Sepsis mortality among patients with haematological malignancy admitted to intensive care 2000-2022: a binational cohort study.

BACKGROUND: Sepsis occurs in 12-27% of patients with haematological malignancy within a year of diagnosis. Sepsis mortality has improved in non-cancer patients in the last two decades, but longitudinal trends in patients with haematological malignancy are not well characterised. We aimed to compare outcomes, including temporal changes, in patients with and without a haematological malignancy admitted to ICU with a primary diagnosis of sepsis in Australia and New Zealand over the past two decades. METHODS: We performed a retrospective cohort study of 282,627 patients with a primary intensive care unit (ICU) admission diagnosis of sepsis including 17,313 patients with haematological malignancy, admitted to 216 intensive care units (ICUs) in Australia or New Zealand between January 2000 and December 2022. Annual crude and adjusted in-hospital mortality were reported. Risk factors for in-hospital mortality were determined using a mixed methods logistic regression model and were used to calculate annual changes in mortality. RESULTS: In-hospital sepsis mortality decreased in patients with haematological malignancy, from 55.6% (95% CI 46.5-64.6%) in 2000 to 23.1% (95% CI 20.8-25.5%) in 2021. In patients without haematological malignancy mortality decreased from 33.1% (95% CI 31.3-35.1%) to 14.4% (95% CI 13.8-14.8%). This decrease remained significant after adjusting for mortality predictors including age, SOFA score and comorbidities, as estimated by adjusted annual odds of in-hospital death. The reduction in odds of death was of greater magnitude in patients with haematological malignancy than those without (OR 0.954, 95% CI 0.947-0.961 vs. OR 0.968, 95% CI 0.966-0.971, p < 0.001). However, absolute risk of in-hospital mortality remained higher in patients with haematological malignancy. Older age, higher SOFA score, presence of comorbidities, and mechanical ventilation were associated with increased mortality. Leukopenia (white cell count < 1.0 × 109 cells/L) was not associated with increased mortality in patients with haematological malignancy (p = 0.60). CONCLUSIONS: Sepsis mortality has improved in patients with haematological malignancy admitted to ICU. However, mortality remains higher in patients with haematological malignancy than those without.

Immunoglobulin replacement vs prophylactic antibiotics for hypogammaglobulinemia secondary to hematological malignancy.

ABSTRACT: Immunoglobulin replacement and prophylactic antibiotics are commonly used to prevent infections in patients with secondary hypogammaglobulinemia due to hematological malignancies but have never been directly compared. In this randomized controlled feasibility trial conducted in 7 hospitals in Australia and New Zealand, we enrolled patients with secondary hypogammaglobulinemia with either a history of recurrent/severe infection or an immunoglobulin G level <4 g/L. Participants were randomized in a 1:2 ratio to immunoglobulin (0.4 g/kg per 4 weeks IV) or daily antibiotics (trimethoprim-sulfamethoxazole 160 mg/800 mg or, if contraindicated, 100 mg doxycycline) for 12 months. Participants allocated to antibiotics were allowed to crossover after grade ≥3 infections. The primary outcome was proportion of patients alive on the assigned treatment 12 months after randomization. Between August 2017 and April 2019, 63 patients were randomized: 42 to antibiotics and 21 to immunoglobulin. Proportion of participants alive on allocated treatment at 12 months was 76% in the immunoglobulin and 71% in the antibiotic arm (Fisher exact test P=.77; odds ratio, 0.78; 95% CI, 0.22-2.52). The lower quartile for time to first major infection (median, not reached) was 11.1 months for the immunoglobulin and 9.7 months for the antibiotic arm (log-rank test, P=.65). Three participants in the immunoglobulin and 2 in the antibiotic arm had grade ≥3 treatment-related adverse events. A similar proportion of participants remained on antibiotic prophylaxis at 12 months to those on immunoglobulin, with similar rates of major infections. Our findings support the feasibility of progressing to a phase 3 trial. Trial registration #ACTRN12616001723471.

Economic evaluation: immunoglobulin vs prophylactic antibiotics in hypogammaglobulinemia and hematological malignancies.

