Daily sedentary time and physical activity as assessed by accelerometry and their correlates in older adults

Physical activity intensities

Mean accelerometer wear time was 849 min per day, or approximately 14 h per day. On average, participants spent 9.2 h (551 min) of total wear time sedentary (65%), and 4.6 h (277 min) in light (33%) and 14 min in moderate physical activity (2%). Vigorous activity was only performed by 108 participants with a range between 1 and 47 min per day.

The percentage of sedentary time was inversely (linearly) related with both the intensity categories light-high (≥ 760 cpm) and vigorous MVPA (≥2020) expressed as percentage of wear time (Fig. 1). Sedentary time was strongly correlated with the intensity categories light low and light high r = − 0.81 and r = − 0.85, respectively, while sedentary time was only modestly correlated with the moderate intensity physical activity r = − 0.48, P-for all < 0.01.

Sedentary time and physical activity according to age, sex, education and BMI

Correlates of sedentary time and physical activity

The correlates of physical activity (≥ 760 cpm light-high intensity and higher) were mostly in the opposite direction compared to sedentary time. A lower amount of its variance was explained by the demographic, lifestyle and health measures (R² 0.176, 0.203, and 0.258, respectively, P < 0.001, as compared to sedentary time. The correlates of MVPA were in line with light-high physical activity, (Additional file 1: Tables S1–S2).

Sensitivity analysis

Excluding participants with potentially biased data due to lower wear time or a break in wear time did not substantially change the results (Additional file 1: Tables S3–S8).

Strengths and limitations

An important strength of this study is the use of hip-accelerometry in a large population-based sample to objectively characterize minute-by-minute sedentary time and physical activity, and allowing to study different intensity levels (sedentary, light-low, light-high, moderate, and vigorous). This provides detailed data that cannot be gathered by traditional self-reported methods, particularly at low levels of physical activity. Also, accelerometry does not have recall-bias and social desirability bias and does not depend on literacy, which are the major limitations of self-report methods. Another strength is that accelerometer information was collected for 7 days for the majority (90%) of participants, which therefore truly reflects habitual behavior of older adults.

Some limitations have to be acknowledged as well. Accelerometers do not capture all types of physical activity, particular activities with a strong upper body component such as raking leaves or activities without a strong vertical component such as riding a bike (a common activity in the Netherlands) [25]. It is possibly that the intensity of bicycling was underestimated by the accelerometer, but it is not possible to determine if this resulted in underestimation of MVPA, and/or under or overestimation of light PA, and it is not feasible to correct objective physical activity data with self-reported physical activity data. Still the majority of participants (70%) reported bicycling activities in the past two weeks and therefore we included self-reported biking as a possible correlate in our models. Furthermore, water activities such as swimming during their data collection period were not registered as the accelerometer was removed, this led to some underestimation of physical activity with 9% of participants reporting swimming in the previous two weeks. Another limitation is possible selection bias (over- or under-recruitment of certain subgroups) for the cohort study as a whole, as well as for this particular sub-study. So the generalizability of the prevalence results to the general older population might be somewhat limited, while the reported associations are less prone to selection bias. There is also a risk for possible reactivity effects. Although, we cannot rule out that participants modified their activity behavior due to participation in the study, no studies to date have reported such an effects using 7-day accelerometer assessments.

In addition, the cut-off values for the higher intensity categories were developed for a general adult population, not for older adults [5, 27]. Particularly, the cut-point for MVPA may not apply to older populations and the optimal cut-off points may vary for different age-groups, due to dissimilar activity patterns, mechanical efficiency and the contrasting nature of movements at different life stages [28]. It is not unlikely that the used accelerometer cut-points misclassified moderate intensity physical activity as light intensity physical activity. For this reason, we divided the light intensity physical activity into light-low and light-high to better study the activity patterns of older adults. As MVPA levels are very low in older adults, the balance between sedentary time and light intensity activity is increasingly important when it comes to health and daily functioning. Which is another important reason why we looked at light intensity activity in more detail by splitting it into two groups. This increased focus on sedentary time and light intensity physical activity, next to MVPA, might be important to develop more effective guidelines, especially for older adults [15].

