Abstract
Background:
Solid organ transplant recipients have several risk factors for peri-operative multi-drug–resistant infection: their immune system is dampened as a result of critical illness and surgical stress that may be further impaired by induction immunotherapy and broad-spectrum antibiotic prophylaxis promotes selection for resistant pathogens. Infection with multi-drug–resistant organisms (MDRO) results in morbidity and mortality for solid organ transplant recipients.
Patients and Methods:
To assess in-hospital mortality and hospitalization duration associated with these infections, we analyzed cross-sectional, retrospective data from the 2016 Agency for Healthcare and Quality, Healthcare Cost and Utilization Project's National Inpatient Sample. Our analysis included 31,105 index admissions records for liver, kidney, heart, lung, and pancreas transplant recipients in the United States. Outcomes were assessed by multivariable regression analysis adjusting for covariables.
Results:
One percent (355/29,451) of patients with diagnosis of no MDRO infections died, 3% (40/1491) with diagnosis of one MDRO infection died, and 15% (25/166) with diagnosis of two MDRO infections died. Diagnosis of one MDRO infection was associated with a 20-day increase in hospitalization duration (95% confidence interval [CI], 17–22) but not increased odds of death (odds ratio [OR], 1.2; 95% CI, 0.5–2.5). Diagnosis of two MDRO infections was associated with an increased odds of death (OR, 9.6′ 95% CI, 3.3–27.9) and a 41-day increase in hospitalization duration (95% CI, 34–49).
Conclusions:
Strategies to decrease peri-operative MDRO infection may improve survival and decrease duration of hospitalization for solid organ transplant patients.
In the last half decade, more than 35,000 patients per year received solid organ transplants to treat end-stage organ disease in the United States and the number continues to increase [1–6]. Patients receiving solid organ transplant have several risk factors for infection with multi-drug–resistant organisms (MDRO): their immune system is dampened due to varying degrees of critical illness and surgical stress; induction immunosuppression further impairs their immune response to acute infection; peri-operative broad-spectrum antibiotic therapy promotes selection for multi-drug–resistant pathogens; and increased length of stay after transplantation increases exposure to pathogens. Infection with MDRO after solid organ transplantation has been associated with increased mortality, transplant organ complications, and prolonged hospitalizations [7–11].
Epidemiology of MDRO pathogens differ by both transplant center and organ transplanted. Based on estimates from multiple single-center studies, MDRO infections constitute 37%–51% of nosocomial pneumonia cases in lung transplant patients, 51% of infections acquired after heart transplantation, and >50% of infections by gram-negative rods after liver transplantation [7,12–17]. Of patients receiving a kidney transplant, 3%–14% acquired MDRO infections post-transplantation [8,18]. Pathogens commonly problematic for transplant recipients—including methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococci spp. (VRE), extended spectrum β-lactamase–producing (ESBL), and carbapenem-resistant gram-negative organisms, multi-drug–resistant Pseudomonas aeruginosa, and Clostridioides difficile—are among the MDRO identified in the 2019 U.S. Centers for Disease Control's report as urgent or serious threats requiring prompt and sustained action [19–23].
Because of their high susceptibility to MDRO infections, transplant patients frequently experience multiple MDRO infections during hospitalization [24]. However, previous studies of solid organ transplant recipients (liver, heart, lung, and kidney) failed to differentiate patients with a single MDRO infection from patients with multiple MDRO infections [7–11]. We sought to clarify the effects of multiple MDRO infections on solid organ recipients. We hypothesized that increasing numbers of MDRO infections would be associated with prolonged hospitalization and in-hospital mortality.
Patients and Methods
Data source
In this cross-sectional, retrospective study, we used data from the Agency for Healthcare and Quality, Healthcare Cost and Utilization Project's (HCUP) National Inpatient Sample (NIS) [25]. Data from 2016 were used because these were the most recent data set available at time of analysis. When analyzed, NIS is weighted to provide national estimates for inpatient utilization, charges, quality, and outcomes for more than 97% of the U.S. population. This sample is built from approximately 20% of all discharges from 4,573 U.S. community hospitals [26].
