1 Introduction

Individuals’ physiological, cognitive, and social functioning changes with advancing age, and that is reflected in various lifestyle adjustments. There are multiple factors operating at all levels, from cellular to social, that affect individuals’ daily behaviours. Changes in the human brain (Dickstein et al. 2007) may lower the ability for cognitive processing of information (Cerella et al. 1980; Eckert 2011). Shifts in circadian rhythms readjust the ‘internal clock’ and affect sleep patterns (de Feijter et al. 2020; Yoon et al. 2003). Shrinking social networks may lead to fewer social interactions (Cornwell 2011; Marcum 2013), and increasing functional limitations may increase the degree to which passive leisure replaces other activities (Cho et al. 2018).

These and other changes are reflected in the duration and diversity of behaviours, and how they are intertwined in the sequence of daily tasks. Studies using time-use data report major similarities in how ageing affects individuals’ time allocation in western societies (Gauthier and Smeeding 2003), and provide information on the duration of different activities and its relationship with individual’s health and wellbeing (Steptoe and Fancourt 2019). However, little attention has been paid to how ageing may change the dynamics of entire 24 h-long time-use sequences including how complex they are.

Analysing the complexity of individuals’ behaviours or life trajectories is a relatively novel idea, which has so far been applied, among other areas, to research on life course and career progression, (Boissonneault 2021; Struffolino and Raitano 2020). It gives an overview of entire trajectories as opposed to focusing on isolated events (Jackle and Kerby 2018). The approach is relevant to this study as we expect that it is not just particular behaviours, but also the dynamics of daily lives that change as individuals age.

Ageing is likely to affect the complexity of individuals’ time-use sequences. First of all, the pool of possible activities shrinks with age, among other factors, due to increasing physical limitations (Cho et al. 2018). Secondly, the number of activities and transitions between them may also decline as the ability to effectively switch between tasks generally declines with age (Wasylyshyn et al. 2011). Consequently, the process of ageing may be accompanied by a progressive simplification of daily routines. However, some individuals may experience faster loss of complexity in their daily lives than others. This may have important implications for individuals’ overall functioning. Higher levels of complexity of behaviours and surrounding environment have been associated with multiple positive outcomes for older adults, including greater wellbeing and better cognitive functioning (Andel et al. 2007; 2001). Conversely, low complexity can lead to a faster decline in intellectual functioning (Schooler 1984).

We look at the complexity of daily behaviours using sequences from individuals’ time-use diaries. These show what activities a person has engaged in during a 24-h period. Typically, they start from getting up in the morning and personal hygiene, and end with lying down for the night to sleep. Such sequences may be highly complex—that is, they have many different elements (activities) and feature multiple transitions between them—or quite simple—that is, they have a few unique activities that are stretched over a long time. The main objective of the present study is to explore the association between the complexity of everyday time-use sequences and individuals’ age and other key sociodemographic characteristics. We run the analyses using nationally representative data from four European countries.

Thus far, the assumption about the association between sequence complexity and age in later life has not been empirically tested on large samples. It is not known when sequence complexity may start declining and if the steepness of this decline differs depending on the particular country context, or how complexity in later life is related to individuals’ other sociodemographic characteristics. The present study addresses these questions. We also discuss the meaning and possible implications of complexity for older adults’ overall functioning and quality of life [QoL].

2 Background

2.1 Daily activities in late adulthood

Many older adults attach great importance to keeping themselves busy in order to feel useful and avoid despondency (Gabriel and Bowling 2004). There is substantial research evidence showing that filling one’s life with various meaningful activities helps sustain the optimal use of time in later life, improves wellbeing, and promotes healthy ageing (Steptoe and Fancourt 2019). In the course of ageing, older adults may endorse different activities to maintain a positive outlook on life (Jopp et al. 2008). Regardless of their nature, involvement in different ‘personal projects’ adds meaning to life and makes it feel worthwhile despite the many challenges of later life (Lawton et al. 2002).

