- Email: [email protected]
Failure of central venous catheter insertion and care bundles in a high central line–associated bloodstream infection rate, high bed occupancy hospital
ARTICLE IN PRESS American Journal of Infection Control 000 (2019) 1−7
Contents lists available at ScienceDirect
American Journal of Infection Control journal homepage: www.ajicjournal.org
Major Article
Failure of central venous catheter insertion and care bundles in a high central line−associated bloodstream infection rate, high bed occupancy hospital Amalia Karapanou RN, MSc a, Anna-Maria Vieru MD b, Michail A. Sampanis RN, MSc a, Angeliki Pantazatou MD c, Ioannis Deliolanis MD c, George L. Daikos MD b, Michael Samarkos MD b,* a
Infection Control Committee, Laikon Hospital, Athens, Greece First Department of Medicine, Medical School, National and Kapodistrian University of Athens, Athens, Greece c Microbiology Laboratory, Laikon Hospital, Athens, Greece b
Key Words: Patient care bundles Central venous catheters Bacteremia Catheter-related infections Nursing staff Patient safety
Background: Our hospital has several characteristics different from the settings in which the central venous catheter (CVC) care bundle has been implemented so far, that is, care bundles or protocols are not systematically used, and the prevalence of central line−associated bloodstream infections (CLABSI) is high, as is bed occupancy rate. We examined the effectiveness of CVC care bundles. Methods: Modified CVC bundles were implemented across all settings of our hospital. During both phases of the study, we collected data on CLABSI, and we monitored CVC insertion and management practices with direct observation audits. Results: We have studied 913 CVC insertions (454 in PRE and 459 in POST) for 11,871 catheter-days. The incidence of CLABSI was 8.3 per 1,000 catheter-days PRE, and 7.6 per 1,000 catheter-days POST (incidence rate ratio, 0.92; 95% confidence interval, 0.60-1.40). Compliance with the CVC insertion bundle increased from 8.4%-74.3% (P < .0001). The CVC management bundle compliance also increased from 11.4%-57.7% (P < .0001). Conclusions: Despite improved compliance after the intervention, implementation of a modified CVC bundle failed to decrease CLABSI incidence. Higher bundle compliance rates may be necessary for a significant decrease in the incidence of CLABSI, along with the appropriate organizational culture and levels of staffing. © 2019 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
Central line−associated bloodstream infections (CLABSI) represent a major challenge to patient safety and a significant burden to health systems. More than 20% of bloodstream infections in acute care hospitals in Europe are CLABSI.1 These infections carry an attributable mortality of up to 30%,2 and they are probably the most costly health care−associated infections.3 It has been estimated that up to 65%-70% of CLABSIs are preventable,4,5 and several factors, such as catheter insertion site, type of catheter, and type of dressing are important for their prevention.6 Care bundles are small sets—usually 3 to 5—of evidence-based interventions that, when implemented collectively and reliably, improve the delivery of health care and improve patient outcomes.7 Care bundles for central venous catheter (CVC) insertion and management *Address correspondence to Michael Samarkos, MD, First Department of Medicine, Medical School, National and Kapodistrian University of Athens, Laikon Hospital, 17 Agiou Thoma St, Athens, Greece 11527. E-mail address: [email protected] (M. Samarkos). Conflicts of interest: None to report.
have decreased the incidence rate of CLABSIs significantly.8 However, the effectiveness of these interventions might not be the same across all settings. As Speroff et al9 remarked, incorporating improvement methods is complex, highly context-dependent, and may take long. Thus there might be differences depending on the level of care (ie, intensive care unit [ICU] vs patient wards), on the experience in implementation and use of care bundles, as well as on differences in safety culture.10 Our hospital has several characteristics that differ from the setting in which the CVC care bundle has been implemented so far. First, care bundles or protocols are not systematically used in patient care. Second, the prevalence of CLABSI in the European Centre for Disease Prevention and Control point prevalence survey of 2012 was high, as bloodstream infections accounted for 45.4% of all hospital-acquired infections, and 50% of which were catheter-related (hospital data). Third, the bed occupancy rate of our hospital is very high, with medical wards systematically registering occupancy rates well over 100%. Fourth, our hospital is understaffed in registered nurses.11 Finally,
https://doi.org/10.1016/j.ajic.2019.11.018 0196-6553/© 2019 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
ARTICLE IN PRESS 2
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
chlorhexidine gluconate 2% is not available in Greece, and transparent semipermeable dressings are in short supply in our hospital. Consequently, the aim of our study was to evaluate the effectiveness of a modified CVC bundle on the incidence of CLABSI in a setting lacking in safety culture, with high CLABSI incidence, high bed occupancy rate, and relatively limited resources. METHODS Setting We have conducted a quasiexperimental pre−post study at Laikon Hospital, a tertiary care, university-affiliated hospital in Athens, with 574 beds that includes hematology−oncology wards and both solid organ and bone marrow transplantation units. All wards (medical, surgical, and ICU), except the emergency department, were included in the study. Procedures The study consisted of 3 phases: preintervention (PRE, January 1 to June 30, 2016), intervention (INT, July 1 to September 30, 2016), and postintervention (POST, October 1, 2016 to March 31, 2017). During the preintervention phase, we collected baseline data on CLABSI. For the CLABSI surveillance, we used the Centers for Disease Control and Prevention/National Healthcare Safety Network definition.12 In brief, CLABSI is defined as a laboratory-confirmed bloodstream infection not related to an infection at another site that develops within 48 hours of a central line placement. If >1 CLABSI occurred on the same catheter, we recorded only the initial one. Data collected included dates of catheter insertion and removal, site of catheterization, number of lumens, urgency of insertion, and setting were the insertion occurred. We have also collected data on whether the index catheterization was the initial one, or it was performed to replace another CVC. Finally, we also collected data on patient demographics, ward, admission diagnosis, presence of comorbidities, interventions such as mechanical ventilation and parenteral nutrition, pathogens isolated from blood cultures, and patient outcomes. The intervention phase consisted of the implementation of the CVC insertion and maintenance care bundles. The bundles were based on the ones issued by the Greek Centre for Control and Prevention of Diseases (KEELPNO) with some modifications.13 The elements of the insertion bundle included hand hygiene, maximal barrier precautions (total body drape and personal protective equipment), aseptic technique throughout the insertion, and use of transparent dressing. The management bundle consisted of hand hygiene, hub antisepsis with alcohol-based handrub, and aseptic technique during the procedure. In both cases, bundle undertaking required compliance with all bundle elements. All wards prepared and stocked CVC insertion kits, which included all materials required by the bundle, as well as the CVC insertion checklist. We did not use antimicrobial-coated catheters. The insertion bundle, however, lacked 2 elements. The assisting nurse did not have the option to interrupt the CVC insertion if the bundle protocol was violated. Thus violations were not recorded. In addition, a checklist included in the CVC insertion kit served as a reminder and it was not to be filled for each insertion. The management bundle did not include daily evaluation of ongoing need of the CVC. The intervention included training of all health care workers involved in the insertion and maintenance of CVCs, with small group teaching sessions with 2 of the researchers acting as facilitators (A.K. and M.S.). During the intervention phase, we did not collect any data. During the postintervention phase, we collected the same data as with the preintervention phase. We also audited CVC insertion and management practices, with direct observation in all phases of the study. Bundle violations were noted in the audit. The audit forms
