Prospective cohort study of central venous catheters among internal medicine ward patients

Prospective cohort study of central venous catheters among internal medicine ward patients William E. Trick, MD, Julio Miranda, MD, Arthur T. Evans, MD, MPH, Marjorie Charles-Damte, RN, Brendan M. Reilly, MD, and Peter Clarke, MD Chicago, Illinois Background: Central venous catheter (CVC) use is less well described for patients outside the intensive care unit. We evaluated CVCs and the associated bloodstream infection rate among patients admitted to the general medical service. Methods: We performed a prospective cohort study of patients who had a CVC on admission or inserted during their stay on the general medical service in a public teaching hospital, November 15, 2004, to March 31, 2005. Results: We identified 106 CVCs, 52 were present on admission and 54 were inserted; there were 682 catheter-days. The primary bloodstream infection rate was 4.4 per 1000 catheter-days (95% CI: 0.9-13): highest for catheters inserted in the emergency department compared with those inserted on other units (24 vs 1.7 per 1000 catheter-days), P 5.045. By multivariable analysis, inadequate dressings were more likely among patients with a body mass index $30 kg/m 2, adjusted odds ratio, 3.4 (95% CI: 1.4-8.0). Conclusions: Many CVCs had previously been inserted in the emergency department or intensive care unit; therefore, strategies to reduce bloodstream infections that focus on ward insertion practices may not dramatically reduce bloodstream infection rates. Intervention strategies should target improved dressing care and consideration of early removal or replacement of catheters inserted in the emergency department. (Am J Infect Control 2006;34:636-41.) Central venous catheters often are essential to patient care. Unfortunately, patients who have these catheters occasionally experience catheter-related mechanical or infectious complications, 1 resulting in increased costs and patient morbidity. 2,3 Certain factors increase the risk of a catheter-related complication; for example, compared with subclavian venous insertion, use of the femoral vein confers a higher risk of infectious and other complications (eg, venous thrombosis or hematoma). 4 Among intensive care unit (ICU) patients, dramatic reductions in central venous catheter-associated bloodstream infections rates have occurred after implementation of various intervention strategies 5-10 ; however, similar intervention strategies have not been reported for the non-icu patient population. The following factors likely have influenced decisions to focus interventions on ICU patients: the epidemiology of central venous catheter use is well described 11,12 ; ICUs are geographically constrained, which facilitates educational sessions and monitoring of catheter use and complications, and patients often have a catheter. 11 Much less is known about catheter use outside of ICUs, despite the possibility of more central venous catheters in patients hospitalized outside an ICU. 13 Also, modifiable practices that increase a patient s risk for infectious complications, such as unnecessary catheter use and inadequate exit site dressings, may be more common among non-icu patients. 14 To better understand central venous catheter use among patients outside the ICU, we performed a prospective cohort study of patients on the general medical service. We identified patients who had a central venous catheter on admission to the general medical service (eg, inserted in the emergency department) or who had a catheter inserted while cared for by the service. We recorded patient and catheter characteristics and prospectively monitored them for infectious complications, and we examined the dressing condition at the catheter exit site. From the Department of Medicine, Stroger Hospital of Cook County and Rush Medical College, Chicago, IL. Address correspondence to William E. Trick, MD, Collaborative Research Unit, Stroger Hospital of Cook County, 1900 W. Polk St, Suite 1600, Chicago, IL 60612. E-mail: wtrick@cchil.org. 0196-6553/$32.00 Copyright ª 2006 by the Association for Professionals in Infection Control and Epidemiology, Inc. doi:10.1016/j.ajic.2006.02.008 METHODS Setting and study design Stroger Hospital of Cook County is a 464-bed public teaching hospital. The only subspecialty internal medicine ward service is the HIV service; all other internal medicine ward patients are cared for by the general medical service. There is a separate critical care team for intensive care unit patients. 636

