Summary
Background:
Venous stasis, which can occur with prolonged sedentary behavior (SB), is associated with venous thromboembolism (VTE) risk, but VTE risk associated with accelerometer-measured SB has not been quantified.
Objectives:
To evaluate accelerometer-based measures of SB in relation to incident VTE.
Methods:
We included 5,591 participants, aged 63–99 years, of the Women’s Health Initiative Objective Physical Activity and Cardiovascular Health cohort study without prior VTE. Between May 2012–2014, participants wore the ActiGraph GT3X+ accelerometer at the hip for 7 days. Three SB measures were classified using the Convolutional Neural Network Hip Accelerometer Posture algorithm: total sitting time, mean sitting bout duration, and total time spent in prolonged (≥30 minute) sitting bouts. VTE events were centrally adjudicated. Multivariable-adjusted Cox models estimated hazard ratios (HRs) for each SB and VTE risk. Women were censored at first VTE, death, loss to follow-up, or February 2023. Mediation by body mass index (BMI) was evaluated.
Results:
Over a mean follow-up of 8.2 years, 229 women experienced a VTE. In adjusted models, longer mean sitting bout duration was associated with greater incident VTE risk (HR per 5-minute increase=1.15; 95% CI: 1.04, 1.28). BMI mediated approximately 30% of this association (p<0.01). We found no significant evidence that total sitting time or total time spent in prolonged sitting bouts were associated with VTE.
Conclusion:
Longer mean sitting bout duration was associated with greater VTE risk, with substantial mediation by BMI. Behavioral efforts to reduce sedentary bout length in older women may reduce their VTE risk.
Keywords: pulmonary embolism, sedentary behavior, venous thromboembolism, women
Introduction
Venous thromboembolism (VTE), which includes both deep vein thrombosis (DVT) and pulmonary embolism (PE), is the third most common cardiovascular disease in the United States.[1,2] Virchow’s Triad proposes that venous stasis, which can occur during prolonged bouts of sedentary behavior (SB), is one of three primary factors contributing to the development of VTE.[3] Self-reported television viewing[4,5], hours spent sitting[6], proportion of time spent sleeping and sitting[7], and travel[8] have been associated with a higher risk of VTE[4,6–8] and PE-related mortality[5] in most, but not all, studies.[9]
In these prior studies of VTE risk, SB information was collected using self-report. However, SB is typically underestimated by self-reported measures.[10] Accelerometers provide an objective measurement of SB, thereby reducing participant recall bias. Accelerometers also allow for examination of SB characteristics (e.g. mean sitting bout duration, total time spent in prolonged sitting bouts), which self-report of SB may not capture. Yet to our knowledge, no studies to date have evaluated accelerometer-measured estimates of SB in relation to VTE risk.
In this prospective cohort study among older ambulatory women set within the Women’s Health Initiative Objective Physical Activity and Cardiovascular Health (OPACH) study, we evaluated how accelerometer-based measures of SB, including total sitting time, mean sitting bout duration, and total time spent in prolonged sitting bouts, relate to incident VTE risk. In secondary analyses, we estimated what proportion of any association present between an SB measure and VTE risk was mediated by body mass index (BMI), given that higher SB has prospectively been associated with increases in BMI over time[11] and that BMI has a dose-response relation with VTE risk that suggests a greater VTE risk with higher BMI.[2] Beyond its possible role as a mediator, BMI may also confound these associations, given that it may be associated with both SB exposures and VTE outcomes. Because of the proposed complex role of BMI in the relation between SB and VTE, it may be both a mediator and a confounder of this association. Further secondary analyses separately evaluated the risks of incident PE and incident DVT.
Methods
Study Population
The Women’s Health Initiative (WHI) recruited and enrolled 161,808 postmenopausal women aged 50–79 at 40 sites throughout the United States between 1993 and 1998, with details previously published.[12] Participants for this analysis were from the Objective Physical Activity and Cardiovascular Health (OPACH) study; an ancillary study to the WHI, for which extensive recruitment details and descriptive characteristics have also been previously published.[13] The OPACH study was originally designed to evaluate accelerometer-measured physical activity in relation to the risk of cardiovascular events among women in later life.[13,14] In brief, OPACH participants were recruited between 2012 and 2014 and included ambulatory participants of the WHI Long Life Study who also consented to wear an ActiGraph GT3X+ triaxial accelerometer and to complete a sleep log and OPACH physical activity questionnaire. The WHI Long Life Study focused on healthy aging and in-home examinations and collection of CVD biomarker data; participants (and thus, OPACH participants) were recruited from all 40 original clinical centers[13] in all 50 states.[15] The OPACH study protocol was approved by the Fred Hutchinson Cancer Research Center institutional review board, and all women provided informed consent.
Between May 2012 and April 2014, consenting ambulatory, community-dwelling WHI OPACH participants aged 63 to 99 years (n=7,058) were instructed to wear an ActiGraph GT3X+ accelerometer for up to 7 consecutive days and to complete a supplementary questionnaire and sleep log. Of these consented participants, 10 died before receiving study materials, 327 did not return accelerometers, and 232 accelerometers were returned without usable data, resulting in 6,489 women with accelerometer data available for analysis (92.1%).[13,16] For the present analysis, we further excluded women with a history of VTE before OPACH baseline (n=300), non-adherent accelerometer wear (<4 days of ≥10 hours/day of wear)[17] (n=343), and who returned accelerometers with data on which the SB algorithm could not be successfully applied (n=255), resulting in a final analytic sample of 5,591 women (86.1% of women with available accelerometer data).
Accelerometer-Based Measurement
OPACH participants wore the ActiGraph GT3X+, a triaxial accelerometer, on their right hip for 24 hours per day for up to 7 consecutive days while both awake and sleeping, except for when bathing or swimming.[18] Participants were not provided any real-time feedback of their sedentary or physical activity behaviors while wearing the accelerometer. ActiGraph GT3X+ accelerometers measured acceleration at 30 Hz,[16] and ActiLife version 6 software was used to aggregate the 30 Hz data into 15-second epochs using the normal filter.[16,19] Accelerometer non-wear periods were identified for removal using the Choi algorithm, applied to vector magnitude acceleration counts with a 90-minute window, 30-minute steam frame, and 2-minute tolerance.[14,20] Participants recorded time in and out of bed using a sleep log, and this information was used to estimate and remove time spent sleeping.
To estimate SB measures, the Convolutional Neural Network Hip Accelerometer Posture (CHAP) algorithm[21] was used to classify OPACH participants’ hip-worn ActiGraph GT3X+ triaxial accelerometer data into minute-level sitting data. CHAP was originally developed in the Adult Change in Thought (ACT) study of older adults, and its development has previously been described in-depth.[19,21] In brief, 709 ACT participants concurrently wore a hip-worn ActiGraph GT3X+ triaxial accelerometer and a thigh-worn activPAL micro3 inclinometer as the criterion measure and a machine-learning approach was used to extract ActiGraph GT3X+ features to classify time-points into sitting or non-sitting.[19,21] Agreement between CHAP algorithm-identified and activPAL inclinometer-measured minute level sitting were high (97.1% sensitivity; 88.6% specificity.[21]
Minute-level CHAP-classified sitting data was used to compute three person-level measures of SB: total sitting time, mean sitting bout duration, and total time spent in prolonged sitting bouts. Total sitting time was estimated by summing total daily sitting time in minutes across adherent wear days and dividing by number of adherent wear days; we reported total sitting time in hours per day units for interpretability. Mean sitting bout duration was estimated using data from all adherent days, with higher bout durations generally indicating more prolonged accumulation patterns and lower bout durations indicating interrupted patterns.[16] Total time spent in prolonged sitting bouts was estimated by summing time across adherent wear days spent in sitting bouts lasting 30 minutes or longer and dividing by the number of adherent wear days.
