Procalcitonin in the assessment of bacteraemia in emergency department patients: results of a large retrospective study

Abstract

Background

The aim of this study was to evaluate the use of procalcitonin in the assessment of bacteraemia in patients in the emergency department, both alone and in conjunction with existing inflammatory markers of bacterial infection.

Methods

We enrolled 3305 cases (range 20–90 years) for which we retrospectively compared procalcitonin concentration, blood culture results, body temperature, absolute neutrophil count, and C-reactive protein concentration. The positive predictive value and the negative predictive value of procalcitonin were established at different cut-off concentrations. Receiver operating characteristic curves were plotted, and the areas under the ROC curves calculated, to allow assessment of the diagnostic accuracy of (a) a combination of three existing inflammatory markers of bacterial infection (body temperature, C-reactive protein, absolute neutrophil count), and (b) this combination with procalcitonin.

Results

Positive predictive values of procalcitonin using 0.1, 1, 2, and 5 ng/mL as the cut-off values were 21.2, 32.2, 34.2, and 37.0%, respectively. Negative predictive values of procalcitonin using 0.1, 1, 2, and 5 ng/mL as the cut-off values were 95.1, 92.2, 91.1, and 89.0%, respectively. Areas under the curve of three inflammatory markers (absolute neutrophil count, C-reactive protein, and body temperature) combined was 0.879; areas under the curve of these markers combined with procalcitonin was 0.932 (p = 0.018).

Conclusions

When procalcitonin is used as a serum marker for ruling out bacteraemia, a cut-off of 0.1 ng/mL may be used. Procalcitonin improves the diagnostic accuracy of existing markers of bacteraemia.

Keywords

Procalcitonin cut-off bacteraemia emergency department

Introduction

Bacteraemia is defined as the presence of bacteria in the blood. The mortality rate associated with bacteraemia is high in patients with systemic clinical symptoms, such as fever, chills, and tremors, even when these patients are actively treated.¹ Thus, it is very important to treat these cases immediately. The diagnosis of bacteraemia should therefore be made promptly and correctly in the emergency department. A blood culture is the gold standard for diagnosing bacteraemia, but the relatively long time required to obtain the results of this test means that diagnosis and treatment are delayed.^2–4 Therefore, in emergency situations, a presumed diagnosis of bacteraemia may be made by taking a blood specimen and determining the white blood cell count (WBC, including the absolute neutrophil count (ANC) and the fraction of immature WBCs) and the concentrations of C-reactive protein (CRP) and procalcitonin (PCT).^5–7

PCT has only recently been identified as a potential diagnostic marker of bacteraemia. Several studies have analysed its performance in the emergency department, and a meta-analysis in 2007 reported that it is moderately useful in this setting.⁸ Other studies have also reported that PCT is a useful marker for ruling out bacteraemia.^9–11 The 2007 meta-analysis⁸ included 2008 cases of suspected bacteraemia. In order to confirm if PCT is a useful marker evaluating this condition, a larger sample size was used in the present study. In addition, we sought to determine if the addition of PCT could improve the diagnostic assessment of bacteraemia if used in combination with other markers of inflammation.

Materials and methods

Materials

The present study analysed 3305 blood samples from patients with suspected bacteraemia (1855 samples from male and 1450 samples from female) who presented at the emergency department in Seoul Asan Hospital from 1 January 2008 through 31 December 2010. ANC, CRP and PCT concentrations and blood culture positivity of each blood sample, and the body temperature of each patient at presentation, were determined by reviewing the medical records. The patients were between 20 and 90 years of age. Data from separate visits were included if the interval between visits was ≥1 month.

