ORIGINAL ARTICLE

Biological markers for predicting bacterial infection,

bacteremia, and severity of infection in the emergency
department
PERE TUDELA1, CRISTINA PRAT2,4, ALICIA LACOMA2,4, JOSEP MARIA MDOL1, JOS DOMNGUEZ2,4,
MONTSE GIMNEZ2, JORDI TOR3
Unitat de Curta Estada-Urgncies, 2Servei de Microbiologia, 3Servei de Medicina Interna, Hospital Universitari
Germans Trias i Pujol, Fundaci Institut dInvestigaci en Cincies de la Salut Germans Trias i Pujol,
Universitat Autonma de Barcelona, Badalona, Barcelona, Spain. 4CIBER Enfermedades Respiratorias, Instituto
de Salud Carlos III, Barcelona, Spain.
1

CORRESPONDENCE:
Pere Tudela
Unitat de Curta Estada-Urgncies
Crta. del Canyet, s/n
08916 Badalona.
Barcelona, Spain
E-mail: ptudela.germanstrias@
gencat.net

RECEIVED:
17-10-2011

ACCEPTED:
2-12-2012

CONFLICT OF INTEREST:
None

Objectives: To analyze the utility of using clinical and laboratory variables (C-reactive
protein [CRP], procalcitonin [PCT], proadrenomedullin [proADM] and neopterin
concentrations) as predictors in cases of bacterial infection in an emergency department.
Methods: The patients were enrolled from the emergency department when blood was
extracted for culture. We recorded age; sex; Charlson index, white blood cell count,
presence of band cells, neutrophil count, microbiology findings and CRP, PCT, proADM,
and neopterin concentrations. Severity of infection was defined by a patients admission
to the intensive care unit, death, or emergency surgery. The data were analyzed by
univariate and multivariate analyses; the area under the receiver operating characteristic
curve and diagnostic yield were calculated for each variable.
Results: We included 412 patients with bacterial infection confirmed by microbiology
(28.3%), possible infection not confirmed by microbiology (39.3%), fever of unknown
origin (9.9%), and no bacterial infection (22.3%). Blood cultures were positive in 53
(12.8%) and 34 infections (8.2%) were considered severe. The independent predictors
of bacterial infection were CRP  70 mg/L, PCT  0.4 ng/mL, and presence of band
cells, although diagnostic precision was limited. The independent variables that best
predicted bacteremia were PCT > 1 ng/mL and proADM > 2 nmol/L; these variables had
negative predictive values of 94% and 93%, respectively. The variables that predicted
severity of infection were PCT > 1 ng/mL and proADM > 1.94 nmol/L, which both had
negative predictive values around 96%.
Conclusions: CRP and PCT concentrations and the presence of band cells can suggest
bacterial infection in emergency patients, although the diagnostic value of these markers
is limited. However, the diagnostic yields are high for PCT and proADM concentrations
and these measurements can be useful for ruling out bacteremia and severity of
infection. [Emergencias 2012;24:348-356]
Key words: Bacteremia. Biological markers. Emergency health services.

Introduction
The diagnosis of patients with fever or suspected acute infectious disease in emergency departments (EDs) is a complex matter yet to be resolved. On the one hand, we know that early
antibiotic treatment improves the prognosis of patients with bacterial infection (BI), but microbiological confirmation is not usually immediate and
348

there is often a need to initiate empirical antibiotic treatment while awaiting laboratory results 1,2.
On the other hand, empirical antimicrobial therapy should not be indiscriminately applied, to
avoid environmental sensitivity problems and unjustified costs. In addition, there is evidence of diagnostic difficulties in the ED, such as the unexpected detection of bacteremia in patients that
have been discharged3,4.
Emergencias 2012; 24: 348-356

BIOLOGICAL MARKERS FOR PREDICTING BACTERIAL INFECTION, BACTEREMIA, AND SEVERITY OF INFECTION IN THE EMERGENCY DEPARTMENT

It is estimated that between 3 and 29% of all

cases of bacteremia detected in the ED were discharged home. Most were of urinary origin, where
appropriate antibiotic treatment would have been
routinely prescribed, and did not produce adverse
events. But in a variable proportion, possibly up to
45%, it was uncontrolled primary bacteremia,
which would force a complete revision of the diagnosis and treatment plan. Diagnoses made via this
clinical circuit are far from insignificant and include
endocarditis, abdominal processes, osteomyelitis,
soft tissue venous catheter infections and bacteremias of uncertain origin, among others3-6. Finally, in recent years the indications for performing
blood cultures, their clinical performance and diagnostic interest have been controversial, especially
with respect to EDs. Different targets for improvement have been established, such as increasing the
percentage of positive blood cultures, reducing the
number of contaminations and limiting the number of blood cultures in patients who are not ultimately admitted to hospital7,8.
All this raises the need for clinical criteria to facilitate and improve decision-making on the indication for microbiological studies, hospital admission
and empirical antibiotic treatment. Different studies
have sought to determine the clinical variables that
may be indicators of BI and/or severity, in an attempt to establish predictive criteria, with mixed
results. In recent years interest has focused on plasma markers such as C-reactive protein (CRP) and
procalcitonin (PCT), among others, which could be
useful and improve diagnostic performance based
on clinical variables such as temperature, tachycardia, hypotension, anemia or leukocytosis9-11. It has
also been postulated that neopterin could be a
marker for viral infection and thus a criterion to
avoid antibacterial treatments9,12. A relatively new
marker is midregional proadrenomedullin
(proADM), which could be useful as a marker of
serious BI13, and this has raised the possibility of
comparing the diagnostic and/or prognostic capabilities of different molecules. Several studies have
evaluated the utility of these markers to detect BI in
critical and pediatric patients, but not in the ED.
The aim of our study was to analyze the capacity
of certain clinical and laboratory variables and the
biomarkers CRP, PCT, pro-ADM and neopterin to
predict BI, bacteremia and severity in the ED.