ABSTRACT: Patients with hematological malignancies are at high risk of developing hypogammaglobulinemia (HGG) and infections. Immunoglobulin (Ig) is one recommended option to prevent these infections, but it is expensive, and its cost-effectiveness compared with other prevention strategies remains unknown. We conducted a trial-based economic evaluation from the Australian health care system perspective to estimate the 12-month cost-effectiveness of prophylactic Ig vs prophylactic antibiotics in 63 adults with HGG and hematological malignancies participating in the RATIONAL feasibility trial. Two analyses were conducted: (1) cost-utility analysis to assess the incremental cost per quality-adjusted life year (QALY) gained; and (2) cost-effectiveness analysis to assess the incremental cost per serious infection prevented (grade ≥3) and per any infection (any grade) prevented. Over 12 months, the total cost per patient was significantly higher in the Ig group than in the antibiotic group (mean difference, AU$29 140; P < .001). Most patients received IVIg, which was the main cost driver; only 2 patients in the intervention arm received subcutaneous Ig. There were nonsignificant differences in health outcomes. Results showed Ig was more costly than antibiotics and associated with fewer QALYs. The incremental cost-effectiveness ratio of Ig vs antibiotics was AU$111 262 per serious infection prevented, but Ig was more costly and associated with more infections when all infections were included. On average and for this patient population, Ig prophylaxis may not be cost-effective compared with prophylactic antibiotics. Further research is needed to confirm these findings in a larger population and considering longer-term outcomes. The trial was registered at the Australian and New Zealand Clinical Trials Registry as #ACTRN12616001723471.

Survey of treatment practices for immunocompromised patients with COVID-19 in Australasia.

Coronavirus disease 2019 (COVID-19) in immunocompromised patients can lead to severe and prolonged illness. Data are limited with regard to management of COVID-19 in this setting, particularly in persistent or recrudescent infection. The authors conducted an online survey among infectious diseases doctors to determine current approaches to treatment across Australasia. There was marked variability in responses relating to the diagnostic modalities and use of antiviral agents in patients with immunocompromise, highlighting the need for high-quality studies to guide treatment decisions in this group.

High risk of infection in 'real-world' patients receiving ibrutinib, idelalisib or venetoclax for mature B-cell leukaemia/lymphoma.

OBJECTIVE: The infection risk in patients receiving ibrutinib, idelalisib or venetoclax for chronic lymphocytic leukaemia (CLL) or B-cell lymphoma treated outside of clinical trials is incompletely defined. We sought to identify the severe infection rate and associated risk factors in a 'real-world' cohort. METHODS: We conducted a retrospective cohort study of adult patients with CLL or lymphoma treated with ibrutinib, idelalisib or venetoclax. RESULTS: Of 67 patients identified (ibrutinib n = 53, idelalisib n = 8 and venetoclax n = 6), 32 (48%) experienced severe infection. Severe infection occurred at a rate of 65 infections per 100 person-years, with a median of 17.8 months of therapy. Median time to first infection (IQR) was 5.4 months (1.4-15.9). Poor baseline Eastern Cooperative Oncology Group (ECOG) performance status and high Charlson Comorbidity Index (CCI) score associated with increased risk of severe infection [hazard ratios (95% CI) 1.57 (1.07-2.31, p = .018) and 1.3 (1.05-1.62, p = .016) respectively]. CONCLUSION: The severe infection rate for patients receiving ibrutinib, idelalisib or venetoclax for lymphoma and CLL exceeded those reported in clinical trials. Patients with poor ECOG or high CCI should be closely monitored for early signs of infection and prevention strategies actively pursued. Further prospective research is required to define optimal antimicrobial prophylaxis recommendations.

Missed Opportunities? A Retrospective Study Into Adults Hospitalized With Invasive Infection From Airway Pathogens.

Background: Invasive disease caused by airway pathogens, including Streptococcus pneumoniae, Haemophilus influenzae, Neisseria meningitidis, and Moraxella catarrhalis, has high morbidity and mortality worldwide, with immunodeficiency being a known association with recurrent disease. The study aimed to describe the frequency of known immunodeficiency and predisposing factors in adult patients presenting with invasive infections and determine the frequency of screening for and detection of immunodeficiency. Methods: A retrospective analysis was conducted at a large tertiary Australian health service, comprising multiple centers. Patients aged 18 years or older, in whom the above pathogens were isolated from sterile sites, were included as identified through a microbiology database, between 2015 and 2020. Using electronic medical records, patient demographics, medical history, outcomes of admission, and pathology results were captured and reviewed to address the aims. Results: In 252 patients, S pneumoniae was the most common culprit, isolated in 73% (185/252), compared to 14.3% (36/252) and 11.5% (29/252) of infections caused by H influenzae and N meningitidis, respectively. Known diagnoses of secondary immunodeficiency were common (31% of patients). Of those presenting with invasive pneumococcal disease, 78% had at least 1 predisposing condition, though only 9 patients (6%) had previously received pneumococcal vaccination. Despite poor screening for immunodeficiency, 12 new diagnoses were made. While the commonest immunodeficiency was secondary, due to hematological and solid organ malignancies, 3 new primary immunodeficiency diagnoses were made. Conclusions: Immunodeficiency is common in this patient population. Screening should be undertaken to ensure timely diagnosis and treatment of the underlying condition to avoid future morbidity and mortality.