Venous Thromboembolism
The primary outcome of interest in this study was VTE, including both DVT and PE. Annually, participants reported inpatient and outpatient DVT and PE events and relevant medical records were obtained to centrally adjudicate potential events. Kappa coefficients for DVT and PE based on physician adjudication were 0.80 and 0.84, respectively, [21] and based on Medicare claims were 0.87 and 0.91, respectively.[22] Follow-up for incident first VTE events continued through February 2023. Details on WHI outcomes ascertainment and adjudication criteria have been published.[23] We separately evaluated DVT with or without PE and PE with or without DVT as secondary endpoints of interest.
Covariates and Other Measures
Questionnaires at WHI study baseline collected self-reported age, race and ethnicity (Black, Hispanic or Latina, White), education (high school graduate/equivalent or less, some college, college graduate or more), and family history of VTE. In WHI, age at menopause was operationally defined as the youngest age of any of the following measures as self-reported at WHI baseline: last menstrual bleeding, bilateral oophorectomy, or initiation of hormone replacement therapy.
During in-home visits conducted at or near OPACH baseline, trained study staff measured height, weight, and systolic blood pressure (SBP; recorded as the average of two measures) and collected fasting blood samples. Physical function, was measured using the Short Physical Performance Battery (SPPB; 0–12, with lower scores indicating lower physical function).[13] Body mass index (BMI) was calculated as weight (in kilograms) divided by height (in meters) squared. Serum levels of high-density (HDL-C) and total cholesterol, triglycerides, and C-reactive protein were measured at the University of Minnesota[18].
Vector magnitudes from accelerometer wear were averaged and reported as objectively measured total physical activity volume.[13] Categories of vector magnitude counts were established during a separate laboratory calibration study and used to estimate intensity-specific physical activity time (minutes/day), with light physical activity defined as 19 to <519 counts/15 seconds and moderate-to-vigorous physical activity (MVPA) defined as ≥519 counts/15 seconds.[13] Steps per 15 second epoch were calculated using ActiLife’s proprietary algorithm and total step count was divided by the number of adherent wear days to calculate steps/day.[13,19]
From the most recent questionnaire completed prior to OPACH baseline, we measured self-reported general health (poor, fair, good, very good/excellent), physical function using the RAND-36 questionnaire (with summary scores ranging from 0 to 100 and higher scores indicating greater physical function)[17], and current smoking status. History of cancer was self-reported and history of hypertension, high cholesterol, and diabetes were defined using participant self-report of a physician diagnosis combined with treatment by medication. Participants self-reported their time spent sitting and separately, lying down. Diet quality was quantified using the Healthy Eating Index-2010.[13]
Statistical Analyses
We described demographic and health-related characteristics of eligible study participants using means and standard deviations (SDs) or percentages across approximate quartiles (quartiles rounded to the nearest minute-level, for interpretability) of mean sitting bout duration.
Participants contributed person-time from OPACH baseline to the first date of VTE diagnosis, death from any cause, loss to follow-up, or end of follow-up (February 2023). In primary analyses, we used Cox proportional hazards models to estimate multivariable-adjusted hazard ratios (HRs) for VTE associated with three separate sedentary exposures of interest: (1) total sitting time; (2) mean sitting bout duration; and (3) total time spent in prolonged sitting bouts. All exposure variables were adjusted for awake wear time using the residuals method.[16,24] We estimated Pearson’s correlation coefficients between the three SB of interest, MVPA (residualized and then winsorized at the 99th percentile), and awake accelerometer wear time.
As decided a priori, each exposure was modeled categorically in approximate quartiles, and separately, continuously. The significance of associations in models using categorical exposures was tested using the Wald test for group differences. In analyses using continuous linear exposure variables, we estimated HRs per unit SD (rounded SD, for interpretability) increase. Evidence of non-linearity was evaluated using adjusted spline models.[25] We found no evidence of non-proportional hazards using Schoenfeld residuals (all p>0.05).
We used a series of models to adjust for demographic characteristics, and potential confounders and mediators identified a priori: Model 1 adjusted for continuous age and race and ethnicity (Black, Hispanic, and White). Model 2 adjusted for model 1 covariates plus potential confounders (education [high school/GED or less, some college, and college graduate or more], smoking status [non-smoker/former smoker vs. current smoker], and history of cancer). Model 3 adjusted for model 2 covariates plus continuous BMI; we adjusted for BMI separately from other covariates in model 2 because BMI could both confound and mediate the association between SB and VTE risk. All analyses using the total time in prolonged sitting bouts exposure additionally adjusted for time spent in sitting bouts <30 minutes. Sensitivity analysis models included continuous MVPA in models 2 and 3. For primary models with significant evidence of an association, we conducted causal mediation analyses[26] that account for exposure-mediator interaction to evaluate the degree to which the associations between SB and incident VTE risk were mediated by BMI by deriving estimates for the direct and indirect effects of BMI on VTE risk.
Further sensitivity analyses adjusted for additional potential mediating factors. Specifically, model 4 adjusted for model 3 covariates plus potential mediating physical function and health-related factors (RAND-36 physical function score [continuous] and self-rated general health [excellent, very good, good, fair and poor]). Model 5 adjusted for model 3 covariates plus potential mediating cardiovascular health-related factors (continuous systolic blood pressure [SBP], HDL-C, triglyceride levels, and logarithmic CRP levels).
In secondary analyses, we evaluated the presence of interaction between each continuous SB exposure of interest and, separate proposed effect modifiers (1) continuous BMI, 2) cancer history, 3) continuous age, and, 4) continuous MVPA) in relation to risk of any incident VTE. To evaluate the presence of statistical interaction, we used the fifth imputed dataset to run models that included Model 2 covariates and were with and without interaction terms included between each continuous SB exposure of interest and each proposed effect modifier (with models including main effects). Likelihood ratio tests were then used to compare models with and without the interaction term, with likelihood ratio p-values <0.05 suggesting significant evidence of statistical interaction. We conducted stratified analyses using categorical variables for any factors with significant evidence of interaction.
Additional secondary analyses used Cox proportional hazards models to estimate multivariable-adjusted HRs for PE, and separately, DVT. Given our interest in incident VTE risk, participants were censored at DVT if they occurred before PE in analyses of PE risk and at PE if they happened before DVT in analyses of DVT risk (in addition to our standard censoring factors already described).
