STRENGTHS AND LIMITATIONS OF THIS STUDY

All participants who wore the accelerometer completed triennial surveys and were linked to state cancer registries and the National Death Index.

Over 91% of participants who were sent an accelerometer returned the device with at least one valid day of wear (ie, ≥10 hours/day), and 87% had at least six valid wear days.

Participants only wore the device at one time point for a maximum of 1 week, which may not represent usual behaviour throughout follow-up.

Data collection spanned 3.5 years before and during the COVID-19 pandemic.

Introduction

Observational studies consistently describe associations between high levels of self-reported moderate-to-vigorous intensity aerobic physical activity (MVPA) with a lower risk of cancer,1 depression,2 cardiovascular disease,3 premature mortality4 and many other adverse health outcomes. Evidence similarly exists, although it is more limited, for relationships between high levels of self-reported sedentary time and increased risk of poor physical and mental health outcomes.5–7 While self-reported measures of MVPA and sedentary time are necessary for gathering information about specific activity types and behavioural contexts, these measures may be influenced by recall bias, social desirability bias, participant comprehension and/or other sources of random and systematic error.8 9 Further, there are dozens of commonly used physical activity surveys, limiting the potential to harmonise self-reported data across studies using different measures. Self-reported measures of physical activity also largely focus on leisure-time MVPA and attempts to measure other important aspects of movement behaviour, such as light intensity activity (LPA), vigorous intermittent lifestyle physical activity (VILPA)10 and patterns or timing of these behaviours, have largely been unsuccessful. As a result, evidence for the associations among LPA, VILPA and physical activity timing with disease incidence and mortality is sparse and inconsistent.10–14 Estimates from device-based measures of MVPA and sedentary time may not be strictly objective, but they are not subject to many of the aforementioned limitations of survey-based estimates.

There is a small number of epidemiologic cohort studies with long-term follow-up (allowing for the accumulation of deaths and incident disease) that have device-based physical activity and sedentary time measures for a subset of their participants, including Women’s Health Study (n=17 061),15 Women’s Health Initiative (n=6382)16 and UK Biobank (n=96 600).17 However, several of these accelerometry studies, especially those within US-based cohorts, are largely made up of non-Latino white women. Further, many accelerometry cohorts are likely too small to study most cancer outcomes without multi-study data pooling.

To address these limitations, accelerometer data were recently collected among a subset of participants in the Cancer Prevention Study-3 (CPS-3), a prospective cohort study of men and women across the USA.18 This cohort profile aims to describe the data collection protocols and the data management methods of the CPS-3 Accelerometry Substudy. Additionally, the baseline characteristics of the subcohort are presented.

Cohort descriptionPatient and public involvement

Patients and/or the public were not involved in the design, conduct, reporting or dissemination of this research.

Study population and setting

The CPS-3 is a prospective cohort study of cancer incidence and mortality initiated by the American Cancer Society (ACS) and has previously been described in detail.18 Briefly, participants were recruited at ACS fundraising events or community enrolment drives in 35 states and Puerto Rico between 2006 and 2013. Over 304 000 participants aged 30–65 years with no history of cancer (except for basal or squamous cell skin cancer) were enrolled. Participants are followed for incident cancers through linkage with state registries and for cause-specific mortality through linkage with the National Death Index. CPS-3 participants provided written informed consent and completed an initial survey at enrolment and are sent follow-up surveys every 3 years (starting in 2015) to update health history and exposure information. All aspects of the CPS-3 are approved by the Emory University Institutional Review Board (#00059007).

From 2019 to 2023, 23 waves of CPS-3 participants were emailed an invitation to register for the Accelerometry Substudy. CPS-3 participants were included in the invite if they: (a) returned a complete 2018 CPS-3 follow-up survey (n=164 810), (b) lived in the USA and (c) had a valid email address on file. To enhance the gender and racial/ethnic diversity of the subcohort, all men and non-white participants were invited first, followed by a random sample of white women (as 64% of the CPS-3 baseline cohort are non-Latina white women). On the registration website, participants were asked about two additional exclusion criteria: (1) the inability to walk without a cane or walker and (2) current pregnancy. Participants were also asked for their pants size so that an accelerometer belt could be appropriately sized. Participants were given 4 weeks from the time of their invite to register for the subcohort.

