Regarding the Effect of Physical Activity on Mortality

Sabia et al.¹ provided a valuable insight into understanding that “it is important to consider both intensity and type of physical activity when examining associations with mortality.”¹ The authors reported that participants with at least 1 hour of moderate physical activity had a one third lower risk of death.¹ Although these findings were in accordance with the current literature regarding the impact of physical activity on all-cause mortality and longevity,² we think that the methodology used by Sabia et al.¹ raises some concerns.

Sabia et al.¹ drew data from the questionnaires of phases 5 and 7 of the Whitehall II Study. Section 2 of the questionnaire (q.2.18 and q.22 in phases 5 and 7 questionnaires, respectively) assessed how the participant's health limited activities ranging from mild to vigorous.³ It could be suggested that the participants with poor health had limited physical activity already, and the observed increased risk of mortality in this group was affected by their health condition, not the intensity or duration of their physical activity. The authors mentioned that they accounted for health measures as a possible confounder, and they concluded that moderate physical activity in late midlife was independently associated with a reduced risk of mortality.¹ However, they only accounted for heart disease, stroke, diabetes, and self-rated fair or poor health. We recognized that the choices of the variables were within the limitation of the collected data, but other health measures such as body mass index (BMI), lung function, blood pressure, and serum cholesterol level were presumably available through medical examination in phases 5 and 7 of the Whitehall II Study.³ Also, some additional data regarding participants’ health conditions could have been extracted from q.2.10 (phase 5, asking whether the participant has ever been diagnosed as having cancer), q.32.a and q.2.1a (phases 5 and 7, asking about long-lasting illnesses or medical conditions in the past 12 months),³ or employment status, which is a major risk factor for morbidity and consequent mortality.⁴

Furthermore, in another study by Sabia et al.,⁵ the authors concluded that multiple pathways link lung function to mortality. It would be important to know the rationale behind the exclusion of lung function and BMI from their multivariate Cox proportional hazard model (model 4). Shortness of breath, a major symptom of common respiratory diseases, including chronic obstructive pulmonary disease (COPD), relates to patients wanting to take it easy and reduce their daily physical activities. Therefore, participants with COPD might have reported low levels of physical activity, whereas at the same time having a significantly increased risk of death because of COPD, which is the third and sixth leading cause of death in high-income and middle-income countries, respectively.⁶ In addition, excess body weight during midlife, including overweight, is known to be associated with an increased risk of death.⁷ Therefore, BMI is an important potential confounder excluded from the model. Finally, the authors’ time frame of analysis did not seem long enough to confidently address questions of mortality in this cohort of mostly healthy, richer men.

We welcome the author's comments on the aforementioned points to better understand the rationale regarding variables selection and methodology.