Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Impact of a strategy based on unique blood culture sampling on contamination rate and detection of bloodstream infections in critically ill patients

Impact of a strategy based on unique blood culture sampling on contamination rate and detection... Introduction there is limited evidence in the existing literature to sup- A bloodculture (BC) collection usually includes one aero- port the concept of intermittent bacteremia, whereas the bic bottle, and one anaerobic bottle; this pair of bottles is volume of blood collected appears to be the most impor- conventionally referred to as a BC set. About 1 to 2.4% tant parameter to identify a BSI [8–12]. Moreover, real- of these sets of BC are false positives (BC contaminants, life use of MS procedures often result in a low number (BCC)) which represent 12 to 34% of positive BC [1–3]. of BC collection. A single BC set (i.e., culture of only 2 To optimize the rate of bloodstream infections (BSI) bottles also named solitary BC) is obtained in up to 41% identification while maintaining a low rate of BCC, sev - to 94% of patients when MS procedure is used [1, 13, eral measures have been recommended. First, inappro- 14]. Another concern with the MS procedure is the rapid priate ordering of BC collection is frequent and sampling and significant loss of pathogen detection after initiation protocols based on the sole criterion of fever are associ- of antibiotic therapy [15]. For patients requiring rapid ated with a high rate of BC. These inappropriate BC sam - initiation of treatment, subsequent BC are frequently plings are thus associated with unnecessary antibiotic sampled after administration of the first antibiotic dose, prescriptions, additional workload for phlebotomists, which decreases their diagnostic yield. additional discomfort for the patient and, overall, addi- A unique BC (UBC) strategy, corresponding to a large tional costs [4, 5]. amount of blood drawn during a unique venipuncture, For decades, strategies to improve the performance may combine the benefits of an optimal blood volume of BC (detection of true bacteremia and limitation of collection with a decreased risk of BCC. In the emer- BCC) have focused on hygiene’s procedures and labora- gency department (ED), a UBC strategy has demon- tory techniques with the development of new BC systems strated a similar performance for the identification of BSI and culture media [6]. However, optimizing the way BC with a lower rate of BCC when compared to MS strategy are collected may be an inexpensive and efficient strat - [1]. However, a UBC strategy has not been evaluated in egy. Given the possibility of intermittent bacteremia, the the intensive care unit (ICU) with the specific problem standard approach is the multi-sampling (MS) strategy, of a high incidence of fever during the ICU stay. Predict- which implies obtaining BC sets from two to three sep- ing a BSI in a febrile and critically ill patient (at admission arate venipunctures within a short period [7]. However, or during ICU stay) is difficult, with only 10% of patients M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 3 of 10 sampled having a BSI [1, 2, 16]. Changing a MS proce- volume collected (up to 10  mL per bottle) and hygiene dure for a UBC strategy without focusing on the order- procedure. ing rule for BC collection would be associated with an The UBC consisted of sampling a large volume of blood important increase in the total number of collected BC (40 mL) through a unique venipuncture and equally dis- bottles. tributed into 2 aerobic bottles and 2 anaerobic bottles In this study, we hypothesized that a multi-faceted pro- (each bottle filled with 10  mL of blood). BC collection gram based on a UBC strategy in the ICU, at admission could only be performed on medical prescription. Col- and all along ICU stay, may optimize the detection of BSI lecting and/or prescribing additional BC was strongly with a reduction of BCC; we thus compared the rate of discouraged for at least 48  h. This point was empha - BSI and BCC during a 6-year period in an ICU, before sized during the training of prescribers and nurses and and after implementing such program. was considered as a major change in the way BC were collected. The educational program was repeated every 6 months Methods (shift of new residents) and administered to every new Setting and population studied paramedic or physician in the ICU (Additional file  1: Fig- The study was conducted in a 24-bed medical ICU from ure S1). January 1st, 2013 to December 31th, 2015 (MS period) and from April 1st, 2016 to December 31th, 2018 (UBC period). All critically ill adult patients with a least one BC definitions and indicators BC collection and hospitalized in the ICU during the A BC collection was defined as a collection of blood dur - study period were included. Data on patients, blood cul- ing a unique sample procedure (one venipuncture) what- tures (number of sets, number of bottles, rate of patho- ever the number of BC bottles collected. A solitary BC gens and BCC) and volume sampled were prospectively was defined as a collection of only 2 bottles during a 24-h monitored. period [1, 13, 14]. The rate and incidence of BSI and BCC for 1000 patient-days were compared between the two periods. Study design Multisampling period A set of BC (1 aerobic + 1 anaerobic bottle) was collected Definition of BC contaminants by ICU nurses in case of fever above 38.5 ℃ (101.3°F), A BCC was defined as a BC with only one bottle grow - and/or chills, and/or on medical order because of patient ing skin organisms (coagulase-negative staphylococci status worsening. Additional BC are drawn without (CoNS), viridans group streptococci, Corynebacterium medical order when fever occurs (above 38.5 ℃) with a spp., Cutibacterium spp. (ex. Propionibacterium spp.), maximum of 3 BC sets/day. This procedure is also imple - Bacillus spp. or Micrococcus spp) with no clinical suspi- mented in the emergency department and in most medi- cion of a specific portal of entry and for which no defini - cal wards. No medical order is needed to collect these BC tive antimicrobial therapy was prescribed [4, 17]. For for most patients. suspicion of central line-associated BSI including a skin organism, medical records were independently reviewed Unique blood culture period by two physicians (RM and AK). Disagreements were The implementation of the UBC strategy was a multi- resolved by a third physician reviewer. faceted educational program including a standardized training program, the UBC protocol and a nurse sur- BC device and estimation of blood volume collected vey on the knowledge on BC (Protocol Additional file  1: The volume of blood sampled during BC collection was Table  S1, Table  S2 and Table  S3). Medical staff (physi - assessed during seven pre-specified 1-month periods. cians and residents) and ICU nurses were trained dur- During the same periods, the blood volume sampled dur- ing a 1-h standardized training program provided by the ing BC collection was also evaluated in the ED, to assess same intensivist (M.R.) during the whole study. The UBC trends unrelated to the study at the hospital level. The protocol was posted in the admission area and in the ICU volume was determined for each filled bottle using the subunits with a reminder of the importance of the blood following formula (considering the density of blood was 1055 g/ml) [18]. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 4 of 10 weight of filled bottle mg − average weight of empty bottles (mg) Blood volume (ml) = blood density (mg/ml) Results BC bottles were incubated using the automated BACT/ Study population ® ® ALERT VIRTUO system (BactAlert FA/FN plus, bio- During the study period, 7,273 ICU patients were hos- Mérieux, Marcy l’Etoile, France). pitalized (Fig.  1). At least one BC was collected in 61.7% of patients (n = 4,491), corresponding to 17,466 BC col- Statistical analysis lections and 35,460 patient-days. The characteristics of All statistical analyses were performed using R Core the patients are presented in Table  1. During the UBC Team software (4.0.3). period, patients were older (63 ± 16 vs 62 ± 17 years), had For descriptive analysis, quantitative variables were a higher SOFA score (7 ± 4.3 vs 6.6 ± 4) and SAPS II score reported as mean ± standard deviation (SD) when their (48 ± 21 vs 44 ± 19), had higher rate of medical reason for distribution can be considered as Gaussian, and with admission (98% vs 94%), longer duration of mechanical median and inter quartile ranges (IQR) otherwise. Quali- ventilation (6 ± 11 vs 5 ± 10) and higher rate of central tative variables were reported using effective (n) and per - line catheter (57% vs 52%). centage (%). Fisher’s exact test were used to study the association BC collection of categorical variables and period. T test were used to A 59.6% reduction (95% CI 56.7–62.3, P < 0.001) of BC study the association of gaussian continuous variables bottles collected per week was observed (Fig.  2). The and period, otherwise when the distribution cannot be number of venipunctures for BC per 1,000 patient-days considered as Gaussian, Mann–Whitney test was per- decreased by 78% (Table 2). The details of BC results are formed. The proportion of BSI and BCC and their 95% presented in Table 2 CI were compared between the MS period and the UBC period, CI95 was also provided. All tests were two-sided. BSI and BCC A P value below 0.05 was considered statistically Overall, a BSI was highly suspected in 21.4% of patients significant. (959 episodes of positive BC for 4,491 patients with at To explain the evolution of binary variables over time least one BC). The rate of BCC and BSI over time are pre - as a function of the period, mixed effects logistic regres sented in Fig.  3. The rate of BC contaminants decreased sion models with fixed effect on period (MS vs UBC) from 11.2 to 3.8 per 100 patients (reduction of 73.4%, with a random effect on time (weeks) were performed. 95% CI 58.1–88.8, P < 0.001) between the MS and UBC To explain the evolution of counting variables over time as a function of the period, generalized Poisson regres- sion models with fixed effect on period (MS vs UBC) with a random effect on time (weeks) were performed. Loess regression is a non-parametric approach using local weighted regression to fit a smooth curve through the points of a scatterplot. Loess curves can reveal trends in the data that might be difficult to fit with a paramet - ric curve. Then, to study the trends in the incidence of bloodculture, bacteremia and contaminants for each study period were summarized using this approach. No imputation of missing data was performed. Holm procedure was used to control the Family Wise Error Rate in the context of multiple testing." Ethical considerations The study obtained approval from Angers University Hospital ethical committee (N°2016/65), and from the French Commission Nationale de l’Informatique et Fig. 1 Enrolment of patients. *A positive BC refers to a bloodstream des Libertés (N° 2018-043). The need for an individual infection or a BC contaminant informed consent was waived. M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 5 of 10 Table 1 Characteristics of the patients Characteristics Overall (n = 4491) MS period (n = 2639) UBC period P value (n = 1852) Age—yr 62 ± 17 62 ± 17 63 ± 16 0.02 Male sex, n (%) 2901 (65) 1678 (64) 1223 (66) 0.1 Simplified Acute Physiology Score II 45 ± 20 44 ± 19 48 ± 21 < 0.001 Sequential Organ Failure Assessment score 6.7 ± 4.1 6.6 ± 4 7 ± 4.3 0.001 McCabe score, n (%) < .001 1 2029 (45) 1348 (51) 681 (37) 2 1727 (38) 841 (32) 886 (48) 3 735 (16) 450 (17) 285 (15) Reason for admission, n (%) <0 .001 Medical 4276 (95) 2469 (94) 1807 (98) Scheduled surgery 25 (0.6) 20 (0.8) 5 (0.3) Emergency surgery 190 (4.2) 150 (5.7) 40 (2.2) Mechanical ventilation, n (%) 2880 (64) 1716 (65) 1164 (63) 0.14 Duration—days 6 ± 10 5 ± 10 6 ± 11 0.007 Central venous catheter, n (%) 2426 (54) 1364 (52) 1062 (57) <0 .001 Duration—days 8 ± 18 7 ± 17 9 ± 19 < 0.001 Indwelling urinary catheter, n (%) 3799 (85) 2222 (84) 1577 (85) 0.4 Outcomes 0.3 In-ICU death 1056 (24) 616 (23) 440 (24) 0.8 periods. Meanwhile, the rate of BSI remained stable Meanwhile, the rate of BCC for these patients decreased at 13.2% and 13.2% during the MS and UBC periods, from 9.4 to 4.3 per 100 patients (reduction of 54.4%, P = 0.98. P = 0.003) between the MS and UBC periods. The yearly variation of BSI and BCC rate are presented Of the deceased, an infection was the reason for ICU in Additional file 2: Figure S2. admission in 39 and 27%, P < 0.001 for the MS and UBC Among positive BC, the proportion of pathogens periods, respectively. increased from 54.1% (95% CI 50.2–58) during the MS period to 77.5% (95% CI 72.2–81.7) during the UBC Blood volume collected and correlation with BSI period (absolute difference of + 23.4% [95% CI 17.1– Four thousand bottles of BC (9.7% of the 41,251 bottles) 29.6], P < 0.001). This proportion of pathogen and con - were manually weighted. During the MS period, the mean taminant among positive BC over time are represented in volume of blood collected per bottle was 2.8 ± 1.8  mL in Additional file 2: Figure S3. the ICU, and 2.3 ± 2.7  mL in the ED, P = 0.11. After the Enterobacteriaceae, Staphylococcus aureus and Strep- implementation of the UBC strategy in the ICU, the tococcus spp. were the most frequent bacteria identi- mean volume of blood increased to 8.2 ± 3.9 mL (P < 0.01 fied with no difference between the two periods, P = 1 for comparison with MS periods), whereas it remained (Additional file  2: Table S4). No significant difference was stable in the ED (Additional file  2: Figure S4). No sig- observed for the ratio of community-acquired and noso- nificant difference in the volume of blood collected per comial infections (P = 0.3) and for the rate of BSI due to bottle was observed between the early (M + 2) and late Gram-negative (42.2% and 41.4%, for the UBC and MS (M+24) period of the UBC strategy in the ICU (P = 0.3). periods, P = 0.8) or Gram positive (48.8% and 50.3%, for A strong positive correlation was observed between the UBC and the MS periods, P = 0.7) between the two the volume of blood collected and the identification of periods. a BSI in the ICU (Additional file  2: Figure S5). For each Among the 1,056 in-ICU deaths, no difference was additional milliliter of blood collected in a bottle of BC, observed for the rate of BSI (18.4 and 14.9 per 100 the rate of positivity increased by 8% (95% CI 5–11), patients for the UBC and the MS periods, P = 0.13). P < 0.001. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 6 of 10 Fig. 2 Weekly variations of BC bottles collected. Mean number of bottles of BC collected per week was significantly lower during the unique blood culture period than during the multi-sampling period (76 (95% CI 73–79) and 196 (95% CI 186–206), P < .001) Table 2 Number, rate and incidence of bloodstream infections and contaminants for each period Characteristics Overall MS period UBC period P value ICU patients 7273 3697 3576 ICU patients with BC collection 4491 2639 1852 Cumulative ICU days 35,460 19,292 16,177 BC sets 17,466 14,735 2731 Bottles of BC 41,251 30,444 10,807 BC sets/patients 3.9 ± 7.2 5.6 ± 9 1.5 ± 1.1 <0 .001 BC sets per 1000 patient-days 411 764 169 < 0.001 Bottles of BC/patients 9 ± 15 12 ± 19 6 ± 5 < 0.001 Bottles of BC per 1000 patient-days 971 1578 668 <0 .001 Positive BC (pathogens and BCC) 959 643 316 <0 .001 BCC Number 366 295 71 – Rate by BC sets, % (95% CI) 2.1 (1.9–2.3) 2.0 (1.8–2.2) 2.6 (2.1–3.2) 0.06 Rate by patient 8.1 (7.4–9) 11.2 (10–12.5) 3.8 (3–4.8) <0 .001 Per 1000 patient-days 10.3 (9.3–11.3) 15.3 (13.6–17.1) 4.4 (3.4–5.5) < 0.001 Rate among positive BC 38.2 (35.2–41.4) 45.9 (42.1–50) 22.5 (18.1–27.6) < 0.001 Rate by bottle of BC 1.1 1.1 0.9 0.1 BSI Number 593 348 245 – Rate by BC sets, % (95% CI) 3.4 (3.1–3.7) 2.4 (2.1–2.6) 9 (7.9–10.1) <0 .001 Rate by ICU stay 13.2 (12.2–14.2) 13.2 (11.9–14.6) 13.2 (11.7–14.9) > 0.99 Per 1000 patient-days 16.7 (15.5–18) 18 (16.1–20.1) 15.1 (13.3–17.1) 0.1 Rate among positive BC 61.8 (58.7–64.9) 54.1 (50.2–58) 77.5 (72.2–81.7) <0 .001 Rate by bottle of BC 4.3 3.6 2.2 < 0.001 M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 7 of 10 Fig. 3 Rate of BCC and BSI per week between the MS and UBC periods. 3A: the rate of BCC per week decreased by 62% from (10.4% to 3.9%) between the MS and UBC periods (P < .001). 