In a recently published study, Dr Communication BiologyUsing data from large-scale studies including the United Kingdom (UK) Biobank and 23andMe, researchers investigated how spouses influence each other’s sleep patterns and circadian preferences.
Study: Correlations between sleep patterns and circadian preferences in women. Image credit: Gorodenkoff/Shutterstock.com
Insufficient and disrupted sleep, including insomnia and short sleep duration, is widespread in society, affecting more than a quarter of United States (US) adults and leading to decreased productivity, occupational accidents, and increased risk of cardiovascular and metabolic diseases, depression, and certain cancers.
Sleep patterns, influenced by age, gender, and lifestyle, show interconnections among couples, potentially affecting family health and providing avenues for targeted interventions.
More research is needed because understanding the interdependence of couples’ sleep patterns is critical to addressing a wide range of sleep problems with important health and social implications, including increased risk of accidents, reduced productivity, and various health conditions.
About the study
The UK Biobank genetic data includes the genotypes of 488,377 individuals using two different arrays. The current study focused on 463,827 individuals of recent European descent, excluding non-European ancestry based on genetic analysis.
At baseline, participants reported their household composition, identifying whether they lived with a spouse, someone else, or alone. Husband-wife pairs were identified using detailed criteria including shared family characteristics and genetic unrelatedness, resulting in a final sample of 47,549 pairs.
At baseline, participants completed a touchscreen questionnaire covering a variety of topics, including sleep. The questionnaire included questions on chronotype, ease of awakening, insomnia symptoms, sleep duration, and snoring, with responses categorized for analysis.
Additionally, 103,711 individuals wore a triaxial accelerometer device several years after baseline, providing detailed sleep data. This data was processed to measure sleep quality, quantity and timing, focusing on Least Active Time, number of sleep episodes, sleep duration and efficiency. Data with problems in recording or calibration were excluded to ensure accuracy.
The UK Biobank study of participants aged 40–70 collected detailed information including age, gender and place of birth, excluding sex mismatches or chromosomal anomalies. It factors in assessment location and season of accelerometer wear, incorporating genetic factors as covariates.
In contrast, the 23andMe dataset consists of customers of a personal genomics company, which focuses on European ancestry to reduce confounding and identify female pairs through genetic analysis.
Both studies surveyed sleep characteristics such as chronotype and insomnia. UK Biobank used categorical responses, while 23andMe used binary variables.
UK Biobank also employed Mendelian randomisation (MR), using genetic risk scores to investigate how an individual’s sleep characteristics affect their spouse. This includes adjusting for confounders and conducting sensitivity analyzes to address horizontal pleiotropy and winner’s curse, ensuring study viability and validity.
Results of the study
The UK Biobank study comprehensively analyzed sleep characteristics in 47,549 spouses. Of these, 47,420 pairs provided self-reported sleep data via a baseline questionnaire and 3,454 pairs had valid accelerometer data, collected between 2.8 and 8.7 years after the initial study. This data allows detailed evaluation of various sleep metrics.
The mean age of female and male spouses at study offset was 56.8 and 58.5 years, respectively. Both groups reported similar sleep durations, with their chronotype preferences varying slightly.
Males were more inclined towards no preference or an evening preference, whereas females showed a stronger morning preference. Women also reported more insomnia symptoms and difficulty waking up, while men were more often reported to be snoring by their partners.
Interestingly, spouses who participated in the accelerometer assessment were older on average than those who did not. They also exhibit healthier lifestyle choices, such as lower smoking rates and alcohol consumption. This cohort reflects a subset of the wider UK biobank population, with significant differences in employment and education levels.
Among UK Biobank participants with genetic data, those living with a spouse were less likely to have extreme evening preferences or difficulty waking up, and experienced less frequent insomnia. However, snoring was more commonly reported, possibly influenced by the nature of the question about snoring.
The 23andMe dataset included slightly older husband-wife pairs than the UK biobank cohort. Similar to UK Biobank findings, sleep duration was consistent between genders, but prevalence of insomnia and snoring differed.
A key finding from both datasets was the correlation of sleep characteristics between spouses. While there was a weak positive correlation for sleep duration and daily activity, an inverse correlation was observed for chronotype. These correlations were generally smaller than those for other sociodemographic and lifestyle factors.
UK Biobank’s MR analysis indicates that one spouse’s sleep duration and activity level may influence the other, with one’s chronotype potentially predisposing them to opposite their partner. Research has revealed complex interplay between various sleep characteristics in couples, underscoring complex genetic and behavioral factors in spousal sleep patterns.
In the UK Biobank study, correlations between genetic risk scores (GRS) for spousal sleep characteristics showed limited evidence of genotypic correlation.
These correlations obtained from single nucleotide polymorphisms (SNPs) associated with sleep characteristics ranged between -0.007 and 0.010. Even when different p-value thresholds from genome-wide association studies (GWAS) were applied, the correlation was subtle, with insomnia showing only a weak, consistent association.
The study also explored whether different factors might modify the effects observed between spouses’ sleep characteristics. The analysis takes into account factors such as age as a proxy for relationship length, location of birth and employment status for potential population composition effects, and lifestyle aspects such as family composition.
Notably, sleep duration effects appeared stronger in older age groups, and activity time effects were more pronounced among childless couples.
Regarding horizontal pleiotropy, which refers to a genetic variant affecting multiple traits, there is little evidence of this phenomenon influencing outcomes. Tests such as Sargan test and MR-Egar intercept test supported this conclusion.
Additionally, analyzes accounting for pleiotropy yield directions of effects consistent with larger confidence intervals, indicating dynamism versus horizontal pleiotropy. However, the presence of weak genetic mechanisms cautions against overinterpreting these results.
Finally, the study addressed potential bias from the winner’s curse, where overlapping GWAS and spouse samples may overestimate SNP effects.
Genetic risk scores based on replicated SNPs are consistent with the results of the original analysis, further confirming the reliability of the results.