were based on the respective checklists that we were planning to implement. Throughout the study we were monitoring hand hygiene compliance monthly. Outcomes The main outcome was CLABSI incidence expressed as CLABSI events per 1,000 catheter-days. Secondary outcomes included days from CVC insertion to CLABSI, duration of catheterization, catheter use rate, and in-hospital mortality. Statistical analysis We compared the PRE with the POST phase. During the INT phase, we did not collect any data, therefore results of the POST phase do not include the INT phase. Continuous variables were expressed as mean and standard deviation or median interquartile range, whereas qualitative variables were expressed as absolute and relative frequencies. We used the x2 test (or the Fisher exact test when appropriate) to compare dichotomous variables, and the Mann−Whitney U test to compare continuous variables. A Poisson distribution was used to compare the incidence rate of CLABSI among different groups, with the calculation of incidence rate ratios (IRR). For the comparison of proportions, the x2 and the Fisher exact tests were used. A stepwise multiple logistic regression analysis (P for removal was set at .1, and P for entry was set at .05) was performed to identify factors independently associated with CLABSI. For time to event variables we performed survival analysis with Kaplan−Meier curves, and Cox regression. All P values reported are 2-tailed. Statistical significance was set at .05 and analyses were conducted using IBM SPSS Statistics version 24.0 (IBM Corporation, Armonk, NY). The study protocol was approved by the institutional review board (approval number 3486/10-3-2015), and permission to create the study database was granted by the Greek Personal Data Protection Authority (decision number 1605-2/18-11-2015). RESULTS Patient characteristics We recorded in total 913 CVC insertions (454, 49.7% in PRE and 459, 50.3% in POST) in 713 patients. The catheters remained in place for 11,871 catheter-days. Mean age in the total patient population was 66.7 § 15.4 years, however, patients in the POST phase were significantly older than those in the PRE phase (68.2 § 14.6 years vs 65.2 § 16.1 years; Mann−Whitney U test, P = .033). The prevalence of major comorbidities in our patients’ population was 22.3% for diabetes mellitus, 8.0% for chronic obstructive pulmonary disease, 5.3% for congestive heart failure, 33.9% for coronary heart disease, and 25.8% for hematologic neoplasms. The prevalence of congestive heart failure and coronary heart disease were significantly higher in the POST phase (3.6% vs 7.1%, P = .039 and 30.3% vs 37.1%, P = .037, respectively). There was a difference in the PRE and POST intervention prevalence of prior admission to hospital (68.1% vs 76.5%; P = .012) and dialysis (15.8% vs 10.8%; P = .046). The most frequently isolated pathogen in the PRE phase was Acinetobacter baumannii (10 of 50, 20.0%), followed by coagulase-negative staphylococci (8 of 50, 16.0%), and Klebsiella pneumoniae (6 of 50, 12.0%). In the POST phase, A baumannii was again the most frequently isolated pathogen (11 of 45, 24.4%), followed by Enterococcus faecium (10 of 45, 22.2%), and coagulase-negative staphylococci (9 of 45, 20.0%). There was no difference in the proportion of gram-positive pathogens isolated in the PRE and POST phase (21 of 50, 42.0% vs 24 of 45, 53.3%; P = 0.3).
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
3
Table 1 Catheter insertion−related variables before and after the intervention
Ward
Site of catheterization
Side of catheterization Number of lumens Urgency of insertion Timing of CVC insertion
Insertion setting
Medical Surgical ICU Femoral* Jugular* Subclavian Right* Left 2 3 Elective* Emergent Initial Catheter malfunction Catheter infection Ward Operating room Dialysis unit Emergency room
PRE, n = 454 n, (%)
POST, n = 459 n, (%)
Total, n = 913 n, (%)
248 (54.6) 125 (27.5) 81 (17.8) 194 (42.7) 150 (33.0) 110 (24.2) 316 (69.6) 138 (30.4) 323 (71.1) 131 (28.9) 367 (80.8) 87 (19.2) 392 (86.3) 39 (8.6) 23 (5.1) 341 (75.1) 101 (22.2) 9 (2.0) 3 (0.7)
226 (49.2) 145 (31.6) 88 (19.2) 76 (16.6) 255 (55.6) 128 (27.9) 394 (85.8) 65 (14.2) 336 (73.2) 123 (26.8) 410 (89.3) 49 (10.7) 409 (89.1) 27 (5.9) 23 (5.0) 320 (69.7) 109 (23.7) 26 (5.7) 4 (0.9)
474 (51.9) 270 (29.6) 169 (18.5) 270 (29.6) 405 (44.4) 238 (26.1) 710 (77.8) 203 (22.2) 659 (72.2) 254 (27.80) 777 (85.1) 136 (14.9) 801 (87.7) 66 (7.2) 46 (5.0) 661 (72.4) 200 (23.0) 35 (3.8) 7 (0.8)
Statistical significance testing for the difference in proportions between PRE and POST. CVC, central venous catheter; ICU, intensive care unit; PRE, preintervention; POST, postintervention. *P <.001.
Catheterization-related variables Table 1 shows variables related with the CVC insertion in the 2 phases of the study and in the total study population. The femoral site insertion decreased significantly from 42.7% (194 of 454) in the PRE to 16.6% (76 of 459) in the POST period (x2, P < .001), whereas jugular site insertion increased significantly (33.0%, 150 of 454 to 55.6%, 255 of 459; P < .001). The urgency of insertion (emergent vs elective) also differed, with significant decrease in emergent CVC insertions from 19.2% (87 of 454) to 10.7% (49 of 459) in the POST phase (x2, P < .001). Outcomes The CLABSI incidence decreased from 8.3 (50 CLABSIs in 5,998 catheter-days) PRE to 7.6 (45 CLABSIs in 5,873 catheter-days) CLABSIs per 1,000 catheter-days POST intervention. The decrease, however, was not significant (IRR, 0.92; 95% confidence interval [CI], 0.601.40). Similarly, the rate of CLABSI was not significantly decreased in subgroups such as medical patients (IRR, 0.64; 95% CI, 0.34-1.17; P = .12), patients with CVC at the subclavian site (IRR, 0.54; 95% CI, 0.19-1.37; P = .15), and patients in which the CVC was inserted in the ward (IRR, 0.87; 95% CI, 0.54-1.38; P = .53). The median time from CVC insertion to CLABSI in the 2 study phases was 12.0 days (95% Cl, 10.0-15.0) for the PRE and 10.0 days (95% CI, 8.0-12.0) for the POST. We have performed survival analyses (Kaplan−Meier curves) to compare the time from CVC insertion to CLABSI between the 2 study phases, however, the difference was not significant (Fig 1, log-rank test x2 = 0.1461, P = .7). Catheter use rate did not differ between the PRE and the POST phase of the study (7.5 per 100 patient-days vs 7.8 per 100 patientdays; IRR, 1.032; exact 95% CI, 0.90-1.17; x2, P = .62). The overall in-hospital mortality was 27.1% (193 of 713) and did not differ between the 2 phases of the study (PRE 24.4%, 86 of 353 vs POST 29.7%, 107 of 360; x2, P = .11). Patients with CLABSI, as expected, had significantly higher mortality (41.2% [35 of 85] vs 25.6% [161 of 628]; Fisher exact test, P = .004).
