Does dialysis modality matter in diabetic patients with end stage renal disease?

By the end of 2007, over 500,000 Americans were afflicted with end stage renal disease (ESRD), and almost 368,000 patients were undergoing dialysis therapy. Diabetes remains the most common cause of ESRD and accounts for over one-half of all new dialysis patients in the United States. Diabetic dialysis patients have poorer outcomes in general compared to nondiabetics. As a result, clinicians and researchers alike are searching for ways to improve outcomes of these patients.

The basic question of whether dialysis modality per se affects the survival of ESRD patients has been debated since the introduction of peritoneal dialysis (PD) for the treatment of ESRD. This issue is particularly important for diabetics because several observational studies have shown a higher risk for death among older diabetic patients treated with PD. Indeed, based on the results of some observational studies, some have questioned whether it is ethical to offer PD as an option for older diabetics.

There are many theoretical advantages and disadvantages to both hemodialysis (HD) and PD (Table 1). Special consideration should be given to diabetics with ESRD since these patients have more severe vascular disease and a higher risk for infection than their nondiabetic counterparts. For diabetics with ESRD, PD may provide certain advantages over HD (Table 2). On the other hand, valid concerns about the adverse effects of glucose-based peritoneal dialysis solutions, including the risks of weight gain and worsened hyperglycemia, may limit PD utilization in diabetics.

Table 1.
Table 2.

Many researchers have sought to answer the question of which dialysis modality is superior. A randomized controlled trial (RCT) would most reliably examine whether any differences in outcomes of patients treated with HD or PD are attributable to the dialysis modality per se, but the existing body of literature is unfortunately lacking in RCTs. The Netherlands Cooperative Study of Dialysis (NECOSAD) is the most recent attempt to perform such an RCT. In this study, 735 of 773 eligible patients refused to be randomized to either HD or PD, thus leaving a study group of only 38 patients—far too few from which to make meaningful conclusions. In the absence of RCTs, data examining the effects of dialysis modality are limited to those originating from either observational or prospective cohort studies.

The observational studies examining survival differences between HD and PD patients over the past two decades are heterogeneous in design but manifest a common theme: PD provides an early survival advantage that varies in magnitude and duration depending on patients’ age, diabetic status, and presence of co-morbidities. Interestingly, during the same period of time, eight prospective cohort studies have been published on this topic, five of which report no difference in survival outcomes between the two modalities. Specifically with regard to diabetics with ESRD, most studies suggest that younger diabetics with no additional co-morbidity have a survival advantage while older diabetics with additional co-morbidity have a survival disadvantage when treated with PD. In other subgroups of diabetics, there appears to be no difference in outcomes between patients treated with either of the two dialysis modalities. In these studies, however, the reported statistical differences in hazard ratios translates into small differences in more clinically relevant measures such as life expectancy of patients treated with HD or PD.

It must also be mentioned that most published studies examining the effect of dialysis modality on survival have drawn their cohorts from patients incident before the year 2000. These older cohorts lacked the benefits of more recent advances in PD such as improvements in connectology, catheter exit-site antibiotic prophylaxis protocols to reduce infection risk, and increased awareness of the importance of solute clearances and careful volume management. All of these PD advances have probably contributed to improvements in PD outcomes, but their benefit has not been well captured in the literature.

A comparison of patient and technique survival in incident HD and PD from 1996 to 2003 has shown that the outcome of incident PD patients progressively improved whereas that of HD patients remained unchanged during the same time period. These improvement trends have continued over the past five years, and a more recent analysis of patients starting dialysis in 2002–04 in the United States shows no significant differences in outcomes between modalities, even among older diabetics treated with either HD or PD.

What can we learn from this varied literature? One must first remember that observational studies, although having the advantage of large patient numbers and therefore excellent statistical power, depend on information gathered from previously existing patient databases that contain limited details about the clinical condition of individual patients. This limitation makes it difficult to know whether reported differences in these studies are truly an effect of dialysis modality or whether they result from unaccounted differences between the patient groups (HD and PD). Such observational studies are also subject to inaccuracies as a result of nonrandom allocation of patients into two dialysis groups and in most parts of the world, do not necessarily reflect patient choice. Thus, despite the use of advanced statistical analyses, one has to be careful in attributing any differences in outcomes of patients treated with different therapies to the dialysis modality itself.

Given the lack of strong and conclusive data favoring one dialysis modality over another plus the differential changes in outcomes of patients treated with HD and PD over time in the United States, it appears reasonable that neither dialysis modality should be automatically excluded for any diabetic patient with advanced CKD; rather, careful evaluation of medical factors are likely best addressed on an individual level.

The diabetic patient who ultimately starts HD should take care to limit interdialytic fluid gains and rapid electrolyte shifts, and venous dialysis catheters should be avoided as far as possible. A diabetic starting PD should be prescribed a regimen that minimizes the use of hypertonic glucose-based solutions. Perhaps more importantly, patient preference should play a major role in the decision-making process, but it necessarily requires dialysis education to be provided in a timely and complete manner. Ultimately, the life-saving benefits of dialysis are of no use if the patient is miserable for lack of being given a choice in their dialysis lifestyle.


[1] Anjali Bhatt Saxena, MD, is with the Santa Clara Valley Medical Center in San Jose, CA, and Stanford University School of Medicine, Palo Alto, CA. Rajnish Mehrotra, MD, is with the division of nephrology and hypertension at Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center in Torrance, CA, and the David Geffen School of Medicine at UCLA in Los Angeles, CA.

[2] Disclosures: Rajnish Mehrotra has received research grant support, served as an ad hoc consultant, and received honoraria from Baxter Health Care. Anjali Saxena has served as an ad hoc consultant to Baxter Health Care.