Variables and outcomes
Index admissions for any patients undergoing organ transplantation were identified by International Classification of Diseases, Tenth Revision Procedure Coding System (ICD-10-PCS) coding [27]. Transplants included were of the liver, kidney, heart, lung, and pancreas (Supplementary Table S1). Multi-drug–resistant organism infections were identified using ICD-10 Clinical Modification (ICD-10-CM) diagnosis codes. Multi-drug–resistant organisms were grouped into composite codes by pathogen for analysis: MRSA [A41.02, A49.02, B95.62, J15.212], VRE [A41.81], multi-drug–resistant gram-negative bacteria [A41.5, A41.59, A41.5, B96.1, J15.0, B96.5, J15.1], Clostridioides difficile infection [A04.7, A04.71, A04.72], and unspecified MDRO [A41.81]. Outcomes studied were death during hospitalization and duration of hospitalization.
Demographic variables assessed in the study were age, gender, race/ethnicity, income, and primary payer. A composite comorbidity score was calculated for each patient as defined by Elixhauser comorbidity.
Statistical methods
Univariable and multivariable analyses were performed to assess relations between variables and outcomes of death during hospitalization and duration of hospitalization. p values were adjusted using Bonferroni correction for eight comparisons. Non-normally distributed continuous variables were analyzed using Wilcoxon rank sum. Variables were analyzed using Wilcoxon rank sum, Fisher exact test, or χ2 as appropriate. Weighted data are reported as a mean with standard deviation (SD) or as counts with percentages. Threshold for significance was predetermined at Bonferroni-adjusted p < 0.05.
Multivariable regression was used to estimate effect of MDRO organism diagnoses on mortality and duration of hospitalization. MDRO diagnoses were modeled as either a binary variable (no infections or any diagnoses of infections) or as independent subgroups (no infections, one infection, or two infections). Effects of MDRO diagnoses were adjusted for age, sex, race/ethnicity, primary payer, income, Elixhauser comorbidity score, and organ transplanted. Residual deviance and Akaike information criterion (AIC) scores were used to compare models for MDRO diagnoses as well as a model excluding MDRO organism altogether. Odds ratios (OR) with 95% confidence intervals (95% CI) are reported. All analyses were performed using RStudio (version 1.1.453). Cells with 10 cases or fewer were reported as <11 in accordance with the HCUP user agreement. This study was classified exempt by the Stanford Institutional Review Board because of a lack of identifiable patient information.
Results
Baseline characteristics
We identified 31,105 admissions for solid organ transplant, of which 420 patients died (Table 1). Mean patient age was 50 years (SD, 16), 19,056 (62%) were male, and mean number of comorbidities was 2.7 (SD, 1.6). Transplanted organs were: kidney, 19,096 (61%); liver, 7,181 (23%); heart, 3,206 (10%); lung, 2,310 (7%); and pancreas, 884 (3%) pancreas. Of these, 1,570 (5%) were combined organ transplants of two organs. Diagnosis of a single MDRO infection was recorded in 1,491 (5%) of admissions, and diagnoses of two infections were recorded in 166 (<1%) of admissions. No patients in our study were diagnosed with three or more MDRO infections.
Baseline Characteristics
SD = standard deviation; MDRO = multi-drug–resistant organism; HCUP = Healthcare Cost and Utilization Project.
Bonferroni corrected for 8 comparisons.
Values <11 are reported as such in accordance with HCUP data use agreements.
Includes admissions for combined organ transplants.
Among variables studied, only the type of organ transplanted (Fisher exact, p < 0.001) and number of MDRO infection diagnoses (χ2, p < 0.001) differed between patients who died during hospitalization (n = 420) and patients who were discharged alive (n = 30,685; Table 1). There was no difference in age, gender, race/ethnicity, income quartile, comorbidity score, or primary payer. Patients who died were more likely to have received a liver (41% of dead vs. 23% of alive; χ2, p < 0.01) or heart transplant (39% of dead vs. 10% of alive; χ2, p < 0.001) and less likely to have received a kidney transplant (12% of dead vs. 62% of alive; χ2, p < 0.001). Of the 166 patients diagnosed with two MDRO infections, 15% died (n = 25) compared with 3% (40/1491) diagnosed with one MDRO infection and 1% (355/29,451) of patients with no MDRO infections (χ2, p < 0.001).