Method
The study was prospective, observational, descriptive and analytical. The study population
Emergencias 2012; 24: 348-356

consisted of patients with suspected BI who consulted the medical area of the ED (excluding surgical, gynecological and pediatric areas) and had
blood samples taken for cultures. Our ED, belonging to a public and university 624-bed hospital, attends about 30,000 medical emergencies
per year. According to the ED medical area protocol, blood cultures were performed for patients
consulting for fever (except those with suspected
viral respiratory infection) or in those with axillary temperature above 37C, and also in those
with nonspecific deterioration (limited expressiveness due to dementia, mental decline, psychiatric disease or advanced age) and the attending physician needed to rule out the
presence of an infectious process. The study was
approved by the hospital ethics committee. Cases were recruited from March 1 to April 30,
2009. We daily recorded the list of patients with
blood culture samples and began follow up via
computerized medical histories. In all cases the
following variables were recorded: age, sex, comorbidity according to the Charlson index14,
presence or absence of leukocytosis > 15 x109/L,
presence or absence of bands (segmented neutrophils), presence or absence of neutropenia
(< 1 x 109/L) and CRP values, all from initial laboratory study in the ED. Then PCT, proADM and
neopterin were measured, using the same samples.
Blood cultures (2 aerobic and 1 anaerobic
vials) were automatically processed using
Bact/Alert (bioMerieux, Durham, NC, USA). The
isolation of coagulase-negative Staphylococcus
was only considered significant when detected in
at least two blood cultures and the clinical picture
was compatible. The determination of serum CRP
was performed using a turbidimetric assay (RCRP,
Siemens Dimension Rxl Max, Siemens), and values
< 3 mg/L were considered normal. Plasma PCT
and proADM were determined by immunofluorescence assay (Kryptor, BRAHMS AG, Germany),
and values < 0.5 ng/mL and 0.39 nmol/L respectively were considered normal. Finally, serum
neopterin was determined using a competitive
ELISA (Neopterin ELISA, IBL, Germany) and values < 10 nmol/L were considered as normal.
With 30-day case follow up, we recorded the
definitive diagnosis according to the report made
by the attending emergency physician, hospital
admission or subsequent out-patient consultations. In doubtful cases the definitive diagnosis
recorded was that reached by consensus of three
observers. Patients were grouped according to
clinical diagnosis and microbiological data: a)
349

P. Tudela et al.

Group I: documented microbiological infection

(bacterial, fungal or viral), b) Group II: possible infection without microbiological documentation, c)
Group III: febrile syndrome of uncertain origin, d)
Group IV: non-infectious disease (negative microbiological tests and the existence of an alternative
diagnosis). The Charlson Index was classified into
four categories according to a previous model15.
Bacteremia was defined as the presence of bacteria in blood cultures. Severe events were recorded, defined as admission to ICU, or death in the
first 7 days, or surgery in the first 48 hours of admission.
Data were expressed as proportions, or medians with interquartile range. We performed ROC
curves with areas under the curve (AUC) and their
corresponding 95% confidence intervals (CI) for
predicting BI, bacteremia and severity, which was
used to determine optimal cutoff points (those
with maximum sum of specificity and sensitivity)
to convert continuous variables into dichotomous
variables. These were compared and analyzed in
three ways: a) between groups with proven BI
and infectious disease, b) between groups with
and without bacteremia, and finally c) between
groups with and without severe events. Comparisons were made using chi-square or Fisher test
for categorical variables, and Student t test for
continuous variables. P values less than 0.05 were
considered statistically significant. Multivariate
analysis was performed using logistic regression.
The results were expressed as odds ratios (OR)
with 95% CI and P values. Finally, we evaluated
the diagnostic performance of the tests by calculating sensitivity, specificity, positive predictive values and negative likelihood ratios. Data were analyzed using the Statistical Package SPSS version
14.0 for Windows.

Results
During the study period, blood cultures were
obtained for 457 patients, of whom 45 were excluded from the analysis for missing data, so 412
(90.1%) cases were finally studied. Clinical characteristics and distribution by diagnostic groups
are shown in Table 1. Documented infection
(Group I), mostly BI, accounted for 28.3%; possible infection (Group II) accounted for 39.3%,
undetermined fever (Group III) accounted for
9.9% and cases with non-infectious disease
(Group IV) accounted for 22.3%. Blood cultures
were positive in 53 (12.8%) of cases. The most
common microorganism identified was E. coli
350

with 18 (33.3%) cases and there were 7 (13.2%)

cases of polymicrobial infection. Severity criteria
were met in 34 (8.2%) cases. Table 2 shows the
distribution of clinical diagnoses, percentage of
cases with documented microbiological bacteremia and the microorganisms detected in
Groups I and II.
Considering that the objective of the study
was BI prediction, statistical analysis was only
performed for the data from patients with documented BI (Group I), excluding the 4 cases of
non-bacterial infection (3 fungal and 1 viral) (Tables 1 and 2). As for the prediction of BI values,
the AUC for each biomarker was: 0.685 (95% CI:
0.613 to 0.757) for PCT, 0.648 (95% CI: 0.572
to 0.725) for CRP, 0.585 (95% CI 0.507 to
0.663) and 0.574 for proADM (95% CI 0.495 to
0.652) for neopterin. Univariate and multivariate
analyzes are shown in Table 3. The independent
variables were the presence of band neutrophils,

