Advertisement

Do we need blood culture stewardship programs? A quality control study and survey to assess the appropriateness of blood culture collection and the knowledge and attitudes among physicians in Swiss hospitals

Open AccessPublished:June 14, 2022DOI:https://doi.org/10.1016/j.ejim.2022.04.028

      Highlights

      • Blood cultures were not collected according to local recommendations in 40%.
      • Inappropriate blood culture collection was most frequently observed in non-severe CAP.
      • The true positivity rate was 0% in infections with low pretest probability.
      • Physicians are uncertain regarding the indication of blood culture collection.
      • Implementation of diagnostic stewardship programs may improve blood culture collection practices.

      Abstract

      Background

      Guidance for blood culture (BC) collection is limited. Inappropriate BC collection may be associated with potentially harmful consequences for the patient such as unnecessary laboratory testing, treatment and additional costs. The aim of the study was to assess the appropriateness of BC collection and related knowledge and attitude of precribers.

      Materials

      We conducted a single-center quality control study to assess the appropriateness of BC collection according to the local guidelines in a Swiss university hospital in 2020 by combining three different approaches: point prevalence, patient-individual longitudinal and diseases-related analysis. Second, we conducted a survey regarding BC collection practices and knowledge among physicians in two non-university and one university hospital using an 18-item electronic questionnaire.

      Results

      We analyzed 1114 BC collected in 344 patients. Approximately 40% of the BCs were collected inappropriately, in particular in diseases with low pretest probability of bacteremia such as non-severe community acquired pneumonia (CAP). Follow-up blood culture (FUBC) collection was inappropriate in 60%. Growth of a relevant pathogen was more frequently observed in appropriately than in inappropriately collected BCs (18% vs. 3%, p < 0.001). In the survey, uncertainty concerning the need of index BC collection was high in non-severe CAP and uncomplicated cellulitis.

      Conclusions

      Almost half of the BCs was not collected according to the guidelines, especially in non-severe CAP and in case of FUBCs. Substantial uncertainty among physicians regarding BC ordering practices was identified. The implementation of diagnostic stewardship programs may improve BC collection practices, increase adherence to local guidelines, and may help reducing unnecessary diagnostics and treatment.

      Keywords

      Abbreviations:

      BC (blood culture), BSI (bloodstream infection), CAP (community acquired pneumonia), CI (confidence interval), CRP (C-reactive protein), ED (emergency department), FUBC (follow-up blood culture), GNB (Gram-negative bacteria), ICD-10 (International Classification of Disease 10), ICU (intensive care unit), ID (Infectious diseases), IQR (interquartile range), OR (Odds ratio), REDCap (Research Electronic Data Capture), SARS-CoV-2 (severe acute respiratory syndrome coronavirus type 2), S. aureus (Staphylococcus aureus), SD (standard deviation)

      1. Introduction

      As bloodstream infections (BSI) are associated with a mortality of 14% and up to 35% in critically ill patients, physicians often have a low threshold to collect blood cultures (BC) [
      • Laupland K.B.
      • et al.
      Population-based assessment of intensive care unit-acquired bloodstream infections in adults: Incidence, risk factors, and associated mortality rate.
      ,
      • Muder R.R.
      • et al.
      Bacteremia in a long-term-care facility: a five-year prospective study of 163 consecutive episodes.
      ,
      • Weinstein M.P.
      • et al.
      The clinical significance of positive blood cultures in the 1990s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults.
      ]. But overall, the yield of BCs is low (4–10%) [
      • Bates D.W.
      • et al.
      Predicting bacteremia in hospitalized patients. A prospectively validated model.
      ,
      • Coburn B.
      • et al.
      Does this adult patient with suspected bacteremia require blood cultures?.
      ] and contaminants (e.g. skin flora) may be identified in up to 40% of positive BCs [
      • Hall K.K.
      • Lyman J.A.
      Updated review of blood culture contamination.
      ]. Decision making around BC collection is multifactorial and influenced by the provider's role, the provider's clinical experience, the comorbidities of the patient and the severity of the infection [
      • Fabre V.
      • et al.
      Prescribers' knowledge, attitudes and perceptions about blood culturing practices for adult hospitalized patients: a call for action.
      ].
      BC collection is recommended in infections with a high (≥ 50%) pretest probability for positive results including septic shock, meningitis, endovascular infections, vertebral osteomyelitis and nontraumatic septic arthritis [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ]. Additionally, it is recommended for infection with intermediate pretest probability such as severe community-acquired pneumonia (CAP), pyelonephritis or cholangitis, prior to the start of antibiotic treatment [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ]. Conversely, in patients with isolated fever and fever within the first 48 h after surgery, the probability of obtaining positive BC results is very low (<5%), and low (<10%) in infections such as cellulitis [
      • Torres J.
      • et al.
      Low yield of blood and wound cultures in patients with skin and soft-tissue infections.
      ,
      • van Daalen F.V.
      • et al.
      Clinical condition and comorbidity as determinants for blood culture positivity in patients with skin and soft-tissue infections.
      ], cystitis/prostatitis [
      • Etienne M.
      • et al.
      Should blood cultures be performed for patients with acute prostatitis?.
      ] and non-severe CAP [
      • Bordon J.
      • et al.
      The presence of pneumococcal bacteremia does not influence clinical outcomes in patients with community-acquired pneumonia: results from the Community-Acquired Pneumonia Organization (CAPO) International Cohort study.
      ,
      • Zhang D.
      • Yang D.
      • Makam A.N.
      Utility of Blood Cultures in Pneumonia.
      ]. When pretest probability is very low, BC collection is not recommended [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ]. It should be considered carefully when pretest probability is only low, as the impact of true positive BCs is limited and conversely, false positive BC results may even lead to unnecessary laboratory testing, inappropriate antibiotic treatment, increased healthcare costs and prolonged hospital length of stay [
      • Coburn B.
      • et al.
      Does this adult patient with suspected bacteremia require blood cultures?.
      ,
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ]. Furthermore, unnecessary BCs may contribute to patient discomfort, adverse events associated with antibiotics started for contaminants and hospital-acquired anemia [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ].
      Follow-up blood cultures (FUBC) are BCs drawn after the initially collected 2–3 BCs and are recommended for the management of Staphylococcus aureus (S. aureus) bacteremia [
      • Habib G.
      • et al.
      2015 ESC Guidelines for the management of infective endocarditis: The Task Force for the Management of Infective Endocarditis of the European Society of Cardiology (ESC). Endorsed by: European Association for Cardio-Thoracic Surgery (EACTS), the European Association of Nuclear Medicine (EANM).
      ] and candidemia [
      • Cornely O.A.
      • et al.
      ESCMID* guideline for the diagnosis and management of Candida diseases 2012: non-neutropenic adult patients.
      ] as persistent bacteremia and fungemia are associated with increased mortality and metastatic spread [
      • Cornely O.A.
      • et al.
      ESCMID* guideline for the diagnosis and management of Candida diseases 2012: non-neutropenic adult patients.
      ,
      • Kuehl R.
      • et al.
      Defining persistent Staphylococcus aureus bacteraemia: secondary analysis of a prospective cohort study.
      ]. In patients with BSI due to Gram-negative bacteria (GNB), routine collection of FUBCs to document clearance is not recommended [
      • Canzoneri C.N.
      • et al.
      Follow-up Blood Cultures in Gram-Negative Bacteremia: Are They Needed?.
      ,
      • Wiggers J.B.
      • Daneman N.
      The culture of follow-up blood cultures.
      ]. Studies assessing the appropriateness of BC collection in hospital setting are scarce and mainly focus on BC collection according to a suspected disease [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ], collection techniques [
      • Garcia R.A.
      • et al.
      Multidisciplinary team review of best practices for collection and handling of blood cultures to determine effective interventions for increasing the yield of true-positive bacteremias, reducing contamination, and eliminating false-positive central line-associated bloodstream infections.
      ] or defined appropriateness of BC testing as a bundle of parameters consisting of indication, technique and antibiotic treatment [
      • Vitrat-Hincky V.
      • et al.
      Appropriateness of blood culture testing parameters in routine practice. Results from a cross-sectional study.
      ].
      The aim of our study was to assess the appropriateness of BC collection in a Swiss university hospital according to the guidelines. Furthermore, we intented to assess differences in knowledge and attitudes among physicians regarding BC collection working in a university hospital or in non-university hospitals in Switzerland.

