Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Oct 1.
Published before final editing as: J Aging Phys Act. 2026 Feb 2:1–13. doi: 10.1123/japa.2024-0233

Associations of Accelerometer-Measured Physical Activity and Sedentary Behavior With Incident Clinical Osteoporotic Fractures Among Older Women: The Objective Physical Activity and Cardiovascular Health Study

Steve Nguyen 1, Carolyn J Crandall 2, John Bellettiere 1, Michael J LaMonte 3, Andrea Z LaCroix 1
PMCID: PMC13626382  NIHMSID: NIHMS2207348  PMID: 41633354

Abstract

Background/Objectives:

Women 65 and older experience nearly three-fourths of the 2 million osteoporotic fractures annually in the United States. We investigated associations of accelerometer-measured total physical activity, light physical activity, moderate-to-vigorous physical activity (MVPA), steps, total sitting time, and mean sitting bout duration with incident osteoporotic fracture.

Methods:

Women from the Objective Physical Activity and Cardiovascular Health cohort (N = 6,122; mean age: 78.7; SD: 6.7) without prior hip fracture wore the ActiGraph GT3X+ up to 7 days and were followed up to 8 years for incident osteoporotic fractures (n = 982). Cox regression models estimated hazard ratios and 95% confidence intervals (CI) additionally containing age, race/ethnicity, education, alcohol, smoking, height, weight, falls, physical functioning, diabetes, thiazide use, and anti-osteoporosis medication use.

Results:

Multivariable HRs (95% CI) were 1 (ref), 1.10 [0.93, 1.31], 0.93 [0.77, 1.13], and 0.87 [0.70, 1.07] across MVPA minutes/quartiles (p-trend = .03) and 1 (ref), 1.10 [0.90, 1.33], 1.21 [1.00, 1.47], and 1.27 [1.04, 1.57] across total sitting time quartiles (p-trend = .02). The hazard ratios (95% CI) for a one-SD increment in MVPA (33 min/day) and total sitting time (103 min/day) were 0.92 [0.85, 0.99] and 1.09 [1.01, 1.18], respectively. The hazard ratios (95% CI) for MVPA and osteoporotic fracture was 0.82 [0.73, 0.93] among women not using anti-osteoporosis medication and 1.06 [0.90, 1.24] among women using anti-osteoporosis medication (p-interaction = .02).

Conclusions:

More MVPA and less total sitting time, but not other accelerometer measures, were associated with lower osteoporotic fracture risk.

Significance/Implications:

These data support interventions and physical activity guidelines aimed at promoting moderate- or vigorous-intensity movement and decreased sitting for osteoporotic fracture prevention.

Keywords: hip fracture, epidemiology, women’s health

Women aged 65 years or older experience 74% of the more than 2 million osteoporotic fractures annually in the United States, accounting for 89% of the $17 billion in total costs (Ensrud, 2013). Hip fractures are strongly associated with increased morbidity and mortality and account for 14% of fractures but 72% of total fracture-related health costs (Burge et al., 2007; Katsoulis et al., 2017; Schnell et al., 2010). Although hip and osteoporotic fracture rates declined from 2002 to 2013, rates plateaued from 2015 to 2017, suggesting little progress in recent years despite the introduction and use of pharmacologic treatments for osteoporosis (Lewiecki et al., 2018, 2020). Given the growing aging population and the accompanying substantial burden of fracture, additional epidemiologic evidence is needed to inform nonpharmacologic fracture prevention strategies among older women.

The 2018 Physical Activity Guidelines for Americans continued to recommend as a broad public health guideline achieving ≥150 min/week of moderate-to-vigorous physical activity (MVPA) and muscle-strengthening activity twice/week and identified a need for future research focused on the associations of light-intensity physical activity (LPA) with aging-related outcomes (2018 Physical Activity Guidelines Advisory Committee, 2018). Importantly, the 2018 Physical Activity Guidelines Advisory Committee noted limited evidence on the dose–response association of MVPA with fracture risk (2018 Physical Activity Guidelines Advisory Committee, 2018). Sedentary behavior (SB), defined as sitting or reclining with low energy expenditure (<1.5 metabolic equivalents [MET]) during waking behavior, is recognized as a risk factor for cardiovascular disease and mortality independent of physical activity (PA) levels (2018 Physical Activity Guidelines Advisory Committee, 2018). Higher amounts of self-reported PA are associated with a lower risk of fracture, whereas higher amounts of self-reported SB are associated with a modestly higher risk of fracture (LaMonte, Wactawski-Wende, et al., 2019). However, a large proportion of movement engaged in by older adults is associated with daily life tasks that, along with SB, are not well captured by PA questionnaires (Colbert et al., 2014). Few studies have examined the prospective associations of accelerometer measures of PA and SB with fracture risk among older women. Furthermore, few studies have examined specific PA measures such as daily steps and SB accumulation patterns, which are not as well captured by questionnaires, with fracture risk.

In the present study, we examined (a) whether higher accelerometer-measured LPA and MVPA and lower SB were prospectively associated with lower fracture risk; (b) the dose–response characteristics of these associations; and (c) the effect modification by age, race and ethnicity, body mass index (BMI), history of falls, thiazide use, and anti-osteoporosis medication use.

Methods

Study Population

The data supporting the present study findings are available upon reasonable request from the Women’s Health Initiative (WHI) Clinical Coordinating Center in accordance with WHI Publications and Presentation Policies (https://sp.whi.org/researchers/data/WHIStudies/StudySites/AS286/Pages/home.aspx).

The Objective Physical Activity and Cardiovascular Health (OPACH) study is an ancillary study of the WHI, which collected accelerometry from 6,489 ambulatory community-living women aged 63–99 at baseline between May 2012 and April 2014. Women were followed through February 28, 2020, for incident fractures (LaCroix et al., 2017). Details on the OPACH study’s protocols are published (LaCroix et al., 2017). In this report, data were excluded from 363 women who returned accelerometers that did not meet the criteria for adherent accelerometer wear (at least 4 days of at least 10 hr/day of wear) and four women who had outlier MVPA and/or steps values, yielding an analytic sample of 6,122 women. In analyses of SB, we additionally excluded 267 (4.4%) women who returned accelerometers containing data to which the Convolutional Neural Network Hip Accelerometer Posture (CHAP) algorithm for classifying SB could not be applied (Greenwood-Hickman et al., 2021). Prior to recruitment, approval of all study protocols was obtained from an institutional research ethics committee, and all women provided written informed consent. This study complies with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (STROBE, n.d.).

Incident Clinical Fracture Ascertainment

In the present study, we examined 4 clinical fracture endpoints: total fracture, osteoporotic fracture, major osteoporotic fracture, and hip fracture. Total fracture included the following sites: hip, vertebral, pelvis, upper and lower leg, upper and lower arm, wrist, coccyx, knee, ankle, foot, clavicle, scapula, elbow, and hand. Osteoporotic fractures consisted of a subset of sites frequently attributable to osteoporosis among women: hip, vertebral, pelvis, upper and lower leg, upper and lower arm, and wrist (Crandall et al., 2014; Warriner et al., 2011). Major osteoporotic fractures consisted of a subset of sites: hip, vertebral, forearm, and humerus (Crandall et al., 2014; Warriner et al., 2011). Hip fractures are a primary outcome in WHI and were adjudicated by physician review of radiology reports, hospital discharge summaries, operative reports, and clinic and progress reports. Fractures at other sites were ascertained by self-report. The mean agreement between self-report and medical records for fractures in the WHI was 71%, varying by site, with high agreement for fractures of the hip (78%), lower arm or wrist (81%), and upper arm (82%), but lower agreement on fractures of vertebrae (51%) and upper leg (28%) (Chen et al., 2004).

Accelerometer Measures of Physical Activity and Sedentary Behavior

During OPACH baseline, women wore the ActiGraph GT3X+ (ActiGraph LLC) over the right hip 24 hr/day for 7 consecutive days excluding water submersion (e.g., bathing and swimming). Periods of accelerometer nonwear were identified for removal using the Choi algorithm (Choi et al., 2011; LaCroix et al., 2017). Sleep time was removed using reported in-bed and out-of-bed times from sleep diaries that women filled out concurrently with accelerometer wear. Missing sleep time was imputed using individual average sleep time if available; otherwise, the entire OPACH cohort average was used.

ActiLife (version 6) software was used to aggregate the 30-Hz data into 15-s epochs, and we applied OPACH Calibration Study vector magnitude-based cutpoints to classify MVPA (>518 counts/15 s) and LPA (19–518 counts/15 s) in minutes per day (Evenson et al., 2015). To classify steps, the GT3X+ data were first processed using the default normal frequency filter. Steps/day was calculated by dividing the total number of steps for each 15-s epoch, determined using the proprietary ActiLife algorithm, by the number of adherent wear days.

To classify SB, we applied the validated CHAP algorithm to estimate total sitting time (minutes/day) and mean sitting bout duration (minutes/bout; Greenwood-Hickman et al., 2021). The CHAP algorithm accurately detects postural transitions from GT3X+ data (sitting vs. standing or walking), whereas traditional accelerometer cutpoint methods define SB below a threshold of low intensity movement, which can misclassify behaviors like standing as SB (Greenwood-Hickman et al., 2021). The CHAP algorithm was developed among 709 older adults (mean age = 77 ± 6.5 years, 59% female) in the Adult Changes in Thought Study and had high agreement with the activPAL micro3 inclinometer at classifying minute-level sitting versus nonsitting (sensitivity = 97.1%, specificity = 88.6%, balanced accuracy = 92.9%), outperforming the commonly used ActiGraph cutpoint of < 100 counts/minute on the vertical axis (sensitivity = 88.2%, specificity = 59.7%, balanced accuracy = 73.9%; Greenwood-Hickman et al., 2021). The CHAP algorithm has been successfully applied to published studies in OPACH (LaMonte et al., 2024; Nguyen et al., 2023, 2024). We applied the residuals method to account for awake-wear time for all PA and SB measures (Willett & Stampfer, 1986).