Trick et al December 2006 637 We performed a prospective cohort study of all central venous catheters inserted into a general medical ward patient or catheters that were present at the time of admission to the general medical ward service during November 15, 2004, through March 31, 2005. We identified catheters through daily discussions among internal medicine house staff, chief residents, and on-call physicians. During these discussions, house staff reported whether they had inserted a central venous catheter or admitted a patient who had a catheter. Chief residents then interviewed the resident to determine the following: venous site of insertion, type of catheter, unit of insertion, and transferring service. The research team monitored catheters from the time of insertion by the medical service or admission to the ward service until 30 days after the first of the following 2 conditions: catheter removal or patient discharge. We monitored catheters after patient transfer to an ICU. The research team reviewed the medical record of each patient who had a central venous catheter and recorded the following patient or catheter characteristics: platelet count, international normalized ratio (INR), body mass index, receipt of prednisone, diagnosis of HIV or AIDS, Charlson comorbidity index, condition of the dressing, venous insertion site, type of catheter (ie,, triple lumen, double lumen, single lumen, or temporary dialysis). The research team monitored each patient for infectious complications. Medical record reviews were performed to evaluate potential bloodstream infections each time an organism was detected by blood culture or a new antibiotic was initiated. We categorized potential bloodstream infection episodes as either secondary (ie, related to infection at another site), primary laboratory confirmed, clinical sepsis, or contaminant using Centers for Disease Control and Prevention definitions. 15 Bedside evaluations were performed to verify presence of the catheter and evaluate the dressing condition. We recorded the exit site dressing condition as determined by visual inspection; the condition was categorized as adequate (ie, adherent and not soiled) or inadequate. Patients were monitored until hospital discharge or catheter removal. If there were any postdischarge emergency department visits or repeat hospitalizations, we reviewed the medical record to determine whether the visit was due to a catheter-related complication. We included temporary dialysis catheters, which could be inserted for temporary dialysis or other procedures, such as leukapheresis, plasmapheresis, or exchange transfusion. We excluded peripherally inserted central catheters (ie, PICCs). After completing data acquisition, we evaluated potential predictors of femoral vein insertion and performed a focus group session with a sample of internal medicine residents who had inserted a central venous catheter during the study period. We received institutional review board approval before study initiation. Statistics We calculated the incidence of primary laboratoryconfirmed bloodstream infection alone and also calculated the incidence including clinical sepsis. The incidence is expressed as the number of infections divided by the number of central venous catheter days (ie, catheter-days), standardized as a rate per 1000 catheter-days. We departed from the CDC surveillance methods in that, if a patient had 2 central venous catheters on the same day, 2 catheter-days were counted. Patients were eligible to have more than 1 infectious event if the episode was determined to be unrelated to the previous episode. We compared the infection incidence by venous insertion site (femoral vs subclavian or internal jugular), unit type (ward vs medical ICU vs emergency department), and catheter type (temporary dialysis vs other). Infections that occurred in a patient who had more than 1 central venous catheter were considered as an infectious episode for each type of catheter or venous insertion site, but only counted once for the overall incidence of infection. We compared incidence rates statistically using exact methods and report the 2-tailed P value. We calculated a device utilization rate (ie, number of catheter-days divided by number of patient-days) 3 100. For calculation of the device utilization rate, we only included catheter-days while the patient was on the general medical service. In a separate analysis, we evaluated whether certain patient or catheter characteristics were associated with the choice of the femoral vein as the venous insertion site compared with the subclavian or internal jugular veins. For patient characteristics, we used the following definitions: coagulopathy, INR. 1.5; thrombocytopenia, platelets,100 3 10 9 per liter; obesity 1, body mass index $30; obesity 2, body mass index $35. We compared the condition of the exit site dressing by patient location (ICU vs ward), venous insertion site (femoral vs subclavian or internal jugular), body mass index, and type of catheter (temporary dialysis vs other). We constructed multivariable logistic regression models to evaluate and adjust for potential confounders. In both bivariable and multivariable analyses, to control for lack of independence in dressing observations (ie, 1 patient could account for several dressing observations), we used the generalized estimating equation. Because there were missing observations of dressing condition, we compared the patient and catheter characteristics for observed versus unobserved dressings. Data were entered into a relational