Table 1. Clinical characteristics of patients

N (%)
Age, median (range)
69 (54-78)
Sex
Men
249 (60.4%)
Women
163 (39.6%)
Charlson Index
0
59 (14.3%)
1-2
232 (56.3%)
3-4
65 (15.8%)
5
56 (13.6%)
Group I. Documented infection
117 (28.3%)
Bacterial
113 (27.4%)
Fungal
3 (0.7%)
Viral
1 (0.2%)
Group II. Probable Infection
162 (39.3%)
Group III. Unclarified Fever
41 (9.9%)
Group IV. Non-infectious disease
92 (22.3%)
Final diagnoses of non-infectious processes
Neoplastic disease
24 (26%)
Liver cirrhosis
14 (15.2%)
Non-infectious arthritis
8 (8.6%)
Inflammatory bowel disease
6 (6.5%)
Autoimmune disease*
5 (5.4%)
Heart Failure
5 (5.4%)
Acute renal failure
4 (4.3%)
Atrial fibrillation
3 (3.2%)
Epilepsy
2 (2.1%)
Hypotension
2 (2.1%)
Miscellaneous**
19 (20.6%)
Episodes with bacteremia
53 (12.8%)
Episodes with severity
34 (8.2%)
Mortality
21 (5%)
Critical
14 (3.3%)
Surgery
2 (0.4%)
*Includes: lupus erythematosus, leukocytoclastic vasculitis, polyarteritis
nodosa, rheumatic polymyalgia, Still's disease. **Includes: anemia,
stroke, atrioventricular block, myocardial infarction, cholestasis,
ischemic colitis, irritable bowel syndrome, diabetes, dementia,
dehydration, pulmonary fibrosis, adrenal insufficiency, back pain,
anxiety, head injury, toxicoderma, pericarditis, insect bite, deep vein
thrombosis.
Emergencias 2012; 24: 348-356

BIOLOGICAL MARKERS FOR PREDICTING BACTERIAL INFECTION, BACTEREMIA, AND SEVERITY OF INFECTION IN THE EMERGENCY DEPARTMENT

Table 2. Description of diagnoses, microbiological documentation and microorganisms for patients in Groups I and II
Diagnoses

Microbiological
documentation

Positive blood
cultures

Other tests

Pneumonia

87

30 (34.8%)

6 (7.1%)

UA: 11, SC: 10, BAL: 2

Microorganisms

S. pneumoniae (18), H. influenzae (2), P. jirovecci (2),

Streptoccocus spp (2), E. coli (1), Moraxella catarrhalis (1),
Klebsiella spp (1), Pseudomonas spp (1)
Bronchial
78
8 (10.2%)
0 (0%)
SC: 8
H. influenzae (3), S. pneumoniae (2), S. aureus (2),
Pseudomonas spp (2)
Urinary
44
42 (95.5%)
18 (40%)
UC: 42
E. coli (32), Klebsiella spp (7), Enterococcus spp (3),
S. anginosus (1) Citrobacter spp (1), Providencia spp (1)
Febrile neutropenia
17
1 (5.8%)
1 (5.8%)

Comamomas testosteroni
Primary bacteremia
14
14 (100%)
14 (100%)

Enterococcus spp (4), S. aureus (3), Pseudomonas spp (3),

S. gallolyticus (3), E. coli (2), Klebsiella spp (1),
Proteus spp (1), Salmonella spp (1)
Cellulitis
9
7 (77.7%)
4 (44.4%)
ExC: 3
Streptoccocus spp (5), S. aureus (3)
Enteritis
7
2 (28.5%)
1 (14.2%)
SC: 1
E. coli (1), Clostridium difficile (1)
Billiary
4
1 (25%)
1 (25%)

E. coli
Meningitis
3
3 (100%)
3 (100%)
CSFC: 1
S. pneumoniae (3)
Venous catheter
3
3 (100%)
3 (100%)

Enterobacter spp (1), Pseudomonas spp (1),

S. epidermidis (1) Stenotrophomonas maltophilia (1)
Endocarditis
1
1 (100%)
1 (100%)

S. epidermidis
Peritonitis
1
1 (100%)
0 (0%)
AFC
Bacteroides spp
Arthritis
1
1 (100%)
0 (0%)
JFC
S. aureus
Empyema
1
1 (100%)
0 (0%)
PLC
Fusobacterium nucleatum
Aspergillosis
1
1 (100%)
0 (0%)
UA & SC
Aspergillus spp
Hepatitis
1
1 (100%)

IgM
Hepatitis A virus
Miscellaneous*
7
0 (0%)

*Includes: pharyngitis (2), dental abscess, Mediterranean spotted fever, viral encephalitis, oral herpes, psoas abscess. UA: antigens in urine, SC: sputum
culture, BAL: bronchoalveolar lavage, UC: urine culture, ExC: exudate culture, SA: stool antigen, CSFC: cerebrospinal fluid culture, AFC: ascitic fluid
culture, JFC: joint fluid culture, PFC: pleural fluid culture.