Methods

Blood culture tests were performed within 12 h of presentation at the hospital. Blood cultures that were determined to have been contaminated were excluded from the positive results. A blood culture was considered contaminated if bacterial growth was only detected in one set of several blood cultures or if skin-residing or contaminating bacteria (Bacillus spp., Coagulase negative Staphylococcus, Corynebacterium spp., Propionibacterium spp., Saccharomyces cerevisiae) grew in the cultures. When bacteria grew in the blood culture and contamination could be excluded, the patient was determined to have bacteraemia. The body temperature was taken upon presentation at the hospital for each case. The PCT and CRP concentrations, and the ANC, were measured immediately or within 12 h of presentation.

Body temperature

Body temperature was assessed orally as the patient arrived at the emergency department. Temperature was used to categorize patients as febrile (≥38℃) or not (<38℃).

PCT

PCT concentrations were measured using the VIDAS B.R.A.H.M.S PCT assay (bioMérieux, Marcy L’Etoile, France). The detection limit for PCT was 0.05 ng/mL. Based on previously published studies,^2,11–14 four PCT cut-off concentrations were used: 0.1, 1, 2, and 5 ng/mL.

CRP

CRP concentrations were measured using a Roche reagent (Roche Diagnostics, Basel, Switzerland) and the Toshiba-200FR biochemical automated analyser (Toshiba Medical Systems Co., Tokyo, Japan). When there is a concomitant bacterial infection, a CRP concentration of ≥100 mg/L is considered diagnostically meaningful.^12,14 Therefore, this was used as the cut-off value.

ANC

The ANC was measured using the Sysmex XE2100 Cell Counter (Sysmex, Kobe, Japan). Blood samples were classified into two groups based on an ANC cut-off value of 7500 cells/mm³, which is considered to be a marker of neutrophilia and is often caused by bacterial infection.¹⁵

Blood culture tests

Blood culture tests were performed using BACTEC PLUS Aerobic/F and BACTEC LYTIC Anaerobic/F Bactec Plus (BD Microbiology, Cockeysville, MD). The presence of bacteria in the cultured blood was determined using the BACTEC 9240 System (BD Microbiology, Cockeysville, MD). After four days in culture, inoculum smearing and Gram staining were performed to look for signs of microorganism growth were detected. Culture-positive bottles were immediately analysed by Gram staining and subcultured using sterile syringes. Five millilitres of Gram-positive isolates were centrifuged into bacterial pellets and bacterial identification was performed using MicroScan Walk-away (Dade Behring, Inc., West Sacramento, CA) via direct inoculation. When micro-organism growth was not detected, sets were checked with acridine orange staining.

Statistical analysis

The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of PCT in relation to the blood culture data were analysed using SPSS (version 17.0) (SPSS Inc., Chicago, IL, USA). Body temperature ≥38℃, CRP ≥100 mg/L, ANC ≥7500 cells/mm³ can be considered indications of inflammation that could be due to a bacterial infection^11,12,14; therefore, analyses of the sensitivity, specificity, PPV, and NPV were performed relative to these cut-offs. Receiver operating characteristic (ROC) curves were plotted for a combination of these three inflammatory markers, and for this combination with PCT, and diagnostic accuracy was assessed by calculating the respective areas under the ROC curves (AUCs).

Results

Of the 3305 blood samples from patients who presented at our emergency department and were suspected of having bacteraemia, 564 (17.1%) were positive according to the blood culture results. Table 1 indicates the sensitivity, specificity, PPV, and NPV values when PCT cut-off concentrations of 0.1, 1, 2, and 5 ng/mL were applied. The sensitivity and NPV were highest when PCT cut-off 0.1 ng/mL was applied, and specificity and PPV were highest when PCT cut-off was 5.0 ng/mL.

Table 1.

Sensitivity, specificity, positive predictive value, and negative predictive value of procalcitonin for predicting bacteraemia using different cut-off values.

