Objectives: In renal transplant recipients, infections can have atypical outcomes due to immunosuppression therapy, and delayed diagnosis causes high mortality. We developed an infection risk score in renal transplant recipients to rapidly predict infection risk in emergency departments.
Materials and Methods: Of 870 renal transplant recipients admitted to the emergency department, we included 608 patients and 262 control cases. Hospital record system data for renal transplant recipients were retrospectively investigated for the period January 2021 through December 2025. Laboratory and vital signs of patients suspected of infection were com-pared versus asymptomatic control cases admitted for routine check-ups. All of our patients met ethical standards and were selected as related donors and recipients.
Results: Demographic characteristics and findings of 608 patients and 262 control cases were compared, and no significant difference was found between the 2 groups in terms of comorbidities (P > .⁰⁵). The most frequent presenting symptom was diarrhea (20.1%); the least frequent symptom was sore throat (6.6%). Leukocytes, neutrophils, neutrophil-lymphocyte ratio, plasma-lymphocyte ratio, C-reactive protein, and sodium and potassium levels, as well as vital signs including fever, pulse, blood pressure, and peripheral oxygen saturation, were associated with infection. This novel infection risk score comprised 6 parameters, including C-reactive protein, neutrophil-lymphocyte ratio, sodium, fever, pulse, and systolic blood pressure, predicted infections with 89.1% (area under the curve) accuracy. Sensitivity, specificity, positive predictive value, and negative predictive value of the score were 0.734, 0.905, 0.947, and 0.594, respectively.
Conclusions: The novel infection risk score developed in our study predicts infections in renal transplant recipients with high accuracy using only routine biochemical and vital parameters. This practical, rapid, and reliable system can contribute to early diagnosis processes in emergency departments. Future external validation through multicenter studies is needed to support its integration into clinical guidelines.

Key words : Emergency room, Immunosuppression, Renal transplantation

Introduction

Renal transplant (RT) is considered the most effective treatment option for patients with end-stage renal failure, significantly reducing mortality and consi-derably improving quality of life.¹ Due to advances in transplant techniques and the effectiveness of immunosuppression treatment regimens, graft and patient survival have improved significantly; in parallel, the number of patients in long-term follow-up after RT has also increased significantly.² Despite these gains, immunosuppression therapies adminis-tered in the post-RT period significantly increase susceptibility to infections, and infections remain a main cause of morbidity and mortality in RT recipients (RTRs).³ Infections are among the most frequent causes of hospitalization in RTRs, and infections are strongly associated with the need for intensive care unit admission, graft dysfunction, and graft loss.^3,4 It has been shown that not only the immune status of the patient but also the infection type and intensity, as well as the duration of immunosuppression administered, are decisive in the development of infection. In Monlezun and colleagues, the risk of infection was revealed to be associated with the specific immunosuppressive agents used, indepen-dent of the total immunosuppression load. Mon-lezun and colleagues also reported that the use of mycophenolate mofetil and alemtuzumab reduced the risk of infection; however, cytomegalovirus antigenemia was one of the strongest predictors of infection development.⁵ Their findings demonstrated that the risk of infection in RTRs is a complex and multidimensional clinical problem that cannot be explained by a single parameter. Emergency departments (EDs) are among the most frequently visited health care facilities for RTRs due to suspected infections. However, infections often present with atypical clinical courses in this patient group. Fever, leukocytosis, and classic inflammatory responses may develop in a suppressed, subtle, or delayed manner due to immunosuppression, which may substantially complicate the process of early diagnosis and accurate risk stratification of the infection.⁶ Delay of infection diagnosis in the ED is directly associated with the development of sepsis, prolonged hospital stay, and increased mortality.⁷ Therefore, early risk assessment is critically important for clinical outcomes in the diagnosis of this patient group. In recent years, several clinical risk scores have been developed to predict the risk of serious infection in RTRs. Dendle and colleagues defined a risk score that included clinical parameters and immune cell subgroups and revealed that estimated glomerular filtration rate, mycophenolate use, CD4-positive T-lymphocyte count, and natural killer cell count were significantly associated with the development of serious infections.⁸ However, the vast majority of such scores target stable outpatients and do not adequately reflect the acute clinical presentations encountered in the ED, the need for a rapid decision-making process, and the urgency for laboratory data within a restricted time. However, in the ED, determination of the risk of infection in RTRs with rapid, reliable, and easily applicable methods is vital to facilitate appropriate decisions regarding hospitalization, intensive care unit requirements, and empirical antimicrobial treatment.^6,9 In this context, a clear need in the ED exists for a specific and practical infection risk score based on routine clinical eva-luation and readily available laboratory parameters. In the published literature, most studies on predictors for the development of infection in RTRs have focused on retrospective risk factor analyses or specific subtypes of infections. There is no syste-matically validated score specific to the ED for prediction of the immediate risk of infection using routine clinical and laboratory data. Therefore, we aimed to provide a rapid and applicable infection risk score for the ED setting.