      2. Material and methods

      This project includes (i) a single center, observational, quality control study conducted at the University Hospital Basel, a 750-bed academic, tertiary care center, and (ii) a multicenter survey conducted in three Swiss hospitals. The study protocol was approved by the Ethics Committee of Northwest and Central Switzerland (EKNZ Project-ID 2020–02,738) with a waiver for informed consent.

      2.1 Quality control study

      We performed three separate analyses to comprehensively assess the appropriateness of BC collection and to capture all aspects of BC collection including FUBCs and disease specific differences.

      2.1.1 Point prevalence analysis

      This analysis aimed to assess the appropriateness of BC collection on two different days before and after the onset of the severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) pandemic (15th of January 2020 and 1st of September 2020). A list of all BCs collected on these two days was extracted from the microbiological laboratory system and all adult patients of the University Hospital Basel (in- and outpatients) who had BCs collected on those days were included.

      2.1.2 Patient-individual, longitudinal analysis

      This analysis aimed to assess BC collection appropriateness per individual patient during the whole hospital stay and was conducted in adult, non-neutropenic patients (since guidelines are primarily defined for this patient population): 40 patients hospitalized in medical wards, 40 patients hospitalized in surgical wards and 50 patients presenting to the emergency department (ED) on the day of inclusion were analyzed. The selection of patients started on 15th of January 2020 and on 1st of September 2020 and was continued by including patients on the subsequent days in chronological order until the sample size was reached. A list of BCs within the defined time frame was extracted from the microbiological laboratory information system.

      2.1.3 Appropriateness of BC collection according to a defined underlying disease

      Patients that were hospitalized at the University Hospital Basel due to non-severe CAP, uncomplicated cellulits, cholangitis or pyelonephritis and discharged between the 1st of April and the 30th of September 2020 were extracted from the local hospital information system by the use of the International Classification of Disease 10 (ICD-10) codes (J13, J14, J15.*, J18.0, J18.1, J18.8 for CAP; L03*, A46 for cellulitis; K80.3*, K83.0 for cholangitis; N10 for pyelonephritis) for the assessment of BC collection appropriateness according to these defined diseases.

      2.1.4 Questionnaire

      We conducted a survey regarding BC collection practices, knowledge and attitudes at the University Hospital Basel, at the City Hospital Triemli Zurich, a 400-bed non-university tertiary care center, and at the Hospital Thun, a 270-bed regional hospital. An 18-item electronic questionnaire (Supplementary Table S1) was sent to physicians working in the Division of Internal Medicine, the Department of Orthopedics, the Department of Infectious Diseases and Hospital Epidemiology, the intensive care unit (ICU) and ED at the University Hospital Basel and to physicians working in the Department of Internal Medicine at the City Hospital Triemli Zurich and the Hospital Thun. Participation was voluntary and anonymous. Answers were provided using a 5-point Likert scale. Answers were summarized into two categories (strongly agree/agree and neutral/disagree/strongly disagree) for analysis purposes.

      2.2 Data collection

      Patient, clinical and microbiological data were extracted from the electronic medical records of the hospital clinical information system and the microbiological laboratory information system. For data collection and management and distribution of the survey, Research Electronic Data Capture (REDCap) [
      • Harris P.A.
      • et al.
      Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support.
      ,
      • Harris P.A.
      • et al.
      The REDCap consortium: Building an international community of software platform partners.
      ] was used. All data were entered by a study physician and double-checked by a board-certified infectious diseases (ID) specialist.

