Introduction
Blood cultures (BCx) are commonly ordered in hospitalized patients for workup of fever, leukocytosis, or other abnormal clinical signs, yet most BCx (85%–95%) will not grow an organism, and among those that are positive, a significant proportion will demonstrate contamination with skin flora. Reference Doern, Carroll and Diekema1–Reference Woods-Hill, Koontz and King4 Studies have estimated that up to 60% of BCx collected in hospitalized adults might not be clinically indicated. Reference Fabre, Klein and Salinas5,Reference Siev, Levy and Chen6 Conversely, a recent study of BCx utilization in Israel, estimated that ∼24% of adult bloodstream infections (BSIs) were missed due to single BCx, lack of anaerobic bottles, or BCx not being performed. Reference Temkin, Biran, Braun, Schwartz and Carmeli7 These data emphasize the need for BCx diagnostic stewardship.
Ordering of BCx is influenced by both clinical (eg, clinical stability and underlying conditions of patients) and nonclinical, socio-behavioral, or process-related factors such as the ordering provider’s healthcare role and years of work experience, shift of the day when BCx are under consideration, and sign-out instructions. Reference Woods-Hill, Koontz and King4,Reference Fabre, Milstone, Keller, Carroll and Cosgrove8–Reference Howard-Anderson, Schwab, Chang, Wilhalme, Graber and Quinn11 Previous surveys have suggested prescribers are unfamiliar with appropriate BCx indications. Reference Fabre, Milstone, Keller, Carroll and Cosgrove8 Little is known regarding providers’ knowledge or practices related to BCx collection. Our objective was to understand healthcare workers’ (HCWs) perceptions toward the utility of BCx in different clinical scenarios, their practices related to BCx ordering and collection, and the potential facilitators and barriers to stewardship interventions to improve BCx stewardship.
Methods
Study design and settings
A mixed-methods study with a convergent parallel design was conducted as part of a larger QI project to optimize BCx practices among inpatient adults. We recruited 8 teaching hospitals through the Centers for Disease Control and Prevention’s Prevention Epicenters Program and the Society for Healthcare Epidemiology of America (SHEA) Research Network to join the collaborative to implement a BCx stewardship QI project based on a prior pilot study. Reference Fabre, Klein and Salinas5 Each participating hospital convened a local QI team composed of 1 or 2 individuals with a leadership position in either the antimicrobial stewardship or the infection control programs as well as 1 or more unit project champions/thought leaders from targeted ICUs and wards (eg, unit director, resident physician, nurse leader).
Johns Hopkins University (JHU) served as the coordinating center for all sites in the collaborative. The Johns Hopkins Medicine Institutional Review Board approved this work as Human Subjects Research under expedited review. Oral consent was required for the interviews. Participation in the survey was taken as consent.
Survey development and administration
A 31-item survey was developed using the Qualtrics survey system (Qualtrics, Provo, UT) to assess HCWs’ knowledge regarding BCx indications, ordering and collection practices and perceptions about these practices, and perceived barriers to improving BCx utilization. A JHU team with expertise in infectious diseases (ID), QI, and diagnostic stewardship developed the survey, which was piloted by 3 ID physicians and modified based on feedback. Site leads received the final survey through an electronic link and distributed the link to key stakeholders including ordering providers and non-ordering HCW roles. Additionally, the project site leads posted QR codes that linked to the survey on common areas in units of interest. A minimum of 30 HCW responses were requested from each site. Surveys were distributed between March 8 and December 4, 2023.
Semi-structured interviews
During baseline assessment, we conducted a group interview with each participating hospital QI team as well as relevant stakeholders such as trainees, bedside nurses using an interview guide developed by JHU team based on the Consolidated Framework for Implementation Research framework. Questions focused on motivation to participate in a BCx stewardship QI, perceptions regarding existing BCx practices, prior/current efforts to improve BCx utilization, and barriers to a BCx stewardship QI intervention (Supplementary Material).
