This study used data from the Scientific Registry of Transplant Recipients (SRTR). The SRTR data system includes data on all donors, wait-listed candidates, and transplant recipients in the US, submitted by the members of the Organ Procurement and Transplantation Network (OPTN). The Health Resources and Services Administration (HRSA), U.S. Department of Health and Human Services provide oversight to the activities of the OPTN and SRTR contractors.

Using the Scientific Registry of Transplant Recipients standard analysis file, we identified all dialysis-dependent primary deceased donor kidney recipients (N = 173,860) between 2005 and 2024. We excluded recipients discharged on maintenance immunosuppression other than tacrolimus and mycophenolate with or without steroids (n = 22,026). Moreover, we excluded recipients of no induction or mixed induction (n = 28,965) and those with a positive or unknown crossmatch (n = 19,191). Thus, our cohort consisted of 103,678 recipients and was divided into two groups of recipients based on the DGF status: the DGF group (n = 31,197) and the No-DGF group (n = 72,481). Delayed graft function (DGF) was defined according to the SRTR/OPTN registry criteria as the need for at least one dialysis treatment within the first week after kidney transplantation.

An additional mate-kidney analysis was performed among recipients with kidneys from the same donor (i.e., mate kidneys) (n = 13,636 kidneys from 6,818 donors). Pairs of kidneys from the same donor who met the above inclusion criteria and were identified where one recipient experienced DGF while the other did not (Figure 1).

The primary outcome of the study was death-censored kidney graft failure and overall graft failure by recipient’s DGF status, with overall graft failure defined as graft loss or death. Secondary outcomes included 6-month and 1-year rejection rate, length of stay, rehospitalization within the first year, BK polyoma virus infection rate, and 1 year creatinine according to recipient DGF status.

Continuous variables were summarized with the mean and standard deviation. Differences between groups were tested using analysis of variance or pooled t-tests. For categorical variables, data were summarized as counts and within-group percentages. Proportions were compared by the chi-square test. Time to event data were summarized with Kaplan-Meier estimates of incidence and curves for death-censored and overall graft survival through 10 years post-transplant. We left-censored the cohort at 90 days to limit the contribution of early complications, including primary non-function and death. Log-rank tests were used to identify statistically significant differences between groups.

Mixed effect Cox proportional hazard models examined the association between DGF and outcomes of interest. Models were adjusted for variables selected a priori based on medical relevance after review by the study team including donor (age, ethnicity, sex, DBD vs. DCD, KDRI), recipient (age, ethnicity, sex, ESKD etiology, dialysis vintage, presence of vascular disease, induction type and maintenance), immunologic (HLA mismatch, cPRA) and procurement factors (preservation). Transplant center was used as a random effect in these models. For the mate analysis, a random effect was also used for the donor. Linearity in all tests was evaluated using splines for continuous variables. Plots of Schoenfeld residuals were used to test the assumption of proportionality. We also performed sensitivity analyses on the full cohort, without left censoring, to assess the robustness of our findings.

To further evaluate the relationship between delayed graft function (DGF) and acute rejection, we performed additional analyses stratified by rejection status within the first-year post-transplant. Separate Cox proportional hazards models were constructed among recipients with and without rejection. In addition, an interaction term between DGF and rejection was included in the full cohort model to assess for effect modification. All models were adjusted using the same covariates as the primary analysis. All analyses were performed using R v4.2.2 (R Foundation for Statistical Computing, Vienna, Austria). P-values less than 0.05 were considered statistically significant.

Our cohort consisted of 103, 678 deceased donor kidney transplant recipients of whom 72,481 (69.90%) recipients did not experience DGF and 31,197 (30.09%) did. Recipients with DGF were older (54.8 vs. 52.7 years, p < 0.001), less likely to be women (31.8% vs. 41.5%, p < 0.001) and more likely to be Black (41.5% vs. 35.4%, p < 0.001). Diabetes mellitus as an etiology of end-stage kidney disease was more frequently observed in recipients with DGF (40.1% vs. 29.8%, p < 0.001), while glomerulonephritis (15.4% vs. 20.4%, p < 0.001) was less frequently observed compared to recipients without DGF. Recipients with DGF had longer waiting time on dialysis and had a higher KDRI (Table 1).

