Prevention of Sepsis during the Transition to Dialysis May Improve the Survival of Transplant Failure Patients

Abstract

Dialysis patients are at risk for sepsis, and the risk may be even higher among transplant failure patients because of previous or ongoing immunosuppression. The incidence and the consequences of sepsis as defined by International Classification of Diseases, Ninth Revision, Clinical Modification hospital discharge diagnoses codes were determined among 5117 patients who initiated dialysis after transplant failure between 1995 and 2004 in the United States. The overall sepsis rate was 11.8 per 100 patient years (95% confidence interval [CI] 11.5 to 12.1). Sepsis was highest in the first 6 mo after transplant failure (35.6 per 100 patient years [95% CI 29.4 to 43.0] between 0 to 3 mo after transplant failure; 19.7 per 100 patient years [95% CI 17.2 to 22.5] between 3 to 6 mo after transplant failure). In comparison, the sepsis rate among incident dialysis patients between 3 and 6 mo after dialysis initiation was 7.8 per 100 patient years (95% CI 7.3 to 8.3), whereas the sepsis rate among transplant recipients between 3 and 6 mo after transplantation was 5.4 per 100 patient years (95% CI 4.9 to 5.9). Patients who were ≥60 yr, obese patients, patients with diabetes, and patients with a history or peripheral vascular disease or congestive heart failure were at risk for sepsis. Transplant nephrectomy was not associated with septicemia. The role of continued immunosuppression and vascular access creation was not assessed and should be addressed in future studies. In a multivariate analysis, patients who were hospitalized for sepsis had an increased risk for death (hazard ratio 2.93; 95% CI 2.64 to 3.24; P < 0.001). Strategies to prevent sepsis during the transition from transplantation to dialysis may improve the survival of patients with allograft failure.

Renal transplant failure is a leading cause of dialysis initiation (^1–5). Despite that 44% of renal transplant failure patients are listed for repeat transplantation, only 15% of transplant failure patients will ever receive another transplant (²). Patients who are treated with dialysis after transplant failure have a high rate of morbidity and mortality (^1–7), yet there is little information regarding factors that may increase the risk for morbidity and mortality in these patients.

Infection is the second leading cause of both morbidity and mortality, behind cardiovascular disease, among dialysis-treated patients (⁸). In fact, the annual mortality rate as a result of sepsis among dialysis patients is reported to be 30 to 45 times higher than that in the general population (⁹). Transplant failure patients may be at particularly high risk for infection because of previous or continued exposure to immunosuppressive medications and the presence of a failed renal allograft. In fact, Kaplan and Meier-Kriesche (⁶) found that transplant failure patients had a nearly four-fold higher rate of sepsis related mortality compared with patients with a functioning allograft (16.3 versus 3.7 per 100 patient years). In this study, we examined the incidence, timing, risk factors, and consequences of septicemia in patients who were treated with dialysis after renal allograft failure.

Materials and Methods

Data Source and Study Population

The data source for the study was the Standard Analysis Files of the US Renal Data System. The study population included patients (≥18 yr) who began their first ESRD treatment after April 1995 and returned to dialysis after failure of a first kidney-only transplant. The study population was further limited to patients with Medicare as the primary payer (66% of all renal transplant failure patients identified during the study period) to ensure complete ascertainment of hospitalized septicemia from institutional claims data.

Statistical Analyses

Medicare hospital claims were used to identify sepsis events on the basis of discharge diagnoses, using the following International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes: Septicemia 0.38.xx (x = 0 to 9 inclusive), Streptococcal (038.0x), Staphylococcal (038.1x), Pneumococcal (038.2), anaerobic (038.3), aerobic Gram-negative (038.4x), other specified septicemia (038.8), and unspecified septicemia (038.9). Patients were followed from the date of transplant failure until death, repeat transplant, or December 2004.

Descriptive statistics (χ² and t test) were used to compare patients who did and did not develop sepsis during the follow-up period. Poisson regression analysis was used to determine the rate of first septicemia events during follow-up and in 6-mo intervals after transplant failure. Poisson regression analysis was also used to test the association between baseline characteristics and first septicemia rates. For the purposes of comparison, first septicemia rates were also determined in (1) incident dialysis patients who were eligible for transplantation (wait-listed for transplantation before start of dialysis or activated to the waiting list during follow-up) and (2) transplant recipients during the study period (April 1995 through December 2004).

