Examining the relationships between individual’s time use and activity participations with their health indicators

In the last decade, there has been a surge in the studies that investigate the positive health effects from the use of physically active travel modes such as walking and cycling. These active travel modes have gained attentions from both transport/urban planners and public health experts because they are considered as of low-emissions, space efficient, and can positively contribute to the travellers’ physical health [1–4]. Hallal et al. [5] estimated that today, globally, 30% of adults were lack of physical activities. Together with an energy-dense diet, insufficient physical activities would lead to obesity epidemic, a condition that is associated with cardiovascular diseases, type 2 diabetes, cancer and impaired mental health [6–9].

Based on this believe, many governments are promoting cycling and walking as an alternative to reduce the health risk that comes out from sedimentary lifestyle (e.g. [10, 11]). The physical health impacts from walking and cycling have now started to be taken into account in various governments’ cost-benefit analysis processes. Various methods were proposed to measure and compare the health impacts of active travel. Typical health impacts models that were used to assess the impact of these physically active transport activities are based on hypothetical scenarios with mostly linear correlation assumption on changes in the amount of active travel and the health gains (e.g. [12, 13]). Systematic reviews on the impacts of physically active travel modes to the travellers’ health conditions can be seen at Mueller et al. [9] and Wanner et al. [14], for example. Most of previous studies agree to each other that higher participations in physically active travel activities, such as walking and cycling, correlate with a better physical health condition in general. The expected impacts, however, vary across different socio-demographic groups, and some are contradicting with each other. For example, some studies such as Vogel et al. [15] highlighted the benefit of physically active travel for older people since the activity is expected to substantially reduce the absolute risk for chronic degenerative disease. Some studies (e.g. [15–22]) also argued that the risk would outweigh the detriments of traffic incidents and air pollution exposure. However, Mueller et al. [9] also pointed out that the argument that older people benefit differently from the same physically active exposure than younger people remains inconclusive. Woodcock et al. [17, 23] argue that older people would benefit a larger mortality risk reduction from physically active participation compared to younger people, whereas Götschi et al. [4] explained that the impacts of physically active travel is also a function of the current conditions of the travellers. 30 min cycling would have relatively smaller impacts to the travellers who have been cycling everyday compared to the ones who hardly cycle at all in their live. Mueller et al. [9] concluded that the benefits of physically active travel participation “are sensitive to the contextual setting and population parameters. Health impact estimations depend on baseline prevalence of active travel participation, baseline exposure to health pathways and the general status of the population.”

Moving to the context of the analysis, most previous studies measured the health effects of active travel based on one day (or “average day”) observation of a given individual’s travel or aggregate, cross-section, based observations. Observing a physically active travel on the given day for a given individual does not necessarily mean he/she has a physically active travel routine on a day-to-day basis (thus difficult to measure his/her baseline exposure to a healthy lifestyle). Furthermore, many studies put a lot of emphasise on a particular indicator of active travel, i.e. frequency or amount of walking and cycling, as individual’s main travel mode on their main trip leg. However, in reality, most travellers are multimodal travellers and, in many cases, the last-mile and short trips modes have a higher frequency and intensity than a regular, medium-long distance, routine trips. Furthermore, since individual needs and desires are not constant from day-to-day, an individual’s activity-travel pattern is neither totally repetitious nor random every day [24–29]. The repetitiveness of mode choices highly correlates with the accessibilities of the activity locations, whereas different types of activity have different patterns of repetition. Moreover, most of the previous studies tend to overlook the benefits and/or dis-benefits that the individuals gain from and the trade-off between their daily in-home and out-of-home activities. This can lead to an over or under estimated benefits of physical active travel to individual’s well-being and also to their physical health conditions, e.g. trade-off between an active travel and an active training in a gym. Therefore, it is important to quantify the benefits that an individual gains from both physically activity travel and their day-to-day in-home and out-of-home activities in order to better understand the real value of physically active travel behaviours; not only to physical health, but also to individual’s social and mental health.

To contribute to these research gaps, using a 3-weeks travel diary which was collected in the Bandung Metropolitan Area, Indonesia, this study investigates the relationships between travellers’ physically active travel participation with the type and intensity of their daily activities and their self-assessed physical, mental and social health conditions. This study not only investigates the impacts of day-to-day variability of different types of in-home and out-of-home activities to individual’s health related aspects, but also reveals such behaviours, correlations, and trends in the context of developing country, i.e. Indonesia, which, to authors’ knowledge, has never been studied in the past.