Materials and Method

Among 1198 RTR patients administered to the ED of the Medical School Hospital of Baskent University in Ankara, Türkiye, we excluded patients with incomplete laboratory data and other test findings and divided the 870 remaining patients into the following 2 groups: the experimental (patient) group (n = 608) and the control group (n = 262). In the control group, we examined blood test findings and drug levels of the patients who visited the transplant outpatient clinic for routine check-ups for the period January 2021 through December 2025 and for whom there no infection-related complaints during these visits. Our aim was to show whether there was a difference in complete blood count parameters and drug levels between periods with and without infection in RTRs. We investigated patient file records and labora-tory findings from the hospital information mana-gement system. From these records, we obtained data on patient age on admission, focus of infection, complete blood count data (leukocytes, erythrocytes, platelets, neutrophils, lymphocytes, mean platelet volume, and red cell distribution width), levels of C-reactive protein (CRP) and immunosuppressant drugs used, graft age, and hospital stay duration (if hospitalized). We investigated the relationship between infec-tions and the level of immunosuppressive drugs (tacrolimus, everolimus, sirolimus, cyclosporine, and mycophenolate mofetil) in RTRs presenting to our ED for the period January 2021 through December 2025, with rejection parameters such as increased serum creatinine, decreased urine output, graft tenderness, fever, edema, hypertension, and sudden weight gain. Renal allograft rejection was defined through histopathological evaluation according to the elevated serum creatinine, presence of donor-specific antibodies, and the Banff classification criteria.¹⁰ The null hypothesis for our study was that no association existed between routinely tested biomarker findings and infection risk scores in RTRs. Our alternative hypothesis was that an association existed between routinely tested biomarker findings and infection risk scores in RTRs.

Statistical analyses
We used Python software (python.org) and R software (r-project.org) for all statistical analyses. We expressed continuous variables as median values (with interquartile range [IQR]) and categorical variables as frequencies (with percentage). The distribution of continuous variables was assessed with the Shapiro-Wilk test, indicating that most variables were nonnormally distributed; accordingly, nonparametric methods were applied. We compared groups (infected vs not infected) using the Mann-Whitney U test for continuous variables and the chi-square test for categorical variables. To identify the factors associated with infection, we initially perfor-med univariate logistic regression analysis. For variables with statistical significance, we further evaluated for multicollinearity using the variance inflation factor, excluding those with high collinearity. We then conducted multivariable logistic regression to identify independent predictors. We assessed correlations between selected clinical and laboratory parameters using the Spearman correlation analysis. We evaluated the diagnostic performance of individual parameters and the constructed model using receiver operating characteristic curve analysis. The area under the curve (AUC) was calculated to determine discriminative ability, and optimal cutoff values were identified using the Youden J index. The AUC values were interpreted as excellent (0.90-1.00), very good (0.80-0.89), and acceptable (0.70-0.79). All analyses were performed within a 95% CI, and P < .⁰⁵ was considered statistically significant.

