Tip:
Highlight text to annotate it
X
>>
MATT: Eric Ding is a nutritionist and epidemiologist; he's a faculty member at Harvard Medical School
and Brigham and Women's Hospital; and founder and chair of the campaign for cancer prevention.
His research primarily focuses on obesity and nutritional risk factors for diabetes,
heart disease and cancer, as well as the translation of research for population wide prevention.
After completing his undergraduate degree at the Johns Hopkins University with honors
in public health and election of Phi Beta Kappa, he earned his dual doctorate in epidemiology
and doctorate in nutrition at the age 23 from Harvard University where he was the youngest
ever graduate from his doctoral programs. At Harvard, Eric has taught and lectured in
more than a dozen graduate and undergraduate courses for which he has received the Derek
Bok Distinction in Technology Award from Harvard College. In addition to teaching, he has published
in the New England Journal of Medicine, in the Journal of American Medical Association,
and researches global disease burden and nutritional risk factors for chronic diseases at Harvard.
His more than three dozen publications have received over 650 external citations, garnering
an H-index scientific impact factor of 11. He currently also serves as senior associate
on the board of the Global Micro-Clinic Project. Eric Ding is an elected fellow of the Paul
& Daisy Soros Foundation for his key role in leading a two-year-long investigation into
the controversial drug safety of Vioxx, a study Dr. Eric Ding published which drew national
media attention. A life-long cancer prevention advocate, he has worked closely with causes
on Facebook to build up the campaign for cancer prevention; currently, with over 5.7 million
members on Facebook. To date, his efforts have raised more than $225,000 for cancer
research. And with that, I would like to welcome Dr. Eric Ding to Google.
>> DING: Thank you, Matt, for the introduction, and thank you. It's a pleasure being here.
Today, I would like to present to you a very interesting study that we did at Harvard.
And it's interesting for many reasons; as all of you know, I'm probably preaching to
the choir, but obesity causes a whole variety of diseases from heart disease to cancer and,
of course, increases mortality. So, if everyone knows this already, then why are we here?
For many reasons. And notably, why are we discussing it in Major League Baseball? And
this is the study we're going to cover that spans 130 years and has--actually has a very
profound impact and results beyond just baseball and beyond just body mass index and mortality,
and you'll see why. So, I have a little disclosure. Babe Ruth, for those who are familiar with
baseball, is one of the most famous baseball players of all time, and he was also famous
for being one of the most obese as well. However, his--he wasn't always quite so obese when
he first joined the Major Leagues. He actually started at 215 pounds, which is a BMI of 27.
We'll go into that in details of the BMI index later. But his--later on he gained substantial
weight and was quite significantly obese at the height of his playing when he was with
the New York Yankees. In more modern times, you may have Barry Bonds. He was very skinny
rookie in 1986, but today, he is known as one of the heavy sluggers of Major League
Baseball. So, body mass index, why does it matter? But there--it's an index of relative
body composition for guiding healthy weight and generally reflects body fat, and generally,
is actually the purpose of our talk here today. However, the calculation is very controversial;
however, the World Health Organization defines normal weight as between 18.5 and 25, overweight
as over 25 and obese as over 30. However, this classification is very controversial
and many people actually dismiss it. So how big of a problem is this? Two in three adults
in America are actually overweight and obese. This is quite a substantial epidemic. And
for those who are familiar with the term "epidemic" but don't actually know the history of obesity
epidemic, let me give you a quick rundown. This is the obesity trends in the U.S. between
1985 and 2009, and it's very self-explanatory. This is in early 80's, late 80's, 90's, the
prevalence of obesity, you can see, is escalating exponentially in every state. In 2001, a new
color is added for Mississippi, the first ever topped over 25%. Then--but it keeps going.
In the past 10 years, we've actually had to add another color, Mississippi leading the
way again. But pretty soon, by 2009, no color is blue anymore, the light blue anymore. And
you can see the comparison, the stark comparison between the early 90's and 20 years later.
And this is where we are, but we all know that. But what's the implication and the linkage
with obesity? So what's the story? What's the story and what's so special about athletes
and baseball players? Here's the problem, BMI is generally recognized as an index of
obesity in the general population, however, everyone knows it's just a simple calculation
of weight and height, and weight is influenced by many things, most notably, muscle mass.