Despite the many voices pointing to the importance of engaging in a variety of activities and maintaining a ‘rich’ life in late adulthood, thus far there have been limited empirical efforts to look at older adults’ daily experiences, other than from a narrow, typically bio-medical perspective. A very common approach has been to use the Instrumental Activities of Daily Living scales (IADL or ADL; Lawton and Brody 1969). IADL evaluates whether an individual is able to perform everyday instrumental activities such as preparing food, self-care or taking medication. While being a very useful tool, in particular for detecting early signs of functional decline (Graf 2008), it has multiple limitations for analysing psychosocial wellbeing. Even with regard to physical functioning, some studies report significant differences in this measure only at the highest stages of functional impairment (Na and Streim 2017). IADL does not capture the more fine-grained changes in older individuals’ lives. Meanwhile, most older adults at some point need to give up on certain activities due to increasing functional or other limitations (Cho et al. 2018). Many are still able to fill up their days, keep themselves busy, and remain involved with life through participation in other activities. Others do not have the capacity or resources to make relevant lifestyle adjustments, or their pool of activities may already be small to begin with, or may be shrinking at a faster pace. This means that, although some age-related decline in complexity of time-use sequences may be expected, older adults will differ in how complex their time-use sequences are.

2.2 The meaning of complexity and why it matters

The concept of complexity has been used to investigate human ageing from the social science perspective (Bengtson et al. 2008), but it has rarely been applied in empirical studies on the topic. In the theory of complexity developed by Kohn and Schooler (1978, 1983), which initially focused on the complexity of work, complex tasks were defined as those which provide a greater number of stimuli, require a greater number of decisions, and demand that more factors be taken into consideration. Related research has established that there exists a positive reciprocal association between involvement in complex tasks or exposure to complex environments and individuals’ cognitive functioning (Caplan and Schooler 2006; Schooler and Mulatu 2001), physical health (Kyriazis 2003), and wellbeing (Kohn et al. 2000). Further studies have shown that, in fact, the complexity of any activity, including housework or leisure, is positively related to individuals’ intellectual flexibility and wellbeing (Kohn et al. 2000). The positive association between complex activities and intellectual functioning is reciprocal and has been established for different social groups, including older adults (Schooler and Mulatu 2001; Andel et al. 2007). Conversely, exposure to simple, unchallenging environments has been shown to result in a decline in cognitive skills (Schooler 1984; Schooler and Mulatu 2001; Schooler et al. 1999). These findings are consistent with biomedical evidence which suggests that exposure to complex environments and maintaining a diversity of stimuli may guard against cognitive and physical decline (Kyriazis 2003). Thus far, it has not been investigated whether complex daily activity schedules would produce a similar effect but certainly a greater variety of activities potentially increases the number of stimuli an individual is exposed to in their daily life. More complex time-use patterns are therefore assumed to provide, on average, greater stimulation.

2.3 Determinants of complexity in later life

While exposure to diverse tasks and environments seems beneficial overall, many older adults face increasing challenges regarding what they are able to do on a daily basis. Loss of diversity of activities may primarily be driven by increasing functional limitations which lead to the replacing of a variety of activities with passive leisure (Cho et al. 2018). For example, having a disability, the likelihood of which increases with age, also affects individuals’ time-use patterns by constraining their activity choices (Pentland and McColl 2002; Pentland et al. 1998).

Individuals’ socioeconomic status is also likely to affect the complexity of their daily lives. Having sufficient income may facilitate some activities for older adults with functional limitations, as it allows for in-home modifications which increase indoor activity options (Li 2006). Furthermore, higher income and advantaged social position have been associated with a more ‘voracious’ leisure consumption (Katz-Gerro and Sullivan 2010), which denotes participation in a variety of leisure activities. Similar findings have also been reported for better-educated working-age individuals who have been found to exhibit more complex leisure behaviours (Jarosz 2016).