Data collection

After each wave of invitations, newly registered participants were shipped an accelerometer kit. All participants were sent a kit within 6 weeks of registration, and approximately 175 kits were shipped out each week. Each kit included (1) an ActiGraph GT3X+ accelerometer and belt, (2) wear protocol instructions, (3) a return checklist, (4) a pre-paid shipping label, (5) two strips of shipping tape for re-sealing the box for return and (6) a four-page paper survey that assessed self-reported physical activity, sedentary time, sleep, smoking, light exposure and medications. On the last page of the survey, participants were asked to log what time they woke up, put the device on, took the device off and went to bed each day.

Participants were instructed to wear the accelerometer on the hip aligning with the midline of the non-dominant thigh for 7 consecutive days during all waking hours, except during bathing or other water-based activities (eg, swimming).19 20 Accelerometers were initialised at a sampling rate of 30 Hz without the low-frequency extension using ActiLife software V.6.13.4.21–23 The ActiGraph GT3X series device has been shown to produce similar values for total physical activity volume, sedentary time, steps and time in each physical activity intensity as direct observation and doubly labelled water.24

Participants were emailed reminders to return their completed kits 21 and 35 days after they were shipped to them. Participants who did not ship their kits back after these reminders were sent a personalised email from the lead study investigator 49 days after their shipment, followed by a phone call 56 days post shipment and a final mailed letter 63 days post shipment.

Accelerometer data were downloaded within a few days of their return to ACS. Raw ActiGraph data were processed using the Choi algorithm to discern accelerometer wear time from non-wear time.21 22 Then, the sojourn-3 axis algorithm was used to estimate daily steps, sedentary time and light, moderate and vigorous-intensity physical activity. The sojourn-3 axis method is a hybrid machine learning/neural network algorithm that uses second-by-second triaxial accelerometer counts to estimate free-living physical activity and sedentary time.23 All raw data were retained to allow for the development and calculation of other measures of physical activity and sedentary behaviour. Wear dates were confirmed with the participant logs to avoid classifying non-human movement, such as shipment time, as wear time.19

Data from all participants were retained, but the main results of this paper include participants with 3 or more days of at least 10 hours of wear per day (ie, at least 3 ‘adherent days’).19 25–27

Findings to date

In total, 109 780 CPS-3 participants were invited to participate in the Accelerometry Substudy, of which 23 111 registered and were shipped an accelerometer kit (figure 1). Among those who were shipped a kit, 91.8% returned an accelerometer and wear log (n=21 219), and 91.1% provided at least one complete day of wear (eg, at least 10 hours of wear/day; n=21 048). The most common reasons for incomplete accelerometer wear were acute illness (eg, COVID-19 infection) or injury and scheduling conflicts.

Figure 1

CPS-3 Accelerometry Subcohort participant invite flow. An adherent day is one with at least 10 hours of wear. CPS-3, Cancer Prevention Study-3.

Participants’ accelerometer kits were returned to ACS in an average of 22.7 days. In total, 1600 accelerometer devices were purchased for the subcohort, and 520 were lost from 2019 to 2023; however, as each device was used by an average of 14 participants, only 2.2% of participants lost a device.

Participants included in analyses (ie, those with at least 3 days of valid wear; n=20 950) were predominantly female (n=16 187, 77.3%) and non-Latino white (n=17 977, 85.8%) with a mean age of 58 years (SD=9.8; table 1). Most participants (58.4%) wore their accelerometers in the year 2021 or 2022. Compared with participants who were invited but did not register or withdrew before receiving their accelerometer, Accelerometer Substudy participants had a lower body mass index, a higher level of formal education, were more likely to live in a metropolitan area, were more likely to be able to carry out activities of daily living and self-reported more MVPA.

Table 1

Participant differences by registration and completion status

Most participants who were sent an accelerometer (87%) wore it for 10 or more hours per day for 6 or 7 days (table 2). Median daily wear time was roughly 15.3 hours/day and did not vary considerably by wear day (median difference between the longest and shortest daily wear time=12 min/day).

