Risk Factors Of Hemodialysis Catheter-related Bloodstream Infection And Prediction Model

Backgrounds and ObjectivesChronic kidney disease(CKD),especially end-stage renal disease(ESRD),has gradually become a global public health problem.Hemodialysis remains the majority of renal replacement therapy(RRT)modality worldwide.The latest report from the China Renal Data System(CNRDS)shows that the number of HD patients in China was 630,000 by the end of 2019.A well-functioning dialysis access is essential for the implementation of HD.Since its application in the 1960s,the arteriovenous fistula(AVF)has been considered to be a more ideal choice because of its longer patency and fewer complications.In contrast,central venous catheters(CVC),including tunnel-cuffed catheters(TCC),are recommended by guidelines/consensus to be avoided or limited due to poor patency rates and complications.Catheter-related infections(CRI),especially catheter-related bloodstream infections(CRBSI),are known to increase healthcare costs and significantly increase the risk of hospitalization and death.However,depending on the timing of dialysis,the condition of the patient at the time of dialysis entry,the patient’s vascular condition,the level of medical technology,and patient acceptance,approximately 0.1 to 45%of patients rely on CVC to maintain HD therapy.Identifying risk factors related to catheter infection or constructing a predictive model of infection risk can help clinical diagnosis of CRBSI and assessment of infection prognosis at an early stage,which has certain guiding significance for improving catheter patency rate and rational use of catheters.In this retrospective study,we collected clinical data of patients,examined the the risk factors of CRBSI in HD patients by univariate and multifactorial logistic regression,and constructed a CRBSI risk prediction model with the aim of providing clinical evidence for the prevention and treatment of CRI.MethodsA total of 254 patients who underwent TCC removal at our hemodialysis access center from July to December 2020 were included in our study.The patients were randomly divided into two groups according to the ratio of 7:3,179 patients in the training set(for risk prediction model construction)and 75 patients in the validation set(for model validation).All patients were more than 18 years,HD treatment for more than 3 months,3 times a week for 4 hours,and had complete clinical data.Right internal jugular vein was the unique route while other alternative routes were excluded.This TCC was in use until its removal.Reasons for extubation included CRBSI,change to arteriovenous access(AVF or arteriovenous graft,AVG),successful renal transplantation,catheter mechanical lesions(catheter breakage and catheter slippage),and catheter dysfunction.Exclusion criteria included exclusion of patients with serious infections of other systems,patients with surgery,trauma,pregnancy status,psychiatric disorders,malignancy,patients on heparin-free hemodialysis,patients on combined dialysis with other accesses within 1 month prior to TCC removal,and patients with incomplete clinical data.All procedures were performed under digital subtraction angiography(DSA)guidance.Retain catheter tip for bacterial culture if the patients were diagnosed with CRB SI.General patient data,laboratory tests and imaging were collected.Data were expressed as mean±SD or median(interquartile range).Counts for categorical variable expressed as percentage of the total.Numerical data were analyzed as appropriate by Mann-Whitney U test.Categorical data were compared using the χ2 test or Fisher’s exact test.Logistic binary regression was applied to examine the association between CRBSI and result variables.P<0.05 was considered statistically significant.Univariate analysis results with P<0.2 were included in multivariate Logistic regression.When the backward LR method was used to screen the variables included in multivariate analysis,a was relaxed to 0.2,and all the other statistical criteria were a=0.05.The OR values of each factor in the multivariate logistic regression were rounded to the assigned scores of the risk model,and the scores of each factor were summed to obtain the risk score when assessing the risk of CRBSI in patients.The sensitivity and specificity of the cut-off value at the maximum of the Youdon index were calculated and verified in the validation set.p<0.05 was considered statistically significant.0.05 was considered statistically significant.Results1.There were 179 HD patients in the validation set,103 males and 76 females,with a mean age of 55.81±15.95 years and a median HD vintage of 18(8,27)months.Primary conditions included diabetic nephropathy in 62 cases(34.6%),hypertensive renal damage in 36 cases(20.1%),chronic glomerulonephritis in 57 cases(31.9%),and other etiology in 24 cases(13.4%).The median TCC retention time was 15(5,24)months.2.All patients underwent central venography,and the results showed that fibrin sheath formation in 87(48.6%)cases,central venous disease in 71(39.7%)patients,central venous thrombosis(CVT)in 34 patients(47.9%)and central venous stenosis(CVS)in 37 patients(52.1%)in the validation set.3.36 patients had positive culture of the catheter tip among the 40 CRBSI patients in the validation set,with a culture positivity rate of 90%.The rate of gram-positive bacterial infection,gram-negative bacterial infection and fungal infection were 52.8%,33.3%and 13.9%respectively.Among the gram-positive bacteria,Staphylococcus aureus was the most common,with 10 strains(27.8%),among the gram-negative bacteria,Pseudomonas aeruginosa was the most common,with 4 strains(11.1%),and the fungi were all Candida parapsilosis.4.In the validation set,compared with non-CRBSI group,the CRBSI group had a higher proportion of diabetes(χ2=5.932,P=0.015),history of previous catheter infection in 3 months(χ2=9.636,P=0.002),more than 4 times ursing interventions in the 1 month before TCC removal(χ2=8.013,P=0.005),central venous disease(χ2=5.062,P=0.024),fibrin sheath(χ2=5.544,P=0.019),and albumin below 40g/L(χ2=3.982,P=0.046).5.Univariate Logistic regression showed that diabetes mellitus(OR=2.445,95%CI,1.177～5.081),history of CRI in 3 months(OR=3.212,95%CI,1.505～6.856),more than 4 times nursing interventions in the 1 month before TCC removal(OR=2.898,95%CI,1.363～6.162),and central venous disease(OR=2.245,95%CI,1.100-4.582),fibrin sheath(OR=2.375,95%CI,1.143-4.933),albumin less than 40g/L(OR=2.070,95%CI,1.005～4.262)was risk factors for CRBSI.6.Multivariate Logistic regression adjusting for gender,catheter indwelling time showed that diabetes mellitus(OR=2.711,95%CI,1.174～6.258),history of CRI in 3 months(OR=3.674,95%CI,1.541～8.760),more than 4 times nursing interventions in the 1 month before TCC removal(OR=3.128,95%CI,1.343～7.283),and central venous disease(OR=2.572,95%CI,1.130～5.854)were independent risk factors for CRBSI.7.The variables in the multivariate regression analysis were assigned scores based on OR values rounded to assess the risk of CRBSI in patients by adding the scores corresponding to each factor to obtain the risk score,constructing a CRBSI risk prediction model in the training set,and plotting the ROC curve with an AUC of 0.761(0.683～0.839).Combining sensitivity and specificity,a cut-off of≥6 was suggested for predicting CRBSI,and it was validated in the validation set,the AUC was 0.794(0.674～0.914),the sensitivity and specificity corresponding to this cut-off were 61.1%and 82.5%.ConclusionsIn patients with indwelling TCC,diabetes mellitus,history of CRI in 3 months,more than 4 times nursing interventions in the 1 month before TCC removal and central venous disease are independent risk factors for CRI,and avoiding or shortening dialysis catheter indwelling is still the most fundamental measure to prevent CRI.