The 10,000 Step Zombie
People only use 10 percent of their brains. Every year, while sleeping, the average person swallows 8 spiders. And we should be walking 10,000 steps per day.
These are examples of "zombie statistics" – figures that continue to be quoted after being repeatedly debunked. Like zombies, they prove hard to kill.
Although the 10,000 step meme has been discredited more than once, a study published last week seems to have revived its credibility. At least that's what national news organizations and prominent health blogs are claiming.
I want to talk about where the 10,000 step meme came from, why it's a zombie statistic, whether we can trust the new study, and how much we should be walking each day.
The birth of a zombie
The 1964 Tokyo Olympic games sparked a fitness craze in Japan. In 1965, the Yamasa Clock and Instrument Company began selling a pedometer called Manpo-kei, which means “10 000 steps meter."
Why did Yamasa choose 10,000? Because the Japanese character for 10,000 bears a slight resemblance to a walking person: 万
In other words, the 10,000-steps-per-day advice began as a marketing ploy.
This doesn't mean it's bad advice. It just means there's no particular reason to trust it. (The kanji for 1,000 looks like a person with outstretched arms (千), but I wouldn't recommend that anyone stretch a thousand times per day,)
A zombie's journey
The phrase "10,000 steps a day" quickly circulated among Japanese walking clubs and spread to Western countries. Nowadays Fitbit and other wearable devices use 10,000 steps as a pre-programmed fitness goal.
This seems healthy – 10,000 steps would be about 5 miles, depending on stride length – but no evidence ever privileged that specific amount.
A Fitbit blog post does cite a "study" that supposedly "reported conclusive health outcomes." I checked. The study didn't look at health outcomes. Rather, it explored how employees felt about a new corporate health program that incorporated a 10,000 step goal. (They mostly liked it.)
As corporate deception goes, this is small stuff, but there it is, right out in the open: Fitbit feeding data to the living dead.
The demise of a zombie
In 2019, Harvard professor Dr. I-Min Lee and colleagues published a study showing that among older women, taking more steps each day is associated with longer life, but the benefits plateau at around 7,500 steps. (A 2022 meta-analysis of longevity research suggested that the ideal number of daily steps varies with age.) Other studies around this time also showed specific health benefits peaking at well under 10,000 steps per day. For instance, Dr. Ed Coyle at UT Austin and his student Heath Burton found that roughly 8,500 may be optimal for fat metabolism.
In short, the 10,000 steps meme began to seem like an overestimate. The national news picked up on this, and for the past five years or so, dozens of articles and blog posts have assured us that we don't need so much daily walking. As a Washington Post headline put it, "that 10,000 step goal is overblown."
So far, so good: Emerging data contradicts a popular meme, the news and social media take note, people presumably become less stressed about counting steps, and the zombie dies.
And yet...
It lives
Although we've heard for five years that 10,000 steps per day is more than we need, it remains a popular meme, and it's still highlighted in wearables technology and countless fitness programs. And now, according to CNN, Newsweek, and other sources, a new study shows that 10,000 steps per day is the optimal amount after all.
Evidently we never quite killed off this particular zombie.
What does the new study actually tell us? Did the media portray the findings accurately? How many steps per day should we be aiming for?
The new study
This study, published in the British Journal of Sports Medicine last week, was led by Dr. Matthew Ahmadi at The University of Sydney.
The methods were relatively straightforward: Just over 70,000 participants were sampled from the UK Biobank, an enormous, ongoing longitudinal project. Daily step counts were obtained from each participant during a one-week period between 2013 and 2015. At that time, participants were around 60 years of age on average and had no cardiovascular disease or cancer.
By 2021, some participants had died or developed cardiovascular disease. The researchers found that the more steps people had previously taken each day, the lower their risk of either outcome. The lowest risk was observed for the range of about 9,000 to 10,500 steps per day. And, walking benefitted people about the same regardless of how sedentary they were.
(In other words, in 2013-2015, some people were spending much of the day sitting or otherwise sedentary, while others were more active, but regardless of activity level, the more steps they took when they walked, the lower their risk of death or cardiovascular disease in 2021.)