Individuals’ time-use patterns are also related to their household composition and close relationships. Marital status affects behaviours and the division of tasks within the family (Habib et al. 2006; Vernon 2010). Simply living with someone else may provide potential partners for social interactions as well as increasing the diversity of those interactions, especially if non-family members, such as carers, also live in the household (Ayalon 2009). Conversely, a lack of partners for positive interaction may greatly impact not only individuals’ time-use patterns but also their cognitive functioning (Pillemer and Holtzer 2016). Finally, area of residence may also affect how complex daily sequences are, as older individuals living in rural areas typically have poorer access to transportation, fewer activity options, and a higher risk of social isolation (Goins et al. 2005).

Maintaining complex sequences of daily activities also involves cognitive resources: time-planning, coordination between activities, self-initiative, and use of prospective memory, all of which rely on central executive function. Age-related impairment in executive functioning has been shown to negatively affect individuals’ time-related behaviours, including time monitoring and time-based prospective memory (d’Ydewalle et al. 2001), shifting, updating, and inhibition (Fisk and Sharp 2004). This results in slower performance and an inability to multitask, as well as problems with starting, organizing, planning or completing tasks (d’Ydewalle et al. 2001). Experimental studies have demonstrated that sequences of activities completed by older respondents have fewer transitions because it takes more cognitive effort for them to switch between tasks (Butler and Weywadt 2013; Löckenhoff et al. 2020; Terry and Sliwinski 2012; Wasylyshyn et al. 2011). The decline in ability to effectively switch between activities has primarily been attributed to cognitive ageing, and specifically, to the impairment of central executive functioning (Wasylyshyn et al. 2011). The implication of these findings for real-life processes is that the frequency of switching between activities in time-use sequences of older individuals may also show progressive decline. It is noteworthy that better-educated individuals typically experience slower cognitive decline in the course of normal ageing (Ardila et al. 2000), with educational attainment being often used as a proxy for cognitive reserve in later life (Meng and D’Arcy 2012). It is therefore possible that highly-educated individuals may be better able to maintain complex time-use sequences in late adulthood.

Overall, lower complexity of everyday time-use patterns may both be caused by existing functional, cognitive, and social limitations, and be a factor that further exacerbates those issues. In other words, sequence complexity could possibly be an indicator of existing issues as well as a predictor of further decline. However, to date there are no studies that would establish an association between complexity and age, complexity and older adults’ functional status, or complexity and older adults’ educational attainment. This study aims to fill these gaps.

2.4 Country context

Complexity of time-use sequences among older adults is likely associated not only with individuals’ characteristics but also with the broader sociocultural context that determines older adults’ QoL and lifestyle practices. Regarding the countries in our sample, studies find that living in Finland, controlling for a set of sociodemographic variables, as well as for health and various types of impairment, was associated with significantly higher QoL among middle-aged and older adults compared to Poland (Raggi et al. 2016).

Concerning particular activities, a comparative time-use study using data collected between 1987 and 1992 for a set of countries including the UK, Germany and Finland, reported little overall difference between these countries in terms of the changes in the proportions of different activities across age categories. However, Finns stood out in terms of their substantially longer time spent on passive leisure, which included listening to music and relaxing, and excluded watching TV. The authors attributed this to possible cultural differences (Gauthier and Smeeding 2003).

Countries in our sample also differ with regard to elderly care models. In countries with high levels of familialism (Leitner 2003), which include Poland, care for older adults is provided by family members. Conversely, in countries with low levels of familialism, the prime examples of which are in the Nordic region, elderly citizens are cared for by the state. The implication of these differences is such that older adults in Poland are likely to be more reliant on their family members, whereas elsewhere public services may step in to provide assistance as long as older adults are able to live independently. This may affect their lifestyle, QoL, and complexity of everyday activities.

Overall, though sequence complexity is expected to be inversely associated with age, as well as with individuals’ other sociodemographic characteristics such as education, it may also reflect living conditions and differences in older adults’ lifestyles and QoL between countries. We therefore expect to find significant country differences in the overall levels of complexity but, notwithstanding these differences, we also expect to detect commonalities between countries with regard to the effects of our independent variables on complexity.

