Insomnia, a new Modifiable Risk Factor for Heart Failure?

Mathieu Berger; Geoffroy Solelhac; Frédéric Roche; Raphael Heinzer


Eur Heart J. 2021;42(40):4177-4179. 

In This Article

Strengths and Limitations

The main strength and innovative aspect of this large population-based study is the use of marginal structural discrete-time survival analysis, allowing adjustment for time-varying insomnia symptoms and selection bias, while previous population-based studies only considered symptoms at baseline.[8,13] In addition, an extensive number of covariates was included and time-varying confounders were also accounted for in the statistical models, providing solid evidence for their findings.

However, some limitations have to be emphasized. First, HF diagnosis and most other health conditions in the HRS were ascertained by self-report of a doctor's diagnosis, which may be prone to misclassification. Indeed, HF is a heterogeneous, complex syndrome, and self-reported HF measures in the HRS are yet to be validated. Another important limitation is the lack of information regarding sleep disorders other than insomnia that could contribute to the development of HF. Unfortunately, the HRS did not record information on sleep duration or obstructive and central sleep apnoea (OSA and CSA), two important covariates that may overlap with insomnia symptoms. Of interest, the authors reported a higher E-value for cumulative insomnia symptoms compared with each symptom taken individually, suggesting more unmeasured confounding. Was sleep apnoea among these unmeasured confounders? That is not impossible. It is important to note that non-restorative sleep, a symptom also frequently reported by patients with OSA and/or CSA, was associated with the greatest risk of incident HF. Although early studies estimated that the overlap of insomnia and OSA was quite low (~5%),[14] our clinical experience and results from the HypnoLaus population-based study suggest that this is much more common, with almost one in two men and one in three women having an apnoea–hypopnoea index >15 events/h (indicating moderate to severe sleep apnoea).[15] A recent review even suggested a bi-directional relationship between insomnia and OSA, with ~30–50% of patients with OSA reporting clinically significant insomnia symptoms, and 30–40% of patients with chronic insomnia fulfilling diagnostic criteria for OSA.[16] Therefore, in the study by Mahmood et al., the possibility that patients reporting cumulative insomnia symptoms also had comorbid sleep apnoea cannot be excluded, which might overlap with the association between insomnia and HF (Graphical Abstract). Another limitation is the observational design of the study which shows associations but does not allow conclusions about causality to be drawn. Although this is inherent in cohort study design, further interventional studies are needed to determine potential causal mechanisms.