Association between Duration of Predialysis Care and Mortality after Dialysis Start

Abstract

Background and objectives 

Early nephrology referral is recommended for people with CKD on the basis of observational studies showing that longer nephrology care before dialysis start (predialysis care) is associated with lower mortality after dialysis start. This association may be observed because predialysis care truly reduces mortality or because healthier people with an uncomplicated course of disease will have both longer predialysis care and lower risk for death. We examined whether the survival benefit of longer predialysis care exists after accounting for the potential confounding effect of disease course that may also be affected by predialysis care.

Design, setting, participants, & measurements 

We performed a retrospective cohort study and used data from 3152 adults with end stage kidney failure starting dialysis between 2004 and 2014 in five Canadian dialysis programs. We obtained duration of predialysis care from the earliest nephrology outpatient visit to dialysis start; markers of disease course, including inpatient or outpatient dialysis start and residual kidney function around dialysis start; and all-cause mortality after dialysis start.

Results 

The percentages of participants with 0, 1–119, 120–364, and ≥365 days of predialysis care were 23%, 8%, 10%, and 59%, respectively. When we ignored markers of disease course as in previous studies, longer predialysis care was associated with lower mortality (hazard ratio_{120–364 versus 0–119 days}, 0.60; 95% confidence interval, 0.46 to 0.78]; hazard ratio_{≥365 versus 0–119 days}, 0.60; 95% confidence interval, 0.51 to 0.71; standard Cox model adjusted for demographics and laboratory and clinical characteristics). When we additionally accounted for markers of disease course using the inverse probability of treatment weighted Cox model, this association was weaker and no longer significant (hazard ratio_{120–364 versus 0–119 days}, 0.84; 95% confidence interval, 0.60 to 1.18; hazard ratio_{≥365 versus 0–119 days}, 0.88; 95% confidence interval, 0.69 to 1.13).

Conclusions 

The association between longer predialysis care and lower mortality after dialysis start is weaker and imprecise after accounting for patients’ course of disease.