select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
A study published in the Annals of Internal Medicine looks at one long walk vs multiple short walks and cardiovascular health.
Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:
“This is an intriguing paper, but one where we have to be careful in interpreting the results. That’s because the researchers themselves draw attention to several limitations stemming from the way the study had to be carried out, and the linked editorial1 mentions more. These limitations are not mentioned in the press release.
“So really, I’d say it’s too early to tell how, if at all, these new findings should feed into public health recommendations on physical activity and step counting.
“The study was observational, using data from participants in UK Biobank, and it is its observational nature that leads to most of the difficulties of interpretation.
“The participants were recruited to UK Biobank between 2006 and 2010. Then in 2013 to 2015, over 100,000 Biobank participants wore an accelerometer for up to 7 consecutive days, and that provided records of the number of steps they took each day, together with when they took the steps. The great majority of participants who provided data for this new research did provide data for 7 days, though some that provided data for fewer days (in extreme cases as few as 3) were also included. The participants were then followed up for an average of about 8 years, and the researchers recorded which of them died (from any cause), and which had diagnoses of cardiovascular disease (CVD, that is, heart attacks, strokes, angina etc.). People who already had diagnosed CVD or cancer were not included in the study, and nor were people who died or had a CVD diagnosis in the first year of follow-up.
“Also the study included only participants who recorded fewer than 8000 steps on average per day. The researchers refer to these as being ‘suboptimally active’, though the linked editorial reminds us that the commonly quoted figure of a minimum of 10,000 daily steps for health was not based on scientific data but, originally, on advertising copy by a Japanese pedometer manufacturer. Also both the research paper and the editorial point out that recommendations on daily stepping targets are emerging only now.
“These exclusions from the study mean that the research is based on data from about 33,500 people.
“The researchers were interested in whether participants accumulated their steps in short or in longer bouts of time. For most of their data analyses, the researchers divided the participants up into 4 groups. The group with the shortest pattern of stepping bouts, referred to as ‘under 5 minutes’, accumulated most of their steps in bouts lasting 5 minutes or less. This doesn’t mean that all their stepping bouts were under 5 minutes long – some would have been considerably longer for some people – but that most of the steps were from bouts of less than 5 minutes. Then there was a group where most steps were in bouts of between 5 and 10 minutes, a group where most were in bouts of 10 to 15 minutes, and finally one where most were in bouts of at least 15 minutes.
“The study is observational because the researchers did not instruct the participants to take given numbers of steps. The participants did what they would do anyway, and the step counts and patterns were observed and recorded. And an issue with pretty well all observational studies is that it’s hard to be sure about what is causing what.
“The researchers found that people in the groups who accumulated more of their daily steps in longer bouts had better health outcomes than those who more commonly took shorter bouts of steps. Those who took their steps in longer bouts had lower death rates during the follow-up period, and lower risks of being diagnosed with CVD.
“But there were many other differences between participants in the four step length groups, apart from the length of stepping bouts. For example, those in the under 5 minutes group walked less, were more sedentary on other measures, were more likely to be overweight, and were less likely to have a university degree, than those in the groups with longer stepping bouts. Those factors have been associated with higher mortality and worse health in other studies. So it could be possible that the association between step bout lengths and health is caused by one or more of these other differences, rather than poorer health being caused by the pattern of lengths of stepping bouts.
“Of course, the researchers were aware of this. It’s possible to make statistical adjustments to try to allow for such possible effects of other factors, and the researchers did this using a commonly used method. They made adjustments for the factors I just mentioned, and many others, including the total number of steps people took daily, their age, their sex, their ethnicity, and more. Even after these adjustments, the association between the patters of stepping bout lengths on the one hand, and mortality and CVD diagnoses on the other hand, still remained.
“Some of the differences in risk are considerable. For instance, after the statistical adjustments, about 44 people per thousand participants in the under 5 minutes group died during the 8-year average follow-up, compared to about 8 people per thousand in the group who took most of their steps in bouts of 15 minutes or more. This seemed not to be just because the 15 minute group took more steps daily in all (though in fact they did), because the researchers had made a statistical adjustment for the number of daily steps.
“However, we still can’t be certain that the differences in mortality and in CVD risk are caused by the difference in step patterns. The researchers adjusted for data on many factors that could have been associated with both step patterns and with health, but we just can’t be at all certain that all important factors were taken into account. The Discussion section of the new paper indeed gives several possibilities of factors that couldn’t be taken into account but might be important. And the linked editorial points out that the findings of this new study aren’t in line with another study2 on a related subject, though that study uses very different methods from the new one.
“The new study uses a method of checking whether all relevant factors have been adequately taken into account, known as a negative control outcome. This involves thinking of a health outcome that is, in the researchers’ view, very unlikely to be affected by the length of stepping bouts, and then carrying out the same statistical analysis as was used for the outcomes (overall mortality and CVD diagnoses) for which the researchers though there might be an association with step bout length. If no association was found between step bout length and the negative control outcome, that’s an indication (though still not a foolproof one) that no important factors have been left out. In this study, the negative control outcome was deaths or hospital admissions from accidents.
“The results of that analysis are intriguing. The researchers did find an association between the step bout length grouping and the risk of accidents, though one that’s much weaker than the associations with all-cause mortality and with CVD diagnoses. That could be because some important factor has been left out (though this method does not tell you what factor it might be), or it could be that in fact there is a reason why the risk of accidents could be partly caused by people’s patterns of step bout lengths, despite the researchers believing that that’s unlikely. So really it just leaves us in the territory we were already in, that this new study provides evidence that different step bout length patterns might affect mortality and CVD risk, but we absolutely can’t be certain about cause and effect.
“Another complication in a study like this is what’s called reverse causality. Is it possible that some types of ill health cause people to have different patterns of step length duration, and that these aspects of ill health lead, later, to higher mortality rates and higher CVD risk? In other words, could the pattern of cause and effect have ill health as the cause and step bout durations as the effect, rather than the other way round?
“In fact the researchers incorporated features in the design of their study that make it less likely that reverse causation could be going on, like excluding participants who already had CVD or cancer, and participants who died or had a new CVD diagnosis in the first year of follow-up. Also, they performed various extra statistical analyses that could possibly have picked up reverse causation if it was happening. This is all good research practice – but it can’t entirely rule out reverse causation as a possibility, as the researchers admit.
“Finally, the researchers draw attention to another limitation. All the observation of each participant went on for a long time. The participants entered the UK Biobank study between 3 and 9 years before their step patterns were recorded, and some of the factors used in the statistical adjustment would have been recorded only when they entered. Their step patterns were recorded only over one period of one week (or sometimes less). Their health was followed up for 8 years after that, but stepping was not recorded again. So the study could not pick up anything about changes in stepping patterns, or in factors that were recorded only on entry. The researchers present an argument that other work has indicated that most of these things do not in fact change much over time – but again it’s hard to be sure.”
1 Sanchis-Gomar, Lavie and Banach, ‘Stepping Up Our Game: Longer Bouts of Activity to Boost Longevity’, doi: 10./7326/ANNALS-25-03529
2 Brown, Henderson, Stern and Carson, doi: 10.1177/15598276241253160
‘Step Accumulation Patterns and Risk for Cardiovascular Events and Mortality Among Suboptimally Active Adults’ by Borja del Pozo Cruz et al. was published in the Annals of Internal Medicine at 21:00 UK time on Monday 27 October 2025.
DOI: 10.7326/ANNALS-25-01547
Declared interests
Prof Kevin McConway: “I have no relevant interests to declare.”