Covariates

At OPACH baseline, questionnaires ascertained age, self-identified race/ethnicity (Black, White, or Hispanic or Latina), education (≤high school equivalent, some college, college graduate), alcohol consumption in the past 3 months prior to baseline (nondrinker, <1 drink/week, ≥1 drink/week, unknown), current smoking status (no, yes), and number of falls in the past 12 months (0–1, ≥2). Participants reported anti-diabetic medication use and physician-diagnosed diabetes mellitus status at OPACH baseline. Physical function was ascertained using the RAND-36 questionnaire; scores ranged between 0 and 100 with higher scores indicating higher functioning (Hays et al., 1993). Trained study staff measured height and weight with a tape measure and calibrated bathroom scale, respectively, with BMI calculated as weight in kilograms divided by the square of height in meters. For women with missing OPACH baseline height, weight, or BMI, their most recent measurements collected in the WHI were used instead. Women reported thiazide and anti-osteoporosis medication use via a mailed medication inventory requesting information transcribed from the product label supplemented with telephone calls when necessary (bisphosphonates, calcitonin, parathyroid hormone and derivatives, receptor activator of nuclear factor kappa-β ligand inhibitors, the selective estrogen receptor modulator raloxifene, and estrogen). The history of total fracture by OPACH baseline was defined as the fracture of any site listed in total fracture that occurred by OPACH baseline.

Statistical Analysis

R 4.3.2 (https://www.r-project.org/) was used for all statistical analyses. To describe participant characteristics, we calculated means and SDs for continuous variables or counts and proportions for categorical variables. We compared these values across MVPA (minutes/day) quartiles using F tests for continuous variables or chi-square tests for categorical variables. We additionally calculated Pearson correlations between all accelerometer variables.

We used the R survival package to fit multivariable Cox proportional hazards regression models and estimated hazard ratios (HRs) and 95% confidence intervals (CI) for fracture endpoints. We specified separate models for each PA or SB variable. We initially modeled PA or SB variables in quartiles to examine potential nonlinear associations with fracture endpoints and to allow for straightforward interpretations of results. We defined follow-up time to the first incident fracture as the number of days from OPACH baseline to the first occurrence of a fracture, date of the last obtained annual medical update, or end of follow-up, whichever occurred first. Participants who did not experience a fracture or who were lost to follow-up were right censored. We assumed noninformative censoring, meaning that the probability of being censored is independent of fracture risk conditional on model covariates. We tested the Schoenfeld residuals to assess the proportional hazards assumption; no violations were noted.

Model 1, the minimally adjusted model, contained PA or SB, age, and race and ethnicity as published OPACH studies showed that PA and SB vary by race and/or ethnicity (Bellettiere et al., 2019; LaCroix et al., 2019). Model 2, the confounder-adjusted model, additionally contained potential confounders identified in the literature to be associated with PA, SB, and fracture risk: educational attainment, alcohol consumption, smoking status, height, weight, physical function, diabetes, falls history, thiazide use, anti-osteoporosis medication use, and history of total fracture (Crandall et al., 2021; LaMonte, Wactawski-Wende, et al., 2019). To assess changes in effect estimates due to potential confounders, we fit models that incrementally contained Model 2 covariates. We calculated p-trend for PA and SB variables as the p value from a Wald test of the PA or SB variable modeled in continuous functional form (Therneau et al., 2024; Therneau & Grambsch, 2000). In sensitivity analyses, we repeated the quartile models and incrementally added physical function, falls history, anti-osteoporotic medication use, and history of total fracture as model covariates to assess changes in effect estimates for PA and SB with fracture risk.

We examined whether the associations of PA and SB with fracture were linear using restricted cubic splines with knots at the 10th, 50th, and 90th percentiles with the R rms package. We computed a Wald test using the anova.rms() function to evaluate whether the model coefficients for the nonlinear spline components differed significantly from zero, with a p value < .05, suggesting a nonlinear association between PA or SB with fracture.

To evaluate the consistency of associations, we examined effect modification a priori by known fracture risk factors that could potentially mask the relationships between PA and SB with fracture risk: age, race and ethnicity, BMI, falls history in the past 12 months before baseline, RAND-36 physical function, diabetes, thiazide use, and anti-osteoporosis medication, which is prescribed for individuals with high fracture risk. We carried out likelihood ratio tests comparing models with cross-product terms between continuous PA or SB measures and each effect modification variable to a reduced model without the cross-product term but containing these variables (Harrell, 2015; Harrell et al., 2017). We specified separate models for each interaction between PA or SB and the variables mentioned. We subsequently carried out stratified analyses according to categories of these variables.

Missing covariate data were imputed using multiple imputation by chained equations (MICE) with the R mice package, including all variables for five iterations and 100 imputations (van Buuren, 2012; van Buuren & Groothuis-Oudshoorn, 2011). Multiple imputation by chained equations accommodates multiple variable types (e.g., continuous and categorical), accounts for relationships between covariates, and is widely used in epidemiological studies (Harel et al., 2018). We fit separate Cox models with each of the imputed data sets and pooled the results using Rubin’s rules to calculate a pooled HR and corresponding standard error (Rubin, 1987; van Buuren, 2012).

We reported nominal p values for tests of effect modification—a total of 216 tests (six PA and SB variables, four study endpoints, and nine stratification variables) were carried out. With a significance threshold of p < .05, 11 (0.05 × 216) of these tests are expected to have nominal p values of less than .05 due to chance, so results should be interpreted with caution.

Results

Study Population Characteristics

At OPACH baseline, women were 78.7 years on average, 33.4% were Black, 16.9% were Hispanic or Latina, and 49.7% were White. Women averaged approximately 336 min/day of total PA, consisting of 287 min/day of LPA and 50 min/day of MVPA, and 622 min/day of total sitting time, and had a mean sitting bout duration of 12.8 min/bout (Table 1). Compared with women with total PA in the lowest quartile, those with total PA in the highest quartile were younger, more likely to be Black or Hispanic, more likely to be a college graduate, less likely to smoke, more likely to consume one or more alcoholic drinks per week, had a lower BMI, had higher RAND-36 physical functioning scores, were less likely to have reported falls in the 12 months before baseline, were less likely to use thiazides, were as likely to report anti-osteoporosis medication use, and were less likely to have a history of total fracture (Table 1). Pearson correlations for total PA with LPA, MVPA, steps/day, total sitting time, and mean sitting bout duration were .94, .67, .64, −.75, and −.69, respectively (Supplementary Figure S1 [available online]).

Table 1.

Mean (SD) or Count (%) of OPACH (n = 6,122) Baseline (2012–2014) Sociodemographic and Health-Related Characteristics Across Quartiles of MVPA

Total MVPA (min/day)a quartiles p
Q1 (<25.8) Q2 (25.8–43.9) Q3 (43.9–68.0) Q4 (>68.0)
Age (years) 78.7 (6.7) 81.0 (6.4) 79.2 (6.6) 78.2 (6.7) 76.5 (6.3) <.001
Race/ethnicity <.001
 White 3,044 (49.7) 949 (62) 787 (51.4) 712 (46.5) 596 (38.9)
 Black 2,046 (33.4) 435 (28.4) 524 (34.2) 528 (34.5) 559 (36.5)
 Hispanic or Latina 1,032 (16.9) 147 (9.6) 219 (14.3) 290 (19) 376 (24.6)
Highest education level .02
 High school or less 1,237 (20.3) 290 (19.1) 310 (20.5) 314 (20.6) 323 (21.1)
 Some college 2,347 (38.6) 635 (41.7) 605 (40.0) 549 (36.1) 558 (36.5)
 College graduate 2,497 (41.1) 596 (39.2) 596 (39.4) 658 (43.3) 647 (42.3)
Health behavior/status, n (%)
 Current smoker
 Alcohol in past 3 months <.001
  Nondrinker 2,089 (34.1) 584 (38.1) 546 (35.7) 498 (32.5) 461 (30.1)
  <1 drink/week 1,910 (31.2) 483 (31.5) 479 (31.3) 496 (32.4) 452 (29.5)
  ≥1 drink/week 1,587 (25.9) 303 (19.8) 364 (23.8) 421 (27.5) 499 (32.6)
  Unknown 536 (8.8) 161 (10.5) 141 (9.2) 115 (7.5) 119 (7.8)
 BMI (kg/m2) 28.2 (5.7) 30.2 (6.1) 28.7 (5.7) 27.5 (5.3) 26.4 (5.0) <.001
 Height (in) 62.9 (2.9) 62.9 (3.0) 62.9 (2.8) 62.9 (2.9) 62.8 (2.8) .85
 Weight (lb) 158.6 (34.3) 170.0 (37.3) 161.8 (34.9) 154.6 (31.5) 148.0 (28.8) <.001
 RAND-36 physical functioning score
 Falls in past 12 months .01
  0–1 5,404 (88.3) 1,326 (86.7) 1,334 (87.2) 1,370 (89.5) 1,374 (89.7)
  ≥2 717 (11.7) 204 (13.3) 196 (12.8) 160 (10.5) 157 (10.3)
 Diabetes 1,246 (20.4) 396 (25.9) 317 (20.7) 329 (21.5) 204 (13.3) <.001
 Thiazide use 3,937 (64.3) 1,081 (70.6) 1,008 (65.9) 979 (64) 869 (56.8) <.001
 Anti-osteoporosis medication use 1,317 (21.5) 310 (20.2) 326 (21.3) 326 (21.3) 355 (23.2) .25
 History of total fracture 642 (10.5) 195 (12.7) 153 (10) 170 (11.1) 124 (8.1) <.001
Accelerometer measures
 Total PA (min/day)a 337.3 (90.1) 225.3 (38.8) 305.2 (17.2) 364.3 (17.1) 454.4 (48.3) <.001
 LPA (min/day)a 287.0 (72.1) 200.2 (35.3) 265.2 (24.9) 309.8 (29.2) 372.7 (47.0) <.001
 MVPA (min/day) a 50.3 (33.3) 25.1 (16.1) 40.0 (21.1) 54.5 (26.3) 81.7 (35.8) <.001
 Steps (n/day)a 3,541 (2,049) 1,973 (1,021) 2,985 (1,354) 3,855 (1,688) 5,352 (2,220) <.001
 Sitting time (min/day)a 621.6 (103.0) 720.3 (61.6) 647.9 (65.9) 598.0 (72.2) 521.5 (89.2) <.001
 MSBD (min)a 12.8 (5.3) 18.2 (6.1) 13.3 (3.4) 11.0 (3.0) 8.9 (2.7)
Open in a new tab