638 Vol. 34 No. 10 Trick et al Table 1. Description of central venous catheter use among general medicine ward patients Variables, categorical database (Microsoft Access, Microsoft Inc., Redmond, WA), and analyses were performed using Stata software, version 8.2 (Stata Corporation, College Station, TX.). RESULTS Number (%), N 5 106 Insertion location CVC inserted on the general 54 (51) medical service ward CVC present at time of transfer to the general medical service Medical intensive care unit 25 (24) Emergency department 17 (16) Other intensive care unit 5 (5) Other facility or unknown 5 (5) Type of catheter Triple lumen 75 (71) Temporary dialysis catheter 28 (26) Single or double lumen 3 (2.8) Venous insertion site Femoral 70 (66) Internal jugular 25 (24) Subclavian 11 (10) Body mass index* (kg/m 2 ),30 69 (67) 30-34 24 (23) $35 10 (10) Laboratory Platelet count,100 k/mm 3 16 (15) Coagulopathy (INR. 1.5) 10 (9) CVC, Central venous catheter. *Body mass index data missing for 3 patients (n 5 103). There were 106 central venous catheters present in 91 ward patients during the study period: 54 (51%) were inserted while the patient was on the ward service, and 52 were transferred to the service with a catheter. These 106 catheters accounted for 682 line-days, 51 days (7.5%) were recorded after patient transfer to an ICU. There were 17,131 general medicine patientdays; therefore, the device utilization rate for ward patients was 4 catheter-days per 100 patient-days. The median duration of catheter insertion was 5 (interquartile [IQ] range, 4-8) days. The median age was 53 (IQ range, 44-63) years; 50 (55%) were female, and the median Charlson comorbidity score was 2. Most catheters had been inserted on the ward, followed by the medical ICU and the emergency department, and were either triple lumen or temporary dialysis catheters; the femoral vein was the most common venous insertion site; most patients did not have thrombocytopenia or coagulopathy; and most were not obese (Table 1). Table 2. Primary laboratory confirmed bloodstream infections among general medical ward patients who had a central venous catheter CVC characteristics Complications No. of BSIs No. of catheter-days BSIs per 1000 catheter-days (95% CI) Overall 3 682 4.4 (0.9-13) Insertion location General medical ward 0 381 0 (0-9.7) Medical intensive 0 161 0 (0-23) care unit Emergency department 2 85 24 (2.8-85)* Other locations y 1 55 18 (4.6-101) Venous insertion site z Femoral 3 443 6.8 (1.3-20) Internal jugular or 1 239 4.2 (0.1-23) subclavian Catheter type Temporary dialysis 1 199 5.0 (0.1-28) No temporary dialysis catheter 2 483 4.1 (0.5-15) BSI, Bloodstream infections; CVC, central venous catheter. *P 5.045 for comparison of bloodstream infection incidence between emergency department insertions and the aggregated incidence for other units (1.7 per 1000 catheter-days). y CVCs also inserted in the trauma, surgical, and cardiac ICUs and another hospital. The single BSI occurred in a patient who had the CVC inserted in the surgical ICU. z One episode occurred in a patient who had 2 CVCs. We counted the infection for each venous insertion site. There were 3 primary laboratory-confirmed bloodstream infections. All 3 infections occurred in patients who had a femoral venous catheter; however, 1 patient also had a subclavian catheter. Because we could not attribute the infection to either catheter, it was counted for each site (Table 2). Catheters inserted in the emergency department had the highest infection rate (Table 2). After excluding temporary dialysis catheters, the bloodstream infection incidence per 1000 patientdays remained higher for the emergency department compared with all other units, 32 vs 0, P 5.01. When we included the single episode of clinical sepsis, the overall bloodstream infection incidence was 5.9 per 1000 catheter-days (95% CI: 1.6-15). Alternatively, the risk of infection for a central venous catheter on our medicine ward service is approximately 4%. The episode of clinical sepsis occurred in a patient who had 2 catheters. One catheter was in the femoral vein for temporary dialysis, and the other was a triple lumen catheter inserted in the internal jugular vein; both catheters had been inserted by the internal medicine team. After inclusion of clinical sepsis in our analysis, the highest rate of infection still was for catheters inserted in the emergency department; however, compared with catheters inserted in other units, the difference was not statistically significant (P 5.09).

