Study population

The ongoing prospective population-based Heinz Nixdorf Recall (HNR) cohort study is carried out in three large adjacent German cities (Mülheim, Essen, Bochum) in the Ruhr district. Details of the aims and the design of the study have been published earlier [22]. Briefly, participants were invited to the study center for the first time between 2000 and 2003 (T0), and 4814 persons (mean age 59.6 years, age range between 45 and 75 years) participated in the baseline examinations. The next visits to the study center took place between 2006 and 2008 (T1) with 4157 participants (mean age 64.4 years, age range 50 to 80 years) and between 2011 and 2015 (T2) with 3089 participants (mean age 68.7 years, age range 55 to 86 years). The median time between T0 and T1 was 5.1 years, and 5.2 years between T1 and T2. During the examination in the study center, participants filled in questionnaires, took part in face-to-face interviews and underwent physical examinations including extensive laboratory tests. Additionally, participants received annual postal questionnaires.

The Heinz Nixdorf Recall MultiGenerationStudy (MGS) comprises partners and children of HNR participants in order to investigate cardiovascular risk factor burden and the related incidence of cardiovascular disease within families. Briefly, from 2013 to 2016, 1237 partners and 1660 adult children of HNR participants, aged 18 to 90 years, underwent a similar examination program as participants of the HNR Study at T2, and completed annual postal questionnaires on health status since then. Fig 1 shows the study design of the HNR Study and MGS. We used data from the third examination (T2) of the HNR Study whose participants served as index persons and data from the MGS.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study design of the Heinz Nixdorf Recall (HNR) Study and Heinz Nixdorf Recall MultiGenerationStudy (MGS).

https://doi.org/10.1371/journal.pone.0252828.g001

Of the 1237 partners in MGS, 1181 had an index person who was examined at T2. Thus, analysis population A included 1181 index persons and their partners. Of the 1660 children in MGS, 1607 were biological children, and of those, 1497 had an index person who took part at T2. 735 index persons had one child in MGS, another 289 index persons had two children in MGS, 52 index persons had three children in MGS, and 7 index persons had four children in MGS. Thus, analysis population B included 1083 index persons and 1497 children.

Assessment of sleep characteristics

Sleep characteristics taken into account in the present study were assessed in an identical way for index persons and their partners / children.

Sleep quality was assessed using an item from the Pittsburgh Sleep Quality Index (PSQI): “How would you rate the overall quality of your sleep during the last four weeks?” (very good / good / poor / very poor) [23]. Sleep disorders were assessed using two other items from the PSQI: „During the last four weeks, how often have you had trouble sleeping because you could not fall asleep within 30 minutes?“, and „During the last four weeks, how often have you slept poorly because you woke up in the middle of the night or early in the morning?”(Not at all during the last four weeks / less than once a week / once or twice per week / three times or more per week). Preferences for rising times were assessed from the following item: „If it were only up to your own well-being and you could plan your day completely freely, when would you get up?”(mark only one hour). Snoring was assessed using the item „Do you snore?”(yes / no). An item on napping was „How frequently do you usually nap?”(never / less than once a week / one to four times per week / five to six times per week / daily).

Statistical analyses

Log-binomial regression models were fitted to estimate relative risks (RRs) with 95% confidence intervals (CI) for associations between sleep characteristics of the index persons as the exposure and sleep characteristics of their partners and children, respectively, as the outcome. We also fitted log-binomial models to estimate RRs (95% CI) for the association between snoring of the index person and sleep characteristics of the partner. To analyze associations between sleep characteristics of the index persons and sleep characteristics of their children, another approach was additionally used to take into account that data of siblings may be correlated: We fitted logistic regression models with random effects to estimate odds ratios (ORs) with 95% CI.

For the regression analyses, sleep characteristics were transformed into binary variables as follows: very poor / poor versus very good / good for sleep quality; three times or more per week versus less often for sleep disorders; once a week or more versus less often for napping; yes versus no for snoring. Persons who preferred getting up before 7 o´clock a.m. were defined as early risers, and those who preferred getting up at 9 o´clock a.m. or later as late risers.

We did several sensitivity analyses:

First, in the analysis of study population A, couples who reported not to sleep in the same room were excluded when the sleep characteristics were snoring, sleep quality or sleep disorders. Second, analyses on the associations between sleep characteristics of index persons and their children were stratified by the median of parental age. Third, although our aim was to analyze associations and not causal inference, we added some adjusted analyses on the association between sleep characteristics of index persons and their children (adjustment for age and sex; adjustment for age, sex, BMI, smoking, chronic diseases and drug intake).

All statistical analyses were performed using SAS version 9.4.

Comparison with earlier studies

In the present study, poor sleep was more frequent in the adult children than in their parents. This is not consistent with earlier results showing that sleep latency increases with age while sleep maintenance and deep sleep decreases [24–26]. However, the offspring belong to another age cohort, and little is known about trends in the prevalence of sleep problems and insomnia in Germany [27]. In addition, employment status may have an impact on sleep quality, and the adult children of the MGS were mostly employed in contrast to their parents.