that it was associated with hospitalization in the ICU (x2, P = .002), with insertion after previous CVC infection (x2, P < .001), and with insertion of CVCs with 3 versus 2 lumens (Fisher exact test, P = .002), whereas insertion in the operating room was associated with significantly lower probability of CLABSIs (x2, P = .031). When we investigated CLABSI rate in relation to possible risk factors (Table 2), we found that CLABSI rate was significantly higher in the ICU (IRR, 2.04; P = .02), and with insertion after previous CVC infection (IRR, 2.65; P = .0027). In addition, duration of catheterization was significantly longer in patients with CLABSI vs patients with no CLABSI (median 16.0 [95% CI, 13.0-19.0] vs 10.0 [95% CI, 10.0-11.0] days; Mann−Whitney U test, P < .001), and so was length of hospital stay (median 21 [95% CI, 203.0-23.0] vs 34 days [95% CI, 30.0-39.0]; Mann−Whitney U test, P < .001). Monthly bed occupancy rates during the study ranged from 49.5%-118.5% (median 83.2%), thus we examined the association of monthly bed occupancy rate of the department in which the patient was hospitalized when the CVC was inserted, with CLABSI. However, there was no difference in bed occupancy rate between patients with and without CLABSI (Mann−Whitney U test, P = .105). To investigate for risk factors, we performed logistic regression in which the outcome was CLABSI and the independent variables were ward, indication for insertion, insertion setting, number of CVC lumens, study phase, and duration of catheterization (Table 3). Only CVC insertion because of previous CLABSI was independently associated with CLABSI (odds ratio, 2.6; P = .017). Duration of catheterization was also statistically significantly associated with CLABSI (P = .001). An odds ratio of 1.03 per extra day with a CVC could be relevant when the catheter remains in place for a long time. We have also performed Cox regression to examine the effect of different variables (eg, ward, number of lumens, and others) to the time from CVC insertion to CLABSI. Only CVC change because of previous CLABSI was independently associated with time to CLABSI (hazard ratio, 2.4 [95% CI, 1.2-4.7]; P = .014).
Bundle compliance Risk factors for CLABSI We examined several variables as risk factor for CLABSI (Table 2). When CLABSI was examined as a dichotomous variable, we found
During the intervention phase, small group teaching sessions regarding the insertion and management of CVC were attended by 139 of 201 nurses (69.1%) and 82 of 93 physicians (88.2%).
ARTICLE IN PRESS 4
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
Fig 1. Kaplan−Meier curve showing the probability of CLABSI according to the duration of catheterization in the 2 study phases. CLABSI, central line−associated bloodstream infection; PRE, preintervention; POST, postintervention.
Compliance with the CVC insertion bundle (n = 153 observations) increased significantly from 8.4% (7 of 83) in the PRE phase to 74.3% (52 of 70) in the POST phase (x2, P < .0001). In the PRE phase, the compliance was lowest for use of maximal barrier precautions (9.6%, 8 of 83) and hand hygiene (27.7%, 23 of 83), whereas in the POST phase, compliance was lowest for maximal barrier precautions (87.1%, 61 of 70) and for the use of transparent dressing (74.3%, 52 of 70). The CVC management bundle compliance (n = 295 observations) also increased from 11.4% (15 of 132) in the PRE phase to 57.7% (94 of 163) in the POST phase (x2, P < .0001). The bundle element with the lowest compliance in the PRE phase was hand hygiene before managing the CVC (31.3%, 42 of 132) and in the POST phase was use of sterile gloves (63.2%, 103 of 163). Details regarding compliance with the individual components of the bundles in the 2 phases of the study are presented in Fig 2. DISCUSSION Implementation of the CVC care bundle in ICU and non-ICU patients reduced CLABSI rates from 8.3 to 7.6 CLABSI per 1,000 catheter-days, however, the reduction was not significant. However, we reduced significantly catheterizations of the femoral vein from 42.7%-16.6%, the insertion bundle compliance increased from 8.4%74.3%, and the management bundle from 11.4%-57.5%. In our analysis of risk factors for CLABSI we could only find that ICU hospitalization, use of 3 versus 2 lumen catheters, previous CLABSI, and duration of catheterization were associated with CLABSI in univariate analyses, however, only previous CLABSI and duration of catheterization were
independently associated with CLABSI in multivariate analysis. We could not associate compliance with the care bundles or with any of their elements with patient outcomes because our study was designed as an ecological one. Compliance was estimated by audit, thus we lacked data on the bundle compliance for each patient as it has been done in other studies.14 Although we have hypothesized that bed occupancy rate would influence the CLABSI rate, we could not establish such an association. In contrast to our study, CVC care bundles have generally succeeded in significantly reducing CLABSI rates.8 There are several possible contributing factors to the failure of our care bundles. Our setting had some unique characteristics. The baseline incidence was a relatively high CLABSI incidence for a mixed setting, however, there have been reported even higher baseline CLABSI rates in the literature.15 The bundle was a modification of the ones widely used.13,16 An important difference was that the assisting nurse was not given the option to interrupt the insertion of the CVC when the bundle was violated because of the reluctance of our nurses to challenge physicians. There is a rather strong hierarchical structure in Greek hospitals, with physicians occupying the higher levels. It is apparent that implementation of care bundles has certain requirements in terms of organizational culture, which unfortunately, we lacked. Furthermore, a checklist served as a reminder only, and it did not have to be filled in each CVC insertion, as the nursing staff suggested that it would add to their already increased workload. In our hospital, there are on average 3 registered nurse to 40 patients in the wards during the morning shift and 2 during the afternoon and night shift (personal observation). In general, the nurse−patient ratio in Greece is the
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
5
Table 2 Univariate risk factors analysis for CLABSI Variable Ward
Site of catheterization
Number of lumens Urgency of insertion Indication for CVC insertion
Insertion setting
Prior admission Infection on admission Surgery Antimicrobial administration Chemotherapy Immunosuppression Mechanical ventilation Parenteral nutrition Nasogastric tube Urinary catheter Dialysis
CLABSI (n, %) Medical Surgical* ICU Femoral Jugular Subclavian* 2 3 Elective Emergent Initial* Catheter malfunction Catheter infection Operating room*,k Ward Dialysis unit ER No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes
51 (10.8) 16 (5.9) 28 (16.6)y 27 (10.0) 46 (11.4) 22 (9.2) 55 (8.3) 40 (15.7)z 85 (10.9) 10 (7.4) 74 (9.2) 8 (12.1) 13 (28.3)x 10 (4.8) 80 (12.1) 5 (14.3) 0.0 22 (9.7) 73 (10.6) 7 (8.3) 88 (10.6) 61 (11.2) 34 (9.3) 14 (12.8) 81 (10.1) 79 (10.9) 16 (8.4) 70 (11.0) 25 (9.0) 68 (10.4) 27 (10.3) 78 (10.2) 17 (11.4) 52 (10.3) 43 (10.6) 22 (9.1) 73 (10.9) 81 (10.3) 14 (10.9)
CLABSI/1,000 catheter-days
IRR (95% CI), P value
7.73 5.63 11.51 7.64 9.05 6.76 7.30 9.22 8.02 7.87 7.22 8.46 19.17 4.73 8.76 9.16 0.00 8.78 7.79 5.56 8.29 8.06 7.90 10.57 7.68 8.37 6.57 8.42 7.02 8.06 7.86 7.92 8.42 7.93 8.10 7.17 8.29 7.88 8.82
1.38 (0.77-2.57), .28 N/A 2.04 (1.06-4.04), .02 1.13 (0.62-2.08), .77 1.34 (0.78-2.33), .32 N/A 1.26 (0.90-1.72), .15 0.98 (0.45-1.89), .99 N/A 1.17 (0.49-2.43), .69 2.65 (1.35-4.82), .0027 N/A 1.85 (0.96-4.0), .07 1.93 (0.51-6.21), .20 N/A 0.89 (0.54-1.50), .6 1.49 (0.69-3.81), .4 0.98 (0.62-1.51), .9 0.73 (0.41-1.39), .2 0.78 (0.43-1.35), .4 0.83 (0.50-1.33), .5 0.97 (0.60-1.54), .9 1.06 (0.59-1.81), .7 1.02 (0.66-1.56), .9 1.16 (0.71-1.96), .6 1.12 (0.58-1.99), .6
Statistical significance testing for the difference in proportions between number of CLABSI and CLABSI. CI, confidence interval; CLABSI, central line−associated bloodstream infection; CVC, central venous catheter; ER, emergency room; ICU, intensive care unit; IRR, incidence rate ratios; N/A, not applicable. *Baseline category for rate comparison. y P = .002. z P = .002. x P <.001. k P = .031.