PCR  70 mg/L and PCT  0.4 ng/mL. The diagnostic performance values of these latter three
variables, considered separately and in combination (Table 4) was very limited.
For predicting bacteremia, AUC values for
each biomarker were: 0.799 (95% CI: 0.734 to
0.864) for PCT, 0.711 (95% CI: 0.625 to 0.796)
for pro- ADM, 0.617 (95% CI: 0.526 to 0.708)
for neopterin, and 0.558 (95% CI 0.475-0.640)
for CRP. The univariate and multivariate analyses
are shown in Table 5. The independent variables
were PCT > 1 ng/mL and proADM > 2 nmol/L.
In regard to diagnostic performance, PCT and
pro-ADM showed a specificity of 84.1% and
85.2% respectively, with specially high negative
predictive values of 94.1% for PCT and 92.7%
for ProADM (Table 4).
For the prediction of severity, the AUC values
for each biomarker were: 0.729 (95% CI: 0.627
to 0.832) for proADM, 0.716 (95% CI: 0.612 to
0.820) for PCT, 0.668 (95% CI: 0.567 to 0.770)
for neopterin and 0.661 (95% CI: 0.565 to
0.756) for CRP. The univariate and multivariate
analyses are shown in Table 6. The independent
variables were male sex, PCT > 1 ng/mL and
proADM > 1.94 nmol/L. With respect to diagnostic performance, PCT and proADM showed
specificities of 80.7% and 82.5% respectively,
with specially high negative predictive values of
95.9% for PCT and 96% for proADM (Table 4).
Emergencias 2012; 24: 348-356

Discussion
Our results, in terms of the general profile of
the sample, are similar to other series and seem
representative of the usual scenario in hospital
EDs. In general, suspected infection was subsequently confirmed in 57-83% of cases, although
microbiological confirmation was only obtained in
34-49%. This means that 16-46% of the patient
sample did not have infection, which suggests
that the level of suspicion of infection in the ED is
very high. The frequency of bacteremia was 1519% and 6-12% died16-19.
As for the capacity to predict BI, none of the
markers separately showed sufficient performance
in our study. Variable results have been obtained
in previous studies, although comparison with
them is complicated by different inclusion criteria
(fever, leukocytosis, suspected infection), the cutoffs for CRP or PCT, or study areas (ED, critical
care, hospitalization) 16-20. Regarding markers in
combination, both PCR and PCT combined with
band neutrophils achieved greater sensitivity but
low specificity. Overall, this limited diagnostic performance suggests they only have orientative value, without diagnostic or therapeutic capacity.
In general, BI in the ED is difficult to predict,
probably because infection is a heterogeneous and
multifactorial phenomenon. The great variability in
biomarker behavior may be attributable to biologi351

P. Tudela et al.

Table 3. Comparison of cases with and without bacterial infection

Variables

Univariate Analysis
Total
N = 205
n (%)

No infection
(Group IV)
N = 92
n (%)

Infection
(Group I)
N = 113
n (%)

Age
Gender

Median [IQR]
69 [54-78]
68.5 [54-76]
60.5 [71-79]
Male
116 (56.6)
46 (50)
70 (61.9)
Female
89 (43.4)
46 (50)
43 (38.1)
Charlson Index
0
33 (16.1)
15 (16.3)
18 (15.9)
1-2
109 (53.2)
45 (48.9)
64 (56.6)
3-4
36 (17.6)
15 (16.3)
21 (18.6)
5
27 (13.2)
17 (18.5)
10 (8.8)
Leukocytes ( 15,000/mL)
No
142 (69.3)
70 (76.1)
72 (63.7)
Yes
63 (30.7)
22 (23.9)
41 (36.3)
Band neutrophils
No
127 (62)
69 (75)
58 (51.3)
Yes
78 (38)
23 (25)
55 (48.7)
Neutropenia (<1.000/mL)
No
197 (96.1)
88 (95.7)
109 (96.5)
Yes
8 (3.9)
4 (4.3)
4 (3.5)
CRP (mg/L)
< 70
98 (47.8)
58 (63)
40 (35.4)
 70
107 (52.2)
34 (37)
73 (64.6)
PCT (ng/mL)
< 0.4
117 (57.1)
67 (72.8)
50 (44.2)
 0.4
88 (42.9)
25 (27.2)
63 (55.8)
proADM (nmol/L)

2.1
157 (77)
79 (86.8)
78 (69)
> 2.1
47 (23)
12 (13.2)
35 (31)
Neopterin (nmol/L)

19
77 (37.6)
41 (44.6)
36 (31.9)
> 19
128 (62.4)
51 (55.4)
77 (68.1)
CRP: C reactive protein, PCT: procalcitonin, proADM: proadrenomedullin, IQR, interquartile range.

cal characteristics of the different bacterial species,

inflammatory response in each patient and each
source of infection, and finally kinetic aspects,
which entail limitations of spot measurements.
Considering all this, the limited diagnostic performance of the biomarkers studied could be explained by the large number of cases with little
systemic impact and/or the fact that measurements were made very early, at initial stages of the
infectious process10,21,22. In this regard, the possibility of at least two determinations (at 12 or 24
hours) should be considered in order to rule out BI
more safely. In contrast, the diagnostic performance of biomarkers in the detection of bacteremia
was significantly higher. While CRP showed little
predictive capacity, as in previous studies23, the

Multivariate Analysis
P

OR [95% CI]

2.383 [1.258-4.514]

0.008

< 0.001

2.254 [1.195-4.251]

0.012

< 0.001

2.254 [1.168-4.350]

0.015

0.138
0.086
0.231

0.056
0.001
0.999

0.003
0.062

PCT cutoff point of 1 ng/mL showed high specificity and negative predictive values, which indicates its usefulness as an exclusion test. Previous
studies have indicated the clinical utility of PCT,
proposing cutoffs of 0.2 or 0.4 ng/mL to exclude
bacteremia, with 97-99% negative predictive value24,25. In a recent study, Mller et al26 suggested
that PCT is a good predictor of predicting bacteremia in patients with community-acquired
pneumonia. These authors suggest that blood cultures should be performed only when PCT exceeds
0.25 mg/L. However, some meta-analyses indicate
that PCT is only moderately predictive of bacteremia and suggest using it in combination with
other clinical or laboratory data27,28. In this regard,
our team recently proposed a predictive model