3 Methods

3.1 Data

This study has used the most recent nationally-representative time-use surveys from four European countries: the 2012–2013 German Time Use Survey (GTUS) collected by Destatis; the 2013–2014 Polish Time-Use Survey (PTUS) collected by Statistics Poland; the 2009–2010 Finish Time-Use Survey (FTUS) collected by Statistics Finland; and the 2014–2015 United Kingdom Time-Use Survey (UKTUS) collected by the Office for National Statistics. Time-use surveys provide reliable and accurate data on an individual’s activities over 24 h (Kan and Pudney 2008). All surveys interviewed independently living individuals aged 10 or 15 and above with no upper age limit. Institutionalized individuals were not interviewed. Each respondent typically completed two diary days, including one weekday and one weekend day. Each diary day was divided into 144 ten-minute time slots and respondents were requested to write down an activity they carried out within each slot. This study has used the diaries of individuals aged 65 and above. The total sample size of such diary days in the pooled dataset was 25,495, of which the majority, that is 16,324, were from PTUS, 4,500 were from GTUS, 3,357 were from UKTUS, and 1,314 were from FTUS.

All of the surveys were carried out within the Harmonized European Time-Use Survey (HETUS) framework, so the lists of activities in each country were very similar. Nonetheless, as there were certain differences in the codes, all activities were harmonized post-hoc based on guidelines from the Multinational Time-Use Study (Gauthier et al. 2006), which is the largest database of national time-use surveys that have been harmonized for the purpose of comparative research. The final list of activities comprised the following categories: sleep; self-care; eating; paid work; housework; shopping and using services; watching TV; reading or listening; other leisure and productive leisure at home; other leisure and productive leisure outside of the home; social life; physical activity; learning; using the Internet/computers; help, civic activities and religious activities; care for others; travel; and a separate category for missing records. The number of missing values in each dataset was typically around 1%.

3.2 Key measures

Complexity of a sequence of behaviours is operationalized as proposed by Gabadinho et al. (2011), that is as a function of: (1) sequence entropy, representing the number of unique elements in the sequence—so, in other words, how diverse the activities were—and (2) the number of transitions between different elements—that is, how many times one switches from one activity to another. Sequence complexity is represented by the complexity index (Gabadinho et al. 2010, 2011) computed as follows:

$$ C(x) = \sqrt {\frac{{\ell_{d} (x)}}{\ell (x)}\frac{h(x)}{{h_{\max } }}} $$

where hmax represents the theoretical maximum value of the entropy based on the total size of the activity repertoire. Complexity reaches its maximum value of 1 if a sequence x is such that: i) x contains each of the states from the entire repertoire of possible activities, ii) the same time l(x)/a is spent in each state/activity, and iii) the number of transitions equals l(x) − 1 (Gabadinho et al. 2011). Conversely, the minimum value of complexity, that is 0, is reached when an individual spends the entire period of observation in the same state (a single activity, no transitions). The unique benefit of the complexity index is that it takes into account both the diversity and the sequencing of the elements, the latter through its measure of transitions. In the case of two sequences containing the same elements, e.g. AABB and ABAB, a higher score is given to the latter because an individual changes their state more often, even though the pool of elements is the same. For the computation of the complexity index and its components we used the TraMineR package for R (Gabadinho et al. 2011). The same package was used to visualize the sequences.

3.3 Analytical approach

Descriptive analyses start with illustrating time-use sequences and comparing the share of different harmonized activities over 24 h in the population of older adults in each country. Mean sequence complexity is then computed and plotted by age. Following that, the two components of complexity (within-sequence entropy and the number of transitions) are analysed.

Multivariable OLS regression models using mean complexity from all respondents’ diaries as outcome variable are fitted separately for each country for individuals aged 65 and above. Explanatory variables include: age, gender, educational attainment, marital status, household income, household size, disability status, and area of residence. Age uses three categories: (1) 65–74, representing the young-old, (2) 75–84, representing middle old age, and (3) the older old, aged 85 and above. Education is recoded based on ISCED categorization into the following general categories: (1) primary, (2) secondary, (3) tertiary. Marital status uses the following categories: (1) single, (2) married, (3) widowed, (4) divorced or separated. Household income uses four categories based on the income distribution in the given population: (1) bottom 25%, (2) 25–50%, (3) 50–75%, and (4) top 25%. Household size is a categorical variable denoting the number of individuals living in the household with the last category standing for “four or more”. Disability status is a binary indicating whether a respondent had a disability. Area of residence uses two categories: (1) urban, and (2) rural, with semi-rural areas being coded as rural.

