Tag: statistics

Posted on

The best story that wasn’t nominated.

Its oscar season in nephrology. The RFN is winnowing down the top stories in nephrology of 2013 but maybe the most important isn’t even nominated. It came out too late. On November 27, Kidney International released this study in advance of print:

The bullet point I learned in fellowship was that the rate of renal failure in kidney donors was no higher than the general public, of course the kidney donors are screened to be much healthier than the general public, so the fact that the rate of renal failure is not significantly lower than the general public is an important signal.

Have I mentioned how
much I love MedCalc?

Donors don’t have obesity, diabetes, hypertension. Of course most of them also have family history of kidney disease, so when they do develop kidney failure attributing the cause cause can be tricky, certainly some of these people would progress to renal failure even if they didn’t donate.

Into this statistical Gordian knot wades the Norwegians.

  • 15 years of follow up for donors that were 46 years old at the time of donation. That is adequate follow-up in my mind.
  • Their center had no preoperative mortality. They did 2,269 living donations with out losing a donor. Great work by the surgical team.
  • They found an increased risk of death by any cause, HR 2.49 (CI 2.13-2.91). 
  • Adjusted HR for all-cause mortality fell to 1.48 (CI 1.17-1.88). 
  • I used their raw mortality data to calculate the Number Needed to Harm. It was a frightening 23 (4.3% absolute risk increase). I do not think this is a valid use of NNH, see below.
  • The CV Mortality HR was 1.4 (CI 1.03-1.91).
  • ESRD was increased from 0.01% per year to 0.03% a relative risk of 3 but a number needed to harm of 5000. (i.e. you would have to do 5,000 living transplants a year to see one additional cases of dialysis in a donor)
The lines only begin to separate after 10 years. This indicates that we should probably ignore studies with less than a decade of follow-up
The data is eye opening, but I would really be interested in seeing what the difference between first degree relatives and people not related. I think much of the increased danger comes from being related to the patient who needed the transplant in the first place.
I also recommend ignoring all the unadjusted data (including the number needed to harm) because the control group was nearly a decade younger than the donors. A ten year difference in age when the total follow-up is only 16 years makes the unadjusted data deceptive, IMHO.
I would love to see a study where the control group was made up of other people who were evaluated and cleared to donate a kidney but ended up deferring, due to the recipient getting another organ or dying or going off the list for some other reason. That would be the ideal control population.
The reality is that people want to donate a kidney to their loved ones. Telling them there maybe a small increase risk of death when you are in your late 50’s or early sixty’s. Will probably not eliminate many close relatives, but altruistic donors and family members should be aware of the risks as understood today. It is probably time to bury the old line about “No increased risk of renal failure than the general public.”

Another study on this from 2010 can be read here.