Reducing blood culture contamination by a simple information intervention. Department of Medical Microbiology, Skåne University Hospital, Sweden

Transcription

1 JCM Accepts, published online ahead of print on September 0 J. Clin. Microbiol. doi:./jcm.00- Copyright 0, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved. Reducing blood culture contamination by a simple information intervention Roth A MD PhD 1, Wiklund AE 1, Pålsson AS, Melander EZ MD PhD, Wullt M MD PhD, Cronqvist J MD, Walder M MD PhD 1, Sturegård E MD PhD 1 Department of Medical Microbiology, Skåne University Hospital, Sweden Department of Infectious Diseases, Skåne University Hospital, Sweden Department of Infection Control, Skåne University Hospital, Sweden Corresponding author: Adam Roth Department of Medical Microbiology, Skåne University Hospital, SUS Malmö tel:+1 adam.roth@med.lu.se (The address may be published). Running head: Reducing blood culture contamination Downloaded from on December, 01 by guest 1

2 ABSTRACT Compared to true-negatives, false-positive blood cultures not only increase laboratory work but prolong length of stay and use of broad-spectrum antibiotics, both likely to increase antibiotic resistance and patient morbidity. The increased patient suffering and surplus costs caused by blood culture contamination motivates substantial measures to decrease the rate of contamination, including the use of dedicated phlebotomy teams. The present study evaluated the effect of a simple information intervention aimed at reducing blood culture contamination at Skåne University Hospital (SUS) Malmö, Sweden, during. months focusing on departments collecting many blood cultures. The main outcome of the study was contamination rates pre- and post-intervention, analyzed in a multivariate logistic regression model, adjusting for relevant determinants of contamination. 1 blood culture sets were drawn from 1 patients during the study-period (January 00 December 00). The blood culture contamination rate pre-intervention was.% and decreased to.% postintervention (odds ratio 0.; % confidence interval 0.-0.). A similar decrease in relevant bacterial isolates was not found post-intervention. Contamination rates at three auxiliary hospitals did not decrease during the same time period. The effect of the intervention on phlebotomists knowledge of blood culture-routines was also evaluated with a clear increase in level of knowledge among interviewed phlebotomists post-intervention. The present study shows that a relatively simple information intervention can have significant effects on the level of contaminated blood cultures, even in a setting with low rates of contamination where nurses and auxiliary nurses conduct phlebotomies. Downloaded from on December, 01 by guest

3 INTRODUCTION Blood cultures are commonly contaminated, often representing as much as 0% of positive cultures. (1) Compared to true-negatives, false-positive (contaminated) blood cultures not only increase laboratory work but also prolong length of stay of patients and increase the use of broad-spectrum antibiotics, with negative consequences for antibiotic resistance and patient morbidity. Furthermore, false-positive results can cause confusion regarding antibiotic regime, endangering patient safety. (,,1) The dominating organism in blood culture contamination, Coagulase-negative staphylococci (CoNS), is also an increasingly important pathogen which is a significant clinical problem, since there is no true gold standard for determining contamination from relevant pathogens.(1,,1,) Although not applicable for clinical use in individual patients, a laboratory assessment definition of contamination for comparison of rates between institutions has been developed. Target rates should not exceed %, () but many teaching hospitals have contamination rates exceeding % or more. (,1,) Considering the potential savings in resource utilization it is justified to invest considerable resources in reducing blood culture contamination. () Since contamination most probably is a result of personnel introducing exogenous bacteria into the blood culture, education of phlebotomists is central to prevention. Many studies advocate the effectiveness of teams of specialized phlebotomists in reducing contamination rates. (,1,) Feedback, intense education in correct phlebotomy routines and long-term monitoring programs has also been shown to be effective. (,,,) In general, interventions seem less effective and Downloaded from on December, 01 by guest

4 contamination rates higher among nursing personnel conducting phlebotomy compared to specialized phlebotomy teams.(,,1,1,) In addition to presenting internationally comparable rates of blood culture contamination from a Swedish university hospital, the present study evaluates the effect of a simple information intervention aimed at reducing blood culture contamination in a setting where phlebotomy for blood culture is conducted by nurses and auxiliary nurses and not by dedicated phlebotomy teams. The major outcome of the study was intervention-effect on blood culture contamination rates. We also monitored the effect in a novel way, by assessing the impact of the intervention on phlebotomists knowledge of disinfection and phlebotomy routines. Downloaded from on December, 01 by guest