      2.3 Definitions, assessment of appropriateness and endpoints

      One BC was defined as one pair of BC bottles consisting of one aerobic and one anaerobic bottle (BactAlert FA/FN plus, bioMérieux, Marcy l'Etoile, France). The local protocol recommends to collect each BC from a separate venipuncture and to separate BC collections by at least 15–30 min.If growth of a pathogen was observed in one of the bottles, the BC was considered positive. Index BCs were defined as two BCs collected within a 2-hour interval in the context of a new event. A third BC was also considered as index BC if it was collected within 24 h of the first two BCs in the absence of a new event. A new event was defined as (i) first presentation of a patient in the ED or (ii) alteration of the clinical presentation of a hospitalized patient (e.g. new onset of fever, cough or hypotension) or (iii) change in working diagnosis leading to BC collection. FUBCs were defined as one or two BCs collected at least 48 h after the collection of the index BCs until a new event occurred. Growth of a relevant pathogen was defined as growth of a pathogen other than contaminants (e.g. coagulase negative staphylococci) as judged by the authors [
      • Elzi L.
      • et al.
      How to discriminate contamination from bloodstream infection due to coagulase-negative staphylococci: a prospective study with 654 patients.
      ]. For the calculation of the true positivity rate only positive BCs with growth of a relevant pathogen were included. Patients were defined as immunocompromised if they met one or more of the following criteria: treatment with corticosteroids (prednisone equivalent of >10 mg per day for ≥4 weeks), biologicals (e.g., TNF-α-inhibitors), immunosuppressive drugs (e.g., calcineurin-inhibitors, mTOR-inhibitors, monoclonal antibodies), chemotherapy within the preceding four weeks, presence of neutropenia (neutrophils <0.5 G/L), primary immunodeficiency, asplenia, solid organ, or hematological stem cell transplantation or liver cirrhosis [
      • Osthoff M.
      • et al.
      Low-Dose Acetylsalicylic Acid Treatment and Impact on Short-Term Mortality in Staphylococcus aureus Bloodstream Infection: A Propensity Score-Matched Cohort Study.
      ]. Severity of illness was assessed using the Pitt score [
      • Korvick J.A.
      • et al.
      Prospective observational study of Klebsiella bacteremia in 230 patients: outcome for antibiotic combinations versus monotherapy.
      ]. Non-severe CAP was defined as mild or moderate CAP (i.e., admitted to a non-ICU ward) [
      • Schweitzer V.A.
      • et al.
      Narrow-spectrum antibiotics for community-acquired pneumonia in Dutch adults (CAP-PACT): a cross-sectional, stepped-wedge, cluster-randomised, non-inferiority, antimicrobial stewardship intervention trial.
      ].
      The primary endpoint was the appropriateness of BC collection at the University Hospital Basel according to the local guidelines (Version December 2019), an algorithm published by Fabre et al. [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ] and the judgement of two board-certified ID specialists. Secondary endpoints included the association of routinely collected parameters of patients (such as age, comorbidities, etc.) with the rate of inappropriately collected BCs and with the percentage of true positive BCs. The percentage of BCs with low (0–10%), intermediate (>10–50%) and high (>50%) pretest probability of bacteremia were evaluated according to the algorithm of Fabre et al. [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ] (Supplementary, Table S2). The average amount of blood collected per BC was defined as 14 mL according to an unpublished in-house analysis demonstrating a filling of approximately 7 mL per BC bottle.

      2.4 Statistics

      All continuous variables were analyzed using the Mann-Whitney-U test with median and interquartile ranges (IQR), or the Student's t-test with mean and standard deviation (SD), where appropriate. We used the chi-square test or the Fisher's exact test for comparison of categorical variables, where appropriate. For the analysis of the association of appropriateness with variables in the prevalence analysis, patients with multiple collected BCs were assigned a score of 1 (appropriate) if all BCs were collected appropriately and a score of 0 (inappropriate) if at least one BC was not collected according to the guidelines. For the analysis of variables with the growth of relevant pathogens in the patient-individual, longitudinal analysis, only index BCs were included. Multivariable regression models that included potential predictors with an univariate p-value of less than 0.1 were built to analyze associations between patient variables and true positive index BCs in the patient-individual, longitudinal analysis, and are presented as odds ratio (OR) with their 95% confidence interval (CI). The results were considered statistically significant if the two-sided p-value was less than 0.05. SPSS 25 software (SPSS; IBM; Chicago, IL) was used for all analyses.

      3. Results

      Overall, 1114 BCs collected in 344 patients were analyzed.

      3.1 Point prevalence analysis

      A total of 69 patients and 127 BCs, 92 index BCs (72.4%) and 35 FUBCs (27.6%), were included in this analysis (Table 1).
      Table 1Patient characteristics of the point prevalence and the patient-individual, longitudinal analysis.
      Point prevalence analysis
      For the point prevalence analysis, all BC collected on two days were analyzed.
      Patient-individual, longitudinal analysis
      For the patient-individual, longituindal analysis, all BC collected in 130 patients were analyzed.
      VariableNumber of patients (n = 69)Number of patients (n = 130)
      15th January 2020, n (%)41 (59.4)N/A
      1st September 2020, n (%)28 (40.6)N/A
      Sex: Male, n (%)43 (62.3)76 (58.5)
      Age in years, median (IQR)61.5 (51.3–74.5)64 (52–76)
      Length of hospital stay in days, median (IQR)12 (5.5–34)10 (4.8–22.3)
      Inpatients, n (%)61 (88.4)109 (83.8)
      Comorbidities/risk factors:
      • Charlson comorbidity index, median (IQR)
      3 (2–5)3 (1–5)
      • Immunosuppression, n (%)
      22 (31.9)29 (22.3)
      • Intravenous drug use, n (%)
      2 (2.9)7 (5.4)
      • Endovascular foreign body, n (%)
      11 (15.9)21 (16.2)
      • Prosthetic joint, n (%)
      10 (14.5)22 (16.9)
      Wards:
      • Medical departments, n (%)
      24 (34.8)40 (30.8)
      • Emergency Department, n (%)
      19 (27.5)50 (38.5)
      • Surgical departments, n (%)
      14 (20.3)40 (30.8)
      • Intensive care unit, n (%)
      12 (17.4)N/A
      Number of BCs
      One BC is defined as one pair of BC bottles consisting of one aerob and one anaerobic bottle.
      collected per patient:
      Number of BCs collected per patient on the days of the point prevalence analysis.
      Number of BCs collected per patient during the hospitalization.
      • 1 BC, n (%)
      17 (26.6)2 (1.5)
      • 2 BCs, n (%)
      48 (69.6)57 (43.8)
      • 3 BCs, n (%)
      2 (2.9)10 (7.7)
      • 4 BCs, n (%)
      2 (2.9)17 (13.1)
      • 5–9 BCs, n (%)
      0 (0)31 (23.8)
      • ≥10 BCs, n (%)
      0 (0)13 (10.0)
      IQR: interquartile range; SD: standard deviation; BC: blood culture; N/A: not applicable.
      low asterisk For the point prevalence analysis, all BC collected on two days were analyzed.
      low asterisklow asterisk For the patient-individual, longituindal analysis, all BC collected in 130 patients were analyzed.
      § One BC is defined as one pair of BC bottles consisting of one aerob and one anaerobic bottle.
      # Number of BCs collected per patient on the days of the point prevalence analysis.
      ## Number of BCs collected per patient during the hospitalization.
      Overall, 51/127 (40.2%) BC were not collected according to local recommendations [index BCs n = 33/92 (36%); FUBCs n = 18/35 (51%)]. Of all BCs, 45/127 (35.4%) and 22/127 (17.3%) were collected in the high and intermediate pretest probability group. They were collected in accordance to the local guidelines in 77.8% and 90.9%, respectively. In index BCs, the majority of inappropriately collected BCs was obtained in the low pretest probability group [n = 24/32 (75%)], especially in non-severe CAP [n = 20/32 (62.5%)]. The true positivity rate (positive BCs with growth of a relevant pathogen) was 8.7% (n = 11/127). Relevant pathogens were only detected in appropriately collected BCs [n = 11/76 (14.5%)].
      Accordance with the local guidelines was highest in the surgical departments [n = 20/24 (83.3%)] followed by the ICU, the medical departments and the ED [n = 16/27 (59.3%); n = 19/34 (55.9%); n = 20/40 (50%), respectively]. Appropriateness was associated with a higher Pitt bacteremia score (median 1 vs. 0, p = 0.014) and a higher C-reactive protein (CRP) concentration (median 90.3 vs. 22.9 mg/L, p = 0.047), and was not different before or during the SARS-CoV-2 pandemic (p = 0.85). A mean of 63.5 BCs was collected at the University Hospital Basel per day. Assuming a BC positivity rate of 10%, inappropriately collected BCs amount to 130 L of avoidably collected blood annually.