Note. p value: chi-square test for categorical variables and F tests for continuous variables. BMI = body mass index; LPA = light-intensity physical activity; MSBD = mean sitting bout duration; MVPA = moderate-to-vigorous physical activity; PA = physical activity; OPACH = Objective Physical Activity and Cardiovascular Health.

a

All accelerometer measures are adjusted for accelerometer awake wear time using the residuals method.

Associations of PA and SB With Incident Fractures

Over a median follow-up of approximately 6.2 years, there were 1,401 total fractures, of which 982 were osteoporotic fractures, 703 were major osteoporotic fractures, and 205 were hip fractures. Unadjusted rates of total fractures and osteoporotic fractures were generally lower across higher quartiles of total PA and MVPA and were higher across higher quartiles of total sitting time and mean sitting bout duration, with a similar pattern observed for other fracture endpoints (Table 2 and Supplementary Table S1 [available online]). However, this pattern was not observed for LPA or steps/day quartiles (Table 2 and Supplementary Table S1 [available online]). In minimally adjusted analyses, higher quartiles of MVPA were associated with lower risk of total fracture and osteoporotic fractures while higher quartiles of total sitting time were associated with higher risk of total fracture and osteoporotic fractures (Table 2).

Table 2.

Associations of LPA, MVPA, Total Sitting Time, and MSBD With Incident Osteoporotic Fracture in the OPACH Cohort 2012–2020 (n = 6,122)

Total physical activity (min/day) quartiles
Q1 (<273.9) Q2
[273.9, 334.7]		 Q3
[334.7, 395.3]		 Q4
(>395.3)		 HR for 1-SD (90.1)
increment		 p trenda
Total fracture events (rate)b 376 (48.5) 370 (43.7) 341 (39.5) 314 (34.9)
 Model 1c 1 (ref) 0.97 [0.84, 1.12] 0.92 [0.79, 1.07] 0.88 [0.75, 1.03] 0.96 [0.91, 1.01] .139
 Model 2c, 1 (ref) 0.97 [0.84, 1.12] 0.90 [0.77, 1.06] 0.87 [0.73, 1.03] 0.96 [0.90, 1.02] .143
OF events (rate)b 266 (33.2) 259 (29.7) 235 (26.4) 222 (24.0)
 Model 1c 1 (ref) 0.99 [0.84, 1.18] 0.94 [0.79, 1.13] 0.95 [0.79, 1.15] 0.98 [0.92, 1.04] .504
 Model 2c 1 (ref) 0.98 [0.83, 1.17] 0.92 [0.76, 1.1] 0.92 [0.75, 1.12] 0.96 [0.89, 1.04] .301
Hip fracture events (rate)b 58 (6.9) 56 (6.1) 49 (5.2) 42 (4.4)
 Model 1c 1 (ref) 1.12 [0.77, 1.62] 1.13 [0.77, 1.66] 1.22 [0.81, 1.84] 1.09 [0.94, 1.26] .254
 Model 2c 1 (ref) 1.04 [0.71, 1.52] 0.99 [0.66, 1.49] 1.01 [0.64, 1.58] 1.00 [0.85, 1.18] .953
MVPA (min/day) quartiles
Q1
(<25.8)		 Q2
[25.8, 43.9]		 Q3
[43.9, 68.0]		 Q4
(>68.0)		 HR for 1-SD (33.3)
increment		 p-trenda
Total fracture events (rate)b 386 (50.7) 380 (45.5) 317 (36.1) 318 (35.0)
 Model 1c 1 (ref) 0.97 [0.84, 1.12] 0.84 [0.72, 0.97] 0.85 [0.73, 1.00] 0.93 [0.88, 0.99] .018
 Model 2c, 1 (ref) 1.05 [0.91, 1.22] 0.90 [0.77, 1.05] 0.91 [0.77, 1.08] 0.95 [0.89, 1.01] .095
OF events (rate)b 276 (35.0) 274 (31.9) 225 (25.0) 207 (22.0)
 Model 1c 1 (ref) 1.02 [0.86, 1.21] 0.88 [0.73, 1.05] 0.83 [0.68, 1.00] 0.91 [0.85, 0.98] .011
 Model 2c 1 (ref) 1.1 [0.93, 1.30] 0.93 [0.77, 1.12] 0.86 [0.70, 1.06] 0.92 [0.85, 0.99] .024
Hip fracture events (rate)b 65 (7.8) 63 (6.9) 45 (4.8) 32 (3.3)
 Model 1c 1 (ref) 1.20 [0.84, 1.70] 1.03 [0.69, 1.52] 0.88 [0.57, 1.38] 0.91 [0.77, 1.08] .297
 Model 2c 1 (ref) 1.23 [0.86, 1.76] 1 [0.66, 1.50] 0.81 [0.50, 1.30] 0.87 [0.72, 1.04] .116
LPA (min/day) quartiles
Q1
(<236.5)		 Q2
[236.5, 285.9]		 Q3
[285.9, 334.6]		 Q4
(>334.6)		 HR for 1-SD (72.1)
increment		 p-trenda
Total fracture events (rate)b 361 (45.6) 381 (45.0) 328 (37.8) 331 (37.7)
 Model 1c 1 (ref) 1.05 [0.90, 1.21] 0.92 [0.79, 1.07] 0.96 [0.83, 1.12] 0.98 [0.93, 1.03] .439
 Model 2c 1 (ref) 1.03 [0.89, 1.20] 0.90 [0.77, 1.06] 0.94 [0.80, 1.10] 0.97 [0.91, 1.03] .311
OF events (rate)b 251 (30.7) 265 (30.3) 230 (25.7) 236 (26.2)
 Model 1c 1 (ref) 1.08 [0.90, 1.28] 0.98 [0.82, 1.18] 1.06 [0.88, 1.27] 1.01 [0.95, 1.08] .774
 Model 2c 1 (ref) 1.05 [0.88, 1.25] 0.96 [0.79, 1.15] 1.01 [0.83, 1.23] 0.99 [0.92, 1.06] .813
Hip fracture events (rate)b 50 (5.8) 51 (5.6) 54 (5.8) 50 (5.3)
 Model 1c 1 (ref) 1.16 [0.78, 1.71] 1.40 [0.95, 2.06] 1.46 [0.98, 2.17] 1.14 [0.99, 1.30] .074
 Model 2c 1 (ref) 1.07 [0.72, 1.60] 1.23 [0.82, 1.85] 1.23 [0.80, 1.88] 1.06 [0.91, 1.24] .460
Steps (n/day) quartiles
Q1
(<2,116)		 Q2
[2,116, 3,151]		 Q3
[3,151, 4,491]		 Q4
(>4,491)		 HR for 1-SD (2,049)
increment		 p-trenda
Total fracture events (rate)b 385 (51.9) 350 (41.4) 360 (41.0) 306 (33.3)
 Model 1c 1 (ref) 0.84 [0.73, 0.97] 0.88 [0.76, 1.02] 0.77 [0.65, 0.90] 0.95 [0.89, 1.01] .089
 Model 2c, 1 (ref) 0.92 [0.79, 1.08] 0.96 [0.82, 1.13] 0.84 [0.69, 1.01] 0.98 [0.92, 1.06] .652
OF events (rate)b 280 (36.6) 237 (27.0) 260 (28.8) 205 (21.7)
 Model 1c 1 (ref) 0.80 [0.67, 0.96] 0.92 [0.77, 1.09] 0.77 [0.63, 0.93] 0.95 [0.88, 1.02] .150
 Model 2c 1 (ref) 0.87 [0.73, 1.05] 0.98 [0.81, 1.18] 0.81 [0.65, 1.02] 0.97 [0.89, 1.05] .477
Hip fracture events (rate)b 66 (8.2) 54 (5.9) 55 (5.8) 30 (3.1)
 Model 1c 1 (ref) 0.97 [0.67, 1.39] 1.16 [0.80, 1.69] 0.85 [0.53, 1.34] 0.99 [0.83, 1.17] .864
 Model 2c 1 (ref) 0.98 [0.67, 1.43] 1.08 [0.72, 1.63] 0.71 [0.42, 1.19] 0.92 [0.76, 1.11]
Total sitting time (min/day) quartiles
Q1
(<555.9)		 Q2
[555.9, 625.6]		 Q3
[625.6, 695.6]		 Q4
(>695.6)		 HR for 1-SD (103.0)
increment		 p-trenda
Total fracture events (rate)b 320 (37.2) 325 (38.8) 345 (42.6) 354 (48.2)
 Model 1c 1 (ref) 1.03 [0.88, 1.21] 1.08 [0.92, 1.26] 1.19 [1.02, 1.39] 1.06 [1.00, 1.12] .048
 Model 2c, 1 (ref) 1.05 [0.90, 1.23] 1.08 [0.92, 1.27] 1.15 [0.97, 1.36] 1.04 [0.98, 1.11] .181
OF events (rate)b 211 (23.7) 221 (25.5) 251 (30.2) 254 (33.5)
 Model 1c 1 (ref) 1.05 [0.87, 1.27] 1.16 [0.96, 1.40] 1.24 [1.02, 1.49] 1.08 [1.01, 1.15] .030
 Model 2c 1 (ref) 1.09 [0.90, 1.33] 1.21 [0.99, 1.46] 1.27 [1.03, 1.56] 1.09 [1.01, 1.17] .026
Hip fracture events (rate)b 38 (4.1) 51 (5.7) 54 (6.2) 56 (7.0)
 Model 1c 1 (ref) 1.22 [0.80, 1.86] 1.06 [0.69, 1.61] 1.01 [0.66, 1.54] 1.01 [0.87, 1.17] .874
 Model 2c 1 (ref) 1.35 [0.88, 2.07] 1.17 [0.76, 1.82] 1.17 [0.74, 1.84] 1.08 [0.92, 1.27] .360
Mean sitting bout duration (min/bout) quartiles
Q1 (<9.3) Q2 [9.3, 11.9] Q3 [11.9, 15.2] Q4 (>15.2) HR for 1-SD (5.3) increment p-trenda
Total fracture events (rate)b 295 (35.8) 325 (39.1) 349 (42.4) 375 (49.0)
 Model 1c 1 (ref) 0.97 [0.83, 1.13] 0.99 [0.85, 1.16] 1.07 [0.92, 1.25] 1.02 [0.97, 1.08] .450
 Model 2c, 1 (ref) 0.97 [0.83, 1.14] 0.98 [0.84, 1.15] 1.07 [0.91, 1.26] 1.01 [0.95, 1.07] .835
OF events (rate)b 208 (24.6) 238 (27.8) 241 (28.3) 250 (31.5)
 Model 1c 1 (ref) 0.99 [0.82, 1.19] 0.95 [0.78, 1.14] 0.95 [0.78, 1.14] 0.97 [0.90, 1.03] .326
 Model 2c 1 (ref) 0.99 [0.82, 1.19] 0.94 [0.78, 1.14] 0.96 [0.78, 1.17] 0.96 [0.89, 1.03] .274
Hip fracture events (rate)b 46 (5.2) 48 (5.3) 50 (5.6) 55 (6.6)
 Model 1c 1 (ref) 0.84 [0.56, 1.26] 0.82 [0.54, 1.22] 0.76 [0.51, 1.13] 0.90 [0.77, 1.04] .152
 Model 2c 1 (ref) 0.86 [0.57, 1.29] 0.86 [0.57, 1.30] 0.87 [0.56, 1.33] 0.93 [0.8, 1.10] .402
Open in a new tab