So, if you're an athlete, you run a lot or you weight lift a lot and you have substantial
amount of body mass, you can look like the guy on the left, but if you're sedentary,
you can look like the guy on the right. However, just a doctor taking your height and weight
would not be able to differentiate the two and that's the limitation of the body mass
index because you can have the same weight and same height and same BMI but look completely
differently. And obviously, we would know who is more likely to die. But the question
is, should the athlete on the left ignore his body mass index? Because everyone says,
"Oh, it's all--I'm heavy because I have a lot of, you know, muscle mass. It doesn't
bother me. It doesn't affect me. I don't need to listen to BMI guidelines of obesity." However,
is that true? So the controversy is it's been--long been a measure of obesity, however, it's not
well--generally well-accepted among athletes, and especially those with high lean body mass.
So is it even relevant for athletes in terms of health? So, whether you are a professional
baseball athlete or you're a runner or you're a weight lifter, I think this question applies
to many more of us than we think. So, the objectives of the study. So scientific jargon
is laid out on the left of why we did the study. But plainly, we want to ask, "Are baseball
players getting fatter over time? Does BMI increase the risk of death among these major
league athletes? How does homerun hitters play?" Because the results we get are quite
striking and the homerun hitters, for those who are not familiar, are notoriously the
largest players in the MLB. So the study design--so what we basically did, we took the entire
history of Major League Baseball from 1870's, there's been five major leagues, and we compile
them all together. And we know--basically, baseball, it's a very unique sport where everyone
is a statistics nut and tabulates basically every single statistic ever recorded during
play including height, weight, date of birth, date of debut in the major leagues. And so,
we tabulate from all these major sources including many official databases used by Major League
Baseball, and from there, we calculated body mass index from height and weight in the historical
records. We know exactly the date of birth, day of debut, and that's our baseline. And
we've--then we just ascertained the mortality in all these databases from historical records,
obituaries, and Social Security Death Index. And basically, we have a near perfect follow-up
for baseball players even dating back as to the 1800's because baseball does have that
complete of a history. So in terms of analysis, we did some multivariate regressions, but
most notably, we did time to event analysis where we asked specifically, "Do baseball
players who are more overweight die sooner?" and that's what the relative risk answers.
Basically, does it increase a risk of dying and dying sooner? And again, the baseline
was 1876 and onwards for whenever the baseball player debuted into the major league. And
we have end of follow-up through the end of 2007. The raw counts; over 131 years of Major
League Baseball, we have 7,400 deaths with--among 15,000 Major League Baseball players, which
is pretty much almost all of Major League Baseball, and we have a total of half a million
person-years of follow-up across the 130 years. So the general characteristics, well, body
mass index, 24 like medium weight; you don't expect baseball players to be skinny. But
we, overall, had 43% overweight, but that's across all years. And for the most part, they
were in their 20's and retired just before they're 30. So this is the raw distribution
of BMI in the players across all these years. It's a pretty normal distribution, but the
story is there's much--is much more complex once you break it down by year. However, how
overweight are baseball players? Some people say baseball players are--baseball, you don't
have to be the fittest Olympian to play baseball. But in terms of other sports, we see the NFL,
National Football League, the prevalence of overweight is basically 95% or higher compared
to NHANES which is the U.S. national population, it's about 60%; and then 30% obese. So in
terms of baseball players--baseball players are about 60% overweight and only 2% obese.
So, if anything, baseball players are not only healthier than football players, no big
surprise, but they're also much healthier in terms of body mass index than general population.
So they're very--in general, you can pretty much say this is a very lean population and
this is their height and weight at their date of debut, which is basically when they are
the most fit, in the peak of their performance. So, some quick stats, well, this is the actual
tabulation of the body mass index across 130 years. That is a scatter plot of all 15,000
people and the red dots are the mean BMI. And you can see 25 being the threshold for
being overweight, it was--it hugged under 25 for many years, but in the recent 20 years,
it's definitely surpassed a mean BMI of 25 and hence, you can see the obesity--the overweight
prevalence is substantially increasing. So temporal trend, just quickly summing up the
last slide, 32% were overweight prior 1880, but by the most recent decade, we have over
55% who are now overweight. And homerun hitters, notably, were more than twice as likely to
be overweight than non-homerun hitters and with the doubling of odds of being overweight
for every 10 homeruns batted per season. And the homerun hitting story does not end there,
it actually gets much more interesting. >> Excuse me.
>> DING: Yes. >> How do you [INDISTINCT] homerun hitting?
>> DING: Homerun hitting is--well, we use a continuous variable, so we can either categorize.
