Fifty-nine patients (43 men, 16 women) with a mean age of 55 ± 13 years (min 22 years, max 76 years) completed the protocol. Thirty-four patients received a kidney graft from a living donor, and 25 from a deceased donor. On average, time since transplantation was 7 years (± 6yrs 10 months; min 3 months, max 29 years), and patients had been on dialysis 3 years and 10 months (± 3yrs 9 months; min 0 months, max 14 years) before transplantation. All of the patients had been on a calcineurin inhibitor (95% tacrolimus, 5% cyclosporine) since transplantation. Mean Mini-Mental Status Examination was 28.93 (± 1.23). There were no missing data in the variables included in the analyses.

Twenty-six patients did not show any cognitive impairment, 9 were mildly impaired, 15 moderately impaired, and 9 severely impaired as defined by our criteria above (see Fig. 1).

Proportion of patients per degree of cognitive impairment

For descriptive values of socioeconomic and medical variables and associations with the degree of cognitive impairment see Tables 1 and 2.

In the sub-group of kidney transplant-recipients with any cognitive impairment (n = 33), there was a significant difference in deficit severity between verbal memory, attention, and executive functions (χ2(2) = 7.11, p = 0.029), see Fig. 2. From the three compared domains, cognitively impaired patients, on average, showed the lowest deficit in verbal memory, and the highest in executive functions (mean rank verbal memory deficit = 1.74, mean rank attention deficit = 1.97, mean rank executive function deficit = 2.29). However, neither the difference between deficit severity in verbal memory and deficit severity in attention (z = − 0.23, adjusted p = 1.00), nor the difference between deficit severity in attention and deficit severity in executive function (z = − 0.32, adjusted p = 0.589) was statistically significant. The difference between deficit severity in verbal memory and deficit severity in executive functions was significant by trend (z = − 0.55, adjusted p = 0.080, r = 0.10, small effect).

Number of patients with deficit per cognitive domain in patients with any cognitive impairment (n = 33)

As can be seen in Table 1, an ordinal logistic regression with degree of cognitive impairment (no, mild, moderate, severe) as outcome and type of transplantation as predictor, showed a by-trend association between the two (OR = 0.39; 95% CI [0.148; 1.025]; Wald χ2(1) = 3.65, p = 0.056). A binomial logistic regression using any cognitive impairment (yes/no) as outcome showed a significant effect: patients with a graft from a deceased donor were, on average, three times more likely to have any cognitive impairment than patients with a graft from a living donor (OR = 3.26, 95% CI [1.08,9.83], Wald χ2(1) = 4.39, p = 0.036; model parameters: χ2(1) = 4.65, p = 0.031; model explained 10% of variance (Nagelkerke R2) and correctly classified 63% of the cases). However, when controlled for the effect of age, the association just barely failed to reach statistical significance at the 0.05 level (OR = 3.03, 95% CI [0.99,9.32], Wald χ2(1) = 3.74, p = 0.053; model parameters: χ2(1) = 6.77, p = 0.034; model explained 15% of variance (Nagelkerke R2) and correctly classified 70% of the cases). In a next step, we also included time since transplantation, which was found to be significantly associated with degree of cognitive impairment in our prior analysis (see Table 1). This model explained 16% of variance (Nagelkerke R2) and correctly classified 68% of the cases. Again, type of transplant was only associated with cognitive impairment (yes/no) by trend in this model (OR = 2.86, 95% CI [0.92,8.90], Wald χ2(1) = 3.30, p = 0.069; model parameters: χ2(1) = 7.26, p = 0.064).

There was a significant association between the type of transplantation and time on dialysis: Patients who received a graft from a living donor had spent less time on dialysis than patients who received a graft from a deceased donor (OR = 1.04, 95% CI [1.02,1.06]; model parameters: χ2(1) = 22.57, p < 0.001; model explained 43% of variance (Nagelkerke R2) and correctly classified 76% of the cases). However, although the cognitively unimpaired group, on average, spent less time on dialysis (no impairment: M = 40.61(± 42.38) months, impairment: M = 50.50(± 48.41) months), this effect was not statistically significant (OR = 1.00, 95% CI [0.99, 1.02]; model parameters: χ2(1) = 0.70, p = 0.404). Therefore, time on dialysis before transplantation does not explain the association between type of transplantation and cognitive impairment in our sample [24].

Similarly, the type of transplantation was significantly associated with creatinine levels: Patients who received a kidney graft from a living donor had lower creatinine levels than patients who received a kidney graft from a deceased donor (linear regression, F(1,57) = 5.10, β = 0.29, p = 0.028, R2 = 0.80). There was, however, no association between creatinine levels and cognitive impairment (logistic regression, OR = 0.91, 95% CI [0.94, 1.51]; model parameters: χ2(1) = 0.14, p = 0.709), which rules out current kidney function as a mediator between type of graft and cognitive impairment in our sample [24]. Type of transplantation did not predict creatinine clearance (F(1,45) = 1.64, β = -0.19, p = 0.207), nor did creatinine clearance predict cognitive impairment (OR = 1.01, 95% CI [0.99,1.03]; model parameters: χ2(1) = 0.70, p = 0.403).