Nephrology Literature Watch†

Citation: Mehrotra R, Chiu Y-W, Kalanatar-Zadeh K, et al. Similar out-comes with hemodialysis and peritoneal dialysis patients with end-stage renal disease. Arch Intern Med. 2010;171: 110-118.

Analysis: A scarcity of suitable donor organs for kidney transplantation necessitates that the majority of patients with end-stage renal disease (ESRD) must undergo some form of dialysis instead of or prior to transplantation. The two most common modalities of dialysis employed are in-center hemodi-alysis (HD) and home peritoneal dialy-sis (PD). Although peritoneal dialysis is associated with significantly lower costs, and despite the financial incen-tives from CMS, only 7% of dialysis patients use this modality. A number of reasons for this lack of acceptance have been postulated including the per-ception that peritoneal dialysis leads to poorer patient outcomes.

The perception that PD when com-pared with HD might lead to lower overall survival for ESRD patients requiring renal replacement therapy had emerged from epidemiologic observa-tions in the 1980 s and 1990 made from the U.S. Renal Data Systems (USRDS) database. The survival advantage for HD versus PD appeared to be greater for patients with certain clinical char-acteristics including for the elderly, the obese, for those with cardiovascular disease, and for those with diabetes. In 2002, Collins and colleagues report-ed that survival outcomes in elderly patients treated with peritoneal dialysis were significantly lower than in those treated with HD, even after adjustment.1 Stack and coworkers demonstrated in the USRDS cohort of patients initiated on dialysis in 1995-1997 that survival for those with patients with high BMI was better with hemodialysis than with PD.2 And more recently, Johnson and colleagues showed, using data from patients in Australia and New Zealand, an increased occurrence of cardiovas-cular events after one year of treat-ment with PD compared with HD.3 The CHOICE study demonstrated a significantly higher risk of death for patients undergoing PD versus HD only the second year of follow-up, even after adjustments. Furthermore, the risk of death was nearly twice as high in perito-neal dialysis patients with cardiovascu-lar disease versus the same population of patients receiving hemodialysis.4

Against these and other observa-tions are recent findings that suggest that in the current era that patient sur-vival with PD and HD are similar.5 The current studies attempt to overcome the limitations described previously with the earlier studies.6, 7, 8 These studies use methods that are increasingly employed in population-based studies of com-parative effectiveness. This study by Mehrotra and colleagues uses these advanced methods.

Validity and threats to validity: These authors used USRDS data to examine survival trends for new dialy-sis patients initiating treatment with peritoneal dialysis and hemodialysis in three 3-year cohorts (1996-1998, 1999-2001, and 2002-2004). Using a non-proportional hazard marginal struc-tural model, the authors observed no difference in survival for patients on PD versus HD for the latest time peri-od. The study showed that patients on peritoneal dialysis experienced a progressive reduction in mortality rates such that for the 2002-2004 compari-son there was no signifi cant difference in the risk of death between the two modalities through fi ve years of follow-up. Furthermore, when compared with earlier time periods, these authors dem-onstrated greater improvements in sur-vival for those patients treated with PD for each strata of the ESRD population that they studied separately. They strati-fied the population according to diabetic status, age less than or greater than 65, and the number of other comorbidi-ties. Furthermore, the survival rates in all follow-up periods showed a greater increase over time in peritoneal dialy-sis patients. Although they observed a persistent survival disadvantage for PD among diabetic patients with at least one comorbidity and/or who were older than 65, even this effect attenuated over time.

Advanced methods, including mar-ginal structural model, were used in this study to analyze the data. Additionally, this study represents the largest study completed to date on this topic. It has excellent external validity to the United States. Furthermore, it has been adjusted for the probability of censor-ing, especially where this is due to renal transplant, which further reduces bias. The study analyzes the cumulative effects of the initial choice of dialysis modality adjusted by propensity score. Thus, by these methods, the peritoneal dialysis and hemodialysis groups will be similar in important measured and unmeasured baseline effect modifi ers and confounders.

This study does have limitations. The ideal study design to evaluate a therapeutic intervention on patient-cen-tered outcomes is a randomized control trial. However, randomized controlled trials in this setting have been unsuc-cessful. Thus, we are dependent on observational databases analyzed with adjustment for comorbidities, etc. There is, however, likely an underestimation of comorbidities from the information obtained from Medical Evidence Form 2728, and therefore one cannot exclude residual confounding. The authors note that this study does not provide evi-dence for an explanation as to why there was a differential improvement in outcomes for patients treated with peri-toneal dialysis compared to hemodialy-sis. Possibilities might be related to true improvements in care for PD patients such as reductions in risk for infection or better prescription management in peri-toneal dialysis patients. Alternatively, one cannot entirely exclude the infl u-ence of selective assignment of healthi-er patients to PD therapy.

Clinical bottom line: Previous studies comparing outcomes between peritoneal dialysis and hemodialysis have provided inconsistent results, mostly due to the populations studied, the era, and the analytical methods used. These authors, by employing propensity scoring and marginal struc-tural analysis applied to observational cohorts from three eras, provide a more robust measure of the infl uence of renal replacement therapy modality on changes in survival over time and sup-port a view that survival for patients on peritoneal dialysis and hemodialysis in the current era is in fact similar. These findings in turn may have healthcare cost implications in terms since PD is about $20,000 less expensive per year than HD. Questions remain that are relevant to patient choice of modality: Is quality of life better with PD than HD further amplifying the cost-effec-tiveness of PD versus HD? Were the improvements in outcomes seen in this study related to patient selection? Would the results have been the same if more patients were initiated on PD and only converted to HD when they failed PD? Does the adequate training of nephrologists in peritoneal dialysis and the structure of PD programs affect survival outcomes? We should expect sophisticated comparative effective-ness studies using the newer methods exemplified by this study to address many of these questions in the absence of large long-term randomized con-trolled trials.