Table 4. Diagnostic performance in predicting bacterial infection, bacteremia and severity

Sensitivity
(95% CI)

Specificity

+PV

PV

Prediction of bacterial infection

Band neutrophils
48.7 (39.2-58.2) 75.0 (64.7-83.2) 70.5 (59.0-80.0) 54.3 (45.3-63.1)
CRP  70 mg/L
64.6 (55.0-73.2) 63.0 (52.3-72.7) 68.2 (58.4-76.7) 59.2 (48.8-68.9)
PCT  0.4 ng/mL
55.8 (46.1-65.0) 72.8 (62.4-81.3) 71.6 (60.8-80.4) 57.3 (47.8-66.3)
CRP  70 mg/L &/or band neutrophils
80.5 (71.8-87.1) 50.0 (39.5-60.5) 66.4 (57.8-74.1) 67.6 (55.1-78.2)
PCT  0.4 ng/mL &/or band neutrophils
73.5 (64.2-81.1) 58.7 (47.9-68.7) 68.6 (59.4-76.6) 64.3 (53.0-74.2)
Prediction of bacteremia
PCT > 1.0 ng/mL
64.2 (49.7-76.5) 84.1 (79.8-87.7) 37.4 (27.6-48.2) 94.1 (90.8-96.3)
proADM > 2 nmol/L
54.7 (40.6-68.2) 85.2 (81.0-88.6) 35.4 (25.3-46.8) 92.7 (89.2-95.2)
Prediction of severity
PCT > 1 ng/mL
61.8 (43.6-77.3) 80.7 (76.3-84.5) 22.3 (14.7-32.3) 95.9 (92.9-97.7)
proADM > 1.94 nmol/L
61.8 (43.6-77.3) 82.5 (78.2-86.1) 24.1 (15.9-34.7) 96.0 (93.1-97.8)
PV: predictive value, LR: likelihood ratio, CRP: C reactive protein, PCT: procalcitonin, proADM: proadrenomedullin.

352

+LR

LR

1.9
1.7
2
1.6
1.7

0.6
0.5
0.6
0.3
0.4

4.2
3.9

0.4
0.5

3.2
3.5

0.4
0.4

Emergencias 2012; 24: 348-356

BIOLOGICAL MARKERS FOR PREDICTING BACTERIAL INFECTION, BACTEREMIA, AND SEVERITY OF INFECTION IN THE EMERGENCY DEPARTMENT

Table 5. Comparison of cases with or without bacteremia

Variables

Univariate Analysis
Total
N = 412
n (%)

No bacteremia
N = 359
n (%)

Bacteremia
N = 53
n (%)

Age
Gender

Median [IQR]
69 [54-78]
68 [53-78]
72 [62-79]
Male
249 (60.4)
221 (61.6)
28 (52.8)
Female
163 (39.6)
138 (38.4)
25 (47.2)
Charlson Index
0
59 (14.3)
52 (14.5)
7 (13.2)
1-2
232 (56.3)
203 (56.5)
29 (54.7)
3-4
65 (15.8)
55 (15.3)
10 (18.9)
5
56 (13.6)
49 (13.6)
7 (13.2)
Leukocytes ( 15.000/mL)
No
313 (76)
281 (78.3)
32 (60.4)
Yes
99 (24)
78 (21.7)
21 (39.6)
Band neutrophils
No
271 (65.8)
245 (68.2)
26 (49.1)
Yes
141 (34.2)
114 (31.8)
27 (50.9)
Neutropenia (< 1.000/mL)
No
385 (93.4)
335 (93.3)
50 (94.3)
Yes
27 (6.6)
24 (6.7)
3 (5.7)
CRP (mg/L)

65.5
194 (47.1)
177 (49.3)
17 (32.1)
> 65.5
218 (52.9)
182 (50.7)
36 (67.9)
PCT (ng/mL)

1.0
321 (77.9)
302 (84.1)
19 (35.8)
> 1.0
91 (22.1)
57 (15.9)
34 (64.2)
proADM) (nmol/L)

2
329 (80)
305 (85.2)
24 (45.3)
>2
82 (20.0)
53 (14.8)
29 (54.7)
Neopterin (nmol/L)

71.9
372 (90.3)
335 (93.3)
37 (69.8)
> 71.9
40 (9.7)
24 (6.7)
16 (30.2)
CRP: C reactive protein, PCT: procalcitonin, proADM: proadrenomedullin, IQR, interquartile range.

that combines PCT < 0.4 ng/mL with Charlson index, which allowed distinguishing a group with
very low probability of bacteremia (0 to 2.9%)
and a negative predictive value of 95%29. Recent
studies have shown similar results30-32.
In addition, others have shown that PCT may

Multivariate Analysis
P

OR [95% CI]

0.039
0.225
0.929

0.004
0.006
0.999
0.019
< 0.001

5.887 [2.904-11.932] < 0.0001

< 0.001

3.090 [1.518-6.291]

0.002

< 0.001

be useful and safe to improve empiric antibiotic

prescription33, which is often done intuitively and
is sometimes unjustified. Hausfater et al17 found
that antibiotic treatment was prescribed in 83%
of cases when only 71% had BI, and was even
prescribed in 35% of patients with viral infection.

