Methods

The WHISH trial protocol (Appendix A in the Supplementary Materials) was approved by the Institutional Review Boards of the WHI Clinical Coordinating Center (CCC) at the Fred Hutchinson Cancer Center and at Stanford University, and by the NHLBI-appointed Data Safety and Monitoring Board. The authors attest to the accuracy of the data analysis and fidelity of the intervention protocol.

Study population

As previously described,22,23 WHISH trial participants were among 161 808 women who provided written consent, at ages 50–79 years (1993–1998) at 40 clinical centers across the United States, to be randomized to active versus control treatment arms within the WHI Clinical Trial (CT) or to enroll in the WHI Observational Study (OS) through 2005.24 In 2005, 115 407 (77% of eligible) CT/OS participants consented to continued follow-up through 2010, at which time 93 567 (87% of eligible) women consented to ongoing follow-up in the WHI-Extension Study 2 (WHI-ES2; 21). The WHI-ES2 consent enabled collection of health outcomes and all data needed for the WHISH trial.

In 2015, 61 444 of WHI-ES2 participants were alive and active and assessed for eligibility for the WHISH trial. Exclusion criteria included known dementia, living in a nursing home, self-reported inability to walk, or indicating Spanish language preference, based on WHI data available at WHISH randomization in 2015. To be eligible for WHISH, participants also had to have cardiovascular outcomes available through either WHI-ES2 outcomes adjudication21 or linkage to the Centers for Medicare and Medicaid Services (CMS) database (see Consort Diagram, Appendix B). Using a randomized consent design,25 49 331 eligible participants were randomly assigned, within this community-based trial, stratified by 2015 age tertiles (66–76, 77–82, 83–99 years), U.S. census regions (Northeast, Southeast, Midwest, and West), and WHI-ES2 outcomes data source (adjudicated Medical Records or Self-report; 21) to Intervention (N = 24 657) or “usual activity” Comparison (N = 24 674) arms in parallel groups.

The WHI Clinical Coordinating Center at the Fred Hutchinson Cancer Center, Seattle, WA mailed all participants randomized to the WHISH Intervention a study description, an initial “_WHISH_ful Actions” newsletter and the NIA “Go4Life: Workout to Go” pamphlet.26 The mailing included a letter informing Intervention participants that the Stanford University WHISH Team would be contacting them to provide more information unless they called the CCC to let them know that they preferred not receiving any other WHISH materials. Passive consent to receive the PA-I was thereby obtained from 23 653 (95.6%) of WHISH Intervention participants. Names and contact information of Intervention participants who “opted out” of receiving the WHISH PA-I were not provided to the Stanford team. The Stanford team tracked the PA-I participants closely to ensure that PA-I mailings and any other contact with the Stanford team were stopped if a participant requested discontinuation at any time or was learned to have died. Participants in the comparison group were not informed of their role in the WHISH trial.

Data collection and physical functioning assessments

Sociodemographic and lifestyle data were collected by questionnaire at WHI baseline (1993–1998) and during the main WHI study and WHI-ES2 follow-up. History of prior clinical CVD and use of antihypertensive and anti-hyperlipidemic medications were based on data collected at WHI enrollment and through follow-up to randomization in WHISH (2015). All WHISH participants received annual WHI-ES2 mailings with medical history update forms and a form with the 10-item RAND-36 Physical Functioning (PF) scale,27 for which scores ranged from 0 to 100 (in 5-point increments) and were categorized as lower (< 65), moderate (65–89), and higher (90–100) PF.

Primary physical activity assessment and outcomes

Total recreational PA (presented roughly as tertiles: < 3, 3–< 14, ≥ 14 metabolic equivalent hours per week; MET-hrs/wk) and walking (MET-hrs/wk), described in detail previously,28 were collected on a WHI-ES2 form a median of 3.1 years (QR: 2.8–3.3) before randomization into the WHISH trial. After randomization, WHISH participants in both groups received annual PA questionnaires from the WHI CCC, labeled as WHI forms to maintain blinding of randomized trial arms (Appendix C1; revised in 2020, Appendix C2) which ascertained walking,28 sitting (hours per day), and other PA behaviors based on a modified Community Healthy Activities Model Program for Seniors (CHAMPS) questionnaire,29,30 including aerobic and total PA (MET-hr/wk), muscle strengthening (sessions per week), and sitting (hours per week).

The physical activity intervention and annual PA-I surveys

The WHISH PA-I was a remotely-delivered “light touch” intervention, ie with no individualized customization or in-person physical activity coaching, in keeping with the pragmatic trial design. It was informed by state-of-the science behavioral theories, including Social Cognitive Theory,31 the Transtheoretical Model,32 and Self-determination Theory.33 Intervention messages aimed to motivate participants to increase their regular PA by any amount and informed them of national physical activity recommendations for older adults, ie, to accumulate 150 minutes a week of moderate-intensity aerobic activity, if possible, or aim to achieve PA levels corresponding to their current physical function status.1,2

Multi-component physical activity was emphasized with strong encouragement to walk and reduce sedentary behavior, particularly (inactive) sitting, every day, and do both muscle-strengthening sessions, including upper body and lower body exercises, and balance and flexibility activities at least two days a week. The PA-I was delivered primarily through quarterly (seasonal) newsletters, usually accompanied by inserts that provided instructions on specific exercises that were recommended or reviewed and approved by a physical therapist with geriatric expertise. Other communication channels included additional print materials, monthly automated outbound telephone messages, monthly emails (if participants provided email addresses), and a WHISH website.