Trick et al December 2006 639 Table 3. Association between central venous catheter dressing condition and patient characteristics Characteristic (number of catheters*) Predicted probability y Adjusted odds ratio (95% CI) z P value Bivariable analyses Location Intensive care unit (n 5 6) 0% Reference category Outside intensive care unit (n 5 67) 32% Indeterminate.16 Body mass index (kg/m 2 ),30 (n 5 46) 19 Reference category $30 and,35 (n 5 17) 38 2.6 (1.0-7.0).052 $35 (n 5 8) 66 8.4 (2.2-32).002 Dialysis catheter No (n 5 54) 24 Reference category Yes (n 5 19) 43 2.4 (1.0-5.7).056 Femoral catheter No (n 5 27) 22 Reference category Yes (n 5 46) 34 1.8 (0.7-4.6).2 Multivariable analysis No dialysis catheter 27 Dialysis catheter 36 1.5 (0.6-3.9).36 BMI,30 21 BMI $30 47 3.4 (1.4-8.0).006 BMI, Body mass index. *Dressing condition was observed for 73 CVCs; patient BMI was not available for 2 catheters. y Probability of inadequate dressing condition. z Adjusted for multiple observations of the same patient (same catheter) over multiple days using the generalized estimating equation. Venous insertion site The femoral vein was the most common venous insertion site, regardless of patient location at the time of insertion. The percentage of catheters inserted in femoral veins was similar for the medical ward (68%), medical ICU (72%), and emergency department (67%). Physicians were more likely to insert catheters in the femoral vein if it was a temporary dialysis catheter, RR 5 1.5 (95% CI: 1.1-1.8; P 5.01); however, no patient characteristics (eg, coagulopathy, thrombocytopenia, or body mass index) influenced the venous insertion site. Exit site dressings The dressing condition was observed at least once for 73 (68%) of the central venous catheters and 206 (30%) of the 682 catheter-days. Dressings were more likely to be inadequate among general medical ward patients (65/190; 34%) compared with intensive care unit patients (0/16; 0%); however, after adjustment for repeat observations of dressings for the same catheter, this difference was not statistically significant (Table 3). Also, we found that obesity was associated with worse dressing conditions and that there was a dose-response effect, ie, the patients who had the highest body mass index were most likely to have a dressing in poor condition (Table 3). Furthermore, although not statistically significant, temporary dialysis catheters were less likely to have a dressing in good condition. By multivariable analysis, high body mass index remained an independent predictor for poor dressing condition (Table 3). Observations of dressing condition occurred less often among patients who were transferred to the medicine service or who had a shorter median duration of catheter insertion (6.5 vs 3 days). Missing dressing observations were not associated with venous insertion site, type of catheter, severity of illness, obesity, coagulopathy, or thrombocytopenia (data not shown). DISCUSSION We evaluated central venous catheter use among patients hospitalized on a general medical service. These patients often had their catheter inserted while on a unit other than the medical ward, most commonly the medical ICU or emergency department. All primary laboratory-confirmed bloodstream infections occurred among patients who had their catheter inserted on a unit other than the general medical ward. In particular, catheters inserted in the emergency department had the highest infection rate. Also, the femoral vein was the preferred venous insertion site; this choice was associated only with placement of a temporary dialysis catheter, not risk of bleeding, unit location at the time of insertion, or body mass index. Furthermore, catheter dressings were in better condition in the intensive care unit and among patients who were not obese. We found a primary bloodstream infection rate (6 per 1000 catheter-days) similar to the mean rate reported among medical intensive care unit patients