4 Results

4.1 Descriptive analyses

Figure 1a illustrates the data for all of the respondents’ 24-h long time-use sequences stacked on top of one another for each country separately. The x axis represents time; that is the 24-h duration of the diary, which starts at 4 am and lasts until 3:59 am on the next day. Activities are presented on this dimension in the order in which they appeared in the diary. As expected, nearly all sequences start and end with sleep. Overall, older individuals followed quite structured meal patterns, with lunch and dinner taking place at a similar time across the entire sample of older adults, in particular in France and Germany. Meals are visible as bright lines running across the entire pool of stacked sequences. Most leisure activities were located in the second part of the day, with errands, housework and other unpaid work being done earlier, which suggests that older adults, including those who do not work, maintain a meaningful pattern of time organization during the day, resembling that of the working population.

Fig. 1
figure 1figure 1

a Stacked 24-h’ time-use sequences in the population aged 65 and above. b Randomly selected 24-h’ time-use sequences of older adults; Polish sample

Full size image

Figure 1b zooms in on the earlier picture and shows ten randomly selected sequences from Poland. These sequences illustrate the entire 24-h period for ten individuals from our sample. Besides sleep at the beginning and end of each sequence, these individuals engaged in a variety of activities, some of which were carried out in long spells (e.g. TV viewing; reading and listening), while others were carried out in fairly brief but evenly spaced episodes (e.g. eating, self-care). Similar patterns were detected in other countries.

As regards particular activities that filled up older adults’ days, there were substantial similaries in how much time respondents spent in different types of activities (Fig. 2). Sleep accounted for the highest share of the time for all samples, followed by TV viewing, housework and eating. Notable country differences included the following: older Poles spent more time sleeping than individuals from the other countries (which shortened the duration of their ‘active’ waking time); older adults in the UK spent a relatively long time on indoor lesiure (leisure in) and relatively little time on outdoor leisure (leisure out); older Finnish adults spent a longer time reading compared to individuals in other countries. The Finnish respondents also spent a relatively high share of their time exercising compared to older adults in the other samples.

Fig. 2
figure 2

Share of time (within 24 h) spent in different activities, by country, weighted

Full size image

The complexity index had a normal distribution in each country. The mean values of the complexity index (C) and its components, entropy (E) and the number of transitions (T), for each country were as follows: Germany (C = 0.30, E = 0.63, T = 25), Finland (C = 0.32, E = 0.62, T = 27), United Kingdom (C = 0.31, E = 0.61, T = 27), Poland (C = 0.28, E = 0.57, T = 23). Mean sequence complexity varied significantly across age categories, though the general pattern of the association was similar in all countries (Fig. 3). Complexity was usually highest for the youngest individuals in the sample, that is those aged 65–74 (the ‘young-old’), and then declined, with the decline being most pronounced for respondents in their eighth decade of life and older. Finland differed from the other countries in the sample. The sequences of respondents aged 75–84 were just as complex as those of the ‘young-old; the sequences of individuals aged 85 and above did not differ significantly either, but the standard errors were large in this case.

Fig. 3
figure 3

Mean levels of sequence complexity by age group and country, weighted. Notes Error bars represent 95%

Full size image

The mean values of the two components of the complexity index, the within-sequence entropy and the number of transitions, are plotted by respondents’ age category (Fig. 3). Entropy was consistently lower for older age groups, suggesting that the age-related decline in complexity was primarily driven by the decline in the variety of activities in the sequence (Fig. 3). The differences in mean entropy between age categories were not significant in Finland but that was likely due to larger standard errors for the oldest age group. The decline in entropy was the steepest in the UK and Poland (Fig. 4).