      3.2 Patient-individual, longitudinal analysis

      130 patients were included in this analysis. 76/130 (58.8%) of the patients were male and 29/130 patients (22.3%) were immunocompromised (Table 1). A total of 625 BCs, with a median of 3 (IQR 2–6) BCs per patient was collected (Fig. 1). Of note, 121/625 (19.4%) BCs were collected in only 5/130 (4%) patients, ranging from 18 to 33 BCs per patient (Table 1, Fig. 1).
      Fig 1
      Fig. 1Number of blood cultures collected per patient in the patient-individual, longitudinal analysis. BC: blood culture.
      All BCs were collected appropriately in 30/130 (23.1%) patients, whereas the opposite, i.e. only inappropriately collected BCs, was observed in 44/130 (33.8%) patients. BCs were collected according to local recommendations in 323/625 (51.7%) patients, with a higher adherence in index BCs than in FUBCs [n = 237/370 (64.1%) vs. n = 86/255 (33.7%), p<0.001]. High rates of inappropriately collected FUBCs were observed in CAP [n = 34/43 (79.1%)], in genitourinary tract infections [n = 15/17 (88.2%)] and in isolated fever [n = 4/4 (0%)]. The true positivity rate was 11.0% (n = 69/625) and varied between the departments and according to the pretest probability (Fig. 2).
      Fig 2
      Fig. 2Appropriateness and growth of significant pathogens in index BCs according to the pretest probability for bloodstream infection and department in the patient-individual, longitudinal analysis. ED: emergency department; ICU: intensive care unit.
      Skin commensals (e.g. coagulase-negative staphylococci) were most frequently identified [n = 24/98 (24.5%)], followed by S. aureus [n = 23/98 (23.5%)] and GNB [n = 19/98 (19.4%)] (Supplementary Table S3). The true positivity rate was 18.6% in appropriately and 3.0% in inappropriately collected BC, respectively (p < 0.001). The majority of BCs with low pretest probability of bacteremia was collected in the ED [n = 48/114 (42.1%)] and was collected in non-severe CAP [n = 62/114 (54.4%)]. Parameters associated with growth of a relevant pathogen in the index BCs are summarized in Table 2.
      Table 2Predictors of growth of a relevant pathogen in index BCs in the patient-individual, longitudinal analysis; univariable and multivariable analysis. OR: Odds ratio; CI: confidence interval; CRP: C-reactive protein; BSI: bloodstream infection; OR: odds ratio.
      VariableUnivariable OR (95% CI)p-valueMultivariable OR (95% CI)p-value
      Intravenous drug use6.92 (2.69–17.80)<0.0014.25 (1.23–14.70)0.022
      Shaking chills3.06 (1.37–6.83)0.006
      Prosthetic joint2.58 (1.28–5.19)0.0085.75 (2.25–14.73)<0.001
      Endovascular foreign body2.05 (1.03–4.09)0.0423.27 (1.32–8.09)0.010
      Body temperature below 36.1 °C or above 38.9 °C4.63 (2.2–9.67)<0.001
      CRP, per 10 mg/L increase1.03 (1.00–1.07)0.048
      Pretest probability of BSI<0.001<0.001
      Low/not classifiableReferenceReference
      Intermediate15.41 (3.29–72.27)7.07 (1.36–36.77)
      High42.61 (9.99–181.85)26.34 (5.91–117.37)

      3.3 Analysis of BC collection according to a defined underlying disease

      160 patients with common infections identified by the use of ICD-10 codes were included. Among these patients 145/160 (90.6%) had BCs collected, totaling 310 index BCs and 52 FUBCs. 89/145 (61.4%) patients had infections with a low pretest probability of bacteremia (uncomplicated cellulitis, n = 45 and non-severe CAP, n = 44) and 56/145 (38.6%) had infections with an intermediate pretest probability of bacteremia (pyelonephritis, n = 40 and cholangitis, n = 16). FUBCs were collected in 37/145 (25.5%) patients, most frequently in pyelonephritis [n = 16/40 (40%)] or in non-severe CAP [n = 7/44 (15.9%)]. The true positivity rate was 8.8% (n = 32/362). The most frequently identified pathogen was Escherichia coli [n = 28/43 (65.1%)] in patients with pyelonephritis (n = 23) and beta-hemolytic Streptococcus spp. (n = 4) in uncomplicated cellulitis (Supplementary Table S3). The true positivity rates were 0% in non-severe CAP, 3.7% in uncomplicated cellulitis, 11.4% in cholangitis and 20.7% in pyelonephritis. In only two FUBC (3.8%) growth of a pathogen was observed (one contaminant, one pathogen identical to the index BC).

