Back

Device-measured movement behaviours and cancer incidence in a population sample of UK adults: Dual 24-hour analyses of postural and intensity compositions

Mitchell, J. J.; Biswas, R.; Koemel, N. A.; Ahmadi, M.; Blodgett, J. M.; Canfell, K.; Lee, I.-M.; Friedenreich, C.; Teixeira-Pinto, A.; Cistulli, P. A.; Dumuid, D.; Okely, A. D.; Fisher, A.; Hamer, M.; Stamatakis, E.

2025-10-03 oncology
10.1101/2025.10.02.25337148 medRxiv
Show abstract

IMPORTANCEInsufficient physical activity (PA) and excessive sedentary behaviour is associated with several cancers. Personalised approaches to increasing healthy movement behaviours and reducing unhealthy behaviours may be more effective than a one-size-fits-all approach. Exploring doses, posture allocation and intensities across the 24-hour spectrum of behaviours are important to develop broader guidelines. OBJECTIVEUsing a novel dual-compositional approach, we assessed how time reallocation between 24-hour daily movement (postures and intensity bands) and sleep are associated with PA-related cancer incidence (a composite of 13 sites linked with physical inactivity). DESIGN, SETTING, AND PARTICIPANTSProspective analysis of adults from the UK Biobank accelerometry subsample, followed-up by health linkage. EXPOSURESParticipants daily movement was classified into two 24-hour compositions based on intensity and posture. Composition 1 (posture-focused): sleep duration, sedentary behaviour (SB), standing, moving at any intensity. Composition 2 (intensity-focused): sleep duration, sedentary time (ST), light PA (LPA), moderate PA (MPA), and vigorous PA (VPA). Secondary analysis combined VPA and MPA as moderate-to-vigorous PA (MVPA). MAIN OUTCOMES AND MEASURESPA-related cancer diagnoses were captured from health linkage. Cox-proportional hazards models were adjusted for age, sex, education, smoking, alcohol, diet, parental cancer history, cardiovascular disease and medication use. RESULTSAnalyses included 59,218 (55% female) participants (mean [SD] age: 61.7 [7.8] years (y)), with a median follow-up of 8.0y [IQR: 7.4-8.5y; 464,640 person years] with 2,385 (4%) incident cancer events. Even among the average, active participant, replacing any behaviour with moving was associated with lower cancer risk, e.g. theoretically replacing 15 min of sleep or SB with 15 min of moving was associated with hazard ratios (HR) of 0.98 (95% Confidence Interval (95%CI): 0.97-0.99) and 0.98 (95%CI: 0.97-0.99), respectively. Similar risk reduction required 30 min of additional standing in place of sleep or ST. Regarding intensity, replacing even 3 min of any other behaviour with equivalent VPA was associated with lower risk (e.g. LPA HR: 0.96 (95%CI: 0.93-0.99); MPA HR: 0.96 (95%CI: 0.93-0.99). Greater MVPA too, in place of any behaviour, was robustly associated with lower risk. CONCLUSIONS AND RELEVANCEMoving at any intensity, in place of other postures, was associated with reduced risk of incident cancer. VPA, or MVPA in place of other movement intensities, were most strongly associated with lower cancer risk. Key pointsO_ST_ABSQuestionC_ST_ABSHow may reallocations of time between movement behaviours (postures and intensities) change cancer risk? FindingsAmong 59,218 middle-aged UK Biobank participants, 2,385 individuals developed cancer at any of 13 sites linked to physical activity (PA) over 8 years. Even among the average, active participant, replacing 15 minutes of standing, sedentary behaviour (SB) or sleep with moving, or 90+ minutes SB or sleep with standing, was associated with small (2-3%) cancer risk reduction, while even 3 minutes vigorous PA in place of light PA, sedentary time or sleep was associated with 1-7% lower risk. MeaningSmall changes to daily movement (posture/intensity) may reduce cancer risk.

Published in BMC Medicine (predicted rank #6) · training set

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.