repeating analyses after excluding donors with eGFR o80 ml/min per 1.73 m2 . DISCUSSION Our analysis demonstrated a significant increase in ESRD, cardiovascular, and all-cause mortality during long-term follow-up after living kidney donation compared with a selected population of non-donors who would have met the criteria for donation. Living kidney donation is associated with a known perioperative mortality risk3 and a small risk of major complications related to the surgical intervention,4 but the use of living donor kidneys has nevertheless become routine. Most published studies have used unselected general populations as the control group,4,10,15,16 skewing results in favor of the donor cohorts as controls with coexisting medical conditions such as cardiovascular disease, malignancy, diabetes, hypertension, and chronic kidney disease were included that would have made them ineligible for kidney donation. Clearly, living donors are very carefully selected and an appropriate comparison group should be selected in a similar manner. Lin et al.17 demonstrated that applying donor health Table 2a | Hazard ratio for death by any cause in kidney donors versus controls pressure. a Adjusted for age, gender, year of inclusion, systolic BP, smoking, and BMI. b After multiple imputation. 164 Kidney International (2014) 86, 162–167 clinical investigation G Mjen et al.: Risks for kidney donors criteria to an unselected population would yield a group with a lower mortality rate. Such a group would clearly be more relevant for mortality comparisons with living donors, but so far only three studies have been published with appropriate controls.3,13,14 Garg et al.13 used health administration data to identify a control group, excluding individuals with relevant diseases. During a follow-up period of 6 years, they found no survival difference between kidney donors and a selected control group. More recently, Garg et al.14 studied more complete data from follow-up of donors in a region of Canada, which confirmed the findings of their initial report. Segev et al.3 compared a cohort of kidney donors with a control group from the third National Health and Nutrition examination Survey study (NHANES III study). Controls were initially selected according to self-reported medical history to fit donor criteria and then a refined match for age, gender, and education. For the overall cohort of donors and controls, the median follow-up time was approximately 6 years. When assessing the influence of BP, body mass index (BMI), and smoking, a smaller cohort was followed for a relatively short time (median 2.1 years).3 Donation was not found to be associated with an increase in all-cause mortality. We also observed no increase in all-cause mortality during the initial 5–10 years after donation, but thereafter the survival curves began to deviate (Figure 2). In the general population, there is a robust association between reduced kidney function and mortality and premature vascular death. Several studies have reported that increased BP, albuminuria, and reduced renal function are associated with increased all-cause and cardiovascular mortality.7–9 These risk factors are more prevalent in kidney donors following nephrectomy.5,6 As a result, concerns have been expressed that kidney donation could lead to increased cardiovascular morbidity and mortality.13 Despite the inevitable reduction in renal function after nephrectomy, and evidence from the general population, it appears that the medical community has not regarded loss of renal function in living kidney donors as a long-term risk factor for mortality and cardiovascular disease. The incidence of ESRD in our study was comparable to that seen in some previous studies, although the incidence of ESRD in kidney donors is assumed to be lower than that in the general population.11,15,16,18,19 In our study, the donors had a substantially increased risk for developing ESRD compared with selected controls. The causes of ESRD were different in donors and controls. Seven out of the nine donors requiring renal replacement therapy had a primary renal disease. This was less common in controls. A likely explanation for the increased risk in donors may be linked to genetic factors, as the majority were immediate family members. It is well known that relatives of patients with chronic kidney disease have an increased incidence of renal disease.20 Accordingly, the increased incidence of ESRD in our cohort of kidney donors could be related to hereditary factors and not only nephrectomy. There are some limitations to our study. All controls lived within one county, whereas kidney donors were drawn from all over Norway. We had no data on renal function in the control group, although this information would be likely to increase the observed risks in donors. Longer follow-up time in controls may have influenced detection of ESRD cases. Due to missing data, we could not adjust for the level of education. However, we have previously shown that about one-third of Norwegian donors receive higher education.21 Only 10% of controls are educated to this level. Adjusting for level of education would therefore most likely increase the risk estimates for donors. However, we could not fully adjust for unmeasured confounders, for example, unknown comorbidities or differences in health-related