The time to the first septicemia episode was determined using the Kaplan-Meier method. Factors that were associated with septicemia were determined in a Cox multivariate regression analysis. Variables were entered into the model when they were associated with septicemia in the univariate Cox regression analyses (P < 0.20) and met the proportional hazards (PH) assumption. The PH assumption was tested using log-negative-log plots of the within-group survivorship probabilities versus log-time. The following variables were tested for inclusion: Age at transplant failure, gender, body mass index, race, diabetes, dialysis modality after transplant failure, duration of allograft survival, donor type, peripheral vascular disease (PVD), congestive heart failure (CHF) and ischemic heart disease, use of induction immunosuppression (classified as depleting when OKT3, thymoglobulin, antithymocyte globulin (ATG), and antithymocyte globulin (ATGAM) were used and nondepleting when basiliximab and daclizumab were used), and transplant nephrectomy (identified from ICD-9-CM codes and treated as a time-dependent covariate).

The independent association of septicemia with patient survival was determined in an extended Cox PH model. Factors that were considered for inclusion into the model were sepsis, having one or more sepsis episodes; age at transplant failure; gender; body mass index; race; diabetes; dialysis modality after transplant failure; duration of transplant function; donor type; use of induction immunosuppression; and history of PVD, CHF, ischemic heart disease, and cerebrovascular accident.

Results

A total of 5117 patients met the inclusion criteria. Patients were followed for a median of 2.3 yr (25th and 75th percentiles 1.3 to 3.8 yr).

Characteristics of the Study Population and Comparison of Patients with and without Septicemia

The characteristics of the study population are shown in Table 1. Septicemia was more likely in older, female, obese, and failed transplant patients who had diabetes and were treated with hemodialysis after transplant failure. Sepsis was also more common in transplant failure patients with a shorter duration of allograft function and patients who received a transplant or had graft failure in an earlier study year. Comorbidities that were associated with an increased likelihood of septicemia included CHF, PVD, and a subnormal serum albumin (<3.4 g/dl). The use of induction immunosuppression was lower in patients who developed sepsis.

Septicemia Frequency, Onset, and Rates

Among the study patients, 1177 (23%) had 1826 septicemia episodes. Of the 1177 patients who had a septicemia episode, 791 (67%) had one episode, 245 (21%) had two episodes, and 141 (12%) had three or more episodes. Among patients who developed sepsis, the median time from transplant failure to first hospitalized septicemia was 160 d (25th and 75th quartiles 53 and 447 d, respectively).

The causative organisms for septicemia episodes were varied. Staphylococcus was the causative organism in 34.7% of all episodes, Gram-negative organisms in 19.2%, Pneumococcus in 1%; anaerobic organisms in 0.5%, and other organisms in 6.6% of cases. The causative microbe was not specified in 38% of septicemia episodes.

The overall septicemia rate in patients with a failed transplant was 11.8 per 100 patient years (95% confidence interval [CI] 11.5 to 12.1). The rate of first septicemia was significantly higher in the first 6 mo after transplant failure (35.6 per 100 patient years [95% CI 29.4 to 43.0] in the first 3 mo after transplant failure; 19.7 per 100 patient years [95% CI 17.2 to 22.5] between 3 and 6 mo after transplant failure), compared with 6 to 12 mo (9.8 per 100 patient years; 95% CI 8.5 to 11.4), 12 to 24 mo (7.4 per 100 patient years; 95% CI 6.4 to 8.5), 24 to 36 mo (5.5 per 100 patient years; 95% CI 4.5 to 6.8), and >36 mo after transplant failure (4.9 per 100 patient years; 95% CI 3.9 to 6.1; Figure 1). The sepsis rate between 3 and 6 mo after transplant failure (19.7 per 100 patient years; 95% CI 17.2 to 22.5) was higher than the sepsis rate in new transplant recipients 3 to 6 mo after transplantation (5.4 per 100 patient years; 95% CI 4.9 to 5.9) and higher than the sepsis rate among incident dialysis patients 3 to 6 mo after dialysis initiation (7.8 per 100 patient years; 95% CI 7.3 to 8.3).

