Results

There were no differences between Intervention (N = 24 657) and Comparison (N = 24 674) arms in participant characteristics at the time of randomization, including mean age (79.7 years) or race (86.8% White; 8.8% Black; 1.9% Asian/Pacific Islander; 0.2% American Indian/Alaska Native) or ethnicity (3.7% Hispanic/Latina). Nor did groups differ by pre-WHISH mean total recreational PA (11.5 MET-hr/wk) and walking (4.1 MET-hr/wk), mean RAND-36 PF score (69.9), most recent BMI (mean 28 kg/m2), current smoking (2.5%), history of CVD (13.9%), or use of antihypertensive (73.6%) or antihyperlipidemic (53.8%) medications (Table 1).

 | Intervention (n = 24 657) | Comparison (n = 24 674)
Characteristic | N | % | n | %
Age, mean (SD) | 79.8 | (6.2) | 79.7 | (6.2)
66–76 | 8421 | 34.2 | 8457 | 34.3
77–82 | 8122 | 32.9 | 8098 | 32.8
83–99 | 8114 | 32.9 | 8119 | 32.9
Ethnicity |  |  |  | 
Not Hispanic/Latina | 23 700 | 96.1 | 23 729 | 96.2
Hispanic/Latina | 915 | 3.7 | 900 | 3.6
Unknown/Not reported | 42 | 0.2 | 45 | 0.2
Race |  |  |  | 
American Indian/AK Native | 48 | 0.2 | 56 | 0.2
Asian | 452 | 1.8 | 423 | 1.7
Native Hawaiian/Other Pacific Islander | 16 | 0.1 | 22 | 0.1
Black | 2190 | 8.9 | 2161 | 8.8
White | 21 398 | 86.8 | 21 439 | 86.9
More than one race | 291 | 1.2 | 311 | 1.3
Unknown/not reported | 262 | 1.1 | 262 | 1.1
Region |  |  |  | 
Northeast | 6010 | 24.4 | 6040 | 24.5
South | 7063 | 28.6 | 7045 | 28.6
Midwest | 5737 | 23.3 | 5751 | 23.3
West | 5847 | 23.7 | 5838 | 23.7
Education |  |  |  | 
≤High school/GED | 4249 | 17.2 | 4126 | 16.7
School after high school | 8730 | 35.4 | 8677 | 35.2
≥College degree | 11 523 | 46.7 | 11 746 | 47.6
Body mass index, kg/m2, mean (SD) | 28.0 | (5.8) | 28.0 | (5.8)
< 25 | 8438 | 34.2 | 8418 | 34.1
25–< 30 | 8770 | 35.6 | 8754 | 35.5
≥ 30 | 7434 | 30.1 | 7490 | 30.4
Physical function (0–100), mean (SD) | 69.8 | (25.6) | 70.0 | (25.6)
< 65 | 8240 | 33.4 | 8142 | 33.0
65–89 | 8248 | 33.5 | 8230 | 33.4
≥ 90 | 8023 | 32.5 | 8143 | 33.0
Smoking |  |  |  | 
Never | 12 627 | 51.2 | 12 648 | 51.3
Past | 11 295 | 45.8 | 11 270 | 45.7
Current | 607 | 2.5 | 609 | 2.5
Alcohol drinks/wk |  |  |  | 
< 1 | 14 736 | 59.8 | 14 632 | 59.3
1–< 7 | 6293 | 25.5 | 6404 | 26.0
≥ 7 | 3628 | 14.7 | 3638 | 14.7
Physical Activity (MET-hr/wk), mean (SD) | 11.5 | (12.3) | 11.6 | (12.3)
< 3 | 7117 | 28.9 | 7064 | 28.6
3–< 14 | 7758 | 31.5 | 7657 | 31.0
≥ 14 | 7698 | 31.2 | 7842 | 31.8
Walking (MET-hr/wk),a mean (SD) | 4.1 | (5.8) | 4.1 | (5.8)
CVD | 3469 | 14.1 | 3396 | 13.8
MI | 1159 | 4.7 | 1087 | 4.4
CABG/PTCA | 1683 | 6.8 | 1619 | 6.6
Stroke | 1017 | 4.1 | 1042 | 4.2
HF | 891 | 3.6 | 868 | 3.5
Antihypertensive use | 18 148 | 73.6 | 18 182 | 73.7
Antihyperlipidemic use | 13 238 | 53.7 | 13 318 | 54.0

Demographics are presented by age strata [66–76, 77–82, and 83–99 years] for all WHISH participants (eTable S1 in the Supplementary Materials). The 23 653 PA-I participants who consented to receiving the WHISH PA-I closely resembled the full Intervention group and showed similar patterns across the three age strata as the entire WHISH cohort (eTable S2 in the Supplementary Materials). Follow-up completeness was comparable between randomized groups based on the proportion of returned physical activity questionnaires (Expected 7.1; Intervention 5.4 [73.1%] versus Comparison 5.6 [75.8%] and the proportion of participants returning at least 5 questionnaires (76.8% vs 78.9%)).

