Essential Science IndicatorsSMfrom
Reuters, the paper, "Prevalence of chronic kidney
disease in the United States" (Coresh J, et al.,
JAMA-J. Am. Med. Assn. 298: 2038-47, 7 November
2007), was among the 20 most-cited Hot Papers over all
fields. It is now a Highly Cited Paper in Clinical Medicine
with 355 citations to its credit up to December 31,
Lead author Dr. Josef Coresh's record in the database
includes 164 papers, the majority of which are classified
under Clinical Medicine, cited a total of 9,736 times
between January 1, 1999 and December 31, 2009. Dr. Coresh
is a Professor at the Johns Hopkins Bloomberg School of
Public Health, where is the Director of the Cardiovascular
Epidemiology Training Program and the George W. Comstock
Center for Public Health Research and Prevention.
In the interview below, he talks with ScienceWatch.com
correspondent Gary Taubes about this paper and his work in chronic
What prompted your initial research interest in
chronic kidney disease and how did it involve into epidemiological
I was working in heart diseases in the 1980s, when Paul Whelton made me
aware that the epidemiology of kidney disease was really just beginning. I
started getting involved in that in the early 1990s, and it turns out there
were really fundamental things that had to be done.
By 2002, I was the vice-chairman of a working group that was developing
standardized criteria for chronic kidney disease. We defined the broader
range of kidney disease in the population and the 2007 JAMA paper
is an update of the trends for that.
Why do you think the epidemiology of kidney
disease was just beginning in the 1980s, or roughly 20-30 years behind
heart disease epidemiologies?
Until then, nephrologists were focused on doing dialysis, which is an
amazing technology that keeps people from dying and was a technological
advancement in the 1950s and 1960s. That was the core of the discipline,
but that's really about the sickest people. It's like looking at heart
disease and focusing only on heart failure, while saying, "Well, heart
attacks, they happen early and they're kind of hard to track."
Still, the number of people on dialysis has grown exponentially in the last
40 years. Now there's half a million people on dialysis in the US, but
that's only the tip of the iceberg for chronic kidney disease. Those are
the most severe cases.
Another problem is that because kidney disease is less common than heart
disease, if we want to do a prospective cohort study, we need the cohort to
be larger. But the less common the disease, the less funding there is. So
it's hard to do the larger cohort, because you can't get the money.
It turns out to be no coincidence that people like me came into the field
from studying chronic heart disease, because the cohorts we had built for
heart disease, the bigger ones, could also tell us about kidney disease. So
this national effort in learning about heart disease can reap additional
dividends in learning about kidney disease, particularly since much of
kidney disease has a vascular basis.
What cohorts did you use for your highly cited
2007 JAMA paper?
That paper is about prevalence of chronic kidney disease, and the
prevalence is best estimated from the National Health and Nutrition Surveys
(NHANES). We used NHANES 3 as the baseline. That ran from 1988 to 1994. And
then we used NHANES 1999-2004 as the follow-up data a decade later.
How did the prevalence change between the two
What we found was that the prevalence of chronic kidney disease, which was
already high in our NHANES 3 work, was actually increasing over the
subsequent decade. It also turns out that there are methodological nuances
that require careful calibration of creatine. Depending exactly how you do
that calibration, the rise we saw could be somewhere between a large rise
and a small rise.
"Chronic kidney disease is complex enough that it's not
going to be a one-pill intervention."
There seems to be a rise either way, both in the prevalence of low kidney
function, estimated from high creatine, and from high albumin in the urine,
which is an indicator of early kidney damage. Both of those were
How do you explain that increase?
The increase in the prevalence of albuminuria can be explained by the
increasing prevalence of diabetes. That was the biggest factor. Together we
looked at diabetes, obesity, hypertension, and age and demographic
distribution of the population, minorities as well. African-Americans, for
instance, have more albuminuria.
So the distribution of albuminuria can be explained, although that doesn't
mean it's not important. If we have more diabetes, then we're going to have
more kidney disease. But that doesn't mean it's not a problem. The
increasing prevalence of low kidney function—signified by high levels
of creatine—we can explain some of that, but not all of it.
Would you consider the creatine calibration issue
to be the most challenging aspect of the research?
Yes, that was the hardest part. I've been involved in that since the late
1990s. When we got our very first estimates, the assay values seemed too
high to be consistent with what we thought was reasonable. So we got
permission to get specimens from the freezer and we analyzed them and
compared them to what we'd get with current laboratory methods.