5 METHODS Setting Skåne University Hospital (SUS) is a tertiary care teaching hospital in Sweden with two separate facilities, SUS Lund (formerly Lund University Hospital, USIL) and SUS Malmö (formerly Malmö University Hospital, UMAS). SUS Malmö consists of around 0 beds and 000 staff. The intervention was carried out at SUS Malmö during. months from February to May 00 focusing on phlebotomists (nurses and auxiliary nurses) at departments collecting many blood cultures. Effects of the intervention on phlebotomists knowledge of disinfection- and phlebotomy-routines as well as on contamination rates, was evaluated comparing periods before the intervention with periods during and after intervention (postintervention). At SUS Malmö phlebotomy for blood culture is carried out by nurses and auxiliary nurses and not by dedicated phlebotomy teams. The standard order for blood culture was two blood culture sets from different phlebotomies with no specified delay between the sets. Each blood culture set consisted of one aerobic BacT/ALERT FA (BioMérieux) and one anaerobic BacT/ALERT FN (BioMérieux) bottle containing -ml of blood. For children, and in some cases adults, the BacT/ALERT PF bottle containing 0.-mL blood was used. In the present study a blood culture set was defined as all bottles containing blood drawn from the same phlebotomy (typically bottles). Under certain conditions, physicians may decide to draw blood from intravascular lines, the recommendation in these cases was to draw a blood culture set by standard phlebotomy as well. The blood culture bottles were kept in room temperature and transferred as soon as possible (>% within hours) to the microbiology laboratory for analysis, where the patient identity and analysis order was manually entered Downloaded from on December, 01 by guest

6 into the computerized microbiology laboratory system. Blood cultures were incubated in the BacT/ALERT D instrument (BioMérieux) for. days or until microbial growth was detected. Detection of microbial growth was automatically registered in the computerized microbiology laboratory system. The incubation period could be extended to days on special indications (endocarditis, fungal infection, animal bites). Definition of contamination A blood culture set was defined as the bottles obtained from one blood sample (1- bottles) and considered contaminated if one of the following organisms were present in <0% of all blood culture sets obtained from one patient on the same day: Coagulase-negative staphylococci, alpha-hemolytic streptococci, Micrococcus species, Propionibacterium species, Corynebacterium species and Bacillus species. Since this definition of contamination could not be applied when only one bottle or one blood culture set was drawn for culture, pediatric blood culture bottles and single set samples were excluded from the present study. The recommendation for blood culture in children is to only draw blood (0,- ml) into a single pediatric bottle. If more than one organism was present in the culture, they were scored individually when calculating contamination rates, such that if <0% of blood culture sets in one patient had both a contaminant and a relevant organism, the set was still counted as contaminated. However, if two different contaminants were present in <0% blood culture sets, it was only counted as one contaminated set. The number of blood culture sets per patient was accumulated during the whole follow-up period, regardless of number of admissions. The intervention Downloaded from on December, 01 by guest

7 The problem of rising contamination rates at some departments was discussed with medical doctors and nurses at the departments of Hospital Hygiene, Infectious Diseases and Clinical Microbiology. Common experience was that phlebotomists did not always adhere to the guidelines on skin disinfection and phlebotomy, factors well known to affect contamination rates. Several reasons were suggested, such as a stressful working environment, impractical phlebotomy guidelines and lack of information and feedback. We decided to re-evaluate our guidelines and to perform an information intervention aimed at phlebotomists. After literature search on evidence for phlebotomy practices, our guidelines were re-written to better suit the current clinical situation. The recommendation to carefully disinfect the skin at the phlebotomy site for two minutes was changed to instead recommend one minute active disinfection before allowing to dry. This change was motivated by a general opinion that the previous recommendation was too time-consuming to be adhered to in a stressful working environment. A 1 minute long structured presentation on blood cultures, for usage at weekly staff-meetings by different discoursers, was composed. The presentation included 1 powerpoint-slides and aimed to deliver main messages: 1) What is a contamination/relevant pathogen? ) Describe how we recommend disinfecting and drawing blood for blood culture, but also why these are our recommendations ) Phlebotomist skill can make a difference for contamination rates ) Feedback on phlebotomy practice to phlebotomists is important for contamination rates. ) The department s contamination rates compared to the whole hospital ) Contamination of blood cultures may be very costly, not only for the laboratory but also for other clinics ) A promise to report back to the departments. The presentation was commented upon by doctors and nurses at the three involved departments. Directors of departments at SUS Malmö were informed of the study. During February to May 00, teams consisting of a nurse and a doctor subsequently conducted presentations at departments. The departments comprised the emergency ward, intensive care unit, infectious disease ward, Downloaded from on December, 01 by guest

