The first main aim of the study is to investigate the association of age with mean values and intraindividual variability (IIV) of sleep parameters in everyday life. It is well-known that age is an important determinant of individual differences in sleep – on average, older adults sleep shorter, more fragmented and less deep than younger adults (Ohayon et al., 2004). Moreover, sleep timing also changes with age with older adults tending to go to sleep earlier and wake earlier than young adults (Evans et al., 2021). According to a recent meta-analysis on actigraphy-assessed sleep, these shifts in sleep timing might even be a more prominent feature of adult age differences than changes in sleep duration and efficiency (Evans et al., 2021).

Yet, sleep does not only vary between individuals (i.e., interindividual differences), but also from night to night within the same person, resulting in IIV. During recent years, this IIV of sleep has been shown to predict health outcomes above and beyond mean values (Slavish et al., 2019; Chaput et al., 2020). Several factors have been associated to IIV of sleep, with age being one prominent factor (Bei et al., 2016).

A recent study using a pooled data set confirmed that sleep duration is more stable from day to day in older adults than in younger adults when measured by actigraphy or sleep diary (Messman et al., 2022). For sleep efficiency, only actigraphy data but not sleep diary data showed significant differences (Messman et al., 2022). Interestingly, those results also indicate an increase rather than a decrease in actigraphy-assessed sleep efficiency in the older age groups (Messman et al., 2022), which is at odds to a recent meta-analysis (Evans et al., 2021). To extend on previous research we will here include several sleep related outcome measures based upon on actigraphy and self-reports. This will allow a more direct comparison on how age is associated to mean and IIV of these sleep measures, as presumptive differences between outcomes and studies may in part be due to heterogeneities between studies. As discussed by Bei et al. (2016) the close association of age and IIV of sleep could in part also be associated to unexamined common causes, such as that age is associated to increased morningness. As part of the analysis strategy we plan therefore to also account for covariates such as, e.g., diurnal type, insomnia, retirement status, anxiety and depression. Moreover, differences in weekend versus weekend sleep may need to be taken into account. Generally, individuals tend to sleep longer on weekends than weekdays, but this difference consistently decreases with increasing adult age (Åkerstedt et al., 2019; Jonasdottir et al., 2021). This would imply that adult age may moderate the effect of weekend vs. weekday on IIV of sleep.

Specifically, we will investigate the following hypotheses, which are based upon previous research results:

As secondary hypotheses, we will test whether weekend vs. weekday moderates the effect of age on the outcome variables (H1.6 – H1.10).

The second main aim of this study is to investigate the effect of day-to-day variations in sleep on cognitive performance in young and older adults. Declines in cognitive performance are a hallmark of sleep loss. Experimental laboratory studies show that both total and partial sleep deprivation cause deterioration across a wide range of cognitive functions (Pilcher and Huffcutt, 1996; Lim and Dinges, 2010; Lowe et al., 2017) whereas simple attention and sustained attention are most dramatically affected (Lim and Dinges, 2010). While older adults usually have slower reaction times than younger adults when sleep is not manipulated, their performance decrements in response to sleep deprivation are less pronounced than in younger adults (Philip et al., 2004; Blatter et al., 2006; Duffy et al., 2009). However, knowledge on how day-to-day changes in sleep are related to cognitive performance is scarce. One recent study found that within-person variations in day-to-day sleep duration had very limited effects on working memory performance in older adults (Lücke et al., 2022). To the best of our knowledge, no study has yet compared young and older adults. Here we will use an established mobile version of the Digit Symbol Substitution Task implemented in the PsyMate^™ mobile application (Verhagen et al., 2019; Daniels et al., 2020), which is considered a measure of processing speed. Performance in the Digit Symbol Substitution tasks is known to be affected by sleep loss (Van Dongen et al., 2003; Lim and Dinges, 2010; Honn et al., 2020).

While the analysis will include both between and within person effects, the hypotheses focus on the within-person effects and the cross-level interaction between adult age and the within-person effect. Building on previous sleep deprivation research, we will test the following hypotheses.

