This conservative vs standard approach to initiating dialysis for acute kidney injury is such an interesting trial, that I will introduce it today and discuss again in podcast form this Wednesday
The study designers made the very common but serious mistake of completely ignoring close calls (e.g. only 13 days off dialysis) when constructing the outcome variable. This simultaneously inflates the sample size while still reducing statistical power, a lose-lose situation detailed in a paper we published last week: https://onlinelibrary.wiley.com/doi/10.1002/sim.70402. A thorough discussion of endpoint characteristics and information loss may be found at https://hbiostat.org/endpoint and https://hbiostat.org/bbr/info.
Statistical analysis should always be based on the most fundamental, finest grained, outcome. Here this would be a daily ordinal outcome indicating not on dialysis, on dialysis, dead, and possibly other levels. A multistate ordinal Markov model would be ideal for this - see https://hbiostat.org/rmsc/markov. This model can encode the treatment effect as a single odds ratio, but also provide much richer clinical readouts such as the reduction in expected days off dialysis and alive and the probability of going at least x days in a row alive and off dialysis, for any x of your choosing.
The fact that one is interested in a certain clinical readout should not be misconstrued that raw data needed to be wastefully transformed to that metric pre-analysis. The full information-preserving longitudinal analysis will have power that far exceeds the power of the binary endpoint the authors used.
As with many medical interventions, there is a financial conflict of interest. The decision to dialyze is made by the nephrologist, who is compensated for performing dialysis.
Should have done it in South Africa, public sector. There's no financial incentive for nephrologists to dialyse. In fact, they seem to do everything they can to avoid it. That's the current 'standard of care'
The commentary is quite good and cites some points JM was unable to cover. "[...] the criteria are perhaps too strict and may not resemble clinical practice; LIBERATE-D may arguably be considered a trial of sensible discontinuation vs unnecessary continuation rather than conservative vs conventional dialysis strategies."
I wish I knew the answer because my best friend who I met when I was six years old is now with AKI after mitral and tricuspid valve repair last week. So far Cr 5 and they’re watching. Preop non-diabetic non-hypertensive, normal renal function. EF is normal and there is no significant regurgitation. I might ask is ICU.if they head in this direction- about this study
I wonder if in an acute scenario, such as the one your friend is facing, it would matter much whether a conservative approach or a more standard approach is chosen. As in, does the outcome of this trial improve clinical outcomes or is it more about unnecessary interventions. I don't know.
Ironically, he status post mitral valve repair because of the data saying a more aggressive approach is better. I agree with that when the echo notes compromise, despite lack of symptoms, which could be subtle. Sadly, he gets the one in 100 complication. I might opt for conservative but the creatinine is improving so it may mute thankfully
I’m confused by this study. When I was an IM resident (late in the last century), on in-pt renal rotation we offered RRT for pts with persistent hyperkalemia, acidemia, and/or volume overload refractory to med Rx, or uremic symptoms. The “conservative strategy” was THE ONLY strategy at the time. When I’m on CCU now, I consult Nephro for IHD for those same reasons. And our nephrologists only offer it for those indications. What exactly is the “standard of care” in those 4 hospitals? Are they dialysing people just for a serum Cr number? If so, that seems insane.
I agree a “non-inferiority” design would have been more appropriate. Less invasive, less cost….”not worse” would have been plenty good enough to adopt this over the “conventional” standard.
Thank you for writing about this study. More medical and surgical interventions need to follow this approach: “Some smart experts have proposed a conservative approach to dialysis, where it is deferred unless it has to be done.”
The financial savings of not doing dialysis: another procedure on a sick patient to put some type of dialysis catheter in, all the equipment and supplies, staffing needed, and the trauma to the already compromised patient—these are significant considerations.
I think this study gives comfort to those who dialyze only when it is absolutely necessary to prevent life threatening complications, like the AEIOU I learned in training.