Results

Of 870 cases, 608 individuals were in the patient group and 262 individuals were in the control group. The patient group comprised 31.7% (n = 193) female patients and 68.3% (n = 415) male patients. The control group consisted of 30.5% (n = 80) female patients and 69.5% (n = 182) male patients. Mean ages of patient and control groups were 42 and 40.5 years, respectively, and no significant differences were found between the groups in terms of sex and age (Table 1). Individuals in the patient group were distributed as follows: 36.7% (n = 223) with low medication levels, 61.5% (n = 374) with moderate medication levels, and 1.8% (n = 11) with high medication levels. Individuals in the control group were distributed as follows: 37% (n = 97) with low medication levels, 61.8% (n = 162) with moderate medication levels, and 1.1% (n = 3) with high medication levels. The number of individuals using tacrolimus was 66% (n = 401) in the patient group and 67.9% (n = 178) in the control group, and the number of individuals using cyclos-porine was 24% (n = 146) in the patient group and 24.8% (n = 65) in the control group. However, 10% (n = 61) of individuals in the patient group and 7.3% (n = 19) of individuals in the control group used sirolimus, with no significant difference between the patient group and control group in terms of the use of immunosuppressant drugs (Table 1). Rates of graft age (time since transplant) in the patient and control groups were 10.7% (n = 65) and 12.6% (n = 33), respectively, for graft age between 0 and 6 months; 8.6% (n = 52) and 9.2% (n = 24), respectively, for graft age between 6 and 12 months; 9.7% (n = 59) and 11.1% (n = 29), respectively, for graft age between 1 and 2 years; 16.8% (n = 102) and 13% (n = 34), respectively, for graft age from 3 to 5 years; and 51.6% (n = 314) and 51.9% (n = 136), respectively, for graft age ≥5 years. No significant difference was shown between the 2 groups in terms of graft age. No significant difference was found between the patient and control groups in terms of comorbid diseases (Table 1). Given the complaints on admission, the most common complaint was diarrhea at 20.1% (n = 122), and the least common complaint was sore throat/upper respiratory tract infection at 6.6% (n = 40). The most frequent source of infection was acute gastroenteritis with a rate of 25.3% (n = 154), and the least frequent source of infection was soft tissue infection with a rate of 3% (n = 18) (Figure 2). Also, the number of hospitalized patients was 53.6% (n = 326), and the number of discharged patients was 46.4% (n = 282). The average hospital stay duration was 4 days (minimum-maximum. 3-8 days), and the rate of patient death during hospital stay was 2.8% (n = 9) (Table 2). We compared laboratory findings of the patient group versus the control group and found significant differences between the 2 groups, including leukocytes, neutrophils, platelets, lymphocytes, sodium, potassium, CRP, systolic blood pressure (BP), diastolic BP, pulse, fever, peripheral capillary oxygen saturation, neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (Table 3, Figure 1). In the clinical model developed to create the in-fection risk score, sodium, CRP, systolic BP, pulse, fever, and NLR were all significant (Table 4, Figure 3). The fact that the highest AUC value was in the NLR parameter suggested that a ratio-based indica-tor of inflammation carries a very strong signal in differentiating infections in RTRs. The fact that the AUC value of CRP was close to the AUC value of NLR suggested that the acute phase response is similarly a valuable component; however, when used alone, CRP may miss several cases. The fact that, although the AUC value of sodium remained moderate, the directional information was assessed to be “low risk,” suggesting that hyponatremia may be a marker more related to the systemic response/severity spectrum and gain when consi-dered in conjunction with other parameters. The moderate AUC values for systolic BP and fever indicated that vital signs alone do not act as diagnostic tests; however, vital signs can be beneficial as a component of a clinical score. The relatively higher AUC for pulse suggested that tachycardia may be beneficial in capturing the physiological response to infections in immunosuppressed patients; however, the specificity/sensitivity balance may vary at the single-parameter level (Table 5). The AUC values of individual variables are shown in Figure 4 from highest to lowest (NLR = 0.801, CRP = 0.793, lymphocyte = 0.725, neutrophil = 0.721, pulse = 0.708, sodium = 0.694, fever = 0.673, and platelet-to-lymphocyte ratio = 0.670). However, in the individual receiver operating characteristic curve analysis, each parameter was presented separately in terms of AUC, sensitivity, and specificity (Table 6). The scoring method of our novel infection risk score for RTRs, which we named IRS-RT, is presented in Table 7. Very-high-risk scores (>8 points) were 99.065% (n = 106) and 0.935% (n = 1) in the patient and control groups, respectively, and low-risk scores (0-2 points) were 35.85% (n = 119) and 64.15% (n = 213) in the 2 groups, respectively. Moderate risk scores (3-5 points) were 80.9% (n = 199) and 19.1% (n = 47) in the patient and control groups, respectively, and high-risk scores (6-8 points) were 99.46% (n = 184) and 0.54% (n = 1) in the 2 groups, respectively (Figure 5). The IRS-RT score was developed to predict the presence of serious infections earlier in RTRs who presented to the ED and to achieve the following objectives via routinely obtainable parameters during the initial assessment by ED physicians: (1) predict the likelihood of infections earlier, (2) estimate the need for hospitalization/close monitoring, and (3) support the empirical antibiotic decision. The IRS-RT tool can predict the risk of infections in RTRs presenting to the ED with 89.1% accuracy (AUC) using 6 routine parameters (CRP, NLR, sodium, fever, pulse, and systolic BP). The IRS-RT can be quickly calculated from complete blood count and biochemistry findings without requiring additional tests, supporting the decision-making process of ED physicians (Figure 1). The sensitivity of the score was measured to be 0.734; the specificity, positive predictive value, and negative predictive value were detected as 0.905, 0.947, and 0.594, respectively (Figure 6).

