Sage Journals: Discover world-class research

Abstract

Objectives

To assess retrospectively the diagnostic value of procalcitonin (PCT) in excluding suspected bloodstream infection, establish cut-off values for PCT levels, and compare PCT with other clinical markers.

Methods

The predictive accuracy of different continuous parameters was estimated by univariate analysis of the area under the receiver operating characteristic curve. Optimized cut-off points for the parameters were selected according to the maximum Youden index values, which in turn were used to define positive and negative predictive values of different parameters in diagnosing bloodstream infection.

Results

The PCT level yielded a statistically significant area under the receiver operating characteristic curve of 0.765, with a best cut-off value of 0.80 ng/ml (83% sensitivity; 65% specificity, Youden index, J = 0.48). Positive and negative predictive values at this cut-off value were 38% and 94%, respectively. Mann–Whitney U-test revealed significantly higher values for PCT, C-reactive protein and percentage of neutrophils, but not for white blood cell count, in patients with bloodstream infection.

Conclusions

The serum PCT level can potentially be used as surrogate marker to exclude bacteraemia and to inform critical management decisions regarding antibiotic usage, in patients admitted with suspected bloodstream infection.

Keywords

Procalcitonin bloodstream infection C-reactive protein percentage of neutrophils predictive value area under the curve

Introduction

Bloodstream infection is one of the most important causes of morbidity and mortality in hospitalized patients,^1,2 although its incidence can be substantially reduced by administration of appropriate antibiotic therapy if it is detected in a timely manner.³ However, a large number of cases of suspected bloodstream infection occur in patients with systemic inflammatory response syndrome (SIRS), and biomarkers are needed to distinguish between the two conditions.³ Rapid diagnosis of bloodstream infection is usually difficult, because obtaining traditional blood cultures is a slow process. Alternative laboratory tests for establishing bloodstream infection – such as determining the erythrocyte sedimentation rate, serum C-reactive protein (CRP) level, white blood cell (WBC) count or the percentage of neutrophils (PON), and performing the polymerase chain reaction (PCR) – are slow, tedious and lack the desired sensitivity and specificity.³ In fact, except for PCR, if used alone, the other tests cannot confirm a diagnosis of bloodstream infection.^3–5

Interleukins, proadrenomedullin, proatrial natriuretic peptide, copeptin or provasopressin, interferon-γ, resistin and procalcitonin (PCT) have been studied as potential biomarkers for sepsis,^3–9 of which PCT is perhaps the most studied. Many of these studies reported that levels of serum PCT were low in healthy individuals and elevated in patients with bacteraemia.^6,8–10 Other investigators reported inconsistent and variable study findings, and questioned the diagnostic and prognostic performance of PCT in detecting bacteraemic conditions, compared with alternative parameters.^11,12 Therefore, more studies would be required for PCT to be routinely recommended for use as a diagnostic or predictive parameter of bloodstream infection.³

The aims of the present retrospective study were to evaluate the diagnostic accuracy of PCT in confirming suspected bloodstream infection, to establish cut-off values for serum PCT levels and to compare PCT with other clinical markers (CRP, PON, WBC) commonly evaluated in the laboratory.

Patients and methods

Study setting and population

This retrospective study was conducted at the General Hospital of The People’s Liberation Army, Beijing, China using data obtained from critically ill patients treated between June 1 2009 and August 1 2010. Data were obtained from participants within 2 weeks after enrolment. The Institutional Scientific and Ethics Committees of the General Hospital of The People’s Liberation Army approved the study. All patients enrolled in the study had to provide written, informed consent.

Bloodstream infection due to coagulase-negative staphylococci (CoNS) was diagnosed if at least two SIRS criteria were present and at least one of the following criteria was fulfilled: (i) at least two positive blood cultures with coagulase-negative staphylococci; (ii) time to positivity was <15 h after blood culture collection.^13,14 An infectious disease specialist (D.S.), who was not aware of the PCT values, classified episodes into either blood contamination due to CoNS or bloodstream infection due to CoNS (without other focus of infection). The infectious disease specialist was blinded to the numbers of bottles positive for blood cultures and the time to positivity. Only data from patients for whom PCT, CRP and complete blood count assessments were conducted concomitantly with blood culture were considered eligible for inclusion in the present analysis.

