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 accelerometer data were processed with ActiLife 6.13.3 (Actigraph, Pensacola, USA). Accelerometer data was collected using 1 s epochs and aggregated to 60 s epochs for data reduction. Data periods with consecutive zero counts for ≥60 min, with allowance for 1–2 min of counts between 0 and 100, were considered as non-wear time periods. A minimum of four valid days was needed for a participant to be included in the analyses. Physical activity intensity categories were defined according to the following existing accelerometer cut-points for activity counts per minute (cpm) [5, 27, 30].
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).