Kidney transplant recipients' attitudes toward COVID-19 vaccination and barriers and enablers to vaccine acceptance.

OBJECTIVE: To identify barriers and enablers to COVID-19 vaccination in renal transplant recipients who are undecided about vaccination. METHODS: An online survey was distributed to 876 adult kidney transplant recipients at a tertiary referral service, who had not been vaccinated against COVID-19. The survey assessed willingness to be vaccinated, attitudes toward COVID-19 vaccines, and barriers and enablers to proceeding with vaccination. RESULTS: The survey response rate was 54% (473/876). Three hundred and forty-six (73.1%) participants planned to receive vaccination (yes group), 105 (22.2%) were undecided, and 22 (4.7%) refused vaccination. The undecided group were younger but were not different in other demographic characteristics to the yes group. The undecided group were less positive toward (34.29% vs. 91.3%, p < .001) and more concerned about (93.3% vs. 25.1%, p < .001) vaccination than the yes group. Their concerns related to vaccine safety (including harm to their transplant), poor efficacy, and a lack of rigorous testing in transplant recipients. Undecided recipients had received less vaccine-specific information from medical specialists than the yes group. Most undecided participants (95.1%) were willing to proceed with vaccination with appropriate supports. The most desired supports were information and a recommendation to proceed with vaccination from their treating transplant specialist and team. CONCLUSION(S): Concerns about vaccine safety (including harm to transplant), poor vaccine efficacy, and lack of rigorous testing were barriers to vaccine uptake. Most undecided recipients would proceed with vaccination with specific recommendations and vaccine information provided by their transplant specialist/team. These simple interventions can be readily implemented to optimize vaccine uptake.

Consensus guidelines for improving patients' understanding of invasive fungal disease and related risk prevention in the haematology/oncology setting, 2021.

Patients with invasive fungal disease (IFD) are at significant risk of morbidity and mortality. A productive partnership between patients, their carers/families, and the multidisciplinary team managing the infection and any underlying conditions, is essential. Sharing information and addressing knowledge gaps are required to ensure those at risk of IFD avoid infection, while those with suspected or confirmed infection optimise their therapy and avoid toxicities. This new addition to the Australian and New Zealand consensus guidelines for the management of IFD and antifungal use in the haematology/oncology setting outlines the key information needs of patients and their carers/families. It specifically addresses risk factor reduction, antifungal agents and adherence, and the risks and benefits of complementary and alternative therapies. Knowledge gaps are also identified to help inform the future research agenda.

SARS-COV-2 vaccine acceptance in patients with rheumatic diseases: a cross-sectional study.