In the next section, the study area is described and so as the used datasets. It is followed by a section which presents the descriptive analyses on individual’s different time use allocation on different day towards different type of in-home and out-of-home activities. Section 4 describes the estimation results on the influence of one’s activity participations with ones’ self-reported physical, mental, social health conditions. Conclusions will be presented in Section 5.

To understand the trade-off between in-home and out-of-home activities on a daily basis, the distributions of time use across different physically, mentally, and socially intense activities, both in-home and out-of-home activities, across different days of the week, were further examined. It can be seen from the distribution of the variables (not shown in this paper) that individuals have longer out-of-home physical activity durations on weekdays, especially those of medium intensity, than on weekends. On contrary, the duration of out-of-home mentally medium intense activities is longer on weekends than on weekdays; but it is the other way around for out-of-home mentally high intense activities. However, there is no substantial difference between weekdays and weekends in terms of out-of-home social activity participation. It is also as expected that the individuals spend more time on in-home low intense activities (physically, mentally and socially) on weekends than on weekdays. How these time use allocations vary across different main travel mode users and across individuals with different levels of self-assessed healthiness can be seen at Figs. 2 and 3, respectively. The classification of activities and assumed level of intensities that were used in this analysis are shown in Table 3.¹ Whilst the level of intensity of each physical, mental, and social activity arguably can be higher or lower as shown in Table 3, the current assumptions of activity intensity were categorised based on criteria on typical local’s activity engagements that is assumed as described on the lower part of Table 3.

	Type of Activities	Level of intensity of engagement
		Physically	Mentally	Socially
A	Sleeping	L1	L1	L1
B	Personal care: taking a bath, brushing teeth, self-care etc.	L1	L1	L1
C	Eating and drinking at house	L1	L1	M1
D	Relaxing activities, such as watching tv, listening to radio, listening to music, reading newspaper/magazine/comic etc., browsing internet etc.	L1	L1	L1
E	Social and family activities, such as chatting with other family members/friends in person or via phone, walking/biking with other family members/friends, visiting relatives/friends, weekly praying etc.	M2	M2	H2
F	Household activities, such as house cleaning, cooking, baking a cake, washing clothes/dishes, ironing, prepare a drink, etc.	M1	M1	L1
G	Babysitting activities, including playing together and feeding your children	H1	H1	M1
H	Indoor working activities, such as working at office desk, doing indoor research or experiment in laboratory, “meeting” with clients at phone, etc.	M2	H2	M2
I	Driving vehicle to other places	L2	H2	L2
J	Outdoor working activities, such as operating machine or heavy vehicle at outdoor environment, outdoor inspection and other related activities	H2	H2	M2
K	Sales activities from door to door, delivery and purchasing related activities	H2	H2	H2
L	Indoor school activities	M2	M2	M2
M	Outdoor school activities, including visiting zoo/museum/park, camping, and other related activities	H2	H2	H2
N	Eating and drinking outside home	L2	M2	H2
O	Shopping activities, including both local grocery shopping and shopping at a shopping centre	M2	M2	M2
P	Participating in organization/volunteer/political activities, such as boy scout and youth/political/religious meeting activities	H2	H2	H2
R	Sport activities	H2	H2	H2
S	Maintenance activities, including going to hospital/health centre/medical doctor, bank and post office	M2	H2	H2
T	Pick up and drop off children/other family members/friends/business partner and others	M2	M2	H2
U	Holiday (away trip)	M2	M2	M2
V	Waiting for public transport	L2	L2	L2
Level of intensity	Physical	Mental	Social
Low	not require any specific physical engagement	not require any in-purpose thinking	not involve any specific intension to interact with others
Medium	activities with some physical engagements	need to be aware/alert of surrounding, but not in intense manner	involve activity that somewhat require interactions with others
High	require intense physical engagements	need always to be alert (on-guard) and/or at intensive thinking	involve intense interaction activities with others

Note: 1 = in-home activities, 2 = out-of-home activities; with assumption of level of intensity:

In terms of differences among travellers’ main mode groups (see Fig. 2), those who walk and cycle most in their daily travel spend longer time on in-home low and medium intense activities (around 1.5 h physically, 2 h mentally and 2 h socially), such as sleeping, eating, drinking at home and relaxing activities than those who use private motorised modes and public transport. This is because they are likely to be retirees or housewives. On the other hand, the ones who use private motorised modes and public transport spend more time on high intense physical activities than the non-motorised travellers (motorised travellers: 0.42 h; public transport travellers: 0.55 h). These activities include sports, participating in organization/volunteer/political activities and outdoor working/school activities. This indicates trade-off effects between the uses of physically intense travel modes with high intense activity participation at destination locations. Overall, slow mode users have a shorter duration of out-of-home medium and high intense activities (physically and mentally) than private motorised modes and public transport users, though there is no such clear difference in time allocation for different types of social activities.