Note. Model 1 is age and ethnicity adjusted; Model 2 = Model 1 + education + alcohol use + smoking + height + weight + falls in past 12 months + RAND-36 physical functioning + diabetes + thiazide use + anti-osteoporosis medication use, and history of total fracture. Model results were estimated with missing covariate data imputed using multiple imputation by chained equations from the R mice package. The analytic sample size for models with total physical activity, MVPA, LPA, and steps was 6,122. The analytic sample size for models with total sitting time and mean sitting bout duration was 5,855. BMI = body mass index; CI = confidence interval; HR = hazard ratio; IQR = interquartile range; LPA = light-intensity physical activity; MVPA = moderate-to-vigorous physical activity; MSBD = mean sitting bout duration; OF = osteoporotic fracture; PA = physical activity.

a

p values from Cox multivariable linear regression models including accelerometer measures in models in continuous form.

b

Crude incidence rate per 1,000 person-years.

c

Data are HR [95% CI].

In multivariable-adjusted analyses, higher MVPA quartiles remained significantly associated with lower risk of osteoporotic fractures, and higher quartiles of total sitting time remained associated with higher risk of osteoporotic fractures (Table 2). After adjustment for potential confounders in Model 2, the HR (95% CI) for incident osteoporotic fractures across MVPA quartiles were Q1: 1.00 (reference), Q2: 1.10 [0.93, 1.30], Q3: 0.93 [0.77, 1.12], and Q4: 0.86 [0.70, 1.07]; p-trend = .02 (Table 2). The fully adjusted HR (95% CI) for osteoporotic fractures across total sitting time quartiles were Q1: 1.00 (reference), Q2: 1.09 [0.90, 1.33], Q3: 1.21 [0.99, 1.46], and Q4: 1.27 [1.03, 1.56]; p-trend = .03 (Table 2). HRs for MVPA and total sitting time with total fracture remained consistent in magnitude but were no longer significant after adding physical function in models that incrementally contained Model 2 covariates (Supplementary Table S1 [available online]). Otherwise, MVPA and total sitting time were not associated with risk of total fracture or other fracture endpoints, and total PA, LPA, steps/day, and mean sitting bout duration were not associated with fracture risk (Table 2 and Supplementary Table S1 [available online]).

Results of restricted cubic spline regression analyses indicated that the associations of total PA, LPA, MVPA, total sitting time, and mean sitting bout duration with fracture endpoints were linear (all p-nonlinear > .18); therefore, we proceeded with modeling using continuous linear functional forms (Figure 1). The interquartile range and corresponding HR (95% CI) for osteoporotic fracture was 121.4 min/day and 0.95 [0.86, 1.05] for total PA, 98.1 min/day and 0.99 [0.90, 1.09] for LPA, 42.2 min/day and 0.89 [0.81, 0.99] for MVPA, 2,375/steps a day and 0.97 [0.88, 1.06] for steps, 139.7 min/day and 1.12 [1.01, 1.24] for total sitting time, and 5.9 min/bout and 0.96 [0.88, 1.04] for mean sitting bout duration (Table 3).

Figure 1 —

Figure 1 —

Open in a new tab

Continuous dose–response associations of total PA, MVPA, LPA, steps/day, total sitting time, and MSBD with incident total fracture, osteoporotic fracture, major osteoporotic fracture, and hip fracture among women in the OPACH Cohort 2012–2020. Models adjusted for age, race and ethnicity, education, alcohol use, smoking status, height, weight, falls history in the last 12 months, RAND-36 physical function, diabetes, thiazide use, anti-osteoporotic medication use, and history of total fracture. PA and SB measures (total PA, MVPA, LPA, steps, total sitting time, and MSBD) were modeled using restricted cubic splines with knots placed at the 10th, 50th, and 90th percentiles. Results were trimmed at the first and 99th percentiles. The reference was set at the 10th percentile. Bottom panels (histograms) show the distributions of PA and SB measures. The vertical dashed lines in the histograms correspond to the arithmetic mean. OPACH = Objective Physical Activity and Cardiovascular Health; CHAP = Convolutional Neural Network Hip Accelerometer Posture; HR = hazard ratio; LPA = light-intensity physical activity; MOF = major osteoporotic fracture; MVPA = moderate-to-vigorous PA; OF = osteoporotic fracture; PA = physical activity; MSBD = mean sitting bout duration; SB = sedentary behavior.

Table 3.

Associations of the Interquartile Range in Total PA, LPA, MVPA, Steps/Day, Total Sitting Time, and MSBD With Incident Osteoporotic Fracture Stratified by Baseline Characteristics in OPACH Cohort 2012–2020 (n = 6,122)