Response rates to PA-I annual surveys (corresponding to year of PA-I data collection) were as follows: Survey 1 (2015), 49.1%; S2 (2016), 53.3%; S3 (2017), 63.0%; S4 (2018), 71.8%; S5 (2019), 70.8%; S6 (2021), 67.7%; S7 (2022–2023), 63.8%; and S8 (2024), 64.8%. Response rates were consistently and considerably higher in the youngest pre-randomization age tertile than older age tertiles and considerably lower in the oldest tertile (eTable S3 in Supplementary Materials).

Over a median of 8.7 years, Intervention participants reported 3.24% higher MET-hr/wk of walking compared to Comparison participants (Table 2), which equated to a mean difference of 0.11 (95% CI: 0.06, 0.17) MET-hr/wk, and 1.5% lower hours/week of sitting (p < .001 for both), which equated to a mean difference of −0.76 (95% CI: −1.13, −0.39) hours/week. The Intervention group also reported modestly higher aerobic exercise (3.4%) and total exercise MET-hr/wk (3.7%), and higher muscle strengthening exercise sessions per week (4.4%) (p < .001 for all).

 | Intervention (n = 24 657) | Comparison (n = 24 674) | Intervention vs comparison
Trial average physical activity | Meana | SD | Mean | SD | % difference (95% CI)b | Difference at comparison mean (95% CI)c | p-Valueb
Walking, MET-hr/wk | 3.60 | 4.58 | 3.40 | 4.37 | 3.24 (1.63, 4.87) | 0.11 (0.06, 0.17) | < .001
Sitting, hr/wk | 51.46 | 19.35 | 52.27 | 19.60 | −1.46 (−2.17, −0.74) | −0.76 (−1.13, −0.39) | < .001
CHAMPS physical activity |  |  |  |  |  |  | 
Total exercise, MET-hr/wk | 16.37 | 15.70 | 15.52 | 14.99 | 3.71 (1.78, 5.67) | 0.58 (0.28, 0.88) | < .001
Aerobic exercise, MET-hr/wk | 4.05 | 6.17 | 3.86 | 5.98 | 3.40 (1.58, 5.26) | 0.13 (0.06, 0.20) | < .001
Strength training, sessions/wk | 0.97 | 1.34 | 0.86 | 1.24 | 4.35 (3.30, 5.41) | 0.04 (0.03, 0.05) | < .001

As seen in Figure 1, walking was higher in the Intervention versus Comparison group by the first post-randomization assessment, a median of 3.1 (IQR: 2.8–3.3) years after the pre-WHISH assessment, at which time there were no differences between trial arms. PA declined and sitting time increased steadily in both groups over the median 8.7 years after randomization, with noticeable changes in most measures coinciding with the 2020–2021 COVID-19 pandemic. In fact, walking increased in 2020 in both groups, then resumed its steady decline, while sitting time increased more steeply in 2020, followed by a decrease in both groups, and subsequent resumption of steadily increasing sitting time in the final two years. However, Intervention participants reported modestly higher MET-hr/wk of walking, aerobic and total PA and strength training sessions/week and fewer hours/week of sitting than Comparison across the full length of the trial, including during the 2020–21 COVID-19 era (Figure 1).

Figure 1: Self-reported physical activity outcomes by intervention arm. 1Walking Pre-WHISH measurement is a median (IQR) of 3.1 (2.8–3.3) years prior to WHISH randomization.

There was a significant interaction between the linear trend across age at randomization group and the effectiveness of the intervention in reducing average aerobic exercise and sitting time (Figure 2). In the youngest age group (66–76 years), the Intervention group reported 6.9% more aerobic exercise than the Comparison group, as opposed to the reported I-C differences of 4.2% and −1.5% in the middle (77–82) and oldest (≥ 83) age groups, respectively (p <.001 for interaction). Also in the youngest age group, the Intervention group reported 2.5% less sitting time than the Comparison group, as opposed to the reported I-C differences of 1.4% and 0.3% less sitting time for the middle and oldest age groups, respectively (p .02 for interaction).

Figure 2: Physical activity by intervention arm by age at WHISH enrollment subgroups. 1Mean and standard deviations by arms derived from average participant physical activity/sitting over the course of the trial, weighted on follow-up time. 2Intervention–comparison percentage differences and corresponding 95% confidence intervals are from a linear regression model with the log-transformed trial average outcome as a function of intervention arm, age group and their interaction, stratified by age strata, region strata, and WHI-ES2 sub-cohort. 3Interaction p-Values from a separate model with the log-transformed trial average outcome as a function of intervention arm, linear trend over age group and their interaction.

As seen in Figure 3, the intervention appeared to be effective in increasing walking and all PA measures in women with higher and middle pre-WHISH PF scores (PF ≥ 90 and PF = 65–89, respectively), but not in participants with lower pre-randomization PF scores. The intervention also appeared to be more effective in increasing walking and PA measures in women with higher pre-WHISH PA levels 3–< 14 and ≥ 14 MET-hr/wk, but not in participants who reported lower (< 3 MET-hr/wk) pre-randomization PA levels (eFigure S1 in the Supplementary Materials). Overall, the intervention appeared to increase the number of strength training sessions/week in all PF groups (Figure 3) and in all PA groups (eFigure S1 in the Supplementary Materials) over the course of the trial.

Figure 3: Self-reported physical activity outcomes by intervention arm and physical function (0–100) at WHISH enrollment. 1Walking Pre-WHISH measurement is a median (IQR) of 3.1 (2.8–3.3) years prior to WHISH randomization.