640 Vol. 34 No. 10 Trick et al (5 per 1000 catheter-days). 16 Our data suggest that, after standardizing bloodstream infection rates by the number of catheter-days, bloodstream infection rates are similar among ICU and non-icu patients; however, given the small number of bloodstream infections and catheter-days in our study, our estimate should be interpreted with caution. Despite observing few primary bloodstream infections, the infection rate was greatest among catheters inserted in the emergency department. This finding supports the Healthcare Infection Control Practices Advisory Committee Guidelines, which recommend that emergently placed catheters be removed as soon as possible. 17 Although there is subjectivity in the definition of emergently placed, placement of a central venous catheter in the emergency department may indicate poor adherence to recommended insertion practices. The central venous catheter utilization rate that we report (4 catheter-days per 100 patient-days) is dramatically less than the mean rate reported by medical ICUs to the Centers for Disease Control and Prevention s National Nosocomial Infection Surveillance (NNIS) system (50 catheter-days per 100 patient-days). 11 Because it is possible that we failed to identify every central venous catheter, the device utilization rate that we report should be considered the lower bound for our patient population. An alternative method of estimating the device utilization rate outside of ICUs would be to perform serial cross-sectional surveys, which would be simpler and, because of the difficulty in detecting catheters over a sustained period of time, possibly more accurate. In our hospital, physicians were most likely to choose the femoral vein for catheter insertion. We expected to find that certain patient characteristics would alter the choice of venous insertion site (eg, coagulopathy, patient care unit at the time of insertion, or thrombocytopenia) but found only that physicians were more likely to choose the femoral vein when placing a temporary dialysis catheter. In our focus group session, house staff indicated their desire to preserve the subclavian and internal jugular veins for placement of long-term cuffed intravascular dialysis catheters. We found that all patients who had a primary bloodstream infection had a femoral central venous catheter; 1 patient also had a subclavian catheter, and we could not attribute the infection to 1 catheter. Since we observed few bloodstream infections and because we performed an observational study, we could not provide a valid comparison of bloodstream infection rates between venous insertion sites. Previous studies suggest that use of the subclavian vein for catheterization reduces the bloodstream infection rate. 4,18 Improving postinsertion exit site care has been a component of several successful bloodstream infection prevention programs, 5-7,10 and improved dressing care has improved infection rates for other invasive devices. 19 Similar to our findings from a cross-sectional survey of central venous catheter use, 14 exit site dressings were in good condition in the ICU but often in poor condition outside the ICU. Also, inadequate dressings were more likely among obese patients, and there was evidence of a dose-response effect, ie, as the body mass index increased, dressing care worsened. Improved dressing care for general medical ward patients is amenable to intervention and should reduce patients risk of infection. Our study has several limitations. We detected few bloodstream infections. Despite the small number of events, we found a significantly higher primary laboratory-confirmed bloodstream infection rate for catheters that were inserted in the emergency department; however, the small number of events precludes adjustment for potential confounders. In addition, although dressing condition was not observed for many patients, the only differences between observed and unobserved patients were the duration of catheter insertion and whether the patient had been transferred from another service. We believe it is unlikely that the association between poor dressing condition and obesity or non-icu patient location were biased by these missing data. Interventions that have successfully reduced central venous catheter associated infections among ICU patients should be exported and adapted for general medical ward patients. In particular, postinsertion dressing care should be improved, and physicians should consider removing or replacing emergently placed central venous catheters. We believe that the dramatic successes reported for ICU patients can be reproduced for the non-icu patient population. The challenge will be in devising valid, sustainable surveillance strategies to monitor central venous catheter use and complication rates outside the ICU and subsequently evaluate the effectiveness of infection prevention activities. The authors thank Kathleen Murray for data collection form development and data collection; Donald Blom for assistance with bloodstream infection determinations; Laura Sadowski for leading the focus group session; Oksana Barilyak, Anand Despande, and Saurabh Sharma for assistance with data collection; and the following chief residents for interviewing residents regarding catheter insertions: Roni Ghaoui, Sean Halleran, Priya Kansal, and Sunita Nathan. References 1. McGee DC, Gould MK. Preventing complications of central venous catheterization. N Engl J Med 2003;348:1123-33. 