Because homerun hitting in 1920, you know, if you hit like 20, it's considered very high,
but if you hit just 20 today, it's not very high. So it's not--we didn't use an absolute
cut-off because it's such a century difference, so we used for every 10 additional homeruns
hit per season, so you have every 10 additional, you have doubling of the odds ratio of being
overweight. Make sense? So this is the--basically the main result, the main result that body
mass index of players at the playing debut and the risk of mortality. For many people
who've been following the news, you know the few years ago there's been studies that says
overweight is associated to lower deaths. Not true whatsoever in many, many, many other
studies; in very--also not true among athletes. Because athletes, if anything, many suspected
that being overweight as an athlete does not increase your risk and only does it get--when
you get to the obese level does it--may your increase risk be present because overweight
among athletes generally reflects mean body mass and muscle mass. So this is actually
much stronger than we anticipated because, if anything, we only anticipated increased
risk in the overweight--in the obese range above 30, the last category. However, compared
to 18.5 to 22 BMI, you can see the increased risk start as low as 25 to 27 which is the
low end of overweight category, and this is very significant with higher BMI increasing
the risk of death starting at 25. So the key findings; BMI is strongly associated with
risk of total mortality. Well, in other words, it's associated risk of dying early. And these
results are adjusted for the age and the decade that they played because players are fatter
today than they were in the decades past. But this reflects not simply the decade or
phenomenon because we adjusted for age, decade and a wide variety of co-variants. So this
truly does reflect BMI, not a temporal trend. Risk increases seems as low as a modest 25,
which is very surprising because my BMI is actually a little over 25 myself--I'm ashamed
to say that--and with almost a doubling of risk to of 94% increase risk among obese players.
But also, someone once imposed this question, well, there are half the players who are really
obese in--really out of shape in baseball, because all you have to do is run one base
and you can be a fast sprinter and still be very out of shape because that's all you have
to run is, like, basically a baseline. But if we stratify those--these--all these results
and we look among those with high number of stolen bases, which basically are the leanest,
fittest sprinters in Major League Baseball, these super lean fit sprinters, actually,
you saw the same phenomenon. So it doesn't matter if you're just a designated hitter
who may be quite overweight and out of shape or if you're actually a sprinter with high
stolen bases, this increased risk still holds up. Now, the homerun hitting is when the results
get extra interesting but also disturbing. The increased risk of death with overweight
was actually stronger among homerun hitters. Well, earlier, I said, "Homerun hitters are
more overweight," and you know, of course some of you would easily think, "Of course,
if they're more overweight, they're more likely to die." But the story is not that simple.
They--not only did they have stronger risk of dying, but they also--there was effect
modification. In other words, the effect differed among--for BMI and mortality whether or not
you are a homerun or non-homerun hitter. Specifically, homerun hitters with one additional homerun
average per game had 15-fold higher risk of dying sooner compared to overweight runners,
overweight players, who hit fewer homeruns. This is quite surprising because you're comparing
overweight individuals to overweight individuals, but per additional homerun, they have a 15
fold increased risk of mortality. This is basically saying that this is the general
trend among all baseball players. But this slope is actually much steeper among homerun
hitters. So, in terms of what that means, we can only speculate, some people suggest
steroids. However, steroid use, you know, however hypothetically prevalent it is today,
it's only been a phenomenon the past 20 years. So--but in terms of mortality across 130 years,
this reflects something, I think, much deeper, that we are not quite sure yet. But the strength
as a magnitude is very surprising and I don't think it should be ignored. And again, no
such homerun hitting mortality risk seen among non-overweight, so if you're non-overweight
and you hit more homeruns, you have no increased risk of death. But--so it's a double edged
sword, it's a double whammy in to--for--to be overweight and homerun hitter. So, in terms
of the Major League Baseball, what they really care about is, like, are baseball--bigger
baseball players better baseball players? Because if they're bigger, for all they care,
you know, you can be as big as you want. If you can hit all the homeruns that your team
needs, basically, your salary would go up and no one would care about your health risk.
But in terms of actual implementation--in terms of actual salary and compensation and
other performance measures, does a bigger player mean you are a better player? And so,
this is just for Major League Baseball curiosity's sake. And when we graph it, this is total
hits and total runs batted in RBI's by body mass index. The total hits is blue, total
RBI's is orange. And you can see that the higher your BMI, it actually goes back down.