8 orthopaedic ward, general medicine ward, general surgery ward and paediatric ward. The larger units were visited several times and we estimate that over half of all personnel conducting phlebotomies at the included departments attended the presentations, Sterilization procedure The sterilization procedure recommended when drawing blood for blood culture included: a) Hand disinfection (ethanol 0-%) b) Swab the top of the BacT/ALERT bottles (0,% chlorhexidine in 0% isopropyl alcohol) after removing the protective plastic from the bottles and allow to dry c) Swab the injection site (0,% chlorhexidine in 0% isopropyl alcohol) during two (pre-intervention)/one (post-intervention) minutes and allow to dry d) Do not palpate the vein again after skin disinfection e) Use protective gloves. Knowledge among phlebotomists about blood culture contamination In order to evaluate the effect of the intervention we assessed the level of knowledge of correct blood culture practice among phlebotomists. A semi-structured interview with a questionnaire (Appendix), comprising four main questions, was developed: 1) Are phlebotomy routines for blood culture different from other blood samples, for instance for chemical analysis? ) How are the routines different? ) Why are the routines different? ) Describe the disinfection routines before drawing blood for blood culture. The interviews were performed orally with in total phlebotomists prior (n=1) and - weeks after (n=1) intervention at comparable departments. The answers to the questions were recorded simultaneously in the questionnaire by the interviewer. The phlebotomists had to have been present at the intervention lecture in order to be included in the group being interviewed after the intervention. In addition, adherence to blood culture routines recommending > sets/blood Downloaded from on December, 01 by guest

9 culture order was investigated by comparing rates of single set cultures pre- and post- intervention. Contamination rates Contamination rates from (including January 00) prior to the intervention that started in February 00 were used as baseline value. Within the years (00-00) prior to the intervention we assessed determinants of contamination in a multivariate logistic regression model. The variables investigated were; year of blood culture (00, 00, 00), age at blood culture (<=0, 1-0, 1-0, >0 years), sex (male, female), number of blood culture sets taken (, -, -, > sets/patient). The year of blood culture was included in this analysis to assess whether there were any trends or variance over time at SUS Malmö in contamination rates. If another variable was related to contamination rates at % level of significance it was included in the final model for analysis of intervention effect. Year of contamination could not be included in the final analysis due to co-linearity with the intervention period. Three auxiliary hospitals to SUS Malmö were also analyzed for contamination rates in blood cultures in a similar way in order assess whether the hospitals were of interest for comparison, albeit with a small number of blood cultures. Contamination rates prior to intervention (January 00 to January 00) were compared to rates during and after the intervention (February 00 to December 00). The contamination rates from May 00 to December 00 were tested for trends to assess if there was a trend of increasing contamination rates in the time following the intervention period using nonparametric test for trends. Since there is no gold standard for when a CoNS is a contaminant or a pathogen, we also presented data for CoNS even when not classified as contamination by Downloaded from on December, 01 by guest

10 our definition: CoNS present in more than 0% of the blood culture sets obtained from one patient on the same day (typically / sets). Furthermore, data on three major blood stream infection organisms proven to almost always be pathogenic were presented; Escherichia coli, Staphylococcus aureus and Streptococcus pneumonia (). The analyses were set up in a multivariate logistic regression model, adjusting for relevant determinants of contamination. All analyses were done using Stata/SE (.0 edition) software. Ethics All phlebotomists agreed voluntarily to participate in the interviews. Downloaded from on December, 01 by guest

