A 10% increase in risk of XYZ means if you have a 0.001% of XYZ happening to you, then a 10% increase takes you to 0.0011%.
Also, what is up with "Participants were followed for a median of 12.38 years" and "methodological limitations were... measurement imprecision due to having only 7 days of accelerometer wear." Does that mean this study only looked at 7 days of data then 12 years later they checked on who died? Wow. Sort of makes the whole thing meaningless (and if so then fuck you to all the news articles implying it was 12 years of data). As if people never change their habits nor activity levels from week to week.
I bet the data is skewed to people who recently died too. Someone dying from cancer isn't going to be very active (though how many wear health monitors?). If those people were bed ridden that week then died within the next year or two and everyone else lived, that could skew the results. Though they claimed to factor in "health-status factors" so maybe they took that into account?