3B: the rate of BSI per week was similar between the MS and UBC periods (P = .91). MS: multi-sampling; UBC: unique blood culture Nurse survey: knowledge on blood culture The concept that most cases of bacteremia were continu - Seventy eight percent of ICU nurses (68/87) were evalu- ous rather than intermittent emerged, and the first study ated on the two baseline surveys and 79% of these (54/68) of a unique venipuncture for detecting BSI was published 6 months after the educational program. Overall, the rate in 1996, showing a good identification rate of patho - of correct answers increased from 68 to 83% between gens [21, 22]. More recently, the UBC strategy (40  mL baselines and 6-month surveys, P < 0.001 (Additional of blood equally distributed into two aerobic bottles and file 2: Figure S6). two anaerobic bottles) was compared to a MS strategy in the ED with a 97.4% and 95.5% rate of pathogen detec- Discussion tion (each patient being his own control) [1]. No studies In this 6-year before–after study in patients hospitalized on UBC have been conducted to date in intensive care in ICU, a multi-faceted program based on a unique sam- wards. In our study, the incidence of BSI was identical pling of a large volume of blood for diagnosing BSI (UBC) between the two periods at 13.2%, close to the 8.2–11% instead of a conventional multiple sampling (MS) proce- BSI rate found in the literature in the ED or for ICU dure was associated with a major decrease in the number admission [1, 23–25]. of venipunctures per patient, with a similar rate of BSI Notably, the BSI rate did not decrease despite a major identified, while the rate of contaminated blood cultures reduction of BC sets obtained. A similar result was found per patient decreased from 11.2 to 3.8%. The increase by Dargère et al. [1], showing that the UBC strategy had in the volume of blood collected and a lower number of a net clinical benefit illustrated by a higher sum of BSI venipunctures per patient might explain the performance missed and BCC with the MS strategy). of the UBC strategy. Volume Sampling strategy and pathogen identifications Most BSI in adults have a low bacterial inoculum and Seminal studies in the 1970s reported that serial blood the volume of blood collected for culture appears to be cultures taken within 24–48  h were able to confirm the most important parameter to identify bacteremia 99% of BSI [19, 20]. Li et  al. (1994) observed that, for a [21, 26]. Overall, about one half of patients with a BSI similar volume of blood collected, no advantage was have a bacterial concentration below one CFU/mL [26]. observed with any particular interval of collection [11]. For example, a bacterial density below one CFU/mL has Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 8 of 10 been described in 38% and 65% of patients with S. aureus blood cultured. In a quite logical way in this study, the and E. coli BSI [21]. The sensitivity of BC collection rate of BCC by BC set was not different between the two increases with the number of BC collected with 70–90% periods. This result probably reflects that the number of of BSI identified with the first BC set and an additional venipunctures is the parameter that predict the most the yield of 5–20% and 2.4–15.3% for the second and third contamination risk. BC set [1, 8, 10–12, 19, 27]. Because most studies did not monitor the volume of blood collected, the number of BC Eec ff t of the educational program sets may not represent adequately the volume of blood For a new procedure, a multifaced approach has been cultured. In our study, the median volume of blood col- associated with a higher rate of successful implementa- lected before implementing the UBC strategy was very tion [29]. Compared to the MS period, the rate of BSI and low, with a median volume of 2.8  mL in the ICU and BCC in the UBC period may have been impacted differ - 2.3 mL in medical wards. Despite the apparent low qual- ently by each of the measure included in this multifaced ity of these BC collections, this result is consistent with a program. The weight of each intervention has not been large study conducted in 10 U.S hospitals, with an aver- evaluated and some of them may have strongly impacted age volume of 2.3 mL per bottle and no hospital collect- the rate of BCC. However, we think that a comparison of ing an optimal volume of blood [28]. The lack of previous each measure one by one would have been out of touch educational program and the workload associated with with reality. an optimal filling of BC (estimated to 7  min per sam - pling for an adequate volume collected) probably explain Strengths and limitations of the study these results [1]. An adequate volume of 40–60  mL (i.e., The 6-year period of the study allowed studying a large two to three set of BC when bottles are correctly filled number of critically ill patients with a large number of up) has been associated with an optimal sensitivity and suspected BSI. No previous study has been conducted may be recommended [1, 9]. During the UBC period, we in the ICU to compare different sampling procedures observed an increase in the blood volume collected in the for BC collection. The prolonged period of active sur - ICU department (8.2 mL in the ICU vs 2.5 mL during the veillance and the number of episodes of suspected BSI same period in medical wards) corresponding to a three- analyzed may provide a sufficient body of evidence to fold increase in the volume of blood collected per bottle. support a UBC strategy in the ICU. This increase in the volume of blood collected per bot - The level of adherence in our study was high as shown tle might have compensated the decrease of the number by the high volume of blood sampled in each bottle (with of bottles collected. Indeed, we confirmed the increasing sustained performance at the M + 24 evaluation) and the yield of bacteremia with increasing volume sampled for low rate of solitary BC collections. The compliance of culture (Additional file 2: Figure S5). health care workers to the new procedure usually require a multifaceted intervention program with repeated per- BC contaminants formance assessments and feedback [29]. Indeed, the The BCC rate markedly decreased from 11.2% to 3.8% implementation of our procedure was time consuming, per patient during the MS and UBC periods (P < 0.001). and labor intensive. Whether this organization and feed- The consequence of an increased number of unnecessary back program can be conducted in other ICUs remains to BC in the ICU is well-illustrated by the study of Verboom be determined. et  al. who implemented a routine BC collection in all The before/after study design may be responsible patients (regardless of infection suspicion) in attempt to of some limitations in this study. Some differences in improve the rate of early detection of BSI [24]. They esti - patient’s characteristics were noted between the two mated that drawing blood from 17 patients was needed to study periods, such as for age and severity scores. How- identify one additional patient with BSI. However, a 4.2- ever, these differences were numerically small (Table  1) fold increase in BCC was observed after their protocol and tended to favor the MS group. A significant differ - implementation [24], from 2.3% of patients when BC was ence was observed for the deceased in the rate of admis- ordered only for clinically suspected infection to 9.6% sion for infection reason without a particular explanation. during the routine BC collection period. This higher rate A potential overall improvement in BC sampling proce- reflects the increased risk of contamination with each dures over time may also explain a lower rate of BCC. additional BC collection and venipuncture. Conversely, if However, we controlled for the volume of blood collected additional BC bottles are sampled during the same veni- for BC in the medical wards and noted no change during puncture (such as in the UBC procedure), the BCC rate the 3-year period of UBC in the ICU. Similarly, there was per patient decreases for a similar or higher amount of no difference in the volume of blood collected in the ICU M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 9 of 10 Competing interests between the three baseline assessments conducted before All authors declare that the research was conducted in the absence of any the implementation of the UBC strategy. commercial or financial relationships that could be construed as a potential Finally, like all study on BC, a bias in the classification conflict of interest. of the patient having BSI or not is possible, because the Author details BC is its own gold standard and some true bacteremia Department of Infectious Diseases, Angers University Hospital, 4 Rue Larrey, may have been missed in both periods. 