Patients who were ≥60 yr at the time of graft failure had a significantly higher rate of septicemia (16.1 per 100 patient years; 95% CI 15.5 to 16.7) compared with patients who were younger than 60 yr at the time of graft failure (11.2 per 100 patient years; 95% CI 11.0 to 11.4). Patients with diabetes had a higher rate of septicemia than patients without diabetes (20.0 per 100 patient years [95% CI 19.0 to 21.0] versus 9.2 per 100 patient years [95% CI 9.0 to 9.4]). White and black patients had more episodes of septicemia than patients of other ethnic origins (white 11.9 per 100 patient years [95% CI 11.5 to 12.3]; black 11.9 per 100 patient years [95% CI 11.3 to 12.5]; other 8.9 per 100 patient years [95% CI 7.6 to 10.2]). Septicemia rates were also higher in women (12.4 per 100 patient years; 95% CI 12.1 to 12.7) compared with men (11.3 per 100 patient years; 95% CI 11.1 to 11.5).

Factors Associated with Septicemia after Transplant Failure

Patients who were ≥60 yr, obese patients, patients with diabetes, patients with a history of PVD or CHF, and patients who were treated with hemodialysis after transplant failure were at increased risk for septicemia. Patients who received induction immunosuppression with nondepleting antibodies had a lower risk for sepsis (P = 0.06). There was no association between transplant nephrectomy and sepsis (Table 2).

Consequences of Septicemia and Patient Survival

The median length of hospitalization for each sepsis admission was 8 d (25th to 75th percentile 4 to 16 d). CHF and myocardial infarction accompanied sepsis in 17.7 and 3.2% of cases, respectively. During hospital admissions for sepsis, 13.2% of failed transplant patients died, and the cumulative mortality was 21.3% at 30 d from admission. The median time from a septicemia episode to death was 1.64 yr (25th and 75th percentiles 0.23 and 5.56 yr, respectively).

An extended Cox regression multivariate model showed the effect of first septicemia on patient survival (Table 3). Transplant failure patients who developed septicemia had an independently increased risk for death (hazard ratio 2.93; 95% CI 2.64 to 3.24). Multiple sepsis episodes were also associated with increased mortality (hazard ratio 1.47; 95% CI 1.26 to 1.73). The risk for mortality with septicemia was greater than that associated with patient age or diabetes.

Discussion

In this study, we found that transplant failure patients have a high rate of septicemia (23%). Sepsis rates were highest during the transition from dialysis to transplantation (first 6 mo after transplant failure) and were much higher during similar transition periods to new forms of renal replacement therapy in de novo dialysis patients and new transplant recipients. Patients who were ≥60 yr, obese patients, patients with diabetes, patients with PVD or CHF, and patients who were treated with hemodialysis after transplant failure had an increased risk for developing septicemia. Septicemia was also a strong independent risk factor for mortality. These findings suggest that efforts to eliminate sepsis, particularly during the early transition to dialysis, may improve the survival of transplant failure patients and should be further studied.

Although many reports have described rates of septicemia in the general dialysis population (^9,10), few have addressed the incidence of septicemia in the patients who have ESRD with a failed renal allograft. The annual mortality rate as a result of sepsis is reported to be 100- to 300-fold higher in general dialysis patients compared with the general population (⁹). Among incident dialysis patients, Foley et al. (¹⁰) found that the admission rate for septicemia rose by 51% between 1991 and 1999 (from 11.6 to 17.5 per 100 patient years, respectively). We found that the sepsis rate among transplant failure patients dramatically exceeds these rates during the first 6 mo of dialysis (35.6 per 100 patient years from 1 to 3 mo and 19.9 per 100 patient years in 3 to 6 mo after transplant failure). These findings are particularly startling given that, in general, transplant patients are younger than general incident dialysis patients and have fewer comorbid conditions (¹¹). After the initial 12 mo of dialysis treatment, failed renal transplant patients had a sepsis rate that was similar to that in general incident dialysis patients who were eligible for transplantation.

The elevated early septicemia rate that was shown in our study may be due to several factors. Exposure to immunosuppressive medications increases the risk for infection as well as other morbidities, including cardiovascular disease and malignancy (^6,12–15). We previously showed that antibody induction at the time of transplantation was not associated with increased mortality after transplant failure (³). Consistent with our previous work, we did not identify an increased risk for mortality among transplant failure patients who were treated with induction immunosuppression. Indeed, we found that the use of nondepleting antibodies was associated with a decreased risk for death after transplant failure (Table 3). We were not able to determine the type of immunosuppression that was maintained after transplant failure. Judicious tapering of immunosuppression may be an important strategy to decrease septicemia in transplant failure patients. Similarly, aggressive treatment of acute rejection in patients who sustained early allograft failure may be an important cause of sepsis. Our analysis did not include specific treatment for acute rejection as a potential factor contributing to the high sepsis rates early after transplant failure. We found no association between transplant nephrectomy and sepsis. However, we cannot draw definitive conclusions regarding the role of transplant nephrectomy from our observational study. Further studies regarding the routine use of transplant nephrectomy are clearly needed.