Table 3 Multivariable analysis (logistic regression) of risk factors for central line−associated bloodstream infection Variable
Odds ratio (95% CI)
P
Study phase Insertion because of catheter malfunction Insertion because of catheter infection Three-lumen catheter Duration of catheterization Insertion in the ward Insertion in the dialysis unit ICU
0.90 (0.58-1.40) 0.99 (0.43-2.27) 2.60 (1.19-5.72) 1.28 (0.79-2.08) 1.04 (1.02-1.06) 1.71 (0.77-3.81) 2.14 (0.63-7.21) 1.31 (0.73-2.34)
.636 .976 .017 .322 .001 .190 .221 .370
CI, confidence interval; ICU, intensive care unit.
lowest in the European Union (344 per 100,000 population in 2014 compared with an European Union average of 864).11 Although bundle compliance in our hospital increased after the intervention, it was still relatively low. In the systematic review of CVC care bundles in critically ill patients by Ista et al,8 the majority of the included studies reported a preintervention bundle compliance that was at least comparable to our POST phase levels. There is not an established critical
compliance threshold above which bundles are effective, however, there are such data for other infection control interventions such as hand hygiene.17 Importantly, compliance with the maintenance bundle in the POST phase was <60%. This might be critical, given that maintenance bundles should be performed several times daily, whereas the insertion bundle must be performed only once. Thus noncompliance with the maintenance bundle may have a cumulative effect larger than that of noncompliance with the insertion bundle. We assessed compliance rates by a direct observation, noncovert audit. It is possible that our audit results do not reflect actual practice. First, we have observed only 153 of the 913 CVC insertions (16.8%), and second because of the so-called Hawthorne effect. There are studies on hand hygiene compliance suggesting that the Hawthorne effect might lead to significantly inflated compliance rates and this might be the case in our study too.18 In addition, the violations of the insertion bundle were not recorded, except in the context of the compliance audit (discussed earlier). This is an important limitation in our study, as we cannot fully explain our findings. Our study had relatively short duration, so that there was not adequate time for the intervention to be incorporated into everyday practice, as this might take long.15,19
ARTICLE IN PRESS 6
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
Fig 2. Compliance with individual elements and overall bundle compliance for the CVC insertion (upper panel) and the CVC maintenance bundle (lower panel) before (PRE) and after (POST) the intervention. CVC, central venous catheter; PPE, personal protective equipment.
Understaffing may have played a role in our results. We do not have accurate data regarding the nurse staffing per department during our study, however, we have already discussed the influence of understaffing on the study procedures, and in addition there is evidence that the level of nurse staffing may affect CLABSI rates in surgical ICUs.20 CONCLUSIONS In our setting, which is characterized by high CLABSI rates, high bed occupancy rates, understaffing, and a hierarchical organizational culture, implementation of CVC care bundles failed to
significantly reduce CLABSI rates. We suggest that care bundles are more than checklists of actions, and their success is, at least in part, context-dependent. Requirements in terms of organizational culture and staffing are probably critical to successful care bundle implementation. References 1. European Centre for Disease Prevention and Control. Point prevalence survey of healthcare-associated infections and antimicrobial use in European acute care hospitals 2011-12. Stockholm: ECDC; 2013.
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7 2. Ziegler MJ, Pellegrini DC, Safdar N. Attributable mortality of central line associated bloodstream infection: systematic review and meta-analysis. Infection 2015;43:29-36. 3. Zimlichman E, Henderson D, Tamir O, Franz C, Song P, Yamin CK, et al. Health careassociated infections: a meta-analysis of costs and financial impact on the US health care system. JAMA Intern Med 2013;173:2039-46. 4. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ. Estimating the proportion of healthcare-associated infections that are reasonably preventable and the related mortality and costs. Infect Control Hosp Epidemiol 2011;32:101-14. 5. Making health care safer: reducing bloodstream infections. CDC Vital Signs 2011, Available from: https://www.cdc.gov/vitalsigns/pdf/2011-03-vitalsigns.pdf. Accessed April 20, 2019. 6. Bell T, O’Grady NP. Prevention of central line-associated bloodstream infections. Infect Dis Clin North Am 2017;31:551-9. TagedP 7. Institute for Healthcare Improvement. What is a bundle? Available from: http:// www.ihi.org/knowledge/Pages/ImprovementStories/WhatIsaBundle.aspx. Accessed April 20, 2019. 8. Ista E, van der Hoven B, Kornelisse RF, van der Starre C, Vos MC, Boersma E, et al. Effectiveness of insertion and maintenance bundles to prevent central-line-associated bloodstream infections in critically ill patients of all ages: a systematic review and meta-analysis. Lancet Infect Dis 2016;16:724-34. 9. Speroff T, Ely EW, Greevy R, Weinger MB, Talbot TR, Wall RJ, et al. Quality improvement projects targeting health care-associated infections: comparing virtual collaborative and toolkit approaches. J Hosp Med 2011;6:271-8. 10. Rhee Y, Heung M, Chen B, Chenoweth CE. Central line-associated bloodstream infections in non-ICU inpatient wards: a 2-year analysis. Infect Control Hosp Epidemiol 2015;36:424-30.
7
11. Economou C, Kaitelidou D, Karanikolos M, Maresso A. Greece: health system review. Health Syst Transit 2017;19:1-166. 12. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health careassociated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008;36:309-32. 13. Greek Center for Disease Prevention and Control. Care bundles for the prevention of central line associated bloodstream infections. Athens: KEELPNO; 2014. 14. Lee KH, Cho NH, Jeong SJ, Kim MN, Han SH, Song YG. Effect of central line bundle compliance on central line-associated bloodstream infections. Yonsei Med J 2018;59:376-82. 15. Apisarnthanarak A, Thongphubeth K, Yuekyen C, Warren DK, Fraser VJ. Effectiveness of a catheter-associated bloodstream infection bundle in a Thai tertiary care center: a 3-year study. Am J Infect Control 2010;38:449-55. 16. Joint Commission. Preventing central line−associated bloodstream infections: a global challenge, a global perspective. Oak Brook (IL): Joint Commission; 2012. 17. Sypsa V, Psichogiou M, Bouzala GA, Hadjihannas L, Hatzakis A, Daikos GL. Transmission dynamics of carbapenemase-producing Klebsiella pneumoniae and anticipated impact of infection control strategies in a surgical unit. PLoS One 2012;7:e41068. 18. McLaws ML, Kwok YLA. Hand hygiene compliance rates: fact or fiction? Am J Infect Control 2018;46:876-80. 19. Duane TM, Brown H, Borchers CT, Wolfe LG, Malhotra AK, Aboutanos MB, et al. A central venous line protocol decreases bloodstream infections and length of stay in a trauma intensive care unit population. Am Surg 2009;75:1166-70. 20. Fridkin SK, Pear SM, Williamson TH, Galgiani JN, Jarvis WR. The role of understaffing in central venous catheter-associated bloodstream infections. Infect Control Hosp Epidemiol 1996;17:150-8.
