Timing of Dialysis Initiation: The Billion Dollar Question

One of the true arts learned during renal fellowship is the timing of initiation of dialysis in outpatients with advanced chronic kidney disease. This a crucial skill to develop as the beginning of dialysis has major implications for the patient’s lifestyle, finances and well being. In this regard, the publication of the IDEAL trial (reviewed by Matt and one of our top stories of 2010) was a landmark in helping to guide management in these situations, highlighting the safety of waiting for signs and symptoms of advanced chronic kidney disease rather than pursuing early start dialysis solely based on eGFR in closely followed individuals.

Does this differ from what’s been going on in practice for the last 10-15 years? A quick look at the USRDS shows that back in 1996 less than 20% of patients started dialysis with an eGFR of greater than 10 ml/min/1.73m2. In stark contrast by 2009 a full 20% were starting with an eGFR of over 15 ml/min/1.73m2 and greater than 50% were starting above the 10 ml/min/1.73m2 mark.

Why the heck did this happen when uremic signs and symptoms, at least in the IDEAL trial, in general seem to occur at GFRs below 10 ml/min/1.73m2?

A variety of explanations have been put forward (well reviewed by Rosansky and colleagues here and here). A commonly proffered argument is that perhaps uremic signs and symptoms are occurring earlier in our aging and increasingly ill population. However, in a study of 2402 nursing home residents who initiated dialysis, the authors found that seven signs and symptoms commonly associated with declining kidney function, such as volume overload and cognitive or functional decline, accounted for less than one third of all cases of early dialysis initiation including less than half of all cases of early outpatient dialysis initiation, suggesting that dialysis initiation may be primarily prompted by laboratory values in a large number of patients.

Regardless of the reasons, the trend is clear, early start was the rule rather than the exception building through the late 90s into the 2000s.

What has this meant for patients and US societal economic cost? A very nice paper from O’Hare and colleagues tells part of the story, using information from a large integrated health care system in Seattle, the group was able to estimate the rate of eGFR decline prior to the onset of dialysis in a subset of patients. This in turn allowed them to predict how many additional days on dialysis patients spent in 2007 versus 1997 by looking at the mean eGFR at time of dialysis initiation in patients listed in the USRDS and then back calculating (this relied on the assumption that the rates of decline were similar in these groups).

These estimates showed that dialysis was initiated on average of five months earlier in 2007 compared to 1997 and almost 8 months earlier in patients over 75! Doing some back of the envelope calculations the group guessed that early initiation in the US had cost an additional $1.5 billion dollars during 2007. This is slightly higher but similar to the over $1 billion dollar additional yearly cost estimate made using slightly different assumptions by the AJKD editorialists for the economic analysis of the IDEAL trial (this was the US estimate – the IDEAL trial was conducted in Australia and New Zealand).

Given that early initiation of dialysis in asymptomatic closely followed patients has no proven benefit (and may even be harmful if you look at recent retrospective data) part of our charge as emerging nephologists is helping to reverse this expensive and unhelpful trend.

(Photo: Stanford Chapel at Night)


  1. In some countries you cannot be listed (and commence your wait time) on the deceased donor list until you start dialysis.
    The temptation is obvious for the physician with a young patient desperate for a transplant.
    In the young perhaps an advantage to starting dialysis early.

  2. Thank you to Graham and to Dr. Klenzak for your replies. From my own, LONG experience from dealing with fsgs, I do understand the CV risks, and I do appreciate the difficulty that comes with deciding when to initate dialysis. This is where good communication between patient and doctor is of utmost importance.

  3. I have few points to make:

    The saving analysis that you posted in the original blog does not take into consideration the cost of recurrent hospitalizations and ED visits that may arise from patients (even closely followed up) may utilize.. for volume control for example.

    On the other hand, the use of the MDRD formula and change in CMS requirements for starting patients on dialysis are to be partly blamed for this increase in early starts, I believe.

    There is certainly a tremendous difference between academic studies and clinical practice. In reality, not every patient can take all these "preventative" medications. I attended conferences where presenters said that they will keep their patients on an ACEi or ARB even till near ESRD by using Kayexalate to prevent hyperkalemia. Try to convince a patient to take kayexalate everyday! Others recommend to stop ACEi or ARB near ESRD to allow for some improvement in eGFR to help affording the patients few extra months off dialysis.

    Yes, Not every patient is suitable for dialysis and that is something our Nephrology community need to take a strong stand on. I truly believe that initiating dialysis should have some more stricter criteria than the loose numeric criteria used now. Criteria that take into account the patient functional status, comorbidities, and mental status among others.

  4. To Moosemom,

    As a practicing nephrologist, I wanted to address the issue of initiation from a real-life, practical perspective. But first, to allay any conspiratorial fears Moosemom may have about doctors withholding life-saving care to patients for financial profit – THAT DOES NOT HAPPEN. The vast majority of my ESRD patients come to me for our first clinic visit with a GFR less than 25. The opportunity for prevention is long gone. My main concern at this point is their high risk of dying of cardiovascular disease, which is actually more likely than living long enough to start dialysis.

    Regarding those patients with treatable/reversible disease – this is what we live for as nephrologists. I know I am speaking for every nephrologist I know, there is nothing better than saving kidneys!

    In any event, given mild uremia and medical compliance, I will wait until my hand is forced to start dialysis, oftentimes with a GFR less than 5, especially in patients who put out good urine volumes. But the timing of initiation is so complex and patient specific, it cannot be reduced to a GFR threshold. I have seen patients who were started too soon, in my opinion, and I have taken them off to "see if they'll fly". Sometimes we can get another 6 months before re-initiating.

