Risk of Kidney Stones in Living Kidney Donors
Risk of Kidney Stones in Living Kidney Donors
We conducted a population-based matched retrospective cohort study using Ontario's healthcare databases held at the Institute for Clinical Evaluative Sciences (ICES). The province of Ontario, Canada, currently has approximately 13 million residents who have universal access to hospital care and physician services. We conducted this study based on a prespecified protocol that was approved by the research ethics board at the Sunnybrook Health Sciences Centre (Toronto, Ontario, Canada). The reporting of this study follows guidelines set out for observational studies (Appendix A).
We ascertained baseline characteristics, covariate information and outcome data from the records in six databases. Trillium Gift of Life Network is Ontario's central organ and tissue donation agency, and collects information on living kidney donors in the province at the time of kidney donation. We used the Trillium database to identify all adult living kidney donors who had donated between 1992 and 2009 at one of the five major transplant centers in Ontario. We then manually reviewed each of the predonation medical charts of over 2000 living kidney donors to ensure accuracy of the information in the Trillium database to identify living kidney donors. The Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD), Same Day Surgery (SDS) and National Ambulatory Care Reporting System (NACRS) databases have demographic, diagnostic and procedural information for all inpatient, outpatient and emergency department visits. The Ontario Health Insurance Plan (OHIP) database contains information on inpatient, outpatient and laboratory services based on billing claims from Ontario physicians. We used OHIP diagnostic codes to identify baseline conditions and procedural codes to identify surgical interventions for kidney stones. The Registered Persons Database (RPDB) contains demographic information on Ontario residents including their sex, date of birth, postal code and vital status. These databases have been used extensively to research health outcomes and health services including outcomes of living kidney donors. These databases were complete for variables used in this study.
Donors. We included all living kidney donors who were permanent residents of Ontario and donated between July 1, 1992 and March 31, 2009 at any of the five major transplant centers in Ontario. The date of nephrectomy served as the start date for follow-up and was designated the index date.
Healthy Nondonors. Choosing the best type of nondonors to whom donors can be compared is central to any study of relative risks associated with nephrectomy. Donors undergo a detailed selection process and are inherently healthier than the general adult population. We used techniques of restriction and matching to identify the healthiest segment of the general population allowing us to create our nondonor cohort. We randomly assigned an index date to the entire adult general population according to the distribution of index dates in donors. We then identified comorbidities and measures of access to healthcare from the beginning of available records (July 1, 1991) to the index date. This provided an average of 11 years of medical records for baseline assessment, with 99% of individuals having at least 2 years of baseline data for review. Among the general population we excluded any adult with any medical condition before the index date that could preclude donation. This included diagnostic, procedural or hospital visit codes for any genitourinary disease, diabetes, hypertension, cancer, cardiovascular disease, pulmonary disease, liver disease, rheumatologic conditions or chronic infections. We excluded anyone with evidence of nephrectomy, renal biopsy or nephrology consultation. We also excluded individuals with evidence of either frequent physician visits (more than four visits in the previous 2 years) or infrequent physician visits (less than one visit in the previous 2 years), given that Ontario has a shortage of physicians and to ensure that nondonors had evidence of access to routine healthcare). From a total of 17 792 616 adult Ontarians during the period of interest, our selection process resulted in the exclusion of 92% of adults, leaving 1 434 439 individuals available for matching.
Matching. Historically, a history of kidney stones (symptomatic or seen on imaging) precluded an individual from becoming a living kidney donor. More recently, centers have accepted individuals with small unilateral stones as living kidney donors, and data are needed to guide this practice. However, in this study, before matching we excluded 13 donors and 4420 healthy nondonors who had evidence of a kidney stone before their index date. We did this to ensure we assessed de novo kidney stones in follow-up. Also there were too few donors (n = 13) to meaningfully look at outcomes for those with a predonation history of kidney stones. From the remaining adults in the general population we matched 10 nondonors to each donor. We matched on age (within 2 years), sex, index date (within 6 months), rural (population <10 000) or urban residence and income (five categories representing average neighborhood income on the index date).