Cut-off value (ng/mL)	Sensitivity (%)	Specificity (%)	Positive predictive value (PPV) (%)	Negative predictive value (NPV) (%)
0.1	93.2	27.7	21.2	95.1
1.0	75.6	64.5	32.2	92.2
2.0	66.1	73.2	34.2	91.1
5.0	50.6	82.2	37.0	89.0

The sensitivity, specificity PPV and NPV of different PCT cut-offs relative to body temperature, CRP and ANC are summarized in Table 2. The sensitivity of CRP ≥100 mg/L combined with PCT cut-off 0.1 ng/mL was highest, and the specificity of body temperature ≥38℃ combined with PCT cut-off 5.0 ng/mL was highest.

Table 2.

Sensitivity, specificity, and positive and negative predictive values using different procalcitonin cut-off values for predicting bacteraemia relative to the presence of other markers of inflammation.

Condition (cut-off)	0.1 ng/mL	1.0 ng/mL	2.0 ng/mL	5.0 ng/mL
Body temperature (38℃)
Sensitivity (%)	65.6	54.0	51.3	50.9
Specificity (%)	80.3	90.6	93.2	93.6
PPV (%)	25.3	35.0	39.2	39.9
NPV (%)	95.2	90.3	85.2	85.0
CRP (100 mg/L)
Sensitivity (%)	98.2	77.0	67.6	52.1
Specificity (%)	29.4	62.7	80.6	91.2
PPV (%)	23.5	31.6	35.6	39.7
NPV (%)	93.0	92.9	91.9	91.7
ANC (7500/mm³)
Sensitivity (%)	99.8	72.1	65.2	57.1
Specificity (%)	60.5	78.2	83.5	89.7
PPV (%)	24.1	32.3	34.8	41.5
NPV (%)	95.2	94.5	93.7	92.1

CRP: C-reactive protein; ANC: absolute neutrophil count.

The PPVs of our patient groups that had 0, 1, 2, or 3 of these markers were analysed at each cut-off concentration. When all three markers of inflammation were present, the PPVs were 13.5–23.9%. When none of these markers were present, the PPVs were 28.7–45.8% (Table 3). Moreover, NPVs were significantly higher when there were ≥2 markers of inflammation (Table 3).

Table 3.

Positive and negative predictive values of different procalcitonin cut-off values for predicting bacteraemia.

No. of increased inflammatory markers	0.1 ng/mL	1.0 ng/mL	2.0 ng/mL	5.0 ng/mL
0
PPV (%)	13.5	21.2	25.0	23.9
NPV (%)	96.8	95.6	95.6	94.7
Total number	83	117	78	104
1
PPV (%)	18.0	27.9	34.9	32.3
NPV (%)	95.8	95.0	94.4	93.0
Total number	161	138	143	152
2
PPV (%)	24.4	31.8	35.1	39.0
NPV (%)	95.1	91.4	90.2	88.1
Total number	137	172	234	190
3
PPV (%)	28.7	38.7	43.8	45.8
NPV (%)	85.3	86.3	83.5	79.9
Total number	183	176	120	119

ANC: absolute neutrophil count; CRP: C-reactive protein. 0 Body temperature <38℃, CRP < 100 mg/L, ANC < 7500 cells/mm³ 1 Body temperature ≥38℃, CRP < 100 mg/L, ANC < 7500 cells/mm³ Body temperature <38℃, CRP ≥ 100 mg/L, ANC < 7500 cells/mm³ Body temperature <38℃, CRP < 100 mg/L, ANC ≥ 7500 cells/mm³ 2 Body temperature ≥38℃, CRP ≥ 100 mg/L, ANC < 7500 cells/mm³ Body temperature ≥38℃, CRP < 100 mg/L, ANC ≥ 7500 cells/mm³ Body temperature <38℃, CRP ≥ 100 mg/L, ANC ≥ 7500 cells/mm³ 3 Body temperature ≥38℃, CRP ≥ 100 mg/L, ANC ≥ 7500 cells/mm³

Figure 1 shows ROC curves for the combination of three inflammatory markers and for this combination with PCT. The AUC for the combination of three inflammatory markers was 0.879. When PCT was added, the AUC was 0.932 (p = 0.018).