11 RESULTS Contamination rates During the complete study period, January 00 to December 00, 1 blood culture sets were drawn from 1 patients at SUS Malmö. The rate of contamination was.0% of all blood culture sets, representing % of positive blood culture sets. The dominating contaminant was CoNS (Table 1). Determinants of contamination During the three years prior the intervention, 00-00, there was no change in contamination rates over time. Furthermore, there was no association between sex and contamination rate, whilst age above 0 years and having many cultures taken were associated with high rates of contamination (Table ). Pooling blood culture sets from three auxiliary hospitals (n=1) displayed larger variance in contamination rates between the years, with contamination rates ranging from.% 00 to.1% 00. Assessment of the effect of the intervention on phlebotomy and disinfection practice Phlebotomists interviewed after the intervention consequently scored better on the questions concerning both knowledge on blood culture routines and disinfection practice. Phlebotomists interviewed both pre- and post-intervention were better at describing why than how procedures were different between phlebotomy for blood culture versus a regular sample. However, post-intervention interviewees scored higher both on stating which routines were different and also in explaining why they were different. (Figure 1) In addition, a significant reduction of single set blood cultures was seen post-intervention, Odds Ratio (OR) 0. (% Downloaded from on December, 01 by guest

12 Confidence Interval (CI) ), demonstrating better adherence to the guidelines recommending > sets/blood culture order Effect of the intervention on contamination rates Of the 0 blood cultures taken before the intervention (pre-intervention: January 00 to January 00).% were classified as contaminated. Post-intervention (including intervention period: February 00 to December 00) the contamination rate was.%, which was a significant decrease also when adjusting for relevant determinants of contamination OR 0. (% CI 0.-0.) (Table ). There was no significant trend of increasing rates of contamination (p=0.) during the months following the intervention (June 00 to December 00). The rate of CoNS not classified as contaminants were not reduced by the intervention, OR 0. (% CI ). There was an increase of S. pneumoniae during the intervention period (Table ). Pooling the three relevant pathogens E. coli, S. aureus and S. pneumoniae, there was no significant increase during the intervention period OR 1. (% CI 0.-1.) whereas there was a significant decrease of contaminants (Table ). Pooling the contamination rates for the three auxiliary hospitals there was no difference in contamination rates comparing pre-intervention to post-intervention period, OR 0. (% CI ) adjusting for relevant determinants of contamination. Downloaded from on December, 01 by guest 1

13 DISCUSSION The rate of positive blood cultures at SUS Malmö were similar to studies from other hospitals while contamination rates of blood cultures before our intervention were comparatively low by international standards. Contamination rates did not increase during the four years included in the study. Having many blood cultures taken and increasing age of patients were both independent risk factors for blood culture contamination. The information intervention had apparent effects on nurses and auxiliary nurses knowledge and understanding of disinfection and phlebotomy routines. Despite a low rate of contamination, the intervention also had a significant effect on contamination rates at the intervention hospital but there was no observable effect during the corresponding time periods at three auxiliary hospitals. The intervention did not decrease the rates of E. coli, S. aureus, S. pneumoniae or CoNS not defined as contaminants. Compared to other studies using a similar definition of contamination, the present study was considerably larger with lower pre-intervention baseline rates of contamination than those presented in most recent studies. (-,,1) However, our contamination rates preintervention were comparable to the mean rates of large multi-center studies from the US, (,1) while previous studies from Sweden or similar settings to ours have not been presented. For optimal detection of pathogens it is recommended by the Swedish reference group for clinical microbiology to collect 0ml of blood per blood culture set. (0) Compared to some of the previous studies, we therefore routinely collect larger volumes of blood (usually -0ml) which may partly explain the low contamination rates since these are inversely related to blood volume. () Some authors have suggested that blood culture contamination rates are increasing, possibly due to better identification instruments and more Downloaded from on December, 01 by guest 1

