Comparison between white blood cell count ...

Clin Chem Lab Med 2014; aop

Laura Magrini*, Giulia Gagliano, Francesco Travaglino, Francesco Vetrone, Rossella Marino, Patrizia Cardelli, Gerardo Salerno and Salvatore Di Somma, on behalf of the GREAT Network

Comparison between white blood cell count, procalcitonin and C reactive protein as diagnostic and prognostic biomarkers of infection or sepsis in patients presenting to emergency department Abstract Background: Procalcitonin (PCT) is currently the most studied infection biomarker and its blood levels seem to mirror the severity of illness and outcome. PCT is widely used together with other biomarkers, such as white blood cells (WBC) count and C reactive protein (CRP), in order to guide antibiotic therapy. This study aimed to verify the diagnostic and prognostic power of WBC, CRP and PCT in patients with suspected infection in emergency department (ED). Methods: A total of 513 patients presenting to the ED with signs/symptoms of local infections or sepsis were enrolled. APACHEII score and in-hospital death were recorded. Patients were subdivided into quartiles by age, and the biomarkers were measured at baseline. Receiver operating characteristics (ROC) curves for evaluating diagnostic and prognostic role of PCT, CRP and WBC were calculated for each variable alone and combined. Results: When compared each other for PCT, CRP, and WBC there was no significant difference between the four subgroups. A direct correlation between PCT and WBC was found in the II, III, and IV quartiles (the highest correlation, r = 0.34, p < 0.0003). PCT alone or when combined with WBC showed the best diagnostic and prognostic power at ROC analysis.

*Corresponding author: Dr. Laura Magrini, MD, Emergency Medicine, Medical-Surgery Sciences and Translational Medicine, Sapienza University of Rome, Sant’Andrea Hospital, Via di Grottarossa 1035-1039, 00189 Rome, Italy, Phone: +39 0633775754, Fax: +39 0633775018/5890, E-mail: [email protected] Giulia Gagliano, Francesco Travaglino, Francesco Vetrone, Rossella Marino and Salvatore Di Somma: Department of Medical-Surgery Sciences and Translational Medicine, Sant’Andrea Hospital, Sapienza University of Rome, School of Medicine and Psychology, Emergency Medicine, Rome, Italy Patrizia Cardelli and Gerardo Salerno: Department of Molecular and Clinical Medicine, Sant’Andrea Hospital, Sapienza University of Rome, School of Medicine and Psychology, Rome, Italy

Conclusions: Our data demonstrate that WBC, but more CRP and PCT are reliable diagnostic and prognostic biomarkers, when considered in combination and with severity clinical score. PCT confirms its stronger usefulness as a diagnostic marker of sepsis. A multi-diagnostic tools approach is fundamental to perform a correct and rapid diagnosis of infection and sepsis in ED. Keywords: C reactive protein; diagnosis; emergency department; infection; procalcitonin; prognosis; white blood cells. DOI 10.1515/cclm-2014-0210 Received February 26, 2014; accepted April 8, 2014

Background Despite progress over the past half century in treating patients with sepsis, the incidence of infections, sepsis, and septic shock is still increasing [1]. Several factors could explain the increased incidence of sepsis, including the aging population, the increasing survival of patients who have cancer or are immune-depressed, the increasing numbers of invasive medical interventions, and the rising of multidrug-resistant bacteria [2, 3]. For patients with sepsis in-hospital mortality ranges from 28.3% to 41.1% in North America and Europe [4]. As a consequence, an early diagnosis is crucial to its successful treatment [5, 6]. Biomarkers have a pivotal role in this process because they can indicate the presence of the severity of infections or sepsis [7–9], can differentiate bacterial from viral and fungal infection, and distinguish systemic sepsis from local infection. Other potential uses of biomarkers include the decision making to start antibiotic treatment and its response [10]. Today a multitude of biomarkers has been proposed in the field of sepsis, but the most validated