We also worked with NIST, the National Institute of Standards and
Technology, to get them to assign a value to a reference sample, and then
we measured that reference sample using what we thought was a very good
method—an enzymatic creatine method—and we then compared that
enzymatic method on several hundred samples frozen from both the NHANES
surveys. We got frozen specimens and again recalibrated the NHANES survey
It turns out that if you look carefully, any estimates over time are hard
to get consistent. The advantage we had is we tested samples from both
surveys at once, so we were calibrating both of them at the same time. The
results should be as comparable as anybody can get them, but even then
small differences could remain—and these are the nuances in this kind
of research—that's why we did a conservative trend analysis as well.
On a graph, the mean creatine level in the population goes up as well as
the tail of those people with the highest creatine levels and low kidney
function. What if the change in the mean wasn't the whole population
shifting, but a statistical artifact? What if we shifted that down, what's
left? It turns out what's left is about half of what we had seen before.
So the unadjusted trend would have been a prevalence of low
GFR—glomerular filtration rate—with an odds ratio of 1.47. The
conservative trend ratio has that down at 1.17. I worked on that for two
years wanting to make sure we had those numbers as right as possible.
What do you see as the major trends in kidney
disease research now that it's well-established and
There's been a big trend in the literature to define the full range of
outcomes of kidney disease. In the last decade that has gotten increasing
recognition. Early reports on chronic kidney disease emphasized some
outcomes, but there's been recognition that the ranges of outcome are much
wider than we thought.
The most obvious outcome is kidney failure and going on dialysis, but
that's still relatively infrequent, which is a good thing. But chronic
kidney disease patients, even those not on dialysis, die more frequently
than people without kidney disease. Part of that is that kidney disease is
correlated with heart disease, diabetes, hypertension, and stroke. Some of
that is a marker; some might be a cause. Teasing that out is quite
If you prevent the complications of kidney
disease, do you prevent or delay the deaths?
One of the things we've looked at is kidney disease anemia—low
hematocrit (a low red cell count) caused by kidney disease. We hoped that
correcting anemia would undo some of the complications of kidney disease,
but the recent clinical trials taking that as a strategy have been
If you raise hematocrit all the way to normal, you get an increase in
strokes. So hematocrit is a complicated variable to correct, because of the
side effects. It could be helpful if we protect from some of the other
metabolic complications of kidney disease, including abnormalities of
calcium and phosphate. And kidney disease clearly aggregates with
hypertension, so treatment of hypertension is an incredibly important
mainstay of preventing further complications. Treatment of diabetes will
obviously be helpful, because kidney disease is a microvascular
complication of diabetes.
We still need a lot of clinical trials to test all these strategies. We're
still very much where hypertension and hypercholesterolemia were in the
1970s, in that the biggest trials are still yet to come for a lot of these
strategies. And we still have to fully define the problem before we can
define all the solutions.
What other aspects of the problem remain to be
There are now articles by many of my colleagues looking at the wider range
of complications, from heart disease and kidney failure to acute kidney
injury and the complications of drug toxicity and what's called contrast
toxicity. When people get a CT scan or a coronary angiogram, they usually
get contrast to make the imaging better, but many of those contrast agents
are toxic to the kidney. So that's a known complication, but it hasn't been
completely quantified yet as to how many people have that and how severe a
complication it is.
Infections are also a complication in patients with kidney disease. The
immune system is somewhat compromised. There are some interesting articles
about the higher rates of pneumonia and the higher rates of mortality for
pneumonia in patients with kidney disease.
Another unfortunate problem we're dealing with is that patients with kidney
disease seem to be at elevated risk of morbidity and mortality but they
very frequently receive less treatment than the average patient. I think
that's partially explained by these patients being of a lower socioeconomic
status, and it's partly because of the physicians' concerns for
"Now there's half a million people on dialysis in the
US, but that's only the tip of the iceberg for chronic
It's also partly because some of the original clinical trials tended to
exclude people with kidney disease because the investigators were worried
about complications. So we don't have as much evidence for these
highest-risk patients and they're getting the least treatment. That in turn
elevates their risk even more.
Is this a common problem with chronic kidney
disease—that patients don't get the treatment they
There's been a wave of papers recently discussing that, documenting levels
of treatment and of awareness of patients with kidney disease. In our 2007
JAMA paper, we did a little of this.