      3.4 Questionnaire

      The survey was sent to 453 physicians, 353 working at the University Hospital Basel, 75 at the City Hospital Triemli Zurich and 25 at the Hospital Thun, respectively. The overall response rate was 38.9% (n = 176/453). Half of the respondents were female. The majority of participants were residents [n = 100/169 (59.2%)] and had a work experience of up to 5 years [n = 83/175 (47.2%)].
      The majority of responders [n = 132/176 (75.0%)] thought that too many BCs are collected in daily practice and 34/176 (19.3%) reported to be unfamiliar with the indication of BC collection. A higher likelihood of obtaining a positive BC result if fever is accompanied by shaking chills, was acknowledged by 112/176 (63.6%) providers. The internal medicine physicians in non-university hospitals [n = 27/44 (61.4%)] as well as the ED physicians [n = 6/13 (46.2%)] and surgeons [n = 7/23 (30.4%)] were less aware of this in the literature described association than the internal medicine physicians at the university hospital [n = 48/61 (78.7%)] (Table 3).
      Table 3Comparison of survey answers between internal medicine physicians working in the university hospital and in the non-university hospitals. BC: blood culture; CRP: C-reactive protein; FUBC: follow-up blood culture. *p-value<0.05 derived from the Chi-square test or the Fisher's exact test where appropriate.
      Internal medicine physiciansNon-university hospitals, n = 44University hospital, n = 61
      Questions(Strongly) agree, n (%)(Strongly) agree, n (%)p-value*
      I have the impression that too many BCs are collected from my patients.34 (77.3)50 (82)0.553
      The indication for BC collection is often unclear to me.9 (20.5)17 (27.9)0.493
      The likelihood for a positive BC is higher if fever is accompanied by shaking chills.27 (61.4)48 (78.7)0.053
      BCs should be collected in hospitalized patients with a new onset of fever (after 48 h of fever absence).28 (63.6)58 (95.1)<0.001
      BCs should be collected in hospitalized patients with new leukocytosis and an increase in CRP.3 (6.8)9 (14.8)0.352
      FUBCs should be collected in patients with Staphylococcus aureus blood stream infections.35 (79.5)56 (91.8)0.085
      FUBCs should be collected in all blood stream infections to document clearance.7 (15.9)15 (24.6)0.281
      At the emergency department, BCs should be collected in patients with uncomplicated cellulitis.5 (11.4)35 (57.4)<0.001
      At the emergency department, BCs should be collected in febrile patients with moderate, community acquired pneumonia.2 (4.5)21 (34.4)<0.001
      In a patient with Staphylococcus aureus blood stream infection and hip prosthesis who is under appropriate antibiotic treatment, fever spikes again after 5 days of fever absence. BCs should be collected.39 (88.6)55 (90.2)1.000
      A patient with suspected urosepsis is admitted to the general ward. Before initiation of antibiotic treatment 2 BCs and a urine culture were drawn. 12 h later, fever spikes again. BCs should be collected.1 (2.3)4 (6.6)0.396
      BC collection in patients with a new onset of fever was stated to be necessary by 28/44 (63.6%) of internal medicine physicians working in non-university hospitals and in 58/61 (95.1%) of internal medicine physicians working at the university hospital (p < 0.001) (Table 3), whereas the surgeons at the university hospital were more reluctant [n = 16/23 (69.6%), p = 0.015].
      An increase in CRP or in leukocytes was more often perceived as trigger for BC collection by physicians working in the ICU [n = 6/21 (28.6%)] compared to professionals working in other departments or in non-university hospitals [n = 9/61 (14.8%) in the Department of Internal Medicine (university hospital), n = 1/13 (7.7%) in the ED and n = 3/44 (6.8%) in non-university hospitals]. None of the ID specialists recommended BC collection in this setting (n = 10).
      Regarding the clinical scenarios, 161/176 (91.5%) of all respondents agreed that BCs should be collected in a patient with confirmed S. aureus BSI, hip prosthesis and a new onset of fever. Collection of FUBCs in S. aureus BSI were deemed necessary by 142/176 (80.7%) of the respondents. 44/176 (25%) responders agreed with the documentation of clearance in all BSI.
      There was a significant difference between the hospitals in managing BC collection in patients presenting to the ED with uncomplicated cellulitis or non-severe CAP, both low pretest probability scenarios for bacteremia. Whereas at the university hospital 35/61 (57.4%) and 21/61 (34.4%) of internal medicine physicians voted to collect BCs in these two scenarios, only 5/44 (11.4%) and 2/44 (4.5%) would do so in non-university hospitals, respectively (p < 0.001) (Table 3).