n n
event		 Total PA
 LPA
 MVPA
 Steps
 Total sitting time
 MSBD
HR
[95% CI]a 		p-int HR
(95% CI)a 		p-int HR
(95% CI)a 		p-int HR
(95% CI)a 		p-int n n
event		 HR
(95% CI)a 		p-int HR
(95% CI)a 		p-int
Total sample 6,122 982 0.95
[0.86, 1.05]		 0.99
[0.90, 1.09]		 0.89
[0.81, 0.99]		 0.97
[0.88, 1.06]		 5,855 937 1.12
[1.01, 1.24]		 0.96
[0.88, 1.04]
Age .11 .14 .50 .55 .48 .03
 <80 years 3,072 341 0.87
[0.74, 1.03]		 0.92
[0.78, 1.08]		 0.85
[0.73, 0.99]		 0.94
[0.81, 1.09]		 2,956 328 1.15
[0.96, 1.37]		 1.08
[0.94, 1.24]
 ≥80 years 3,050 641 0.99
[0.88, 1.12]		 1.02
[0.90, 1.14]		 0.94
[0.83, 1.06]		 1.00
[0.88, 1.13]		 2,899 609 1.10
[0.97, 1.25]		 0.92
[0.83, 1.01]
BMI .77 .64 .06 .17 .17 .62
 <30 kg/m2 4,168 743 0.94
[0.84, 1.05]		 0.99
[0.89, 1.11]		 0.87
[0.78, 0.97]		 0.97
[0.88, 1.08]		 3,992 711 1.14
[1.02, 1.28]		 0.94
[0.85, 1.04]
 ≥30 kg/m2 1,954 239 0.94
[0.76, 1.16]		 0.93
[0.76, 1.14]		 1.02
[0.82, 1.26]		 0.98
[0.76, 1.27]		 1,863 226 1.02
[0.81, 1.28]		 1.00
[0.87, 1.16]
Falls .88 1.00 .70 .18 .80 .76
 0 or 1 5,405 821 0.94
[0.84, 1.05]		 0.98
[0.89, 1.09]		 0.89
[0.80, 0.98]		 0.93
[0.84, 1.03]		 5,182 788 1.14
[1.02, 1.28]		 0.97
[0.88, 1.06]
 2 or more 717 161 0.99
[0.77, 1.27]		 1.02
[0.80, 1.29]		 0.93
[0.73, 1.19]		 1.19
[0.92, 1.53]		 673 149 1.02
[0.79, 1.31]		 0.92
[0.77, 1.10]
Physical function .83 .36 .260 .34 .31 .99
 <75 2,808 516 0.90
[0.78, 1.04]		 0.91
[0.79, 1.04]		 0.96
[0.83, 1.11]		 1.05
[0.88, 1.25]		 2,684 492 1.11
[0.96, 1.28]		 0.98
[0.89, 1.09]
 ≥75 3,314 466 1.00
[0.87, 1.15]		 1.07
[0.94, 1.23]		 0.87
[0.76, 0.99]		 0.96
[0.85, 1.08]		 3,171 445 1.15
[1.00, 1.34]		 0.94
[0.81, 1.08]
Diabetes .50 .43 .75 .12 .16 .89
 No 4,876 790 0.97
[0.87, 1.08]		 1.00
[0.90, 1.12]		 0.91
[0.82, 1.01]		 1.01
[0.91, 1.12]		 4,671 757 1.08
[0.96, 1.21]		 0.96
[0.87, 1.05]
 Yes 1,246 192 0.86
[0.68, 1.09]		 0.90
[0.72, 1.12]		 0.83
[0.65, 1.08]		 0.77
[0.58, 1.01]		 1,184 180 1.33
[1.03, 1.70]		 0.96
[0.81, 1.13]
Thiazide use .99 .60 .36 .67 .34 .73
 No 2,185 362 0.98
[0.84, 1.15]		 1.04
[0.89, 1.21]		 0.88
[0.76, 1.03]		 0.99
[0.86, 1.15]		 2,057 338 1.18
[1.00, 1.39]		 0.97
[0.85, 1.11]
 Yes 3,937 620 0.93
[0.82, 1.06]		 0.96
[0.85, 1.08]		 0.91
[0.80, 1.03]		 0.95
[0.83, 1.08]		 3,798 599 1.08
[0.95, 1.23]		 0.95
[0.86, 1.05]
Anti-osteoporosis medication .15 .43 .02 .01 .65 .17
 No 4,805 676 0.90
[0.80, 1.01]		 0.96
[0.86, 1.08]		 0.82
[0.72, 0.92]		 0.86
[0.76, 0.97]		 4,602 644 1.09
[0.97, 1.24]		 0.99
[0.90, 1.09]
 Yes 1,317 306 1.06
[0.89, 1.27]		 1.04
[0.87, 1.23]		 1.06
[0.90, 1.24]		 1.21
[1.03, 1.42]		 1,253 293 1.18
[0.99, 1.41]		 0.89
[0.75, 1.04]
History of total fracture .02 .02 .03 .01 .07 .06
 No 5,480 827 0.97
[0.87, 1.08]		 1.01
[0.91, 1.12]		 0.90
[0.81, 1.00]		 0.96
[0.86, 1.07]		 5,242 792 1.09
[0.98, 1.22]		 0.95
[0.87, 1.03]
 Yes 642 155 0.84
[0.64, 1.11]		 0.89
[0.68, 1.15]		 0.85
[0.66, 1.10]		 0.98
[0.76, 1.27]		 613 145 1.31
[1.00, 1.72]		 1.01
[0.81, 1.25]
Race/ethnicity .37 .43 .35 .21 .69 .02
 White 3,044 681 0.98
[0.87, 1.10]		 1.00
[0.89, 1.13]		 0.94
[0.84, 1.06]		 1.02
[0.91, 1.15]		 2,915 647 1.12
[0.99, 1.27]		 0.90
[0.82, 1.00]
 Black 2,046 165 0.93
[0.73, 1.17]		 1.01
[0.81, 1.26]		 0.74
[0.57, 0.97]		 0.76
[0.58, 0.99]		 1,944 159 1.07
[0.83, 1.37]		 1.06
[0.89, 1.26]
 Hispanic or Latina 1,032 136 0.85
[0.65, 1.11]		 0.90
[0.69, 1.18]		 0.85
[0.67, 1.07]		 0.96
[0.76, 1.21]		 996 131 1.18
[0.90, 1.56]		 1.20
[0.91, 1.57]
Open in a new tab

Note. IQRs for total PA, LPA, MVPA, steps/day, total sitting time, and MSBD were 121.4 min/day, 98.1 min/day, 42.2 min/day, 139.7 min/day, and 5.9 min/bout, respectively. Associations were adjusted for age, race/ethnicity, educational attainment, alcohol use, smoking status, height, weight, falls in the past 12 months, physical function, thiazide use, anti-osteoporosis medication, and prevalent fracture. Model results were estimated using multiple imputation by chained equations from R mice package. BMI = body mass index; CI = confidence interval; HR = hazard ratio; IQR = interquartile range; LPA = light-intensity physical activity; MVPA = moderate-to-vigorous physical activity; MSBD = mean sitting bout duration; PA = physical activity; OPACH = Objective Physical Activity and Cardiovascular Health.

a

HR (95% CI) = hazard ratio (95% confidence interval).

Effect Modification of Associations of PA and SB With Incident Fracture

In stratified analyses, results for MVPA and total sitting time with osteoporotic fracture risk were generally consistent in direction across most cohort subgroups (Table 3). There was nominal evidence of effect modification of the association between MVPA and osteoporotic fracture risk by anti-osteoporosis medication use (p-interaction = .02), suggesting lower risk for women who reported no anti-osteoporosis medication use (HR = 0.82, 95% CI [0.72, 0.92]) but not for those who reported anti-osteoporotic medication use (HR = 1.06, 95% CI [0.90, 1.24]; Table 3). Higher steps/day was not associated with osteoporotic fracture risk overall; however, there was nominal evidence of effect modification by anti-osteoporotic medication use (p-interaction = .01), suggesting lower risk for women who reported no anti-osteoporotic medication use (HR = 0.86, 95% CI [0.76, 0.97]) and higher risk for those who reported anti-osteoporotic medication use (HR = 1.21, 95% CI [1.03, 1.42]). Higher total PA was not associated with higher osteoporotic fracture risk overall, however, there was nominal evidence of effect modification by history of total fracture (p-interaction = .02). The HR (95% CI) was 0.84 [0.64, 1.11] among those with a history of total fracture and 0.97 [0.87, 1.08] among those without a history of total fracture. Similarly, higher LPA was not overall associated with osteoporotic fracture risk; however, there was nominal evidence of effect modification by history of total fracture (p-interaction = .02). The HR (95% CI) was 0.89 [0.68, 1.15] among women with a history of total fracture and 1.01 [0.91, 1.12] among those without a history of total fracture. Although longer mean sitting bout duration was not overall associated with osteoporotic fracture risk, there was nominal evidence of effect modification by age (p-interaction = .03). The HR (95% CI) was 1.08 [0.94, 1.24] among those aged less than 80 years and 0.92 [0.83, 1.01] among those aged 80 or older. There was nominal evidence of effect modification for mean sitting bout duration with osteoporotic fracture risk by race/ethnicity (p-interaction = .02). The HR (95% CI) was 0.90 [0.82, 1.00] among White women, 1.06 [0.89, 1.26] among Black women, and 1.20 [0.91, 1.57] among Hispanic or Latina women.

Results from stratified analysis for the associations of PA and SB measures with other fracture endpoints were generally consistent in direction across cohort subgroups. Higher total PA, LPA, and steps/day appeared to be associated with lower risk of total fracture among women who reported no anti-osteoporosis medication use but not for those who reported anti-osteoporosis medication use (p-interaction = .03, .05, .04, respectively; Supplementary Table S2 [available online]). Higher total PA, LPA, MVPA, and steps/day appeared to be associated with lower risk of total fracture among women with a history of total fracture but not for those without (p-interaction = .001, .001, .002, .003; Supplementary Table S2 [available online]). Higher total sitting time and longer mean sitting bout duration appeared to be associated with higher total fracture risk among women with a history of total fracture but not for those without (p-interaction = .01; Supplementary Table S2 [available online]). Higher MVPA and steps/day appeared to be associated with lower risk of incident major osteoporotic fracture among women who reported no anti-osteoporosis medication use but not for those who reported anti-osteoporosis medication use (Supplementary Table S3 [available online]). However, these HR 95% CIs were relatively wide and contained the value “1,” indicating that the observed hazard ratio was not statistically significant. Results from stratified analysis for the associations of PA and SB measures with incident hip fracture were consistent across cohort subgroups (Supplementary Table S4 [available online]). However, these results should be interpreted with caution given the number of effect modification tests (six accelerometer measures, nine stratification variables, and four endpoints for 216 tests).