Recipients who experienced DGF were more likely to have received kidneys from older donors (42.0 vs. 38.1 years p < 0.001), donation after cardiac death (DCD) donors (DCD, 32.5% vs.16.8%, p < 0.001), and imported kidneys (37.3% vs. 28.6%, p < 0.001). These kidneys had longer cold ischemia times (19.9 vs. 17.4 h, p < 0.001), and their donors were less frequently supported with inotropic agents before donation (34.3% vs. 41.4%, p < 0.001). A higher number of HLA-mismatches was noted among the recipients with DGF. The induction type was similar between the two groups (Table 1).

The DGF rate increased over the study period. Before the KAS was implemented in 2014, the DGF rate averaged around 26.3%, while after KAS the DGF rate averaged around 30.6%. In March of 2021, KAS250 was implemented, and the rate of DGF has since averaged 35.0% (Figure 2). The increase in DGF rates over time was most pronounced following implementation of the KAS and KAS250 allocation policies, with a relative increase of approximately 16% from the pre-KAS to post-KAS250 era. This trend likely reflects evolving donor utilization practices, including increased use of donation after circulatory death (DCD) donors and longer ischemia times.

On univariable analysis, DGF was associated with worse short-term outcomes (Table 2). The length of stay was longer in the DGF group (8.2 days, vs. 5.7 days) compared to the recipients without DGF (p < 0.001). The rejection rate was also higher in DGF recipients at 6 months (6.0% vs. 4.7% p < 0.001). This difference remained through the 1-year mark (8.9% vs. 7.3% p < 0.001). BK infection rate within the first year was not different between groups (8.8% vs. 9.4%, p = 0.14) in recipients with and without DGF, respectively, although BK data was only available on 28,147 recipients (27.1%). The 1-year creatinine level was higher in the DGF group, at 1.58 mg/dL (±0.7) compared to 1.37 mg/dL (±0.6) (p < 0.001).

In the Kaplan-Meier analysis, DGF was associated with worse overall and death-censored graft survival (both log-rank p < 0.001) (Figure 3). Overall graft survival at 5 and 10 years in patients without DGF was 77.7% (CI 77.3, 78.0) and 48.7% (CI 48.1, 49.2) vs. 68.9% (CI 68.3, 69.6) and 37.4% (CI 36.5, 38.3) in those with DGF, respectively.

To further examine the impact of DGF, an adjusted analysis was performed. In the Cox proportional hazards model, DGF was associated with a 40% increased risk of death-censored graft failure [HR 1.40, CI (1.33,1.47), p < 0.001] and a 27% increased risk of overall graft loss by 10 years post-transplant [HR 1.27, CI (1.23,1.31), p < 0.001]. Older recipient age was associated with lower death-censored graft failure [HR 0.97, CI (0.97, 0.97), p < 0.001] but higher overall graft failure [HR 1.02, CI (1.02, 1.02), p < 0.001]. Death-censored and overall graft failure were slightly lower in recipients of Hispanic or other ethnicities compared to white recipients. Black recipients had an increased risk of death-censored graft failure [HR 1.42, CI (1.34, 1.51), p < 0.001], but not overall graft failure [HR 0.99, CI (0.95, 1.02), p = 0.42] compared to white recipients. Compared to diabetes as a cause of ESRD, other ESRD etiologies were associated with lower death-censored and overall graft failure (Table 3).