Discussion

In our present study, we developed the IRS-RT tool as a novel, easily applicable, and clinically significant risk scoring method for the prediction of the presence of infections in RTRs at ED presentation. The main finding of our study was the ability of IRS-RT to predict the risk of infections in RTRs with 90% accuracy using 6 routine parameters consisting of simple biomarkers and clinical characteristics. The ED is the most frequent setting for patients presenting with suspected infections, especially in RT patient groups, due to immunosuppression. Thus, the EDs are directly related to the development of sepsis and mortality in cases for which diagnosis and treatment of such patients are delayed, which creates subsequent delays in admission. Therefore, early risk assessment in this population is critically important for clinical outcomes.^6,7 These data can help clinicians identify patients at high risk for infections and also guide clinical trials that investigate strategies to reduce infections in RTRs. The primary objective of our IRS-RT tool was to predict the presence of serious infections at the ED presentation in RTRs. Furthermore, the score can predict the likelihood of infections using several parameters routinely obtained during the initial assessment by the ED physicians, can estimate the requirement for hospitalization/close monitoring, and can also support empirical antibiotic decisions. Of note, for the period January 2021 through December 2025, 870 participants were retrospectively evaluated during the development process of the IRS-RT. Our study findings demonstrated that infections can be predicted with high accuracy using the combination of routinely obtainable biochemical and vital parameters. In particular, the high discri-minatory power of the developed score (AUC = 0.891) and similar performance in internal validation supported the potential of the IRS-RT for clinical practice. One of the most noticeable findings of our study was that, compared with other individual biomarkers, the combined score approach provides a significant advantage. Indeed, although the AUC value of NLR, which is the parameter with the highest individual performance, was 0.801, the AUC value of the combined score reached 0.891 and thereby reflects the multifactorial nature of the infection, indicating that the combined assessment of the inflammatory response, hemodynamic alterations, and electrolyte imbalances will provide more reliable outcomes in the clinical decision-making process. In 2019, Dendle and colleagues published a prospective study of 168 patients in 2015 for which they had calculated the infection score based on 4 components, including CD4-positive T-lymphocyte count, natural killer cell count, estimated glomerular filtration rate, and mycophenolate use; they conclu-ded that the developed immunosuppression level score has sufficient accuracy to predict the risk of serious infection within 2 years and could be used to identify high-risk patients.¹¹ Unlike the 4 components in the study by Dendle and colleagues, the 6 biomarkers included in the scoring criteria of our present study were CRP, NLR, sodium, fever, pulse, and systolic BP, which yielded significant outcomes. Each of these parameters in the score represents the pathophysiology of infections from different perspectives. Although CRP and NLR stand out as strong indicators of systemic inflammatory response, hyponatremia is associated with increased antidiuretic hormone (ADH) secretion and cytokine response during infection.¹² Vital signs, such as fever, pulse, and systolic BP, reflect the systemic physiological response to infection and are among the early and reliable indicators of the systemic physiological response to infection. In particular, parameters such as fever, pulse, and systolic BP reflect the host organism’s response to inflammatory processes. Although fever is consi-dered an indicator of the acute phase response, triggered by the action of proinflammatory cytokines, particularly interleukin-1 (IL-1), interleukin-6 (IL-6), and tumor necrosis factor-alpha (TNF-α), tachycardia develops as a result of increased metabolic needs and systemic inflammation. Alterations in systolic BP are particularly associated with vasodilation and increa-sed capillary permeability observed in the later stages of infection and may be an early sign of hemodynamic instability. When evaluated together, these parameters play a crucial role for deter-mination of the severity of infection and prediction of clinical prognosis. Indeed, the existence of vital signs such as fever and heart rate in the systemic inflam-matory response syndrome (SIRS) criteria also supports the idea that these parameters are essential components of the systemic response to infection. Likewise, the central role of vital signs is emphasized in presently established sepsis definitions and early warning score systems, and the importance of vital signs in clinical decision-making processes is steadily increasing.