It's bell shaped instead of a positive slope that, you know, you would expect from more
overweight players that--who can supposedly hit more homeruns and other hits. So there's
no performance correlation between bigger size and total hits and runs batted in. In
terms of homeruns, this is what everyone is hoping the payoff would be; the bigger the
player, the better the slugger but that's not true at all. There's a simple bell-shaped
association and there's no general correlation or association whatsoever. So are larger homerun
players being overpaid? I don't know. In terms of batting average, this is as null as it
gets. Basically, there's absolutely no association, correlation between body mass index and playing--and
batting statistics. So again, you're not for paying any premiums by hiring and retaining
more overweight players on your team. But what is the bottom line? Body mass index and
player performance, no relationship whatsoever. So in terms of summary, baseball players have
become significantly bigger across 130 years of professional baseball history with homerun
hitters being substantially more overweight than players with fewer homeruns, and higher
the BMI associated higher risk of death in as modest as a BMI of 25, which is actually
very modest by overweight standards with homerun hitters again getting the brunt of the mortality
risk. Because for homerun hitters who are overweight, they are substantially--and by
substantially, I mean 15 folds greater risk per additional homerun a game, which is quite
significant. So in terms of future work in this area, well, in term--there's much to
be done because the mortality I presented were only for all cause mortality, not mortality
broken down by specific causes. We want to break it down by cardiovascular causes, cancer
causes, deaths from other cause, accidents and so forth. However, this requires, obviously,
substantial funding staff and National Death Index search, and there's also a wealth of
other data about player performance, injury rates as well as, you now, left handedness,
right handedness, and of course, there's many other sports we should verify this. And if
it holds among baseball players, then surely, among football players with 95% prevalence
of overweight and five-fold greater prevalence of obesity, there--the risks are probably
much higher and worse among professional football players. However, these are things that we
look forward to studying in the near future. Thank you very much, and that's my link for
the campaign for cancer prevention if you're interested. And I understand we have some
time for questions. Yes. >> [INDISTINCT] the raw data.
>> DING: The raw data is available. It's online actually. And so this is what's so interesting
about this project because the data was derived from online resources, they're open access.
However, the playing data is not in the same data as the mortality data and to get them
to merge--so you can access them player by player, and you have--you can access the raw
data except for the playing. But the mortality data, the raw data set is not--it's by request.
However, once requested, you have to merge the data set together which takes up quite
a bit of time compiling from the different data sets and merging all 15,000 players.
But that's actually how we accessed it. >> Hi. Eric, I was wondering if you've shared
the results from the study with the Major League Baseball and what their reaction has
been? Or you know, have individual baseball players--you know, if you've heard any comments
on this, like, in the media. >> DING: Well, I--when I presented this abstract
previously, it did get some media coverage. However, the commentary from--I don't believe--we
didn't get direct commentary from them. However, one of their team physicians for one of the
Major League teams commented and he says, "Well, this tells us what we know all along,
that you should keep your weight down." But I don't think he--but when reading the article,
I don't think he did quite grasp that were talking about, not only that obesity is increasing
among baseball players but also that the mortality risk is increasing about baseball players.
That the mortality risk actually come from the 130 years of baseball players and that,
you know, as low as BMI 25, you see increased risk which basically covers 70% of all baseball
players today have a BMI of over 25; which is quite disturbing. And basically, they're
all at elevated risk. So in terms of direct commentary, no, but in terms--we have gotten
few small commentary, but not official commentary. Any other questions? Yes.
>> Is there a similar study for women? >> DING: In terms of women, among women athletes,
there's not such a long-standing cohort in existence, because most female professional
leagues didn't emerge until the past two decades at the earliest and so, those players who
were playing then, almost no one has died yet. So in baseball, we are able to benefit
because baseball has such a long standing history and you have--and can accrue 7,000
deaths to do such a study. You can't even actually do it with National Football League
actually, because in terms of total number of deaths, NFL is actually a very young league.
And the same for basketball, it only dates back to 1930's, 1940's, so the total number
of deaths that's accrued to study the mortality risk is--very limited for other sports and
very, very limited for women, unfortunately. However, there's--the findings definitely
do hold up in women because they can stand--the nurses health study, for example, based at
Harvard, which has over 150,000 women in each cohort, so 300,000 women for the past 20,
30 years, and, you know, BMI definitely increases risk similarly in women as well as men. But
in terms of actual athletes, unfortunately, we don't have women. Yeah.
>> I was wondering, just early on in your presentation you showed slides of how in the
United States, you know, state by state, the obesity rate has just been going through the
roof, you know, each decade. And then I'm just--I'm asking about the relationship between,
basically, the general public's obesity rate rising and the rise in obesity among baseball
players and I was wondering, is the baseball player right--rate rising faster than America
as a whole? Or is it like--or is the baseball player rate just a reflection of Americans
getting fatter? >> DING: Well, that's a very good question.