Data analysis
Survey data were analyzed using descriptive statistics to summarize respondents’ answers by unit (intensive care unit [ICU] vs non-ICU), hospital, or healthcare role. Questions with a 5-point Likert scale were categorized into 3 groups by combining “strongly agree” and “agree” or “almost always” and “often” or “very likely” and “likely” and by combining “disagree” and “strongly disagree” or “unlikely” and “very unlikely” or “rarely” and “almost never”. Reference Woods-Hill, Koontz and King4 Separately, we conducted a qualitative content analysis of the interview data. Reference Graneheim and Lundman12 Through an iterative process, 2 JHU study team members (AX, VF) independently reviewed each interview transcript to identify recurring phrases, ideas, and concepts and discussed to achieve consensus and create common themes. Two additional JHU study team members (SEC and AMM) reviewed and provided feedback on the node structure to ensure the credibility, dependability, and confirmability of the analysis. Reference Mays and Pope13,Reference Devers14 Quantitative survey data and qualitative interview data were then triangulated to better understand HCWs’ knowledge, attitudes, and perceptions regarding BCx practices. For each section, we present survey results first followed by the interview findings.
Results
Cohort characteristics
Characteristics of survey and interview participants are displayed in Table 1. Three-hundred and fourteen HCWs from 8 hospitals located in California, Nebraska, Michigan, Tennessee, Pennsylvania, Maryland, Virginia, and North Carolina took the electronic survey; 282 reported their affiliation with a median number of respondents per hospital of 35 (range 16–58). Most survey participants were physicians (67.4%, of which 45.8% were physicians in training), followed by nurses (15.6%) and advanced practice practitioners (14.5%). Most indicated they were involved in ordering BCx only (73.7%), with 8.5% involved in ordering and collecting BCx and 16.7% in collecting BCx only.
Characteristics of survey and semi-structured interview participants
*Includes bedside nurse, phlebotomist, and infection preventionist. **314 individuals took the survey, and 282 completed questions related to demographics.
Semi-structured group interviews were conducted with 36 participants averaging 5 participants per hospital. Most participants were physicians (53% attendings and 8% residents), followed by the grouped category of bedside nurses, infection preventionists, and phlebotomists (31%).
Perceptions about blood culture ordering practices
Half (52.2%) of survey participants perceived that there were too many BCx ordered in their units, and 65.3% agreed that they could safely reduce the number of BCx obtained (responses by hospital are shown in Figure 1 and by unit in Table 2). Most respondents felt that clinicians have a low threshold to order BCx for isolated fever (84.4%) and that not all patients with a new fever should get BCx (72.3%). Fewer agreed that newly isolated leukocytosis triggered BCx in clinical practice (27.7%). There were variations in the temperature threshold respondents used to consider BCx, 45.5% used >100.4°F (38.0°C), 29.8% used >101.0°F (38.3°C), and 12.1% reported different thresholds depending on the patient population. This lack of standardization regarding BCx indications and temperature thresholds to obtain BCx was confirmed by interviewees who reported lower temperature thresholds were used for immunocompromised patients and those requiring extracorporeal membrane oxygenation support (quotes [Q] 1 and 2, Table 3).
Participants’ perceptions regarding blood culture (BCx) use and safety of reducing BCx use in their units by participating hospitals (n = 314 survey respondents).
Healthcare workers’ perceptions regarding blood culture indications and practices. The table shows the proportion of respondents who strongly agreed/agreed with the statement stratified by unit (n = 314)
Quotes from interviewed healthcare workers regarding perceived practices related to blood culture (BCx) ordering and collection
Q, quote.
Regarding the decision to order BCx, most reported a combination of practices (eg, making decisions on their own, following sign-out instructions, consulting institutional guidelines). Although 74.1% of respondents indicated decisions regarding BCx were made as a team (74.1%), only 42.4% indicated these discussions occurred in daily practice. Notably, interviewees overwhelmingly agreed decisions to order BCx were predominately made by trainees (Q3, Table 3).
Factors influencing blood culture ordering
Most survey respondents indicated clinicians usually review the electronic medical record (EMR) (74.8%) but do not necessarily evaluate the patient (78.7%) prior to ordering BCx. Interviewees described bedside evaluation of the patient prior to ordering a BCx as particularly challenging at night due to the cross-covering provider being responsible for many patients. Interviewees felt less experienced trainees were more likely to order unnecessary BCx (Q4, Table 3).
Regarding HCW knowledge of BCx indications, 49% of survey participants received training on when to draw BCx. Although most survey participants correctly responded cellulitis was unlikely to be associated with bacteremia (69.0%), a lower proportion correctly indicated meningitis was likely associated with bacteremia (52.8%) (Suppl. Table 2). Most survey respondents correctly agreed that all patients with Staphylococcus aureus bacteremia but not all patients with uncomplicated E. coli bacteremia required repeat BCx to document bacteremia resolution (80.0% and 64.3%, respectively). A minority of survey participants had received feedback regarding BCx utilization for their units (21.7%), and more than half were unaware if there was an institutional policy addressing fever workup or BCx indications (Table 2). During interviews, only 3 institutions reported the existence of some guidance regarding BCx indications (eg, when to repeat for uncomplicated gram-negative bacteremia, when to order BCx for fever workup). Regarding nursing involvement in BCx ordering decision-making, interviewees reported that it was usually limited to specific scenarios such as upon activation of sepsis alerts in the EMR but that there had been increasing engagement of nursing in BCx stewardship (Q5 and Q6, Table 3).
Perceptions about blood culture collection practices
Survey responses related to BCx collection practices are presented in Tables 2 and 4. Only 59% of survey participants agreed that 2 sets of BCx increased the likelihood of bacteremia detection, and 32.5% correctly identified 8–10 mL as the recommended volume to fill in each bottle. Half of the survey participants agreed that central-line BCx were more likely to yield BCx contaminants compared to BCx obtained through peripheral venipuncture, and ∼10% reported central-line BCx rarely occurred. Interview participants shared their institution’s approach to minimize central-line BCx including requiring prior authorization and restricting to specific circumstances (Q7, Table 3), although most did not track the number of central-line BCx in clinical units, usually due to unreliable documentation of BCx source. They reported variation in their institutions’ approach to diagnose catheter-related BSIs (eg, 2 sites performed cultures of catheter tips, and 1 site performed time to positivity) as well as increased engagement of nurses in discussing opportunities to draw peripheral BCx when central-line BCx were ordered (Q8, Table 3).
Knowledge and practices related to blood culture (BCx) collection practices, stratified by BCx collection role (n = 279)
While 65.6% of survey respondents reported familiarity with strategies to prevent BCx contamination (BCC), most were unaware of the recommended BCC threshold in the United States (65.7% were unsure, and 11% selected a threshold other than 3% or 1%). Interviewees reported BCx were usually obtained by either phlebotomists or nurses, depending on patient location, and most hospitals provided collector-specific feedback regarding BCC results with re-training for repeat BCC. Interviewees reported a range of interventions to reduce BCC such as dedicated kits and blood diversion devices, although limited compliance with new protocols often compromised the sustainability of interventions.
Perceived barriers and strategies to improving blood culture utilization
Barriers to reducing unnecessary BCx cited most frequently by survey respondents included concern for missing an infection (82.9%), lack of guidelines for when BCx are indicated (55.4%), and not receiving feedback on BCx utilization (51.2%). Creation of an electronic BCx algorithm in the EMR was perceived as the potentially most helpful strategy to improve BCx indications (selected by 60.6% of survey respondents), followed by integration of a BCx algorithm in institutional treatment guidelines (57.1%), and a bacteremia risk calculator based on patient data (55.1%) (Suppl. Table 4).
Interviewees reported frequent trainee turnover as a major challenge to sustainable changes in BCx practices (Q9, Table 3). Although they viewed a BCx algorithm very positively, they had concerns about its use at the point of care as trainees prefer clinical-decision support (CDS) tools in the EMR and most QI teams reported building CDS tools in the EMR can take a long time (Q10 and Q11, Table 3). Although most hospitals had prior experience with diagnostic stewardship initiatives (most commonly around urine cultures or Clostridioides difficile testing), interviewees found BCx stewardship to potentially be more challenging due to competing priorities such as the need to quickly obtain BCx to meet the Centers for Medicare & Medicaid Services SEP-1 core measure quality metric (Q13 and Q14, Table 3) and lack of incentives to avoid potential harm from unnecessary test utilization. They also viewed a multicenter QI intervention as an important nudge to implement a BCx stewardship program at their institutions. The primary reasons to participate in a BCx stewardship multicenter collaborative were the impact on patients (eg, avoiding adverse events associated with BCC) (Q15, Table 3), the potential to decrease central-line-associated BSI (CLABSI) (Q16, Table 3), and the need to protocolize BCx ordering (Q17, Table 3) and to reduce unnecessary nursing workload.
Discussion
Although studies have identified a need to improve when BCx are ordered and how they are collected, Reference Doern, Carroll and Diekema1,Reference Fabre, Carroll and Cosgrove2,Reference Warren, Yarrington, Polage, Anderson and Moehring15 little is known about HCWs’ knowledge, perceptions, and practices related to BCx ordering and collection for hospitalized adults. We surveyed 314 and interviewed 36 HCWs working in adult ICUs or adult wards from 8 large academic hospitals from 4 regions in the United States. We found many HCWs did not understand when BCx are useful or what are the collection parameters that improve BCx sensitivity. Furthermore, there was a discrepancy in how HCWs perceived opportunities to improve BCx indications and whether BCx were overused. A summary of knowledge and practice gaps and barriers to improving BCx use is summarized in Table 5.
Summary of knowledge gaps, variability in practices, and commonly perceived barriers identified through the survey and semi-structured interviews conducted among healthcare workers (HCWs)
Diagnostic stewardship aims to optimize clinical management and healthcare resources. Reference Fabre, Davis and Diekema16 The increasing rates of healthcare products and drug shortages in recent years related to climate change highlight the importance of stewardship initiatives.
Implementation of evidence-based algorithms to guide clinicians’ decisions to order BCx, reduced BCx utilization by 20%–40% without increased mortality or readmissions in both adult and pediatric patients in US centers; Reference Fabre, Klein and Salinas5,Reference Wang, Zhou and Shay17–Reference Woods-Hill, Colantuoni and Koontz19 however, large-scale multicenter QI projects to improve BCx utilization in adult patients have not been conducted. Understanding HCWs’ knowledge and practices about BCx can inform the development of effective interventions. In this survey, participants showed a strong perception that BCx could be safely reduced although overuse of BCx was felt to be less of a problem, despite most participants believing BCx are ordered reflexively for fever evaluation, a common driver of BCx orders. The perception of BCx overuse ranged from 39% to 64% among participating institutions, which is lower than the 75% estimate reported by hospitals in Switzerland. Reference Drager, Giehl and Sogaard20 This may be due to variation in practices as BCx inappropriateness based on indication can vary significantly between hospitals. Reference Fabre, Klein and Salinas5,Reference Siev, Levy and Chen6 Factors that may be contributing to this discrepancy in how HCWs perceive BCx use in their units and HCWs’ approach to fever may be multifactorial and related to their uncertainty about appropriate indications for BCx, unawareness of local BCx utilization rates, or response bias as most survey respondents were ordering providers. We also found knowledge gaps regarding parameters that influence BCx sensitivity (eg, adequate blood volume per bottle, number of sets, factors increasing blood culture contamination), even among those who are only involved with the collection of BCx. A major limitation to improving HCW knowledge around these issues according to interviewees was frequent staff turnover. Lack of professional society guidance on BCx indications likely contributes to HCW’s knowledge gap in BCx indications. Incorporating education in pre-graduate training may help overcome this challenge. Our findings also emphasize opportunities to enhance multidisciplinary discussions about BCx, including better integration of nurses and the microbiology laboratory in BCx stewardship initiatives. Reference Monsees, Tamma, Cosgrove, Miller and Fabre21
Although several studies have shown a poor correlation between isolated fever or leukocytosis and bacteremia, Reference Fabre, Sharara, Salinas, Carroll, Desai and Cosgrove3 participants highlighted more data on special populations such as immunocompromised patients were needed to reduce unnecessary BCx in these populations. Participants identified guidance built in the EMR as the most potentially useful strategy to help guide BCx ordering decisions, yet they also acknowledged oversensitive EMR alerts for sepsis were perceived drivers of unnecessary BCx use as many patients meeting sepsis criteria have noninfectious conditions, up to 75% based on published literature. Reference Rhee, Chiotos and Cosgrove22 Another favored potential strategy was the development of a bacteremia risk score. Notably, bacteremia prediction tools have been developed for use in the emergency department setting but have not been widely implemented. Reference Fabre, Carroll and Cosgrove2 A recent study from Japan compared attending physicians’ gestalt with 2 existing bacteremia prediction tools in adults hospitalized with a suspected infection. Reference Fujii, Takada and Kamitani23 Although physicians overestimated the probability of bacteremia compared to the prediction tools, they were less likely to miss a bacteremia case (22% of patients were identified as low risk of bacteremia by physician gestalt vs 53% by the Shapiro prediction tool with 4% of patients identified as low risk of bacteremia by physician gestalt vs 6% by the prediction tool developing bacteremia). Hence, while clinical-based algorithms are helpful, risk-score tools might be more efficient in reducing unnecessary BCx. More research is needed to define the utility of scoring tools based on users’ years of clinical experience and in evaluating hospital-onset events.
We found variations in how institutions approached the diagnosis of CLABSI, central-line blood draws, and prevention of BCC. A previous study evaluating BCx practices among patients with central lines at a university hospital in Iowa showed central-line BCx were infrequent, and those who received them were more likely to have positive BCx for skin contaminants than those who did not. Reference Kovoor, Kobayashi and Sheeler24 These findings emphasize the need to better define the role of central-line BCx and their potential implications in patient care and hospital metrics in future updates to guidelines.
There are limitations to this study. Participating institutions were academic centers which may limit generalizability to non-teaching settings; however, they were located across the 4 US regions providing diversity to the sample. We were unable to calculate a response rate due to the distribution method (a link to the survey was both shared via email to relevant stakeholders and posted as QR codes in common areas in units of interest), which may have also led to selection bias. Additionally, 10% of respondents did not indicate their institution; however, based on the 282 respondents for which we had institution affiliation, 5 of the 8 participating institutions contributed a minimum of 30 responses per site. The target audience for this survey was unit staff in ICUs and wards, phlebotomists and HCWs from other areas with high BCx utilization such as emergency medicine and oncology were under and not represented, respectively. Finally, the study preceded the recent national BCx bottle shortage, which may have changed HCWs’ perceptions and knowledge about BCx.
In summary, using mixed methods, we evaluated knowledge, perceptions, and practices related to BCx ordering and collection among HCWs of adult ICUs and wards from 8 academic centers in the US. We identified several opportunities for improvement including increasing HCWs’ knowledge of BCx indications and collection practices, improving access to BCx quality indicators data such as BCC, and standardization of indications through CDS tools.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ice.2024.208.
Acknowledgments
We thank Zunaira Virk for assistance with the coordination of this project. We thank the SHEA Research Network for facilitating the recruitment of hospitals for participation in the project.
Financial support
This work was supported by the Centers for Disease Control and Prevention’s Prevention Epicenters Program (grant nos. 6U54CK000617-01-01, 5U54CK000617-02-00, and 5U54CK00617-03-00). The content is solely the responsibility of the authors and does not necessarily represent the official view of the funding agency.
Competing interests
All authors report no conflicts of interest relevant to this article.