Multi-drug–resistant organism infection diagnoses and in-hospital mortality vary by organ transplanted
Number of infections diagnosed, and frequency of in-hospital death also varied by type of organ transplanted (Table 2). Liver, heart, and lung transplant patients were more frequently diagnosed with MDRO infections: at least one MDRO infection was diagnosed in 9% of liver recipients, 12% of heart recipients, and 21% of lung recipients (all χ2, p < 0.001). Among patients receiving these transplants, number of in-hospital deaths was also increased: death occurred in 2% of all liver recipients, 5% of heart recipients, and 2% of lung recipients (all χ2, p < 0.001). Kidney and pancreas recipients, in contrast, had lower rates of both infection and death during the index admission: 2% of kidney recipients were diagnosed with one or more MDRO infections (χ2, p < 0.001) and <1% of all recipients died (χ2, p < 0.001); 3% of pancreas recipients were diagnosed with MDRO infection (χ2, p = 0.06) and none died (Fisher exact, p < 0.001).
Infections by Organ Transplanted
MRSA = methicillin-resistant Staphylococcus aureus; CDI = Clostridioides difficile infection.
Values <11 are reported as such in accordance with HCUP data use agreements.
Outcomes associated with diagnosis of none, one, or two MDRO infections
We evaluated odds of death associated with any diagnosis of MDRO infection as a binary variable and did not find an association between diagnosis of MDRO infection in hospitalized solid organ transplant patients with increased odds of death (OR, 1.9; 95% CI, 0.9–3.5). Even after subdividing diagnosis of MDRO infection into subgroups by number of infections, diagnosis of one MDRO infection was not associated with increased odds of death (OR, 1.2; 95% CI, 0.5–2.5). However, diagnosis of two MDRO in solid organ transplant patients was independently associated with a 9.6 (95% CI, 3.3–27.9) increased odds of death (Table 3). These results were adjusted for demographics, transplanted organ, and number of comorbidities. In a comparison of residual deviances and AIC between three models—the first excluded MDRO infections as a predictor, the second included infections as binary predictor, and the third included number of infections as independent subgroups—modeling number of MDRO infections as independent subgroups was superior to the other two methods (residual deviances were 636, 633, and 624, respectively; Akaike information criterion [AIC] were 676, 675, and 668, respectively).
Mortality and Diagnosis of MDRO Infection
OR = odds ratio; CI = confidence interval; MDRO = multi-drug–resistant organism.
Bonferroni corrected for 8 comparisons.
OR adjusted for age, gender, race, income, primary payer, transplanted organ, Elixhauser comorbidities.
Diagnosis of MDRO infection—treating any diagnosis of MDRO as a binary variable—was associated with a 22-day (95% CI, 19–24 day) increased duration of hospitalization. When we evaluated diagnosis with one or more than one MDRO infections as independent subgroups, diagnosis of one MDRO infection was associated with a 19-day increase (95% CI, 17–22 day) in hospitalization duration compared with solid organ transplant patients without diagnosis of MDRO infection, and diagnosis with two MDRO infections was associated with a 41-day increase (95% CI, 34–49 day) in duration (Table 4). Compared with a model excluding MDRO infections (residual deviance 2475351, AIC 49887) and with a model including MDRO infections as a binary predictor (residual deviance 2345306, AIC 49588), evaluating single versus multiple MDRO infections as independent subgroups produced the best model (residual deviance 2331779, AIC 49558).
Prolonged Hospitalization and MDRO Infection
CI = confidence interval; MDRO = multi-drug–resistant organism.
β coefficient from multivariable linear regression.
Bonferroni corrected for 8 comparisons.
Adjusted for age, gender, race, income, primary payer, transplanted organ, Elixhauser comorbidities.
Discussion
We found that diagnosis of two MDRO infections in solid organ transplant patients was associated with increased in-hospital mortality and prolonged length of stay compared with diagnosis with one infection or no infections. Diagnosis of one MDRO infection was associated with prolonged duration of hospitalization, but not increased mortality in the peri-operative period. These results were not attributable to organ transplanted or to differences in comorbidities.