Covariates with missing data (smoking status [7.7%], education [0.6%], physical function [1.0%], self-rated general health [0.3%], BMI [6.1%], SBP [5.4%], HDL-C [20.6%], triglyceride levels [20.6%], and logarithmic CRP levels [20.6%]) were multiply imputed using a chained equations approach[27] with 10 imputed datasets. In a sensitivity analysis, missing covariate data was not imputed, meaning that only participants with complete data were included in these sensitivity analyses. All analyses were performed in Stata 15.0.[28]
Results
We identified 229 incident VTE events during a mean follow-up of 8.2 years. At OPACH baseline, participants had a mean age of 78.6 years (SD 6.7 years), with a mean BMI of 28.0 kg/m2 (SD 5.7), with 90.3% of participants self-rating their health as “good” or better (Table 1). On average, women in higher quartiles of mean sitting bout duration were slightly older, had a higher BMI, and a greater proportion had a history of conditions including cancer, hypertension, and diabetes than women in lower quartiles of mean sitting bout duration. In Supplemental Table 1, we report additional participant characteristics by quartiles of mean sitting bout duration at OPACH baseline.
Table 1.
Characteristics of participants by quartiles of accelerometer-measured mean sitting bout duration (minutes), at OPACH baseline.
| Mean sitting bout duration (minutes) |
|||||
|---|---|---|---|---|---|
| Total n=5,591 | Quartile 1 ≤9 minutes n=1,270 |
Quartile 2 >9 to 12 minutes n=1,612 |
Quartile 3 >12 to 15 minutes n=1,279 | Quartile 4 >15 minutes n=1,430 |
|
|
| |||||
| Characteristic** | |||||
|
| |||||
| Age, years (mean, SD) | 78.6 (6.7) | 76.8 (6.6) | 78.1 (6.6) | 79.1 (6.5) | 80.4 (6.5) |
| Race and ethnicity (n [%]) | |||||
| White | 2,785 (49.8) | 513 (40.4) | 759 (47.1) | 666 (52.1) | 847 (59.2) |
| Black | 1,850 (33.1) | 485 (38.2) | 552 (34.2) | 392 (30.6) | 421 (29.4) |
| Hispanic or Latina | 956 (17.1) | 272 (21.4) | 301 (18.7) | 221 (17.3) | 162 (11.3) |
| College graduate or more (n [%]) | 2,315 (41.4) | 538 (42.4) | 672 (41.7) | 549 (42.9) | 556 (38.9) |
| BMI, kg/m2 (mean, SD) | 28.0 (5.7) | 26.6 (4.9) | 27.4 (5.4) | 28.1 (5.6) | 29.8 (6.2) |
| Current smoker (n [%])** | 128 (2.3) | 34 (2.7) | 32 (2.0) | 32 (2.5) | 30 (2.1) |
| Age at menopause, y (mean, SD) | 48.1 (6.5) | 48.0 (6.3) | 48.1 (6.6) | 48.2 (6.6) | 48.1 (6.4) |
| History of cancer (n [%]) | 919 (16.4) | 166 (13.1) | 250 (15.5) | 219 (17.1) | 284 (19.9) |
| History of hypertension (n [%]) | 4,002 (71.6) | 855 (67.3) | 1,121 (69.5) | 920 (71.9) | 1,106 (77.3) |
| History of diabetes (n [%]) | 1,121 (20.1) | 185 (14.6) | 302 (18.7) | 266 (20.8) | 368 (25.7) |
| History of high cholesterol requiring meds (n [%]) | 2,325 (41.6) | 449 (35.4) | 662 (41.1) | 574 (44.9) | 640 (44.8) |
| Healthy Eating Index (mean, SD) | 72.1 (10.3) | 73.0 (10.1) | 72.4 (10.1) | 72.5 (10.2) | 70.5 (10.7) |
| Self-rated general health (n [%]) | |||||
| Excellent | 569 (10.2) | 170 (13.4) | 183 (11.4) | 130 (10.2) | 86 (6.0) |
| Very good | 2,317 (41.4) | 572 (45.0) | 671 (41.6) | 547 (42.8) | 527 (36.9) |
| Good | 2,166 (38.7) | 431 (33.9) | 633 (39.3) | 498 (38.9) | 604 (42.2) |
| Fair | 497 (8.9) | 84 (6.6) | 120 (7.4) | 101 (7.9) | 192 (13.4) |
| Poor | 25 (0.4) | 7 (0.6) | 1 (0.1) | 3 (0.2) | 14 (1.0) |
| Not Reported | 17 (0.3) | 6 (0.5) | 4 (0.2) | 0 (0.0) | 7 (0.5) |
| SPPB Score (n [%]) | |||||
| 0–4 | 401 (7.2) | 45 (3.5) | 87 (5.4) | 96 (7.5) | 173 (12.1) |
| 5–8 | 1,972 (35.3) | 389 (30.6) | 549 (34.1) | 471 (36.8) | 563 (39.4) |
| 9–12 | 2,436 (43.6) | 651 (51.3) | 780 (48.4) | 552 (43.2) | 453 (31.7) |
| Unknown | 782 (14.0) | 185 (14.6) | 196 (12.2) | 160 (12.5) | 241 (16.9) |
| Systolic blood pressure (mean, SD)** | 126.2 (14.8) | 124.1 (13.7) | 125.4 (14.6) | 127.0 (15.3) | 128.0 (15.3) |
| HDL cholesterol (mean, SD)** | 60.5 (15.0) | 63.3 (15.8) | 61.8 (14.6) | 60.2 (15.1) | 57.3 (14.3) |
| Family history of VTE (n [%]) | 444 (7.9) | 103 (8.1) | 113 (7.0) | 104 (8.1) | 124 (8.7) |
BMI = body mass index; CRP = C-reactive protein; d=day; h = hours; m = meters; MET = metabolic equivalent of task; MSBD = mean sitting bout duration; min = minutes; kg = kilograms; MVPA = moderate-to-vigorous physical activity; PA = physical activity; SD = standard deviation; SPPB = short physical performance battery; VTE = venous thromboembolism.
All percentages presented are among non-missing values. Characteristics with missingness were: current smoker (7.7%); RAND SF-36 physical function score (1.0%); CRP (20.6%); systolic blood pressure (5.4%); HDL cholesterol (5.4%); triglycerides (20.6%).
Data has been residualized
Data has been residualized then winsorized at 99th percentile
Measures of SB were moderately to strongly correlated with one another, with total time in prolonged sitting bouts and total sitting time most strongly correlated (r=0.88) (Supplemental Table 2). MVPA was moderately negatively correlated with all three SB. Awake accelerometer wear time was not correlated with SB or MVPA (Supplemental Table 2).
Sedentary Behaviors and Risk of Any VTE
We found no evidence of an association between total sitting time (measured in hours/day) nor total time spent in prolonged sitting bouts (i.e. bouts ≥30 minutes) in relation to incident VTE risk, when exposures were modeled with multivariable adjustments using categorical quartiles or continuous linear exposures (all p-values for group difference, p-values for linear trend, and p-values for non-linearity >0.05) (Table 2). There remained no evidence of association in sensitivity analyses that further adjusted for MVPA (Supplemental Table 3), for potentially mediating physical function and health-related factors, and for cardiovascular health factors (Supplemental Table 4). In complete case sensitivity analyses of total sitting time and prolonged sitting bout exposures, results were similar (Supplemental Table 5).
Table 2.