49933 Cedex Angers, France. CRCINA, Inserm, Univ Angers, Université́ de Nantes, SFR ICAT , 49000 Angers, France. Laboratoire de Bactériologie, Centre Hospitalier Universitaire d’Angers, Angers, France. Emergenc y Depar tment, Conclusion Angers University Hospital, Univ Angers, Angers, France. UMR MitoVasc CNRS A strategy based on UBC appears to be a safe and effec - 6015 - INSERM 1083, Health Faculty, Univ Angers, FCRIN, INNOVTE, Angers, France. Department of Medical Intensive Care, University Hospital, Angers, tive strategy for BC sampling in critically ill patients. France. MINT, UMR INSERM 1066, UMR CNRS 6021, UNIV Angers, Micro Et Provided less venipunctures are performed but a large Nano Médecines Translationnelles, Angers, France. Methodology and Biosta- amount of blood is sampled for each BC set to optimize tistics Department, Delegation to Clinical Research and Innovation, Angers University Hospital, 49100 Angers, France. the yield of BC, it markedly limits the risk of BC contami- nation and associated unnecessary consequences. Received: 16 December 2022 Accepted: 5 February 2023 Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s13613- 023- 01107-y. References 1. Dargère S, Parienti J-J, Roupie E, Gancel P-E, Wiel E, Smaiti N, et al. Unique Additional file 1: Table S1. Educational program content. Table S2. blood culture for diagnosis of bloodstream infections in emergency Comparison between Multi-sampling and Unique blood culture protocols. departments: a prospective multicentre study. Clin Microbiol Infect. Table S3. Knowledge survey about blood cultures, bloodstream infections 2014;20:O920-927. and contaminants (multiple choice questions). Figure S1. Schema of our 2. Rupp ME, Cavalieri RJ, Marolf C, Lyden E. Reduction in blood culture con- multifaceted educational program for a unique blood culture strategy. tamination through use of initial specimen diversion device. Clin Infect Dis. 2017;65:201–5. Additional file 2: Figure S2. Evolution of BSI and BC contamination rate 3. Lee C-C, Lee N-Y, Chuang M-C, Chen P-L, Chang C-M, Ko W-C. The impact during the MS and UBC periods. Figure S3. Proportion of pathogen and of overcrowding on the bacterial contamination of blood cultures in the contaminant among positive BC. Table S4. Site of acquisition, source of ED. Am J Emerg Med. 2012;30:839–45. infection and pathogen identifications of BSI. Figure S4. Comparative 4. Dargère S, Cormier H, Verdon R. Contaminants in blood cultures: impor- changes in the mean volume of blood collected per bottle during the tance, implications, interpretation and prevention. Clin Microbiol Infect. MS and UBC periods in the ICU and in the emergency department (ED). 2018;24:964–9. Figure S5. Proportion of bacteremia identified according to different 5. Alahmadi YM, Aldeyab MA, McElnay JC, Scott MG, Darwish Elhajji FW, volume of blood collected. Figure S6. Mean rate of correct answer for Magee FA, et al. Clinical and economic impact of contaminated blood each question of the survey about bloodculture. cultures within the hospital setting. J Hosp Infect. 2011;77:233–6. 6. Kirn TJ, Weinstein MP. Update on blood cultures: how to obtain, process, report, and interpret. Clin Microbiol Infect. 2013;19:513–20. Acknowledgements 7. Miller JM, Binnicker MJ, Campbell S, Carroll KC, Chapin KC, Gilligan PH, Professor Christian Brun-Buisson for useful advice during the preparation of et al. A guide to utilization of the microbiology laboratory for diagnosis this manuscript. of infectious diseases: 2018 update by the infectious diseases society of America and the American society for microbiology. Clin Infect Dis. Author contributions 2018;67:e1-94. RM, AK have designed the study. All the authors (except DD, RM and PMR) 8. Cockerill FR, Wilson JW, Vetter EA, Goodman KM, Torgerson CA, Harmsen were part of the group of experts and contributed greatly to the quality of this WS, et al. Optimal testing parameters for blood cultures. Clin Infect Dis. work through their valuable comments. RM, AK drafted the manuscript, and 2004;38:1724–30. all authors contributed substantially to its revision. RM and JT performed the 9. Collazos-Blanco A, Pérez-García F, Sánchez-Carrillo C, de Egea V, Muñoz P, statistical analysis. RM and AK take responsibility for the paper as a whole. All Bouza E. Estimation of missed bloodstream infections without the third authors read and approved the final manuscript. blood culture set: a retrospective observational single-centre study. Clin Microbiol Infect. 2019;25:469–73. Funding 10. Khatib R, Simeunovic G, Sharma M, Fakih MG, Johnson LB, Briski L, et al. None. Blood culture series benefit may be limited to selected clinical condi- tions: time to reassess. Clin Microbiol Infect. 2015;21:332–6. Availability of data and materials 11. Li J, Plorde JJ, Carlson LG. Eec ff ts of volume and periodicity on blood The data sets generated during and/or analyzed during the current study are cultures. J Clin Microbiol. 1994;32:2829–31. available from the corresponding author on reasonable request. 12. Patel R, Vetter EA, Harmsen WS, Schleck CD, Fadel HJ, Cockerill FR. Optimized pathogen detection with 30- compared to 20-milliliter blood Declarations culture draws. J Clin Microbiol. 2011;49:4047–51. 13. Choi J, Ensafi S, Chartier LB, Van Praet O. A quality improvement initiative Ethics approval and consent to participate to decrease the rate of solitary blood cultures in the emergency depart- The study obtained approval from Angers University Hospital ethical commit- ment. Acad Emerg Med. 2017;24:1080–7. tee (N°2016/65), and from the French Commission Nationale de l’Informatique 14. Denny KJ, Sweeny A, Crilly J, Maloney S, Keijzers G. Is it time for a culture et des Libertés (N° 2018-043). The need for an individual informed consent change? Blood culture collection in the emergency department. Emerg was waived. Med Australas. 2018;30:575–7. 15. Scheer CS, Fuchs C, Gründling M, Vollmer M, Bast J, Bohnert JA, et al. Consent for publication Impact of antibiotic administration on blood culture positivity at the Not applicable. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 10 of 10 beginning of sepsis: a prospective clinical cohort study. Clin Microbiol Infect. 2019;25:326–31. 16. Rothe K, Spinner CD, Ott A, Querbach C, Dommasch M, Aldrich C, et al. Strategies for increasing diagnostic yield of community-onset bacterae- mia within the emergency department: a retrospective study. PLoS ONE. 2019;14: e0222545. 17. Schifman RB, Strand CL, Meier FA, Howanitz PJ. Blood culture contamina- tion: a college of American Pathologists Q-Probes study involving 640 institutions and 497134 specimens from adult patients. Arch Pathol Lab Med. 1998;122:216–21. 18. Bouza E, Sousa D, Rodríguez-Créixems M, Lechuz JG, Muñoz P. Is the volume of blood cultured still a significant factor in the diagnosis of bloodstream infections? J Clin Microbiol. 2007;45:2765–9. 19. Washington JA. Blood cultures: principles and techniques. Mayo Clin Proc. 1975;50:91–8. 20. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults. I. Laboratory and epidemiologic observations. Rev Infect Dis. 1983;5:35–53. 21. Jonsson B, Nyberg A, Henning C. Theoretical aspects of detection of bacteraemia as a function of the volume of blood cultured. APMIS. 1993;101:595–601. 22. Arendrup M, Jensen IP, Justesen T. Diagnosing bacteremia at a Danish hospital using one early large blood volume for culture. Scand J Infect Dis. 1996;28:609–14. 23. Shapiro NI, Wolfe RE, Wright SB, Moore R, Bates DW. Who needs a blood culture? A prospectively derived and validated prediction rule. J Emerg Med. 2008;35:255–64. 24. Verboom DM, van de Groep K, Boel CHE, Haas PJA, Derde LPG, Cremer OL, et al. The diagnostic yield of routine admission blood cultures in criti- cally Ill patients. Crit Care Med. 2021;49:60–9. 25. Niven DJ, Stelfox HT, Shahpori R, Laupland KB. Fever in adult ICUs: an interrupted time series analysis*. Crit Care Med. 2013;41:1863–9. 26. Arpi M, Bentzon MW, Jensen J, Frederiksen W. Importance of blood vol- ume cultured in the detection of bacteremia. Eur J Clin Microbiol Infect Dis. 1989;8:838–42. 27. Weinstein MP. Current blood culture methods and systems: clinical concepts, technology, and interpretation of results. Clin Infect Dis. 1996;23:40–6. 28. Khare R, Kothari T, Castagnaro J, Hemmings B, Tso M, Juretschko S. Active monitoring and feedback to improve blood culture fill volumes and positivity across a large integrated health system. Clin Infect Dis. 2020;70:262–8. 29. Bouadma L, Mourvillier B, Deiler V, Le Corre B, Lolom I, Régnier B, et al. A multifaceted program to prevent ventilator-associated pneumo- nia: impact on compliance with preventive measures. Crit Care Med. 2010;38:789–96. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Intensive Care Springer Journals

Impact of a strategy based on unique blood culture sampling on contamination rate and detection of bloodstream infections in critically ill patients

