Temporal changes in dialysate [Na+] prescription from 1996 to 2018 and their clinical significance as judged from a meta‐regression of clinical trials

Over the last two decades, the clinical care of dialysis patients has refocused sharply on fluid volume control. Dialysate [Na+] is a key, albeit under‐investigated, clinical tool for manipulation of fluid volume on dialysis. In the article, we firstly use data from the Dialysis Outcomes and Practice Patterns Study to document the global decrease in dialysate [Na+] that has occurred from 1996 to 2018, and demonstrate the virtual disappearance of [Na+] profiling from routine dialysis practice over the same period. Second, we used data from previously synthesized randomized clinical trial evidence combined with that of a more recently published trail to assess the clinical significance of these changes, estimating the effects of different levels of low dialysate [Na+] on key clinical outcomes. Our analyses suggest that current levels of dialysate [Na+] in some health jurisdictions are possibly causing harm to many patients, especially given that real world populations are significantly less robust and more vulnerable than clinical trial ones. To quote a recent editorial, “more evidence needed before lower dialysate sodium concentrations can be recommended.” That evidence is coming, and no further changes should be made to default customary practice until it is available.


| INTRODUC TI ON
There is ample evidence supporting a causal link between fluid overload and the development of morbidity and mortality in dialysis populations. Although there have always been proponents of aggressive fluid volume control as a means of improving patient outcomes, 1 it has only been more recently that there has been a more systematic focus in this area. 2,3 Recently, there has been renewed interest in lower dialysate [Na + ] as a means for controlling fluid volume, supported by positive qualitative and quantitative reviews of cumulative clinical experience. 4,5 This has provided a much needed evidence base to support the various "calls to action" from clinical governance groups that have made specific clinical recommendations in this area. Nonetheless, definitive clinical trials of lower dialysate [Na + ] are still needed to ascertain the effect of this intervention on harder clinical outcomes, such as hospitalization and mortality.
In this report, we used data from the Dialysis Outcomes and

| Clinical significance analysis
To estimate dose-response relationships in relation to different levels of low dialysate [Na+], we examined effect modification of treatment effectiveness using meta-regression of data from published randomized clinical trials. We extracted data from a recent Cochrane systematic review on the subject, 4 which made comparisons between control groups receiving conventional dialysate [Na+] (the weighted average being 140.8 mmol/L) and intervention groups receiving a lower concentration (the weighted average being 133.9 mmol/L). We combined these data with that from a clinical trial that has been published since the Cochrane review (the SoLID trial), which compares a variety outcomes over one year between patients randomized to dialysate [Na+] of 140 vs 135 mmol/L. [6][7][8][9] In these 13 clinical trials, a range of lower dialysate [Na+] were trialed, spanning from 130 to 137 mmol/L. In addition, a range of conventional dialysate [Na+] were used in the trials, meaning overall differences in dialysate [Na+] between the control and intervention groups spanned 3 to 15 mmol/L across the different studies.
We therefore modeled the effect of these two treatments (the first being level of low dialysate [Na+], the second the level of difference in dialysate [Na+] between groups) on key clinical outcomes using meta-regression. In terms of treatment safety measures, we used intra-dialytic hypotension and cramps as markers for adverse effects from the intervention.
We created meta-regression linear prediction (  These data are illustrated graphically in Figures 1 and 2, and provided in Tables S1-S3.

| RE SULTS
The meta-regression of data from the Cochrane review and the recently published SoLID trial are illustrated in Figures 3-6. Figure 3 illustrates dose-response for the treatment efficacy measure of inter-dialytic weight gain. In those trials reporting this measure, there is a progressive decrease over the entire span of lower dialysate [Na+] and reduction in dialysate sodium, indicating a monotonic dose-response relationship. Figure 4 illustrates the corresponding relation-

| D ISCUSS I ON
The mortality risk of dialysis patients has improved over the last 10 years, although their survival in absolute terms remains similar or worse than many common cancers. 10 The dominant cause of death in this population is cardiovascular in nature; approximately half of modern end-stage kidney disease patients start dialysis with a diagnosis of cardiovascular (CV) disease, and the majority of naïve patients develop it over their remaining lifetime. 11 The most common modality of demise is sudden cardiac death due to bradyarrhthmia, 12,13 and a panacea for this appears some way off given our limited understanding of specific risk factors and casual pathways.
Against this background, one of the most accepted surrogate markers of CV risk is left ventricular (LV) hypertrophy. 14 As importantly, what is the optimal approach for hemodialysis providers? The overall picture suggests that they should not be continuing to decrease default dialysate [Na + ] without further evidence or assessment for benefit and harm. Intradiaytic hypotension-despite its clinical importance-is possibly the most poorly collected adverse event and clinical performance indicator for providers, impeding granularity of research and precision of patient care. 28 At the present time, given the current levels dialysate [Na + ] around the world and the ongoing downward trend, it is possible or even likely that some patients are already experiencing harm from this prescribing pattern. However, we will only be certain about the presence and extent of that harm once the RESOLVE trial is reported.
In essence, Dr Hecking and colleagues are correct-"More Evidence