Authenticated | [email protected] author's copy Download Date | 5/11/14 7:08 PM

2 Magrini et al.: WBC, PCT, CRP and infections

remains procalcitonin (PCT) that has become part of the International Guidelines for Management of Severe Sepsis and Septic Shock 2012 [11]. Nevertheless, in everyday clinical practice, worldwide, white blood cells (WBC) count and C reactive protein (CRP) still represent the cornerstones on which is based the diagnosis and prognosis of infected patients together with other signs and symptoms like fever, tachycardia, and tachypnea, when a documented or suspected infection is detected [11, 12]. In emergency department (ED) diagnosis of sepsis can be delayed by the overcrowding [13] and by the complexity of patients, often elderly and with multiple comorbidities. All these variables can alter signs and symptoms [14]. However, these are also the most relevant risk factors for developing sepsis [1]. Moreover, in early stages of the process, the source of infection may be unclear and the related systemic response indistinguishable from non-infectious diseases [15]. The aim of the present prospective study was to clarify the reliability of WBC count, CRP and PCT alone or coupled together as diagnostic and prognostic biomarkers of infection and sepsis in patients referred to the emergency setting in different age groups.

Materials and methods This was a retrospective, observational study conducted in a university 400-bed hospital, S. Andrea Hospital of Rome. We studied 513 patients (Table 1) (263 males and 250 females; mean age 71.7 ± 15.9

years) referred to our ED with symptoms of infection, and in which a diagnosis of infection or sepsis was formulated. The diagnosis was performed in agreement with the International Guidelines for Management of Severe Sepsis and Septic Shock 2012 criteria. Infection was documented, as appropriate, by chest X-ray, ultrasound exams, computed tomography, and also by laboratory tests, urine culture, blood culture, and cultural examination of buffer. We collected blood samples to measure PCT, CRP, and blood cell count at baseline. Arterial blood gas analysis was performed. Prognostic risk stratification was performed using the Acute Physiology and Chronic Health Evaluation (APACHE) II score. Charlson index was also calculated (score ranging from 0 to ≥ 3). In order to standardize data collection, the worst physiologic and laboratory values from the last 24 h before study inclusion while patients were still in the ED were used to calculate the APACHE II scores. Clinical data of patients were recorded, as well as in-hospital death. Entire population was subdivided into quartiles by age (I: 19–69 years, II: 70–77 years, III: 78–83 years, IV: 84–102 years). This retrospective clinical trial was designed following the criteria of the Declaration of Helsinki and approved by the ethical committee of the hospital.

PCT assay PCT was performed at the arrival in the ED (T0). Venous blood withdrawal was performed for each patient with lithium heparinized tubes. After clotting time, each sample was analyzed without storage and performed on the mini-Vidas (Bio-Merieux, Marcy L’Etoile, Craponne, France), using the ELFA technique (Vidas, Brahms PCT). The lowest concentration measured by the assay is 0.05 ng/mL (95% percentile). This is considered the cut-off for healthy individuals on the basis of manufacturer’s instructions. Following manufacturer’s instructions the repeatability (inter-assay precision) and interassay reproducibility (inter-assay precision) were calculated using a

Table 1 Patients’ characteristics in total population, septic group, and in non-septic group.

Total population n = 513

Septic patients n = 221

Non-septic patients n = 292

p-Valuea

Age, yearsb Gender M/F, n HR, bpmb RR, breaths/minb T, °Cb WBC, 103/μLb PCT, ng/mLb CRP, mg/dLb Charlson Index score Past historyc DM CD Neoplasms COPD CKD

71.78 ± 15.90 263/250 95.58 ± 19.48 21.54 ± 5.55 37.31 ± 1.14 13,229.38 ± 7083.80 8.57 ± 31.61 15.75 ± 13.73 1.86 ± 1.56

73.88 ± 15.47 105/116 102.25 ± 18.54 26.6 ± 5.16 37.58 ± 1.19 15,969 ± 8323.87 17.16 ± 46.88 20.19 ± 14.87 2.08 ± 1.74

70 ± 16.08 142/140 88.79 ± 17.62 19.25 ± 4.99 36.83 ± 0.9 11,155.29 ± 5102.55 2.37 ± 6.79 12.34 ± 11.81 1.69 ± 1.39

ns < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.005

23.8% 59.1% 5.8% 25.1% 12.9%

11.3% 31.2% 4% 16.6% 8.1%

22.6% 53% 2.4% 28.7% 10.7%

ns < 0.002 ns ns ns

p-Value is between septic vs. non-septic groups; bvalues expressed as mean ± SD; cvalues expressed as percentage. bpm, beats per minute; CD, cardiovascular disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CRP, C reactive protein; DM, diabetes mellitus; HR, heart rate; ns, not significant; PCT, procalcitonin; RR, respiratory rate; SD, standard deviation; T °C, body temperature in Celsius; WBC, white blood cells. a


Magrini et al.: WBC, PCT, CRP and infections 3

protocol based on the recommendations of CLSIEP5-A2 document23. Intra-assay coefficients of variation as determined in our laboratory (15 human serum samples in duplicates in 5 parallel determinations) were, depending on the sample concentration, between 2.0% and 3.2%. Assay sensitivity (95th percentile on 200 normal subjects) is < 0.05 ng/mL.

