Hemodialysis vs. Peritoneal Dialysis

My attention was caught by the recent article in CJASN which compared the mortality of peritoneal dialysis (PD) and hemodialysis (HD) patients in the first 2 years of dialysis therapy. When comparing survival outcomes of PD and HD patients, the data we have so far is based on observational studies. A randomized controlled study has never been successfully completed because of difficulties in randomization. The only randomized study so far - the NECOSAD study (Netherlands) managed to randomize only 5% of the eligible subjects3.  

Most of the observational studies looking at survival had methodological limitations like suboptimal adjustment for modality switch over time (PD patients more likely to switch to HD), inability to account for time-varying confounding by laboratory values and inappropriate adjustment for the differential longitudinal censorship of transplantation across modalities (PD patients more likely to get a transplant). While analyzing such time-varying covariates which are simultaneously confounders as well as predictors of outcome and subsequent exposure, traditional methods like logistic or proportional hazards regression are biased and hence they pose unique analytical challenges. Hence a new statistical model – a Marginal Structural Model (MSM) which employs  inverse probability weights (IPWs) to determine the effects of these time varying covariates on the primary outcome (which was survival  in this study) was utilized  in this study. In order to adjust for the effect of each dialysis modality and censorship from transplantation, a combination of inverse probability of treatment weights (IPTWs) and inverse probability of censoring weights (IPCWs) was used. The IPTW (or IPCW) will have estimated probabilities of treatment (or censorship) using baseline covariates as the numerator and estimated probabilities of treatment (or censorship) using baseline and time-dependent covariates as the denominator. The MSM helped to derive meaningful survival data, adjusting for the above mentioned confounders

The study used information from two large databases viz. USRDS and Da Vita, providing a large cohort of almost 24000 incident dialysis patients. Separate analysis was conducted using a Kaplan–Meier survival curve, cox proportional hazards and the MSM model. A 48% greater survival for incident PD patients at 2 years was found using the MSM model. These findings were in contrast to findings in other studies in the past which showed either no difference in survival or marginal survival advantage especially in the first year for PD compared to HD5. Additionally, a comparison between the cox model and the MSM   demonstrated that the time-dependent confounders determined the difference in survival. Changes in modality during the first 2 years of dialysis affected the survival patterns over time and the reason for this trend is not completely understood at this point. This study supports greater use of PD in the treatment of ESRD patients especially in US where less than 8% of prevalent patients with ESRD are on PD.A comprehensive dialysis modality education program should be encouraged to expand the practice of PD.

See these two previous posts on the debate between PD and HD.

Posted by Bijin Thajudeen

Time for a change?

This month's issue of NDT has an interesting debate concerning whether or not clinical laboratories should start reporting CKD-EPI GFR instead of MDRD. The pro side is here while the con side is here. Basically, the argument for changing is that there is less bias in the CKD-EPI equation, it is more accurate at higher GFRs and more accurately classifies patients as stage 3 as opposed to stage 2 (in terms of overall prognosis). The counter-argument is that, although there is a slight decrease in bias associated with the use of CKD-EPI, it is not any more precise than MDRD - this is more a fault of creatinine as a test of renal function rather than a specific problem with the equations. It should also be mentioned that the CKD-EPI equation is not necessarily better in all circumstances - as documented by Leo in this post about renal transplant recipients.

To (perhaps) settle the argument on one side, the moderator of the debate wrote a commentary and came down on the side of changing to the CKD-EPI equation. The argument is that, even if the improvement is slight, we, as a nephrology community, should not settle for something that is clearly inferior in most circumstances. MDRD was developed in a population of patients with CKD and therefore does not accurately reflect GFR in healthy populations. For this reason, in the research community, there has been a move towards more use of CKD-EPI in the recent past as it is more appropriate for epidemiologic research. The switch to CKD-EPI would not require the use of any new analytes - a simple change in coding in the computers reporting results. In fact, a number of organizations have already switched.