Risk of venous thromboembolism, including PE and/or DVT, associated with sedentary behavior (n=5,591).
| Model 1 | Model 2 | Model 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||
| Age & race/ethnicity adjusted* | Primary confounder-adjusted** | Primary confounder + BMI-adjusted*** | |||||||
|
| |||||||||
| Total n | VTE events | Rate/1000 person- years | HR | 95% CI | HR | 95% CI | HR | 95% CI | |
|
| |||||||||
|
Sitting time (hours/d)
| |||||||||
| Q1 (≤9 hours) | 1,180 | 39 | 3.73 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>9 to 10.5 hours) | 1,732 | 71 | 4.83 | 1.16 | (0.79, 1.72) | 1.14 | (0.77, 1.69) | 1.06 | (0.71, 1.57) |
| Q3 (>10.5 to 11.5 hours) | 1,212 | 55 | 5.58 | 1.22 | (0.80, 1.84) | 1.18 | (0.78, 1.78) | 1.01 | (0.66, 1.55) |
| Q4 (>11.5 hours) | 1,467 | 64 | 5.96 | 1.19 | (0.79, 1.79) | 1.11 | (0.74, 1.68) | 0.89 | (0.58, 1.37) |
| p-value for group difference | 0.805 | 0.887 | 0.811 | ||||||
| Linear: Per unit SD increase (100 minutes) | 1.08 | (0.95, 1.24) | 1.06 | (0.93, 1.21) | 0.97 | (0.84, 1.12) | |||
| p-value for linear trend | 0.238 | 0.405 | 0.706 | ||||||
| p-value for non-linearity | 0.510 | 0.470 | 0.368 | ||||||
|
Mean sitting bout duration (min) | |||||||||
| Q1 (≤9 minutes) | 1,270 | 40 | 3.70 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>9 to 12 minutes) | 1,612 | 55 | 4.06 | 1.06 | (0.70, 1.59) | 1.05 | (0.70, 1.58) | 1.00 | (0.67, 1.51) |
| Q3 (>12 to 15 minutes) | 1,279 | 57 | 5.40 | 1.37 | (0.91,2.06) | 1.32 | (0.88, 1.99) | 1.22 | (0.81, 1.84) |
| Q4 (>15 minutes) | 1,430 | 77 | 7.12 | 1.67 | (1.13, 2.46) | 1.59 | (1.08, 2.34) | 1.36 | (0.91,2.04) |
| p-value for group difference | 0.021 | 0.047 | 0.297 | ||||||
| Linear: Per unit SD increase (5 minutes) | 1.17 | (1.05, 1.30) | 1.15 | (1.04, 1.28) | 1.10 | (0.98, 1.23) | |||
| p-value for linear trend | 0.003 | 0.008 | 0.111 | ||||||
| p-value for non-linearity | 0.279 | 0.355 | 0.511 | ||||||
|
Total time spent in prolonged sitting bouts (≥30 minutes) (hours/d) + | |||||||||
| Q1 (≤4.5 hours) | 1,377 | 42 | 3.46 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>4.5 to 6 hours) | 1,527 | 61 | 4.73 | 1.22 | (0.82, 1.82) | 1.20 | (0.81, 1.79) | 1.13 | (0.76, 1.69) |
| Q3 (>6 to 7.5 hours) | 1,320 | 57 | 5.29 | 1.24 | (0.81, 1.89) | 1.20 | (0.79, 1.82) | 1.07 | (0.70, 1.63) |
| Q4 (>7.5 hours) | 1,367 | 69 | 6.97 | 1.44 | (0.90, 2.29) | 1.34 | (0.84, 2.15) | 1.08 | (0.66, 1.77) |
| p-value for group difference | 0.507 | 0.663 | 0.945 | ||||||
| Linear: Per unit SD increase (130 minutes) | 1.08 | (0.91, 1.29) | 1.05 | (0.88, 1.25) | 0.95 | (0.79, 1.14) | |||
| p-value for linear trend | 0.371 | 0.585 | 0.577 | ||||||
| p-value for non-linearity | 0.384 | 0.373 | 0.317 | ||||||
Model 1 adjusted for age and race and ethnicity;
Model 2 adjusted for model 1 covariates + education, smoking status, history of cancer;
Model 3 adjusted for model 2 covariates + BMI.
Models additionally adjusted for time spent in sitting bouts <30 minutes
d=day; DVT = deep vein thrombosis; HR = hazard ratio; min = minutes; PE = pulmonary embolism; Q = quartile.
In contrast, mean sitting bout duration (in minutes) was associated with incident VTE risk in confounder-adjusted models (model 2), with mean sitting bout durations of >15 minutes associated with a 59% greater risk of incident VTE than ≤9 minutes (Q4 vs. Q1 HR=1.59; 95% CI: 1.08, 2.34; p-group difference=0.047) (Table 2). In models with mean sitting bout duration modeled linearly, every 1-SD increase in mean sitting bout duration (i.e., every 5-minute increase) was associated with a 15% greater risk of incident VTE (HR=1.15; 95% CI: 1.04, 1.28; p-trend=0.008). Results from causal mediation analyses suggested that BMI mediated approximately 30% of the association between mean sitting bout duration and incident VTE risk (p<0.01). When BMI was included as a covariate in Cox proportional hazards models (model 3; Table 2), estimates shifted toward the null and were no longer statistically significant (Q4 vs. Q1 HR=1.36; 95% CI: 0.91, 2.04; p-group difference=0.297; and per-SD increase HR=1.10; 95% CI: 0.98, 1.23; p-trend=0.111) (Table 2). Most results from complete case sensitivity analyses of mean sitting bout duration were not meaningfully different such that they changed interpretation of results from those of primary analyses (Supplemental Table 5) but point estimates shifted slightly away from the null. Results from complete case causal mediation analyses suggested that BMI mediated approximately 21% of the association between mean sitting bout duration and incident VTE risk (p=0.03). When BMI was included as a covariate in complete case analyses (model 3; Supplemental Table 5), estimates still shifted toward the null as they had in primary analyses and were no longer statistically significant in quartile-based analyses but remained statistically significant in analyses of continuous exposures (per-SD increase HR=1.17; 95% CI: 1.03, 1.33; p-trend=0.019).
In sensitivity analyses that further adjusted for MVPA for the exposure of mean sitting bout duration, point estimates were not meaningfully different from those resulting from primary models of interest; however, estimates shifted slightly toward the null, and were no longer statistically significant (all p-values for group difference, trend, and non-linearity >0.05) (Supplemental Table 3). Point estimates from additional sensitivity analyses that further adjusted for potentially mediating physical function and health, and separately, cardiovascular health factors were similar to estimates from model 3 that included adjustment for BMI, and there was no statistically significant evidence of associations (Supplemental Table 4). We found no evidence of a non-linear association between any of the three SB exposures evaluated and incident VTE risk (all p-values for non-linearity>0.05).
There was no evidence of interaction between any of the three continuous SB exposures and, separately: cancer history, continuous age, or continuous MVPA, with incident VTE risk, adjusting for model 2 covariates. There was some suggestion of interaction by continuous BMI in the associations between mean sitting bout duration and separately, total time spent in prolonged sitting bouts in relation to incident VTE risk, although p-values were above the threshold for significance of <0.05 (likelihood ratio p-values = 0.06 and 0.05, respectively), with adjustment for model 2 covariates. However, in analyses stratified by obesity status for mean sitting bout duration and separately, time spent in prolonged sitting bouts in relation to incident VTE risk (Supplemental Table 6), we observed no significant evidence of linear associations.
Sedentary Behaviors and Separate Risks of PE and DVT
In secondary analyses with PE and DVT modeled as separate outcomes, we found no evidence of an association between total sitting time nor time spent in prolonged sitting bouts in relation to incident PE (Table 3) nor DVT (Table 4) risk.