Figure 1.

Receiver operating characteristic (ROC) curves are shown for the combination of PCT and other increased inflammatory markers. Top (red) line indicates area under ROC curved (AUC) of PCT + 3 inflammatory markers (0.932, 95% CI: 0.907–0.966) and middle (green) line of three inflammatory markers (0.879, 95% CI: 0.835–0.924).

Discussion

This is a large-scale retrospective study of 3305 blood samples with suspected bacteraemia in our emergency department. When PCT cut-offs of 0.1–5 mg/dL were used to indicate bacteraemia, the PPVs were 21.2–37.0% and NPVs were 89.0–95.1%. Similarly, Chirouze et al. have reported that when PCT cut-off concentrations of 0.07–6.20 ng/mL were used to indicate bacteraemia, the PPVs were 14.3–31.3% and NPVs were 89.0–100%.¹⁰ In addition, Kim et al. have reported that when PCT cut-off concentrations of 0.05–5.98 ng/mL were used, the PPVs were 14.0–48.0% and NPVs were 94.0–99.0%.¹⁶ These two studies and two additional reports indicate that the PPV at which the PCT concentrations indicate bacteraemia is generally about 50%, although this will differ depending on the cut-off concentration, the patient group, and the hospital.^9,11,16,17 In a recent meta-analysis of studies on bacteraemia markers, Jones et al. reported only moderate diagnostic performance for PCT as a single test for diagnosing bacteraemia in emergency department patients, demonstrating a sensitivity of 76% and a specificity of 73% at a cut-off concentration of 0.4–0.5 ng/mL.⁸ PCT performed similarly in our present study, although the specificity of 0.5 ng/mL was lower.

Since the utility of PCT alone to predict bacteraemia was reported to be only moderate,⁸ several other studies have also examined whether it can serve as an auxiliary marker. A study on 211 patients with a body temperature of ≥38℃ who presented at the department of internal medicine of a college hospital revealed that compared with PCT alone, PCT when used together with body temperature and CRP significantly increases the sensitivity at which bacteraemia could be diagnosed, although the PPVs did not differ significantly.¹⁸ In the current study, various PCT cut-off values were used to assess its ability to predict bacteraemia on their own. The use of additional markers (body temperature, CRP, and ANC) increased the sensitivity, specificity, and PPVs compared with PCT alone; ANC ≥ 7500 cells/mm³ with PCT showed best sensitivity. Moreover, PCT improved the diagnostic discrimination of a combination of three existing markers of bacterial infection. The combination of PCT with body temperature, CRP, and ANC could differentiate between confirmed bacteraemia by blood culture and non-infectious fever with an AUC of 0.932.

Several studies have reported that PCT is a useful biochemical marker for ruling out bacteraemia.^10,11,13,19 Riedel et al. reported that bacteraemia could be ruled out using a PCT cut-off value of <0.1 ng/mL with an NPV value of 96.3%.⁹ This was also observed by Kim et al. (NPV 96.2%)¹⁶ and Albrich and Mueller (NPV = 98.2%).¹⁰ Consistent with these studies, the NPV in the present study was 95.1% at a PCT cut-off value of 0.1 ng/mL; the number of samples with PCT <0.1 ng/mL was 564 out of 3305.

In summary, when dealing with patients with suspected bacteraemia who present at the emergency department, it is important to determine promptly whether or not to administer antibiotics empirically; hence the need for clinical signs and serum markers that can permit accurate and rapid prediction of bacteraemia. The current study shows that when PCT is used as a serum marker for ruling out bacteraemia in adult patients, a cut-off value of 0.1 ng/mL can be used; if PCT is <0.1 ng/mL, then antibiotics may be withheld. When screening for bacteraemia, the addition of PCT to a combination of body temperature, CRP, and ANC improves diagnostic discrimination.

