Fracture Risk in Living Kidney Donors

• Summary *

Background: Individuals with chronic kidney disease have an increased risk of fracture, but the long-term fracture risk among living kidney donors remains unclear. Prior studies in donors have shown changes in bone biomarkers, including elevated levels of parathyroid hormone; however, follow-up has been limited (≤5 years), and these studies have not evaluated fracture outcomes. Only one prior study has assessed fracture risk in living kidney donors and found no increased risk, with up to 17 years of follow-up. Leveraging linked administrative healthcare data across three provinces, fracture risk can be evaluated in living kidney donors with more than 25 years of follow-up and improve upon prior methodologies through rigorous matching strategies and control selection.

Methods: The investigators will conduct a matched population-based retrospective cohort study of living kidney donors using provincial healthcare administrative databases from Ontario, Alberta, and British Columbia. Using linked databases from each province, individuals who have donated a kidney between 1992 and 2024 will be identified and compared to matched nondonors (1:10) with similar baseline health status and demographic characteristics. Donors and nondonors will be matched at the time of donation for age, sex, year of nephrectomy (simulated nephrectomy in nondonors), residency (urban or rural), income quintile, and previous fracture. The primary outcome will be a composite measure of fragility fractures, including fractures of the hip, femur, humerus, wrist/forearm, and pelvis. Secondary outcomes will involve analyzing each of these fracture types separately. In a tertiary outcome analysis, vertebral fractures will be examined as part of the composite fragility fracture outcome and independently, given the nature of how this fracture type is captured in health administrative data. Vertebral fractures are often underrecognized because they can be asymptomatic and may only be identified incidentally through imaging. Privacy-preserving methods will be used to combine results and estimate risks across these three provinces, as individual-level data cannot be shared due to privacy constraints.

Study outcome: This study aims to determine whether living kidney donors have an increased risk of fragility fracture (defined as a composite of hip, femoral shaft, humeral, wrist/forearm, and pelvic fractures) compared with carefully selected nondonors of comparable baseline health. Secondary objectives include identifying risk factors for fracture in both donors and nondonors and evaluating fracture risk within prespecified subgroups.

• Literature Review *

Individuals with reduced kidney function, defined as an estimated glomerular filtration rate (eGFR) of 15-59 mL/min/1.73 m², have a 1.5- to 3-fold higher risk of fragility fractures compared with those with normal kidney function (eGFR ≥60 mL/min/1.73 m²). Among living kidney donors, approximately 40% have an eGFR of 60-80 mL/min/1.73 m² and 10% have an eGFR of 30-59 mL/min/1.73 m² ten years post-donation, raising concern about potential long-term skeletal consequences.

To date, only two studies have examined fracture risk in living kidney donors. A population-based matched retrospective cohort study by Garg and colleagues (2012), using ICES administrative data, found no significant difference in fracture risk between donors and nondonors after accounting for baseline health differences, with a maximum follow-up of 17 years. More recently, Kremers and colleagues (2024) used a survey-based approach and reported a lower overall fracture risk among donors but an increased risk of vertebral fractures compared with matched controls after a mean follow-up of 25 years. Together, these findings suggest that if an association between living kidney donation and fragility fractures exists, it may emerge over longer follow-up periods, particularly as donors age and kidney function declines below an eGFR of 60 mL/min/1.73 m². To ensure that living kidney donation remains safe across the donor lifespan, updated population-based analyses with extended follow-up and improved methodology are warranted.

• Research Setting *

The research will take place at ICES (ices.on.ca), a non-profit organization recognized under Ontario's health privacy legislation as a prescribed entity. This designation allows ICES to access and use health and demographic data without individual consent for the purpose of evaluating and improving the health system. The use of data in this project falls under Section 45 of Ontario's Personal Health Information Protection Act (PHIPA) and it does not require approval from a Research Ethics Board.

In Ontario, data will be obtained from several linked administrative databases, including the Trillium Gift of Life Network (TGLN), the Registered Persons Database (RPDB), the Ontario Health Insurance Plan (OHIP) database, and the Canadian Institute for Health Information (CIHI) Discharge Abstract Database (CIHI-DAD), National Ambulatory Care Reporting System (NACRS), and Same Day Surgery (SDS) databases. For most hospitalization and physician billing data, information is available from 1991 onward; therefore, July 1, 1992, will mark the start of the accrual period. Near-complete capture for all study variables is anticipated. Living kidney donors will be identified primarily using TGLN, with CIHI-DAD used for donor identification in years where TGLN data are unavailable.

To ensure consistency across provinces, equivalent administrative databases will be used in Alberta and British Columbia. In Alberta, data will be accessed through the Alberta Kidney Disease Network (AKDN) and Alberta Health Services (AHS), subject to analyst assignment and data availability. Living kidney donors will be identified using CIHI-DAD. Demographic characteristics and vital statistics will be obtained from the Alberta Provincial Registry and Vital Statistics databases, while information on hospital admissions, diagnoses, and healthcare encounters will be drawn from CIHI-DAD, NACRS, and the Alberta Practitioner Claims database.