Table 3.
Risk of pulmonary embolism associated with sedentary behavior (n=5,591).
| Model 1 | Model 2 | Model 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||
| Age & race/ethnicity adjusted* | Primary confounder-adjusted** | Primary confounder + BMI-adjusted*** | |||||||
|
| |||||||||
| Total n | PE events | Rate/1000 person-years | HR | 95% CI | HR | 95% CI | HR | 95% CI | |
|
Sitting time (hours/d) | |||||||||
| Q1 (≤9 hours) | 1,180 | 20 | 1.91 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>9 to 10.5 hours) | 1,732 | 39 | 2.66 | 1.24 | (0.72, 2.13) | 1.19 | (0.69, 2.05) | 1.11 | (0.65, 1.92) |
| Q3 (>10.5 to 11.5 hours) | 1,212 | 27 | 2.74 | 1.15 | (0.64, 2.06) | 1.10 | (0.61, 1.98) | 0.96 | (0.53, 1.74) |
| Q4 (>11.5 hours) | 1,467 | 39 | 3.63 | 1.39 | (0.80, 2.41) | 1.28 | (0.74, 2.23) | 1.04 | (0.58, 1.85) |
| p-value for group difference | 0.674 | 0.826 | 0.943 | ||||||
| Linear: Per unit SD increase (100 minutes) | 1.16 | (0.97, 1.39) | 1.13 | (0.94, 1.36) | 1.04 | (0.86, 1.27) | |||
| p-value for linear trend | 0.113 | 0.198 | 0.661 | ||||||
| p-value for non-linearity | 0.153 | 0.152 | 0.129 | ||||||
|
Mean sitting bout duration (min) | |||||||||
| Q1 (≤9 minutes) | 1,270 | 18 | 1.66 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>9 to 12 minutes) | 1,612 | 24 | 1.77 | 1.03 | (0.56, 1.90) | 1.02 | (0.55, 1.87) | 0.98 | (0.53, 1.81) |
| Q3 (>12 to 15 minutes) | 1,279 | 34 | 3.22 | 1.83 | (1.03, 3.25) | 1.74 | (0.98, 3.09) | 1.63 | (0.91,2.92) |
| Q4 (>15 minutes) | 1,430 | 49 | 4.53 | 2.38 | (1.38, 4.11) | 2.22 | (1.28, 3.84) | 1.98 | (1.12, 3.49) |
| p-value for group difference | 0.001 | 0.003 | 0.018 | ||||||
| Linear: Per unit SD increase (5 minutes) | 1.25 | (1.10, 1.42) | 1.22 | (1.08, 1.38) | 1.17 | (1.02, 1.35) | |||
| p-value for linear trend | 0.001 | 0.002 | 0.021 | ||||||
| p-value for non-linearity | 0.025 | 0.033 | 0.051 | ||||||
|
Total time spent in prolonged sitting bouts ≥30 minutes (hours/d) + | |||||||||
| Q1 (≤4.5 hours) | 1,377 | 16 | 1.32 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference |
| Q2 (>4.5 to 6 hours) | 1,527 | 35 | 2.71 | 1.80 | (0.99, 3.26) | 1.75 | (0.96, 3.17) | 1.67 | (0.92, 3.04) |
| Q3 (>6 to 7.5 hours) | 1,320 | 28 | 2.60 | 1.47 | (0.78, 2.79) | 1.41 | (0.75, 2.66) | 1.30 | (0.68, 2.47) |
| Q4 (>7.5 hours) | 1,367 | 46 | 4.65 | 2.08 | (1.06, 4.06) | 1.93 | (0.98, 3.79) | 1.64 | (0.81, 3.32) |
| p-value for group difference | 0.134 | 0.187 | 0.324 | ||||||
| Linear: Per unit SD increase (100 minutes) | 1.12 | (0.88, 1.41) | 1.08 | (0.85, 1.36) | 0.99 | (0.77, 1.27) | |||
| p-value for linear trend | 0.357 | 0.533 | 0.934 | ||||||
| p-value for non-linearity | 0.124 | 0.128 | 0.108 | ||||||
Model 1 adjusted for age and race and ethnicity;
Model 2 adjusted for model 1 covariates + education, smoking status, history of cancer;
Model 3 adjusted for model 2 covariates + BMI.
Models additionally adjusted for time spent in sitting bouts <30 minutes
d=day; DVT = deep vein thrombosis; HR = hazard ratio; min = minutes; PE = pulmonary embolism; Q = quartile.
Table 4.
Risk of deep vein thrombosis associated with sedentary behavior (n=5,591).
| Model 1 | Model 2 | Model 3 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||
| Age & race/ethnicity adjusted* | Primary confounder-adjusted** | Primary confounder + BMI-adjusted*** | ||||||||
|
| ||||||||||
| Total n | DVT events | Rate/1000 person-years | HR | 95% CI | HR | 95% CI | HR | 95% CI | ||
|
Sitting time (hours/d) | ||||||||||
| Q1 (≤9 hours) | 1,180 | 24 | 2.30 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference | |
| Q2 (>9 to 10.5 hours) | 1,732 | 45 | 3.06 | 1.20 | (0.73, 1.97) | 1.19 | (0.73, 1.96) | 1.10 | (0.66, 1.81) | |
| Q3 (>10.5 to 11.5 hours) | 1,212 | 42 | 4.26 | 1.51 | (0.91,2.52) | 1.48 | (0.89, 2.47) | 1.26 | (0.75, 2.11) | |
| Q4 (>11.5 hours) | 1,467 | 42 | 3.91 | 1.27 | (0.76, 2.12) | 1.21 | (0.73, 2.03) | 0.94 | (0.55, 1.62) | |
| p-value for group difference | 0.432 | 0.474 | 0.606 | |||||||
| Linear: Per unit SD increase (100 minutes) | 1.10 | (0.93, 1.29) | 1.08 | (0.91, 1.27) | 0.98 | (0.82, 1.16) | ||||
| p-value for linear trend | 0.266 | 0.372 | 0.798 | |||||||
| p-value for non-linearity | 0.460 | 0.431 | 0.377 | |||||||
|
Mean sitting bout duration (min) | ||||||||||
| Q1 (≤9 minutes) | 1,270 | 25 | 2.31 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference | |
| Q2 (>9 to 12 minutes) | 1,612 | 37 | 2.73 | 1.13 | (0.68, 1.88) | 1.13 | (0.68, 1.89) | 1.07 | (0.64, 1.79) | |
| Q3 (>12 to 15 minutes) | 1,279 | 37 | 3.51 | 1.41 | (0.84, 2.34) | 1.38 | (0.83, 2.30) | 1.25 | (0.75, 2.10) | |
| Q4 (>15 minutes) | 1,430 | 54 | 5.00 | 1.84 | (1.14, 2.97) | 1.80 | (1.11,2.92) | 1.51 | (0.91,2.49) | |
| p-value for group difference | 0.041 | 0.056 | 0.323 | |||||||
| Linear: Per unit SD increase (5 minutes) | 1.19 | (1.05, 1.35) | 1.19 | (1.05, 1.35) | 1.12 | (0.98, 1.29) | ||||
| p-value for linear trend | 0.006 | 0.008 | 0.109 | |||||||
| p-value for non-linearity | 0.343 | 0.402 | 0.556 | |||||||
|
Total time spent in prolonged sitting bouts ≥30 minutes (hours/d)+ | ||||||||||
| Q1 (≤4.5 hours) | 1,377 | 30 | 2.47 | 1.00 | Reference | 1.00 | Reference | 1.00 | Reference | |
| Q2 (>4.5 to 6 hours) | 1,527 | 35 | 2.71 | 0.99 | (0.60, 1.62) | 0.98 | (0.60, 1.60) | 0.91 | (0.55, 1.49) | |
| Q3 (>6 to 7.5 hours) | 1,320 | 45 | 4.18 | 1.39 | (0.85, 2.28) | 1.36 | (0.83, 2.22) | 1.18 | (0.71, 1.94) | |
| Q4 (>7.5 hours) | 1,367 | 43 | 4.34 | 1.32 | (0.75, 2.35) | 1.25 | (0.70, 2.21) | 0.95 | (0.52, 1.75) | |
| p-value for group difference | 0.402 | 0.471 | 0.665 | |||||||
| Linear: Per unit SD increase (100 minutes) | 1.11 | (0.90, 1.39) | 1.09 | (0.87, 1.35) | 0.96 | (0.77, 1.21) | ||||
| p-value for linear trend | 0.325 | 0.460 | 0.748 | |||||||
| p-value for non-linearity | 0.456 | 0.427 | 0.424 | |||||||
Model 1 adjusted for age and race and ethnicity;
Model 2 adjusted for model 1 covariates + education, smoking status, history of cancer;
Model 3 adjusted for model 2 covariates + BMI.