In terms of the trend, we don't have quite good data on the prevalence of obesity back
in the 1800's. However, if we answer today, for example NHANES here showing here is a
nationally representative U.S. sample for US adults and so here, you can see that overweight
is similar, however, obesity is 25% among NHANES; however, only 2% among these baseball
players. So, baseball players are much leaner than the general population as expected, although
it's opposite for NFL, however, in terms of actual trend, it's hard to say because the
actual trend, we don't have the trend data for as many years for national prevalence.
But it's likely that the epidemic is very, very similar. However, though, in the past
20 years, I would say looking at graphs, the evolution of obesity, it's gone up substantially
faster in the general population then in baseball. Yes.
>> I was wondering about the causes of obesity and particularly the long term trend. So for
example, I can imagine between, you know, kind of the 1800's just maybe 2000 or whatever
that we're just--you know, there's no prosperity and we're healthier, but I can't just imagine
that between--you didn't say the last decade or the last two decades, that prosperity has
changed that much that people are, you know, the diets are changing or whatever that much,
because of just economic reasons. So what--to what do you speculate or attribute causes
of obesity currently? >> DING: Among the general population or among
baseball players? >> Well, either and--or lets say both, and
particularly, why it continues to go up recently. >> DING: Well, of course, obesity, there's--it's
many--it's a multifactorial disease. There are many causes, contributions, such as greater
sedentary activity, we're all in front of the computers much more than we were in the
80's; we don't exercise as much. There's a whole host of environmental reasons such as
bike paths, the urban--the suburbanization of cities and having to drive more; and then
there's dietary reasons, of course, like, potentially sugar-sweetened beverages like
Coca Cola, proliferation of Coca Cola in the past 20 years. There's many other risk factors
and basically, a lot of foods have gotten cheaper. I mean, it also, you know, the proliferation
of fast food. There's--in terms of actual obesity, the causes are endless. But in terms
of Major League Baseball--so partly, I grew--it definitely reflects an increasing trend among
the general population. However, again, perhaps there's more motivation now to use steroids
in Major League Baseball. Not that all baseball players do it, but there's much more propensity
to want to use it and the rewards for hitting more homeruns have increased ever since--because,
you know in the past, basically, 100 years before the recent two decades, baseball playing
was a pretty much amateur paying sport. It's a professional sport, but the pay of it was
mostly like a middle class, middle income class. Well, today is to the level of super
stardom and the compensations and basically, the desire to be bigger and hit more has also
increased. So, I think in terms of general population, diets, physical activity, our
environment has substantially changed, but in terms of baseball, I think also there's
many other potential and motivating factors that we can speculate more on. Any questions?
>> Or did you just [INDISTINCT] I heard you do a lot of research on diabetes and [INDISTINCT].
>> DING: So the question was, "Do I have any data on diabetes risk and incidence of diabetes?"
Well, it's a good question. However, again, we only have mortality and we only have total
mortality because we only know the day someone died. To get cause of death, we need to go
to death certificates of individual--of all 7,000 players and counting who have died.
So we do not have the resource to fund such a huge endeavour at the moment. In terms of,
like, diabetes incidence and linkage with medical records, first of all, most medical
records are not electronic--linked in the United States and getting access to current
players who have not yet died, the medical records, you would need informed consent and
for past players, you have to track down the families, so it's unlikely we can get disease
incidence data. The best we can do is get disease mortality data and that even takes
substantial funding and support. However, obesity is one of the strongest predictors
of diabetes. With increased BMI of 35 of 40, you have basically, 40, 50-fold increase risk
of diabetes. And so, with increasing BMI, it's a forgone conclusion there's also increased
risk of diabetes, cardiovascular disease and cancer and many other obesity-related diseases.
Yes. >> How trustworthy are the heights and weights
in people's stats? I can imagine a lot of players may lie to seem like a better athlete
in the way a lot of people would lie at the DMV about their weight.
>> DING: That's a very good question. In terms of how trustworthy the weight is, we actually
got height and weight from two different sets of data and they had perfect--near perfect
correlation of 9.98, .99, so they're independently verified. In terms of actual reporting, this
is their debut weight, and so, in terms of historical records, I'm sure there's not as
much fudging the numbers in the old history books. In the current history books, even
if there is a little fudging downward--thing is--you know, among baseball players, do you
want to list your weight your weight higher or lower? It depends. And it's unlikely to
bias the results, because those who fudged their numbers do not know their future mortality
outcomes. And basically, it's a--it's random misclassification, it's non-differential misclassification
compared to the disease outcome. So, only if they actually knew they were going to die
sooner--or higher risk of dying and they fudged their numbers would there induce any bias,
and that's very unlikely in this kind of data set, but a really good question. Well, thank
you all for coming and please contact me if you have further questions. It was a pleasure.