Media coverage of the study treated it as supportive of the 10,000 step-per-day meme. This is slightly misleading, given that the optimal value ranged from 9,000 to 10,500, but it's not altogether inaccurate.
Should we conclude then that 10,000 steps, more or less, is a worthwhile goal?
Measurement issues
I have concerns with the new study, as you may have gathered from the zombie references.
I'll describe two measurement-related issues here, and two more in the Appendix.
1. The snapshot issue.
Daily steps were measured by an accelerometer, a wrist-worn device that records accelerating movement in any direction. Accelerometer data, appropriately massaged, is assumed to estimate step counts.
In this case, the data were recorded during a one-week period sometime between 2013 and 2015. During this time period, participants were asked to wear an accelerometer 24/7 for exactly seven days. Average daily step counts were then used to predict cardiovascular disease and mortality in 2021.
How can you generalize about a person's daily step counts over a period of years, based on a single week's worth of data?
I reached out to the lead author, Dr. Matthew Ahmadi, for clarification. Although acknowledging that walking and other daily activities change with age and in response to certain life events, Dr. Ahmadi also asserted that
"Overall, among adults physical activity patterns remain relatively consistent because day-to-day routines remain consistent."
He added that, as in earlier studies, his team demonstrated such consistency among a subset of study participants:
"There were a subset of participants (n=3,400) with a follow-up measurement 4 years later and we saw strong consistency in their daily step count averages (Kendall’s W = 0.74)."
In short, a week's worth of daily step counts should extrapolate fairly well across a period of years, according to Dr. Ahmadi.
I believe the data are more squishy than he suggests. Day-to-day routines don't remain consistent for everyone, and that Kendall’s W value of 0.74 does allow for quite a bit of variability. (Kendall's W evaluates rankings – higher values in this case only tell us that more active people at Time A tend to be more active at Time B, relatively speaking.)
2. The ontological issue.
Even if the data are squishy, the researchers did find a link between daily steps and later health. Doesn't that mean their approach to measuring steps was good enough?
Well, what exactly did they measure? Accelerometers don't record steps directly (see here). Maybe what these devices picked up on was general health, or "vigor", or something like that. (Ahmadi and colleagues didn't use exercise or any other indicator of activity as a covariate.)
Let's turn it around: The researchers found that people who recorded the least number of steps were at greater risk for cardiovascular disease or death. Is that because they walked less, or because they had some other, unmeasured health problem responsible for both walking less and their later health problems?
Zombies are not good for public health
The concerns I've raised here (and two more discussed in the Appendix) question the accuracy and meaning of step counts in this study.
I would describe my concerns as minor if the researchers simply concluded that the more people walk, the better the their health. Instead, they tied their conclusions much more concretely to step counts (e.g., by stressing the optimality of the 9,000–10,500 range). At the end, they make it clear that they want their results to be taken concretely:
"These findings provide tangible targets that can be easily implemented in future steps-based and sedentary time-based interventions, and can inform the first generation of device-based guidelines."
With great power comes great responsibility. Only by using statistical techniques developed in the 20th century (and computers developed in recent decades) could the researchers have even begun to link step counts to health outcomes among 70,000+ people. But if we can't trust their measurements of step counts, it seems unwise to recommend those counts as a basis for actual fitness guidelines.
Conclusion and practical advice
In spite of what some news reports and blogs have claimed, the new study doesn't show that 10,000 steps per day is optimal for cardiovascular health and longevity, or even that 9,000 to 10,500 steps per day is optimal. I don't think we can trust the measurement of step counts.
What seems more credible is the vaguer conclusion that up to a point, the more steps one takes, the better.
I realize this is the sort of conclusion that pleases no one. Obviously Dr. Ahmadi and his team prefer their more concrete and precisely quantified interpretation. The harshest critics of the study would say that the data tell us nothing. In between those extremes, one might say: We already know that moving around is good for people. What's the point of concluding, vaguely, that up to a point, more is better?
In this instance, vagueness is more faithful to the data, and probably better for your health.