Discussion

In this prospective study of racially and ethnically diverse older women, higher amounts of accelerometer-measured MVPA were associated with lower risk of incident osteoporotic fracture. Higher amounts of total sitting time were associated with higher risk of osteoporotic fracture. These associations were independent of several relevant covariates including physical function, history of total fracture, and medication use. Associations were inverse and linear for MVPA and linear for total sitting time, suggesting that higher MVPA and lower total sitting time could contribute to lower fracture risk regardless of how much MVPA and SB one accrues. Each 33 min/day increment in MVPA was associated with an 8% lower risk of osteoporotic fracture, whereas each 103 min/day increment in total sitting time was associated with a 9% higher risk of osteoporotic fracture. These associations were generally consistent in direction across cohort subgroups, enhancing confidence in the primary study findings. Otherwise, MVPA and total sitting time were not overall associated with risk of other incident fracture endpoints. Total PA, LPA, steps/day, and mean sitting bout duration were not overall associated with incident fracture risk. To our knowledge, this is the first prospective study to investigate accelerometer measures of PA and SB with fracture risk among older women.

The present study results have significant public health implications. Older adults face barriers to engaging in higher amounts of PA, particularly MVPA, including fear of falling, safety, and cost (Kerr et al., 2012; Spiteri et al., 2019). Much of the movement older adults engage in is associated with daily living activities, which are not well captured by questionnaires and are composed mainly of LPA with relatively less MVPA (Colbert et al., 2014). In addition, older adults can accumulate substantial amounts of SB during awake time (e.g., screen use, leisure activities, transportation, and in the home environment; Jain et al., 2021; Nguyen et al., 2022; Owen et al., 2011). PA and SB are not merely the inverse of each other as individuals can accrue both high PA and high SB throughout the day. The present study results show that higher amounts of accelerometer-measured MVPA are associated with lower risk of osteoporotic fracture, whereas higher amounts of total sitting time are associated with higher osteoporotic fracture risk.

Three stratified analysis results in the present study warrant further attention. First, higher MVPA and steps were associated with lower osteoporotic fracture risk among women who reported no anti-osteoporosis medication use. This could be due to the potency of anti-osteoporosis medications, particularly bisphosphonates, to the extent that they could obscure the smaller effects of MVPA and steps. The findings for steps/day and osteoporotic fracture risk among women who reported no anti-osteoporosis medication use are noteworthy as steps are recorded by a variety of wearable devices and could be more readily adopted than other PA metrics such as MVPA volume. Second, higher total PA and LPA was associated with lower osteoporotic fracture risk among women with a history of total fracture. Similarly, higher amounts of total PA, LPA, MVPA, and steps; lower total sitting time; and shorter mean sitting bout duration were associated with lower risk of total fracture among women with a history of total fracture. However, there were relatively few women with a history of total fracture potentially resulting in wide HR 95% CIs. Larger studies are needed to examine whether higher PA and lower SB could contribute to lower fracture risk among women with a history of fracture. Importantly, there was no evidence that higher PA was associated with greater fracture risks among women with a history of total fracture. Finally, although higher LPA did not contribute to fracture risk overall, LPA was also not associated with higher fracture risk. Taken together, the present study results suggest that higher PA and lower SB can be encouraged for a lower risk of subsequent total fracture and general health benefits without concern of increased fracture risks and that higher PA intensities can be encouraged for lower osteoporotic fracture risk.

The present study extends the literature on the prospective associations of PA and SB with fracture risk. A published study in WHI showed that compared with women with 0 MET-hr/week of mild PA, those with at least 3 MET-hr/week had a lower risk of total fracture (HR = 0.91, 95% CI [0.87, 0.94]; p-trend < .001) and hip fracture (HR = 0.82, 95% CI [0.73, 0.93]; p-trend = .003) (LaMonte, Wactawski-Wende, et al., 2019). In addition, compared with women with less than 9 MET-hr/week of self-reported MVPA, those with at least 9 MET-hr/week did not have a lower risk of total fracture (HR = 1.01, 95% CI [0.99, 1.04]; p-trend = .38) but had a lower risk of hip fracture (HR = 0.88, 95% CI = 0.81, 0.96; p-trend = .002; LaMonte, Wactawski-Wende, et al., 2019). In the Million Women Study, women who reported engaging in strenuous exercise at least once per week had a lower risk of hip fracture compared with those who reported never or rarely engaging in strenuous exercise (relative risk = 0.63, 95% CI [0.59, 0.67]; p-trend < .001; Armstrong et al., 2012). In contrast, the present study found that higher amounts of accelerometer LPA and MVPA were not overall associated with lower risk of total fracture; however, though the results for MVPA with hip fracture were directionally consistent with the literature, the 95% CIs contained 1. These differences could be due to the PA and SB measures used as the Pearson correlation between self-reported and accelerometer measured total PA, LPA, MVPA, and sitting in the present study analytic sample were .26, .05, .16, and .33, respectively. Thus, questionnaires may not be capturing all PA and SB as accurately as accelerometry and may not be measuring the same behaviors (LaMonte, Lee, et al., 2019). Importantly, the present study showed that higher amounts MVPA, whether from daily living activities or leisure activities, and lower total sitting time could modestly contribute to fracture prevention.

Relatively few studies have examined accelerometer measures of PA with incident fracture. In the Osteoporotic Fractures in Men Study, the highest quintile of accelerometer-measured MVPA, expressed as active energy expenditure, was associated with a lower risk of nonspine fracture (HR = 0.55, 95% CI [0.33, 0.91]; p-trend = .04) relative to the lowest quintile. This association is consistent in direction with but stronger in magnitude than the present study findings for MVPA, possibly attributable to differences in study population; accelerometer model; and its placement, endpoints, and PA operationalization (Cauley et al., 2013). In the UK Biobank Study, a 150 min/week increment in MVPA was associated with higher risk of fracture of the upper limb (HR = 1.11, 95% CI [1.03, 1.19]) and lower risk of fracture of the lower limb (HR = 0.89, 95% CI [0.79, 0.99]) classified using hospital records among those aged 65 or older, but it did not consider hip fractures or combined fracture endpoints such as major osteoporotic fracture as in the present study. Importantly, the present study extends the literature on accelerometer measures of PA with fracture by also examining steps and SB in relation to several fracture endpoints.

Few studies have examined the relationship of accelerometer-measured SB with fracture. In Osteoporotic Fractures in Men Study, higher amounts of device-measured total sitting time were nonsignificantly associated with higher risk of nonspine fracture (HR for highest vs. lowest quintile = 1.39, 95% CI [0.83, 2.33]; p-trend = .09; Cauley et al., 2013). The present study examined the SB-fracture association among a diverse study of older women, a group at higher risk of osteoporosis and osteoporotic fracture. In addition, we classified sitting using the CHAP algorithm, which accurately captures sitting and postural transitions compared with accelerometer cutpoint methods that could misclassify low-intensity and nonsitting behaviors like standing. Thus, the present study results are relevant and could inform future SB guidelines.

The U.S. 2018 Physical Activity Guidelines (PAG) recommends PA to reduce the risk of injuries from falls, including those that result in bone fracture (2018 Physical Activity Guidelines Advisory Committee, 2018). However, the PAG does not provide recommendations regarding SB and fracture. In addition, the PAG identified the need for future research to examine the dose–response relationship between PA and fracture risk as well as whether these associations differ by race and ethnicity and by existing impairments (e.g., decline in physical function; 2018 Physical Activity Guidelines Advisory Committee, 2018). The present study supports the PAG to increase PA for fracture prevention and contributes relevant information for addressing research gaps in these identified areas to thereby inform future PA and SB guidelines for fracture prevention in older adults.

Several mechanisms could explain the present study results for PA. Movements and behaviors such as walking involve weight-bearing and impact (ground reaction forces), which could improve balance, strength, and bone mineral density, thereby reducing fracture risk (2018 Physical Activity Guidelines Advisory Committee, 2018; Garber et al., 2011; Shojaa et al., 2020; Thomas et al., 2019). A meta-analysis of 20 randomized trials of exercise interventions totaling 7,704 older adults and 428 fracture events showed that exercise interventions were associated with lower risk of fracture (relative risk = 0.74, 95% CI [0.59, 0.92]; p = .007; Wang et al., 2020). However, the lower intensity movement engaged in by older women during daily living activities or leisure-time PA may not contain sufficient weight-bearing or impact to support bone health. These movements may reduce fracture risk at some but not all sites, potentially explaining why the present study observed an inverse association of MVPA with osteoporotic fracture, due to the inclusion of leg fracture, but not with major osteoporotic fracture or total fracture. Individuals may be at higher risk of falls and injuries resulting in a fracture when engaging in PA (Bea et al., 2017). Higher amounts of SB are associated with lower physical function and bone mineral density due to a lack of muscle activation and bone unloading, which could increase fracture risk (Chastin et al., 2014; Seguin et al., 2012). However, when individuals are sedentary, they are at a lower risk of a fall or other injurious event potentially resulting in a fracture. These mechanisms could have contributed to the modest associations for MVPA and total sitting time with osteoporotic fracture risk in the present study. Future research on measuring ground reaction forces during movement and SB with accelerometers, and their associations with fracture risk, is warranted (Cauley & Giangregorio, 2020).