WBC assay WBC was performed at arrival in ED (T0). Each sample was analyzed in central hospital laboratory by a Sysmex XE-2100 used to determine full blood counts and WBC counts (WBC linearity: 0–440.00 × 103/μL; body fluid linearity: WBC-BF: > 0.050 × 103/μL).

CRP assay CRP was assessed at baseline (T0). Each sample was analyzed in central hospital laboratory. CRP was measured on VITROS® Chemistry Systems (Ortho-Clinical Diagnostics) using a “MicroSlide” method. The reference range for CRP measurements was 0.01–1.50 mg/dL.

Statistical analysis Data were normally distributed and they are expressed as mean ± standard deviation (SD) or percentage (%). Comparison of two means was performed using the Student’s t-test, and between dichotomous variables it was assessed by χ2-test. Spearman’s rank test has been used for coefficients of correlation. To determine the prognostic value of biomarkers for death in total population, and for diagnosis of sepsis, the receiver operating characteristics (ROC) curve and area under the curve (AUC) were constructed and the p-value was obtained. ROC curves were calculated for each variable alone, and for PCT, CRP and WBC combined each other. Multiple regression analysis was performed with PCT, WBC, gender and age in quartiles as prognostic independent factors. A p-value < 0.05 was considered to be statistically significant in all tests. Statistical analysis was performed using MedCalc Statistical Software version 13.0.4 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2014).

Results Table 1 describes the characteristics of patients. Patients studied were mainly older than 65 years of age; there was no difference for gender. When the studied population was subdivided into septic and non-septic groups, there was a statistical significant difference for heart rate, respiratory rate, body temperature, WBC, PCT, and CRP between the two groups. On the basis of past history most patients were affected by cardiovascular diseases and chronic obstructive pulmonary disease (COPD). On the basis of Charlson index score, patients’ groups results were expressed as follows: 0 = 15.8%; 1 = 29.4%; 2 = 28.6%; ≥ 3 = 25.5%. Mean score value was significantly higher in septic patients than in non-septic patients (p < 0.005) (Table 1). There was no statistical difference between the four quartiles groups when the mean values were compared with each other, either for PCT, CRP, and for WBC (Table 2). For APACHE score there was a statistical significant difference between the quartiles I vs. III (p < 0.01), and vs. IV (p < 0.0001); II vs. IV (p < 0.0007); II vs. IV (p < 0.006) (Table 2). A direct significant correlation between PCT and WBC mean values was not found in patients with age lower than 70 years but in the last three quartiles of age (the highest correlation, r = 0.34, p < 0.0003), while for PCT and CRP was never found any correlation (Table 3). ROC curve analysis showed a significant diagnostic and prognostic role of PCT, and diagnostic for CRP when considered separately (for diagnosis of sepsis: PCT AUC 0.79, p < 0.0001–WBC AUC 0.53, p = 0.78–CRP AUC 0.72, p < 0.001 (Figure 1A); for prognosis: PCT AUC 0.72, p < 0.0001–WBC AUC 0.63, p = 0.18–CRP AUC 0.53, p = 0.23) (Figure 1B), and for diagnosis of sepsis in combination (for diagnosis of sepsis: PCT+WBC AUC 0.79, p < 0.0001–CRP+WBC AUC 0.71, p < 0.001–PCT+CRP+WBC AUC 0.80, p < 0.0001 (Figure 2A); for prognosis: PCT+WBC AUC 0.65, p = 0.06–CRP+WBC

Table 2 Laboratory values (mean ± SD) of white blood cells, procalcitonin, C reactive protein, and APACHE score in each quartile of age.