The first three “_WHISH_ful Actions” newsletters were mailed within a 2-month window. These mailings included the NIA Go4Life® Exercise & Physical Activity: Your Everyday Guide book,34 WHISH-branded pedometers and easy-to-use PA tracking calendars with instructions on how to use them. Subsequent newsletters with “seasonal” themes were mailed every 3–4 months. The themes and content summary of the 36 _WHISH_ful Actions newsletter packets mailed over the 9-year intervention appear in Appendix D. Mailings also provided WHISH-branded resistance bands with specific exercise recommendations, pedometer belts, sun visors, neckerchiefs designed to promote specific exercises, fanny packs, and “Stick-with-it” notepads of postable logs, designed to create intentional weekly action plans over the course of the intervention. Participants could request replacements of any items at any time.

Brief (< 1 minute) motivational phone messages were delivered monthly by an automated telephone system using a familiar human voice, “Adriana,” unless a participant requested to not get these calls (requested by fewer than 2%). PA-I participants who provided email addresses, ie, 25% at the outset of the intervention, increasing to 35% by the final 2–3 years, received monthly emails with informational messages and links to vetted sites, such as _Go4Life®_35 and Move Your Way.36 A WHISH website provided additional resources and exercises and a PA self-tracking tool. Participants could also access an algorithm-driven interactive voice response (IVR) system, designed specifically for WHISH PA-I participants to offer tailored advice and support from “Adriana” and provided personal goal-setting features. Participants could also reach staff through email, phone, and postal mail throughout the intervention.

Participants were encouraged to complete action plans regularly, do self-assessments of physical function (and do specific exercises to improve their scores), come up with 2 minutes of an activity to replace sitting time throughout the day, complete 24-hour activity graphs (of their hours spent sleeping, resting or inactive sitting, active sitting and light activity, and moderate-to-vigorous activity), design exercises to be done in each room of their home, map out walking destinations of various distances from their home, complete Bingo-like cards of multi-component activities, lay out values that motivate them to move, and other projects designed to help them create personalized physical activity programs.

An array of motivational strategies to bring about PA change were encouraged through the above remote communication channels,37,38 including providing informational content (eg, news and research stories), emphasizing affective (thoughts and feelings) and social (friends, family, and caregiver activities) themes, and suggesting environmental cues and contexts (eg, coping with weather, strategically posting PA reminders, etc.). PA-I participants also served as “coping” role models by sharing personal stories, photos, and videos about overcoming barriers to PA (eg, recovery from injuries, illnesses, and personal losses) in “Life Happens” newsletter features and on the WHISH website.

The intervention was adapted in response to participant input, obtained primarily through annual PA-I surveys that were designed to solicit information on PA-I participant activity and exercise preferences, personal motivational drivers, barriers to physical activity and reactions to intervention materials and messages. The number (and percent) of respondents to each mailed or online survey, postcard, and other engagement materials were tracked closely, and reports were generated by age strata, race and ethnicity, and other factors to fine-tune the intervention, as warranted. Survey data were shared with PA-I participants in “All About You” newsletters.

Early feedback suggested that exercise materials were too challenging for some participants and not challenging enough for others. Therefore, starting in Fall 2018, alternative inserts were developed to present either less challenging or more challenging exercises than the “generic” insert. The 19 201 PA-I participants who were still receiving the WHISH PA-I at that time were triaged, based on self-reported PF and PA levels on recent WHI/WHISH forms, to receive the “Lower” (36.8%), “Middle” (31.9%), or “Higher” (31.3%) level PA-I inserts. Survey response rates were subsequently tracked by these “targeted” PF/PA insert groups.

Another example of adapting the intervention was a greater focus on balance after the majority of respondents to the 2021 Survey #6 indicated that they would like to make changes in their current physical activity. Improving balance was the most common specific change women in all three target groups checked, among five (“all that apply”) choices, followed by, in the same order for all target groups, strength, endurance, flexibility, and time spent sitting. From December 2022 through 2023, newsletters and targeted exercise inserts emphasized balance and how balance, strength, flexibility, and endurance related to mobility.

Statistical analysis of physical activity by randomized arm

Demographics at WHISH randomization by trial arm, and for the PA-I subset of women, are presented with means and standard deviations for continuous characteristics and frequencies and percentages for categorical characteristics, with differences between arms assessed using t-tests and chi-square tests, respectively.

Trial PA and sitting per participant were calculated as the average of all yearly measurements over the course of the trial. Differences in participant average PA measures and sitting time are presented by trial arm with Intervention—Comparison percent differences and corresponding 95% confidence intervals from a linear regression model with the log-transformed trial average participant PA or sitting time outcome as a function of trial arm, adjusted for age strata, census region, and WHI-ES2 outcomes data source.21 The resulting absolute outcome difference between arms is calculated using the model-derived percent differences at the mean comparison outcome level.

To determine how differences in trial average PA and sitting outcomes by trial arm varied by age, subgroup models were fit to log-transformed outcomes as a function of trial arm, age group, and their interaction. Interaction p-values are from a separate linear regression model with the log-transformed outcome as a function of trial arm, linear trend over age groups, and their interaction. All linear models were adjusted for census region strata and WHI-ES2 sub-cohort.

Trial average PA and sitting outcomes were analyzed overall and by pre-randomization PF and PA level tertiles using least squares means at each calendar year, with means derived from a linear regression model with the outcome as a function of trial arm, adjusted for age strata, region strata, and WHI-ES2 outcomes sub-cohort.

Analyses were conducted in SAS for Windows version 9.4 (SAS Institute, Cary, NC) and R version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria).