INTRODUCTION
Sepsis, the systemic inflammatory response to infection, continues to be a major cause of morbidity and mortality in critically ill patients, resulting in frequent diagnostic testing, increased antibiotic administration and prolonged length of hospitalization [Reference Shafazand and Weinacker1, Reference Ak2]. The major health and economic impacts of bacteraemia highlight the importance of early detection and treatment of healthcare-acquired infections (HAIs), particularly in critically ill patients [Reference Shafazand and Weinacker1–Reference Akhtar3]. Blood cultures are the most important and most commonly performed tests used to identify the presence of bacteria in the bloodstream. When properly performed they enable therapy to be directed against the causative microorganism, helping to reduce mortality and the selection of multiresistant bacteria [Reference Hall and Lyman4, Reference Bamber5]. However, blood culture contamination (BCC) leading to false-positive blood cultures, with microorganisms which were not actually present in the bloodstream can limit the value of this approach.
Rates of BCC vary widely between healthcare settings and have been estimated to range from 1·5% to >9% in several teaching hospital-based studies [Reference Weinbaum6–Reference Hall10]. Although it has been recommended that target rates for BCC should not exceed 3% [Reference Ernst11], in many institutions it actually exceeds 7% [Reference Ernst11, Reference Roth12]. Blood cultures are generally incubated for 5 days for bacterial growth but positive test results are usually available within 48 h. If culture results are negative, administering antibiotics may be unnecessary and other potential diagnoses can be considered [Reference Weddle13]. When a positive culture is recognized as a likely contaminant, the usual practice is to obtain additional blood cultures which adds to costs and potentially delays targeted treatment of the patient [Reference Bryant and Strand14, Reference Bates, Goldman and Lee15]. In the interests of patient safety, many physicians will treat patients once any growth in the blood culture is observed, with the potential for continuing empirical treatment, especially in immunocompromised or critically ill patients [Reference Ruge16].
An earlier prospective case-control study at the Antrim Area Hospital to determine the clinical and economic impact of contaminated blood cultures showed increased lengths of hospital stay and total costs per patient of 5·4 days and £5002, respectively [Reference Alahmadi17]. Moreover, the study showed a particularly high rate (13·6%) of false-positive cases in the intensive care unit (ICU). These findings provided the rationale for targeting the ICU to assess the impact of a staff educational intervention programme (a poster, DVD, monthly feedback of BCC rates) on decreasing BCC rates to a target value of <3% in the ICU at Antrim Area Hospital.
METHODS
The project was designed in the form of an audit of blood culture-taking practice within the hospital, as part of a continuing quality improvement programme. Approval was obtained from the Northern Health and Social Care Trust (NHSCT) research governance committee, and the study was registered within the Trust research and development department.
Study setting
The study was conducted in the ICU at the Antrim Area Hospital in Northern Ireland, a 426-bed district general teaching hospital serving a population of ~420 000. The hospital provides all acute, general medical and surgical services, supports a range of outpatient facilities and acts as a centre for the coordination of health service provision throughout a defined geographical area in Northern Ireland. The ICU consists of eight beds (adult general medical and surgical), and admits about 35 patients per month.
Definitions
BCC was defined using a predictive model that is part of the standard operating procedure of the Microbiology Laboratory. The laboratory standard for blood cultures is a minimum draw of two sets of blood cultures from separate peripheral venous puncture sites in the same time-frame. For intravascular devices the standard is a minimum draw of two sets of blood cultures from separate peripheral venous puncture sites followed by drawing blood from each of the lumens of intravascular catheters in the same time-frame; in practice the laboratory usually only receives samples from a single catheter lumen. The multiple sets of blood cultures required by the standard is to increase the volume of blood cultured, thereby increasing the chance of recovering the pathogen as in adult bacteraemia there may be <1 organism/ml of blood sampled. In addition, the laboratory uses the multiple draws to establish the significance of any organism isolated. Bacteria and yeasts are classified into groups: (i) organisms that always or nearly always represent true infection (⩾90%) and (ii) common contaminants but which may be significant in specific clinical circumstances. The number of blood culture sets that grow a particular microorganism, when measured as a function of the total number of blood culture sets obtained is widely accepted to be a useful tool in the interpretation of the clinical significance of positive blood cultures. If organisms from the first group are isolated from any blood culture set they are considered to be clinically significant. However organisms form the second group are only considered to be clinically significant if they are isolated from a minimum of two blood culture sets and they yield organisms of the same biochemical profile and antibiogram; otherwise they are considered to be likely contaminants and reported as ‘possible contamination please repeat if clinically indicated’. If the contamination rate is 3%, then the probability that identical organisms recovered from two sets of blood cultures from a patient are contaminants is <1/1000 (0·03 × 0·03 = 0·0009) [Reference Hall and Lyman4, Reference Weinstein18, Reference Stohl19].
Data collection
In order to determine the BCC rate, the number of contaminated blood cultures and the total blood cultures taken in the ICU on a monthly basis were recorded through evaluation of microbiology laboratory records (i.e. retrospectively during the period from January 2006 to August 2010 and prospectively during the period from September 2010 to December 2012). This latter time period in the study includes evaluation of blood culture request form completion from September 2010 to March 2012 and the educational intervention from June 2011 to December 2012.
To assess the impact of the intervention on the appropriate completion, by staff, of the information requested on the blood culture request forms, the following details were recorded during the second period for each form completed: patient's name, gender, hospital number, date of birth, ward, time and date sample taken, clinical details, current antibiotic use, sample site and signature of person taking the sample, and expressed as the percentage of fully completed request forms.
Study design
The impact of the educational intervention on the BCC rate was evaluated by comparing the rate of BCC during the pre-intervention period (January 2006 to May 2011) with the rate during the educational intervention (i.e. June 2011 to December 2012; Fig. 1).
Fig. 1. Study design. Impact of intervention on blood culture contamination (BCC) rate and completion of blood culture (BC) request forms timeline.
The educational intervention
The educational intervention was designed to increase the awareness of ICU clinical staff of BCC and the use of proper techniques for taking blood culture samples. This consisted of a poster (Supplementary Fig. S1), a 13-min structured DVD on blood culture sampling (produced by NHSCT Antimicrobial Management Team and Iskus Health; available at www.iskushealth.com) and the provision of monthly feedback to ICU staff via email on BCC rates with a run chart of the results which was displayed on the ward and updated on a monthly basis.
The DVD and poster were devised based on the Trust's policies and procedures for optimal blood culture collection practices. The poster was displayed on the main noticeboard in the ICU throughout the educational intervention from June 2011. A register of all ICU staff involved in taking blood cultures was established (16 physicians, 54 nurses), and each individual was requested to view the DVD in September 2011 or November 2011. The DVD focused on (a) the definition of BCC; (b) the targeted contamination rate (i.e. <3%); (c) the increased healthcare costs due to BCC; (d) the most common organisms involved in the contamination of blood cultures; and (e) the proper technique for taking blood culture samples. The latter was consistent with the seven steps outlined in the poster presentation, i.e. gather components; wash hands; wipe rubber septum with alcohol swab; decontaminate hands with alcohol gel; disinfect patient's skin with ChloraPrep® (CareFusion, UK); transfer blood to the blood bottles; send labelled bottles with fully completed request form to laboratory. Feedback of results to ICU staff began in December 2011.
Economic evaluation
The estimated potential annual cost savings were determined based on the analysis previously published from this hospital [Reference Alahmadi17] and included antibiotic, laboratory, and daily hospital costs (‘hotel’ charges) for each directorate in the hospital.
Statistical analysis
Non-normal distribution data were analysed using the Mann–Whitney U test with continuous variables expressed as medians and interquartile ranges (IQRs), and Pearson's χ 2 test with categorical variables expressed as percentages. A run chart was constructed showing the monthly BCC rate and the contamination rate was determined from the number of BCCs × 100/number of total blood cultures taken. The impact of the implemented intervention was evaluated utilizing the analysis of segmented regression of an interrupted time-series [20]. This allowed the estimation of the intervention effect while taking account of time trend and autocorrelation existing in consecutive observations. Data for the period June 2011–December 2011 were omitted from the analysis as the full effect of the intervention was not present during this period. Data were coded as described elsewhere [20] (0, before educational intervention; 1, post-educational intervention), and a P value of <0·05 was considered to be statistically significant. All statistical analyses were performed using SPSS software v. 20·0 (SPSS Inc., USA).
RESULTS
During the whole period (i.e. both pre-intervention and during the educational intervention, January 2006 to December 2012) 6421 blood culture sets were obtained from the ICU, of which 508 were categorized as contaminated. During the baseline phase (to May 2011), 5002 blood culture sets were obtained, of which 451 (9·0%) were categorized as contaminated. By contrast, during the intervention period (June 2011 to December 2012), 1419 blood culture sets were obtained, of which 57 (4·0%) were categorized as contaminated.