When two or more sets of the above parameters were reported for a patient, such as tests undertaken pre- and post-treatment, only the first set of parameters was enrolled for analysis; this was undertaken to avoid false-negative culture results (resulting from factitious interference such as antibiotic administration). A positive blood culture was used as the reference standard for the diagnosis of bloodstream infection.

Haematological disease and vaccination, or antimicrobial therapy in the days before hospital admission, were considered as exclusion criteria for potential study participants.

Data collection

A standardized retrospective case report form was used to retrieve demographic, clinical, microbiological, prior intake of antibiotics and laboratory data, including data that were used to determine symptoms as defined by the SIRS criteria.¹⁵ The number of CoNS-positive blood cultures and the total number of collected blood cultures were also recorded.

Blood culture

Blood samples (10 ml in duplicates) were routinely obtained from each patient by sterile venipuncture within 48 h of suspected bloodstream infection and were inoculated into a pair of aerobic or anaerobic blood culture bottles (bioMérieux Inc., Durham, NC, USA) containing charcoal suspension, to neutralize antibiotics. The blood culture bottles were incubated for ≤7 days in an automatic analyser (BacT/ALERT® 3D; bioMérieux SA, Marcy l’Etoile, France). In cases of positive bacterial or fungal growth; the blood culture system automatically demonstrated the time to positive blood culture. Meanwhile, Gram stains were prepared, and the organisms were subcultured for identification either manually or by the VITEK® 2 automated system (bioMérieux Inc.).

Quantification of PCT, CRP, WBC, and PON levels

Blood samples for PCT level determination were obtained as previously mentioned and immediately centrifuged at 3000 rpm for 10 min at room temperature (Bai Yang 600 A type centrifuge, Bai Yang Medical Centrifuge LLC, Beijing, China). The supernatant serum was used for measurement of the PCT level by an automated rapid quantitative immunoluminometric assay, the VIDAS® BRAHMS PCT assay (bioMérieux SA).

Whole blood samples containing 6% ethylenediaminetetra-acetic acid were simultaneously collected for CRP through venipuncture, and complete blood count analysis was done immediately using a Horiba ABX Micros CRP 200 Hematology Analyzer (HORIBA ABX SAS, Montpellier, France). The lower and higher detection limits of PCT were 0.05 ng/ml and 200 ng/ml, whereas for the CRP assay, the limits were 0.2 mg/dl and 20 mg/dl respectively. If CRP or PCT was not detectable, a value equal to the lower limit of detection for the assay was assigned.

Statistical analyses

All statistical analyses were performed using the SPSS® software package, Version 17 (SPSS Inc., Chicago, IL, USA), for Windows®. For univariate description of non-normally distributed variables, median values (50^th percentile) and interquartile ranges (25^th–75^th percentile) were used. A comparison of the plasma levels of more than two independent samples according to the blood culture diagnosis was made with the nonparametric Kruskal–Wallis one-way analysis of variance for continuous variables, while Mann–Whitney U-test was applied for comparison of two independent samples. Univariate analysis of the predictive accuracy of continuous predictors (PCT, CRP, WBC and PON) was performed using receiver operating characteristic (ROC) curves, with an area under the curve (AUC) as a measure of overall diagnostic value. An AUC of 0.5 represents a low discriminatory ability of a test and a value of 1.0 represents perfect discriminatory ability. Such analysis simultaneously provided different cut-off points, followed by different sensitivities and specificities, for different diagnostic tests. The optimized cut-off points of parameters were selected according to the maximum Youden’s index value (J = sensitivity + specificity − 1). Subsequently, the positive and negative predictive values for the PCT test, (at cut-off values of 0.5 ng/ml, 0.8 ng/ml, 1.0 ng/ml, 2 ng/ml) and CRP concentration (at cut-off values of 4.0 mg/l, 5.0 mg/l, 10.0 mg/l and 10.45 mg/l were calculated. A Youden index of 1 indicated a perfect test and a Youden index of 0 indicated a nondiscriminatory test. Differences with P-values < 0.05 (for a two-sided test) were considered statistically significant.