Citation: Kinsella SM, Coyle JP, Long EB, et al. Maintenance hemodialysis patients have high cumulative radiation exposure. Kidney Int. 2010;78:789-793.

Analysis: In this recent era of med-icine, there has been a rapid increase in technology with a steady growth in the options and modalities of diag-nostic imaging. Concurrent with the increased use, significant attention has been focused on the risk of ionizing radiation and the potential harms asso-ciated with its frequent use.9 It has been shown with multiple data sources that there is an association with radiation exposure and malignancy, birth defects, and heart disease and stroke.10 Low dose ionizing radiation (in atom bomb survivors and workers in the nuclear industry) has been associated with an excess relative risk for these adverse outcomes, even at total exposure doses less than 100 millisievert (mSv).11, 12 Thus, guidelines have been established to monitor and limit exposure in work-ers in the healthcare and nuclear industries to less than 20 mSv per year.13 Patients, however, are often exposed to higher doses of radiation, in the belief that the benefit of accurate and early diagnosis of potentially harmful dis-eases outweighs the risk of low doses of ionizing radiation.

Estimating the risk of cancer or other potential outcomes, such as car-diovascular disease due to radiation dose, is difficult given the infl uence of age, gender, exposure dose, absorptive dose, and multiple genetic factors on these outcomes. The National Research Council issued a comprehensive report in 2006 detailing the biological effects of ionizing radiation and providing an estimate of risk for such exposures.9 Thus, a chest x-ray delivers an effective dose of approximately 0.1 mSv, with a CT scan of the abdomen having a dose of 10 mSv, a dose that is a hundred times greater than that with a chest x-ray.14 This has led some to estimate that 1.5-2% of all solid cancers in the U.S. might be due to the widespread use of CT scans.15 Reports have indicated that radiation exposure from medical imaging has increased by a factor of nearly six-fold since the early 1980 s, and extrapolations from one large data-base of nearly one million non-elderly patients suggest that up to four million Americans are exposed to more than 20 mSv of radiation per year from medical imaging.9

While these numbers are concern-ing, they do not adequately describe the exposure risk for certain segments of the population. The elderly and those with chronic diseases might be expected to have much higher rates of expo-sure. The study by Kinsella and col-leagues represents the fi rst comprehen-sive assessment of exposure in ESRD patients. This study is a retrospective analysis of 100 maintenance hemodi-alysis patients from a single Irish uni-versity-based dialysis center evaluated for their cumulative effective radiation dose over a median follow-up of 3.4 years. The authors report that a substan-tial number of prevalent ESRD patients have levels of radiation exposure in excess of 20 mSv and that there is a sig-nificant association between radiation dose and cause of end-stage renal dis-ease, history of ischemic heart disease, transplant waitlist status, number of in-patient hospital days over the follow-up period, and risk of death.

Validity and threats to validity: This study is a retrospective analysis of a relatively small number of dialysis patients who were, however, followed over a significant period of time–– thereby increasing the patient years of observation and hence the precision of the exposure estimates. A prolonged time frame of observation is extremely important when exposure is the fac-tor of concern, especially in the case where the intensity and frequency of the exposure may vary widely. Since this study was designed to quantify cumula-tive exposure for ESRD patients, the optimal study design is a prospective cohort. Threats to validity that might be introduced when a cohort study is performed retrospectively include the misclassification of exposure from sig-nificant opportunities for unmeasured exposures in other healthcare settings, the selective exclusion of patients, who by virtue of their disease severity, might have differing degrees of exposure, and the exclusion of significant numbers of patients from analysis because of being lost to follow-up. Each of these poten-tial threats to validity are minimized in this study by virtue of the healthcare systems structure that required that all ESRD patients in the region be followed by a single unit for their dialysis and their imaging studies.

When considering applicability of this study to other dialysis settings, certain features of this study must be considered. The population studied was entirely of white race with a high per-centage of patients whose etiology of renal failure was glomerulonephritis (28%). While this may be representative of a certain subset of dialysis patients, it may not fully represent the North American dialysis population where exposure may be influenced by such factors as ethnicity and co-morbidities. The study cannot account for the infl u-ence of the structure of the healthcare systems on the use of imaging modali-ties and, hence, exposure. Additionally, most study participants were enrolled in the study at the time of hemodi-alysis initiation. Twenty-two percent, however, were prevalent patients with a mean vintage of 1.8 years. Exposure may have been more limited in this lat-ter class of patients given the fact that a significant number of procedures and interventions are performed at the time of dialysis initiation.

Nevertheless, this study is the fi rst to quantify ionizing radiation expo-sure in hemodialysis patients, a unique subset of patients who are generally sicker and have more co-morbidities than the general population. CT scans accounted for 66% of the total exposure dose, while vascular access imaging contributed to 14% of the total dose. Thirteen percent of the patients in this study received a cumulative radiation dose of greater than 75 mSv, a level associated with a 7% increase in cancer-related mortality.12 Not surprisingly, this group of dialysis patients received a higher cumulative radiation dose than other reported databases of non-dialysis patients.9

Clinical bottom line: This study should raise awareness and prompt future studies examining the impact of low-level ionizing radiation dosage on the health of hemodialysis patients. In the United States, where the lack of a unified medical record may result in more unnecessary repetition of imaging studies, exposure may be even higher. In future studies, it will be most impor-tant to focus on patient-centered out-comes addressing such questions as: Given this specialized subset of patients, do higher levels of exposure increase mortality or morbidity? Would restriction or withholding of necessary imaging studies due to excess concerns about radiation cause harm because of delayed or inaccurate diagnoses? These questions will likely require a prospective cohort or for rare outcomes a nested case-control study.