    But there are also patients who have been started with GFRs slightly above 15 who demonstrated great benefits from initiation – hospitalizations came down and overall health improved. These are rare cases, though, and often the primary problem for them was access (transportation, interactions with providers, self-neglect and abuse, literacy) coupled with volume.

    In any event, I do believe that dialysis should be viewed as a bridge to transplant for all candidates. For my patients who are not candidates, finding the best fit for their physiology and their lifestyle is the great challenge. I have 12 patients on the NXStage machine and the majority of them are leading fairly regular lives without the post dialysis fatigue, cramping, weakness, with better nutrition and fewer mood disturbances.

    The timing of initiation is not a simple calculation. I am only 6 years into private practice and still struggling with this issue on an almost daily basis. One of the downsides to late starts is oftentimes the patient hits a wall without warning, ends up in the ED at an outside hospital, or on a Sunday night with my partner, and is initiated, often emergently, by another physician. The hospitalization could have been avoided had I initiated him after our last clinic visit. I have learned the hard way that late starts who are planning on PD can be fraught with problems as well – PD doesn't work for everybody and we often don't know until it's tried – should I be putting fistulas in my PD patients, too?

    There is no great conspiracy out there – it's just one patient at a time, trying to do what's best for each individual, with limited powers and limited options.

    Jennifer Stoddard Klenzak, MD

  5. MooseMom,

    There are certainly several well accepted strategies for decreasing the rate of decline of GFR in a variety of types of chronic kidney disease (remembering that CKD has a large number of causes including diabetes, FSGS, IgA nephropathy, ADPKD and so on – with more or less evidence for different strategies in different types of disease).

    Is there an opportunity to reduce the incidence of advanced renal replacement requiring CKD by 70% or more using current medications and or lifestyle changes?

    If one could turn the tide and prevent a large fraction of new cases of obesity, diabetes and hypertension the answer is probably yes as these are the major drivers of a large fraction of ESRD.

    If we are talking about progression delay strategies in prevalent CKD cases there is certainly missed opportunity due to lack of insurance, poor care coordination and failure to systematically employ proven strategies for every appropriate patient every time.

    The strategies cited by Anonymous in the last post center around dual ACE/ARB use along with a collection of cardiovascular risk reduction and proteinuria reduction treatments.

    None of these are top secret or ignored.

    Dual ACE/ARB use has received a massive amount of attention and unfortunately through trials such as ONTARGET and the now withdrawn COOPERATE has not lived up the hope of enhanced progression delay. The study by Remuzzi and colleagues is interesting but has limited generalizability to the CKD population at large. As stated, it is small and nonrandomized and conducted in a highly selected population.

    So is there good evidence that a current protocol or medication combination could reduce the incidence of advanced renal replacement requiring CKD by 70% here in the US? No.

    Are progression delay strategies left out of renal fellowship? No.

    Are fellows withholding progression delay advice so that they and dialysis organizations can profit? No.

  6. Anonymous, I've been reading this same post from you all over the web. I'd like to take this opportunity to ask neph fellows if this poster is correct, that none of you are being taught his methods for eradicating 70 – 80% of ESRD with existing meds and demonstrated protocols. Are you fellows in league with the LDOs, withholding essential treatment to vast numbers of CKD patients so that they will have to eventually progress onto dialysis? Is this true? Or do you think there is something, and for some conspiratorial reason, very lacking in your education?

  7. 70-80% of current ESRD is now preventable within normal lifespan using existing drugs, and demonstrated protocols.

    How would 80% less incident ERSD change the world if such protocols were universally adopted? Within ONE year, enough surplus capacity would exist to support every-other-day dialysis at zero increase in taxpayer costs. Within FIVE years, enough surplus would exist to support daily dialysis, and within 10, the supply of donated organs would roughly match the number of people waiting.

    The long waits for transplant (currently getting WORSE, not better) -would be a thing of the past. Dialysis would primarily be used only as a short-term bridge to transplant (at perhaps 10% of current rates), since pre-emptive transplant would be the common rather than rare occurrence. The harsh decisions regarding who is "worthy" of receiving a donated organ would be dramatically relaxed.

    In other words, a 30 billion dollar industry would collapse. The analysts at the LDO's have run these projections, which is largely why demonstrated techniques for preventing CKD progression are so conspicuously absent from the protocols everyone so complacently follows.

    The MAIN reason dialysis is initiated so much earlier today despite no evidence of benefit is dialysis ownership and shared-ownership/profit sharing business models. Keep playing the game. Follow your company protocols rather than actual patient symptoms and progress, and rest assured you will have a very "successful" practice.

  8. Anonymous,

    Your point about not dialyzing, is an excellent one. The dialysis decision is one of shared decision making and based on my own experience and emerging data I believe there is a subset of patients who derive little or no benefit from dialysis. Even for those that do, the decision to adopt dialysis into one's life is a major one best made by the patient armed with sound advice.

  9. As you all graduate and start practicing, there will be subtle and not so subtle pressure to put patients on dialysis. I have been told that the success of a nephrologist is defined by the number of dialysis patients that he/she has.

    With the advances in the care of CKD the GFR decline can be stablilized or slowed down. patients can maintain their CKD status for a very long time and not become ESRD.

    I think another art that needs to be learned or acquired during fellowship that may also be as important is when NOT to start patients on dialysis.

Leave a Reply