^13,14 This multidimensional approach increases diagnostic accuracy, especially in condi-tions in which infection can have an atypical course in immunosuppressed patients. C-reactive protein is an acute-phase reactant and was one of the strongest biomarkers in our score (AUC = 0.793). The diagnostic value of CRP in transplant recipients has been proven.¹⁵ In our study, NLR had the highest AUC value (AUC = 0.801), reflecting the systemic inflammatory response. The NLR is also a reliable indicator of sepsis and infection severity,¹⁶ and decreased sodium levels are commonly seen in infections. Due to the relationship between inflammatory cytokines and ADH, the body releases proinflammatory cytokines, such as IL-1, IL-6, and TNF-α, to stimulate the immune system during infection; these cytokines increase ADH by releasing stimulants to the hypothalamus, and this condition, called hyponatremia, is associated with infection severity.¹⁷ High fever is the classic cardinal sign of infection and is the most basic clinical indicator of infection. Fever is a primary clinical manifestation of infection and represents one of the most significant indicators of the host’s immune response. Fever is also primarily mediated by proinflammatory cytokines, such as IL-1, IL-6, and TNF-α, and constitutes a key component of the systemic inflammatory response in infectious diseases.^13,18,19 On the other hand, tachycar-dia reflects the systemic physiological response to infection and is a key component of the SIRS criteria and the quick sepsis-related organ failure assessment (qSOFA) score. Although not included in the qSOFA score, heart rate remains an important clinical para-meter in the early recognition of sepsis and systemic inflammatory response.^13,14,20 Systolic BP, an indicator of stress response/hemodynamic changes in infection, reflects the hemodynamic response to infection and may indicate underlying circulatory dysfunction. In infectious conditions, especially sepsis, inflammatory mediators give rise to vasodilation and capillary leakage, resulting in hypotension, which is a key marker of disease severity and is incorporated into clinical scoring systems, such as qSOFA.^14,20-22 Analyses based on the cutoff points indicated that the IRS-RT score can be used in different clinical scenarios. Obtaining high sensitivity at low cutoff values allows IRS-RT to be used as a screening tool, and very high specificity obtained at high cutoff values provides reliable support, especially in hospitalization and aggressive treatment decisions. In this respect, the IRS-RT score offers a dual benefit as both an early diagnosis tool and a clinical decision support system.²³ The significant increase in infection rates under the risk classification strengthens the clinical sig-nificance of the IRS-RT score. The fact that the probability of infection is almost certain, especially in RTRs with a score ≥6, highlights the requirement for early antibiotic initiation and close monitoring in this population, which can contribute to quick and accurate decisions in time-critical settings such as the ED. The first and most important limitation of our study was its single-center, retrospective design. A second limitation was that the control group consisted of individuals who presented to the transplantation outpatient clinic. Another limitation was the small number of different infection foci. Finally, immuno-supElements of the Infection Risk Scorepressive drugs were not included in the score because no significant difference was found between the 2 groups.

Conclusions

Our study presents a practical, rapid, and reliable scoring system for the early diagnosis of infections in RTRs. Future multicenter and prospective studies may allow the achievement of external validation of the IRS-RT score, the evaluation of its performance in different patient groups, and its integration into clinical guidelines.