14 extensive use of indwelling intravenous catheters for drawing blood culture (). There was however no increase in blood culture contamination during the four years of follow-up in the present study, which is in line with the findings from a major previous multi-center study. () In the same study, comparing contamination rates between phlebotomists with different educational background, nurse staff had higher rates than non-nurses and the use of specialized dedicated phlebotomy teams is advocated. In the present study all phlebotomies were conducted by nurses and auxiliary nurses, and considering our lower contamination rates both pre- and post-intervention we could probably only marginally, if at all, have improved contamination rates by introducing specialized phlebotomy teams. The current study, specifically investigating age as a factor associated with contaminated blood cultures, clearly demonstrated an association between increasing age of patients and contamination of blood cultures which is in line with a previous study showing nonsignificant trend to such an association.() A likely explanation to this finding is that drawing blood may be more difficult in older patients and therefore result in more contaminated blood cultures In addition to increasing age, multiple sampling was found to be an independent risk factor of having a contaminated blood culture. Similar to earlier studies (,), no association between sex and blood culture contamination could be found. Data analyzed in the present study was obtained from the database at the Clinical Microbiology Laboratory at SUS Malmö and no clinical characteristics were obtained for included patients. However, both high age and multiple testing are factors likely to be larger with increasing severity of disease, which has previously been shown to be related to blood culture contamination. () The main analysis, comparing contamination rates before and after intervention, was adjusted for relevant determinants of contamination in a multivariate analysis. Through adjusting for multiple sampling, cultures from one patient will not impact the results disproportionately. Downloaded from on December, 01 by guest 1

15 In the current study all departments conducting many blood cultures were included and we compared rates of contamination levels before and after the intervention. This study design has the inherent weakness that factors changing over time may have caused the observed effect rather than the intervention itself. We tried to compensate this effect through adjusting for relevant determinants of contamination but there could still be uncontrolled time effects. We do not however, consider it likely that the effect on contamination rates observed could be caused by other factors than the intervention, since the level of contamination at SUS Malmö varied little during the three years prior to the intervention. Furthermore, comparing contamination rates during the same periods of time at the three auxiliary hospitals, there was no corresponding drop in contamination rates; however, this sub-analysis was hampered by the small numbers of blood cultures. There was no drop in relevant pathogens during the corresponding time periods, which further supports a true intervention effect (Table ). The rate of S. pneumoniae in blood cultures actually increased during the time period corresponding to the information intervention, a finding that may not related to the intervention as such. Although there may have been an actual increase in invasive S. pneumoniae the past years as suggested also in other parts of Sweden, (1) this may well be a random finding considering that the number of cultures positive with S. pneumoniae were few (n= during intervention). The information in the intervention did not reach all phlebotomists (especially night personnel) which probably weakened the impact of the intervention. The number of phlebotomists participating in the evaluation of the effect of the intervention on knowledge of phlebotomy routines was small. However, we considered it likely that major differences in knowledge would be noticed on quite a limited number of interviews. It is not Downloaded from on December, 01 by guest 1

16 likely that the result would have been different with a larger number of participants since all participants answered in a similar way. An effective way of testing knowledge would be to observe and comment on actual practice of phlebotomy (,), a however much more timeand resource-consuming practice than the intervention used in the present study. Since the questionnaire in itself can be considered an intervention, we interviewed different persons before and after the information intervention. Similar departments and phlebotomists with similar previous education were selected before and after the intervention, and we do therefore not think it likely that pre-existing differences between the compared groups explains the difference in knowledge rather than the intervention itself. The follow-up for this part of the study is however short and in order to predict whether increased knowledge will decrease blood culture contamination rate, it would be desirable to do follow-up interviews when more time has passed after the intervention. A clear effect of the intervention on knowledge among phlebotomists further supports that the observed intervention-effect on contamination rate is true. Importantly, our definition of contamination should not be used for clinical purposes considering that the organisms defined as contaminants under certain circumstances may cause disease. The two major contaminants, CoNS and alpha-hemolytic streptococci, are both to a significant extent considered to be true pathogens. () Alpha-hemolytic streptococci are often included in the definition of contamination when comparing institution rates although it is known that they almost as often constitutes true pathogens (). Excluding all alphahemolytic streptococci from our contamination group did not affect the analysis of intervention-effect (data available on request). CoNS are definitely most often contaminants but have become increasingly clinically important as the etiologic agent of bacteremia in for instance patients with intravascular devices. (,1,1) The common practice of determining Downloaded from on December, 01 by guest 1

