Background End‐stage kidney disease patients have increased mortality compared to the general population. Haemodialysis (HD) of more frequent and of longer duration has been proposed to improve survival but it remains unclear if this is attributed to increased frequency, duration, or both. We aimed to examine the independent effects of session frequency and duration on mortality in incident HD patients. Methods A retrospective cohort study was performed using data from the Australian and New Zealand Dialysis and Transplant Registry examining non‐Indigenous patients aged ≥18 years who initiated HD of ≥3 sessions/week in Australia from 2001 to 2015. Initial dialysis prescription was categorized as session duration >5 h/session compared to ≤5 h/session and session frequency as >3 sessions/week compared to 3 sessions/week. Survival analysis was performed using Cox regression analysis, with multivariable analysis controlling for available covariates. Results We examined 16 944 patients of whom 757 (4.5%) received >3 sessions/week and 518 (3.1%) received >5 h/session. After controlling for frequency, patients initiated on HD sessions >5 h had a significantly reduced risk of mortality compared with patients with HD session ≤5 h (adjusted hazard ratio (HR) = 0.57; 95% confidence interval (CI) = 0.44–0.74). In contrast, patients initiated on >3 sessions/week of HD had a similar risk of death when compared with patients on 3 sessions/week of HD (adjusted HR = 0.97; 95% CI = 0.84–1.13), after controlling for duration. Limitations include potential residual confounding and changes in exposure over time. Conclusion Longer duration rather than increased frequency of treatment appears to reduce mortality in HD patients. This has implications for management and requires further study.

Introduction: In South Africa, 84% of the population rely on public sector healthcare and 16% use mainly private facilities. Since 2012, the South African Renal Registry (SARR) has reported on the treatment of end-stage renal disease (ESRD) in South Africa. Overall rates of renal replacement therapy (RRT) are low, with a prevalence of 183 per million population (pmp). Large disparities exist between the private and public sectors (798 versus 68 pmp), between provinces and between ethnic groups. Private sector patients on RRT are approximately 10 years older, have more diabetes and other comorbid illnesses, and are less likely to be transplanted or use peritoneal dialysis. The aim of this analysis was to explore, for the first time, survival of all incident and prevalent patients on RRT in South Africa. Methods: SARR data were used to estimate 1-year survival in patients on RRT by the Kaplan Meier method. The incident cohort included patients who started treatment between 1 January 2013 and 30 September 2016 and survived 90 days; the prevalent cohort included all those on RRT on 31 December 2015. Data were collected on potential risk factors, including ethnicity, renal diagnosis, first RRT modality, HIV status and public/private sector provision of RRT. Failure events included death and discontinuation of RRT without recovery of function. Patients were censored if they recovered renal function, emigrated or were lost to follow-up. Multivariable Cox regression was used to estimate adjusted hazard ratios. Results: In the incident cohort (n=6187), the mean age was 51.8 years, 54% were Black, 44% were diabetic, 10% were HIV positive and 82% had haemodialysis as their first modality. Overall 1-year survival from day 90 was 90.4% (95%CI 89.6-91.2%). On multivariable analysis, survival was associated with age, renal diagnosis, diabetes and province. The 18-40-year age group and non-diabetics had improved survival. There was no association with ethnicity, first modality or healthcare sector. HIV status was missing in 20% of the cohort and was therefore omitted from the multivariable analysis. The crude 1-year survival rates of HIV-positive and negative patients were 95.7% and 93.8%, respectively, and much lower at 74.8% in patients without data on HIV. The absence of this optional data element seemed to identify sicker patients; those who survived for longer would be more likely to have their virological status captured by the registry. In the December 2015 prevalent cohort (n=10155), the mean age was 51.5 years, 52% were Black, 37% were diabetic, 8% were HIV positive, and 74% were on haemodialysis as their treatment modality. The overall 1-year survival was 90.1% (CI 89.4-90.6%). The crude survival rates of public and private sector patients were 93.4% and 88.4%, respectively. This was not significant on multivariable analysis, where survival was associated with age, diabetes, RRT vintage, and province. The 18-40-year age group, non-diabetics and those with RRT vintage >2.5 years had improved survival. Conclusions: The s...