EARLY ALERT FREQUENCY PREDICTS TECHNIQUE FAILURE IN AUTOMATED PERITONEAL DIALYSIS

Introduction:

The first month of peritoneal dialysis (PD) is a critical time of training and consolidation. Therapy alerts during automated PD (APD) can cause significant disruption to patients’ sleep and quality of life. The relationship of alert frequency during this early period to longer term technique failure is an important one to understand to improve outcomes.

Methods:

Following probabilistic linkage of the Baxter Sharesource database to the ANZDATA registry, we examined the relationship of therapy alarm frequency with the risk of technique failure. We included all adult patients in Australia and New Zealand who commenced APD with the Baxter Homechoice Claria APD cycler over 2019–2023 and continued for at least 30 days (N = 1,927). We calculated the average number of therapy alerts per treatment in the first 30 days and divided the cohort into quartiles (<0.45, 0.45-1.1, 1.2-2.5, ≥2.6 alerts per night). We modelled: overall technique failure (including death, but censored for transplant), death censored technique failure and cause-specific infective and non-infective technique failure. Analyses used Cox regression, and competing risk regression for infective and non-infective technique failure. Models were conditional on continuation of dialysis to 30 days, adjusted for covariates: age, sex, BMI, state/territory, smoking status, comorbidities, and late referral.

Results:

The cohort was  65% male. Median age at PD commencement was 58 (interquartile range 45-68)  years. Median BMI was 27.1 kg/m2 (interquartile range 23.7-31.1 kg/2). Diabetes was reported for 47% as a comorbidity; 34% had a form of vascular disease. Overall technique survival was 78 [95% CI 76-80]% at 1, and 41 [38-45] at 3 years. Death-censored technique survival was 83 [81-85]% at 1 year and 53 [49-56]% at 3 years.

The relationship between alert number and risk of technique failure varied over followup time. Progressively higher rates of technique failure within 12 months of PD start were associated with a higher average alert number, but there was no relationship of alert frequency with risk of technique failure beyond 12 months.

This relationship was seen in univariate analyses, and remained so after adjustment for potential confounders. The trend was seen across all-cause and death-censored technique failure, and also among non-infective and infective causes of technique failure (Figure).

Conclusions:

The number of early alerts reported predicts risk of technique failure among people receiving APD to 12 months, but not beyond. Although this study does not define the mechanism of this relationship, it does lend support interventions targeting interventions to reduce the alert burden in the early stages of therapy. Further analyses will explore relationships of individual alert types and timing with outcomes.

I have potential conflict of interest to disclose.
This work was support by an investigator initiated grant from Baxter Healthcare.

I did not use generative AI and AI-assisted technologies in the writing process.