The primary outcome was evidence of a kidney stone with surgical intervention (i.e. shockwave lithotripsy, ureteroscopy or percutaneous nephrolithotomy; see Appendix B for codes used to define this outcome; these codes are listed in claims that result in surgeon reimbursement, with expected high sensitivity and positive predictive value as shown with other service payment codes). The secondary outcome was a hospital encounter with kidney stone (emergency room or hospital admission as recorded in our data sources, Appendix B; similar codes have a high positive predictive value [95.9%] but a low expected sensitivity underestimating the true number of events).
All participants were followed up from index date until: (i) death, (ii) emigration from the province or (iii) the end of study period (March 31, 2012). Of the individuals who reached the end of the study, those whose most recent healthcare encounter was more than 3 years before the end of study were classified as having emigrated from the province. These individuals were censored at 1 year following their last healthcare encounter.
We assessed differences in baseline characteristics between donors and matched nondonors using standardized differences. This metric describes differences between group means relative to the pooled standard deviation with differences >10% reflecting the potential for meaningful imbalance. We used a negative binomial model stratified on matched sets to estimate the rate ratio and 95% confidence interval (CI). This model also accounts for the possibility of a person having more than one stone event in follow-up (defined by events separated by at least 90 days). We repeated the primary analysis in three prespecified subgroups defined by age (≤40 vs. >40 at index date), sex and index date (1992–2001 [median follow-up 13.3 years, interquartile range (IQR) 11.4–15.8] vs. 2002–2009 [median follow-up 5.9 years, IQR 4.3–7.8]). We examined whether subgroup-specific rate ratios differed among subgroups using a series of pair-wise standard z-tests. We repeated the primary analysis using Cox proportional-hazards regression stratified on matched sets to examine the first stone event in follow-up for both the primary and secondary outcomes. We examined the characteristics associated with stone events separately in donors and nondonors using negative binomial regression models. All analyses were performed at ICES with SAS software version 9.2 (SAS Institute Inc., Cary, NC).
Methods
Design and Setting
We conducted a population-based matched retrospective cohort study using Ontario's healthcare databases held at the Institute for Clinical Evaluative Sciences (ICES). The province of Ontario, Canada, currently has approximately 13 million residents who have universal access to hospital care and physician services. We conducted this study based on a prespecified protocol that was approved by the research ethics board at the Sunnybrook Health Sciences Centre (Toronto, Ontario, Canada). The reporting of this study follows guidelines set out for observational studies (Appendix A).
Data Sources
We ascertained baseline characteristics, covariate information and outcome data from the records in six databases. Trillium Gift of Life Network is Ontario's central organ and tissue donation agency, and collects information on living kidney donors in the province at the time of kidney donation. We used the Trillium database to identify all adult living kidney donors who had donated between 1992 and 2009 at one of the five major transplant centers in Ontario. We then manually reviewed each of the predonation medical charts of over 2000 living kidney donors to ensure accuracy of the information in the Trillium database to identify living kidney donors. The Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD), Same Day Surgery (SDS) and National Ambulatory Care Reporting System (NACRS) databases have demographic, diagnostic and procedural information for all inpatient, outpatient and emergency department visits. The Ontario Health Insurance Plan (OHIP) database contains information on inpatient, outpatient and laboratory services based on billing claims from Ontario physicians. We used OHIP diagnostic codes to identify baseline conditions and procedural codes to identify surgical interventions for kidney stones. The Registered Persons Database (RPDB) contains demographic information on Ontario residents including their sex, date of birth, postal code and vital status. These databases have been used extensively to research health outcomes and health services including outcomes of living kidney donors. These databases were complete for variables used in this study.
Population
Donors. We included all living kidney donors who were permanent residents of Ontario and donated between July 1, 1992 and March 31, 2009 at any of the five major transplant centers in Ontario. The date of nephrectomy served as the start date for follow-up and was designated the index date.