This study investigated both correlates of various physical activity intensities and joint combinations of different sedentary time and physical activity profiles across strata of sex, age, education and BMI groups. Sedentary time was strongly related with the intensity categories light-low and light-high, while sedentary time was only modestly related with moderate intensity physical activity meaning that these are activity patterns that are not strongly correlated or the inverse of each other. The high correlation between sedentary time and the two light intensity categories suggests that these are more likely to replace each other in daily routines. Changes in light intensity activities during daily routines as opposed to increase sports or other moderate to vigorous activities could lead to more substantial reductions in sedentary time and might be easier to implement. Nonetheless, the health benefits of moderate to vigorous intensity physical activity should not be overlooked and improving physical activity across the intensity spectrum is recommended.

Study population

We used data from the Longitudinal Aging Study Amsterdam (LASA), which is an ongoing population-based study in the Netherlands that started in 1992 to determine predictors and consequences of aging. A description of the cohort sampling and data collection procedures has been described elsewhere [29]. Briefly, 11 municipality registers from three geographical areas in the Netherlands were used to recruit men and women aged 55–85 years.

Since the start of LASA, two additional cohorts of participants aged 55–65 years were recruited using the same sampling frames exactly 10 (2002–2003) and 20 years (2012–2013) after the initial sample. For the current study, we used data from the 2015–2016 examination with participants of all three cohorts. At each examination, two interviews were conducted: a main interview and a medical interview with clinical measurements about 4–6 weeks apart.

All 1770 LASA participants who participated in the main interview were invited for the accelerometry ancillary study (See flow diagram Fig. 2). A total of 1412 participants indicated to be willing to participate and were sent an accelerometer for a 7-day period of whom 1218 wore and sent back the accelerometer (response 86%). Reasons of non-participations were: acute health problems (n = 151), lost accelerometer (n = 24), measurement error/broken accelerometer (n = 4), too frail (n = 2), too busy (n = 2), or unknown (n = 11).

For the present analysis, we excluded (n = 17) participants who wore the accelerometer < 4 days. Of all participants, 4 occasionally used a wheelchair, 10 used a walking stick and 7 used a walker, but all were able to walk. Altogether 1201 LASA participants had valid accelerometry data and formed the analytic sample. The LASA study is conducted in line with the Declaration of Helsinki, and was approved by the medical ethics committee of the VU University medical center.

Objective measurement of physical activity

An accelerometer is a small device that records the acceleration of the body. The Actigraph tri-axial accelerometer (Model GT3X+; ActiGraph, Pensacola, USA) was used to objectively measure participants’ physical activity intensities. The accelerometer together with an instruction brochure with pictures how to properly wear the accelerometer was sent to the participants by regular mail. The accelerometer was worn around the waist with an elastic belt and placed above the right iliac crest for comparability with most other studies that assessed objective physical activity. Participants were instructed to wear the accelerometer for a consecutive 7-day period during waking hours with the exception of water based activities such as bathing, showering and swimming.

Participants completed a daily log diary to record the time after waking up that the accelerometer was put on and the time the accelerometer was taken off just before going to bed, as well as the times and the reason when the accelerometer was taken off during the day. The log diaries were used to indicate participants with non-wear time due to other activities. Participants were instructed to wear the accelerometer only during waking hours, however some participants forgot to or experienced difficulties taking-off the accelerometer (n = 19). For these participants, wear time and sedentary time was adjusted using the sleep information from the log diary. Adjusted wear time was calculated as: 1440 (total minutes per day) – reported sleep time assuming that a participant wears the accelerometer during all wear hours. The adjusted sedentary time for this small group was calculated as: adjusted wear time – low light – light high – moderate – vigorous physical activity.

The intensity categories moderate and vigorous physical activity were also summed (MVPA). We added the light-low and light-high categories for a better insight in the distribution of the different intensity levels of older participants. In this population, light-high activities might have been of a moderate intensity for some participants.

