Adherence to a Healthy Sleep Pattern and Incident Heart Failure: A Prospective Study of 408,802 UK Biobank Participants

Heart failure affects more than 26 million people worldwide and is increasing alarmingly.¹ Emerging evidence has implicated various sleep problems in the development of heart failure.² Notably, sleep behaviors are multifaceted and individual sleep behaviors usually correlate with each other in a compensatory fashion.^3,4 However, the majority of previous studies did not consider the complex nature of the overall sleep pattern. We recently generated and validated a healthy sleep pattern that incorporated five different sleep behaviors, which was associated with a lower risk of coronary heart disease in two independent cohorts.⁵ In the current study, we prospectively investigated the relationship between this healthy sleep pattern and the risk of heart failure.

The UK Biobank is a prospective cohort of 502,505 participants aged 37-73 at the time of recruitment (2006-2010). We excluded participants with heart failure (N=2,418) and those with missing information on the sleep behaviors (N=91,810) at baseline, leaving a total of 408,802 participants for the current analysis. All participants provided written informed consent. The study was approved by the NHS National Research Ethics Service and the Institutional Review Board of Tulane University. Data sharing: requests to access the dataset from qualified researchers trained in human subject confidentiality protocols may be sent to UK Biobank (https://www.ukbiobank.ac.uk/researchers/).

Sleep behaviors were collected through the touchscreen questionnaire. Healthy sleep score was created by combining chronotype, sleep duration, insomnia, snoring, and excessive daytime sleepiness.⁵ Sleep duration was categorized into 3 groups: short (<7h/d), recommended (7-8 h/d), and prolonged (≥9 h/d). Healthy sleep factors were defined as: early chronotype (‘morning’ or ‘morning than evening’); sleep 7-8 h/d; reported never/rarely or sometimes insomnia symptoms; no snoring; and no excessive daytime sleepiness (‘never/rarely’ or ‘sometimes’). Each sleep factor was coded 1 if meeting the healthy criterion and 0 if not. The healthy sleep score was obtained by summing up the five individual sleep factors. A higher score indicated a healthier sleep pattern.

Incidence of heart failure was collected until April 1^st, 2019, and defined based on International Classification Diseases edition 10th (ICD 10: I11.0, I13.0, I13.2, I50.X), hospital procedure codes, and self-reported information confirmed by nurses.

Among the 408,802 study participants, a total of 5,221 cases of heart failure were recorded during a median follow-up of 10.1 years. The association between the healthy sleep score and heart failure incidence was estimated by Fine-Gray subdistribution hazard models, accounting for competing risk of mortality.

The healthy sleep score was inversely associated with the incidence of heart failure across all the three models. The HR (95% CI) for heart failure was 0.85 (0.83, 0.87) for 1 point increase in the healthy sleep score (Table, model 2). The association did not appreciably change with additional adjustment for prevalent diabetes, hypertension, medication use, and genetic variations (Table, model 3, HR (95% CI): 0.87 (0.84, 0.90)). Compared to participants with a healthy sleep score of 0-1, the HR (95% CI) for heart failure was 0.58 (0.49, 0.70) for those with a healthy sleep score of 5 in model 3. The restricted cubic spline analysis showed a linear relationship between the healthy sleep score and incidence of heart failure (data not shown). When analyzing each binary (low risk vs. high risk) component of the healthy sleep pattern and the risk of heart failure, early chronotype, sleeping 7-8 h/d, no frequent insomnia, and no frequent daytime sleepiness was each independently associated with 8%, 12%, 17%, and 34% lower risk of heart failure, respectively, in the multivariable-adjusted model (Table). The sensitivity analysis excluding patients who developed heart failure within 2 years of follow-up yielded similar results, providing evidence against reverse causation (data not shown).

Table.

Hazard ratios of heart failure by healthy sleep score and its individual component among 408,802 UK Biobank participants.

Healthy sleep	Model 1		Model 2		Model 3
	HR (95% CI)	p-value	HR (95% CI)	p-value	HR (95% CI)	p-value
Healthy sleep score	0.81 (0.78, 0.83)	<.001	0.85 (0.83, 0.87)	<.001	0.87 (0.84, 0.90)	<.001
0-1	ref	-	ref	-	ref	-
2	0.72 (0.61, 0.85)	<.001	0.77 (0.65, 0.92)	0.003	0.85 (0.71, 1.01)	0.07
3	0.56 (0.48, 0.65)	<.001	0.63 (0.54, 0.74)	<.001	0.72 (0.61, 0.84)	<.001
4	0.45 (0.38, 0.53)	<.001	0.53 (0.46, 0.63)	<.001	0.62 (0.53, 0.73)	<.001
5	0.40 (0.34, 0.47)	<.001	0.49 (0.41, 0.58)	<.001	0.58 (0.49, 0.70)	<.001
Individual component*
Early chronotype	0.90 (0.85, 0.97)	0.002	0.92 (0.84, 0.99)	0.03	0.92 (0.84, 0.99)	0.04
Sleep 7-8 h/day	0.76 (0.71, 0.81)	<.001	0.86 (0.79, 0.94)	<.001	0.88 (0.81, 0.96)	0.002
No frequent insomnia	0.77 (0.72, 0.82)	<.001	0.80 (0.74, 0.87)	<.001	0.83 (0.76, 0.91)	<.001
No self-report snoring	0.92 (0.86, 0.98)	0.009	0.91 (0.84, 0.98)	0.02	0.97 (0.89, 1.05)	0.44
No frequent daytime sleepiness	0.56 (0.49, 0.64)	<.001	0.62 (0.52, 0.73)	<.001	0.66 (0.55, 0.78)	<.001

Model 1: adjusted for age, ethnicity, and sex;

Model 2: model 1+ Townsend deprivation index, center, alcohol intake, smoking status, physical activity, sedentary hour, healthy diet score, average household income, education;

Model 3: Model 2+ diabetes (y/n), hypertension (y/n), medications for anti-hypertension, cholesterol-lowering, and insulin, first 10 genetic principle components, and genetic risk score of heart failure.

^*Each individual component was modeled as binary variable: met or unmet the healthy criterion. All the five individual components were included in the model simultaneously.

Our study is the first prospective investigation on the relation between adherence to an overall healthy sleep pattern and the risk of heart failure. Sleep behaviors are inter-correlated and the human body regulates sleep in a holistic way to maintain an overall constancy of sleep intensity, quality, and duration. Thus, our study extends the prior findings on individual sleep behaviors by jointly evaluating multiple sleep behaviors. In addition, the healthy sleep pattern included not only the sleep behaviors during nighttime, but also the chronotype and daytime sleepiness. Several potential limitations should be acknowledged. First, sleep behaviors were self-reported, thus, recall bias and misclassifications were inevitable. Second, the information on changes in sleeping behaviors during the follow-up was not available. Third, the observational nature of the study precludes causal inference. Fourth, although we have carefully controlled potential confounders in the analysis, unmeasured or unknown residual confounding might be possible.

In summary, our results provide new evidence to indicate that adherence to the healthy sleep pattern is associated with a lower risk of heart failure, independent of the conventional risk factors. Our findings highlight the potential importance of overall sleep behaviors in the prevention of heart failure.