OBJECTIVES: To evaluate severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccine acceptance among patients with rheumatic diseases (RMD). METHODS: All rheumatology patients attending a large suburban health network were invited to participate in an anonymized online survey. The primary outcome of interest was SARS-COV-2 vaccine acceptance. RESULTS: The mean (SD) age of respondents (n = 641) was 52.7 (15.1) years and 74.4% (n = 474) were female. Sixty-five percent were willing to have a SARS-COV-2 vaccine, while 34.4% were vaccine-hesitant (unwilling or undecided). On multivariate analysis, vaccine acceptance was associated with smoking (OR: 2.25 [95% CI: 1.22-4.15; p = .009]), history of malignancy (OR: 2.51 [95% CI: 1.19-5.26; p = .015]), influenza or pneumococcal vaccination in the preceding year (OR: 2.69 [95% CI: 1.78-4.05; p < .001]) and number of COVID-Safe measures practiced (OR: 1.54 [95% CI: 1.05-2.26; p = .027]). Vaccine acceptance correlated with positive beliefs regarding vaccine efficacy (r = 0.40; p < .001) and safety (r = 0.36; p < .001). Vaccine acceptance correlated negatively with concerns regarding side-effects (r = -0.30; p < .001) and vaccine-associated RMD flare (r = -0.21; p < .001). In vaccine-hesitant respondents, 39.2% were more likely to accept vaccination if given a choice of which vaccine they receive and 54.5% if their rheumatologist recommended it. Twenty-seven percent of patients on immunomodulators were willing to withhold medications while 42.1% were willing if advised by their rheumatologist. CONCLUSION: SARS-COV-2 vaccine hesitancy is prevalent amongst RMD patients and associated with concerns regarding vaccine safety, efficacy, side effects and RMD flare. Clinician recommendation, vaccine choice and communications targeting patient concerns could facilitate vaccine acceptance.Significance and Innovations Vaccine hesitancy is prevalent in RMD patientsVaccine acceptance is associated with beliefs regarding vaccine safety and efficacy and concerns regarding RMD flare and vaccine-associated side effectsVaccine choice and clinician recommendation have the potential to improve vaccine acceptance in patients who are hesitant.

A Description of the Type, Frequency and Severity of Infections Among Sixteen Patients Treated for T-Cell Lymphoma.

BACKGROUND: Infections are an important cause of morbidity and mortality in T-cell lymphomas. Factors contributing to increased risk of infection include the nature of the underlying disease, as well as treatment-associated immunosuppression. Currently there are few reports describing the types of infections, including preventable infections, in this cohort of patients. The aim of the study was to identify the type, frequency and severity of infection in patients with T-cell lymphoma undergoing treatment. METHODS: A case series was performed on all patients with T-cell lymphoma over a 5-year period from 2011 to 2016 at a tertiary Australian hospital. Information was collected from medical record review regarding patient demographics, lymphoma treatment and outcomes, and infectious outcomes. Severe infections were recorded, defined as infection requiring hospitalization. RESULTS: Sixteen patients were identified with a diagnosis of T-cell lymphoma who received treatment at our institution. There were 42 discrete episodes of severe infections in total. Severe infections occurred in 81% of patients, with over 40% having more than one infection. The median length of hospital stay was 13 days, 33% required intensive care admission and 14% of infectious episodes resulted in death. Only 50% of infectious episodes were microbiologically proven, with the most common etiology being bacterial. The most commonly isolated organism overall was Staphylococcus aureus, with the most common source of infection being skin and soft tissue. There was one case of cytomegalovirus (CMV) infection and five cases (12%) of invasive fungal infection. The highest rates of infection occurred during progressive disease. Rates of prophylaxis were highest with antiviral agents, and comparatively lower with antibacterial and antifungal agents. CONCLUSION: Infections are frequent, opportunistic and severe in patients with T-cell lymphoma. Our data suggests that fungal prophylaxis may be indicated with T-cell lymphoma.

Severe infections remain common in a real-world rheumatoid arthritis cohort: A simple clinical model to predict infection risk.

OBJECTIVE: This study aimed to investigate the incidence of severe infections in patients of a dedicated rheumatoid arthritis (RA) clinic, identify the associated risk factors, and derive an infection risk screening tool. METHODS: Between January and July 2019, 263 eligible patients with a diagnosis of RA were recruited retrospectively and consecutively from an RA clinic of an Australian tertiary hospital. The primary outcome was severe infection (requiring hospital admission) between January 2018 and July 2019. We collected data from medical records and pathology results. We used validated scores, such as the disease activity score of 28 joints (DAS28) and the Charlson comorbidity index, to assess the disease activity and comorbidity burden. Multivariable logistic regression was used for statistical analysis. RESULTS: A total of 45 severe infection episodes occurred in 34 (13%) patients, corresponding to 10.8 infections per 100 patient-years. Respiratory (53%) and urinary (13%) tract infections were the most common. In the multivariable analysis, significant risk factors included low lymphocyte count (odds ratio [OR], 4.08; 95% confidence interval [CI], 1.16-14.29), severe infection in the past 3 years (OR, 3.58; 95% CI, 1.28-9.97), Charlson comorbidity index >2 (OR, 2.69; 95% CI, 1.03-7.00), and higher DAS28 (OR, 1.35/0.5-unit increment; 95% CI, 1.10-1.67). A model incorporating these factors and age had an area under receiver operating characteristic curve of 0.82. CONCLUSION: To the best of our knowledge, this was one of the first Australian studies to evaluate severe infection rates in a real-world RA cohort. The rates remained high and comparable with those of the older studies. Lymphopenia, disease activity, comorbidity burden, and previous severe infection were the independent risk factors for infection. A model comprising easily assessable clinical and biological parameters has an excellent predictive potential for severe infection. Once validated, it may be developed into a screening tool to help clinicians rapidly identify the high-risk patients and inform the tailored clinical decision making.