Fig. 4
figure 4

Sequence entropy and fragmentation (transitions) by age group and country, weighted. Notes Error bars represent 95%

Full size image

Though sequence entropy was typically the lowest for adults aged 85 and above in all countries, that was not always the case for the number of transitions. There was no uniform trend in how the number of transitions were linked to age, though overall it also seemed to be lower for the oldest age categories. This means that while some older individuals might have reported a relatively larger number of episodes of activities over the day, the variety of these activities generally declined with age. This finding also suggests that decline in overall sequence complexity in later life is likely due to the decreasing variety of activities, rather than to increasing difficulties with switching between tasks.

4.2 Multivariable analyses

The covariates of sequence complexity were analysed using multivariable regression models fitted separately for each country (Table 1). In most countries, age, gender, educational attainment, and household size were significantly associated with the outcome variable. Older age was associated with lower complexity in Germany, Poland, and the UK but not in Finland. Being female and tertiary-educated were associated with having more complex time-use sequences in all countries.

Table 1 Covariates of sequence complexity in the population of older adults, by country
Full size table

Household size was negatively associated with complexity in all countries. While this was assumed to represent possible coupling options implying that a larger household might allow for more activities, the results suggest that the opposite is the case: living with others—controlling for marital status—was associated with having less complex time-use sequences. Having a disability had a moderate negative effect on complexity in Poland, and a very strong negative effect in the UK—but only when household size was excluded from the model (results not shown). Accounting for household size rendered this effect insignificant, suggesting there is a link between living with others and having a disability.

Lastly, living in a rural area was associated with having less complex time-use patterns in Poland but not elsewhere.

5 Discussion

Daily time-use sequences reflect activity arrangements that require planning, organization and coordination between tasks. The present study has shown that sequence complexity generally declines with age, but this decline is not uniform across countries. The most significant decline, especially around the eighth decade of life, was reported for Poland. Conversely, there was no significant decline in Finland. The decline in complexity in all samples was mostly driven by loss of diversity in activities, rather than a lower number of transitions between them. This means that while the older adults sampled might still have been switching between activities quite often, the variety of these activities was lower among older age groups.

In multivariable models, sequence complexity is negatively associated with individuals’ ages, having a disability, and living in a larger household. While the negative effects of age and disability were expected, the negative effect of household size came as a surprise. Earlier findings about an association between living with others and disability risk (Khongboon et al. 2016) help to interpret this result. In this context, living with others in later life may denote a loss of independence. Becoming unable to live alone, an individual either moves in with their family members, or to a care home. Institutionalized respondents were not interviewed in the national time-use surveys. However, dependent individuals living with their families could have been included in the sample as long as they were able to complete the diary. A major question arising from this finding is whether loss of sequence complexity precedes moving in with others, or whether moving in with others causes complexity to decrease. Perhaps it is both. Individuals experiencing a decline in health become increasingly limited in their daily activities. The repertoire of these activities further shrinks when other members of the household take over certain tasks. Importantly, lost activities are not replaced with other tasks. This study does not allow the question of whether or not such a simplified life has the potential to further speed up an individual’s functional or cognitive decline to be answered. More research is needed about the effects of home-based care on older adults’ daily functioning and their QoL.

Complexity is substantially higher among tertiary-educated individuals in all countries. The mechanism through which education may affect later-life behaviours is not known but, as noted earlier, tertiary-educated older adults have been shown to have a greater cognitive reserve (Meng and D’Arcy 2012) and better-educated individuals have been shown to experience slower cognitive decline (Ardila et al. 2000). Education may therefore be associated with sequence complexity in terms of its link with older adults’ cognitive abilities. Alternatively, it may be the case that in late adulthood complexity declines at a similar pace for more and less educated individuals, but this decline starts at different levels. There are studies that evidence more complex time-use patterns among better-educated individuals at earlier life stages (Jarosz 2016). These two explanations are not exclusive and they may work in conjunction.

The positive effect of female gender is consistent across countries. The mechanisms through which gender might be affecting sequence complexity were not identified. However, earlier studies using aggregated measures of time-use have pointed to gender differences with regard to the allocation of time at older age (Gauthier and Smeeding 2003). The differences in levels of complexity may also reflect gender-specific patterns of time-use that appear earlier in life and are preserved into older age.