Impact of the intervention
The analysis of time series showed that the educational intervention (R 2 = 0·27) impacted upon the BCC rate, with a level change reduction (coefficient = −4·7, P = 0·034) being observed 2 months post-intervention. The run chart (Fig. 2) shows a reduction in the monthly BCC rates during the educational intervention period to a mean of 3·7%, compared to 9·5% during the baseline period. There was no significant difference, using the Mann–Whitney U test, between the monthly number of blood culture sets obtained during the pre-intervention (median 72, IQR 61–84) and educational intervention (median 72, IQR 55–86) periods (P = 0·63).
Fig. 2. Run chart of monthly blood culture contamination (BCC) rates over the whole study period (ICU; January 2006–December 2012), completion of blood culture request forms during pre- and educational intervention period (September 2010 to May 2011 and June 2011 to March 2012).
Request form completion rates during the pre-intervention and educational intervention periods were almost identical for the following parameters: patient's name, gender, date of birth, hospital number, ward name, signature of person taking the sample, and date and time sample taken. There was, however, a significant increase in the percentage of fully completed forms during the educational intervention period in relation to sample site, clinical details and current antibiotic use (P = 0·008, 0·006, <0·001, respectively). Absolute improvements in the completion of those variables were 8·0%, 7·4% and 14·2%, respectively, as shown in Table 1.
Table 1. Comparison of proper completion of blood culture request forms during pre-intervention and educational intervention period (September 2010 to May 2011 and June 2011 to March 2012)
* Pearson's χ 2 test.
Sites associated with BCC
Data for this aspect of the study were collected over a 19-month period (September 2010 to March 2012). Of the 1509 blood culture sets obtained during this period, 619 (41·0%) were taken via peripheral venepuncture, 503 (33·3%) from a central venous catheter, 134 (8·9%) from a dialysis line, and 52 (3·4%) from an arterial line. A total of 201 (13·3%) sample sets did not have the site recorded. Among the false-positive blood culture results, the most common sample site was the central venous catheter (54·5%), while the least common site was the arterial line (3·4%). The peripheral venepuncture site was the most common site for both the true-positive and negative blood culture results (43·7% and 41·3%, respectively). Table 2 shows that blood cultures obtained via a central venous catheter yielded a contamination rate of 48/503 (9·5%), compared to 30/619 (4·8%) for peripheral venepuncture. Moreover, according to the odds ratio (OR) values, blood culture samples obtained through a central venous catheter were more than twice as likely to be contaminated [OR 2·1, 95% confidence interval (CI) 1·3-3·3] than samples taken via peripheral venepuncture. Samples taken via a dialysis line and arterial lines showed no significant relationships with BCC (Table 2).
Table 2. Relationships between blood culture sample sites and BCC (September 2010 to March 2012).
BCC, Blood culture contamination; OR, Odds ratio; CI, confidence interval.
Economic analysis
Using data collected in the previous study in Antrim Area Hospital [Reference Alahmadi17], it was found that patients with false-positive blood cultures had additional hospital costs of £5002 compared to the patients with negative blood cultures. A median of 72 blood cultures were taken in the ICU per month during the pre-intervention and educational intervention periods, and false positives before and after the intervention were 9·5% and 3·7%, respectively. Using these data the estimated potential annual cost saving was approximately £250 100 (Table 3).
Table 3. Estimated economic benefit flowing from the educational intervention
ICU, Intensive care unit.
* Based on a previous study in Antrim Area Hospital [Reference Alahmadi17].
DISCUSSION
The rate of contamination of blood cultures should not exceed 3% according to the standards of the American Society for Microbiology [Reference Ernst11, Reference Roth12]. Previously, in the study site hospital, the BCC rate was found to have a mean value of 4·7% [Reference Alahmadi17], with a high percentage of the false-positive cases occurring in the ICU (BCC rate 13·6%).
The intervention put in place in the current study aimed to reduce the BCC rate in the ICU of the study site hospital to the recommended standard. It is a particular challenge to reduce BCC in an ICU as previous studies have demonstrated that BCC rates are generally higher in such units compared to general wards, due to the high workload of staff, high severity and urgency of illness, and poor vascular condition of patients [Reference Kim9, Reference Little21, Reference Lee22]. Therefore, only a few studies have attempted to decrease BCC rates in adult ICUs through educational interventions [Reference Larkin8, Reference Roth12, Reference Robert23]. The run chart approach was used in the present study to determine the effect of the educational intervention as it is considered to be of value for application in healthcare quality development [Reference Perla, Provost and Murray24].