Figure 3, however, shows that the relationship between individuals’ self-reported health conditions and their activity time use seems more confounding. Those who reported good physical health, the last quantile group (75%–100%), seem to spend more time on in-home (0.5 h) and out-of-home (0.06 h) physically high intense activities than those who reported bad physical health conditions, the first quantile group (0%–25%). Similarly, those who reported good mental health also have higher time use in out-of-home mentally high intense activities than those who reported bad mental health (0.71 h), however, they have a lower time use in out-of-home mentally medium intense activities (−0.57 h). From reported social health perspectives, those who reported better social health have longer time use in out-of-home socially low and medium intense activities (1.2 h) and shorter time use in in-home socially low and medium intense activities (1.07 h). However, some confounding relationships may be due to the differences in individual socio-demographics and residential environment. To be able to measure the impacts of different intensities of physical, mental, and social activities to the individual reported health conditions, after all other variables being controlled, a path analysis is employed in the next section.

In order to investigate the complex relationships between in-home and out-of-home participation to the individual’s self-reported physical, mental and social health conditions, a path analysis is adopted. The path model that is adopted in this study is structured as in Fig. 4 below.

In this model structure, the individual/household social demographic variables and residential environment variables are treated as exogenous and they influence the time use variables. The time use groups (physical/mental/social activities) are believed to affect individuals’ health conditions (physical/mental/social health factor scores) as well as travel time. At the same time, activity duration will also impact the travel time as travel is a derived demand of activity participation. It is worth noting that the model structure takes into account the endogeneity effect. And therefore, the direct/indirect/total effects are reported instead of the estimated coefficients (weights represented as arrows in Fig. 4). Most previous studies examined the relationship between non-motorised travel time and health (the arrows from travel time to self-reported health scores), while this study takes one step forward by considering activity durations as mediation variables as well as the trade-off effect between activity and travel time use. Therefore, the confounding health effects between an intensive activity time use and an active travel could be revealed through the direct and indirect effects of travel time and activity time use (activity duration) on the self-reported health scores.

As shown in Table 4, individuals’ activity participation significantly affects their self-reported health conditions, while travel time does not show any significant effects on the self-reported health conditions after controlling for the endogeneity. For self-reported physical health condition, out-of-home physically low intense activity duration positively correlates with self-reported physical health condition due to its strong DE, while the IE (activity duration influences travel time thus indirectly influences self-reported health conditions) is negative but insignificant. However, out-of-home physically medium intense activity duration shows a negative correlations self-reported physical health condition. For self-reported mental health, both in-home and out-of-home mentally high intense activities seem to be positively correlated with self-reported mental health conditions. Out-of-home mentally low intense activities show a significant DE (−0.368) but an insignificant TE. Longer out-of-home mentally low intense activities are associated with longer public transport travel time but shorter motorised travel time which indirectly improves the self-reported mental health condition. For self-reported social health, out-of-home medium intense activities/high intense activities show a positive/negative DE but both show insignificant TE. These findings again reveal that activity participations are strongly related with individuals’ self-reported health conditions but some effects are offset by the time use trade-offs between activities and travel. However, one should be aware that those findings need to be interpreted with caution. As intensity level and duration may have a non-linear effect and reference dependency on health condition [4]. For instance, increasing duration of out-of-home physically high intense activity when the given individual already has an intensive physical activity would be less beneficial, or even physically harmful, compared to the same situation for the individual who has not exercised for a long time. In other words, how long the duration of physically high/low intense activity is considered good for physical health (reference dependent) should depend on the individual characteristics as well as their habit, past activity participation, etc.

In terms of travel time spent on the given, non-motorised travel time, however, shows no significant correlation on self-reported health conditions, which differs from evidences from developed countries (e.g. [23, 50, 51]) who found a positive correlation between using slow modes and physical health condition. This finding may be attributed to the fact that activity time use explains most of the variations in the self-reported health conditions and travel time does not show any significant effects due to its close relationship with activity time use. In other words, activity time use should be more suitable predicting the self-reported health conditions than travel time.