In two clinical studies, familial aggregation of insomnia was found [18, 19]. In one study which lacked a control group, 35% of insomnia patients reported that at least one family member had insomnia also; in the other study, familial insomnia was reported most often by patients with primary insomnia, less often by patients with psychiatric insomnia, and least often by control persons without insomnia (73%, 43%, and 24%, respectively). Contrary to the present study, the relatives were not interviewed in these clinical studies. This also applies to a population-based cross-sectional study which showed that persons who ever suffered from insomnia had an increased chance of having relatives with insomnia compared to persons without a history of insomnia (OR = 1.57, 95% CI: 1.19–2.07) [20]. In a further study with a reconstructed cohort design, biological relatives of index persons had a higher risk of insomnia when index persons were affected with insomnia but results were not unambiguous for parents and children (RR = 1.80 (p = 0.04) for all biological relatives, 4.96 (p = 0.08) for siblings, 0.84 (p = 0.71) for parents, 1.65 (p = 0.38) for offspring) [21]. In the same study, partners of index persons were also taken into account, and the corresponding relative risk was increased but not estimated precisely (RR = 2.13, p = 0.28). In line with the present study, Gunn et al. found that into-bed times were positively correlated within couples [28].

Contrary to earlier studies, we also did age-stratified analyses on the associations between sleep characteristics of index persons and their children. For sleep quality, napping and snoring, these associations were stronger for parents younger than the median parental age of 69 years than for older parents. This is plausible because these associations may partly result from shared life styles or shared environments and because such factors may match less the longer parents and children live apart.

Our finding that partners of snorers are more often affected from sleep disorders and poor sleep quality is in accordance with earlier results from a study in Sweden, which showed that women of heavy snorers more often suffer from insomnia and daytime sleepiness than women whose partners do not snore [29]. In a recent study on snorers, it was reported that sound levels of snorers exceeded 45 dB(A) in two thirds of the participants so that the authors considered snoring as a “noise pollution in the bedroom” [30]. Thus, there is evidence that snoring bedpartners are a risk factor for poor sleep.

We did not adjust for potential confounders because we were interested in associations and not in causal inferences. For associational questions, a mere comparison is sufficient [31]. For example, we wanted to know whether partners of index persons were more likely to have sleep disorders if the index person was affected. We did not want to investigate whether difficulties falling asleep in the index persons cause this sleep disorder in the partners. To answer the latter question, potential confounders like the noise level in the bedroom or shared problems which cause stress in both partners would have to be taken into account. Additionally, we adjusted the analysis on sleep characteristics of parents and children for age, sex, BMI, smoking, chronic diseases and drug intake. This adjustment had only little impact on the strength of the associations. This seems plausible as the proportion of individuals taking drugs or having chronic diseases was much lower in the adult children than in their parents. Thus, the observed associations do not result from confounding paths where, e.g., drug intake of the index person causes sleep problems in the index person on the one hand, and causes drug intake in the children on the other hand which in turn causes sleep problems in the child.

Public health relevance

Poor sleep is a risk factor for adverse health events [1–12]. Our study confirms that adult children are more likely to suffer from sleep disorders and poor sleep quality when their parents are affected. As past studies have shown familial aggregation also in other sleep related diseases like obstructive sleep apnea or restless leg syndrome [32, 33], physicians should generally give consideration to first-degree relatives of patients who are affected with sleep disorders.

Snoring is strongly associated with obstructive sleep apnea which has a high incidence in the older population [34]. Our findings on the effects of snoring on bed partners’ sleep is a second reason why snoring should be given consideration. An earlier study suggested that bed partners of snorers may benefit from treating snorers: continuous positive airways pressure (CPAP) was used to eliminate snoring, and bed partner’s sleep quality improved strongly [35].

Limitations and strengths

In this study, sleep characteristics were assessed by self-report. This is not a limitation with regard to preferences for rising times or frequency of napping. However, in an earlier study, agreement between self-reports and partners’ reports of snoring was rather poor [36], and, thus, self-report of snoring may lead to misclassification. Furthermore, we only assessed sleep disorders during the last four weeks so that we could not distinguish between early and late onset sleep problems. Earlier research suggested that familial aggregation was stronger for early than for late onset of insomnia [17]. Our study also has strengths. It is population-based with fairly large study groups. Contrary to some other studies, sleep characteristics of offspring and partners were directly assessed. Thus, reporting bias could be avoided because in earlier studies there was an indication that persons with insomnia had a tendency to overreport insomnia in relatives [21].

Conclusions

There are only few studies on familial aggregation of sleep characteristics. We found aggregation of sleep disorders and poor sleep quality in parents and their offspring, but not in couples. Moreover, we showed that bed partners of snorers are more often affected from poor sleep and sleep disorders.