Contents lists available at ScienceDirect
American Journal of Infection Control journal homepage: www.ajicjournal.org
Major Article
Failure of central venous catheter insertion and care bundles in a high central line−associated bloodstream infection rate, high bed occupancy hospital Amalia Karapanou RN, MSc a, Anna-Maria Vieru MD b, Michail A. Sampanis RN, MSc a, Angeliki Pantazatou MD c, Ioannis Deliolanis MD c, George L. Daikos MD b, Michael Samarkos MD b,* a
Infection Control Committee, Laikon Hospital, Athens, Greece First Department of Medicine, Medical School, National and Kapodistrian University of Athens, Athens, Greece c Microbiology Laboratory, Laikon Hospital, Athens, Greece b
Key Words: Patient care bundles Central venous catheters Bacteremia Catheter-related infections Nursing staff Patient safety
Background: Our hospital has several characteristics different from the settings in which the central venous catheter (CVC) care bundle has been implemented so far, that is, care bundles or protocols are not systematically used, and the prevalence of central line−associated bloodstream infections (CLABSI) is high, as is bed occupancy rate. We examined the effectiveness of CVC care bundles. Methods: Modified CVC bundles were implemented across all settings of our hospital. During both phases of the study, we collected data on CLABSI, and we monitored CVC insertion and management practices with direct observation audits. Results: We have studied 913 CVC insertions (454 in PRE and 459 in POST) for 11,871 catheter-days. The incidence of CLABSI was 8.3 per 1,000 catheter-days PRE, and 7.6 per 1,000 catheter-days POST (incidence rate ratio, 0.92; 95% confidence interval, 0.60-1.40). Compliance with the CVC insertion bundle increased from 8.4%-74.3% (P < .0001). The CVC management bundle compliance also increased from 11.4%-57.7% (P < .0001). Conclusions: Despite improved compliance after the intervention, implementation of a modified CVC bundle failed to decrease CLABSI incidence. Higher bundle compliance rates may be necessary for a significant decrease in the incidence of CLABSI, along with the appropriate organizational culture and levels of staffing. © 2019 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
Central line−associated bloodstream infections (CLABSI) represent a major challenge to patient safety and a significant burden to health systems. More than 20% of bloodstream infections in acute care hospitals in Europe are CLABSI.1 These infections carry an attributable mortality of up to 30%,2 and they are probably the most costly health care−associated infections.3 It has been estimated that up to 65%-70% of CLABSIs are preventable,4,5 and several factors, such as catheter insertion site, type of catheter, and type of dressing are important for their prevention.6 Care bundles are small sets—usually 3 to 5—of evidence-based interventions that, when implemented collectively and reliably, improve the delivery of health care and improve patient outcomes.7 Care bundles for central venous catheter (CVC) insertion and management *Address correspondence to Michael Samarkos, MD, First Department of Medicine, Medical School, National and Kapodistrian University of Athens, Laikon Hospital, 17 Agiou Thoma St, Athens, Greece 11527. E-mail address: [email protected] (M. Samarkos). Conflicts of interest: None to report.
have decreased the incidence rate of CLABSIs significantly.8 However, the effectiveness of these interventions might not be the same across all settings. As Speroff et al9 remarked, incorporating improvement methods is complex, highly context-dependent, and may take long. Thus there might be differences depending on the level of care (ie, intensive care unit [ICU] vs patient wards), on the experience in implementation and use of care bundles, as well as on differences in safety culture.10 Our hospital has several characteristics that differ from the setting in which the CVC care bundle has been implemented so far. First, care bundles or protocols are not systematically used in patient care. Second, the prevalence of CLABSI in the European Centre for Disease Prevention and Control point prevalence survey of 2012 was high, as bloodstream infections accounted for 45.4% of all hospital-acquired infections, and 50% of which were catheter-related (hospital data). Third, the bed occupancy rate of our hospital is very high, with medical wards systematically registering occupancy rates well over 100%. Fourth, our hospital is understaffed in registered nurses.11 Finally,
https://doi.org/10.1016/j.ajic.2019.11.018 0196-6553/© 2019 Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.
ARTICLE IN PRESS 2
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
chlorhexidine gluconate 2% is not available in Greece, and transparent semipermeable dressings are in short supply in our hospital. Consequently, the aim of our study was to evaluate the effectiveness of a modified CVC bundle on the incidence of CLABSI in a setting lacking in safety culture, with high CLABSI incidence, high bed occupancy rate, and relatively limited resources. METHODS Setting We have conducted a quasiexperimental pre−post study at Laikon Hospital, a tertiary care, university-affiliated hospital in Athens, with 574 beds that includes hematology−oncology wards and both solid organ and bone marrow transplantation units. All wards (medical, surgical, and ICU), except the emergency department, were included in the study. Procedures The study consisted of 3 phases: preintervention (PRE, January 1 to June 30, 2016), intervention (INT, July 1 to September 30, 2016), and postintervention (POST, October 1, 2016 to March 31, 2017). During the preintervention phase, we collected baseline data on CLABSI. For the CLABSI surveillance, we used the Centers for Disease Control and Prevention/National Healthcare Safety Network definition.12 In brief, CLABSI is defined as a laboratory-confirmed bloodstream infection not related to an infection at another site that develops within 48 hours of a central line placement. If >1 CLABSI occurred on the same catheter, we recorded only the initial one. Data collected included dates of catheter insertion and removal, site of catheterization, number of lumens, urgency of insertion, and setting were the insertion occurred. We have also collected data on whether the index catheterization was the initial one, or it was performed to replace another CVC. Finally, we also collected data on patient demographics, ward, admission diagnosis, presence of comorbidities, interventions such as mechanical ventilation and parenteral nutrition, pathogens isolated from blood cultures, and patient outcomes. The intervention phase consisted of the implementation of the CVC insertion and maintenance care bundles. The bundles were based on the ones issued by the Greek Centre for Control and Prevention of Diseases (KEELPNO) with some modifications.13 The elements of the insertion bundle included hand hygiene, maximal barrier precautions (total body drape and personal protective equipment), aseptic technique throughout the insertion, and use of transparent dressing. The management bundle consisted of hand hygiene, hub antisepsis with alcohol-based handrub, and aseptic technique during the procedure. In both cases, bundle undertaking required compliance with all bundle elements. All wards prepared and stocked CVC insertion kits, which included all materials required by the bundle, as well as the CVC insertion checklist. We did not use antimicrobial-coated catheters. The insertion bundle, however, lacked 2 elements. The assisting nurse did not have the option to interrupt the CVC insertion if the bundle protocol was violated. Thus violations were not recorded. In addition, a checklist included in the CVC insertion kit served as a reminder and it was not to be filled for each insertion. The management bundle did not include daily evaluation of ongoing need of the CVC. The intervention included training of all health care workers involved in the insertion and maintenance of CVCs, with small group teaching sessions with 2 of the researchers acting as facilitators (A.K. and M.S.). During the intervention phase, we did not collect any data. During the postintervention phase, we collected the same data as with the preintervention phase. We also audited CVC insertion and management practices, with direct observation in all phases of the study. Bundle violations were noted in the audit. The audit forms
