How good are we doctors in conveying the concept of risk to our patients and our community?

I recently met someone who donated a kidney to a person who was not a relative. My impression was that the recipient was previously a stranger, but I did not press that issue. I was overwhelmed by the staggering generosity of that gift. I am embarrassed to say that I do not know how I would respond if I were asked to donate.

Shortly afterwards in the impossibly large volume of medical journals that crosses the desk of even a retired physician, and proving the hypothesis that a person only sees what they are prepared to find, I noticed a research paper estimating the risk of kidney failure in a donor following the removal of one of the normal pair. The risk does not appear to be zero. I want to use that paper as a base to continue writing about conveying the results of medical studies and of risk to the public.

Why did Mother Nature Give Us Two Kidneys?
In medical school, and in common public wisdom, it is universally recited that we only need one of our two kidneys to live. That is true. As it happens though, there is little or no information about whether we would live as long or as well! It is known that as people age, there is an expected gradual decline in kidney function. It happens sooner and faster in people with hypertension, diabetes, and so on. It is reasonable to speculate that if starting out from a half-normal position, that a single remaining kidney might run out of steam sooner without it’s helpmate. The best way to definitively settle the matter would be to randomly divide in half a group of people who had committed to donate, only allow one group to do so, and then follow the lives of both groups to see if the donors develop kidney failure more frequently than the donor wanna-bees. I do not see that experiment happening, but the information would be relevant to a person deciding whether or not to donate.

Is there any way to make an educated guess?
As with everything else in medicine, we do the best we can. Just last month Dr. Abimereki Muzaale and his colleagues published the results of an ingenious study (1). They looked at 96,217 living-kidney donors for up to 15 years after donation. Studying such a large number was possible only because a national transplantation registry is built in to our healthcare system. (One huge benefit of a coordinated national health program!) Not unexpectedly, especially given that some of the donors were older, 99 of the donors developed kidney failure during a median followup of only 7.6 years. Is that bad? Compared to what? There lies the rub. A comparison to people in the general population is not valid because the donors would have to be in better health to be eligible in the first place. What is needed is a comparison group of demographically similar individuals in the same general health as the donors.

What the researchers did was match donors to individuals in a large, long-term observational study of regular people– the well known Third National Health and Nutrition Examination Survey. Donors were matched one-for-one with survey members for things like age, sex, race, education, blood pressure, obesity, smoking, and the like. In this way, an attempt was made to compare people in similar states of health. I did not expect the results.

Do the numbers speak for themselves? Or not?
Of the 96,217 individuals, 99 of the donors (0.103%) and 36 of the “normal” controls (0.037%) developed kidney failure in less than 15 years. Correcting for the fact that not every patient or control was followed for the maximum of 15 years, the estimated risk of kidney failure at 15 years for donors was 30.8 per 10,000 individuals vs. 3.9 per 10,000 for the comparison population – a 7.9-fold increase. The lifetime risk of kidney failure was projected to be 90 per 10,000 donors but only 14 per 10,000 in the controls. This more than 6-fold difference was highly statistically significant. Granted, the rates for these two relative healthy groups were much lower than the baseline lifetime incidence of kidney failure in the unhealthier general population which was estimated to be 326 per 10,000.

There were some differences among the donor groups. Men had double the incidence of kidney failure at 15 years then women donors. Black people had three times the incidence of white people– Hispanics were intermediate. Older donors did less well than younger by several fold. Donors who were biologically related to the recipient had twice the incidence of those who were unrelated.

What are we to make of this?
At first blush, donating looks like risky business. Should we insist that all kidney donors be dead first? Should black and Hispanic people be discouraged from donating? Should there be an age limit? Should potential live donors be promised that they will receive a kidney of their own should their remaining one fail? (No such protection exists today. We harvest from everyone to supply the privileged.) What kind of informed consent should donors be expected to sign?

“If you torture the data long enough, it will tell you what you want to hear.”
Sixteen years ago, five of my physician colleagues and I went through a year-long fellowship program in Washington DC to introduce us to how health policy was made. On one of our first briefings, we visited a public relations firm where we were shown how the results of a medical research paper could be turned on its head to show exactly the opposite of what the authors intended. We were in shock and awe at what these masters of deception could do– and in fact be proud of!

How our current study of post-donor kidney failure might be communicated to the public might vary depending on the goals of a given organization. Transplant programs and patient-advocacy groups desperately beating the bushes for badly needed organs might tend to stress the relative infrequency of kidney failure or point out issues intrinsic to the study design that moderated its conclusions. A research oversight committee would want the unvarnished risk to appear in its informed-consent document, or at the very least, point out that the risk is unknowable. Hypothetically, a manufacturer of dialysis equipment or a dialysis center might want to ramp up the risk of kidney transplantation. Naturally, for such a humanitarian activity no one would knowingly manipulate the data, but you get the idea.

May the odds be always in your favor.
One of the best trends in medicine that I have witnessed in my professional lifetime is what might be called “evidence-based medicine.” We physicians were inheritors of a large body of medical wisdom and treatment options that were just plain wrong– ineffective if not harmful. To serve the obvious need to better evaluate the effectiveness and risk of what we were doing, newer (if not always more understandable) statistical mathematical methods of confirming meaningful differences in mountains of information evolved. This branch of medical science can get very complicated, sometimes giving us results that we do not know what to do with. For now, allow me to explore a few concepts that are essential to medical decision-making for both doctors and patients.