In British Columbia, data will be accessed through Population Data BC (PopDataBC) and the Healthcare Data Platform BC (HDPBC). Living kidney donors will be identified using data from BC Transplant and the Patient Records and Outcome Management Information System (PROMIS), accessed via BC Renal; if this data is unavailable, CIHI-DAD will be used for donor identification. Demographic and vital statistics data will be sourced from Population Data BC's Consolidation File/Central Demographics database. Hospital admissions, diagnoses, and health care visit data will be obtained from CIHI-DAD, CIHI-NACRS, and Population Data BC's Medical Services Plan database.

This retrospective cohort study will use existing administrative healthcare data. In accordance with best practices for observational research, the study description, design, and statistical analysis plan will be publicly registered on ClinicalTrials.gov prior to outcome analyses.

• Statistical Analysis *

Baseline characteristics: Continuous variables will be presented as mean (standard deviation) or median (interquartile range [IQR]), as appropriate, and categorical variables as frequency (percentage). Due to our large expected cohort size and the influence of sample size on statistical testing, the characteristics of donors and nondonors at the time of cohort entry will be compared prior to and after matching using standardized mean differences, where a value >0.10 is considered meaningfully different. Characteristics specific to the donors will also be reported depending on data availability (e.g., pre-nephrectomy kidney function and donor relationship to the recipient).

Period of observation: Participants will be accrued between July 1, 1992, and March 31, 2024, subject to data availability. Follow-up will begin at cohort entry and will be censored at the earliest occurrence of the relevant study outcome (time to first event), death, emigration from the province, or the end of the observation period (March 31, 2025).

Assessing risk of the primary, secondary, and tertiary outcomes: The primary analysis will evaluate the association between living kidney donation and the risk of hospitalization for a composite fragility fracture, defined as hip, femoral shaft, humeral, wrist/forearm, or pelvic fracture. Only the first fracture event will be considered. Hazard ratios (HRs) and 95% confidence intervals (CIs) will be estimated using Cox proportional hazards regression with robust variance estimation to account for correlation within matched sets. The proportional hazards assumption will be evaluated using a Kolmogorov-type supremum test based on 1,000 simulated patterns. If the proportional hazards assumption is violated, a stratified log-rank test will be used to assess statistical significance, and differences in restricted mean survival time at 20 years with corresponding 95% CIs will be reported. This analytical approach will be repeated for each fracture site examined separately (secondary outcomes: hip, femur, humerus, wrist/forearm, and pelvis) and for a composite outcome that additionally includes vertebral fractures and vertebral fractures separately (tertiary outcome).

Risk factor analysis: Risk factor analyses will be conducted to identify factors associated with fracture risk among living kidney donors and nondonors. Age, sex (male vs female), rurality of residence (urban vs rural), neighborhood income quintile (per quintile increase), year of donation (simulated donation in nondonors), and history of fracture within five years prior to cohort entry (yes vs no) will be examined as predictors of fracture risk. Associations will be estimated using Cox proportional hazards regression models, with results reported as hazard ratios and 95% confidence intervals.

Subgroup analysis: Prespecified subgroup analyses will be conducted to assess whether the association between living kidney donation and fracture risk varies across clinically relevant subgroups. Analyses will be stratified by age at cohort entry (<55 vs ≥55 years), sex (male vs female), and index period (1992-2001, 2002-2012, and 2013-2024). In addition, a combined age-sex subgroup will be examined using four categories (younger females, older females, younger males, and older males). For each subgroup, hazard ratios and 95% confidence intervals will be estimated using Cox proportional hazards regression with robust variance estimation, consistent with the primary analysis approach.

Combining outcome results from Ontario, Alberta and British Columbia: To improve statistical power and generalizability, results from Ontario, Alberta, and British Columbia will be combined for the outcomes. A privacy-preserving Cox regression approach will be used for multisite studies, where individual-level data cannot be shared. This method requires a single transfer of summary-level outputs from each province and yields results identical to those obtained from pooled individual-level data. The approach assumes province-specific baseline hazards and accommodates survival and binary outcomes, with confounding control (e.g., matching or weighting) conducted within each province. Summary-level risk-set tables generated by each province will be securely transferred to the coordinating site to estimate combined hazard or risk ratios with 95% confidence intervals. To comply with privacy regulations, all cell sizes of five or fewer will be suppressed (reported as ≤5) in publications, and all team members will adhere to required data confidentiality and data use agreements.

Additional prespecified analyses (Ontario only): Supplementary descriptive analyses will be conducted in the Ontario cohort to characterize additional laboratory and medication variables related to bone health.

(i) The number and proportion of individuals with multiple fractures during follow-up.

(ii) Bone mineral density (BMD) testing, reporting the distribution of the number of tests during follow-up. These data reflect whether a test was performed, not the test results.

(iii) Among a sub-cohort aged ≥66 years at index, the Ontario Drug Benefit (ODB) database will be used to identify new osteoporosis-related prescriptions, excluding individuals with relevant prescriptions in the year prior to index to capture new users of these medications. Medications of interest include bisphosphonates, systemic estrogen/raloxifene, denosumab (after February 2012), prescription vitamin D/calcium, and teriparatide (after February 2006).

(iv) If an adequate amount of measurements is available in a sub-cohort, bone metabolism markers will be assessed, including parathyroid hormone, corrected calcium, alkaline phosphatase, 25-hydroxyvitamin D, and phosphate. Limitations related to laboratory data capture in our data sources may affect the completeness of follow-up test counts.