Results

A total of 5532 patients admitted to the hospital between June 1 2009 and August 1 2010 were clinically suspected of having a bloodstream infection, based on their symptoms and the physician’s experience; these patients underwent blood culture analyses. Of these 5532 patients, 628 had blood culture, CRP, routine blood and PCT examinations performed on the same day. Of these 628 patients (median age 54 years; range 11–92 years, 65% male), 162 had positive blood cultures; 466 cases were negative. Of the 162 patients with positive blood cultures, 57 were positive for CoNS and 105 were positive for other bacteria. Of the 57 patients with positive blood culture due to CoNS, 15 were confirmed as bloodstream infections due to CoNS and 42 were false positives for CoNS; these 42 cases were deemed to be false positives due to blood contamination, and were not enrolled in further analyses, except for measurement of contamination. The remaining 15 samples were confirmed as true-positive blood cultures due to CoNS after further evaluations, including assessment of clinical symptoms in addition to fulfilment of diagnostic criteria.^13,14 Consequently 586 cases (628 less the 42 false-negative cases) were analysed in the current study. Hence, 20.5% (120/586) of the enrolled patients had a positive blood culture report. The enrolled patient population (n = 586) consisted of cases from the general internal medicine ward (n = 200, 34%), surgical ward (n = 95, 16%), intensive care units (n = 120, 20%), respiratory intensive care unit (n = 86, 15%), and other wards (n = 85, 15%). It was beyond the remit of the present study to determine whether the patient population had community-acquired or healthcare-associated infections. Of note, there was no prior history of use of antibiotics in the period immediately preceding the study for any of the enrolled patients.

Serum levels of PCT showed a strongly positive correlation with bloodstream infections in blood culture-positive patients (Figure 1). Serum levels of PCT had a high negative predictive value (NPV) but a low positive predictive value (PPV). By comparison, the highest AUC of PCT was 0.765 (range, 0.723–0.808) with a best cut-off value of 0.80 ng/ml. Out of the 120 positive cases of blood culture, PCT was >0.80 ng/ml in 100 cases (83.3%), whereas in cases of negative blood culture PCT was >0.80 ng/ml in 162 cases (34.8%) (Table 1). Figure 1 also details the ROC curves for WBC, CRP, and PON markers assayed in the enrolled cases.

Figure 1.

Receiver operating characteristic (ROC) analysis of procalcitonin (PCT), C-reactive protein (CRP), white blood cell counts (WBC) and percentage of neutrophils (PON) for diagnosis of bloodstream infection in 120 patients with positive blood cultures. The colour version of this figure is available at: http://imr.sagepub.com.

Table 1.

Results of procalcitonin (PCT) and blood culture tests in 586 patients enrolled in a study to examine a correlation between PCT and blood culture positivity.

PCT, ng/ml	Blood culture
PCT, ng/ml	Positive	Negative	Total
Positive, ≥0.80 ng/ml	100	162	262
Negative, <0.80 ng/ml	20	304	324
Total	120	466	586

The sensitivity, specificity, positive predictive value, negative predictive value and Youden index of PCT concentration were determined in the 586 patients analysed (Table 2). At the best cut-off value, the sensitivity and specificity of PCT were 83% and 65%, respectively. However, the positive and negative predictive values at this cut-off were 38% and 94%, respectively. Of note, serum CRP levels showed a lower positive predictive value than PCT and had a similar negative predictive value to PCT.

Table 2.

Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of serum procalcitonin (PCT) and C-reactive protein (CRP) levels in predicting bloodstream infection in 586 patients with suspected bloodstream infection.

Parameter	Cut-off value	Sensitivity, %	Specificity, %	PPV, %	NPV, %	Youden index
PCT	0.5 ng/ml	86	46	29	93	0.32
PCT	0.8 ng/ml	83	65	38	94	0.48
PCT	1.0 ng/ml	76	69	39	92	0.45
PCT	2.0 ng/ml	61	75	38	88	0.36
CRP	4.0 mg/dl	87	31	24	90	0.19
CRP	5.0 mg/dl	82	35	25	88	0.17
CRP	10.0 mg/dl	55	64	28	84	0.19
CRP	10.45 mg/dl	55	67	29	85	0.22

Levels of PCT, CRP and PON were significantly higher in patients with confirmed bloodstream infections than in those without (all P < 0.002;), whereas WBC levels showed no significant difference (Table 3). Of note, the median levels of PON, WBC, and CRP in the blood culture-negative group were either higher than the normal range or at the upper limit of normal (Table 3).

Table 3.

Diagnostic performance of procalcitonin (PCT), C-reactive protein (CRP), white blood cells (WBC) and percentage of neutrophils (PON) in blood cultures (BC) of patients with suspected bloodstream infection.

Parameter*	BC-positive, n = 120	BC-negative, n = 466	P-value	AUC (95% CI)
PCT, ng/ml	3.65 (1.00–19.38)	0.41 (0.10–2.05)	0.000	0.765 (0.723, 0.808)
CRP, mg/dl	11.00 (6.04–14.6)	7.40 (3.4–12.1)	0.000	0.613 (0.559, 0.666)
WBC, × 10⁹/l	9.87 (5.9–15.72)	9.71 (6.27–14.07)	NS	0.517 (0.456, 0.577)
PON	0.860 (0.800–0.917)	0.828 (0.733–0.829)	0.002	0.591 (0.536, 0.646)

Data presented as median (range).

Reference ranges: PCT, 0 – 0.05 ng/ml; CRP, 0 – 0.8 mg/dl; WBC, 3.5 – 10 × 10⁹g/l; PON, 0.50 – 0. 70

Mann–Whitney U-test was performed for two independent samples. P-values are for AUC as a predictor of bacteraemia.

Values for parameters indicated by median (25^th–75^th) percentiles.

AUC, area under the receiver operating characteristic curve.

CI, confidence interval.

NS, not statistically significant (P ≥ 0.05).

Micro-organisms isolated from the blood culture-positive group (n = 120) were divided into four categories (Table 4). Escherichia coli (n = 13) was the most common Gram-negative isolate, while CoNS (n = 15) and Enterococcus spp. (n = 15) were the most common Gram-positive isolates. Candida spp. were responsible for all fungaemia cases (n = 19). Besides the cases described above, eight mixed-infection cases were also observed. There were no significant differences among serum levels of PCT, CRP, WBC, and PON levels (Table 4). However, among organisms responsible for ≥10 cases of bloodstream infection, relatively low PCT levels were associated with Acinetobacter baumanii, CoNS, and Candida spp., whereas E. coli, Klebsiella pneumoniae and Enterococcus were associated with considerably higher levels of PCT (Table 4).

Table 4.

Relationship between procalcitonin (PCT), C-reactive protein (CRP), white blood cells (WBC), and percentage of neutrophils (PON) levels based on pathogens isolated from 120 patients with positive blood cultures.