were based on the respective checklists that we were planning to implement. Throughout the study we were monitoring hand hygiene compliance monthly. Outcomes The main outcome was CLABSI incidence expressed as CLABSI events per 1,000 catheter-days. Secondary outcomes included days from CVC insertion to CLABSI, duration of catheterization, catheter use rate, and in-hospital mortality. Statistical analysis We compared the PRE with the POST phase. During the INT phase, we did not collect any data, therefore results of the POST phase do not include the INT phase. Continuous variables were expressed as mean and standard deviation or median interquartile range, whereas qualitative variables were expressed as absolute and relative frequencies. We used the x2 test (or the Fisher exact test when appropriate) to compare dichotomous variables, and the Mann−Whitney U test to compare continuous variables. A Poisson distribution was used to compare the incidence rate of CLABSI among different groups, with the calculation of incidence rate ratios (IRR). For the comparison of proportions, the x2 and the Fisher exact tests were used. A stepwise multiple logistic regression analysis (P for removal was set at .1, and P for entry was set at .05) was performed to identify factors independently associated with CLABSI. For time to event variables we performed survival analysis with Kaplan−Meier curves, and Cox regression. All P values reported are 2-tailed. Statistical significance was set at .05 and analyses were conducted using IBM SPSS Statistics version 24.0 (IBM Corporation, Armonk, NY). The study protocol was approved by the institutional review board (approval number 3486/10-3-2015), and permission to create the study database was granted by the Greek Personal Data Protection Authority (decision number 1605-2/18-11-2015). RESULTS Patient characteristics We recorded in total 913 CVC insertions (454, 49.7% in PRE and 459, 50.3% in POST) in 713 patients. The catheters remained in place for 11,871 catheter-days. Mean age in the total patient population was 66.7 § 15.4 years, however, patients in the POST phase were significantly older than those in the PRE phase (68.2 § 14.6 years vs 65.2 § 16.1 years; Mann−Whitney U test, P = .033). The prevalence of major comorbidities in our patients’ population was 22.3% for diabetes mellitus, 8.0% for chronic obstructive pulmonary disease, 5.3% for congestive heart failure, 33.9% for coronary heart disease, and 25.8% for hematologic neoplasms. The prevalence of congestive heart failure and coronary heart disease were significantly higher in the POST phase (3.6% vs 7.1%, P = .039 and 30.3% vs 37.1%, P = .037, respectively). There was a difference in the PRE and POST intervention prevalence of prior admission to hospital (68.1% vs 76.5%; P = .012) and dialysis (15.8% vs 10.8%; P = .046). The most frequently isolated pathogen in the PRE phase was Acinetobacter baumannii (10 of 50, 20.0%), followed by coagulase-negative staphylococci (8 of 50, 16.0%), and Klebsiella pneumoniae (6 of 50, 12.0%). In the POST phase, A baumannii was again the most frequently isolated pathogen (11 of 45, 24.4%), followed by Enterococcus faecium (10 of 45, 22.2%), and coagulase-negative staphylococci (9 of 45, 20.0%). There was no difference in the proportion of gram-positive pathogens isolated in the PRE and POST phase (21 of 50, 42.0% vs 24 of 45, 53.3%; P = 0.3).
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
3
Table 1 Catheter insertion−related variables before and after the intervention
Ward
Site of catheterization
Side of catheterization Number of lumens Urgency of insertion Timing of CVC insertion
Insertion setting
Medical Surgical ICU Femoral* Jugular* Subclavian Right* Left 2 3 Elective* Emergent Initial Catheter malfunction Catheter infection Ward Operating room Dialysis unit Emergency room
PRE, n = 454 n, (%)
POST, n = 459 n, (%)
Total, n = 913 n, (%)
248 (54.6) 125 (27.5) 81 (17.8) 194 (42.7) 150 (33.0) 110 (24.2) 316 (69.6) 138 (30.4) 323 (71.1) 131 (28.9) 367 (80.8) 87 (19.2) 392 (86.3) 39 (8.6) 23 (5.1) 341 (75.1) 101 (22.2) 9 (2.0) 3 (0.7)
226 (49.2) 145 (31.6) 88 (19.2) 76 (16.6) 255 (55.6) 128 (27.9) 394 (85.8) 65 (14.2) 336 (73.2) 123 (26.8) 410 (89.3) 49 (10.7) 409 (89.1) 27 (5.9) 23 (5.0) 320 (69.7) 109 (23.7) 26 (5.7) 4 (0.9)
474 (51.9) 270 (29.6) 169 (18.5) 270 (29.6) 405 (44.4) 238 (26.1) 710 (77.8) 203 (22.2) 659 (72.2) 254 (27.80) 777 (85.1) 136 (14.9) 801 (87.7) 66 (7.2) 46 (5.0) 661 (72.4) 200 (23.0) 35 (3.8) 7 (0.8)
Statistical significance testing for the difference in proportions between PRE and POST. CVC, central venous catheter; ICU, intensive care unit; PRE, preintervention; POST, postintervention. *P <.001.
Catheterization-related variables Table 1 shows variables related with the CVC insertion in the 2 phases of the study and in the total study population. The femoral site insertion decreased significantly from 42.7% (194 of 454) in the PRE to 16.6% (76 of 459) in the POST period (x2, P < .001), whereas jugular site insertion increased significantly (33.0%, 150 of 454 to 55.6%, 255 of 459; P < .001). The urgency of insertion (emergent vs elective) also differed, with significant decrease in emergent CVC insertions from 19.2% (87 of 454) to 10.7% (49 of 459) in the POST phase (x2, P < .001). Outcomes The CLABSI incidence decreased from 8.3 (50 CLABSIs in 5,998 catheter-days) PRE to 7.6 (45 CLABSIs in 5,873 catheter-days) CLABSIs per 1,000 catheter-days POST intervention. The decrease, however, was not significant (IRR, 0.92; 95% confidence interval [CI], 0.601.40). Similarly, the rate of CLABSI was not significantly decreased in subgroups such as medical patients (IRR, 0.64; 95% CI, 0.34-1.17; P = .12), patients with CVC at the subclavian site (IRR, 0.54; 95% CI, 0.19-1.37; P = .15), and patients in which the CVC was inserted in the ward (IRR, 0.87; 95% CI, 0.54-1.38; P = .53). The median time from CVC insertion to CLABSI in the 2 study phases was 12.0 days (95% Cl, 10.0-15.0) for the PRE and 10.0 days (95% CI, 8.0-12.0) for the POST. We have performed survival analyses (Kaplan−Meier curves) to compare the time from CVC insertion to CLABSI between the 2 study phases, however, the difference was not significant (Fig 1, log-rank test x2 = 0.1461, P = .7). Catheter use rate did not differ between the PRE and the POST phase of the study (7.5 per 100 patient-days vs 7.8 per 100 patientdays; IRR, 1.032; exact 95% CI, 0.90-1.17; x2, P = .62). The overall in-hospital mortality was 27.1% (193 of 713) and did not differ between the 2 phases of the study (PRE 24.4%, 86 of 353 vs POST 29.7%, 107 of 360; x2, P = .11). Patients with CLABSI, as expected, had significantly higher mortality (41.2% [35 of 85] vs 25.6% [161 of 628]; Fisher exact test, P = .004).