Models additionally adjusted for time spent in sitting bouts <30 minutes
d=day; DVT = deep vein thrombosis; HR = hazard ratio; min = minutes; PE = pulmonary embolism; Q = quartile.
In secondary analyses of mean sitting bout duration in relation to risk of incident PE, longer mean sitting bout durations were associated with greater incident PE risk in primary confounder-adjusted models (model 2), with mean sitting bout durations of >15 minutes associated with more than a 2-fold greater risk of incident PE than ≤9 minutes (Q4 vs. Q1 HR = 2.22; 95% CI: 1.28, 3.84; p-group difference = 0.003) (Table 3). In continuous models, every 1-SD increase in mean sitting bout duration was associated with a 22% greater risk of incident PE (HR=1.22; 95% CI: 1.08, 1.38; p-trend=0.002). Results from mediation analyses suggested that BMI mediated approximately 20% of this association, but evidence of this mediation was not statistically significant (p=0.08). After further adjustment for BMI as a confounder (model 3), point estimates shifted slightly toward the null, and associations remained statistically significant (Table 3). In analyses with incident PE as the outcome of interest, although p-values were statistically significant for continuous linear trends, there was also significant evidence of non-linearity (p-values for non-linearity = 0.03 and 0.05 before and after adjustment for BMI, respectively) (Table 3).
In secondary analyses of continuous linear mean sitting bout duration in relation to incident DVT risk, every 1-SD (5 minute) increase in mean sitting bout duration was associated with a 19% greater risk of DVT in primary confounder-adjusted models (HR=1.19; 95% CI: 1.05, 1.35; p-trend=0.008); however, after further adjustment for BMI, the point estimate shifted toward the null and was no longer statistically significant (Table 4).
Discussion
In this cohort study of older ambulatory women with an average age of nearly 80, we found no evidence of an association between total sitting time nor time spent in prolonged sitting bouts in relation to risk of any VTE, PE, or DVT. However, longer mean sitting bout durations were associated with a greater risk of any incident VTE, and separately, with the risk of incident PE and DVT. Approximately 30% of the association appeared to be mediated by higher BMI associated with longer sitting duration. Estimates were not meaningfully different after adjustment for MVPA, suggesting that this association is independent of MVPA. Furthermore, there was no evidence of interaction by MVPA, cancer history, or age.
To our knowledge, this is the first study of accelerometer-measured SB in relation to incident VTE risk. A wide-range of self-reported SB have been evaluated in prior studies, including self-reported sitting or immobility[6,29], television viewing[4,5,9,30], sedentary occupation[31], and immobility-related risk factors[32] in relation to VTE risk. Broadly, self-reported SB are associated with VTE risk in most[4–6,29–32] but not all[9] studies. In general, SB exposures employed across studies varied widely in definition and categorization, making it difficult to directly compare results across studies or to our study’s three accelerometer-measured SB exposures.
In this study, mean sitting bout duration, which better accounts for breaks from sitting, was more strongly associated with risk than were measures that accounted for SB in a cumulative manner (i.e., sitting time and time spent in prolonged sitting bouts). Together, these findings may suggest the importance of recommending breaks from sitting bouts, because more frequent breaks would result in a shorter mean sitting bout duration, given the same total sitting time.
This discrepancy between mean sitting bout duration and total sitting time differs somewhat from comparable evalations of accelerometer-measured SB and other cardiovascular events. In prior WHI OPACH studies, both higher sitting time and longer mean sitting bout duration were associated with greater CVD risk[16] and higher all-cause and cardiovascular disease mortality risk,[33] but there was no evidence of an association between these exposures and mild cognitive impairment or probable dementia.[19] When taken together, these results suggest that clinical advice to older women may need to include lesser overall sitting time to reduce CVD risk and morality,[33] but also shorter bouts of sitting to most effectively prevent VTE.
The effect size associated with mean sitting bout duration was similar across our three related outcomes of any VTE, PE, and DVT. However, we observed a linear dose-response relationship for associations of mean sitting bout duration with any VTE and with DVT outcomes. In contrast, results for PE suggested a threshold effect, with an approximately 2-fold greater risk of incident PE associated with mean sitting bout durations over 12 minutes. If confirmed, targeting mean sitting durations above this threshold may offer a useful clinical strategy for PE risk reduction.
Elevated BMI and obesity are strong risk factors for VTE.[34–38] Furthermore, higher SB has been associated with increases in BMI over time[11], and the association between SB and BMI is one mechanism by which SB may be associated with VTE. In causal mediation analyses, BMI mediated about one-third of the association between mean sitting bout duration and VTE risk, suggesting that while this is likely an important mediator, there are also other likely mechanisms through which mean sitting bout duration is associated with VTE risk. This proportion mediated by BMI is similar to that in a Tromsø-based analysis of repeated self-reported physical activity and incident VTE risk (14–36% mediated by BMI)[40], which suggests that BMI is likely an important mediator of the physical activity and VTE association but that it also does not explain the entirety of this association. Indeed, multiple linking mechanisms between SB and VTE risk have been proposed, including increased systemic inflammation, increased plasma viscosity and platelet aggregation, poorer endothelial function, and adverse changes in other intermediate cardiovascular risk factors such as hemostatic factors (e.g. fibrinogen)[39], blood pressure, and cholesterol.[30] While we also conducted analyses with BMI included as a covariate, and estimates shifted toward the null and were no longer statistically significant; we believe that its inclusion as a covariate likely over-adjusts for BMI, given its complicated role as a likely confounder and mediator. Studies with repeated measures of both BMI and accelerometry would be needed to clarify this issue definitively.