Persistently positive culture of antimicrobial-susceptible Legionella pneumophila despite appropriate therapy.

Legionella pneumophila is a well-known cause of pneumonia that is infrequently cultured in clinical practice. We report a case of an immunocompromised patient with persistently positive L. pneumophila cultures from invasive respiratory samples despite prolonged treatment with appropriate antibiotics. In vitro testing showed that the isolate remained susceptible to ciprofloxacin and azithromycin.

Evaluating the sustained effectiveness of a multimodal intervention aimed at influencing PIVC insertion practices in the emergency department.

INTRODUCTION: Multimodal interventions (MMI) are frequently used in various healthcare settings to encourage change in healthcare personnel practices and improve patient safety. In 2013, an MMI conducted in an Australian metropolitan ED used clinician champions, guidelines, education sessions and promotional materials to encourage a reduction in unused and inappropriate peripheral intravenous cannulas (PIVC). A 60-day postintervention demonstrated a successful reduction in the number of unused PIVCs without changes in appropriate insertions. We aimed to investigate if this MMI produced a sustained effect in reducing the frequency of unused PIVCs inserted in this ED. METHODS: A single-centre retrospective cohort study of adult patients presenting to the above ED in Victoria, Australia, was conducted in April 2018. A random sample of 380 patients with a PIVC inserted in ED was assessed to determine if the PIVC was used (termed 'Long-term follow-up'). The appropriateness of unused PIVCs was assessed. Our findings were compared with previously collected data in 2013 ('Pre-Intervention' and 'Immediately Post-Intervention') to determine a sustained reduction in the frequency of unused PIVC insertions was achieved. Long-term analysis of the MMI, including the overall frequency of PIVC insertions in ED before and after the MMI, was also collected. RESULTS: In our Long-term follow-up cohort, 101 of 373 (27.1%, 95% CI 22.6% to 31.9%) PIVCs were unused (seven cases excluded). This was significantly lower than the Pre-Intervention cohort (139/376, 37.0%, 95% CI 32.1% to 42.1%). While not significant, the frequency of unused PIVCs in the Post-Intervention cohort was lower in comparison (73/378, 19.3%, 95% CI 15.4% to 23.7%). No significant change in the appropriateness of unused PIVCs was observed between the Post-Intervention and Long-term follow-up. The overall proportion of patients receiving a PIVC has remained low since the MMI. CONCLUSION: An MMI aimed at reducing unused PIVC insertions in ED has been effective in eliciting sustained change. Unused but appropriately inserted PIVCs seem unaffected by the intervention.

Incidental mucocutaneous cytomegalovirus detection and its predictive value for systemic disease.

Multiplex polymerase chain reaction (PCR) testing has revolutionised microbiological practice but also increased the number of positive results of uncertain significance. This phenomenon has been seen in the increasing detection of cytomegalovirus (CMV) in mucocutaneous swabs for herpesviruses, the microbiological significance of which is a priori unclear. The aim of our study was to determine if an incidental finding of a positive CMV result represented CMV disease, if it facilitated a timely diagnosis of CMV disease or whether there were any deleterious outcomes. We performed a retrospective review of patients with an incidentally positive PCR result for CMV on external and mucosal swabs, including medical comorbidities and presence of immunosuppression, subsequent investigations, whether a diagnosis of CMV disease was made, and treatment. CMV detection was infrequent, detected in 158 (3.4%) of 4626 herpes multiplex PCR tests performed. The majority (60.4%) of patients were immunocompromised, and amongst these patients a positive swab represented a new diagnosis or already known CMV disease in 14%. In seven patients (5%), all of whom were immunocompromised, the positive CMV PCR on a swab led to further investigation and subsequent diagnosis and treatment of CMV disease. Whilst not recommended for diagnosis of CMV disease, if CMV is detected on a mucocutaneous swab in an immunocompromised patient, further assessment and investigation for CMV disease should be undertaken.