17 clinical significance of CoNS through the number of bottles in a series with growth (the more bottles with growth, the more relevant) has been questioned (), which is supported by our data since the effect of the intervention seemed even larger on rates of CoNS not classified as contamination when using bottles instead of blood culture sets for contamination definition OR 0. % CI ( ) versus 0. (0.-1.0) (Table ). A true intervention effect is also supported by the fact that the intervention affected contaminant-cons rates but not rates of CoNS present in more than one blood culture set (Table ). The information intervention constructed for our study is very economic in resource utilization compared to other feedback and education studies described where phlebotomists are given feedback at several occasions in a one-to-one manner.(,) Approximately nurse staff hours and 0 hours for a medical doctor were used in total, including preparation and the information visits. It would therefore be interesting to test the information intervention as a means of controlling contamination in a setting where contamination rates are higher than in the present study. One study with comparably low baseline rates of contamination however showed a larger effect of feedback to individual specialized phlebotomists than we demonstrated with our information intervention. () Thus, with further effort and resources invested into the information intervention we may be able to further decrease contamination rates, for instance through pursuing a higher rate of attendance at the presentations and reaching night-personnel. There was no trend of increased rates of contamination during follow-up, but we expect that the information campaign would have to be repeated yearly to maintain an effect. Downloaded from on December, 01 by guest It is evident from several studies from the US that there are considerable costs associated with contaminated cultures in adult patients (,,1,1) that can be related mainly to prolonged 1

18 1 1 length of stay and unnecessary antibiotic regimes, with estimated additional hospital costs per contaminated culture ranging from $00 to $00. It may be that cost of contamination in countries with universal health-care systems differ from the US, for instance due to different incentives for shortening length of stay of a patient. Due to the high excess costs of contamination, many studies advocate the use of dedicated phlebotomy teams to collect all blood cultures. (,,,) Considering the low contamination rates and the apparent effect of the information effort, it does not seem necessary to use phlebotomy teams instead of nurses and auxiliary nurses in our setting. The present study shows that a relatively simple information intervention can have significant effects on the level of contaminated blood cultures, even in a setting with low rates of contamination. Even though the actual costs of contaminated blood cultures have not been investigated in our health-care system it is probable that they exceed the relatively low costs of the information intervention used in the present study. Downloaded from on December, 01 by guest 1

19 REFERENCES 1. Aronson, M. D. and D. H. Bor. 1. Blood cultures. Ann.Intern.Med. : Bates, D. W., L. Goldman, and T. H. Lee.. Contaminant blood cultures and resource utilization. The true consequences of false-positive results. JAMA :-.. Bekeris, L. G., J. A. Tworek, M. K. Walsh, and P. N. Valenstein. 00. Trends in blood culture contamination: a College of American Pathologists Q-Tracks study of institutions. Arch.Pathol.Lab Med. 1:1-1.. Eskira, S., J. Gilad, P. Schlaeffer, E. Hyam, N. Peled, I. Karakis, K. Riesenberg, F. Schlaeffer, and A. Borer. 00. Reduction of blood culture contamination rate by an educational intervention. Clin.Microbiol.Infect. 1:1-1.. Gander, R. M., L. Byrd, M. DeCrescenzo, S. Hirany, M. Bowen, and J. Baughman. 00. Impact of blood cultures drawn by phlebotomy on contamination rates and health care costs in a hospital emergency department. J.Clin.Microbiol. :1-.. Gibb, A. P., B. Hill, B. Chorel, and R. Brant. 1. Reduction in blood culture contamination rate by feedback to phlebotomists. Arch.Pathol.Lab Med. :0-0.. Hall, K. K. and J. A. Lyman. 00. Updated review of blood culture contamination. Clin.Microbiol.Rev. 1:-0. Downloaded from on December, 01 by guest 0 1. Lee, C. C., W. J. Lin, H. I. Shih, C. J. Wu, P. L. Chen, H. C. Lee, N. Y. Lee, C. M. Chang, L. R. Wang, and W. C. Ko. 00. Clinical significance of potential 1