      4. Discussion

      The present study provides a comprehensive assessment of the appropriateness of BC collection in a Swiss university hospital by combining three different approaches: point prevalence, patient-individual longitudinal, and diseases-related analysis. To the best of our knowledge there is no study focusing on the appropriateness of BC collection according to local guidelines, not least because they are often nonexistent [
      • Vitrat-Hincky V.
      • et al.
      Appropriateness of blood culture testing parameters in routine practice. Results from a cross-sectional study.
      ,
      • Chen A.I.
      • et al.
      Blood culture utilization at an academic hospital: Addressing a gap in benchmarking.
      ]. In addition there are no international BC collection guidelines available since most guidelines provide BC collection recommendations according to established diagnoses but poorly define clinical findings that should trigger BC collection.
      We found that at least 40% of index BCs and of FUBCs were not collected according to the guidelines which is higher compared to published data (25%) [
      • Vitrat-Hincky V.
      • et al.
      Appropriateness of blood culture testing parameters in routine practice. Results from a cross-sectional study.
      ]. Direct comparison of these rates is limited however, since different definitions of “appropriateness” were used. Furthermore, we found that in one third of the patients, all BCs were inappropriately collected. Conversely, all BCs were collected appropriately only in a minority of patients (23%). In particular, the rate of unnecessary collected FUBCs was pronounced (approx. 60%). The collection of 3 BCs or more in at least 50% of the patients implies that FUBCs are drawn in the majority of patients, whereas this is indicated in only a small number of settings. Indeed, FUBCs were collected in clinical settings with a low pretest probability of BSI in 25% of the patients.
      About one third of index BCs was collected in infections with a low probability of BSI, most frequently in non-severe CAP in the ED. Growth of a relevant pathogen was not observed in any of these BCs. These results are in line with studies showing a low yield and low impact on further antibiotic management of BC collection and a number needed to test of up to >270 to obtain one relevant BC result in non-severe CAP [
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ,
      • Huber L.C.
      • Schibli A.
      • Yield Low
      High Costs: The Futility of Blood Cultures in Pneumonia.
      ,
      • Forstner C.
      • et al.
      Rate and Predictors of Bacteremia in Afebrile Community-Acquired Pneumonia.
      ,
      • Kennedy M.
      • et al.
      Do emergency department blood cultures change practice in patients with pneumonia?.
      ]. However, especially elderly patients may present with unspecific symptoms in the ED that do not immediately point to diagnosis of CAP, and hence, physicians may tend to collect BCs more liberally when the diagnosis is unclear.
      Interestingly, physicians working in non-university hospitals reported significantly less or even no BC collection in the setting of non-severe CAP compared to their colleagues at the university hospital. This might be related to strong local recommendations and teaching to avoid collection of BCs in non-severe CAP, but instead to focus on immunocompromised patients or on patients with severe CAP. As physicians at university hospitals are more frequently facing severely ill and immunocompromised patients, they may be more inclined to collect BCs in every patient with CAP without differentiating the disease severity or the comorbidities. In patients with uncomplicated cellulitis, a second disease with low pretest probability of BSI, results were similar to non-severe CAP. Likewise, international guidelines recommend BC collection only in patients with severe cellulitis or concomitant immunosuppression [
      • Stevens D.L.
      • et al.
      Practice guidelines for the diagnosis and management of skin and soft tissue infections: 2014 update by the Infectious Diseases Society of America.
      ], as the yield of BC in uncomplicated disease is low and the cost effectiveness and impact on further management of these patients limited [
      • Perl B.
      • et al.
      Cost-effectiveness of blood cultures for adult patients with cellulitis.
      ].
      In the great majority of patients included in the study, growth of relevant pathogens was mainly detected in appropriately collected BCs. Regarding FUBCs, if a pathogen other than S. aureus was identified, it was either already evident in index BCs or was a contaminant. These results are reassuring and confirm previous study results showing a high rate of negative FUBCs (84%) and reporting S. aureus as relevant pathogen most frequently identified [
      • Chen A.I.
      • et al.
      Blood culture utilization at an academic hospital: Addressing a gap in benchmarking.
      ]. Our slighthly higher rate of positive BCs (14.1%) compared to previous studies observing a positivity rate of 5–10% [
      • Bates D.W.
      • et al.
      Predicting bacteremia in hospitalized patients. A prospectively validated model.
      ,
      • Zwang O.
      • Albert R.K.
      Analysis of strategies to improve cost effectiveness of blood cultures.
      ], may be related to a greater amount of collected FUBCs in patients with S. aureus BSI, which are frequently positive and were often collected every 24 h in contrast to a 48–72 h interval as recommended in international guidelines [
      • Habib G.
      • et al.
      2015 ESC Guidelines for the management of infective endocarditis: The Task Force for the Management of Infective Endocarditis of the European Society of Cardiology (ESC). Endorsed by: European Association for Cardio-Thoracic Surgery (EACTS), the European Association of Nuclear Medicine (EANM).
      ,
      • Baddour L.M.
      • et al.
      Infective Endocarditis in Adults: Diagnosis, Antimicrobial Therapy, and Management of Complications: A Scientific Statement for Healthcare Professionals From the American Heart Association.
      ].
      The recommendations for the collection of FUBCs are limited to endovascular infections [
      • Wiggers J.B.
      • Xiong W.