The present study has several notable strengths. OPACH is a large study of racially and ethnically diverse older women, an understudied population with regards to aging-related outcomes. A week of accelerometer data for PA and SB were successfully collected, which provides substantial quantification of movement in everyday life. Extensive health information was also available, allowing for detailed assessment of potential confounding factors including health behaviors, health status risk factors, and medication use. We addressed missing data using multiple imputation by chained equations to reduce selection bias and enhance the precision of study results.

We acknowledge limitations when considering the results of this study. Only hip fractures were adjudicated; fractures at other sites were self-reported. Although there was high agreement between self-report and medical record review for fractures of the lower arm or wrist (81%) and lower leg or ankle (78%), there was lower agreement for fractures of vertebrae (51%) and upper leg (28%), resulting in potential misclassification of combined fracture endpoints (Chen et al., 2004). Accelerometer measures of PA and SB were collected only at OPACH baseline, precluding analyses of changes in PA and SB with fracture risk. We applied calibrated accelerometer cutpoints to classify LPA and MVPA; however, some individuals could have experienced relative intensity MVPA even though their absolute intensity was classified as LPA using these cutpoints. In addition, accelerometer data do not distinguish between types of activity, for example, balance training. The results of the present study should be examined in other studies of men and younger women who were not represented in the OPACH cohort.

Our study showed that higher amounts of MVPA were associated with lower risk of osteoporotic fracture, whereas higher amounts of total sitting time were associated with higher osteoporotic fracture risk. Large, randomized trials like the WHI Strong and Healthy are needed to evaluate the effectiveness of PA and SB interventions in relation to fracture among older adults. Healthcare providers and future PA guidelines should continue promoting PA alongside reducing SB for fracture prevention among aging populations.

Supplementary Material

supplementary file

Key Points.

  • Among older women, those with more moderate-to-vigorous physical activity and less sitting were less likely to experience osteoporotic fracture.

  • Importantly, higher amounts of total physical activity, light-intensity physical activity, and steps were not associated with higher risk of osteoporotic fracture, including among women with a history of fracture.

  • Women can be encouraged to move more and sit less as part of a fracture prevention strategy.

Acknowledgments

We thank the WHI participants, staff, and investigators. The short list of WHI investigators can be found at 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: http://www.whi.org/researchers/Documents%2520%2520Write%2520a%2520Paper/WHI%2520Investigator%2520Long%2520List.pdf.

Funding:

This work was supported by the National Institute on Aging (P01 AG052352 (5T32AG058529-03, 1K99AG082863-01) and the National Heart, Lung, and Blood Institute (R01HL105065, 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, and 75N92021D00005). The funding agencies were not involved in the design of the study, data collection, analysis, data interpretation, or manuscript writing.

Funding for the Parent Study:

This work was supported by the National Heart, Lung, and Blood Institute (contracts HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, HHSN271201100004C) and the National Heart, Lung and Blood Institute (grant R01HL105065).

