Practice Point: One-two days per week of walking at least 8,000 steps per day might be enough to decrease all-cause mortality.
EBM Pearl: Hazard ratios are more appropriate than adjusted risk differences (aRDs) when analyzing time-dependent observational data.
Getting people to exercise is a challenge. Common sense dictates that any exercise is better than none, but does the weekend warrior approach work? We still don’t have good evidence on how many days per week one must walk to optimize long-term health. To that point, a study in JAMA looked at the association between days per week walked and all-cause mortality risk at 10 years.
Researchers linked 2005/2006 National Health and Nutrition Examination Survey (NHANES) data to the National Death Index through 2019 for 3,101 adult participants who wore an accelerometer for at least four days during the study period. Participants were categorized into groups based on the number of days (0, 1-2, and 3-7) they walked at least 8,000 steps during the one week of accelerometer data collection and were evaluated for all-cause and cardiovascular disease mortality. That’s correct — this whole study examining ten-year mortality trends is based on a single week of self-reported data. The dose-response association was also examined for the number of days per week participants walked 8,000 steps and mortality.
At ten years, all-cause mortality risk was significantly lower among those who took at least 8,000 steps one to two days per week (adjusted risk difference [aRD] = -14.9%) and three to seven days per week (aRD = -16.5%) on the absolute scale, compared with no days per week. Similarly, cardiovascular disease-related mortality risk was 8.1 percent lower among those who took at least 8,000 steps on one or two days each week and 8.4 percent lower in those who took this number of steps on three to seven days per week.
The relationship between walking frequency and mortality risk followed a curvilinear pattern, with a rapid decrease in risk up to three or four days of walking per week, after which the risk plateaued.
So, do we think walking 8,000 steps one or two days per week is sufficient to tack on a few extra years to your life? Maybe, but not necessarily based on this study due to a few important problems.
The first is “the chicken and the egg” conundrum: when we observe a population, those who exercise are already healthier and more likely to live longer. This was not a randomized trial, and the initial observation was effectively looking at three different populations. Participants who had a higher frequency of those 8,000 or more-step-days at the beginning of the study tended to be younger, nonsmoking, less likely to have obesity or other medical conditions, and/or rate their health well, all factors known to be associated with better long-term health independent of activity level.
Second, adjusted risk differences (aRDs) simply aren’t the appropriate statistical tool to use here. Hazard ratios would have been better since they account for time-to-event, whereas aRDs quantify the absolute disparity in risk between the exposed and unexposed groups.
We all know that correlation does not equal causation. The three groups studied seem to be fundamentally different, which makes them hard to compare. It’s also hard to say that one week of data is truly a reflection of a person’s step count over the next ten years. That said, there is likely little harm in recommending patients to strive to walk at least 8,000 steps once weekly. And citing this study to our patients might be a good sell to help get them moving.
For more information, see the topic Adult Preventative Health in DynaMed.
Reference: JAMA Netw Open. 2023 Mar 1;6(3):e235174