In terms of socio-demographics and residential environment characteristics, older respondents (age > 55), as expected, have a lower self-reported physical and social health condition than the reference age group (age 36–55), after all other variables being controlled. Older respondents in general may be more aware of social needs, thus tend to have a lower self-reported social health even when they spend the same amount of time on social activities as adults. Young people (age < 25) shows a higher self-reported physical health condition than the reference age group (age 36–55). Those who live in large households also tend to have higher self-reported social health compared to those living in small households. Similarly, part-time workers and non-workers show lower self-reported physical health conditions compared to workers. Surprisingly, income shows no significant correlation with self-reported physical health condition, although part of income effect may be absorbed in the effect of motorised travel time. Population density positively correlates to self-reported physical health condition, indicating a potential health benefit of urban area densification in developing countries, e.g. better accessibility to hospital. Being a part-time worker would have a lower self-reported mental health condition compared to a full-time worker. Harsh work environment, unstable work time, etc. may be the underlying reason of this effect. Population density also positively correlates to all three self-reported health conditions. This indicates the potential benefits of urbanization on individuals’ health conditions in general.

As can be seen from Table 4, after controlling activity time use, travel time, socio-demographics and residential environment, the remaining error terms still show considerable correlations. The three self-reported health conditions are all positively correlated with each other. Therefore, health conditions in these three dimensions should not be treated independently but positively interrelated. Other factors such as well-being, happiness, etc., which are not captured by the existing model may appear in the error terms and contribute to these positive correlations.

Using a multiday household activity-travel survey (2013 BMA dataset), this paper explored the relationship between individuals’ self-reported health conditions and their activity-travel time use. The self-reported health conditions are hypothesized to be affected by individuals’ activity participation pattern and usual travel mode use, described by the observed activity-travel time use in 21 consecutive days. General descriptive analysis shows clear day-to-day variability of individuals’ activity-travel time use. On average, individuals only spent around 4–5% of their time on travel. In-home activities such as sleeping, in-home preparation and eating activities take more than 75% of individuals’ daily time expenditure, highlighting the importance of understanding the physical activity intensity that is caused by in-home activities as well. The self-reported health conditions show, however, more confounding relationships. The trends of in-home and out-of-home activity time use may be affected by the fact that self-reported health groups consist individuals with different socio-demographics and residential environment.

The model results reveal that individuals’ activity participation and socio-demographic characteristics significantly affects their self-reported health conditions, while self-reported travel time spent does not show any significant effects on the self-reported health conditions after controlling for the endogeneity (some effects are offset by the time use trade-offs between activities and travel). Surprisingly, income shows no significant correlation with self-reported physical health condition, although part of income effect may be absorbed in the effect of motorised travel time. This study, however, did not find evidence of positive relationship between cycling and walking and self-reported physical health condition, which has been found in many developed countries. Presumably because in developing countries like Indonesia the individuals who walk and cycle are likely to be a part of economically disadvantaged and less educated groups who may have low appreciation and less awareness to oneself condition [52, 53].

Beside activity and travel time use factors, age and working status were found significantly affect the self-reported health conditions, regardless of respondents’ gender and income. Population density also found positively correlates to self-reported respondents’ health conditions, indicating benefits such as better health care in urbanised areas on individuals’ health conditions in general.

However, one should also be aware of several limitations of this study. The self-reported health condition is likely to be as well related to one’s well-beings, past habit, etc. and not a detailed measured medical report. Self-reported scores are reference-dependent, meaning that individuals with almost the same actual (true) health condition may judge their own health conditions very differently. Twenty-one consecutive days activity-travel participation may be enough to represent their current habit but may not describe their past habit. Thus, the self-reported health condition may also differ significantly from the true health condition due to individual’s cognitive limits, that individual do not realize their health condition, and behavioural failure, that individuals themselves do know the negative effect of certain habit (e.g. smoking) on their health but do not admit it. The mismatch between self-reported health and true health may affect the estimated variable effects. It is also important that this study reveals the correlations between activity participations and individual self-reported health conditions, not necessarily the causal relationships. Thus, further interpretation need to be done with great care. Further investigation on how individual self-reported health conditions and physically active travel and activities influenced by individual attitudes and psychological characteristics would be one of plausible research directions of this study.