We found more modest effects of single MDRO infection than prior studies. A 2018 study demonstrated infection after liver transplantation led to 13.3 increase in odds of death (95% CI, 4.88–35.92) [9]. However, they had a much higher infection rate: 32% versus the 5% seen in our study. In a second single-institution longitudinal study of liver transplant patients, VRE infection was associated with 2.65 increased odds of death (95% CI, 1.53–4.58) and MRSA infection was not associated with death (OR, 1.00; 95% CI, 0.43–2.30). Although the effect of MDRO infection in this latter study was more modest, it remains difficult to compare their results with ours, because NIS is limited to data from index admissions and lacks outcomes from outpatient follow-up.
Several factors may explain differences in effect of MDRO infections on mortality between our work and those previously published. First, our study analyzes cross-sectional, retrospective data and may be less likely to capture MDRO diagnoses. As a result, misclassification bias in our analysis may have led to an underestimation of the true effects of MDRO infections in our patients. Second, as detection and treatment of MDRO infections has improved, outcomes in transplant patients may have improved in parallel [28–30]. Last, improvements in management of single type of MDRO infections may contribute to diverging trends in mortality for patients with one versus two infections.
Our finding that increased duration of hospitalization is associated with MDRO infection among solid organ recipients is consistent with previous reports. The two most common infections in our study were MRSA and Clostridioides difficile. Among all transplant patients, infection with Clostridioides difficile has been associated with prolonged hospitalization courses [19]. In a study of liver and kidney recipients, MRSA infection was associated with protracted stay in the intensive care unit but not overall hospitalization [31]. Our results show the relation between MDRO infections on duration of hospitalization cannot be fully captured by treating MDRO infection as a binary variable. Interestingly, the increase in length of stay associated with a second MDRO infection diagnosis is roughly double the first (41 vs. 22 days), suggesting an additive rather than synergistic effect.
There are limitations to this study. First, severity of critical illness was not able to be directly quantified given limitations of NIS. To address this, we adjusted effects using Elixhauser comorbidity scores as proxy. Second, incomplete or variable ICD-10 documentation between hospitals potentially introduced misclassification bias of MDRO infection. To minimize these errors, we generated organism composites for MDRO infections and included infection diagnoses without resistance modifiers. As a result, impact of MDRO infections on death and duration of hospitalization are likely to be underestimated by our results and could explain the lack of clear association between single infection diagnosis and increased odds of death. Third, multiple infections of the same category during an admission are not identified by the NIS data set. This limitation further contributes to potentially underestimating impact of MDRO infections on mortality and morbidity in our results. Fourth, susceptibility to MDRO infection of individuals in our study was unknown, and it may have been false to assume these patients are representative of the general transplant population. Fifth, our cross-sectional study lacked temporal data associating MDRO infection and outcomes, and we were unable to investigate causal relationships. It is possible that peri-operative complications related to transplantation led to prolonged hospitalizations and subsequently to increased diagnoses of MDRO infections in our data, not vice versa. Sixth, the coding system we used to identify presence of MDRO organism diagnosis could not differentiate community-acquired infections from hospital-acquired infections. Seventh, we lacked data about patients after discharge and were unable to assess outcomes from infections diagnosed in the outpatient setting. Last, we did not specify combined organ transplants in our analysis, which could result in an overestimate of mortality and duration of hospitalization for single organ transplant recipients.
Conclusions
Diagnosis of two MDRO infections is associated with increased odds of death and prolonged hospitalizations. These results suggest that strategies to decrease perioperative MDRO infection may improve survival and decrease duration of hospitalization for solid organ transplant patients.
Footnotes
Acknowledgments
The authors thank Dr. Kristin Sainani, Ph.D. for review of statistical methods.
Authors' Contributions
Siqi Cao contributed to study design, data analysis, and writing of this manuscript. Lakshika Tennakoon contributed to data collection, study design, and manuscript preparation. Aleah Brubaker contributed to manuscript preparation. Joseph Forrester contributed to the research question, study design, and manuscript preparation.
Data Availability
HCUP NIS data are available for purchase online through the Online HCUP Central Distributor.
Funding Information
No funding was received for this work.
Author Disclosure Statement
The authors of this manuscript have no conflicts of interest to disclose. This manuscript has been submitted solely to this journal and has not been published or submitted elsewhere.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