The third main aim of this project is to address the relationship between sleep and positive and negative affect. A wealth of research during recent years has shown that sleep and lack thereof is associated with changes in several aspects of affective functioning (Goldstein and Walker, 2014). A recent meta-analysis concluded that sleep deprivation decreases positive mood, and to a somewhat lesser extent increases negative mood (Tomaso et al., 2021). Yet, the effect of sleep deprivation on mood appears to be diminished in older adults (Schwarz et al., 2019). Longitudinal studies addressing the relationship between day-to-day changes in sleep and affect overall also concur that poor sleep worsens mood, and vice versa (for review Konjarski et al., 2018). Those associations are stronger if both sleep and mood are measured using self-report compared to when objective indicators of sleep are used, possibly suggesting a common method bias.

There is hardly any knowledge whether adult age moderates the relationship between sleep and affect in everyday life. To our knowledge, only three experience sampling studies have so far included older adults, confirming an association between poor sleep and poor affect in older adults (Mccrae et al., 2008; Wrzus et al., 2014; Blaxton et al., 2015). One study suggests differences between middle-aged and older adults (Blaxton et al., 2015), but no comparison to young adults was made. Wrzus et al. (2014) reported that adolescents well-being was better the longer their sleep duration was, while adults and in particular older adults benefited most from sleeping their usual sleep-length. This study used only subjective sleep reports, which are known to be biased by age (Akerstedt et al., 2016). Thus, overall, it remains unclear whether there are notable differences between young and older adults in the association between sleep and affect in everyday life.

A further important aspect that has not been directly addressed is whether day-to-day changes in sleep also impact on dynamic aspects of affect the next day. Affect is not static, but transient and varies over time. Higher variability in affect has been associated to poorer psychological health (Houben et al., 2015).

Cross-sectional analyses show that poorer sleep (Leger et al., 2019) and sleep variability (Ying et al., 2022) are associated with increased variability in positive affect. Aggregate measures of affect reactivity across 8 days have also been linked to lower sleep efficiency (Ong et al., 2013). These findings suggest that poor sleep not only is associated to poorer affect on average, but also impacts on dynamic aspects of affect in everyday life. However, this has so far only been addressed in cross-sectional analyses which do not allow for distinguishing the direction of the relationship. Therefore, it is important to extend the research and focus also on intra-individual associations (Gillett et al., 2021). Here, we will use an intensive longitudinal setup to investigate whether daily changes in sleep are related to changes in both mean levels and variability of positive and negative affect the next day.

We will also investigate whether effects differ between young and older adults, because older adult age is associated to changes of affect in daily life (Carstensen et al., 2011), lower variability in positive affect (Röcke and Brose, 2013) and smaller effects of total sleep deprivation on mood (Schwarz et al., 2019). The results will contribute to a deepened understanding of how sleep is linked to affective well-being.

Specifically, we will investigate the following hypotheses.

This is an intensive longitudinal study with a mixed design. Repeated mobile ESM and actigraphy during 21 days is the within subject factor, adult age group is the between subjective factor.

Individuals in the age ranges 18–30 years (young adults) and 55–75 years (older adults) are eligible for inclusion in the study. We plan to include n = 160 in the young adult group and n = 160 in the older adult group.

Participants need to have a Swedish social security number, be fluent in Swedish, be able to travel to Stockholm University twice for the briefing and debriefing meeting, have no indication of cognitive impairment (Mini Mental State Examination >24) and have a smartphone. The smartphone needs to be compatible with the mobile application used for the ESM. No one but the participant should have regular access to the smartphone. Participants should also have the possibility to wear an actigraph 24/7.

Individuals are not eligible for the study if they lack the ability or confidence to use a smartphone or if their smartphone screen has a functional damage. Individuals whose work requires wakefulness between 2:00 and 5:00 once a month or more often with the exception for on-call work, are also not eligible for the study and neither are individuals who report interest in the study but later on cannot be contacted or who are no longer interested when contacted. Individuals with health conditions and/or medical treatments that likely impact daily physical activity, sleep, psychological well-being or performance ability to a great extent are also not eligible for the study. Examples of medical conditions that may lead to exclusion if present during the past 6 months are sleep disorders and/or a score of >21 on the Insomnia Severity Index (Morin et al., 2011), self-reported depression diagnosis, anxiety diagnosis, exhaustion disorder or any other psychiatric diagnoses, fibromyalgia, autism or current cancer. Examples of chronic diseases that may lead to exclusion are Parkinson’s disease or dementia. Examples of medicaments that warrant exclusion are antidepressants, or prescribed sleep medicine taken >1/week.