The type of vascular access may have been an important factor that contributed to the high rate of septicemia during the first 6 mo after transplant failure and may explain the increased rate of septicemia in hemodialysis versus peritoneal dialysis patients (^10,16,17). Sepsis rates are known to be higher in patients who initiate dialysis with temporary hemodialysis access (^8,9,18–21). We were not able to determine the type of vascular access in all patients who initiated hemodialysis after transplant failure. Adherence to published recommendations for vascular access may also help to decrease septicemia in transplant failure patients (^22–24).

A third potential explanation for the initially elevated sepsis rates in the failed renal transplant patients who were on dialysis may be related to the predialysis care of the failing transplant patients (^1,4). It has been shown that patients with failed kidney transplants start dialysis at levels of hematocrit, serum albumin, and GFR that may be suboptimal (¹). Whether more aggressive chronic kidney disease management during the period of allograft function can improve outcomes after transplant failure should be studied further. Finally, fragmentation of patient care during the transition from transplantation to dialysis may be an important contributing issue. It is tempting to speculate that strategies such as initiating dialysis in the transplant center, before transferring care to peripheral centers, may improve outcomes (^25,26).

The strong association of septicemia with death in transplant failure patients is consistent with reported observations among general dialysis patients (¹⁰). The mechanisms by which septicemia contributed to mortality cannot be determined from our study. We found that 13% of sepsis hospitalizations resulted in death and increased to 21% when deaths within 30 d of admission for septicemia were included. It is possible that, in some cases, patients were admitted to the hospital for another reason that was complicated by septicemia and death. Alternatively, the presence of immunosuppression may have complicated septicemia, resulting in high mortality rates. We also found that sepsis had long-term consequences for patient survival (median time to death after septicemia 1.64 yr). This delayed association of sepsis with mortality has also been reported in the general dialysis population, where mortality rates in septic patients remained more than twice that in the nonseptic patients up to 5 yr after the sepsis episode (¹⁰).

Readers of our study should consider the inherent limitations of this retrospective analysis of administrative data. The study was limited to patients who had Medicare as the primary payer, which may limit the applicability of our findings to other patient populations. In addition, the relatively short duration of graft survival in our study patients may limit the ability to extend our finding to patients who return to dialysis after longer durations of graft survival. Comorbid conditions were determined at the time of initiation of the first ESRD treatment from the CMS Medical Evidence Report Form (form 2728), which may underreport comorbid conditions (²⁷). Finally, we were not able to validate the ICD-9-CM codes for septicemia. Our study provides only limited information regarding other conditions that may have precipitated hospital admissions that were coded as sepsis. For example, a patient who was admitted for myocardial infarction may have subsequently developed sepsis in the hospital and died. It is possible that the high rates of septicemia that were identified from ICD-9-CM codes could be related to an ascertainment bias. For example, sepsis may be the default diagnosis in an unwell patient with no clear diagnosis.

Conclusion

Transplant failure patients are at very high risk for septicemia, particularly during the first 6 mo of dialysis treatment, and septicemia is a strong risk factor for death. Efforts to reduce sepsis, particularly during the transition to dialysis after transplant failure, may improve survival in these high-risk patients with ESRD and should be further studied.

Disclosures

None.

Figure 1: Comparison of first septicemia rates in failed transplant patients, incident dialysis patients who were eligible for transplantation, and transplant recipients. The x axis denotes time in months after (1) return to dialysis in transplant failure patients, (2) dialysis initiation in incident dialysis patients, and (3) transplantation in transplant recipients. For ensuring complete capture of septicemia events from Medicare claims, septicemia events in the 0- to 3-mo period were determined only for a subset of transplant failure patients who had graft loss within 3 yr of transplantation and maintained ESRD Medicare coverage.

Table 1: Characteristics of the study population and a comparison of patients with and without sepsisa

Table 2: Factors associated with sepsis after transplant failurea

Table 3: Factors associated with death after transplant failure

O. Johnston is supported by a grant from the Canadian Institute of Health Research and Michael Smith Foundation for Health Research. J.S. Gill is supported by the Michael Smith Foundation for Health Research.

This study was presented in part at the World Transplant Congress; July 23, 2006; Boston, MA.