Determinants of physical activity adoption and maintenance in older adults: a dual process approach

Key points

1. 01Behavior-specific reflective determinants (PA intentions, self-efficacy, plans) were the most consistent within-person predictors of physical activity across both adopters and maintainers
2. 02Associations between PA intentions, self-efficacy, and activity were stronger among PA maintainers than adopters
3. 03Automatic determinants (affect, physical context, routine) more consistently predicted how much PA occurred rather than whether PA occurred at all
4. 04Being outdoors at a given moment was positively associated with both MVPA and steps for all participants; being alone was linked to more MVPA but fewer steps
5. 05General reflective factors (demands, deliberation, self-control) were more predictive of step counts than MVPA

How it was conducted

Design

Ecological Momentary Assessment (EMA) measurement burst study with three 2-week data collection waves spaced approximately 6 months apart over one year

Participants

202 community-dwelling older adults aged 60 years or older, residing in Guilford County, NC, who reported at least 30 minutes of MVPA per week

Physical activity measurement

Actigraph GT3X accelerometer (MVPA) and ActiPAL micro 4 device (step counts), worn for 14 days per wave

EMA protocol

10 randomly delivered prompts per day on 4 days per wave, each with 20 items assessing reflective and automatic determinants

Group classification

PA adopters: insufficiently active or meeting guidelines for less than 6 months; PA maintainers: meeting guidelines for 6 months or more, confirmed by device data

Analysis

Multilevel modeling with disaggregated within- and between-person variables; zero-inflated PA data modeled with logistic and gamma regression; false discovery rate adjustment applied

What they found

• Wave 1 sample was 63.2% PA adopters and 36.8% PA maintainers; EMA response rate was 92.1%; retention from Wave 1 to Wave 2 was 89.1% and Wave 2 to Wave 3 was 87.6%
• Within-person PA intentions (beta=0.008, SE=0.001, OR=1.008, p_adj<.001), self-efficacy (beta=0.008, SE=0.001, OR=1.008, p_adj<.001), and plans (beta=0.006, OR=1.006, SE=0.001, p_adj<.001) were positively associated with likelihood of engaging in MVPA regardless of group
• Maintainers engaged in more minutes of MVPA on average compared to adopters (beta=0.329, SE=0.053, IRR=1.390, p_adj<.001)
• Within-person PA intentions and MVPA minutes association was stronger for maintainers vs adopters (p_adj=.005); between-person PA intentions and MVPA minutes was significant for maintainers but not adopters (p_adj=.007)
• Positive affect was positively associated with MVPA minutes among maintainers at the within-person level but not adopters; between-person positive affect was also positively associated with MVPA among maintainers (beta=0.023, SE=0.008, IRR=1.025, p_adj=.007)
• Being outdoors was positively associated with likelihood of MVPA among maintainers at the within-person level (p_adj<.001) but not adopters; being alone was positively associated with MVPA likelihood across all participants (beta=0.182, SE=0.047, OR=1.199, p_adj<.001)
• Maintainers were more likely to engage in stepping behavior compared to adopters (beta=0.538, SE=0.110, OR=1.712, p_adj<.001); when stepping did occur, maintainers took more steps (beta=0.309, SE=0.029, IRR=1.362, p_adj<.001)
• Being outdoors was positively associated with likelihood of stepping (p_adj<.001) and this between-person association was stronger for maintainers (beta=5.563, SE=1.108, OR=260.48, p_adj<.001)
• On occasions when maintainers followed their typical routine more than usual, they engaged in more MVPA minutes (beta=0.005, SE=0.001, IRR=1.005, p_adj<.001) and more steps (beta=0.007, SE=0.001, IRR=1.007, p_adj<.001); for adopters, following routine was negatively associated with steps (beta=-0.056, SE=0.003, IRR=0.994, p_adj<.001)

Limitations

• Sample was predominantly non-Hispanic Caucasian (72.5%) and female (71.5%), limiting generalizability; adults reporting no MVPA were excluded, restricting range
• All EMA determinants were assessed with single items adapted from multi-item scales, which may reduce construct validity
• Classification as PA maintainer required only 6 months of meeting guidelines, which may be insufficient time to develop the dominant automatic processes theorized to regulate maintenance
• Automatic determinants were assessed via self-report of contextual cues rather than objective or implicit measures, and cues were not assessed immediately before PA, potentially attenuating associations

Why it matters

For patients

Older adults trying to build an exercise habit can benefit from strengthening specific intentions and confidence right before planned activity, while those already active may benefit from protecting their routine and monitoring their mood as automatic cues.

For clinicians

Intervention strategies should differ by stage: adopt behavior-specific motivational prompts for new exercisers and focus on routine stability and affective cues for those maintaining activity long-term.

For readers

This study is one of the first to use real-time EMA and accelerometers to map both reflective and automatic moment-to-moment determinants of PA across the adoption-to-maintenance continuum in older adults, providing a novel dual-process framework for intervention design.