      • Daneman N.
      Sending repeat cultures: is there a role in the management of bacteremic episodes? (SCRIBE study).
      ] and BSI with S. aureus [
      • Habib G.
      • et al.
      2015 ESC Guidelines for the management of infective endocarditis: The Task Force for the Management of Infective Endocarditis of the European Society of Cardiology (ESC). Endorsed by: European Association for Cardio-Thoracic Surgery (EACTS), the European Association of Nuclear Medicine (EANM).
      ,
      • Baddour L.M.
      • et al.
      Infective Endocarditis in Adults: Diagnosis, Antimicrobial Therapy, and Management of Complications: A Scientific Statement for Healthcare Professionals From the American Heart Association.
      ] or Candida spp. [
      • Cornely O.A.
      • et al.
      ESCMID* guideline for the diagnosis and management of Candida diseases 2012: non-neutropenic adult patients.
      ] and are debated in GNB [
      • Canzoneri C.N.
      • et al.
      Follow-up Blood Cultures in Gram-Negative Bacteremia: Are They Needed?.
      ,
      • Wiggers J.B.
      • Daneman N.
      The culture of follow-up blood cultures.
      ,
      • Giannella M.
      • et al.
      Follow-up blood cultures are associated with improved outcome of patients with gram-negative bloodstream infections: retrospective observational cohort study.
      ,
      • Maskarinec S.A.
      • et al.
      Positive follow-up blood cultures identify high mortality risk among patients with Gram-negative bacteraemia.
      ]. The benefit ot FUBC collection in S.aureus BSI was acknowledged by the great majority (90%) of survey respondents. However, in other BSI uncertainty was higher, as 25% of the survey respondents voted to collect FUBCs in all BSI to document clearance, despite the lack of evidence [
      • Wiggers J.B.
      • Daneman N.
      The culture of follow-up blood cultures.
      ]. The impact of routinely collected FUBCs on the further management of the patients might be very limited and on the contrary, may even cause more harm than benefit to the patient, including additional diagnostics and prolonged or unnecessary antibiotic treatments, additional costs, working time and unnecessarily collected blood [
      • Bates D.W.
      • et al.
      Predicting bacteremia in hospitalized patients. A prospectively validated model.
      ,
      • Coburn B.
      • et al.
      Does this adult patient with suspected bacteremia require blood cultures?.
      ,
      • Fabre V.
      • et al.
      Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
      ,
      • Zwang O.
      • Albert R.K.
      Analysis of strategies to improve cost effectiveness of blood cultures.
      ]. The latter may be associated with patient discomfort and anemia [
      • Henry M.L.
      • Garner W.L.
      • Fabri P.J.
      Iatrogenic anemia.
      ,
      • Salisbury A.C.
      • et al.
      Diagnostic blood loss from phlebotomy and hospital-acquired anemia during acute myocardial infarction.
      ]. Altogether, BCs are important in the detection of severe infections and shoud not be withhold if indicated, but their collection should in all cases be considered thoroughly, avoiding the collection independly of the suspected or defined disease.
      75% of the prescribers thought that too many BCs are collected in daily practice, a statement that was in accordance to the high rate of innappropriatly collected BCs as observed in the present study. Interestingly, the rate of appropriate BCs was higher in the surgical departments and demonstrated a higher positivity rate than those collected in the medical departments. As surgeons are less often present on the wards and are accustomed to post-operative fever spikes, the urge to collect BCs in every febrile patient may be less pronounced.
      Our results demonstrate that despite the implementation of guidelines at our hospital in 2016, the adherence rate was low. Our results underscore the need for expanding diagnostic stewardship programs especially focusing on unnecessary BC collection in diseases with low pretest probability and on the indication for FUBC collection. Numerous effective approaches aim to reduce unnecessary laboratory testing by targeting the behavior of prescribers: educational efforts to enhance knowledge, audit and feedback strategies, preanalytical algorithms or the implementation of computerized physician decision support systems and order entry systems which target the ordering behavior and the compliance to ordering systems’ recommendations directly [
      • Bindraban R.S.
      • et al.
      Reducing Test Utilization in Hospital Settings: A Narrative Review.
      ,
      • Delvaux N.
      • et al.
      The Effects of Computerized Clinical Decision Support Systems on Laboratory Test Ordering: A Systematic Review.
      ]. A diagnostic stewardship approach that displays guideline concordant indications during the BC ordering process, limiting the collection of BCs to certain indications or even locking the BC ordering panel during 24 to 48 h after the first BC order, may support guideline concordant BC collection and may impact on unnecessary diagnostics [
      • Ferrari R.
      • Prosser C.
      Testing Vitamin D Levels and Choosing Wisely.
      ].
      This study has several limitations including the analysis of a limited number of BCs and patients. As a single center study, its findings may not be representative for other countries, regions or hospitals. Due to its retrospective design, we relied exclusively on the documentation in the hospital electronic health record. This may influence the assessment of appropriateness of BC collection, and the reason for BC collection may not be fully assessable from clinical notes alone. Therefore, our study may have underestimated the necessity for BC collection when the clinical criteria had not been fulfilled. Furthermore, we mostly applied a reverse approach, analyzing patients in whom BCs were collected. In this regard, we were not able to report results on the cases where BC were not collected although they would have been recommended. Hence, our results are mostly relevant for future efforts to reduce the number of unnecessarily collected BCs. Last, the response rate of the survey was moderate, leading to a limited generalisability of the results.

