Peak intensity tracked survival better than volume

Adults who have diabetes and hypertension together carry a substantially elevated risk of dying from any cause, and physical activity is one of the few genuinely modifiable factors in that picture. The problem is measurement: self-reported questionnaires and crude categories such as active versus inactive may miss variation that actually matters clinically. Wrist-worn accelerometers make it possible to separate how much someone moves in total from how hard they move at their best. This study asked whether those two continuous dimensions carry different information about mortality risk in this high-risk group.

The researchers analysed two population-based cohorts with device-measured wrist accelerometry. From UK Biobank they took 4,294 adults with both diabetes and hypertension plus 1,482 with diabetes alone, followed for a median of 85.2 months, during which 731 deaths occurred. From NHANES 2011-2014 they took 530 adults with both conditions and 235 with diabetes alone, followed for a mean of 63.5 months with 123 deaths. Two continuous metrics were derived: daily total activity as a volume measure, and Peak-30 as a peak intensity measure. Analyses used multivariable Cox regression, restricted cubic splines and gradient-boosted models with SHAP attribution.

Both activity dimensions were independently associated with all-cause mortality in the group with both conditions across the two cohorts. In UK Biobank the fully adjusted hazard ratios were 0.66 for daily volume and 0.43 for peak intensity. In NHANES the peak intensity association remained robust at 0.57 with p < 0.001, but the daily volume measure attenuated to non-significance at 0.95 with p = 0.116 after full adjustment. Non-linear dose-response modelling identified lower inflection points for peak intensity in the combined-condition group than in those with diabetes alone, and the SHAP analyses showed peak intensity making stable protective contributions across comorbidity strata.

This supports a growing line of accelerometry research suggesting that how hard someone moves adds information beyond simple daily totals, and it extends that work to a specific high-risk comorbid group. The divergence between the two cohorts on daily volume complicates a simple more-is-better reading and shows how sensitive volume measures are to adjustment and to the metric used. Because the design is observational, people who are already ill may move less and die sooner, which can manufacture exactly this pattern. Whether raising peak intensity actually lowers risk would need a trial, not a cohort.

This is a reason to think that the hardest sustained stretch of your day may carry information that step counts and total movement do not, particularly for people managing diabetes and high blood pressure together. It is not evidence that deliberately pushing intensity harder will extend life, because nobody was randomised to anything here. The finding also hints that risk thresholds may differ depending on which conditions someone has, rather than one activity target fitting everyone. Treat it as a case for wearables capturing more than volume, not as a training prescription.