The study designers made the very common but serious mistake of completely ignoring close calls (e.g. only 13 days off dialysis) when constructing the outcome variable. This simultaneously inflates the sample size while still reducing statistical power, a lose-lose situation detailed in a paper we published last week: https://onlinelibrary.wiley.com/doi/10.1002/sim.70402. A thorough discussion of endpoint characteristics and information loss may be found at https://hbiostat.org/endpoint and https://hbiostat.org/bbr/info.
Statistical analysis should always be based on the most fundamental, finest grained, outcome. Here this would be a daily ordinal outcome indicating not on dialysis, on dialysis, dead, and possibly other levels. A multistate ordinal Markov model would be ideal for this - see https://hbiostat.org/rmsc/markov. This model can encode the treatment effect as a single odds ratio, but also provide much richer clinical readouts such as the reduction in expected days off dialysis and alive and the probability of going at least x days in a row alive and off dialysis, for any x of your choosing.
The fact that one is interested in a certain clinical readout should not be misconstrued that raw data needed to be wastefully transformed to that metric pre-analysis. The full information-preserving longitudinal analysis will have power that far exceeds the power of the binary endpoint the authors used.
As with many medical interventions, there is a financial conflict of interest. The decision to dialyze is made by the nephrologist, who is compensated for performing dialysis.
Should have done it in South Africa, public sector. There's no financial incentive for nephrologists to dialyse. In fact, they seem to do everything they can to avoid it. That's the current 'standard of care'
The commentary is quite good and cites some points JM was unable to cover. "[...] the criteria are perhaps too strict and may not resemble clinical practice; LIBERATE-D may arguably be considered a trial of sensible discontinuation vs unnecessary continuation rather than conservative vs conventional dialysis strategies."
This is the second trial I’ve read in two weeks that designed a study with what I see as overly optimistic priors.
RETREAT-FRAIL was designed for a 25% all cause morality reduction just by stopping HTN medications. It also “failed” to show benefit.
Someone stop me from being cynical, what role are reimbursement incentives playing here?
I wish I knew the answer because my best friend who I met when I was six years old is now with AKI after mitral and tricuspid valve repair last week. So far Cr 5 and they’re watching. Preop non-diabetic non-hypertensive, normal renal function. EF is normal and there is no significant regurgitation. I might ask is ICU.if they head in this direction- about this study
I wonder if in an acute scenario, such as the one your friend is facing, it would matter much whether a conservative approach or a more standard approach is chosen. As in, does the outcome of this trial improve clinical outcomes or is it more about unnecessary interventions. I don't know.
Ironically, he status post mitral valve repair because of the data saying a more aggressive approach is better. I agree with that when the echo notes compromise, despite lack of symptoms, which could be subtle. Sadly, he gets the one in 100 complication. I might opt for conservative but the creatinine is improving so it may mute thankfully
I’m confused by this study. When I was an IM resident (late in the last century), on in-pt renal rotation we offered RRT for pts with persistent hyperkalemia, acidemia, and/or volume overload refractory to med Rx, or uremic symptoms. The “conservative strategy” was THE ONLY strategy at the time. When I’m on CCU now, I consult Nephro for IHD for those same reasons. And our nephrologists only offer it for those indications. What exactly is the “standard of care” in those 4 hospitals? Are they dialysing people just for a serum Cr number? If so, that seems insane.
I agree a “non-inferiority” design would have been more appropriate. Less invasive, less cost….”not worse” would have been plenty good enough to adopt this over the “conventional” standard.
Thank you for writing about this study. More medical and surgical interventions need to follow this approach: “Some smart experts have proposed a conservative approach to dialysis, where it is deferred unless it has to be done.”
The financial savings of not doing dialysis: another procedure on a sick patient to put some type of dialysis catheter in, all the equipment and supplies, staffing needed, and the trauma to the already compromised patient—these are significant considerations.
I’d say it passes the ‘what would I want for my family member?’ test. I would choose conservative all day long.
Wait, WHAT?
1) No scientific support for this care?
"The conventional approach has been ensconced for years—without obvious trial support. "
I think this study gives comfort to those who dialyze only when it is absolutely necessary to prevent life threatening complications, like the AEIOU I learned in training.
Oh wow. Late 50s