Table 6. Comparison of cases with and without severity

Variables

Univariate Analysis
Total
N = 412
n (%)

Age
Gender

With severity
N = 378
n (%)

Without severity
N = 34
n (%)

Median [IQR]
68 [57-74.3]
69 [54-78]
69 [53-8.3]
Male
26 (76.5)
249 (60.4)
223 (59.0)
Female
8 (23.5)
163 (39.6)
155 (41.0)
Charlson Index
0
3 (8.8)
59 (14.3)
56 (14.8)
1-2
19 (55.9)
232 (56.3)
213 (56.3)
3-4
2 (5.9)
65 (15.8)
63 (16.7)
5
10 (29.4)
56 (13.6)
46 (12.2)
Leukocytes (> 15,000/mL)
No
21 (61.8)
313 (76.0)
292 (77.2)
Yes
13 (38.2)
99 (24.0)
86 (22.8)
Band neutrophils
No
17 (50.0)
271 (65.8)
254 (67.2)
Yes
17 (50.0)
141 (34.2)
124 (32.8)
Neutropenia (< 1,000/mL)
No
30 (88.2)
385 (93.4)
355 (93.9)
Yes
4 (11.8)
27 (6.6)
23 (6.1)
Bacteremia
No
20 (58.8)
359 (87.1)
339 (89.7)
Yes
14 (41.2)
53 (12.9)
39 (10.3)
CRP (mg/L)

126.5
15 (44.1)
288 (69.9)
273 (72.2)
> 126.5
19 (55.9)
124 (30.1)
105 (27.8)
PCT (ng/mL)

1
13 (38.2)
318 (77.2)
305 (80.7)
>1
21 (61.8)
94 (22.8)
73 (19.3)
proADM (nmol/L)

1.94
13 (38.2)
324 (78.8)
311 (82.5)
> 1.94
21 (61.8)
87 (21.2)
66 (17.5)
Neopterin (nmol/L)

32.5
12 (35.3)
268 (65.0)
256 (67.7)
> 32.5
22 (64.7)
144 (35.0)
122 (32.3)
CRP: C reactive protein, PCT: procalcitonin, proADM: proadrenomedullin, IQR, interquartile range.
Emergencias 2012; 24: 348-356

Multivariate Analysis
P

OR [95% CI]

0.846
0.046

0.308 [0.127-0.747]

0.009

< 0.001

3.601 [1.499-8.654]

0.004

< 0.001

4.582 [1.902-11.038]

0.001

0.019

0.043
0.043
0.264
< 0.001
0.001

< 0.001

353

P. Tudela et al.

Although the clinical criterion for predicting BI

was acceptable, antibiotic treatment was excessively prescribed, perhaps for lack of other supporting data. In this regard, the use of a biomarker would be of help to rationalize its use.
As for the prediction of severity, PCT and especially proADM showed high specificity and
negative predictive values, and this raises the
possibility of using them to exclude severity in
patients with suspected infection in the ED.
These results are consistent with those of previous studies13,34. Predicting severity in the ED is of
great importance, since it conditions certain decisions such as the route of administration for
the initial treatment, the area where the patient
should be located and the start of support measures. This is really the subgroup where early initiation of antimicrobial treatment provides the
greatest clinical benefit. There are different models for estimating severity, based on clinical variables such as tachycardia, tachypnea, shock or
altered level of consciousness, with acceptable
results 35, but the presence of these conditions
can involve variable interpretation and often reflect an already established serious situation. In
this regard, we believe that predictive biomarkers could help establish an estimate of severity
early in the process and with greater objectivity36.
The present study has certain limitations. Firstly, unlike other studies, it did not include initial
temperature as a predictor, considering it is often
artifacted due to self-medication with antipyretics.
The inclusion criterion, ultimately based on the
clinicians indication for a blood culture, introduces an important subjective component and
possible selection bias. Secondly, most of the diagnoses were made by the attending physician
and some may have been erroneous. Furthermore, in Group II, there may have been cases of
BI that were not recorded as such, for different
reasons (previous intake of antibiotics, absence of
biological samples), which could mean fewer cases in Group I. However, this is a complicated matter since a definitive indicator of BI at this early
stage is not available. When microbiological tests
are not performed in patients with a clinical diagnosis of infection, how to make a definitive diagnosis is an intractable dilemma21,22. In our study,
assigning patients to groups for comparison
seems somewhat exclusive, but we believe this
approach is more rigorous, and there appears to
be little option until a future determination becomes available, and moreover, our study design
was similar to those of other studies16.
354

Cases of viral infection were probably underestimated, and we did not perform a thorough investigation, so some may have been included in
Group II or III without being defined. This may
explain why neopterin was not discriminative. Biomarker values can be altered by the presence of
comorbidities such as renal or cardiac failure, but
this was not taken into account. The Charlson index categorization and the cutoff vale for leukocytosis may be questionable, but these were derived
from previous work and ultimately they were intended to facilitate practical application. We did
not record cases of previous antibiotic treatment,
nor the time of onset of symptoms, and both
these factors could have affected the performance
of some biomarkers. Finally, the criteria for admission to critical care units may differ between hospitals and this impedes extrapolating our results
to other centers.
In conclusion, for the overall prediction of bacterial infection in the ED, the following are indicative, but have limited diagnostic capacity: the existence of band neutrophils in the hemogram,
CRP levels  70 mg/L and PCT  0.4 ng/mL. In
contrast, PCT
1 ng/mL and proADM
1.9
nmol/L showed considerable diagnostic capacity
for more specific prediction, exclusion of bacteremia and severity in patients with suspected infection in the ED. Thus, until new molecular techniques appear that allow the early detection of
bacterial infection, PCT and proADM can help in
making vital clinical decisions in the ED, such as
initiating empirical antibiotic treatment, the need
for hospitalization and the prediction of severity.