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/impact-of-a-strategy-based-on-unique-blood-culture-sampling-on-yrGywLnop2

References (32)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
eISSN
2110-5820
DOI
10.1186/s13613-023-01107-y
Publisher site
See Article on Publisher Site

Abstract

Introduction there is limited evidence in the existing literature to sup- A bloodculture (BC) collection usually includes one aero- port the concept of intermittent bacteremia, whereas the bic bottle, and one anaerobic bottle; this pair of bottles is volume of blood collected appears to be the most impor- conventionally referred to as a BC set. About 1 to 2.4% tant parameter to identify a BSI [8–12]. Moreover, real- of these sets of BC are false positives (BC contaminants, life use of MS procedures often result in a low number (BCC)) which represent 12 to 34% of positive BC [1–3]. of BC collection. A single BC set (i.e., culture of only 2 To optimize the rate of bloodstream infections (BSI) bottles also named solitary BC) is obtained in up to 41% identification while maintaining a low rate of BCC, sev - to 94% of patients when MS procedure is used [1, 13, eral measures have been recommended. First, inappro- 14]. Another concern with the MS procedure is the rapid priate ordering of BC collection is frequent and sampling and significant loss of pathogen detection after initiation protocols based on the sole criterion of fever are associ- of antibiotic therapy [15]. For patients requiring rapid ated with a high rate of BC. These inappropriate BC sam - initiation of treatment, subsequent BC are frequently plings are thus associated with unnecessary antibiotic sampled after administration of the first antibiotic dose, prescriptions, additional workload for phlebotomists, which decreases their diagnostic yield. additional discomfort for the patient and, overall, addi- A unique BC (UBC) strategy, corresponding to a large tional costs [4, 5]. amount of blood drawn during a unique venipuncture, For decades, strategies to improve the performance may combine the benefits of an optimal blood volume of BC (detection of true bacteremia and limitation of collection with a decreased risk of BCC. In the emer- BCC) have focused on hygiene’s procedures and labora- gency department (ED), a UBC strategy has demon- tory techniques with the development of new BC systems strated a similar performance for the identification of BSI and culture media [6]. However, optimizing the way BC with a lower rate of BCC when compared to MS strategy are collected may be an inexpensive and efficient strat - [1]. However, a UBC strategy has not been evaluated in egy. Given the possibility of intermittent bacteremia, the the intensive care unit (ICU) with the specific problem standard approach is the multi-sampling (MS) strategy, of a high incidence of fever during the ICU stay. Predict- which implies obtaining BC sets from two to three sep- ing a BSI in a febrile and critically ill patient (at admission arate venipunctures within a short period [7]. However, or during ICU stay) is difficult, with only 10% of patients M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 3 of 10 sampled having a BSI [1, 2, 16]. Changing a MS proce- volume collected (up to 10  mL per bottle) and hygiene dure for a UBC strategy without focusing on the order- procedure. ing rule for BC collection would be associated with an The UBC consisted of sampling a large volume of blood important increase in the total number of collected BC (40 mL) through a unique venipuncture and equally dis- bottles. tributed into 2 aerobic bottles and 2 anaerobic bottles In this study, we hypothesized that a multi-faceted pro- (each bottle filled with 10  mL of blood). BC collection gram based on a UBC strategy in the ICU, at admission could only be performed on medical prescription. Col- and all along ICU stay, may optimize the detection of BSI lecting and/or prescribing additional BC was strongly with a reduction of BCC; we thus compared the rate of discouraged for at least 48  h. This point was empha - BSI and BCC during a 6-year period in an ICU, before sized during the training of prescribers and nurses and and after implementing such program. was considered as a major change in the way BC were collected. The educational program was repeated every 6 months Methods (shift of new residents) and administered to every new Setting and population studied paramedic or physician in the ICU (Additional file  1: Fig- The study was conducted in a 24-bed medical ICU from ure S1). January 1st, 2013 to December 31th, 2015 (MS period) and from April 1st, 2016 to December 31th, 2018 (UBC period). All critically ill adult patients with a least one BC definitions and indicators BC collection and hospitalized in the ICU during the A BC collection was defined as a collection of blood dur - study period were included. Data on patients, blood cul- ing a unique sample procedure (one venipuncture) what- tures (number of sets, number of bottles, rate of patho- ever the number of BC bottles collected. A solitary BC gens and BCC) and volume sampled were prospectively was defined as a collection of only 2 bottles during a 24-h monitored. period [1, 13, 14]. The rate and incidence of BSI and BCC for 1000 patient-days were compared between the two periods. Study design Multisampling period A set of BC (1 aerobic + 1 anaerobic bottle) was collected Definition of BC contaminants by ICU nurses in case of fever above 38.5 ℃ (101.3°F), A BCC was defined as a BC with only one bottle grow - and/or chills, and/or on medical order because of patient ing skin organisms (coagulase-negative staphylococci status worsening. Additional BC are drawn without (CoNS), viridans group streptococci, Corynebacterium medical order when fever occurs (above 38.5 ℃) with a spp., Cutibacterium spp. (ex. Propionibacterium spp.), maximum of 3 BC sets/day. This procedure is also imple - Bacillus spp. or Micrococcus spp) with no clinical suspi- mented in the emergency department and in most medi- cion of a specific portal of entry and for which no defini - cal wards. No medical order is needed to collect these BC tive antimicrobial therapy was prescribed [4, 17]. For for most patients. suspicion of central line-associated BSI including a skin organism, medical records were independently reviewed Unique blood culture period by two physicians (RM and AK). Disagreements were The implementation of the UBC strategy was a multi- resolved by a third physician reviewer. faceted educational program including a standardized training program, the UBC protocol and a nurse sur- BC device and estimation of blood volume collected vey on the knowledge on BC (Protocol Additional file  1: The volume of blood sampled during BC collection was Table  S1, Table  S2 and Table  S3). Medical staff (physi - assessed during seven pre-specified 1-month periods. cians and residents) and ICU nurses were trained dur- During the same periods, the blood volume sampled dur- ing a 1-h standardized training program provided by the ing BC collection was also evaluated in the ED, to assess same intensivist (M.R.) during the whole study. The UBC trends unrelated to the study at the hospital level. The protocol was posted in the admission area and in the ICU volume was determined for each filled bottle using the subunits with a reminder of the importance of the blood following formula (considering the density of blood was 1055 g/ml) [18]. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 4 of 10 weight of filled bottle mg − average weight of empty bottles (mg) Blood volume (ml) = blood density (mg/ml) Results BC bottles were incubated using the automated BACT/ Study population ® ® ALERT VIRTUO system (BactAlert FA/FN plus, bio- During the study period, 7,273 ICU patients were hos- Mérieux, Marcy l’Etoile, France). pitalized (Fig.  1). At least one BC was collected in 61.7% of patients (n = 4,491), corresponding to 17,466 BC col- Statistical analysis lections and 35,460 patient-days. The characteristics of All statistical analyses were performed using R Core the patients are presented in Table  1. During the UBC Team software (4.0.3). period, patients were older (63 ± 16 vs 62 ± 17 years), had For descriptive analysis, quantitative variables were a higher SOFA score (7 ± 4.3 vs 6.6 ± 4) and SAPS II score reported as mean ± standard deviation (SD) when their (48 ± 21 vs 44 ± 19), had higher rate of medical reason for distribution can be considered as Gaussian, and with admission (98% vs 94%), longer duration of mechanical median and inter quartile ranges (IQR) otherwise. Quali- ventilation (6 ± 11 vs 5 ± 10) and higher rate of central tative variables were reported using effective (n) and per - line catheter (57% vs 52%). centage (%). Fisher’s exact test were used to study the association BC collection of categorical variables and period. T test were used to A 59.6% reduction (95% CI 56.7–62.3, P < 0.001) of BC study the association of gaussian continuous variables bottles collected per week was observed (Fig.  2). The and period, otherwise when the distribution cannot be number of venipunctures for BC per 1,000 patient-days considered as Gaussian, Mann–Whitney test was per- decreased by 78% (Table 2). The details of BC results are formed. The proportion of BSI and BCC and their 95% presented in Table 2 CI were compared between the MS period and the UBC period, CI95 was also provided. All tests were two-sided. BSI and BCC A P value below 0.05 was considered statistically Overall, a BSI was highly suspected in 21.4% of patients significant. (959 episodes of positive BC for 4,491 patients with at To explain the evolution of binary variables over time least one BC). The rate of BCC and BSI over time are pre - as a function of the period, mixed effects logistic regres sented in Fig.  3. The rate of BC contaminants decreased sion models with fixed effect on period (MS vs UBC) from 11.2 to 3.8 per 100 patients (reduction of 73.4%, with a random effect on time (weeks) were performed. 95% CI 58.1–88.8, P < 0.001) between the MS and UBC To explain the evolution of counting variables over time as a function of the period, generalized Poisson regres- sion models with fixed effect on period (MS vs UBC) with a random effect on time (weeks) were performed. Loess regression is a non-parametric approach using local weighted regression to fit a smooth curve through the points of a scatterplot. Loess curves can reveal trends in the data that might be difficult to fit with a paramet - ric curve. Then, to study the trends in the incidence of bloodculture, bacteremia and contaminants for each study period were summarized using this approach. No imputation of missing data was performed. Holm procedure was used to control the Family Wise Error Rate in the context of multiple testing." Ethical considerations The study obtained approval from Angers University Hospital ethical committee (N°2016/65), and from the French Commission Nationale de l’Informatique et Fig. 1 Enrolment of patients. *A positive BC refers to a bloodstream des Libertés (N° 2018-043). The need for an individual infection or a BC contaminant informed consent was waived. M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 5 of 10 Table 1 Characteristics of the patients Characteristics Overall (n = 4491) MS period (n = 2639) UBC period P value (n = 1852) Age—yr 62 ± 17 62 ± 17 63 ± 16 0.02 Male sex, n (%) 2901 (65) 1678 (64) 1223 (66) 0.1 Simplified Acute Physiology Score II 45 ± 20 44 ± 19 48 ± 21 < 0.001 Sequential Organ Failure Assessment score 6.7 ± 4.1 6.6 ± 4 7 ± 4.3 0.001 McCabe score, n (%) < .001 1 2029 (45) 1348 (51) 681 (37) 2 1727 (38) 841 (32) 886 (48) 3 735 (16) 450 (17) 285 (15) Reason for admission, n (%) <0 .001 Medical 4276 (95) 2469 (94) 1807 (98) Scheduled surgery 25 (0.6) 20 (0.8) 5 (0.3) Emergency surgery 190 (4.2) 150 (5.7) 40 (2.2) Mechanical ventilation, n (%) 2880 (64) 1716 (65) 1164 (63) 0.14 Duration—days 6 ± 10 5 ± 10 6 ± 11 0.007 Central venous catheter, n (%) 2426 (54) 1364 (52) 1062 (57) <0 .001 Duration—days 8 ± 18 7 ± 17 9 ± 19 < 0.001 Indwelling urinary catheter, n (%) 3799 (85) 2222 (84) 1577 (85) 0.4 Outcomes 0.3 In-ICU death 1056 (24) 616 (23) 440 (24) 0.8 periods. Meanwhile, the rate of BSI remained stable Meanwhile, the rate of BCC for these patients decreased at 13.2% and 13.2% during the MS and UBC periods, from 9.4 to 4.3 per 100 patients (reduction of 54.4%, P = 0.98. P = 0.003) between the MS and UBC periods. The yearly variation of BSI and BCC rate are presented Of the deceased, an infection was the reason for ICU in Additional file 2: Figure S2. admission in 39 and 27%, P < 0.001 for the MS and UBC Among positive BC, the proportion of pathogens periods, respectively. increased from 54.1% (95% CI 50.2–58) during the MS period to 77.5% (95% CI 72.2–81.7) during the UBC Blood volume collected and correlation with BSI period (absolute difference of + 23.4% [95% CI 17.1– Four thousand bottles of BC (9.7% of the 41,251 bottles) 29.6], P < 0.001). This proportion of pathogen and con - were manually weighted. During the MS period, the mean taminant among positive BC over time are represented in volume of blood collected per bottle was 2.8 ± 1.8  mL in Additional file 2: Figure S3. the ICU, and 2.3 ± 2.7  mL in the ED, P = 0.11. After the Enterobacteriaceae, Staphylococcus aureus and Strep- implementation of the UBC strategy in the ICU, the tococcus spp. were the most frequent bacteria identi- mean volume of blood increased to 8.2 ± 3.9 mL (P < 0.01 fied with no difference between the two periods, P = 1 for comparison with MS periods), whereas it remained (Additional file  2: Table S4). No significant difference was stable in the ED (Additional file  2: Figure S4). No sig- observed for the ratio of community-acquired and noso- nificant difference in the volume of blood collected per comial infections (P = 0.3) and for the rate of BSI due to bottle was observed between the early (M + 2) and late Gram-negative (42.2% and 41.4%, for the UBC and MS (M+24) period of the UBC strategy in the ICU (P = 0.3). periods, P = 0.8) or Gram positive (48.8% and 50.3%, for A strong positive correlation was observed between the UBC and the MS periods, P = 0.7) between the two the volume of blood collected and the identification of periods. a BSI in the ICU (Additional file  2: Figure S5). For each Among the 1,056 in-ICU deaths, no difference was additional milliliter of blood collected in a bottle of BC, observed for the rate of BSI (18.4 and 14.9 per 100 the rate of positivity increased by 8% (95% CI 5–11), patients for the UBC and the MS periods, P = 0.13). P < 0.001. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 6 of 10 Fig. 2 Weekly variations of BC bottles collected. Mean number of bottles of BC collected per week was significantly lower during the unique blood culture period than during the multi-sampling period (76 (95% CI 73–79) and 196 (95% CI 186–206), P < .001) Table 2 Number, rate and incidence of bloodstream infections and contaminants for each period Characteristics Overall MS period UBC period P value ICU patients 7273 3697 3576 ICU patients with BC collection 4491 2639 1852 Cumulative ICU days 35,460 19,292 16,177 BC sets 17,466 14,735 2731 Bottles of BC 41,251 30,444 10,807 BC sets/patients 3.9 ± 7.2 5.6 ± 9 1.5 ± 1.1 <0 .001 BC sets per 1000 patient-days 411 764 169 < 0.001 Bottles of BC/patients 9 ± 15 12 ± 19 6 ± 5 < 0.001 Bottles of BC per 1000 patient-days 971 1578 668 <0 .001 Positive BC (pathogens and BCC) 959 643 316 <0 .001 BCC Number 366 295 71 – Rate by BC sets, % (95% CI) 2.1 (1.9–2.3) 2.0 (1.8–2.2) 2.6 (2.1–3.2) 0.06 Rate by patient 8.1 (7.4–9) 11.2 (10–12.5) 3.8 (3–4.8) <0 .001 Per 1000 patient-days 10.3 (9.3–11.3) 15.3 (13.6–17.1) 4.4 (3.4–5.5) < 0.001 Rate among positive BC 38.2 (35.2–41.4) 45.9 (42.1–50) 22.5 (18.1–27.6) < 0.001 Rate by bottle of BC 1.1 1.1 0.9 0.1 BSI Number 593 348 245 – Rate by BC sets, % (95% CI) 3.4 (3.1–3.7) 2.4 (2.1–2.6) 9 (7.9–10.1) <0 .001 Rate by ICU stay 13.2 (12.2–14.2) 13.2 (11.9–14.6) 13.2 (11.7–14.9) > 0.99 Per 1000 patient-days 16.7 (15.5–18) 18 (16.1–20.1) 15.1 (13.3–17.1) 0.1 Rate among positive BC 61.8 (58.7–64.9) 54.1 (50.2–58) 77.5 (72.2–81.7) <0 .001 Rate by bottle of BC 4.3 3.6 2.2 < 0.001 M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 7 of 10 Fig. 3 Rate of BCC and BSI per week between the MS and UBC periods. 3A: the rate of BCC per week decreased by 62% from (10.4% to 3.9%) between the MS and UBC periods (P < .001). 