that it was associated with hospitalization in the ICU (x2, P = .002), with insertion after previous CVC infection (x2, P < .001), and with insertion of CVCs with 3 versus 2 lumens (Fisher exact test, P = .002), whereas insertion in the operating room was associated with significantly lower probability of CLABSIs (x2, P = .031). When we investigated CLABSI rate in relation to possible risk factors (Table 2), we found that CLABSI rate was significantly higher in the ICU (IRR, 2.04; P = .02), and with insertion after previous CVC infection (IRR, 2.65; P = .0027). In addition, duration of catheterization was significantly longer in patients with CLABSI vs patients with no CLABSI (median 16.0 [95% CI, 13.0-19.0] vs 10.0 [95% CI, 10.0-11.0] days; Mann−Whitney U test, P < .001), and so was length of hospital stay (median 21 [95% CI, 203.0-23.0] vs 34 days [95% CI, 30.0-39.0]; Mann−Whitney U test, P < .001). Monthly bed occupancy rates during the study ranged from 49.5%-118.5% (median 83.2%), thus we examined the association of monthly bed occupancy rate of the department in which the patient was hospitalized when the CVC was inserted, with CLABSI. However, there was no difference in bed occupancy rate between patients with and without CLABSI (Mann−Whitney U test, P = .105). To investigate for risk factors, we performed logistic regression in which the outcome was CLABSI and the independent variables were ward, indication for insertion, insertion setting, number of CVC lumens, study phase, and duration of catheterization (Table 3). Only CVC insertion because of previous CLABSI was independently associated with CLABSI (odds ratio, 2.6; P = .017). Duration of catheterization was also statistically significantly associated with CLABSI (P = .001). An odds ratio of 1.03 per extra day with a CVC could be relevant when the catheter remains in place for a long time. We have also performed Cox regression to examine the effect of different variables (eg, ward, number of lumens, and others) to the time from CVC insertion to CLABSI. Only CVC change because of previous CLABSI was independently associated with time to CLABSI (hazard ratio, 2.4 [95% CI, 1.2-4.7]; P = .014).
Bundle compliance Risk factors for CLABSI We examined several variables as risk factor for CLABSI (Table 2). When CLABSI was examined as a dichotomous variable, we found
During the intervention phase, small group teaching sessions regarding the insertion and management of CVC were attended by 139 of 201 nurses (69.1%) and 82 of 93 physicians (88.2%).
ARTICLE IN PRESS 4
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
Fig 1. Kaplan−Meier curve showing the probability of CLABSI according to the duration of catheterization in the 2 study phases. CLABSI, central line−associated bloodstream infection; PRE, preintervention; POST, postintervention.
Compliance with the CVC insertion bundle (n = 153 observations) increased significantly from 8.4% (7 of 83) in the PRE phase to 74.3% (52 of 70) in the POST phase (x2, P < .0001). In the PRE phase, the compliance was lowest for use of maximal barrier precautions (9.6%, 8 of 83) and hand hygiene (27.7%, 23 of 83), whereas in the POST phase, compliance was lowest for maximal barrier precautions (87.1%, 61 of 70) and for the use of transparent dressing (74.3%, 52 of 70). The CVC management bundle compliance (n = 295 observations) also increased from 11.4% (15 of 132) in the PRE phase to 57.7% (94 of 163) in the POST phase (x2, P < .0001). The bundle element with the lowest compliance in the PRE phase was hand hygiene before managing the CVC (31.3%, 42 of 132) and in the POST phase was use of sterile gloves (63.2%, 103 of 163). Details regarding compliance with the individual components of the bundles in the 2 phases of the study are presented in Fig 2. DISCUSSION Implementation of the CVC care bundle in ICU and non-ICU patients reduced CLABSI rates from 8.3 to 7.6 CLABSI per 1,000 catheter-days, however, the reduction was not significant. However, we reduced significantly catheterizations of the femoral vein from 42.7%-16.6%, the insertion bundle compliance increased from 8.4%74.3%, and the management bundle from 11.4%-57.5%. In our analysis of risk factors for CLABSI we could only find that ICU hospitalization, use of 3 versus 2 lumen catheters, previous CLABSI, and duration of catheterization were associated with CLABSI in univariate analyses, however, only previous CLABSI and duration of catheterization were
independently associated with CLABSI in multivariate analysis. We could not associate compliance with the care bundles or with any of their elements with patient outcomes because our study was designed as an ecological one. Compliance was estimated by audit, thus we lacked data on the bundle compliance for each patient as it has been done in other studies.14 Although we have hypothesized that bed occupancy rate would influence the CLABSI rate, we could not establish such an association. In contrast to our study, CVC care bundles have generally succeeded in significantly reducing CLABSI rates.8 There are several possible contributing factors to the failure of our care bundles. Our setting had some unique characteristics. The baseline incidence was a relatively high CLABSI incidence for a mixed setting, however, there have been reported even higher baseline CLABSI rates in the literature.15 The bundle was a modification of the ones widely used.13,16 An important difference was that the assisting nurse was not given the option to interrupt the insertion of the CVC when the bundle was violated because of the reluctance of our nurses to challenge physicians. There is a rather strong hierarchical structure in Greek hospitals, with physicians occupying the higher levels. It is apparent that implementation of care bundles has certain requirements in terms of organizational culture, which unfortunately, we lacked. Furthermore, a checklist served as a reminder only, and it did not have to be filled in each CVC insertion, as the nursing staff suggested that it would add to their already increased workload. In our hospital, there are on average 3 registered nurse to 40 patients in the wards during the morning shift and 2 during the afternoon and night shift (personal observation). In general, the nurse−patient ratio in Greece is the
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
5
Table 2 Univariate risk factors analysis for CLABSI Variable Ward
Site of catheterization
Number of lumens Urgency of insertion Indication for CVC insertion
Insertion setting
Prior admission Infection on admission Surgery Antimicrobial administration Chemotherapy Immunosuppression Mechanical ventilation Parenteral nutrition Nasogastric tube Urinary catheter Dialysis
CLABSI (n, %) Medical Surgical* ICU Femoral Jugular Subclavian* 2 3 Elective Emergent Initial* Catheter malfunction Catheter infection Operating room*,k Ward Dialysis unit ER No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes
51 (10.8) 16 (5.9) 28 (16.6)y 27 (10.0) 46 (11.4) 22 (9.2) 55 (8.3) 40 (15.7)z 85 (10.9) 10 (7.4) 74 (9.2) 8 (12.1) 13 (28.3)x 10 (4.8) 80 (12.1) 5 (14.3) 0.0 22 (9.7) 73 (10.6) 7 (8.3) 88 (10.6) 61 (11.2) 34 (9.3) 14 (12.8) 81 (10.1) 79 (10.9) 16 (8.4) 70 (11.0) 25 (9.0) 68 (10.4) 27 (10.3) 78 (10.2) 17 (11.4) 52 (10.3) 43 (10.6) 22 (9.1) 73 (10.9) 81 (10.3) 14 (10.9)
CLABSI/1,000 catheter-days
IRR (95% CI), P value
7.73 5.63 11.51 7.64 9.05 6.76 7.30 9.22 8.02 7.87 7.22 8.46 19.17 4.73 8.76 9.16 0.00 8.78 7.79 5.56 8.29 8.06 7.90 10.57 7.68 8.37 6.57 8.42 7.02 8.06 7.86 7.92 8.42 7.93 8.10 7.17 8.29 7.88 8.82
1.38 (0.77-2.57), .28 N/A 2.04 (1.06-4.04), .02 1.13 (0.62-2.08), .77 1.34 (0.78-2.33), .32 N/A 1.26 (0.90-1.72), .15 0.98 (0.45-1.89), .99 N/A 1.17 (0.49-2.43), .69 2.65 (1.35-4.82), .0027 N/A 1.85 (0.96-4.0), .07 1.93 (0.51-6.21), .20 N/A 0.89 (0.54-1.50), .6 1.49 (0.69-3.81), .4 0.98 (0.62-1.51), .9 0.73 (0.41-1.39), .2 0.78 (0.43-1.35), .4 0.83 (0.50-1.33), .5 0.97 (0.60-1.54), .9 1.06 (0.59-1.81), .7 1.02 (0.66-1.56), .9 1.16 (0.71-1.96), .6 1.12 (0.58-1.99), .6
Statistical significance testing for the difference in proportions between number of CLABSI and CLABSI. CI, confidence interval; CLABSI, central line−associated bloodstream infection; CVC, central venous catheter; ER, emergency room; ICU, intensive care unit; IRR, incidence rate ratios; N/A, not applicable. *Baseline category for rate comparison. y P = .002. z P = .002. x P <.001. k P = .031.