References
1 Talan DA. Infectious disease issues in the emergency department.
Clin Infect Dis. 1996;23:1-12.
2 Len Gil C, Garca-Castrillo Riesgo L, Moya Mir M, Artigas Raventos
A, Borges Sa M, Candel Gonzlez FJ, et al. Documento de consenso
(SEMES-SEMICYUC). Recomendaciones del manejo diagnstico-teraputico inicial y multidisciplinario de la sepsis grave en los servicios
de urgencias hospitalarios. Emergencias. 2007;31:375-87.
3 Epstein D, Raveh D, Schlesinger Y, Rudensky B, Gottehrer NP, Yinnon
AM. Adult patients with occult bacteremia discharged from the
emergency department: epidemiological and clinical characteristics.
Clin Infect Dis. 2001;32:559-65.
4 Tudela P, Queralt C, Gimnez M, Carreres A, Tor J, Sopena N, et al.
Deteccin de bacteriemia en pacientes dados de alta en urgencias:
estudio de 61 episodios. Med Clin (Barc). 1998;111:201-4.
5 Ramos JM, Masia M, Elia M, Gutirrez F, Royo G, Bonilla F, et al. Epidemiological and clinical characteristics of occult bacteremia in an
adult emergency department in Spain: influence of blood culture results on changes in initial diagnosis and empiric antibiotic treatment.
Eur J Clin Microbiol Infect Dis. 2004;23:881-7.
6 Villamil I, Rodrguez L, Villacin MJ, Van den Eyden A, Garca-Zabarte
M. Bacteriemia en pacientes dados de alta en el servicio de urgencias. Emergencias. 2005;17:62-6.
7 Wilson ML. Outpatient blood cultures: progress and unanswered
questions. Eur J Clin Microbiol Infect Dis. 2004;23:879-80.
8 Ibero Esparza C, Regidor Sanz E, Daz Pedroche C, Garca de Casasola G. Si fiebre, hemocultivos? Rev Clin Esp. 2010;210:559-66.

Emergencias 2012; 24: 348-356

BIOLOGICAL MARKERS FOR PREDICTING BACTERIAL INFECTION, BACTEREMIA, AND SEVERITY OF INFECTION IN THE EMERGENCY DEPARTMENT

9 Herzum I, Renz H. Inflammatory markers in SIRS, sepsis and septic

shock. Curr Med Chem. 2008;15:581-7.
10 Reinhart K, Meisner M, Brunkhorst FM. Markers for sepsis diagnosis:
what is useful? Crit Care Clin. 2006;22:503-19.
11 Simon L, Gauvin F, Amre DK, Saint-Louis P, Lacroix J. Serum procalcitonin and C-reactive protein levels as markers of bacterial infection: a
systematic review and meta-analysis. Clin Infect Dis. 2004;39:206-17.
12 Prat C, Domnguez J, Andreo F, Blanco S, Pallars A, Cuchillo F, et al.
Procalcitonin and neopterin correlation with aetiology and severity
of pneumonia. J Infect. 2006;52:169-77.
13 Christ-Crain M, Morgenthaler NG, Struck J, Harbarth S, Bergmann A,
Muller B. Mid-regional pro-adrenomedullin as a prognostic marker in
sepsis: an observational study. Crit Care. 2005;9:R816-24.
14 Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of
classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-83.
15 Murray SB, Bates DW, Ngo L, Ufberg JW, Shapiro NI. Charlson Index
is associated with one-year mortality in emergency department patients with suspected infection. Acad Emerg Med. 2006;13:530-6.
16 De Kruif MD, Limper M, Gerritsen H, Spek CA, Brandjes DP, ten Cate
H, et al. Additional value of procalcitonin for diagnosis of infection
in patients with fever at the emergency department. Crit Care Med.
2010;38:457-63.
17 Hausfater P, Juillien G, Madonna-Py B, Haroche J, Bernard M, Riou B.
Serum procalcitonin measurement as diagnostic and prognostic
marker in febrile adult patients presenting to the emergency department. Crit Care. 2007;11:R60.
18 Indino P, Lemarchand P, Bady P, de Torrente A, Genne L, Genne D.
Prospective study on procalcitonin and other systemic infection markers in patients with leukocytosis. Int J Infect Dis. 2008;12:319-24.
19 Kofoed K, Andersen O, Kronborg G, Tvede M, Petersen J, Eugen-Olsen J, et al. Use of plasma C-reactive protein, procalcitonin, neutrophils, macrophage migration inhibitory factor, soluble urokinase-type
plasminogen activator receptor, and soluble triggering receptor expressed on myeloid cells-1 in combination to diagnose infections: a
prospective study. Crit Care. 2007;11:R38.
20 Julin A, Palomo MJ, Ortiz R, Pedrosa A, Parejo R, Salcedo R. Utilidad
de la procalcitonina y la protena C reactiva en el paciente con sepsis en urgencias. Emergencias. 2009;21:23-7.
21 Marshall JC,Reinhart K. Biomarkers of sepsis. Crit Care Med.
2009;37:2290-8.
22 Schuetz P, Christ-Crain M, Muller B. Procalcitonin and other biomarkers to improve assessment and antibiotic stewardship in infections-hope for hype? Swiss Med Wkly. 2009;139:318-26.
23 Wyllie DH, Bowler IC, Peto TE. Bacteraemia prediction in emergency

Emergencias 2012; 24: 348-356

medical admissions: role of C reactive protein. J Clin Pathol.