Compared to the pre-KAS era, the KAS era was associated with 15% decreased death-censored graft failure [HR 0.85, CI (0.80, 0.90), p < 0.001] and 4% increased risk of overall graft failure [HR 1.04, CI (1.00,1.08), p = 0.04]. The post-KAS 250 was not associated with a higher risk of death-censored graft failure [HR 0.95, CI (0.72, 1.25), p = 0.72] or overall graft failure [HR 1.04, CI (0.89–1.21), p = 0.64].

KDRI, DBD and donor age were significantly associated with a higher risk of death censored and overall graft failure, whereas vascular disease and dialysis time were only significantly associated with the latter. Compared to rabbit Anti-thymocyte globulin (r-ATG), neither alemtuzumab nor Interleukin 2 Receptor Antagonists (IL-2RA) was associated with increased death-censored graft failure. However, IL-2RA was associated with a 5% increased risk of overall graft failure [HR 1.05, CI (1.01, 1.10), p = 0.021]. Steroid use was associated with an increased risk of overall graft failure but not death-censored kidney graft failure. Compared to conventional single kidney transplants, en-bloc transplants were associated with lower death-censored and overall graft failure; dual kidney transplants were associated with lower overall graft failure but not death-censored graft failure (Table 3).

To further explore the relationship between delayed graft function (DGF) and acute rejection, we performed stratified analyses based on rejection status within the first-year post-transplant. DGF remained significantly associated with both death-censored and overall graft failure in recipients with and without rejection. No significant interaction between DGF and rejection status was observed, suggesting that DGF’s adverse effect on long-term outcomes is independent of rejection status (Table 4).

To control for unaccounted donor factors, we performed a mate-kidney analysis, and we found, in the unadjusted analysis, that those who experienced DGF, had lower death-censored kidney survival (log-rank p < 0.001) and overall kidney graft survival (log-rank p < 0.001), compared to those without DGF (Figure 4). In the mate-kidneys cohort with fully adjusted models (Table 5), delayed graft function (DGF) was associated with an 58% higher risk of death-censored graft failure (HR 1.58; 95% CI 1.39–1.80; p < 0.01) and a 35% higher risk of overall graft failure (HR 1.35; 95% CI 1.25–1.46; p < 0.01).

This study has several limitations inherent to its retrospective, registry-based design. Despite adjustment for a comprehensive set of donor, recipient, immunologic, and procurement variables, residual confounding from unmeasured or incompletely captured factors may persist, and causal inferences cannot be definitively established. Although the mate-kidney analysis strengthens causal inference by controlling for donor-related factors, it does not account for center-level practices or recipient-specific clinical decisions that may influence outcomes.

Delayed graft function (DGF) was defined using the standard SRTR/OPTN definition (dialysis within the first week post-transplant); however, this definition does not capture the severity or duration of DGF, as the registry records it as a dichotomous variable. As such, we were unable to evaluate potential dose-response relationships between DGF severity and long-term outcomes. In addition, variability in clinical practice, particularly differences in dialysis initiation thresholds across centers, may lead to misclassification of DGF and introduce bias.

The SRTR also lacks granular data on immunosuppression management, including the timing of calcineurin inhibitor initiation, dosing, and drug levels. Therefore, we could not assess whether delayed or subtherapeutic tacrolimus exposure contributed to the higher rejection rates or worse outcomes observed in recipients with DGF.

Despite these limitations, our study has several and notable strengths. It includes a large cohort of transplant recipients, long-term follow-up and comprehensive reporting of outcomes and risk factors. Most importantly, the subgroup analysis of mate kidneys allowed us to control for donor-related confounders. Finally, the large sample size provided enough statistical power to detect statistically significant differences.

DGF is a strong, independent, and durable predictor of inferior long-term kidney allograft outcomes. Importantly, the association persists even after controlling for all donor-related factors using mate-kidney analysis. As DGF continues to rise in the current allocation era, strategies focused on prevention, early identification, and enhanced post-transplant management are critical to improving long-term outcomes for deceased donor kidney recipients.