3B: the rate of BSI per week was similar between the MS and UBC periods (P = .91). MS: multi-sampling; UBC: unique blood culture Nurse survey: knowledge on blood culture The concept that most cases of bacteremia were continu - Seventy eight percent of ICU nurses (68/87) were evalu- ous rather than intermittent emerged, and the first study ated on the two baseline surveys and 79% of these (54/68) of a unique venipuncture for detecting BSI was published 6 months after the educational program. Overall, the rate in 1996, showing a good identification rate of patho - of correct answers increased from 68 to 83% between gens [21, 22]. More recently, the UBC strategy (40  mL baselines and 6-month surveys, P < 0.001 (Additional of blood equally distributed into two aerobic bottles and file 2: Figure S6). two anaerobic bottles) was compared to a MS strategy in the ED with a 97.4% and 95.5% rate of pathogen detec- Discussion tion (each patient being his own control) [1]. No studies In this 6-year before–after study in patients hospitalized on UBC have been conducted to date in intensive care in ICU, a multi-faceted program based on a unique sam- wards. In our study, the incidence of BSI was identical pling of a large volume of blood for diagnosing BSI (UBC) between the two periods at 13.2%, close to the 8.2–11% instead of a conventional multiple sampling (MS) proce- BSI rate found in the literature in the ED or for ICU dure was associated with a major decrease in the number admission [1, 23–25]. of venipunctures per patient, with a similar rate of BSI Notably, the BSI rate did not decrease despite a major identified, while the rate of contaminated blood cultures reduction of BC sets obtained. A similar result was found per patient decreased from 11.2 to 3.8%. The increase by Dargère et al. [1], showing that the UBC strategy had in the volume of blood collected and a lower number of a net clinical benefit illustrated by a higher sum of BSI venipunctures per patient might explain the performance missed and BCC with the MS strategy). of the UBC strategy. Volume Sampling strategy and pathogen identifications Most BSI in adults have a low bacterial inoculum and Seminal studies in the 1970s reported that serial blood the volume of blood collected for culture appears to be cultures taken within 24–48  h were able to confirm the most important parameter to identify bacteremia 99% of BSI [19, 20]. Li et  al. (1994) observed that, for a [21, 26]. Overall, about one half of patients with a BSI similar volume of blood collected, no advantage was have a bacterial concentration below one CFU/mL [26]. observed with any particular interval of collection [11]. For example, a bacterial density below one CFU/mL has Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 8 of 10 been described in 38% and 65% of patients with S. aureus blood cultured. In a quite logical way in this study, the and E. coli BSI [21]. The sensitivity of BC collection rate of BCC by BC set was not different between the two increases with the number of BC collected with 70–90% periods. This result probably reflects that the number of of BSI identified with the first BC set and an additional venipunctures is the parameter that predict the most the yield of 5–20% and 2.4–15.3% for the second and third contamination risk. BC set [1, 8, 10–12, 19, 27]. Because most studies did not monitor the volume of blood collected, the number of BC Eec ff t of the educational program sets may not represent adequately the volume of blood For a new procedure, a multifaced approach has been cultured. In our study, the median volume of blood col- associated with a higher rate of successful implementa- lected before implementing the UBC strategy was very tion [29]. Compared to the MS period, the rate of BSI and low, with a median volume of 2.8  mL in the ICU and BCC in the UBC period may have been impacted differ - 2.3 mL in medical wards. Despite the apparent low qual- ently by each of the measure included in this multifaced ity of these BC collections, this result is consistent with a program. The weight of each intervention has not been large study conducted in 10 U.S hospitals, with an aver- evaluated and some of them may have strongly impacted age volume of 2.3 mL per bottle and no hospital collect- the rate of BCC. However, we think that a comparison of ing an optimal volume of blood [28]. The lack of previous each measure one by one would have been out of touch educational program and the workload associated with with reality. an optimal filling of BC (estimated to 7  min per sam - pling for an adequate volume collected) probably explain Strengths and limitations of the study these results [1]. An adequate volume of 40–60  mL (i.e., The 6-year period of the study allowed studying a large two to three set of BC when bottles are correctly filled number of critically ill patients with a large number of up) has been associated with an optimal sensitivity and suspected BSI. No previous study has been conducted may be recommended [1, 9]. During the UBC period, we in the ICU to compare different sampling procedures observed an increase in the blood volume collected in the for BC collection. The prolonged period of active sur - ICU department (8.2 mL in the ICU vs 2.5 mL during the veillance and the number of episodes of suspected BSI same period in medical wards) corresponding to a three- analyzed may provide a sufficient body of evidence to fold increase in the volume of blood collected per bottle. support a UBC strategy in the ICU. This increase in the volume of blood collected per bot - The level of adherence in our study was high as shown tle might have compensated the decrease of the number by the high volume of blood sampled in each bottle (with of bottles collected. Indeed, we confirmed the increasing sustained performance at the M + 24 evaluation) and the yield of bacteremia with increasing volume sampled for low rate of solitary BC collections. The compliance of culture (Additional file 2: Figure S5). health care workers to the new procedure usually require a multifaceted intervention program with repeated per- BC contaminants formance assessments and feedback [29]. Indeed, the The BCC rate markedly decreased from 11.2% to 3.8% implementation of our procedure was time consuming, per patient during the MS and UBC periods (P < 0.001). and labor intensive. Whether this organization and feed- The consequence of an increased number of unnecessary back program can be conducted in other ICUs remains to BC in the ICU is well-illustrated by the study of Verboom be determined. et  al. who implemented a routine BC collection in all The before/after study design may be responsible patients (regardless of infection suspicion) in attempt to of some limitations in this study. Some differences in improve the rate of early detection of BSI [24]. They esti - patient’s characteristics were noted between the two mated that drawing blood from 17 patients was needed to study periods, such as for age and severity scores. How- identify one additional patient with BSI. However, a 4.2- ever, these differences were numerically small (Table  1) fold increase in BCC was observed after their protocol and tended to favor the MS group. A significant differ - implementation [24], from 2.3% of patients when BC was ence was observed for the deceased in the rate of admis- ordered only for clinically suspected infection to 9.6% sion for infection reason without a particular explanation. during the routine BC collection period. This higher rate A potential overall improvement in BC sampling proce- reflects the increased risk of contamination with each dures over time may also explain a lower rate of BCC. additional BC collection and venipuncture. Conversely, if However, we controlled for the volume of blood collected additional BC bottles are sampled during the same veni- for BC in the medical wards and noted no change during puncture (such as in the UBC procedure), the BCC rate the 3-year period of UBC in the ICU. Similarly, there was per patient decreases for a similar or higher amount of no difference in the volume of blood collected in the ICU M ahieu et al. Annals of Intensive Care (2023) 13:13 Page 9 of 10 Competing interests between the three baseline assessments conducted before All authors declare that the research was conducted in the absence of any the implementation of the UBC strategy. commercial or financial relationships that could be construed as a potential Finally, like all study on BC, a bias in the classification conflict of interest. of the patient having BSI or not is possible, because the Author details BC is its own gold standard and some true bacteremia Department of Infectious Diseases, Angers University Hospital, 4 Rue Larrey, may have been missed in both periods. 