2. Renaud B, Brun-Buisson C. Outcomes of primary and catheter-related bacteremia: a cohort and case-control study in critically ill patients. Am J Respir Crit Care Med 2001;163:1584-90. 3. Pittet D, Tarara D, Wenzel RP. Nosocomial bloodstream infection in critically ill patients: excess length of stay, extra costs, and attributable mortality. JAMA 1994;271:1598-601. 4. Merrer J, De Jonghe B, Golliot F, Lefrant JY, Raffy B, Barre E, et al. Complications of femoral and subclavian venous catheterization in

Trick et al December 2006 641 critically ill patients: a randomized controlled trial. JAMA 2001;286: 700-7. 5. Sherertz RJ, Ely EW, Westbrook DM, Gledhill KS, Streed SA, Kiger B, et al. Education of physicians-in-training can decrease the risk for vascular catheter infection. Ann Intern Med 2000;132:641-8. 6. Eggimann P, Harbarth S, Constantin MN, Touveneau S, Chevrolet JC, Pittet D. Impact of a prevention strategy targeted at vascular-access care on incidence of infections acquired in intensive care. Lancet 2000;355:1864-8. 7. Coopersmith CM, Rebmann TL, Zack JE, Ward MR, Corcoran RM, Schallom ME, et al. Effect of an education program on decreasing catheter-related bloodstream infections in the surgical intensive care unit. Crit Care Med 2002;30:59-64. 8. Bijma R, Girbes AR, Kleijer DJ, Zwaveling JH. Preventing central venous catheter-related infection in a surgical intensive-care unit. Infect Control Hosp Epidemiol 1999;20:618-20. 9. Warren DK, Zack JE, Mayfield JL, Chen A, Prentice D, Fraser VJ, et al. The effect of an education program on the incidence of central venous catheter-associated bloodstream infection in a medical ICU. Chest 2004;126:1612-8. 10. Warren DK, Zack JE, Cox MJ, Cohen MM, Fraser VJ. An educational intervention to prevent catheter-associated bloodstream infections in a nonteaching, community medical center. Crit Care Med 2003;31: 1959-63. 11. Richards MJ, Edwards JR, Culver DH, Gaynes RP, the National Nosocomial Infections Surveillance System. Nosocomial infections in combined medical-surgical intensive care units in the United States. Infect Control Hosp Epidemiol 2000;21:510-5. 12. Braun BI, Kritchevsky SB, Wong ES, Solomon SL, Steele L, Richards CL, et al. Preventing central venous catheter-associated primary bloodstream infections: characteristics of practices among hospitals participating in the Evaluation of Processes and Indicators in Infection Control (EPIC) study. Infect Control Hosp Epidemiol 2003;24:926-35. 13. Climo M, Diekema D, Warren DK, Herwaldt LA, Perl TM, Peterson L, et al. Prevalence of the use of central venous access devices within and outside of the intensive care unit: results of a survey among hospitals in the prevention epicenter program of the Centers for Disease Control and Prevention. Infect Control Hosp Epidemiol 2003;24:942-5. 14. Trick WE, Vernon MO, Welbel SF, Wisniewski MF, Jernigan JA, Weinstein RA. Unnecessary use of central venous catheters: the need to look outside the intensive care unit. Infect Control Hosp Epidemiol 2004; 25:266-8. 15. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988;16: 128-40. 16. National Nosocomial Infections Surveillance System. National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2004, issued October 2004. Am J Infect Control 2004;32:470-85. 17. O Grady NP, Alexander M, Dellinger EP, Gerberding JL, Maki DG, McCormick RD, et al. Guidelines for the prevention of intravascular catheter-related infections. MMWR 2002;1(RR10):1-26. 18. Mermel LA, McCormick RD, Springman SR, Maki DG. The pathogenesis and epidemiology of catheter-related infection with pulmonary artery Swan-Ganz catheters: a prospective study utilizing molecular subtyping. Am J Med 1991;91:197S-205S. 19. Trick WE, Kioski CM, Howard KM, Cage GD, Tokars JI, Yen BM, et al. Outbreak of Pseudomonas aeruginosa ventriculitis among patients in a neurosurgical intensive care unit. Infect Control Hosp Epidemiol 2000;21:204-8.