Statistical significance does not equal clinical significance.
Even the public gets to worship statistical significance via the ubiquitous drug advertisements we are exposed to. A “p value” of less than .05 is taken as the gold standard of effectiveness in clinical studies. To put it over-simply, p<.05 indicates that a difference in results has only only a 5% probability of being wrong– one in 20. (Would you play Russian roulette with a 20 chambered gun?  I would not!)

Statistical significance by itself is value-neutral and must be judged along with clinical value. For example, if a new cancer drug is found to extend life for 6 months in every patient who takes it compared to similar untreated patients, that difference in a reasonable sized sample would almost certainly be very highly statistically significant. Should every patient with that cancer then receive the new drug? Should insurance companies pay for it out of our premium dollars? What if four of those last months were usually spent hunched retching over a toilet bowl? What if the drug added $120,000 a year to the cost? (Some new drugs cost more.) What if the prolongation of life was only one week– perhaps still statistically significant– but not clinically significant to a reasonable person who must also consider family finances, quality of life, risk of an earlier death due to the treatment, and the host of other factors underlying the difficult decisions that doctors, patients, and their insurance companies make together every day. Who should decide how to balance these risks to body, soul, and pocketbook? How can information guide us?

Double the risk of a rare occurrence is still a rare occurrence.
I used to take care of people who had diseases with an annual occurrence rate of one in a million people per year. If I was aware that drinking coffee tripled that rate to 3 in a million per year, should I insist that my patient never drink coffee? If I thought the evidence was reliable (how can you ever really know for such rare events) I would certainly be obligated to tell them, but if they enjoyed coffee a lot, I would support their tossing the dice and join them in shouting L’Chaim– to life! My patient would have made a perfectly valid and ethically acceptable value decision. Patients and their doctors do this every day. Many organ donors do too.

No study is perfect nor can the results be applied to all people or groups.
As is virtually always the case, if we dig down into the paper at hand, other factors that might have increased the apparent risk of kidney failure in the donors can be found. (This is one of the ways that scientific knowledge advances– divide and understand.) For example, two-thirds of the donated kidneys went to blood relatives of the donor. It is possible that the same genetic, environmental, or socio-economic factors that caused kidney failure in the recipient were also present in the donor. In fact, the risk of subsequent kidney failure for a related donor was twice as high as that for a non-related donor.

Socio-economic or medical factors predisposing to kidney failure such as hypertension may explain in part the higher risk for both black and Hispanic donors and their control matches. The authors made an attempt to sort out how much of the observed incidence of kidney failure was due to the act of donating, or whether it reflected the spontaneous background incidence of kidney failure. They concluded that for white people, all of the observed kidney failure could be attributed to donation, but for black and Hispanic people this proportion fell to roughly two-thirds. I suspect that this different proportion is not the result of any intrinsic biological difference, but is just another reflection of the medical disparities we accept in our nation.

To a meaningful extent then, where you start determines your ultimate risk of donation. If you are a young white woman with no family history of kidney disease, your absolute risk is much lower than an older black man who is donating to his brother whose kidney failure was due to hypertension.

Then again, these broad generalizations based on subsets of data may not be robust enough on which to base recommendations for real people. The problem with slicing and dicing the results of a study too much is that mining for secondary conclusions that a study was not designed to examine will always reveal some seemingly meaningful results due to chance alone. No study is perfect. Every one has its strengths and limitations and may not have the power to answer every question that is asked of it.

How would I present this information to potential donors?
If I wanted to scare people away from donating, or cover my backside from medical-legal exposure, I might unconsciously stress the apparent several-fold risk of developing premature kidney failure. After all, kidney failure was nearly an order of magnitude higher in donors than for similar healthy people! The study did not even consider the effect on length of life.

If I did not want to discourage donation, and to enable a selfless act, I would point out that the absolute risk of kidney failure was very low, much lower in fact than in the general population, and could probably be mitigated by careful attention to good health habits in the future.

What should I say to a black or Hispanic person, groups that made up 13% each of the live-donor population? For that matter, should I give a different spiel to a man than a woman? The increased risk seems real, but given the limitations in the study design and the imperfect comparison group, might we ethically say that the incidence of kidney failure for donors and healthy non-donors was roughly in in the same ballpark and considerably less than for the general population? I don’t know.

How much is it worth to the donor?
We can make these numbers sound as scary, or as reasonable as we wish. For someone donating a kidney to a family member or loved one, these risks, honestly revealed, would probably not even make that person blink. Even for a person considering donating to a non-relative or stranger and for whom the risk is not negligible, the offsetting benefits to the donor himself or herself may be substantial. Many are called upon to make greater sacrifices for their fellows. This one is not as dangerous as some.

Conclusions.
Even the best possible study may not be definitive. We will never have perfect knowledge. You take the best information you can gather and weigh both advantage and disadvantage, risk and benefits. Benefits may be subjective and highly individual and risks not always easy to discern. The informed potential donor (or patient) always has a say.

I doubt we will have a better design study than this in my lifetime. The authors’ final conclusion seems reasonable. “Compared with a matched cohort of healthy non-donors, kidney donors had an increased risk of end-stage renal failure: however the magnitude of the absolute risk increase was small. These findings may help inform discussions with persons considering live kidney donation.”

Peter Hasselbacher, MD
President, KHPI
Emeritus Professor of Medicine
University of Louisville
March 13, 2014

(1) Risk of End-StageRenal Disease Following Live Kidney Donation
Abimereki D. Muzaale,MD, MPH; Allan B. Massie, PhD; Mei-ChengWang, PhD; Robert A. Montgomery, MD, DPhil; Maureen A. McBride, PhD; Jennifer L.Wainright, PhD; Dorry L. Segev, MD, PhD:  JAMA. 2014;311(6):579-586.