Pathogen	Patients N	PCT ng/ml	CRP mg/dl	WBC 10 × 10⁹	PON
Gram-negative pathogens	53	5.05 (1.49–38.59)	10.9 (5.7–14.6)	12.04 (6.42–16.75)	0.852 (0.800–0.923)
Escherichia coli	13	16.62	12.6	16.74	0.910
Acinetobacter baumanii	11	1.54	7.0	13.73	0.830
Klebsiella pneumoniae	10	12.54	8.0	7.80	0.871
Burkholderia cepacia	3	6.86	5.2	6.75	0.840
Aeromonas hydrophila	1	1.87	14.2	14.00	0.937
Serratia marcescens	2	39.23	15.9	15.96	0.952
Stenotrophomonas maltophilia	1	1.20	2.3	5.79	0.921
Pseudomonas aeruginosa	5	5.23	6.7	4.38	0.825
Salmonella spp.	1	12.54	12.8	8.01	0.66
Proteus mirabilis	2	3.00	16.7	9.56	0.858
Ralstonia pickettii	1	1.25	14.2	9.78	0.807
Brevendimonas vesicularis	1	44.19	11.0	18.63	0.901
Chryseobacterium meningosepticum	1	3.26	6.3	8.73	0.751
Brucella	1	0.16	10.0	4.28	0.350
Gram-positive pathogens	40	2.00 (0.615–11.12)	9.4 (5.50–13.58)	9.515 (5.76–12.92)	0.853 (0.80–0.875)
Coagulase-negative staphylococcus	15	1.46	11.0	7.30	0.852
Enterococcus	15	21.45	12.0	12.17	0.880
Streptococcus viridans	3	0.14	5.6	9.87	0.800
Kocuria	2	11.45	11.5	9.735	0.866
Streptococcus Dysgalactiae equlsimilis	1	0.62	4.8	4.92	0.850
Staphylococcus aureus	2	1.46	11.0	7.03	0.852
Micrococcus	1	1.27	8.6	8.72	0.917
Listeria	1	0.16	0.4	5.77	0.360
Fungi	19	2.39 (0.32–11.02)	10.7 (6.04–13.70)	8.28 (4.75–13.93)	0.86 (0.638–0.909)
Candida spp.	19	2.39	10.7	8.28	0.862
Mixed infection	8	5.54 (0.513–19.40)	13.8 (11.32–19.50)	10.66 (6.70–19.65)	0.904 (0.866–0.919)

Data presented as median.

Data indicated by median (25^th/75^th) percentiles and analysed by Kruskal–Wallis one-way analysis of variance.

Measurement of contamination

Serum levels of laboratory markers in patients with blood contamination and bloodstream infection due to CoNS were compared, to test the merit of such differential identification for the same organism. A total of 57 blood culture-positive CoNS cases were identified during the study period: 42 (74%) were classified as blood contamination and 15 (26%) as bloodstream infection (Table 5). Patients with bloodstream infection had significantly higher PCT levels (1.68 ng/ml) than those with blood contamination (0.356 ng/ml) (P < 0.01), while CRP and PON levels showed no significant difference. The proportion of positive blood culture bottles to total number of blood bottles in patients with bloodstream infection was 58% (1.9 of 3.3 bottles), compared with 27% (1.1 of 4.1 bottles) in patients with blood contamination.

Table 5.

Serum levels of procalcitonin (PCT), together with bloodstream infection and blood contamination findings, in 57 patients who were blood culture-positive for coagulase-negative staphylococci.

PCT, ng/ml	Blood culture
PCT, ng/ml	Bloodstream infection	Blood contamination	Total
Positive, ≥0.58	14	15	29
Negative, <0.58	1	27	28
Total	15	42	57

Data presented as n.

As determined by the ROC curve analysis (Figure 2), PCT was an optimal blood marker for distinguishing between bloodstream infection and blood contamination caused by CoNS, with a best cut-off of 0.58 ng/ml. This figure is considerably lower than the 0.80 ng/ml determined for distinguishing positive and negative blood cultures. Taking 0.58 ng/ml as a cut-off value, the sensitivity, specificity, PPV, and NPV were 93%, 64%, 48%, and 96%, respectively (Table 6).

Figure 2.