Healthy Nondonors. Choosing the best type of nondonors to whom donors can be compared is central to any study of relative risks associated with nephrectomy. Donors undergo a detailed selection process and are inherently healthier than the general adult population. We used techniques of restriction and matching to identify the healthiest segment of the general population allowing us to create our nondonor cohort. We randomly assigned an index date to the entire adult general population according to the distribution of index dates in donors. We then identified comorbidities and measures of access to healthcare from the beginning of available records (July 1, 1991) to the index date. This provided an average of 11 years of medical records for baseline assessment, with 99% of individuals having at least 2 years of baseline data for review. Among the general population we excluded any adult with any medical condition before the index date that could preclude donation. This included diagnostic, procedural or hospital visit codes for any genitourinary disease, diabetes, hypertension, cancer, cardiovascular disease, pulmonary disease, liver disease, rheumatologic conditions or chronic infections. We excluded anyone with evidence of nephrectomy, renal biopsy or nephrology consultation. We also excluded individuals with evidence of either frequent physician visits (more than four visits in the previous 2 years) or infrequent physician visits (less than one visit in the previous 2 years), given that Ontario has a shortage of physicians and to ensure that nondonors had evidence of access to routine healthcare). From a total of 17 792 616 adult Ontarians during the period of interest, our selection process resulted in the exclusion of 92% of adults, leaving 1 434 439 individuals available for matching.
Matching. Historically, a history of kidney stones (symptomatic or seen on imaging) precluded an individual from becoming a living kidney donor. More recently, centers have accepted individuals with small unilateral stones as living kidney donors, and data are needed to guide this practice. However, in this study, before matching we excluded 13 donors and 4420 healthy nondonors who had evidence of a kidney stone before their index date. We did this to ensure we assessed de novo kidney stones in follow-up. Also there were too few donors (n = 13) to meaningfully look at outcomes for those with a predonation history of kidney stones. From the remaining adults in the general population we matched 10 nondonors to each donor. We matched on age (within 2 years), sex, index date (within 6 months), rural (population <10 000) or urban residence and income (five categories representing average neighborhood income on the index date).
Outcomes
The primary outcome was evidence of a kidney stone with surgical intervention (i.e. shockwave lithotripsy, ureteroscopy or percutaneous nephrolithotomy; see Appendix B for codes used to define this outcome; these codes are listed in claims that result in surgeon reimbursement, with expected high sensitivity and positive predictive value as shown with other service payment codes). The secondary outcome was a hospital encounter with kidney stone (emergency room or hospital admission as recorded in our data sources, Appendix B; similar codes have a high positive predictive value [95.9%] but a low expected sensitivity underestimating the true number of events).
All participants were followed up from index date until: (i) death, (ii) emigration from the province or (iii) the end of study period (March 31, 2012). Of the individuals who reached the end of the study, those whose most recent healthcare encounter was more than 3 years before the end of study were classified as having emigrated from the province. These individuals were censored at 1 year following their last healthcare encounter.
Statistical Analysis
We assessed differences in baseline characteristics between donors and matched nondonors using standardized differences. This metric describes differences between group means relative to the pooled standard deviation with differences >10% reflecting the potential for meaningful imbalance. We used a negative binomial model stratified on matched sets to estimate the rate ratio and 95% confidence interval (CI). This model also accounts for the possibility of a person having more than one stone event in follow-up (defined by events separated by at least 90 days). We repeated the primary analysis in three prespecified subgroups defined by age (≤40 vs. >40 at index date), sex and index date (1992–2001 [median follow-up 13.3 years, interquartile range (IQR) 11.4–15.8] vs. 2002–2009 [median follow-up 5.9 years, IQR 4.3–7.8]). We examined whether subgroup-specific rate ratios differed among subgroups using a series of pair-wise standard z-tests. We repeated the primary analysis using Cox proportional-hazards regression stratified on matched sets to examine the first stone event in follow-up for both the primary and secondary outcomes. We examined the characteristics associated with stone events separately in donors and nondonors using negative binomial regression models. All analyses were performed at ICES with SAS software version 9.2 (SAS Institute Inc., Cary, NC).
Source...