In a March 11 New Yorker cartoon, Roz Chast depicts a man sauntering down the street, whistling, and looking utterly carefree. Here's the caption:
"The guy who didn't know his cholesterol, his B.M.I., his net worth, his I.Q., his credit score, his astrological sign, or his ancestry."
Touché. Most of the variables on Chast's list are 20th century constructions that rely to a greater or lesser extent on modern statistics. The premise is that our historically unprecedented access to personal data can be stressful. In fact, I do know people who become a little anxious when they don't get their 10,000 steps in.
Here's what I would tell these people: Relax. Walk or jog when you can, regularly if possible. Don't count steps if it stresses you out. If you do count, taking 6,000 steps and feeling good about it might be healthier than taking 9,000 and worrying that you missed your goal. After all, decades of research show that stress is bad for cardiovascular health. Meanwhile, recent studies suggest health benefits from walking far less than what that 10,000 step zombie advises.
So, what's the minimum number of steps that would yield better health? What's the maximum, beyond which no additional benefits accrue? It's hard to say – estimates vary widely depending on study and type of health outcome. As a New York Times article pointed out this morning, we shouldn't just be counting steps anyway. How fast you walk, what you're carrying, and the flatness of the terrain matter too. Whatever you do, it's probably best done in a peaceful and relatively data-free state of mind.
Thanks for reading!
Appendix: Additional measurement issues
Two further issues raise doubts about the quantification of steps in the new study. (I'm including these in case the zombie hasn't been fully dispatched.)
1. The zero-sum issue.
Dr. Ahmadi and colleagues used accelerometer data to divide the sample into low vs. high in sedentariness. The researchers found that for each group, more steps lowered the risk of cardiovascular disease and mortality, up to an optimal range of 9,000 to 10,500 steps per day.
I found this baffling. When a person is sedentary, they're not walking around. When they're walking around, they're not sedentary. How could you say, for example, that among people who are very sedentary, the more they walk, the better their health? If they're walking more, they're not sedentary.
I reached out to Dr. Ed Coyle at UT Austin, a leading expert on the relationship between fat metabolism and inactivity, and a co-author of research showing, for example, that less than 10,000 steps per day can be metabolically optimal. He seemed baffled too. As he noted,
"Trying to separate effects of the two (steps and sedentariness) would seem difficult in an epidemiological study with high variability."
The only way this makes sense is to assume that some people walk more rapidly than others. For instance, given two highly sedentary people who sit for 14 hours per day and each walk for about half an hour total, one person might take more steps during that half hour period than the other one.
There's some plausibility to this interpretation, but Dr. Ahmadi and colleagues don't elaborate, much less explain how members of the highly sedentary group could manage to get in 9,000 or more steps per day.
2. The uptick issue
Below are two figures obtained from the new study. Without getting into technical issues, the y-axis on each graph is estimated risk, and the x-axis is daily step counts.
A curvilinear relationship between risk and steps is evident in each figure. Starting from the far left (the lowest number of steps per day) and moving to the right, you can see that as people take more steps, risk declines. The lowest risk is at around 9,000 to 10,500 steps. Then the line curves upward again.
The researchers say absolutely nothing about that uptick. I'm not sure what to make of it myself, but it does raise questions.
9,000 to 10,500 steps is roughly equivalent to 4 or 5 miles. I understand that walking this much would have more cardiovascular benefits than, say, walking 1 or 2 miles per day. But why would the benefits start to diminish among people who are walking 5 or 6 miles? Why would 6 or 7 miles be less helpful than 5 or 6?
You might say: It's an older sample. By 2021 the average age was mid- to late-sixties. Additional activity taxes their cardiovascular systems.
This makes sense to me in theory but not in practice. As a runner and a hiker myself, I gather that people who can run or walk 5 miles every day vs. 7 miles every day all tend to be in good shape. It's hard to image that one group incurs more cardiovascular risk than the other one.
I don't know what's responsible for the uptick, but given the concerns I've raised in this newsletter about measurement of step counts, I would not conclude that reasonably healthy people should limit themselves to 10,500 steps per day. Here again, overemphasis on specific numbers has public health implications.