Kidney failure, CKD progression and mortality after nephrectomy

Data were extracted for 5983 patients enrolled in the Queensland CKD Registry (CKD.QLD) from five clinical sites across Queensland, Australia: Royal Brisbane and Women’s Hospital (n = 1643), Logan Hospital (n = 1534), Mackay Base Hospital (n = 680), Toowoomba Hospital (n = 1470), and Townsville University Hospital (n = 656) [12]. Patients were recruited between June 2011 and June 2018; data were extracted from the registry in January 2020. Inclusion criteria for the registry are a diagnosis of CKD and care under a specialist nephrology practice in Queensland.

Patients were excluded if they had an unclear, unknown, or unspecified primary kidney disease (n = 625), dates were ambiguous or missing (n = 106), baseline eGFR data were missing (n = 27), or if they had kidney failure (eGFR < 15 mL/min per 1.73m²) at the time of recruitment (n = 345). There were 4,880 patients in the final sample. Human research ethics approval was obtained from the Royal Brisbane and Women’s Hospital (HREC/15/QRBW/294) and the University of Queensland (2011000029) and a Public Health Act Approval (QCOS/029817/RD006802) for Queensland Health data linkages. All participants provided written informed consent before being included in the CKD.QLD registry.

The registry is informed by an association with the Statistical Analysis and Linkage Unit (SALU), Queensland Health, who link the CKD.QLD registry patient cohort to the following collections under a Public Health Act approval: Queensland Hospital Admitted Patient Data Collection (QHAPDC), Kidney replacement therapy (KRT), Queensland Activity Based Funding (ABF) Model Output Data, Death Registrations from the (Qld) Registry of Births, Deaths and Marriages and Cause of Death Unit Record File from the Australian Coordinating Registry, informing on institution of KRT and deaths.

Age at recruitment was calculated from the date of birth and date of recruitment. Date of recruitment was defined as the date informed consent was given for inclusion, capturing both incident and prevalent cases. Diabetes mellitus and hypertension were classified as yes, no, or missing. Smoker status was classified as never, former or current smoker, or missing. The primary kidney disease was grouped into five categories (diabetic nephropathy, glomerulonephritis, polycystic kidney diseases [PCKD], acquired single kidney, or other) based on coded diagnosis and free-text fields in the database, where the person’s treating nephrologist assigned the primary cause(s) of CKD based on clinical assessment. The acquired single kidney group included those where nephrectomy was listed as the primary cause of CKD. An ancillary chart review of this group was undertaken to determine the date and indication for nephrectomy. Baseline eGFR was determined from serum creatinine taken at the time of recruitment; eGFR at 12 and 24 months after recruitment was recorded if available in the patient’s electronic medical records. The CKD-EPI equation was used to calculate eGFR. Data collection has been described previously [12].

The primary outcome of interest was a composite of time to the development of kidney failure (defined as eGFR persistently < 15 mL/min per 1.73m²) or commencement of KRT. Secondary outcomes included time to all-cause mortality, change in eGFR, and CKD progression. The CKD.QLD registry provided event data for kidney failure, KRT, and death, after performing data linkages of state electronic data sets. Change in eGFR was calculated over at least a 12-month period, and defined as the difference between baseline eGFR at recruitment, and eGFR at either 12 or 24 months, depending on the most recent value available. Progressive CKD was defined as an annual decrease in eGFR of at least 5 mL/min per 1.73m² without subsequent recovery, the development of kidney failure, or the requirement for KRT within two years of recruitment.

Baseline characteristics were presented descriptively. The association between primary kidney disease and time to kidney failure/KRT (primary outcome) and all-cause mortality (secondary outcome) was analysed using Cox proportional hazards models. When evaluating kidney failure/KRT, participants were censored on either the site-specific date that the database was updated, last date the person was seen before being lost to follow-up, or date of death. When evaluating all-cause mortality, participants were censored on the date the database was updated or date lost to follow-up. All analyses were stratified by baseline eGFR (< 45 and ≥ 45 mL/min per 1.73m²), which was considered a clinically significant cut-point [13]. The reference group was defined as the cohort with diabetic nephropathy as primary cause of CKD and a baseline eGFR ≥ 45 mL/min per 1.73m². A sensitivity analysis evaluated time to kidney failure/KRT with death as a competing event, using Fine and Gray’s proportional sub-hazards model. Change in eGFR was analysed using multivariable linear regression, stratifying analysis by baseline eGFR. The reference group was defined as the cohort with an acquired single kidney. Unlike the other analyses, comparison was only made within instead of between eGFR strata, as the decreased accuracy of the CKD-EPI equation at lower values for serum creatinine may introduce bias if directly comparing the two strata using the same definition for eGFR change [14]. Progressive CKD was analysed using multivariable logistic regression, stratified by baseline eGFR. The reference group was defined as the cohort with diabetic nephropathy and an eGFR < 45 mL/min per 1.73m². In all multivariable models, confounders were identified using directed acyclic graphs. A supplementary analysis of patients with an acquired single kidney for whom a date and indication for nephrectomy was able to be determined from chart review; indication and year were compared descriptively by whether or not the patient progressed to kidney failure. Analysis was performed using Stata 14.2 (StataCorp, College Station, TX).