Patient and Technique Survival among a Canadian Multicenter Nocturnal Home Hemodialysis Cohort

Discussion

This study is novel in that it describes patient disposition from the largest NHD cohort assembled for this purpose. To our knowledge, this is also the first report of predictors of patient and technique survival that incorporates patient-level variables that are relevant for home hemodialysis, such as duration and frequency of dialysis, type of vascular access, and degree of patient independence for performing NHD. The results are encouraging with an overall mortality rate of approximately 10% over the 12 year since the cohort start date. Advanced age, diabetes status, CVC use, and dependence on others to perform the dialysis procedure are possible predictors of program exits as a result of adverse events. Technique failure, as defined by a physical or cognitive inability to perform NHD at home either alone or with assistance, is uncommon (<3% at 1 year and <5% at 5 years) and is best predicted by age alone. This study also demonstrated an important relationship between transplantation and NHD: almost one quarter of NHD patients had previously undergone transplantation, and a similar proportion exited NHD programs to undergo transplantation. This cohort also highlights that NHD patients are younger (mean age 45.7 years) and largely without diabetes (12.1% prevalence of diabetes). Our finding that the NHD treatment center significantly affected adverse events suggests that differences in clientele demographics, cultural background, socioeconomic status, education, and other such factors that may be different between the centers likely result in residual confounding. A center effect also highlights the need to examine practice patterns among NHD programs with eventual standardization of procedures and protocols to optimize outcomes.

This investigation builds on previous work by Komenda et al. (10) that outlined the home hemodialysis program outcomes in British Columbia, Canada, between 2004 and 2007. That cohort included 105 patients, approximately 70% of whom were performing NHD. Those authors described 19% program attrition as a result of all causes of program exit after 1 year (this study had a 1-year program attrition of approximately 10%). NHD survival from adverse program exits (death and all other program exits) censored for transplantation was 85% at 1 year, and 1-year technique survival was reported as 94% (97.9% in this study). Given the variable demographic makeup of these populations (e.g., our cohort was 6 to 7 years younger than the British Columbia cohort) and the different rates of transplantation in these groups (12% in the British Columbia cohort versus 22% in our study), these unadjusted statistics are reasonably similar.

Whereas technique failure is explicitly defined in the peritoneal dialysis literature as an inability to continue that modality (usually because of peritonitis, ultrafiltration failure, or inability to perform peritoneal dialysis independently), necessitating a unidirectional modality change to hemodialysis, no consensus definition has yet emerged in the NHD literature. For the purpose of this investigation, we defined technique failure as a physical or cognitive inability to perform home NHD either independently or with assistance and censored our analysis of technique survival not only for death and transplantation but also for program exits unrelated to physical or cognitive functional decline. Given that many NHD patients are young, independent, and employed, the motivation for exiting the program includes reasons unrelated to ESRD management (e.g., geographic relocation for the purpose of employment to a region where NHD is not offered) and would not truly represent a failure in a patient's ability to perform NHD. The highly selected nature of the NHD population precludes meaningful unadjusted comparisons with technique failure reported for other dialysis modalities.

Several important strengths and limitations of this study warrant consideration. Most notably, this study maximized the sample size of a large multicenter Canadian NHD cohort, including patient-level variables related to NHD frequency and duration, as well as vascular access type and the degree of (in)dependence with which NHD is performed at home. Indeed, this analysis suggests that hemodialysis-specific variables such as CVC use and the dependence on a caregiver to perform NHD may predict poor outcomes for these patients; however, we were unable to demonstrate an effect of dialysis duration or frequency on adverse events or technique survival. That these variables were not predictive in univariate analyses or were no longer predictive in multivariable models may be reflective of insufficient statistical power rather than an actual absence of association. In addition, because the event rate for adverse outcomes is low (26 deaths, 10 technique failures), the number of variables that can be incorporated into survival models is limited. Furthermore, this study lacks comorbidity data as well as data for patients who commenced NHD training but never graduated to perform NHD in their home setting. This latter deficit means that this study did not address issues concerning training failure (as distinct from technique failure), arguably an important outcome in the recruitment of patients to NHD. It must also be acknowledged that the mean duration of ESRD before commencing NHD was almost 5 years, so patients who self-administered this therapy were survivors of previous modalities, thereby contributing to the overall excellent survival observed in this cohort. Initiatives such as the International Quotidian Dialysis Registry may eventually be able to help address some of the aforementioned limitations (11).