The association between mean sitting bout duration and VTE risk appeared to be independent of MVPA. Importantly, SB is a distinct behavior from physical inactivity (which describes insufficient amounts of MVPA), and is defined as sitting or reclining-based waking behavior with a low energy expenditure ≤1.5 metabolic equivalents.[41]
Strengths and Limitations
Our study has numerous strengths. To our knowledge, this is the first prospective study of accelerometer-measured SB in relation to incident VTE risk; accelerometer-based measurement of SB may be strongly preferrable to self-report due to difficulties in accurate participant recall of both accumulation and patterns of SB. Furthermore, the WHI OPACH study is a population of older women with substantial racial and ethnic diversity. OPACH participants were highly adherent; 94.4% of participants who returned accelerometers with usable data had adherent wear (≥4 days of ≥10 hours/day of wear).[33] VTE events were centrally adjudicated, increasing validity of the outcome. In addition, the WHI and OPACH studies collected rich health information that allowed for adjustment of possible confounders and the evaluation of potential mediation by BMI and effect modification by multiple factors.
As a limitation of our study, the hip-worn ActiGraph did not measure posture; however, the CHAP algorithm, which was used to estimate sitting time using algorithms developed with the thigh-worn activPAL has been shown to have high levels of accuracy.[21] In addition, accelerometer-wear over a 7-day period may not always capture usual SB. Due to the lack of prior research that evaluate accelerometer-based measures of SB in relation to VTE risk, it is unclear what characteristics of SB (e.g. bout duration, total SB time) most strongly contribute to VTE risk. We included three SB exposures of primary interest in our study, which increases the number of tests conducted in our study; however, this evaluation of these three exposures contributes to our understanding of specific SB characteristics in relation to VTE risk. As with all observational studies, there is the possibility for residual confounding; however, we attempted to mitigate the role of possible confounders by adjusting for covariates defined using the rich data collected in the WHI. Furthermore, although our study is useful for understanding this association in a targeted, specific population of older women, replication among other populations is needed to improve our understanding of objectively measured SB in relation to VTE risk on a larger population-level.
Conclusions
In conclusion, total sitting time and total time spent in prolonged sitting bouts were not associated with risk of any VTE, PE, or DVT in this population of older ambulatory women. In contrast, longer mean sitting bout duration was associated with greater VTE risk, with significant mediation by BMI. Our results suggest that reducing sedentary bout length in older women may reduce VTE risk.
Supplementary Material
Acknowledgements
We thank the Women’s Health Initiative (WHI) participants, staff, and investigators. The short list of WHI investigators can be found at the following site: https://www-whi-org.s3.us-west-2.amazonaws.com/wp-content/uploads/WHI-Investigator-Short-List.pdf. The full list of WHI investigators can be found at the following site: https://s3-us-west-2.amazonaws.com/www-whi-org/wp-content/uploads/WHI-Investigator-Long-List.pdf.
Funding
The present study was supported by the National Heart, Lung, and Blood Institute (grants K01HL139997 to L.B. Harrington and R01HL105065 and R01HL153462 to A.Z. LaCroix) and the National Institute on Aging (K24AG065525 to K. Mukamal and K99AG082863 to S. Nguyen). The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, 75N92021D00005.
Footnotes
Conflicts of Interest
LBH, KLCH, SN, JB, MJL, CBE, MAA, RBW, JEM, MKJ, CK, IL, KJM, AZL report no conflicts of interest. GAW previously served as a consultant for Google, LLC (Mountain View, CA) and currently serves as a consultant for the Health Effects Institute (Boston, MA).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Heit JA, Spencer FA, White RH. The Epidemiology of Venous Thromboembolism. J Thromb Thrombolysis 2016; 41: 3–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lutsey PL, Zakai NA. Epidemiology and prevention of venous thromboembolism. Nature reviews Cardiology 2023; 20: 248–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Previtali E, Bucciarelli P, Passamonti SM, Martinelli I. Risk factors for venous and arterial thrombosis. Blood Transfus 2011; 9: 120–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kubota Y, Cushman M, Zakai N, Rosamond WD, Folsom AR. TV viewing and incident venous thromboembolism: the Atherosclerotic Risk in Communities Study. J Thromb Thrombolysis 2018; . [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Shirakawa T, Iso H, Yamagishi K, Yatsuya H, Tanabe N, Ikehara S, Ukawa S, Tamakoshi A. Watching Television and Risk of Mortality From Pulmonary Embolism Among Japanese Men and Women: The JACC Study (Japan Collaborative Cohort). Circulation 2016; 134: 355–7. [DOI] [PubMed] [Google Scholar]
- 6.Kabrhel C, Varraso R, Goldhaber SZ, Rimm E, Camargo CA. Physical Inactivity and Idiopathic Pulmonary Embolism in Women: Prospective Study. BMJ 2011; 343: d3867. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Engbers MJ, Blom JW, Cushman M, Rosendaal FR, Van Hylckama Vlieg A. Functional Impairment and Risk of Venous Thrombosis in Older Adults. J American Geriatrics Society 2017; 65: 2003–8. [DOI] [PubMed] [Google Scholar]
- 8.Chandra D, Parisini E, Mozaffarian D. Meta-analysis: travel and risk for venous thromboembolism. Ann Intern Med 2009; 151: 180–90. [DOI] [PubMed] [Google Scholar]
- 9.Munger JA, Koh I, Howard G, Howard V, Plante TB, Wilkinson K, Cushman M, Zakai NA. Television viewing, physical activity and venous thromboembolism risk: The Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study. J Thromb Haemost 2021; . [DOI] [PubMed] [Google Scholar]
- 10.Prince SA, Cardilli L, Reed JL, Saunders TJ, Kite C, Douillette K, Fournier K, Buckley JP. A comparison of self-reported and device measured sedentary behaviour in adults: a systematic review and meta-analysis. Int J Behav Nutr Phys Act 2020; 17: 31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mitchell JA, Bottai M, Park Y, Marshall SJ, Moore SC, Matthews CE. A Prospective Study of Sedentary Behavior and Changes in the Body Mass Index Distribution. Medicine & Science in Sports & Exercise 2014; 46: 2244–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Women’s Health Initiative Study Group. Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group. Control Clin Trials 1998; 19: 61–109. [DOI] [PubMed] [Google Scholar]