Two other things to mention. Neither equation has been properly validated in Asian populations and this needs to be remedied. Secondly, the role for cystatin C-based or combination equations is still uncertain. Cystatin C is a better predictor of outcomes than creatinine but there are many non-GFR determinants of cystatin C that are likely biasing this and are not related to renal function. Also, the cystatin C test is expensive and has not been fully standardized. There may be a place for the combination equation in patients with borderline GFRs (45-60) in whom the diagnosis of CKD is uncertain.

Falsification Analysis

Interesting paper this month in JAMA about post-marketing studies of adverse drug effects. Randomized controlled trials are obviously the gold standard for the detection of common adverse events related to treatment. The problem is that, if an adverse event is rare, it is unlikely to be detected by a normal RCT. As a result, there has been a move recently towards conducting post-marketing studies of commonly used drugs to identify rare adverse effects. One such effect mentioned in the study is the association between bisphosphonate use and atypical femoral fractures.

The other commonly cited example recently was the association between PPI use and community acquired pneumonia that has been noted in multiple studies. The putative mechanism is that it is due to a reduction in gastric pH. The problem is the question of residual confounding - is there an alternative reason that these patients have more pneumonia. Are these patients simply sicker overall? Are PCPs who prescribe PPIs more likely to diagnose pneumonia? Just because there is a plausible mechanism doesn't make it true.

One potential solution is to perform a falsification analysis. Once you have determined the primary outcome of the study (in this case pneumonia) with a plausible outcome, you then perform a series of prespecified analyses with other non-plausible outcomes. If all of the outcomes are associated with the use of PPIs, it suggests that the association is more likely related to residual confounding rather than a real effect.

In the study referred to in the JAMA article, the authors, working from registry data, not only found an association between PPI use and pneumonia but also with osteoarthritis, urinary tract infections, rheumatoid arthritis, chest pain, DVTs and skin infections. Thus, they suggested that the association with pneumonia was more likely to be confounded because of the lack of a plausible relationship with these other adverse events. One criticism I would have is that I could think of perfectly reasonable hypotheses for why PPI use could be associated with OA and RA (use of NSAIDs) and chest pain (GERD). Another important point is that if this is not done properly (prespecified adverse events) you could find an association between the use of a drug an some adverse event if you tested enough and it could be used to wrongly refute the association between a drug and a problem.

Still, the whole article is a fascinating insight into the problems with post-marketing studies of drugs in the wider population.

Miracle Drug?

The above figure compares long term survival in a subgroup of a trial that was published in Circulation in 1980. This was a randomized controlled trial of just over 1000 patients with known cardiovascular disease who were treated with medical therapy alone. The patients were randomized to two treatment groups and were followed for 5 years. In the primary analysis, there was no statistically significant difference in survival between the two groups. However, a subgroup analysis that compared only patients with 3-vessel disease and LV dysfunction at baseline (~200 patients in each group) found that the outcomes were significantly better in group B (p=0.025).

So what was this treatment that was so successful in reducing mortality in group B? There was no treatment. The patients were randomized to the two groups and then simply followed with usual therapy. This study was designed to show the danger of subgroup analyses and why they should be taken with a grain of salt. When the authors looked deeper into the data, it became apparent that the patients in group B were not as sick as those in group A and that the survival difference was non-significant in a multivariable analysis. However, when they further stratified the patients by only including those with no history of congestive heart failure, the difference between the groups became more significant and remained significant in the multivariable analysis (p=0.01).

We are often confronted by negative clinical trials in nephrology and other disciplines and there is a natural tendency to try and find something useful when these trials are completed. Like any multiple comparison, if you do enough subgroup analyses, you will eventually find one that is significant. Any good statistician will tell you that this needs to be accounted for in the final analysis but this is not necessarily always done. Think about this trial when you are reading about the wonderful effects of a treatment that was negative for most but efficacious for a small group of patients with very specific attributes.