2005;58:352-6.
24 Chirouze C, Schuhmacher H, Rabaud C, Gil H, Khayat N, Estavoyer
JM, et al. Low serum procalcitonin level accurately predicts the absence of bacteremia in adult patients with acute fever. Clin Infect
Dis. 2002;35:156-61.
25 Prat C, Sancho JM, Domnguez J, Xicoy B, Gimnez M, Ferra C, et
al. Evaluation of procalcitonin, neopterin, C-reactive protein, IL-6
and IL-8 as a diagnostic marker of infection in patients with febrile
neutropenia. Leuk Lymphoma. 2008;49:1752-61.
26 Muller F, Christ-Crain M, Bregenzer T, Krause M, Zimmerli W, Mueller B,
et al. Procalcitonin levels predict bacteremia in patients with communityacquired pneumonia: a prospective cohort trial. Chest. 2010;138:121-9.
27 Becker KL, Snider R, Nylen ES. Procalcitonin assay in systemic inflammation, infection, and sepsis: clinical utility and limitations. Crit Care
Med. 2008;36:941-52.
28 Jones AE, Fiechtl JF, Brown MD, Ballew JJ, Kline JA. Procalcitonin test in the
diagnosis of bacteremia: a meta-analysis. Ann Emerg Med. 2007;50:34-41.
29 Tudela P, Lacoma A, Prat C, Mdol J, Gimnez M, Barallat J, et al.
Prediccin de bacteriemia en los pacientes con sospecha de infeccin en urgencias. Med Clin (Barc.). 2010;135:685-90.
30 Kim MH, Lim G, Kang SY, Lee WI, Suh JT, Lee HJ. Utility of procalcitonin as an early diagnostic marker of bacteremia in patients with
acute fever. Yonsei Med J. 2011;52:276-81.
31 Riedel S, Melndez JH, An AT, Rosenbaum JE, Zenilman JM. Procalcitonin as a marker for the detection of bacteremia and sepsis in the
emergency department. Am J Clin Pathol. 2011;135:182-9.
32 Su CP, Chen TH, Chen SY, Ghiang WC, Wu GH, Sun HY, et al. Predictive model for bacteremia in adult patients with blood cultures
performed at the emergency department: A preliminary report. J Microbiol Immunol Infect. 2012 (en prensa).
33 Tang H, Huang T, Jing J, Shen H, Cui W. Effect of procalcitonin-guided treatment in patients with infections: a systematic review and
meta-analysis. Infection. 2009;37:497-507.
34 Lee CC, Chen SY, Tsai CL, Wu SC, Chiang WC, Wang JL, et al. Prognostic value of mortality in emergency department sepsis score, procalcitonin, and C-reactive protein in patients with sepsis at the emergency department. Shock. 2008;29:322-7.
35 Knott JC, Tan SL, Street AC, Bailey M, Cameron P. Febrile adults presenting to the emergency department: outcomes and markers of serious illness. Emerg Med J. 2004;21:170-4.
36 Shapiro NI, Trzeciak S, Hollander JE, Birkhahn R, Otero R, Osborn TM,
et al. A prospective, multicenter derivation of a biomarker panel to assess risk of organ dysfunction, shock, and death in emergency department patients with suspected sepsis. Crit Care Med. 2009;37:96-104.

355

P. Tudela et al.

Biomarcadores para la prediccin en urgencias de infeccin bacteriana, bacteriemia

y gravedad
Tudela P, Prat C, Lacoma A, Mdol JM, Domnguez J, Gimnez M, Tor J
Objetivos: Analizar la posible utilidad de algunas variables clnicas y de laboratorio, as como la protena C reactiva
(PCR), la procalcitonina (PCT), la proadrenomedulina (proADM) y la neopterina para predecir infeccin bacteriana (IB),
bacteriemia y gravedad en el servicio de urgencias (SU).
Mtodo: Se incluy a los pacientes atendidos en el SU en los que se tomaron muestras para hemocultivos, de quienes
se registr: edad, sexo, ndice de Charlson, leucocitosis, clulas en banda, neutropenia, resultados microbiolgicos y niveles de PCR, PCT, proADM y neopterina. La existencia de gravedad se defini como un ingreso en la unidad de cuidados intensivos, fallecimiento o necesidad de ciruga urgente. Se realiz el anlisis univariado y multivariado, las curvas ROC y se calcul el rendimiento diagnstico.
Resultados: Se incluy a 412 pacientes, que fueron clasificados en cuatro grupos: (I) infecciones microbiolgicamente
documentadas (28,3%), (II) posible infeccin, sin documentacin microbiolgica (39,3%), (III) fiebre no aclarada
(9,9%) y (IV): ausencia de enfermedad infecciosa (22,3%). Los hemocultivos fueron positivos en 53 casos (12,8%), y
34 casos fueron considerados graves (8,2%). Para la prediccin de IB, las variables independientes fueron PCR  70
mg/L, PCT  0,4 ng/mL y presencia de bandas, aunque la precisin diagnstica fue limitada. Para la prediccin de
bacteriemia, las variables independientes fueron PCT > 1 ng/mL y proADM > 2 nmol/L, con valores predictivos negativos (VPN) del 94% y 93%, respectivamente. Para la prediccin de la gravedad, las variables independientes fueron
PCT > 1 ng/mL y proADM > 1,94 nmol/L, con VPN en torno al 96% para ambas.
Conclusiones: En los pacientes con sospecha de infeccin en el SU, la existencia de bandas y los niveles de PCR y PCT
pueden ser orientativos de IB, aunque con un valor diagnstico limitado. En cambio, la determinacin de PCT y proADM muestran un considerable rendimiento diagnstico para la exclusin de bacteriemia y de gravedad. [Emergencias
2012;24:348-356]
Palabras clave: Bacteriemia. Biomarcadores. Urgencias.

356

Emergencias 2012; 24: 348-356