As regards country differences in the mean levels of complexity across age groups, the possible drivers for these are lifestyle differences or differences in the QoL of older adults. Large disparities in QoL between Poland and Finland have been reported in earlier studies (Raggi et al. 2016). With respect to lifestyle differences, an earlier study looking at particular activities of older adults from a comparative perspective (Gauthier and Smeeding 2003) did not show major differences in the shares of different activities across age groups. However, we find that older Finns spend more time overall being physically active and reading. These activities are important because they may be indicative of leading a lifestyle that could be more beneficial for physical and cognitive functioning. Needless to say, they also constitute additional distinctive activities in the sequences.

It is noteworthy that the countries in our sample differ also in the availability of places in long-term care homes, which relates to different elderly care models. The number of such places is high in Germany and Finland, slightly lower in the UK, and much lower in Poland (Eurostat 2020). Therefore, older adults with major impairments or those who require long-term care are less likely to reside in private homes in Germany, Finland and the UK compared to Poland. This could have implications for the composition of our sample in each country, which could have affected the results.

Overall, despite significant differences in mean complexity levels across countries and some differences in the covariates, we do find certain universal patterns which include gender differences, educational gradient and, usually, an inverse association between complexity and age. We therefore conclude that the results of our study could also be relevant for other developed countries not included in our sample.

5.1 Limitations

There is empirical evidence of a reciprocal association between complexity of activities and cognitive functioning, but this cannot be tested with the available data. It also cannot be established whether the positive effects of being female, and of being tertiary-educated on complexity in older age means that those individuals face a slower decline in complexity, or rather their decline starts from higher complexity levels established earlier in life. While the models have accounted for having a disability for the countries for which this variable was provided, other potentially important variables, such as self-reported health, were unavailable and therefore not included in the models.

This study has also used a simplified representation of time-use sequences as it included only the main activity sequence, and did not account for secondary activity sequences (any additional activity that respondents were doing simultaneously with their main activity). This decision was made due to the fact that secondary sequences had a very high number of missing values (accounting for over 80% of the sequences in the pooled sample); with majority of records missing, complexity could not be computed.

6 Conclusions

We conclude that everyday time-use patterns may indeed become simpler as individuals age, because in most countries, complexity of individuals’ time-use sequences is inversely associated with age. The loss of complexity is mostly due to declining entropy—that is, the diversity of activities. However, there were important social differences, with women and tertiary-educated individuals reporting significantly more complex time-use sequences in all countries. This suggests that men and women differ in how they organize their days in later life and that there might be a life-long positive effect of education on daily functioning. Furthermore, individuals who had a disability, as well as those who lived with others, had significantly lower sequence complexity, net of other variables. Lower complexity of daily time-use patterns could thus be an objective marker of functional or other type of impairment, or of decline in physical or psychosocial functioning.

Complexity of behaviours and environments has been positively associated with individuals’ health, wellbeing and intellectual functioning. Older adults typically want to live rich and engaged lives as long as they are able to. Currently there is a lack of any concise measure of older adults’ overall functioning or QoL other than IADL and related bio-medical instruments. Sequence complexity could be used to monitor individuals’ ageing trajectories and to pick out those groups who live more simplified lives and thus could be at an increased risk of some degree of functional, cognitive or social impairment. Based on the present study, we conclude that lower-educated individuals, men and those who live in households with a higher number of people may be at a higher risk of leading simpler and potentially less stimulating and less engaging lives. We also note that the overall risk may differ by country and that some risk factors may be country-specific, such as lower complexity among rural residents in Poland. Individuals with low levels of sequence complexity could benefit from certain lifestyle modifications initiated, for example, using social prescribing (Drinkwater et al. 2019). The objective of such interventions would be to increase exposure to diverse stimuli through a variety of activities and, ultimately, to make their lives richer. As earlier studies have shown, the ability to fill one’s days with a variety of meaningful activities, even within existing limitations, is essential for helping older adults avoid despondency, improve wellbeing and promote positive engagement in life.