Receiver operating characteristic (ROC) analysis of procalcitonin (PCT), C-reactive protein (CRP) and percentage of neutrophils (PON) for distinguishing between systemic bloodstream infection due to coagulase-negative staphylococci (CoNS) and CoNS blood contamination. The colour version of this figure is available at: http://imr.sagepub.com.

Table 6.

Procalcitonin (PCT), C-reactive protein (CRP), white blood cells (WBC), and percentage of neutrophils (PON) levels in patients with bloodstream contamination and bloodstream infection due to coagulase-negative staphylococci.

Parameter	Bloodstream contamination n = 42	Bloodstream infection n = 15	P-value	AUC (95% CI)
PCT, ng/ml	0.356 (0.09–0.84)	1.68 (0.84–3.32)	<0.01	0.825 (0.719, 0.930)
CRP, mg/dl	7.45 (3.98–13.45)	11.00 (7.00–14.10)	NS	0.586 (0.413, 0.759)
Percentage of neutrophils	0.853 (0.727–0.88)	0.854 (0.80–0.87)	NS	0.521 (0.370, 0.673)
WBC × 10⁹/l*	9.37 (6.21–12.52)	10.97 (6.20–16.86)	–	–
Time to positivity of BC, h*	36 (26–60)	23 (21–33)	–	–
No. positive BC bottles/no. total*	1.1 (1–2)/4.1 (2–10)	1.9 (1–4)/3.3 (2–6)	–	–

AUC: area under the curve

CI: confidence interval

Data presented as median (range).

Data represent median values or (25^th/75^th) percentiles.

Where AUC data and P-values are missing, no statistical comparison was performed since the parameter was considered a diagnostic criterion¹³ for bloodstream infection.

Discussion

This retrospective study, performed using data from newly admitted, critically ill patients with suspected bloodstream infection, confirmed the previously described positive correlation between positive blood cultures (sepsis) and serum PCT levels (83.3%),⁵ and found that PCT had a modest positive predictive value and significantly higher negative predictive value as a marker for bloodstream infection. The study also found that CRP, PON, and WBC all had modest discriminative values as indicators of bloodstream infection, while there was a lack of any correlative/predictive value of serum PCT levels and the aetiology of bloodstream infection. Finally, PCT provided good diagnostic accuracy as a differentiator of CoNS as infectants or contaminants in blood samples.

Serum PCT has been evaluated as one of several inflammatory parameters for improving the diagnosis of sepsis. However, optimal cut-off values of PCT in the diagnosis of bloodstream infections were diverse in different studies.^13,16,17 In the present study, the PCT cut-off value (>0.8 ng/ml) for the diagnosis of such infections was different to the manufacturer’s recommended value (PCT >0.5 ng/ml), even though the Youden index for our data was on the lower side (0.48). These differences are unsurprising due to the target population: patients in our study were highly suspected to have bloodstream infections and had diverse underlying diseases such as tumours, hypertension and type II diabetes. The establishment of the cut-off value by manufacturers takes all patients into consideration, including patients without any inflammatory signs or symptoms or other underlying conditions. Such heterogeneity of the patient population also explained why median serum levels of parameters in blood culture-negative patients in our study were relatively high. Our findings demonstrated that the PCT serum level was moderately associated with bacteraemia in patients who underwent blood cultures, and its concentration was found to be the most discriminatory laboratory variable in the diagnosis of bloodstream infection, the AUC of which exceeded those of CRP, PON and WBC.