49933 Cedex Angers, France. CRCINA, Inserm, Univ Angers, Université́ de Nantes, SFR ICAT , 49000 Angers, France. Laboratoire de Bactériologie, Centre Hospitalier Universitaire d’Angers, Angers, France. Emergenc y Depar tment, Conclusion Angers University Hospital, Univ Angers, Angers, France. UMR MitoVasc CNRS A strategy based on UBC appears to be a safe and effec - 6015 - INSERM 1083, Health Faculty, Univ Angers, FCRIN, INNOVTE, Angers, France. Department of Medical Intensive Care, University Hospital, Angers, tive strategy for BC sampling in critically ill patients. France. MINT, UMR INSERM 1066, UMR CNRS 6021, UNIV Angers, Micro Et Provided less venipunctures are performed but a large Nano Médecines Translationnelles, Angers, France. Methodology and Biosta- amount of blood is sampled for each BC set to optimize tistics Department, Delegation to Clinical Research and Innovation, Angers University Hospital, 49100 Angers, France. the yield of BC, it markedly limits the risk of BC contami- nation and associated unnecessary consequences. Received: 16 December 2022 Accepted: 5 February 2023 Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s13613- 023- 01107-y. References 1. Dargère S, Parienti J-J, Roupie E, Gancel P-E, Wiel E, Smaiti N, et al. Unique Additional file 1: Table S1. Educational program content. Table S2. blood culture for diagnosis of bloodstream infections in emergency Comparison between Multi-sampling and Unique blood culture protocols. departments: a prospective multicentre study. Clin Microbiol Infect. Table S3. Knowledge survey about blood cultures, bloodstream infections 2014;20:O920-927. and contaminants (multiple choice questions). Figure S1. Schema of our 2. Rupp ME, Cavalieri RJ, Marolf C, Lyden E. Reduction in blood culture con- multifaceted educational program for a unique blood culture strategy. tamination through use of initial specimen diversion device. Clin Infect Dis. 2017;65:201–5. Additional file 2: Figure S2. Evolution of BSI and BC contamination rate 3. Lee C-C, Lee N-Y, Chuang M-C, Chen P-L, Chang C-M, Ko W-C. The impact during the MS and UBC periods. Figure S3. Proportion of pathogen and of overcrowding on the bacterial contamination of blood cultures in the contaminant among positive BC. Table S4. Site of acquisition, source of ED. Am J Emerg Med. 2012;30:839–45. infection and pathogen identifications of BSI. Figure S4. Comparative 4. Dargère S, Cormier H, Verdon R. Contaminants in blood cultures: impor- changes in the mean volume of blood collected per bottle during the tance, implications, interpretation and prevention. Clin Microbiol Infect. MS and UBC periods in the ICU and in the emergency department (ED). 2018;24:964–9. Figure S5. Proportion of bacteremia identified according to different 5. Alahmadi YM, Aldeyab MA, McElnay JC, Scott MG, Darwish Elhajji FW, volume of blood collected. Figure S6. Mean rate of correct answer for Magee FA, et al. Clinical and economic impact of contaminated blood each question of the survey about bloodculture. cultures within the hospital setting. J Hosp Infect. 2011;77:233–6. 6. Kirn TJ, Weinstein MP. Update on blood cultures: how to obtain, process, report, and interpret. Clin Microbiol Infect. 2013;19:513–20. Acknowledgements 7. Miller JM, Binnicker MJ, Campbell S, Carroll KC, Chapin KC, Gilligan PH, Professor Christian Brun-Buisson for useful advice during the preparation of et al. A guide to utilization of the microbiology laboratory for diagnosis this manuscript. of infectious diseases: 2018 update by the infectious diseases society of America and the American society for microbiology. Clin Infect Dis. Author contributions 2018;67:e1-94. RM, AK have designed the study. All the authors (except DD, RM and PMR) 8. Cockerill FR, Wilson JW, Vetter EA, Goodman KM, Torgerson CA, Harmsen were part of the group of experts and contributed greatly to the quality of this WS, et al. Optimal testing parameters for blood cultures. Clin Infect Dis. work through their valuable comments. RM, AK drafted the manuscript, and 2004;38:1724–30. all authors contributed substantially to its revision. RM and JT performed the 9. Collazos-Blanco A, Pérez-García F, Sánchez-Carrillo C, de Egea V, Muñoz P, statistical analysis. RM and AK take responsibility for the paper as a whole. All Bouza E. Estimation of missed bloodstream infections without the third authors read and approved the final manuscript. blood culture set: a retrospective observational single-centre study. Clin Microbiol Infect. 2019;25:469–73. Funding 10. Khatib R, Simeunovic G, Sharma M, Fakih MG, Johnson LB, Briski L, et al. None. Blood culture series benefit may be limited to selected clinical condi- tions: time to reassess. Clin Microbiol Infect. 2015;21:332–6. Availability of data and materials 11. Li J, Plorde JJ, Carlson LG. Eec ff ts of volume and periodicity on blood The data sets generated during and/or analyzed during the current study are cultures. J Clin Microbiol. 1994;32:2829–31. available from the corresponding author on reasonable request. 12. Patel R, Vetter EA, Harmsen WS, Schleck CD, Fadel HJ, Cockerill FR. Optimized pathogen detection with 30- compared to 20-milliliter blood Declarations culture draws. J Clin Microbiol. 2011;49:4047–51. 13. Choi J, Ensafi S, Chartier LB, Van Praet O. A quality improvement initiative Ethics approval and consent to participate to decrease the rate of solitary blood cultures in the emergency depart- The study obtained approval from Angers University Hospital ethical commit- ment. Acad Emerg Med. 2017;24:1080–7. tee (N°2016/65), and from the French Commission Nationale de l’Informatique 14. Denny KJ, Sweeny A, Crilly J, Maloney S, Keijzers G. Is it time for a culture et des Libertés (N° 2018-043). The need for an individual informed consent change? Blood culture collection in the emergency department. Emerg was waived. Med Australas. 2018;30:575–7. 15. Scheer CS, Fuchs C, Gründling M, Vollmer M, Bast J, Bohnert JA, et al. Consent for publication Impact of antibiotic administration on blood culture positivity at the Not applicable. Mahieu et al. Annals of Intensive Care (2023) 13:13 Page 10 of 10 beginning of sepsis: a prospective clinical cohort study. Clin Microbiol Infect. 2019;25:326–31. 16. Rothe K, Spinner CD, Ott A, Querbach C, Dommasch M, Aldrich C, et al. Strategies for increasing diagnostic yield of community-onset bacterae- mia within the emergency department: a retrospective study. PLoS ONE. 2019;14: e0222545. 17. Schifman RB, Strand CL, Meier FA, Howanitz PJ. Blood culture contamina- tion: a college of American Pathologists Q-Probes study involving 640 institutions and 497134 specimens from adult patients. Arch Pathol Lab Med. 1998;122:216–21. 18. Bouza E, Sousa D, Rodríguez-Créixems M, Lechuz JG, Muñoz P. Is the volume of blood cultured still a significant factor in the diagnosis of bloodstream infections? J Clin Microbiol. 2007;45:2765–9. 19. Washington JA. Blood cultures: principles and techniques. Mayo Clin Proc. 1975;50:91–8. 20. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults. I. Laboratory and epidemiologic observations. Rev Infect Dis. 1983;5:35–53. 21. Jonsson B, Nyberg A, Henning C. Theoretical aspects of detection of bacteraemia as a function of the volume of blood cultured. APMIS. 1993;101:595–601. 22. Arendrup M, Jensen IP, Justesen T. Diagnosing bacteremia at a Danish hospital using one early large blood volume for culture. Scand J Infect Dis. 1996;28:609–14. 23. Shapiro NI, Wolfe RE, Wright SB, Moore R, Bates DW. Who needs a blood culture? A prospectively derived and validated prediction rule. J Emerg Med. 2008;35:255–64. 24. Verboom DM, van de Groep K, Boel CHE, Haas PJA, Derde LPG, Cremer OL, et al. The diagnostic yield of routine admission blood cultures in criti- cally Ill patients. Crit Care Med. 2021;49:60–9. 25. Niven DJ, Stelfox HT, Shahpori R, Laupland KB. Fever in adult ICUs: an interrupted time series analysis*. Crit Care Med. 2013;41:1863–9. 26. Arpi M, Bentzon MW, Jensen J, Frederiksen W. Importance of blood vol- ume cultured in the detection of bacteremia. Eur J Clin Microbiol Infect Dis. 1989;8:838–42. 27. Weinstein MP. Current blood culture methods and systems: clinical concepts, technology, and interpretation of results. Clin Infect Dis. 1996;23:40–6. 28. Khare R, Kothari T, Castagnaro J, Hemmings B, Tso M, Juretschko S. Active monitoring and feedback to improve blood culture fill volumes and positivity across a large integrated health system. Clin Infect Dis. 2020;70:262–8. 29. Bouadma L, Mourvillier B, Deiler V, Le Corre B, Lolom I, Régnier B, et al. A multifaceted program to prevent ventilator-associated pneumo- nia: impact on compliance with preventive measures. Crit Care Med. 2010;38:789–96. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations.

Journal

Annals of Intensive Care – Springer Journals

Published: Mar 3, 2023

Keywords: Unique blood culture; Blood culture; Sampling procedure; ICU; Bacteremia

There are no references for this article.