We pointed out that the awareness of people who had weak or failing kidneys
in 1999 to 2004 was low. Those with stage 3 kidney failure—weak or
failing kidneys—only 11% of the men and 5% of the women were aware of
it. If they had albuminuria on top of that, the awareness rose to 23%. And
if they were at stage 4, which is severe kidney disease, the awareness was
still only 4%. That's the stage right before dialysis, where three-fourths
or more of kidney function is lost. Still, less than half of them reported
knowing having weak and failing kidneys.
Other papers showing that patients with kidney disease are often
under-treated are very important since they indicate that unfortunately
often we are doing just the wrong thing—treating patients at high
risk less rather than more.
What do you think explains the increase in chronic
kidney disease that you reported in 2007?
In terms of protein in the urine, albuminuria, we saw an 18% increase
between the two surveys. When we adjusted for the older age of the
population, for sex, race, diagnosed hypertension and diabetes, and higher
BMI over time, that increase becomes only 3%, and it's no longer
statistically significant. So almost all is explained by these other
factors with the obesity and diabetes epidemic being the dominant factor.
What do you think the causal connection is between
diabetes and higher BMI and kidney disease?
Some of the causal connections are very clear. Obesity clearly causes
diabetes, and since we've had an obesity epidemic, that explains the
subsequent diabetes epidemic. Diabetes and hyperglycemia clearly cause
what's called microvascular disease, and kidney disease is one of those
So diabetes is the leading cause of people being on dialysis, and diabetics
develop albuminuria. That's been known for a long time. In the clinical
trials, when we improve diabetes control, it clearly reduces the level of
In some sense, part of the impetus for that 2007 paper was this idea, given
the fact that we're seeing a diabetes epidemic, are we seeing an increase
in the prevalence of kidney disease? And we are.
The picture is quite complicated, though, because kidney failure itself
increased exponentially for a number of decades. It's still increasing, but
now a lot of the increase is driven by the aging population. If we adjust
for the demographic trend of the last few years, the adjusted rates of
kidney failure and dialysis have flattened out.
This could be due to improved diabetes and hypertension treatment. But it's
only been a few years. It's not clear yet whether the increase in the early
stages of the disease that we're seeing will lead to a further increase in
dialysis rates as well.
After albuminuria appears, it typically takes two decades before you end up
on dialysis. So it will take another decade before we know the full
consequences of the diabetes epidemic in the 1990s.
If we lived in an ideal world and you had
unlimited resources to do your research, what studies would you do
that you can't do now?
I would do randomized trials of comprehensive interventions to improve the
care of chronic kidney disease patients. Chronic kidney disease is complex
enough that it's not going to be a one-pill intervention. We were lucky
with heart disease. If statins knock cholesterol low enough you get an
amazing effect. We have ACE inhibitors that are incredibly good for kidney
disease, but that's just one step. We're going to need more. There are a
lot of treatments that are actually available but they're not necessarily
One of the sad facts of public health and epidemiology is that if we only
did the things we know how to do—if we got all hypertension under
control, for example—we could make major strides. If we got diabetes
under reasonable control, if we had glycemia controlled, and potentially
some physical activity, that would do much better.
I'm a big believer in direct lifestyle interventions, but nobody can patent
them. If you discover that doing all these things is good for you, it's
hard to capitalize on that. If I had unlimited resources, I would test
pretty intensive interventions in chronic kidney disease to see whether we
can reduce the progression at all stages.
It would have to be somewhat tailored to the stage, so we would start at
the earliest stages of diabetes and hypertension. We would treat those, and
use ACE inhibitors to lower proteinuria, and potentially test a number of
other things. It would have to be done one at a time, or together.
At later stages, when there are more complications, we'd have to treat the
complications to see if we can decrease the progression to dialysis; if we
can decrease heart attacks and mortality. We have some trials going for
those kinds of interventions, but some are quite small and they're not very
Josef Coresh, M.D., Ph.D.
Johns Hopkins Bloomberg School of Public Health
Baltimore, MD, USA
Josef Coresh's current most-cited paper in Essential Science
Indicators, with 854 cites:
Coresh J, et al., "Prevalence of chronic kidney disease and
decreased kidney function in the adult US population: Third National Health
and Nutrition Examination Survey," Amer. J. Kidney Dis. 41(1):
1-12, January 2003. Source:
Essential Science Indicators from
Note: this interview pertains the to the paper, Coresh J, et al.,
"Prevalence of chronic kidney disease in the United States," JAMA-J.
Am. Med. Assn. 298(17): 2038-47, 7 November 2007, with 355 cites.