIDEMIOLOGY, Danaei and colleagues1 elegantly estimated both the direct effect and the indirect effect—that is, the effect mediated by blood pressure, cholesterol, glucose, fibrinogen, and high-sensitivity C-reactive protein—of body mass index (BMI) on the risk of coronary heart disease (CHD). They analyzed data from 9 cohort studies including 58,322 patients and 9459 CHD events, with baseline measurements between 1954 and 2001.1 Using sophisticated and cutting-edge methods for direct and indirect effect estimations,2,3 the authors estimated that half of the risk of overweight and obesity would be mediated by blood pressure, cholesterol, and glucose. Few additional percentage points of the risk would be mediated by fibrinogen and hs-CRP. How should we understand these estimates? Can we say that if obese persons reduce their body weight and reach a normal body weight, their excess risk of CHD would be reduced by half through an improvement in these mediators and by half through the reduction in BMI itself? Is that also true if these individuals are prevented from becoming obese in the first place? Can we also conclude that if these mediators are well controlled in obese individuals through other means than a body weight reduction, their excess risk of CHD would be reduced by half? Let us confront these estimates with observations from studies evaluating 2 interventions to reduce body weight, that is, bariatric surgery in patients with severe obesity4–6 and intensive lifestyle intervention in overweight patients with diabetes.7,8 Bariatric surgery is an effective method of reducing body weight in obese patients. In the controlled Swedish Obese Subjects study, the mean changes in body weight 2 and 20 years after surgery were −23% and −18% in patients having undergone surgery (n = 2,010) and 0% and −1% in patients having usual care (n = 2,037), respectively.4 Surgery also has favorable effects on blood pressure, blood glucose, and blood lipids, and in a recent systematic review, it was shown to be associated with a 50% reduction in the risk of cardiovascular diseases, including myocardial infarction and stroke, compared with obese patients not undergoing surgery.6 This effect of surgery on BMI and related outcomes is, at least qualitatively, coherent with the findings of Danaei et al.1 Other observations are more puzzling. Compared with surgery, lifestyle interventions are less efficient at reducing body weight.5,9 Nevertheless, it is commonly recommended in overweight patients with diabetes. The Look AHEAD randomized controlled trial evaluated the effect of intensive lifestyle intervention among overweight patients with diabetes.7,8 Patients in the intervention group had a sustained weight loss (weight was −6.0% vs. −3.5% in the control group at study end) and improvements in fitness, glycemic control, and blood pressure.7 Surprisingly, after 9.6 years of follow-up, there was no difference between the groups in the rate of cardiovascular events, leading to the early interruption of the trial.7,8 These observations are difficult to reconcile with the findings by Danaei et al.1 Indeed, either through a direct effect of the reduction in body weight or through an indirect effect due to the improvements in the mediators blood pressure and blood glycemia, a decrease in cardiovascular diseases incidence should have been observed. It is also worth recalling that several observational studies not restricted to patients with diabetes have shown that intentional weight loss was associated with an increased mortality, predominantly due to cardiovascular diseases.10 Furthermore, in a recent large systematic review, overweight persons had lower all-cause mortality (largely due to cardiovascular diseases) compared with those of normal weight,11 raising a vivid debate on what would be the optimal BMI.12 The results of Danaei et al1 are correct in theory, and their analytical methods to compute direct and indirect effects are surely fine.2,3,13 However, what is missing is that the authors did not specify what they meant by a causal effect of BMI on CHD events. They should have considered the issue of consistency violation in the effect of exposures such as BMI in observational studies.14,15 BMI is indeed probably the result of a complex interplay of various interventions and determinants. Therefore, the effect of BMI in an observational study reflects the effect of a combination of interventions and determinants that are specific to the observed population. This specificity implies that the effect estimate cannot be simply transposed to other populations (eg, the issue of nontransportability of causal inference with compound treatments).15 The fact that this combination differs from one study population to the other is probably one reason for the heterogeneity in the effect estimate of obesity on mortality. Furthermore, BMI is not directly manipulable as is smoking (you can quit smoking) or blood pressure (you can decrease blood pressure by a drug). Although it is possible to modify BMI, there is no intervention that can set BMI to a given level without having in itself an effect on the outcome, whatever the level of BMI reached. For instance, if people try to lose weight by starting smoking, they will not have the same CHD risk as if they increase physical activity, even if the resulting BMI is identical.10,14 The relative contribution of direct and indirect effects may change from one intervention to the other, as they have different effects not only on outcomes but also on mediators. Without a specification of the method implemented to lose weight or to prevent weight gain, estimating direct and indirect effect of BMI is a theoretical exercise of vague causal effect estimations. Consistency violation is a plague for observational studies used to assess the effect (total, direct, or indirect) of BMI on any health outcomes.14 This is also true for other exposures such as socioeconomic status, for which the causal effect is too often vaguely defined. To go beyond vague causal effect estimation, studies assessing the effect of specific, well-defined, and feasible (preventive or therapeutic) interventions aiming to modify BMI are necessary. We would better understand how different interventions can have the same effect on BMI but different total, direct, and indirect effect on health outcomes of interest. This approach would also result in a relatively simple causal interpretation of the effect estimates, that is, what is the causal (total, direct, and indirect) effect of this intervention to change BMI on the outcome of interest. Finally, these effect estimates are directly actionable, with a true relevance for policy makers, physicians, and patients. ABOUT THE AUTHOR ARNAUD CHIOLERO is an Epidemiologist trained at the Lausanne University Hospital, Switzerland, and at the Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Canada. He is a Senior Lecturer at the Institute of Social and Preventive Medicine in Lausanne (www.iumsp.ch), where he leads research on the epidemiology of cardiovascular diseases, and Chief Physician at the Observatoire valaisan de la santé (www.ovs.ch), Sion, Switzerland, where he oversees public health surveillance activities.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.