Participants are economically compensated for their participation using a staggered payment scheme depending on the number of daily questionnaires they fill in. The maximum amount paid is 2,940 SEK (before tax), which requires the participant to fill in a minimum of 7 ESM questionnaires per day.

This study was approved by the Swedish Ethical Review Authority (2022-04683-02).

Participants will be recruited via advertisements in newspapers, social media, online forums, and paper flyers in the larger Stockholm area. Moreover, participants will be recruited from databases of persons who have previously participated in research projects and/or expressed interest in participation in future projects. Participants will also be invited to pass on information on the study to anyone else they know who may be interested in participating. As shown in Figure 1 the study assessments comprise the online screening and background surveys, the briefing module before the first day of ESM; the ESM module spanning 21 days; and the debriefing module after the end of the ESM module. Data collection started in September 2022.

The planned data analysis is described for each of the main research questions below. For all analyses significance is set at α = 0.05, two-tailed tests.

A recent analysis of a pooled data set found a compliance rate of 78% in ESM studies using 10 daily prompts (Rintala et al., 2019). The assessment period in this study is with 21 days longer than the studies included in this analysis and study length was associated to a slight decrease in compliance (Rintala et al., 2019). In order to ensure compliance, we employ a number of strategies: Firstly, participants are individually briefed in person and called up after about 2–3 days to check in (Palmier-Claus et al., 2011). Participants are also welcome to reach out to the study assistants in case of questions during the data collection. Secondly, we use a staggered compensation scheme depending on the contribution of the participants. According to a meta-analysis, incentives are associated to increased compliance (Vachon et al., 2019). Thirdly, the participants are given access to a visualization of parts of their own data at the end of their respective data collection. Fourthly, we only use a limited number of questions per ESM questionnaire (< 30 questions), as previous research showed that a short length of the questionnaire (30 vs. 60) is more important than the frequency of notifications when it comes to both the experience of the participants and the quality of answers (Eisele et al., 2022).

After the data collection is completed, individual data might be excluded from the analysis due to either the interpretability of the data being compromised (e.g., technical dysfunction of the phone, significant life event) or visualization of data, respectively, residuals suggesting that a datapoint is an outlier / influential observation. Exclusion of data and/or transformation of data will be made transparent in publications. While it is a common practice in many ESM studies to include a minimum required response frequency (e.g., 1/3 of the prompts), these thresholds are arbitrary (Viechtbauer, 2022). Thus, we here do not plan to use thresholds per se, but only in conjunction with sensitivity analyses.

In this study we use a combination of different measurements, namely self-reports, cognitive performance tests and actigraphy with each of them having their respective limitations and strengths. A common criticism of self-report data is that they require participants to be aware of their thoughts, feelings and behaviors and truthfully report them, and that self-report data are by nature subjective. Yet, self-report measures allow insights in the state of the participant that cannot be gained otherwise. Moreover, during recent years self-report measures are increasingly shown to actually be linked and predictive of traditional objective measures. For instance, self-reported hedonic well-being is not only bidirectionally linked with health but may even prospectively predict health outcomes in older age (Steptoe et al., 2015). Likewise, the KSS as measure of subjective sleepiness, that we also use in this study, has been shown to be a sensitive and valid indicator of insufficient sleep and impaired waking function (Akerstedt et al., 2014). Cognitive testing via mobile phone app: While the use of a performance test clearly is a strength of this study, and mobile cognitive testing is a promising area, there are certainly also drawbacks of this method (Moore et al., 2017). Differences between the hardware and software and mobile phones may influence the performance on the cognitive tasks. However, here the primary focus of the analysis is on studying intra-individual changes in performance. Moreover, cognitive testing in daily life situations may suffer from interference from the environment. Actigraphy is generally considered a valid and reliable way to objectively measure sleep and wakefulness, but is still less accurate than the gold-standard polysomnography (Sadeh, 2011). Also, actigraphy cannot be used to deduce any information regarding sleep stages.

While we here apply inclusion and exclusion criteria that are less strict than most sleep deprivation studies which hopefully results in a sample that is more representative to the general population, it still needs to be considered that the included sample may differ from the general population as we do not use random sampling from the population and self-selection effects are in play.

The current project utilizes mobile ESM in combination with objective, actigraphic sleep recording to increase the knowledge of day-to-day sleep variability and the pivotal role of sleep for cognitive performance and affective well-being in young and older adults’ everyday life.