Footnotes

Acknowledgements

We would like to thank department of laboratory medicine for assistance and guidance in this research.

Declaration of conflicting interests

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Ethical approval

The ethics committee of Asan Medical Center in South Korea approved this study (REC number: 2012-0302).

Guarantor

W-KM.

Contributorship

S-YK, TDJ, WL, and SC researched literature and conceived the study. W-KM was involved in protocol development, gaining ethical approval, patient recruitment, and data analysis. S-YK wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

References

Leibovici

. Bacteremia in febrile patients. A clinical model for diagnosis. Arch Intern Med 1991; 151: 1801–1806.

Mitaka

. Clinical laboratory differentiation of infectious versus non-infectious systemic inflammatory response syndrome. Clin Chim Acta 2005; 351: 17–29.

Reimer

Wilson

Weinstein

. Update on detection of bacteremia and fungemia. Clin Microbiol Rev 1997; 10: 444–465.

Muñoz

Simarro

Rivera

. Evaluation of procalcitonin as a marker of infection in a nonselected sample of febrile hospitalized patients. Diagn Microbiol Infect Dis 2004; 49: 237–241.

Giannoudis

Hildebrand

Pape

. Inflammatory serum markers in patients with multiple trauma. Can they predict outcome? J Bone Joint Surg (Br) 2004; 86: 313–323.

Reinhart

Bayer

Brunkhorst

. Markers of endothelial damage in organ dysfunction and sepsis. Crit Care Med 2002; 30: S302–S312.

Hur

Moon

Yun

. Comparison of diagnostic utility between procalcitonin and C-reactive protein for the patients with blood culture-positive sepsis. Korean J Lab Med 2009; 29: 529–535.

Jones

Fiechtl

Brown

. Procalcitonin test in the diagnosis of bacteremia: a meta-analysis. Ann Emerg Med 2007; 50: 34–41.

Riedel

Melendez

. Procalcitonin as a marker for the detection of bacteremia and sepsis in the emergency department. Am J Clin Pathol 2011; 135: 182–189.

10.

Albrich

Mueller

. Predicting bacteremia by procalcitonin levels in patients evaluated for sepsis in the emergency department. Expert Rev Anti Infect Ther 2011; 9: 653–656.

11.

Chirouze

Schuhmacher

Rabaud

. Low serum procalcitonin level accurately predicts the absence of bacteremia in adult patients with acute fever. Clin Infect Dis 2002; 35: 156–161.

12.

Liaudat

Dayer

Praz

. Usefulness of procalcitonin serum level for the diagnosis of bacteremia. Eur J Clin Microbiol Infect Dis 2001; 20: 524–527.

13.

Baylan

Balkan

Inal

. The predictive value of serum procalcitonin levels in adult patients with active pulmonary tuberculosis. Jpn J Infect Dis 2006; 59: 164–167.

14.

Hausfater

Juillien

Madonna-Py

. Serum procalcitonin measurement as diagnostic and prognostic marker in febrile adult patients presenting to the emergency department. Crit Care 2007; 11: R60–R60.

15.

Miale

. Laboratory medicine: hematology, 6th ed. St Louis: Mosby, 1982.

16.

Kim

Lim

Kang

. Utility of procalcitonin as an early diagnostic marker of bacteremia in patients with acute fever. Yonsei Med J 2011; 52: 276–281.

17.

van Nieuwkoop

Bonten

van’t Wout

. Procalcitonin reflects bacteremia and bacterial load in urosepsis syndrome: a prospective observational study. Crit Care 2010; 14: R206–R206.

18.

de Kruif

Limper

Gerritsen

. Additional value of procalcitonin for diagnosis of infection in patients with fever at the emergency department. Crit Care Med 2010; 38: 457–463.

19.

Schuetz

Mueller

Trampuz

. Serum procalcitonin for discrimination of blood contamination from bloodstream infection due to coagulase-negative staphylococci. Infection 2007; 35: 352–355.