References

  1. 2018 Physical Activity Guidelines Advisory Committee. (2018). 2018 physical activity guidelines advisory committee scientific report. U.S. Department of Health and Human Services. https://health.gov/sites/default/files/2019-09/PAG_Advisory_Committee_Report.pdf [Google Scholar]
  2. Armstrong MEG, Cairns BJ, Banks E, Green J, Reeves GK, & Beral V. (2012). Different effects of age, adiposity and physical activity on the risk of ankle, wrist and hip fractures in postmenopausal women. Bone, 50(6), 1394–1400. 10.1016/j.bone.2012.03.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bea JW, Thomson CA, Wallace RB, Wu C, Seguin RA, Going SB, LaCroix A, Eaton C, Ockene JK, LaMonte MJ, Jackson R, Jerry Mysiw W, & Wactawski-Wende J (2017). Changes in physical activity, sedentary time, and risk of falling: The women’s health initiative observational study. Preventive Medicine, 95, 103–109. 10.1016/j.ypmed.2016.11.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bellettiere J, Lamonte MJ, Evenson KR, Rillamas-Sun E, Kerr J, Lee I-MM, Di C, Rosenberg DE, Stefanick ML, Buchner DM, Hovell MF, Lacroix AZ, Stefanick M, Buchner DM, Hovell MF, & LaCroix AZ (2019). Sedentary behavior and cardiovascular disease in older women: The OPACH study. Circulation, 139(8), 1036–1046. 10.1161/CIRCULATIONAHA.118.035312 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, & Tosteson A (2007). Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. Journal of Bone and Mineral Research, 22(3), 465–475. 10.1359/jbmr.061113 [DOI] [PubMed] [Google Scholar]
  6. Cauley JA, & Giangregorio L. (2020). Physical activity and skeletal health in adults. The Lancet Diabetes & Endocrinology, 8(2), 150–162. 10.1016/S2213-8587(19)30351-1 [DOI] [PubMed] [Google Scholar]
  7. Cauley JA, Harrison SL, Cawthon PM, Ensrud KE, Danielson ME, Orwoll E, & Mackey DC (2013). Objective measures of physical activity, fractures and falls: The osteoporotic fractures in men study. Journal of the American Geriatrics Society, 61(7), 1080–1088. 10.1111/jgs.12326 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chastin SFM, Mandrichenko O, Helbostadt JL, & Skelton DA (2014). Associations between objectively-measured sedentary behaviour and physical activity with bone mineral density in adults and older adults, the NHANES study. Bone, 64, 254–262. 10.1016/j.bone.2014.04.009 [DOI] [PubMed] [Google Scholar]
  9. Chen Z, Kooperberg C, Pettinger MB, Bassford T, Cauley JA, LaCroix AZ, Lewis CE, Kipersztok S, Borne C, & Jackson RD. (2004). Validity of self-report for fractures among a multiethnic cohort of postmenopausal women: Results from the Women’s Health Initiative observational study and clinical trials. Menopause, 11(3), 264–274. 10.1097/01.GME.0000094210.15096.FD [DOI] [PubMed] [Google Scholar]
  10. Choi L, Liu Z, Matthews CE, & Buchowski MS (2011). Validation of accelerometer wear and nonwear time classification algorithm. Medicine and Science in Sports and Exercise, 43(2), 357–364. 10.1249/MSS.0b013e3181ed61a3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Colbert LH, Matthews CE, Schoeller DA, Havighurst TC, & Kim KM. (2014). Intensity of physical activity in the energy expenditure of older Adults. Journal of Aging and Physical Activity, 22(4), 571–577. 10.1123/JAPA.2012-0257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Crandall CJ, Larson JC, LaCroix AZ, Robbins JA, Wactawski-Wende J, Johnson KC, Sattari M, Stone KL, Weitlauf JC, Gure TR, & Cauley JA. (2021). Risk of subsequent fractures in postmenopausal women after nontraumatic vs traumatic fractures. JAMA Internal Medicine, 181(8), 1055–1063. 10.1001/jamainternmed.2021.2617 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Crandall CJ, Larson JC, Watts NB, Gourlay ML, Donaldson MG, LaCroix A, Cauley JA, Wactawski-Wende J, Gass ML, Robbins JA, & Ensrud KE. (2014). Comparison of fracture risk prediction by the US preventive services task force strategy and two alternative strategies in women 50-64 years old in the women’s health initiative. Journal of Clinical Endocrinology and Metabolism, 99(12), 4514–4522. 10.1210/jc.2014-2332 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ensrud KE (2013). Epidemiology of fracture risk with advancing age. The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences, 68(10), 1236–1242. 10.1093/gerona/glt092 [DOI] [PubMed] [Google Scholar]
  15. Evenson KR, Wen F, Herring AH, Di C, LaMonte MJ, Tinker LF, Lee I-M, Rillamas-Sun E, LaCroix AZ, & Buchner DM. (2015). Calibrating physical activity intensity for hip-worn accelerometry in women age 60 to 91 years: The Women’s Health Initiative OPACH Calibration Study. Preventive Medicine Reports, 2, 750–756. 10.1016/j.pmedr.2015.08.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Garber CE, Blissmer B, Deschenes MR, Franklin BA, Lamonte MJ, Lee IM, Nieman DC, & Swain DP. (2011). Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: Guidance for prescribing exercise. Medicine and Science in Sports and Exercise, 43(7), 1334–1359. 10.1249/MSS.0b013e318213fefb [DOI] [PubMed] [Google Scholar]
  17. Greenwood-Hickman MA, Nakandala S, Jankowska MM, Rosenberg D, Tuz-Zahra F, Bellettiere J, Carlson J, Hibbing PR, Zou J, LaCroix AZ, Kumar A, & Natarajan L. (2021). The CNN Hip Accelerometer Posture (CHAP) Method for classifying sitting patterns from hip accelerometers: A validation study. Medicine & Science in Sports & Exercise, 53(11), 2445–2454. 10.1249/MSS.0000000000002705 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Harel O, Mitchell EM, Perkins NJ, Cole SR, Tchetgen Tchetgen EJ, Sun B, & Schisterman EF (2018). Multiple imputation for incomplete data in epidemiologic studies. American Journal of Epidemiology, 187(3), 576–584. 10.1093/aje/kwx349 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Harrell FE Jr. (2015). Regression modeling strategies with applications to linear models, logistic and ordinal regression and survival analysis (2nd ed.). Springer-Verlag. http://biostat.mc.vanderbilt.edu/rms [Google Scholar]
  20. Harrell FE Jr., Harrell MFE Jr., & Hmisc D. (2017). Package “rms.” Vanderbilt University. [Google Scholar]
  21. Hays RD, Sherbourne CD, & Mazel RM. (1993). The RAND 36-item health survey 1.0. Health Economics, 2(3), 217–227. 10.1002/hec.4730020305 [DOI] [PubMed] [Google Scholar]
  22. Jain P, Bellettiere J, Glass N, LaMonte MJ, Di C, Wild RA, Evenson KR, & LaCroix AZ. (2021). The relationship of accelerometer-assessed standing time with and without ambulation and mortality: The WHI OPACH study. The Journals of Gerontology: Series A, 76(1), 77–84. 10.1093/gerona/glaa227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Katsoulis M, Benetou V, Karapetyan T, Feskanich D, Grodstein F, Pettersson-Kymmer U, Eriksson S, Wilsgaard T, Jørgensen L, Ahmed LA, Schöttker B, Brenner H, Bellavia A, Wolk A, Kubinova R, Stegeman B, Bobak M, Boffetta P, & Trichopoulou A. (2017). Excess mortality after hip fracture in elderly persons from Europe and the USA: The CHANCES project. Journal of Internal Medicine, 281(3), 300–310. 10.1111/joim.12586 [DOI] [PubMed] [Google Scholar]
  24. Kerr J, Rosenberg DE, Nathan A, Millstein RA, Carlson JA, Crist K, Wasilenko K, Bolling K, Castro CM, Buchner DM, & Marshall SJ. (2012). Applying the ecological model of behavior change to a physical activity trial in retirement communities: Description of the study protocol. Contemporary Clinical Trials, 33(6), 1180–1188. 10.1016/j.cct.2012.08.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. LaCroix AZ, Bellettiere J, Rillamas-Sun E, Di C, Evenson KR, Lewis CE, Buchner DM, Stefanick ML, Lee I-M, Rosenberg DE, LaMonte MJ, & for the Women’s Health Initiative (WHI). (2019). Association of light physical activity measured by accelerometry and incidence of coronary heart disease and cardiovascular disease in older women. JAMA Network Open, 2(3), Article e190419. 10.1001/jamanetworkopen.2019.0419 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. LaCroix AZ, Rillamas-Sun E, Buchner D, Evenson KR, Di C, Lee I-M, Marshall S, LaMonte MJ, Hunt J, Tinker LF, Stefanick M, Lewis CE, Bellettiere J, & Herring AH. (2017). The objective physical activity and cardiovascular disease health in older women (OPACH) study. BMC Public Health, 17(1), Article 192. 10.1186/s12889-017-4065-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. LaMonte MJ, LaCroix AZ, Nguyen S, Evenson KR, Di C, Stefanick ML, Hyde ET, Anuskiewicz B, & Eaton CB. (2024). accelerometer-measured physical activity, sedentary time, and heart failure risk in women aged 63 to 99 years. JAMA Cardiology, 9(4), 336–345. 10.1001/jamacardio.2023.5692 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. LaMonte MJ, Lee I-M, Rillamas-Sun E, Bellettiere J, Evenson KR, Buchner DM, Di C, Lewis CE, Rosenberg DE, Stefanick ML, & LaCroix AZ. (2019). Comparison of questionnaire and device measures of physical activity and sedentary behavior in a multi-ethnic cohort of older women. Journal for the Measurement of Physical Behaviour, 2(2), 82–93. 10.1123/jmpb.2018-0057 [DOI] [Google Scholar]
  29. LaMonte MJ, Wactawski-Wende J, Larson JC, Mai X, Robbins JA, LeBoff MS, Chen Z, Jackson RD, LaCroix AZ, Ockene JK, Hovey KM, & Cauley JA. (2019). Association of physical activity and fracture risk among postmenopausal women. JAMA Network Open, 2(10), Article e1914084. 10.1001/jamanetworkopen.2019.14084 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lewiecki EM, Chastek B, Sundquist K, Williams SA, Weiss RJ, Wang Y, Fitzpatrick LA, & Curtis JR. (2020). Osteoporotic fracture trends in a population of US managed care enrollees from 2007 to 2017. Osteoporosis International, 31(7), 1299–1304. 10.1007/s00198-020-05334-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lewiecki EM, Wright NC, Curtis JR, Siris E, Gagel RF, Saag KG, Singer AJ, Steven PM, & Adler RA. (2018). Hip fracture trends in the United States, 2002 to 2015. Osteoporosis International, 29(3), 717–722. 10.1007/s00198-017-4345-0 [DOI] [PubMed] [Google Scholar]
  32. Nguyen S, Bellettiere J, Anuskiewicz B, Di C, Carlson J, Natarajan L, LaMonte MJ, & LaCroix AZ. (2024). Prospective associations of accelerometer-measured machine-learned sedentary behavior with death among older women: The OPACH Study. Journal of the American Heart Association, 13(5), Article e031156. 10.1161/JAHA.123.031156 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Nguyen S, Bellettiere J, Wang G, Di C, Natarajan L, LaMonte MJ, & LaCroix AZ. (2022). Accelerometer-derived daily life movement classified by machine learning and incidence of cardiovascular disease in older women: The OPACH Study. Journal of the American Heart Association, 11(5), Article e023433. 10.1161/JAHA.121.023433 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Nguyen S, LaCroix AZ, Hayden KM, Di C, Palta P, Stefanick ML, Manson JE, Rapp SR, LaMonte MJ, & Bellettiere J. (2023). Accelerometer-measured physical activity and sitting with incident mild cognitive impairment or probable dementia among older women. Alzheimer’s & Dementia, 19(7), 3041–3054. 10.1002/alz.12908 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Owen N, Sugiyama T, Eakin EE, Gardiner PA, Tremblay MS, & Sallis JF. (2011). Adults’ sedentary behavior: Determinants and interventions. American Journal of Preventive Medicine, 41(2), 189–196. 10.1016/j.amepre.2011.05.013 [DOI] [PubMed] [Google Scholar]
  36. Rubin DB. (1987). Multiple imputation for nonresponse in surveys. John Wiley and Sons. [Google Scholar]
  37. Schnell S, Friedman SM, Mendelson DA, Bingham KW, & Kates SL. (2010). The 1-year mortality of patients treated in a hip fracture program for elders. Geriatric Orthopaedic Surgery & Rehabilitation, 1(1), 6–14. 10.1177/2151458510378105 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Seguin R, Lamonte M, Tinker L, Liu J, Woods N, Michael YL, Bushnell C, & Lacroix AZ. (2012). Sedentary behavior and physical function decline in older women: Findings from the women’s health initiative. Journal of Aging Research, 2012, Article 271589. 10.1155/2012/271589 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Shojaa M, Von Stengel S, Schoene D, Kohl M, Barone G, Bragonzoni L, Dallolio L, Marini S, Murphy MH, Stephenson A, Mänty M, Julin M, Risto T, & Kemmler W. (2020). Effect of exercise training on bone mineral density in post-menopausal women: A systematic review and meta-analysis of intervention studies. Frontiers in Physiology, 11, 652–652. 10.3389/fphys.2020.00652 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Spiteri K, Broom D, Bekhet AH, Caro J.X.de, Laventure B, & Grafton K. (2019). Barriers and motivators of physical activity participation in middle-aged and older adults—A systematic review. Journal of Aging and Physical Activity, 27(6), 929–944. 10.1123/japa.2018-0343 [DOI] [PubMed] [Google Scholar]
  41. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE). (n.d.). STROBE. Retrieved June 18, 2024, from https://www.strobe-statement.org/
  42. Therneau TM, & Grambsch PM. (2000). Modeling survival data: Extending the Cox model. Springer-Verlag. 10.1007/978-1-4757-3294-8 [DOI] [Google Scholar]
  43. Therneau TM, Lumley T, Elizabeth A, & Cynthia C. (2024). Survival: Survival Analysis (Version 3.8-3) [Computer software]. https://cran.r-project.org/web/packages/survival/index.html [Google Scholar]
  44. Thomas E, Battaglia G, Patti A, Brusa J, Leonardi V, Palma A, & Bellafiore M. (2019). Physical activity programs for balance and fall prevention in elderly. Medicine, 98(27), Article e16218. 10.1097/MD.0000000000016218 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. van Buuren S. (2012). Flexible imputation of missing data (Vol. 10). CRC Press. https://www.jstatsoft.org/htaccess.php?volume=093&type=b&issue=01&filename=paper [Google Scholar]
  46. van Buuren S, & Groothuis-Oudshoorn K. (2011). MICE: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45, 1–67. 10.18637/jss.v045.i03 [DOI] [Google Scholar]
  47. Wang Q, Jiang X, Shen Y, Yao P, Chen J, Zhou Y, Gu Y, Qian Z, & Cao X (2020). Effectiveness of exercise intervention on fall-related fractures in older adults: A systematic review and meta-analysis of randomized controlled trials. BMC Geriatrics, 20(1), Article 322. 10.1186/S12877-020-01721-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Warriner AH, Patkar NM, Curtis JR, Delzell E, Gary L, Kilgore M, & Saag K. (2011). Which fractures are most attributable to osteoporosis? Journal of Clinical Epidemiology, 64(1), 46–53. 10.1016/j.jclinepi.2010.07.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Willett W, & Stampfer MJ. (1986). Total energy intake: Implications for epidemiologic analyses. American Journal of Epidemiology, 124(1), 17–27. 10.1093/oxfordjournals.aje.a114366 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplementary file
NIHMS2207348-supplement-supplementary_file.docx (171.6KB, docx)

RESOURCES