I quartile 19–69 years II quartile 70–77 years III quartile 78–83 years IV quartile 84–102 years

PCT, ng/mL

WBC, 103/μL

CRP, mg/dL

APACHE score, %

6.83 ± 16.67 6.34 ± 18.45 6.67 ± 18.63 14.72 ± 57.73

12,891.13 ± 7833.10 12,690.36 ± 6277.90 12,767.27 ± 6601.29 13,899.90 ± 7239.41

16.07 ± 14.03 14.68 ± 12.50 16.32 ± 15.58 15.74 ± 11.74

16.37 ± 18.33 22.42 ± 12.57 25.20 ± 15.10 39.41 ± 25.63

APACHE, Acute Physiology and Chronic Health Evaluation; CRP, C reactive protein; PCT, procalcitonin; WBC, white blood cells; SD, standard deviation.



Table 3 Correlation between procalcitonin and white blood cells, and between procalcitonin and C reactive protein in each quartile of age.

PCT and WBC I quartile 19–69 years II quartile 70–77 years III quartile 78–83 years IV quartile 84–102 years PCT and CRP I quartile 19–69 years II quartile 70–77 years III quartile 78–83 years IV quartile 84–102 years

r (95% CI)

p-Value

0.140 (–0.0121–0.285) 0.295 (0.133–0.442) 0.265 (0.103–0.413) 0.340 (0.161–0.498)

0.071

–0.16 (–0.331–0.0024) 0.067 (–0.121–0.250) 0.075 (–0.104–0.251) 0.035 (–0.174–0.242)

< 0.0005 < 0.0016 < 0.0003

0.053 0.48 0.40 0.74

CRP, C reactive protein; PCT, procalcitonin; WBC, white blood cells.

AUC 0.64, p = 0.64–PCT+CRP+WBC AUC 0.65, p = 0.65) (Figure 2B). In multiple regression analysis, including as predictors both WBC and PCT, age and gender, for diagnosis PCT (p < 0.0001), WBC (p < 0.0001), and age (p = 0.004) are all independently significantly associated to the risk of sepsis, while the most important significant predictor for hospital death appeared to be the age, and also WBC. PCT did not reach predictive significance for the risk to develop adverse outcomes (Table 4).

Discussion In our study the reliability of WBC count, CRP and PCT used alone or in combination as diagnostic and prognostic

A 100

B

100

T0_WBC T0_PCT T0_CRP

60 40

Sensitivity

80

80 Sensitivity

biomarkers of infection and sepsis in patients in an emergency setting has been investigated. From our results it seems that there is no difference between the three biomarkers levels as for gender and age, while the presence of comorbidities, such as, mainly cardiovascular and pulmonary diseases, could impact the severity of sepsis as shown in Table 1. Even though a large volume of information has been collected about managing and treating septic patient [16– 19], the lack of knowledge about the diagnostic criteria for sepsis by emergency teams is one of the greatest factors limiting its adequate treatment. Many efforts have been made in recent years to give a standardized definition to this syndrome and consequently make the diagnosis easier. Joint and combined efforts culminated in the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference [2, 20] that paved the way to the most recent International Guidelines for Management of Severe Sepsis and Septic Shock published in 2012. To increase survival, diagnosis must be promptly made in ED and an appropriate treatment should be immediately started. Furthermore, ED setting is not comparable to regular ward or an Intensive Care Unit (ICU), since the length of visit time performed by the emergency physician is limited by the presence of overcrowding [13, 21], while patients boarding to ED is not selected, and subjects with acute diseases are, in many cases, elderly and with multiple comorbidities that could mask infections. As a consequence, in ED more than in other settings, reliable, simple and rapid markers capable to detect infections and sepsis are needed. Currently, the universally accepted criteria for the diagnosis of sepsis include either patient’s general variables (fever, hypothermia, tachycardia, tachypnea, altered mental status, hyperglycemia), either inflammatory variables (leukocytosis or leukopenia or normal WBC count

T0_WBC T0_PCT T0_CRP

60 40 20

20

0

0 0

40 80 100-Specificity

0

40 80 100-Specificity

Figure 1 ROC curve for (A) diagnosis of sepsis and (B) prognosis of in-hospital death. In (A) ROC curve for diagnosis of sepsis: WBC AUC 0.533, p = 0.780; PCT AUC 0.798, p < 0.0001; CRP AUC 0.720, p = 0.001. In (B) ROC curve for outcome (in-hospital death) in total population: WBC AUC 0.635, p = 0.180; PCT AUC 0.723, p < 0.0001; CRP AUC 0.538, p = 0.239. T0, time of arrival in ED; WBC, white blood cells; PCT, procalcitonin; CRP, C reactive protein.