Other variables

LASA interviewers obtained comprehensive data on participants’ demographics, anthropometrics and co-morbid conditions during the main interview. Body height was measured to the nearest 0.1 cm using a stadiometer. Body weight was measured without clothes and shoes to the nearest 0.1 kg using a calibrated bathroom scale (Seca, model 100, Lameris, Utrecht, the Netherlands). When necessary, corrections were made to adjust measured body weight for clothing (− 2 kg) or corset (− 1 kg). Body mass index (BMI) was calculated by dividing body weight by height squared (kg/m²). We defined BMI categories as: underweight < 70 years < 18.5 kg/m², ≥ 70 years < 20 kg/m², normal: ≥20–25 kg/m², overweight ≥25–30 kg/m², and obese ≥30 kg/m² [31].

The interviewer assessed participants’ education level, smoking status, and living situation. Education was reported on a 9-category scale. We distinguished education into 3 categories: low (elementary school or less), medium (lower vocational or general intermediate education) and high (intermediate vocational education, general secondary school, higher vocational education, college or university). Smoking status was categorized as never, former and current smoker. Living situation was defined as living alone or together with a spouse/partner/family member. Self-rated health was assessed as a measure of overall health status with 4 response categories. We dichotomized this question to poor/fair and good/excellent. Season was calculated based on the first day the accelerometer was worn using meteorological seasons divided into autumn, winter, spring and summer. Level of urbanization was assessed based on the number of addresses per km² (rural, < 1000 addresses/km²; intermediate, 1000–2500 addresses/km²; urban, ≥ 2500 addresses/km²).

The number of chronic diseases was based on self-report of the most frequent somatic chronic diseases in the Netherlands and included: chronic non-specific lung disease, cardiac disease, peripheral artery disease, stroke, type 2 diabetes, arthritis and malignancies. Self-reported functional limitations was measured with a questionnaire adapted from the Organization for Economic Cooperation and Development (OECD) questionnaire and validated by Central Bureau of Statistics Netherlands [29]. Participants were asked whether they have difficulty performing 4 common activities related to mobility: 1) walk up and down a 15-step staircase without resting, 2) sit down and get up from a chair, 3) walk 5 min outside without resting, 4) drive or use public transport. The total score ranged from no limitation to limitations for all functions (stairs/transport/chair/walk) score 0–4 and was categorized into 0 limitations and ≥ 1 limitation. Further, self-reported bicycling and swimming in the past two weeks was measured with the LASA physical activity questionnaire [32].

For the 6 m walk test, participants were asked to walk 3 m, turn 180°, and walk back 3 m as fast as possible while the interviewer recorded the time in seconds. A higher walk test time indicates poorer physical performance.

Statistical analyses

Baseline characteristics are presented as mean and standard deviation for continuous variables or number and percentage for categorical variables. We summarized demographics, lifestyle, health measures, sedentary time and physical activity (total and different intensity categories) by sex and calculated Pearson correlation coefficients among various physical activity intensity categories.

We graphically displayed the continuous relationship for sedentary time and physical activity intensities as percentage of total wear time using cubic splines with 95% confidence intervals adjusting for age, sex, BMI and education level. We used analysis of covariance to estimate adjusted means and 95% confidence intervals to study characteristics of 7-day hip accelerometry by categories of age, sex, education and BMI using the F-test as P-for trend over the adjusted means.

Further, we applied hierarchical regression analysis to assess correlates of sedentary time and physical activity with a P-value of 0.10 as inclusion criterion using complete case analysis. The basic model included age, sex and education as a block of covariates and correlates were added one by one thereafter. In total three blocks were defined: block 1 age, sex, education and season; block 2: adds smoking, BMI, and urbanization categories; block 3 adds walking speed, functional limitations and self-rated health. The results are reported as unstandardized regression coefficients with 95% confidence intervals. In addition, the R² was assessed to estimate the explained variance of sedentary time or a specific physical activity intensity.

Next, we estimated adjusted means across combined sedentary and physical activity categories based on the median using analysis of covariance.

Sensitivity analyses

Additionally, we performed a sensitivity analysis to test the robustness of the associations. Participants who wore the accelerometer at night (n = 19), participants with a mean wear time < 600 min for ≥4 days (n = 136) and participants who reported a significant break in wear time based on self-report form the log diary (n = 144) were excluded.

Statistical analyses were performed with SPSS for Windows (version 22.0).