Prevalence and distribution of functional splenic tissue after splenectomy.

BACKGROUND: Individuals splenectomised for trauma have lower infection rates than those splenectomised for other conditions. Residual functional splenic tissue (FST) after splenectomy may provide ongoing immunological protection. AIMS: To quantify the prevalence and volume of residual FST post-splenectomy using standard testing. METHODS: Splenectomised adults were recruited from the Spleen Australia clinical registry. Eligible individuals had been splenectomised at least 1 year prior to their visit and resided in Victoria. Splenic function was identified by evaluating Howell-Jolly bodies and IgM memory B cells. A 99m-Technetium-labelled, heat-denatured erythrocyte scintigraphic scan was performed if splenic function was detected. RESULTS: Initially, 75 splenectomised individuals (all cause) were recruited, with a median of 58 years of age and who were splenectomised a median of 14 years previously. The most common indications for splenectomy were trauma (30.7%) and haematological disease (28.0%). Scintigraphy identified FST in nine individuals (12.0%). Eight had been splenectomised for trauma. In this cohort, 34.8% of individuals splenectomised for trauma had residual FST. To explore our findings further, 45 additional individuals were recruited, predominately individuals splenectomised for trauma. Twenty-five individuals completed assessments by December 2018. An additional 11 individuals had FST, of whom 9 had been splenectomised for trauma. Overall, we identified 20 individuals with residual FST. Volumes ranged from 2.2 to 216.0 cc. We saw individuals with accessory spleens and splenotic nodules and an individual with both. Seventeen individuals had been splenectomised for trauma. CONCLUSIONS: Residual FST is commonly seen in individuals splenectomised for trauma. It can present in varying distributions and of varying volume. The clinical significance is unclear.

A simple score can identify kidney transplant recipients at high risk of severe infection over the following 2 years.

BACKGROUND: The aim of this study was to determine whether a composite score of simple immune biomarkers and clinical characteristics could predict severe infections in kidney transplant recipients. METHODS: We conducted a prospective study of 168 stable kidney transplant recipients who underwent measurement of lymphocyte subsets, immunoglobulins, and renal function at baseline and were followed up for 2 years for the development of any severe infections, defined as infection requiring hospitalization. A point score was developed to predict severe infection based on logistic regression analysis of factors in baseline testing. RESULTS: Fifty-nine (35%) patients developed severe infection, 36 (21%) had two or more severe infections, and 3 (2%) died of infection. A group of 19 (11%) patients had the highest predicted infectious risk (>60%), as predicted by the score. Predictive variables were mycophenolate use, graft function, CD4+, and natural killer cell number. The level of immunosuppression score had an area under the receiver operating curve of 0.75 (95% CI: 0.67-0.83). CONCLUSION: Our level of immunosuppression score for predicting the development of severe infection over 2 years has sufficient prognostic accuracy for identification of high-risk patients. This data can inform research that examines strategies to reduce the risks of infection.

Natural killer cell function predicts severe infection in kidney transplant recipients.

The aim of this study was to determine if natural killer cell number (CD3- /CD16± /CD56± ) and cytotoxic killing function predicts severity and frequency of infection in kidney transplant recipients. A cohort of 168 kidney transplant recipients with stable graft function underwent assessment of natural killer cell number and functional killing capacity immediately prior to entry into this prospective study. Participants were followed for 2 years for development of severe infection, defined as hospitalization for infection. Area under receiver operating characteristic (AUROC) curves were used to evaluate the accuracy of natural killer cell number and function for predicting severe infection. Adjusted odds ratios were determined by logistic regression. Fifty-nine kidney transplant recipients (35%) developed severe infection and 7 (4%) died. Natural killer cell function was a better predictor of severe infection than natural killer cell number: AUROC 0.84 and 0.75, respectively (P = .018). Logistic regression demonstrated that after adjustment for age, transplant function, transplant duration, mycophenolate use, and increasing natural killer function (odds ratio [OR] 0.82, 95% confidence interval [CI] 0.74-0.90; P < .0001) but not natural killer number (OR 0.96, 95% CI 0.93-1.00; P = .051) remained significantly associated with a reduced likelihood of severe infection. Natural killer cell function predicts severe infection in kidney transplant recipients.