      5. Conclusions

      The present analysis of BC collection practices in a Swiss university hospital revealed a high number of inappropriately collected BCs. This was particularly evident in diseases with a low pretest probability of BSI including non-severe CAP and in case of FUBCs. Furthermore, a substantial part of prescribers is uncertain regarding BC ordering practices despite the presence of local guidelines. The implementation of diagnostic stewardship programs, in particular the use of electronic decision support systems, may improve BC collection practices, increase adherence to local guidelines and reduce unnecessary diagnostics and treatment.

      Funding

      This work was supported by departmental funds of the Division of Internal Medicine, University Hospital Basel to MO.

      Declaration of Competing Interest

      The authors declare no conflict of interests.

      Acknowledgement

      None.

      Appendix. Supplementary materials

      References

        • Laupland K.B.
        • et al.
        Population-based assessment of intensive care unit-acquired bloodstream infections in adults: Incidence, risk factors, and associated mortality rate.
        Crit Care Med. 2002; 30: 2462-2467
        • Muder R.R.
        • et al.
        Bacteremia in a long-term-care facility: a five-year prospective study of 163 consecutive episodes.
        Clin Infect Dis. 1992; 14: 647-654
        • Weinstein M.P.
        • et al.
        The clinical significance of positive blood cultures in the 1990s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults.
        Clin Infect Dis. 1997; 24: 584-602
        • Bates D.W.
        • et al.
        Predicting bacteremia in hospitalized patients. A prospectively validated model.
        Ann Intern Med. 1990; 113: 495-500
        • Coburn B.
        • et al.
        Does this adult patient with suspected bacteremia require blood cultures?.
        JAMA. 2012; 308: 502-511
        • Hall K.K.
        • Lyman J.A.
        Updated review of blood culture contamination.
        Clin Microbiol Rev. 2006; 19: 788-802
        • Fabre V.
        • et al.
        Prescribers' knowledge, attitudes and perceptions about blood culturing practices for adult hospitalized patients: a call for action.
        Infect Control Hosp Epidemiol. 2018; 39: 1394-1396
        • Fabre V.
        • et al.
        Does This Patient Need Blood Cultures? A Scoping Review of Indications for Blood Cultures in Adult Nonneutropenic Inpatients.
        Clin Infect Dis. 2020; 71: 1339-1347
        • Torres J.
        • et al.
        Low yield of blood and wound cultures in patients with skin and soft-tissue infections.
        Am J Emerg Med. 2017; 35: 1159-1161
        • van Daalen F.V.
        • et al.
        Clinical condition and comorbidity as determinants for blood culture positivity in patients with skin and soft-tissue infections.
        Eur J Clin Microbiol Infect Dis. 2017; 36: 1853-1858
        • Etienne M.
        • et al.
        Should blood cultures be performed for patients with acute prostatitis?.
        J Clin Microbiol. 2010; 48: 1935-1938
        • Bordon J.
        • et al.
        The presence of pneumococcal bacteremia does not influence clinical outcomes in patients with community-acquired pneumonia: results from the Community-Acquired Pneumonia Organization (CAPO) International Cohort study.
        Chest. 2008; 133: 618-624
        • Zhang D.
        • Yang D.
        • Makam A.N.
        Utility of Blood Cultures in Pneumonia.
        Am J Med. 2019; 132: 1233-1238
        • Habib G.
        • et al.
        2015 ESC Guidelines for the management of infective endocarditis: The Task Force for the Management of Infective Endocarditis of the European Society of Cardiology (ESC). Endorsed by: European Association for Cardio-Thoracic Surgery (EACTS), the European Association of Nuclear Medicine (EANM).
        Eur Heart J. 2015; 36: 3075-3128
        • Cornely O.A.
        • et al.
        ESCMID* guideline for the diagnosis and management of Candida diseases 2012: non-neutropenic adult patients.
        Clin Microbiol Infect. 2012; 18: 19-37
        • Kuehl R.
        • et al.
        Defining persistent Staphylococcus aureus bacteraemia: secondary analysis of a prospective cohort study.
        Lancet Infect Dis. 2020; 20: 1409-1417
        • Canzoneri C.N.
        • et al.
        Follow-up Blood Cultures in Gram-Negative Bacteremia: Are They Needed?.
        Clin Infect Dis. 2017; 65: 1776-1779
        • Wiggers J.B.
        • Daneman N.
        The culture of follow-up blood cultures.
        Clin Microbiol Infect. 2020; 26: 811-813
        • Garcia R.A.
        • et al.
        Multidisciplinary team review of best practices for collection and handling of blood cultures to determine effective interventions for increasing the yield of true-positive bacteremias, reducing contamination, and eliminating false-positive central line-associated bloodstream infections.
        Am J Infect Control. 2015; 43: 1222-1237
        • Vitrat-Hincky V.
        • et al.
        Appropriateness of blood culture testing parameters in routine practice. Results from a cross-sectional study.
        Eur J Clin Microbiol Infect Dis. 2011; 30: 533-539
        • Harris P.A.
        • et al.
        Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support.
        J Biomed Inform. 2009; 42: 377-381
        • Harris P.A.
        • et al.
        The REDCap consortium: Building an international community of software platform partners.
        J Biomed Inform. 2019; 95103208
        • Elzi L.
        • et al.
        How to discriminate contamination from bloodstream infection due to coagulase-negative staphylococci: a prospective study with 654 patients.
        Clin Microbiol Infect. 2012; 18: E355-E361
        • Osthoff M.
        • et al.
        Low-Dose Acetylsalicylic Acid Treatment and Impact on Short-Term Mortality in Staphylococcus aureus Bloodstream Infection: A Propensity Score-Matched Cohort Study.
        Crit Care Med. 2016; 44: 773-781
        • Korvick J.A.
        • et al.
        Prospective observational study of Klebsiella bacteremia in 230 patients: outcome for antibiotic combinations versus monotherapy.
        Antimicrob Agents Chemother. 1992; 36: 2639-2644
        • Schweitzer V.A.
        • et al.
        Narrow-spectrum antibiotics for community-acquired pneumonia in Dutch adults (CAP-PACT): a cross-sectional, stepped-wedge, cluster-randomised, non-inferiority, antimicrobial stewardship intervention trial.
        Lancet Infect Dis. 2022; 22: 274-283
        • Chen A.I.
        • et al.
        Blood culture utilization at an academic hospital: Addressing a gap in benchmarking.
        Infect Control Hosp Epidemiol. 2018; 39: 1353-1359
        • Huber L.C.
        • Schibli A.
        • Yield Low
        High Costs: The Futility of Blood Cultures in Pneumonia.
        Chest. 2020; 158: 1284
        • Forstner C.
        • et al.
        Rate and Predictors of Bacteremia in Afebrile Community-Acquired Pneumonia.
        Chest. 2020; 157: 529-539
        • Kennedy M.
        • et al.
        Do emergency department blood cultures change practice in patients with pneumonia?.
        Ann Emerg Med. 2005; 46: 393-400
        • Stevens D.L.
        • et al.
        Practice guidelines for the diagnosis and management of skin and soft tissue infections: 2014 update by the Infectious Diseases Society of America.
        Clin Infect Dis. 2014; 59: e10-e52
        • Perl B.
        • et al.
        Cost-effectiveness of blood cultures for adult patients with cellulitis.
        Clin Infect Dis. 1999; 29: 1483-1488
        • Zwang O.
        • Albert R.K.
        Analysis of strategies to improve cost effectiveness of blood cultures.
        J Hosp Med. 2006; 1: 272-276
        • Baddour L.M.
        • et al.
        Infective Endocarditis in Adults: Diagnosis, Antimicrobial Therapy, and Management of Complications: A Scientific Statement for Healthcare Professionals From the American Heart Association.
        Circulation. 2015; 132: 1435-1486
        • Wiggers J.B.
        • Xiong W.
        • Daneman N.
        Sending repeat cultures: is there a role in the management of bacteremic episodes? (SCRIBE study).
        BMC Infect Dis. 2016; 16: 286
        • Giannella M.
        • et al.
        Follow-up blood cultures are associated with improved outcome of patients with gram-negative bloodstream infections: retrospective observational cohort study.
        Clin Microbiol Infect. 2020; 26: 897-903
        • Maskarinec S.A.
        • et al.
        Positive follow-up blood cultures identify high mortality risk among patients with Gram-negative bacteraemia.
        Clin Microbiol Infect. 2020; 26: 904-910
        • Henry M.L.
        • Garner W.L.
        • Fabri P.J.
        Iatrogenic anemia.
        Am J Surg. 1986; 151: 362-363
        • Salisbury A.C.
        • et al.
        Diagnostic blood loss from phlebotomy and hospital-acquired anemia during acute myocardial infarction.
        Arch Intern Med. 2011; 171: 1646-1653
        • Bindraban R.S.
        • et al.
        Reducing Test Utilization in Hospital Settings: A Narrative Review.
        Ann Lab Med. 2018; 38: 402-412
        • Delvaux N.
        • et al.
        The Effects of Computerized Clinical Decision Support Systems on Laboratory Test Ordering: A Systematic Review.
        Arch Pathol Lab Med. 2017; 141: 585-595
        • Ferrari R.
        • Prosser C.
        Testing Vitamin D Levels and Choosing Wisely.
        JAMA Intern Med. 2016; 176: 1019-1020