As for the sensitivity of PCT, although there was a strong correlation between positive blood culture tests and high PCT values, some positive blood cultures yielded near-to-normal PCT levels. This might be due to either a potential bacterial contamination other than CoNS or the failure of a specific tissue response (such as liver tissue, which is known to be a major site of production of PCT).¹⁸ Our observations cannot be attributed to previous administration of antibiotics, as there was no history of antibiotic administration immediately before the bloodstream infection was suspected, in the present study; previous administration of antibiotics in bacteraemic patients may be associated with low levels of PCT.¹⁹ Hence, it would be of potential interest to perform PCT analyses with information on prior history of antibiotic administration so that community-acquired bacteraemia/fungaemia could be ruled out at the onset. This is important, since PCT has a half-life of about 24 h,¹³ and PCT levels decrease in some patients who have received prior antibiotic therapy.²⁰

A meta-analysis of published data suggested that PCT is not useful for predicting bacterial sepsis due to its low sensitivity and specificity.¹⁷ Our low positive predictive value appears to corroborate this suggestion. However, high serum PCT levels in patients with negative blood cultures may also be due to prior incidence of stroke, burns, trauma, liver cancer or cardiac surgery.^21–23 Additionally, in the present study, analyses of the micro-organisms commonly associated with bloodstream infection revealed that some were associated with very high serum PCT levels, whereas others were associated with moderate to diagnostic threshold serum PCT levels. Finally, the detection of PCT as an acute phase reactant at 4 h (peaking at 6 h) after intravenous endotoxin injection²⁴ also suggests that there may be multiple mechanisms contributing to high serum PCT levels. Overall, such findings suggest that further randomized studies are required before recommending serum PCT quantitation as a tool for predicting bloodstream infection. Of note, given its high negative predictive value, serum PCT levels can still be used judiciously in ruling out bloodstream infection, and for the empiric introduction of antibiotic therapy to a patient with a putative diagnosis of bloodstream infection.

Coagulase-negative staphylococci are frequent blood culture contaminants,^25,26 hence the diagnosis of true CoNS-related bloodstream infection is a difficult clinical challenge. It is recommended that, for the diagnosis of bacteraemia, a low-virulence pathogen must be isolated in at least two blood cultures, performed on separate occasions using samples from different anatomical sites.^25,26 However, the good negative predictive value found in our study suggested that PCT is a rapid and highly discriminative means to rule out bacteraemia, especially bacteraemia due to CoNS in hospitalized patients, and can thus be used in conjunction with existing tests.

The current study has some potent limitations. Its small sample size and the heterogeneity of CoNS cases are potential biases. Moreover, the study was conducted using data from patients treated between June 1, 2009 and August 1, 2010, but is being reported now. This delay in reporting stems from our efforts to cross-verify the epidemiological and demographic data from the enrolled patients; however, the delay in reporting the results does not affect the overall findings of our work.

In conclusion, the serum PCT level can be used as a negative predictor for excluding bloodstream infection with a higher level of confidence than using this analysis as a positive predictor for including such infections. The serum PCT level – in combination with CRP, PON or WBC – does not increase the additive value of these parameters as diagnostic tools for bloodstream infection. A careful examination of patient history, comprehensive physical examination and evaluation of other laboratory parameters (including time to blood culture positivity) cannot be overemphasized in the clinical setting, unless our results can be reproduced in larger randomized clinical trials. However, measuring the serum PCT level can potentially serve as a speedy and relatively accurate surrogate marker for differentiation of bloodstream infection from blood contamination due to CoNS. Future, homogeneous, large-scale prospective studies on this subject are warranted.

Footnotes

Declaration of conflicting interest

The authors declare that there are no conflicts of interest.

Funding

This work was supported by the Beijing Medicine Research and Development Fund (No. 2009-1023).

References

Angus

Linde-Zwirble

Lidicker

. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med 2001; 29: 1303–1310.

Dombrovskiy

Martin

Sunderram

. Rapid increase in hospitalization and mortality rates for severe sepsis in the United States: a trend analysis from 1993 to 2003. Crit Care Med 2007; 35: 1244–1250.

Kibe

Adams

Barlow

. Diagnostic and prognostic biomarkers of sepsis in critical care. J Antimicrob Chemother 2011; 66(suppl 2): ii33–ii40.

Ballot

Perovic

Galpin

. Serum procalcitonin as an early marker of neonatal sepsis. S Afr Med J 2004; 94: 851–854.