Our findings indicate that the introduction of an educational intervention alone did significantly reduce BCC rates within an ICU in a typical medium-sized hospital in the UK. The rate of contaminated blood cultures was significantly reduced from 9·5% to 3·7%, and the rate fell below the ‘gold standard’ target of 3% for a 7-month period during the 19-month educational intervention period. Moreover, the completion of the blood culture request forms was improved. As expected, it was found that the blood culture samples obtained via central venous catheters were more frequently associated with BCC.
The overall educational intervention, i.e. the poster, DVD and monthly feedback of BCC rates was developed and introduced to ensure there was minimal disruption to the work and staff in the ICU. The DVD was presented to the ICU staff for them to view at a time that suited them best and they were asked to sign a form to confirm that the DVD had indeed been viewed. This resulted in making the intervention more flexible and less costly, compared with other educational interventions, which often need to be performed on a one-to-one basis [Reference Gibb25, Reference Eskira26]. Moreover, the educational intervention focused on improving staff knowledge regarding the proper techniques for taking blood culture samples without changing hospital guidelines or patient services. Other previous studies, in combination with educational interventions, have sought to change guidelines for phlebotomy practice [Reference Roth12], introduce new disinfectants [Reference Larkin8], or adopted new blood culture kits [Reference Bamber5].
Two to three blood culture sets per episode is recommended for the optimal detection of bloodstream infection according to the blood culture standards of the Clinical and Laboratory Standards Institute (CLSI) [Reference Wilson27]. A previous study aimed at reducing BCC rates through an educational intervention and the introduction of new blood culture collection kits, found that the BCC rate decreased, but this was coupled with a reduction in the total number of blood culture sets collected. The latter outcome was suggested to reflect the fear by junior physicians of the possible consequences if they were held responsible for contamination [Reference Thomas28]. The monthly blood culture numbers did not change over the study period. The blood culture request form in the study site hospital contains important information that may be used as an aid by the microbiologists in their interpretation of possible BCC (e.g. sample site, clinical details). Thus, completion of blood culture request forms was used as an indicator of the educational intervention on the awareness of the ICU staff and as such is, to date, the first study to document this outcome measure. Indeed, completion of the forms was found to have improved during the intervention.
The percentage of contaminated blood cultures obtained via central venous catheters (over a 19-month period), was almost double that via peripheral venepuncture (9·5% vs. 4·8%) which is consistent with data from other studies [Reference Bryant and Strand14, Reference Bates, Goldman and Lee15, Reference Everts29]. A recent study compared contamination rates of blood cultures obtained by central venous catheters with samples from peripheral venepuncture over 5.5 years in two ICUs (general and medical) and also found central venous catheter cultures to be more frequently contaminated than peripheral blood cultures (8% vs. 4%) [Reference Little21]. The ICU is among the most expensive wards in hospitals and ICU patients may have long hospital stays. Contamination of blood cultures will increase this cost by extending length of stay inappropriately. We have demonstrated that a simple bespoke educational intervention to reduce BCC rates can result in significant cost savings (Table 3). In addition to the benefits of more appropriate use of resources, significant improvements in patient care and wellbeing are obtained through reductions in unnecessary antimicrobial usage and length of hospital stay, which reduce the risk of patients acquiring HAIs and the development of resistance.
This study has some limitations. First, the educational intervention was carried out in a single ICU and might have benefited if a multicentre trial had been possible. Second, there was no evaluation (e.g. a questionnaire survey) of staff awareness levels regarding adequate blood culture sampling techniques before and after the educational intervention. However, this was indirectly evaluated by documenting the percentage of blood culture request form details that had been comprehensively completed. Moreover, observation of the outcomes of blood culture sampling is considered a more effective method by which to evaluate staff awareness [Reference Gibb25, Reference Eskira26]. Third, data on blood culture drawing sites, determined before and after the intervention, was not available for analysis and it was not possible to assess differences in BCC rates accordingly. Last, although the evaluation of the impact of the intervention might have benefited from a longer follow-up period, the present analysis included sufficient numbers of observations post-intervention.
In conclusion, the educational intervention approach used in this study addressed the problem of false-positive blood cultures in hospital to a level that is consistent with best practice in the area. The observed decrease in BCC rates has important clinical and economic implications. Importantly, the approach was simple in design, flexible in delivery and efficient in outcome, and thus may encourage the translation of this approach into clinical practice in different healthcare settings.
SUPPLEMENTARY MATERIAL
For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S0950268814003008.
ACKNOWLEDGEMENTS
The authors acknowledge the staff at the Department of Medical Microbiology at NHSCT for their input.
DECLARATION OF INTEREST
Iskus Health Ltd partly contributed to the development of the DVD which was used as part of the overall educational intervention.