Table 3 Multivariable analysis (logistic regression) of risk factors for central line−associated bloodstream infection Variable
Odds ratio (95% CI)
P
Study phase Insertion because of catheter malfunction Insertion because of catheter infection Three-lumen catheter Duration of catheterization Insertion in the ward Insertion in the dialysis unit ICU
0.90 (0.58-1.40) 0.99 (0.43-2.27) 2.60 (1.19-5.72) 1.28 (0.79-2.08) 1.04 (1.02-1.06) 1.71 (0.77-3.81) 2.14 (0.63-7.21) 1.31 (0.73-2.34)
.636 .976 .017 .322 .001 .190 .221 .370
CI, confidence interval; ICU, intensive care unit.
lowest in the European Union (344 per 100,000 population in 2014 compared with an European Union average of 864).11 Although bundle compliance in our hospital increased after the intervention, it was still relatively low. In the systematic review of CVC care bundles in critically ill patients by Ista et al,8 the majority of the included studies reported a preintervention bundle compliance that was at least comparable to our POST phase levels. There is not an established critical
compliance threshold above which bundles are effective, however, there are such data for other infection control interventions such as hand hygiene.17 Importantly, compliance with the maintenance bundle in the POST phase was <60%. This might be critical, given that maintenance bundles should be performed several times daily, whereas the insertion bundle must be performed only once. Thus noncompliance with the maintenance bundle may have a cumulative effect larger than that of noncompliance with the insertion bundle. We assessed compliance rates by a direct observation, noncovert audit. It is possible that our audit results do not reflect actual practice. First, we have observed only 153 of the 913 CVC insertions (16.8%), and second because of the so-called Hawthorne effect. There are studies on hand hygiene compliance suggesting that the Hawthorne effect might lead to significantly inflated compliance rates and this might be the case in our study too.18 In addition, the violations of the insertion bundle were not recorded, except in the context of the compliance audit (discussed earlier). This is an important limitation in our study, as we cannot fully explain our findings. Our study had relatively short duration, so that there was not adequate time for the intervention to be incorporated into everyday practice, as this might take long.15,19
ARTICLE IN PRESS 6
A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7
Fig 2. Compliance with individual elements and overall bundle compliance for the CVC insertion (upper panel) and the CVC maintenance bundle (lower panel) before (PRE) and after (POST) the intervention. CVC, central venous catheter; PPE, personal protective equipment.
Understaffing may have played a role in our results. We do not have accurate data regarding the nurse staffing per department during our study, however, we have already discussed the influence of understaffing on the study procedures, and in addition there is evidence that the level of nurse staffing may affect CLABSI rates in surgical ICUs.20 CONCLUSIONS In our setting, which is characterized by high CLABSI rates, high bed occupancy rates, understaffing, and a hierarchical organizational culture, implementation of CVC care bundles failed to
significantly reduce CLABSI rates. We suggest that care bundles are more than checklists of actions, and their success is, at least in part, context-dependent. Requirements in terms of organizational culture and staffing are probably critical to successful care bundle implementation. References 1. European Centre for Disease Prevention and Control. Point prevalence survey of healthcare-associated infections and antimicrobial use in European acute care hospitals 2011-12. Stockholm: ECDC; 2013.
ARTICLE IN PRESS A. Karapanou et al. / American Journal of Infection Control 00 (2019) 1−7 2. Ziegler MJ, Pellegrini DC, Safdar N. Attributable mortality of central line associated bloodstream infection: systematic review and meta-analysis. Infection 2015;43:29-36. 3. Zimlichman E, Henderson D, Tamir O, Franz C, Song P, Yamin CK, et al. Health careassociated infections: a meta-analysis of costs and financial impact on the US health care system. JAMA Intern Med 2013;173:2039-46. 4. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ. Estimating the proportion of healthcare-associated infections that are reasonably preventable and the related mortality and costs. Infect Control Hosp Epidemiol 2011;32:101-14. 5. Making health care safer: reducing bloodstream infections. CDC Vital Signs 2011, Available from: https://www.cdc.gov/vitalsigns/pdf/2011-03-vitalsigns.pdf. Accessed April 20, 2019. 6. Bell T, O’Grady NP. Prevention of central line-associated bloodstream infections. Infect Dis Clin North Am 2017;31:551-9. TagedP 7. Institute for Healthcare Improvement. What is a bundle? Available from: http:// www.ihi.org/knowledge/Pages/ImprovementStories/WhatIsaBundle.aspx. Accessed April 20, 2019. 8. Ista E, van der Hoven B, Kornelisse RF, van der Starre C, Vos MC, Boersma E, et al. Effectiveness of insertion and maintenance bundles to prevent central-line-associated bloodstream infections in critically ill patients of all ages: a systematic review and meta-analysis. Lancet Infect Dis 2016;16:724-34. 9. Speroff T, Ely EW, Greevy R, Weinger MB, Talbot TR, Wall RJ, et al. Quality improvement projects targeting health care-associated infections: comparing virtual collaborative and toolkit approaches. J Hosp Med 2011;6:271-8. 10. Rhee Y, Heung M, Chen B, Chenoweth CE. Central line-associated bloodstream infections in non-ICU inpatient wards: a 2-year analysis. Infect Control Hosp Epidemiol 2015;36:424-30.
7
11. Economou C, Kaitelidou D, Karanikolos M, Maresso A. Greece: health system review. Health Syst Transit 2017;19:1-166. 12. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health careassociated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008;36:309-32. 13. Greek Center for Disease Prevention and Control. Care bundles for the prevention of central line associated bloodstream infections. Athens: KEELPNO; 2014. 14. Lee KH, Cho NH, Jeong SJ, Kim MN, Han SH, Song YG. Effect of central line bundle compliance on central line-associated bloodstream infections. Yonsei Med J 2018;59:376-82. 15. Apisarnthanarak A, Thongphubeth K, Yuekyen C, Warren DK, Fraser VJ. Effectiveness of a catheter-associated bloodstream infection bundle in a Thai tertiary care center: a 3-year study. Am J Infect Control 2010;38:449-55. 16. Joint Commission. Preventing central line−associated bloodstream infections: a global challenge, a global perspective. Oak Brook (IL): Joint Commission; 2012. 17. Sypsa V, Psichogiou M, Bouzala GA, Hadjihannas L, Hatzakis A, Daikos GL. Transmission dynamics of carbapenemase-producing Klebsiella pneumoniae and anticipated impact of infection control strategies in a surgical unit. PLoS One 2012;7:e41068. 18. McLaws ML, Kwok YLA. Hand hygiene compliance rates: fact or fiction? Am J Infect Control 2018;46:876-80. 19. Duane TM, Brown H, Borchers CT, Wolfe LG, Malhotra AK, Aboutanos MB, et al. A central venous line protocol decreases bloodstream infections and length of stay in a trauma intensive care unit population. Am Surg 2009;75:1166-70. 20. Fridkin SK, Pear SM, Williamson TH, Galgiani JN, Jarvis WR. The role of understaffing in central venous catheter-associated bloodstream infections. Infect Control Hosp Epidemiol 1996;17:150-8.