- 13.LaCroix AZ, Rillamas-Sun E, Buchner D, Evenson KR, Di C, Lee IM, Marshall S, LaMonte MJ, Hunt J, Tinker LF, Stefanick M, Lewis CE, Bellettiere J, Herring AH. The Objective Physical Activity and Cardiovascular Disease Health in Older Women (OPACH) Study. BMC public health 2017; 17: 192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.LaCroix AZ, Bellettiere J, Rillamas-Sun E, Di C, Evenson KR, Lewis CE, Buchner DM, Stefanick ML, Lee IM, Rosenberg DE, LaMonte MJ. Association of Light Physical Activity Measured by Accelerometry and Incidence of Coronary Heart Disease and Cardiovascular Disease in Older Women. JAMA network open 2019; 2: e190419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.LaMonte MJ, Manson JE, Anderson GL, Baker LD, Bea JW, Eaton CB, Follis S, Hayden KM, Kooperberg C, LaCroix AZ, Limacher MC, Neuhouser ML, Odegaard A, Perez MV, Prentice RL, Reiner AP, Stefanick ML, Van Horn L, Wells GL, Whitsel EA, et al. Contributions of the Women’s Health Initiative to Cardiovascular Research. Journal of the American College of Cardiology 2022; 80: 256–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bellettiere J, LaMonte MJ, Evenson KR, Rillamas-Sun E, Kerr J, Lee IM, Di C, Rosenberg DE, Stefanick M, Buchner DM, Hovell MF, LaCroix AZ. Sedentary behavior and cardiovascular disease in older women: The Objective Physical Activity and Cardiovascular Health (OPACH) Study. Circulation 2019; 139: 1036–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Migueles JH, Cadenas-Sanchez C, Ekelund U, Delisle Nyström C, Mora-Gonzalez J, Löf M, Labayen I, Ruiz JR, Ortega FB. Accelerometer Data Collection and Processing Criteria to Assess Physical Activity and Other Outcomes: A Systematic Review and Practical Considerations. Sports Med 2017; 47: 1821–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.LaMonte MJ, Lewis CE, Buchner DM, Evenson KR, Rillamas-Sun E, Di C, Lee IM, Bellettiere J, Stefanick ML, Eaton CB, Howard BV, Bird C, LaCroix AZ. Both Light Intensity and Moderate-to-Vigorous Physical Activity Measured by Accelerometry Are Favorably Associated With Cardiometabolic Risk Factors in Older Women: The Objective Physical Activity and Cardiovascular Health (OPACH) Study. J Am Heart Assoc 2017; 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nguyen S, LaCroix AZ, Hayden KM, Di C, Palta P, Stefanick ML, Manson JE, Rapp SR, LaMonte MJ, Bellettiere J. Accelerometer-measured physical activity and sitting with incident mild cognitive impairment or probable dementia among older women. Alzheimer’s & Dementia 2023; 19: 3041–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of Accelerometer Wear and Nonwear Time Classification Algorithm. Medicine & Science in Sports & Exercise 2011; 43: 357–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Greenwood-Hickman MA, Nakandala S, Jankowska MM, Rosenberg DE, Tuz-Zahra F, Bellettiere J, Carlson J, Hibbing PR, Zou J, Lacroix AZ, Kumar A, Natarajan L. The CNN Hip Accelerometer Posture (CHAP) Method for Classifying Sitting Patterns from Hip Accelerometers: A Validation Study. Medicine & Science in Sports & Exercise 2021; 53: 2445–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Burwen DR, Wu C, Cirillo D, Rossouw JE, Margolis KL, Limacher M, Wallace R, Allison M, Eaton CB, Safford M, Freiberg M. Venous thromboembolism incidence, recurrence, and mortality based on Women’s Health Initiative data and Medicare claims. Thromb Res 2017; 150: 78–85. [DOI] [PubMed] [Google Scholar]
- 23.Curb JD, McTiernan A, Heckbert SR, Kooperberg C, Stanford J, Nevitt M, Johnson KC, Proulx-Burns L, Pastore L, Criqui M, Daugherty S. Outcomes ascertainment and adjudication methods in the Women’s Health Initiative. Ann Epidemiol 2003; 13: S122–8. [DOI] [PubMed] [Google Scholar]
- 24.Willett W, Stampfer MJ. Total Energy Intake: Implications for Epidemiologic Analyses. American Journal of Epidemiology 1986; 124: 17–27. [DOI] [PubMed] [Google Scholar]
- 25.Govindarajulu US, Spiegelman D, Thurston SW, Ganguli B, Eisen EA. Comparing smoothing techniques in Cox models for exposure–response relationships. Statistics in Medicine 2007; 26: 3735–52. [DOI] [PubMed] [Google Scholar]
- 26.Valeri L, VanderWeele TJ. Mediation analysis allowing for exposure–mediator interactions and causal interpretation: Theoretical assumptions and implementation with SAS and SPSS macros. Psychological Methods 2013; 18: 137–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.StataCorp. Impute missing values using chained equations. Stata Multiple-Imputation Reference Manual, Release 13. 2013. [Google Scholar]
- 28.StataCorp. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC; [Google Scholar]
- 29.Flinterman LE, van Hylckama Vlieg A, Rosendaal FR, Cannegieter SC. Body height, mobility, and risk of first and recurrent venous thrombosis. J Thromb Haemost 2015; 13: 548–54. [DOI] [PubMed] [Google Scholar]
- 30.Kunutsor SK, Dey RS, Laukkanen JA. Television viewing and venous thrombo-embolism: a systematic review and meta-analysis. European journal of preventive cardiology 2022; 29: e239–42. [DOI] [PubMed] [Google Scholar]
- 31.Suadicani P, Hannerz H, Bach E, Gyntelberg F. Jobs encompassing prolonged sitting in cramped positions and risk of venous thromboembolism: cohort study. JRSM Short Reports 2012; 3: 1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Engbers MJ, Blom JW, Cushman M, Rosendaal FR, van Hylckama Vlieg A. The contribution of immobility risk factors to the incidence of venous thrombosis in an older population. J Thromb Haemost 2014; 12: 290–6. [DOI] [PubMed] [Google Scholar]
- 33.Nguyen S, Bellettiere J, Anuskiewicz B, Di C, Carlson J, Natarajan L, LaMonte MJ, LaCroix AZ. Prospective Associations of Accelerometer-Measured Machine-Learned Sedentary Behavior With Death Among Older Women: The OPACH Study. JAHA 2024; 13: e031156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Cushman M, O’Meara ES, Heckbert SR, Zakai NA, Rosamond W, Folsom AR. Body size measures, hemostatic and inflammatory markers and risk of venous thrombosis: The Longitudinal Investigation of Thromboembolism Etiology. Thromb Res 2016; 144: 127–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Borch KH, Brækkan SK, Mathiesen EB, Njølstad I, Wilsgaard T, Størmer J, Hansen J-B. Anthropometric Measures of Obesity and Risk of Venous Thromboembolism: The Tromsø Study. ATVB 2010; 30: 121–7. [DOI] [PubMed] [Google Scholar]
- 36.Severinsen MT, Kristensen SR, Johnsen SP, Dethlefsen C, Tjønneland A, Overvad K. Anthropometry, Body Fat, and Venous Thromboembolism: A Danish Follow-Up Study. Circulation 2009; 120: 1850–7. [DOI] [PubMed] [Google Scholar]
- 37.Yuan S, Bruzelius M, Xiong Y, Håkansson N, Åkesson A, Larsson SC. Overall and abdominal obesity in relation to venous thromboembolism. Journal of Thrombosis and Haemostasis 2021; 19: 460–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Holst AG, Jensen G, Prescott E. Risk factors for venous thromboembolism: results from the Copenhagen City Heart Study. Circulation 2010; 121: 1896–903. [DOI] [PubMed] [Google Scholar]
- 39.Howard BJ, Fraser SF, Sethi P, Cerin E, Hamilton MT, Owen N, Dunstan DW, Kingwell BA. Impact on hemostatic parameters of interrupting sitting with intermittent activity. Med Sci Sports Exerc 2013; 45: 1285–91. [DOI] [PubMed] [Google Scholar]
- 40.Evensen LH, Braekkan SK, Hansen JB. Regular Physical Activity and Risk of Venous Thromboembolism. Semin Thromb Hemost 2018; . [DOI] [PubMed] [Google Scholar]
- 41.Sedentary Behaviour Research Networ. Letter to the Editor: Standardized use of the terms “sedentary” and “sedentary behaviours.” Appl Physiol Nutr Metab 2012; 37: 540–2. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.