20 contaminants in blood cultures among patients in a medical center. J.Microbiol.Immunol.Infect. 0: Little, J. R., P. R. Murray, P. S. Traynor, and E. Spitznagel. 1. A randomized trial of povidone-iodine compared with iodine tincture for venipuncture site disinfection: effects on rates of blood culture contamination. Am.J.Med. :-1.. Madeo, M. and G. Barlow. 00. Reducing blood-culture contamination rates by the use of a % chlorhexidine solution applicator in acute admission units. J.Hosp.Infect. :0-0.. Mirrett, S., M. P. Weinstein, L. G. Reimer, M. L. Wilson, and L. B. Reller Relevance of the number of positive bottles in determining clinical significance of coagulase-negative staphylococci in blood cultures. J.Clin.Microbiol. : Piette, A. and G. Verschraegen. 00. Role of coagulase-negative staphylococci in human disease. Vet.Microbiol. 1:-. 1. Richter, S. S., S. E. Beekmann, J. L. Croco, D. J. Diekema, F. P. Koontz, M. A. Pfaller, and G. V. Doern. 00. Minimizing the workup of blood culture contaminants: implementation and evaluation of a laboratory-based algorithm. J.Clin.Microbiol. 0:-. 1. Rupp, M. E. and G. L. Archer. 1. Coagulase-negative staphylococci: pathogens associated with medical progress. Clin.Infect.Dis. 1:1-. Downloaded from on December, 01 by guest Schifman, R. B., C. L. Strand, F. A. Meier, and P. J. Howanitz. 1. Blood culture contamination: a College of American Pathologists Q-Probes study involving 0 0

21 institutions and 1 specimens from adult patients. Arch.Pathol.Lab Med. 1: Souvenir, D., D. E. Anderson, Jr., S. Palpant, H. Mroch, S. Askin, J. Anderson, J. Claridge, J. Eiland, C. Malone, M. W. Garrison, P. Watson, and D. M. Campbell. 1. Blood cultures positive for coagulase-negative staphylococci: antisepsis, pseudobacteremia, and therapy of patients. J.Clin.Microbiol. : Strand, C. L., R. R. Wajsbort, and K. Sturmann. 1. Effect of iodophor vs iodine tincture skin preparation on blood culture contamination rate. JAMA : Surdulescu, S., D. Utamsingh, and R. Shekar. 1. Phlebotomy teams reduce bloodculture contamination rate and save money. Clin.Perform.Qual.Health Care : Swedish Institute for Infectious Disease Control. Pneumococci infection (invasive infection) Swedish reference group for clinical microbiology, Swedish Institute for Infectious Disease Control. Referensmetodik:bakteremi-diagnostik Waltzman, M. L. and M. Harper Financial and clinical impact of false-positive blood culture results. Clin.Infect.Dis. :-. Downloaded from on December, 01 by guest Weinbaum, F. I., S. Lavie, M. Danek, D. Sixsmith, G. F. Heinrich, and S. S. Mills. 1. Doing it right the first time: quality improvement and the contaminant blood culture. J.Clin.Microbiol. :-. 1

22 . Weinstein, M. P. 00. Blood culture contamination: persisting problems and partial progress. J.Clin.Microbiol. 1:-.. Weinstein, M. P., M. L. Towns, S. M. Quartey, S. Mirrett, L. G. Reimer, G. Parmigiani, and L. B. Reller. 1. The clinical significance of positive blood cultures in the s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults. Clin.Infect.Dis. :-0.. Zwang, O. and R. K. Albert. 00. Analysis of strategies to improve cost effectiveness of blood cultures. J.Hosp.Med. 1:-. Downloaded from on December, 01 by guest

23 TABLES AND FIGURES TABLE 1. Distribution of organisms in positive blood culture sets classified as contaminants a at University Hospital SUS Malmö Organisms n/total no blood cultures Percent Coagulase-negative staphylococci /1.0 Alpha-hemolytic streptococci 1/1 0. Bacillus species /1 0.0 Corynebacterium species 1/1 0.0 Propionibacterium species /1 0.0 Micrococcus species /1 0.0 All contaminants b 1/1.0 a A blood culture set was defined as the bottles obtained from one blood sample (1- bottles) and considered contaminated if one of the organisms above were present in <0% of the sets obtained from one patient on the same day. b Cultures with multiple organisms are only counted once in the estimate for all Downloaded from on December, 01 by guest contaminants.