Table 2

Description of accelerometry measures from participants with at least three adherent* days (n=20 950)

Daily MVPA time was skewed right (figure 2), with a median time of 29 min/day (IQR=36) in moderate-intensity aerobic physical activity and 3 min/day (IQR=13) in vigorous-intensity aerobic physical activity. The distribution of daily time spent sedentary was normal (figure 3), and the median time spent sedentary was roughly 9.4 hours (564 min, IQR=171 min). Participants took roughly 6474 steps per day (IQR=5348). Median daily MVPA and sedentary time were similar across wear-years and across weekends versus weekdays (table 2).

Figure 2

Mean MVPA min/day. MVPA, moderate-to-vigorous intensity aerobic physical activity. N, number of participants.

Figure 3

Mean sedentary min/day. N, number of participants.

Strengths, limitations, and further details

The CPS-3 Accelerometry Substudy has several notable strengths. First, this cohort adds to the limited sex and racial/ethnic diversity of existing US-based accelerometry cohorts, with at least three complete days of data on 2973 total participants who reported a race/ethnicity other than non-Latino white and 4763 men. Additionally, participants appeared to wear the device for most, if not all, waking hours (median wear=920 min/day) and wore the device consistently across all days of wear. Over 91% of participants who were sent an accelerometer returned the device with at least one valid day of wear, and 87% had at least 6 valid days of wear. Finally, the use of a waist-worn ActiGraph GT3X+ device allows for the potential to pool data with several other cohorts, including the Women’s Health Study and Women’s Health Initiative, to study more rare outcomes beyond all-cause and all-cancer mortality.

Conversely, a waist-worn device generally requires a waking protocol (as it is too uncomfortable to wear the device to bed), limiting the ability to measure sleep. Another limitation of this study is that, like nearly all other epidemiologic cohort studies with accelerometer data, participants only wore the device at one time point for a maximum of 1 week. However, several studies have shown that 1 week, and even as little as 3 days, of accelerometry data estimates long-term patterns of movement behaviours.19 20 25 27 Another potential limitation of this subcohort is that data collection spanned 3.5 years, in part because of a 6-month pause at the start of the COVID-19 pandemic. Timing of data collection around the pandemic may also be a limitation as studies (including one study using self-reported data within CPS-3)28 suggest that the pandemic may have affected usual physical activity and sedentary time patterns. However, there was no evidence of pandemic-related variations in movement behaviours in the current study as median MVPA and sedentary time did not vary considerably by wear year.

With a median of 32 min/day of MVPA (47.5 min/day on average), most participants in the CPS-3 Accelerometry Substudy are meeting the US MVPA guidelines. However, compared with other US-based cohorts with waist-worn ActiGraph data, such as the Women’s Health Study and Women’s Health Initiative, CPS-3 participants are slightly less active. Although there were differences in data processing (eg, different cutpoints applied), device wear (eg, varying mean wear time) and participant demographics that may account for differences in estimated MVPA time, participants in the Women’s Health Study accumulated 91.9 min/day of MVPA and participants in the Women’s Health Initiative accumulated 50.4 min/day of MVPA.16 However, participants in all three cohorts are much more active than the nationally representative participants in The National Health and Nutrition Examination Survey (NHANES), who accumulated about 14 min/day of MVPA on average (2003–2004, ActiGraph 7164).

The CPS-3 Accelerometry Substudy is the largest US-based epidemiologic cohort of accelerometry data to date. Together with the continuous collection of data from surveys, state cancer registries and the National Death Index, the CPS-3 Accelerometer Substudy will be well powered to address complex questions around the relationships between health outcomes (namely cancer incidence, survival and quality of life post diagnosis) with the volume, patterns and timing of physical activity and sedentary behaviour.

Collaboration

The data underlying the findings of this study are restricted by the Emory University Institutional Review Board. Data are available from the ACS by following the ACS Data Access Procedures (https://www.cancer.org/research/population-science/research-collaboration.html) for researchers who meet the criteria for access to confidential data. Please email cohort.data@cancer.org to inquire about access.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Emory University Institutional Review Board (#00059007). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

The authors express sincere appreciation to all Cancer Prevention Study-3 participants and each member of the study and biospecimen management group. The views expressed here are those of the authors and do not necessarily represent the American Cancer Society or the American Cancer Society—Cancer Action Network.