Aikawa

Fujishima

Endo

. Multicenter prospective study of procalcitonin as an indicator of sepsis. J Infect Chemother 2005; 11: 152–159.

Lai

Chen

Wang

. Diagnostic value of procalcitonin for bacterial infection in elderly patients in the emergency department. J Am Geriatr Soc 2010; 58: 518–522.

Liaudat

Dayer

Praz

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

Tsangaris

Plachouras

Kavatha

. Diagnostic and prognostic value of procalcitonin among febrile critically ill patients with prolonged ICU stay. BMC Infect Dis 2009; 9: 213–213.

Schneider

Yilmaz

Kleespies

. Accuracy of procalcitonin for outcome prediction in unselected postoperative critically ill patients. Shock 2009; 31: 568–573.

10.

Assicot

Gendrel

Carsin

. High serum procalcitonin concentrations in patients with sepsis and infection. Lancet 1993; 341: 515–518.

11.

Bogar

Molnar

Kenyeres

. Sedimentation characteristics of leucocytes can predict bacteraemia in critical care patients. J Clin Pathol 2006; 59: 523–525.

12.

Jereb

Kotar

Jurca

. Usefulness of procalcitonin for diagnosis of infective endocarditis. Intern Emerg Med 2009; 4: 221–226.

13.

Jones

Fiechtl

Brown

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

14.

Schuetz

Mueller

Trampuz

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

15.

American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992; 20: 864–874.

16.

Nakamura

Wada

Ikejiri

. Efficacy of procalcitonin in the early diagnosis of bacterial infections in a critical care unit. Shock 2009; 31: 586–591.

17.

Tang

Eslick

Craig

. Accuracy of procalcitonin for sepsis diagnosis in critically ill patients: systematic review and meta-analysis. Lancet Infect Dis 2007; 7: 210–217.

18.

Nijsten

Olinga

The

. Procalcitonin behaves as a fast responding acute phase protein in vivo and in vitro. Crit Care Med 2000; 28: 458–461.

19.

Gendrel

Raymond

Coste

. Comparison of procalcitonin with C-reactive protein, interleukin 6 and interferon-alpha for differentiation of bacterial vs. viral infections. Pediatr Infect Dis J 1999; 18: 875–881.

20.

Charles

Tinel

Barbar

. Procalcitonin kinetics within the first days of sepsis: relationship with the appropriateness of antibiotic therapy and the outcome. Crit Care 2009; 13: R38–R38.

21.

Becker

Snider

Nylen

. Procalcitonin assay in systemic inflammation, infection, and sepsis: clinical utility and limitations. Crit Care Med 2008; 36: 941–952.

22.

Sponholz

Sakr

Reinhart

. Diagnostic value and prognostic implications of serum procalcitonin after cardiac surgery: a systematic review of the literature. Crit Care 2006; 10: R145–R145.

23.

Uzzan

Cohen

Nicolas

. Procalcitonin as a diagnostic test for sepsis in critically ill adults and after surgery or trauma: a systematic review and meta-analysis. Crit Care Med 2006; 34: 1996–2003.

24.

Brunkhorst

Heinz

Forycki

. Kinetics of procalcitonin in iatrogenic sepsis. Intensive Care Med 1998; 24: 888–889.

25.

Richter

Beekmann

Croco

. Minimizing the workup of blood culture contaminants: implementation and evaluation of a laboratory-based algorithm. J Clin Microbiol 2002; 40: 2437–2444.

26.

Weinstein

Reller

Murphy

. 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.

Predictive value of procalcitonin for excluding bloodstream infection: Results of a retrospective study and utility of a rapid,quantitative test for procalcitonin

Abstract

Objectives

Methods

Results

Conclusions

Keywords

Introduction

Patients and methods

Study setting and population

Data collection

Blood culture

Quantification of PCT, CRP, WBC, and PON levels

Statistical analyses

Results

Measurement of contamination

Discussion

Footnotes

Declaration of conflicting interest

Funding

References