24 TABLE. Determinants of contamination at the University Hospital SUS Malmö prior to the intervention, years Contamination Cultures (%) OR (CI) OR (CI) a Year 00, , 0. (0.-1.) 0. ( ) 00 1, 1.0 (0.0-1.) 1.0 (0.0-1.) Age at culture <= (0.-1.0) 1.1 (0.-1.) (1.1-1.) 1. (1.0-1.) > (1.0-1.) 1.0 (1.0-1.) Sex Male 1. 1 Female (0.-1.1) 1.01 (0.-1.1) Number of. 1 Blood Culture (0.-1.) 1. (0.1-1.) sets obtained b ( ) 1.1 (0.-1.) Downloaded from on December, 01 by guest > (1.0-1.) 1. (1.0-1.) a Multivariate analysis, estimates adjusted for all the variables b Blood cultures could span several septic episodes.

25 TABLE. Effect of the intervention on rates of contaminants and relevant pathogens, university hospital SUS Malmö, Organisms Pre-intervention Post-intervention Effect n/total % n/total % OR a % Confidence interval All organisms b 1/0. / ( ) Contamination c 0/0. /1. 0. (0.-0.) Other CoNS d /0 0. 1/ (0.-1.0) E. coli 1/0 1. 1/ (0.-1.1) S. aureus /0 0. / (0.-1.) S. pneumoniae 1/0 0. / (1.1-.1) a Odds ratio by logistic regression adjusting for relevant determinants of contamination (age at culture, number of cultures taken per patient). b Blood cultures where growth of any organism was detected, including organisms classified as contaminants. c Contamination was defined as presence of one of the following organisms in <0% of all blood culture sets obtained from one patient on the same day: CoNS, alpha-hemolytic Downloaded from on December, 01 by guest streptococci, Micrococcus species, Propionibacterium species, Corynebacterium species and Bacillus species. d Other CoNS includes Coagulase-negative staphylococci not classified as contamination due to

26 presence in >0% of blood culture sets obtained from one patient on the same day. 1 FIGURE 1. Knowledge of Routines among phlebotomists pre and post intervention Do routines for blood culture differ from those of other blood-samples? HOW are procedures different? More careful disinfection Larger volume of blood is drawn Do not draw blood from old CVC/PVC Use two different phlebotomy sites Take two bottles at each phlebotomy site Start with the blood culture at multiple testing WHY are procedures Careful disinfection: to kill bacteria performed? Obtain large volume: low concentration of bacteria in blood Do not use old CVC/PVC: larger risk of contamination Use two phlebotomy sites: helps distinguish contamination Obtain two bottles at each phlebotomy site: aerobe/anaerobe and large volume Start with blood culture at multiple testing to avoid contamination DESCRIBE the disinfection routine for phlebotomy for blood culture Blood cultures are obtained to detect bacteraemia Disinfection of own hands Disinfection of bottle tops Careful disinfection of skin Let the skin dry completely Use of gloves No palpation of vein after disinfection 0% 0% 0% Percent correct answers. Downloaded from on December, 01 by guest

27 APPENDIX Questionnaire Do routines differ between regular blood sampling and blood culture? Yes No How do routines differ? Disinfection Blood amount PVC/CVC Two sampling sites Two bottles Start with blood culture New PVC Other: Other: Why are blood cultures obtained? To detect bacteria in blood Other: How Why Correct prompted Incorrect prompted Do not know How should disinfection be performed prior to obtaining a blood culture? Disinfect hands with hand disinfection Disinfect bottle tops using chlorhexidine Carefully disinfect patient s skin for 1 minute using chlorhexidine Let the skin dry Use protective gloves Do not palpate the vein again after disinfection Other correct: Other correct: Other correct: Downloaded from on December, 01 by guest