B 100

100 80 60 40 20 0

WBC_PCT_CRP_Diag WBC CRP Diag WBC_PCT_Diag 0

60

Sensitivity

Sensitivity

A

80 60 40 20 0

WBC PCT out WBC CRP PCT out WBC CRP PCT out 0 40 80 100-Specificity

100-Specificity

Figure 2 ROC curve for (A) diagnosis of sepsis and (B) prognosis of in-hospital death for combination of biomarkers. In (A) ROC curve for diagnosis of sepsis: WBC+PCT AUC 0.799, p < 0.0001; WBC+CRP AUC 0.718, p = 0.001; WBC+PCT+CRP AUC 0.800, p < 0.0001. In (B) ROC curve for outcome (in-hospital death) in total population: WBC+PCT AUC 0.656, p = 0.065; WBC+CRP AUC 0.643, p = 0.097; WBC+PCT+CRP AUC 0.658, p = 0.095. WBC, white blood cells; PCT, procalcitonin; CRP, C reactive protein; Diag, diagnosis; Out, outcome.

Table 4 Multiple regression analysis for (A) diagnosis of sepsis, and (B) risk for in-hospital death.

Coefficient

(A) Independent variables (Constant) –0.03967 Quartiles age 0.05345 0.00002224 WBC PCT 0.003180 Gender –0.01676 (B) Independent variables (Constant) –0.09295 Quartiles age 0.03341 WBC 0.000006101 PCT 0.0007645 Gender 0.04118

Standard error

r partial

t

p-Value

0.01854 0.000002846 0.0006417 0.04064

0.1296 0.3337 0.2191 –0.01896

2.884 7.813 4.955 –0.413

0.0041 < 0.0001 < 0.0001 0.6801

0.01250 0.000001883 0.0004212 0.02722

0.1228 0.1483 0.08370 0.06983

2.673 3.240 1.815 1.513

0.0078 0.0013 0.0702 0.1310

CRP, C reactive protein; PCT, procalcitonin; WBC, white blood cells.

with greater than 10% immature forms, CRP, PCT) both potentially altered especially in elderly patients [11, 20]. Although modification in inflammatory response and WBC are well described and recognized in elderly due to the immune senescence and to the progressive aging of the bone marrow, little is known about PCT modification in this type of subjects. While the correlation between CRP and WBC is easily comprehensible during sepsis, and also their plasma level variations in immuno-depressed and elderly patients, the relationship between PCT and inflammation, even the role played during infections, still remains uncertain. Immune senescence is generally characterized by chronic, low-grade, systemic inflammation and impaired responses to immune challenge. Thymic involution is well known, and there are substantial data in literature on the functional decline of T cells with age. The decreased proliferation is correlated to a decreased release of interleukin-2 and soluble IL-2 receptor. However, soluble IL-2 receptor expression on the cell surface is normal. Interferon-γ as the main T-helper-1 cytokine is produced less by lymphocytes

of the elderly, whereas the T-helper-2 cytokines interleukin-4 and interleukin-10 are produced in higher amounts as compared to stimulated lymphocytes of young donors. Monocyte function seems to be increased in the elderly. Leukocytes of elderly produce higher amounts of interleukins 1, 6, and 8 and tumor necrosis factor-α after induction with lipopolysaccharide than leukocytes from young donors [22, 23]. The production of PCT is induced by pro-inflammatory cytokines, such as interleukin-1, interleukin-6 and tumor necrosis factor-α, and expressed by nearly all kinds of parenchymal cells [24]. It is unknown how the increased inflammation and declined functional reserve of parenchymal cells in elderly would affect the response of PCT, but on the basis of current knowledge, it could be postulated that, unlike CRP and WBC, PCT remains a reliable expression of infection even in the more elderly population. Moreover, it must be taken into account that inflammation (present also in infectious conditions), which takes place on a cellular level, is triggered by a wide variety of factors, such as oxidative stress, ingestion of toxins,



excess exposure to ultraviolet radiation, hormonal changes, that are typical of the aging process. For this reason we evaluated PCT and CRP values and WBC count as positive predictive power in detecting sepsis when used alone and combined together in a wide range of age populations, including the elderly. We did not find relevant differences between age expressed in quartiles and the studied biomarkers (Table 2). From our results in the group of elderly patients there was a significant increase of PCT, but not in CRP. This indicates the different behavior of these two biomarkers in case of infection or sepsis, probably due to the different kinetics, and also to the different metabolism and biochemical origin of the two biomarkers. In a very recent study, Loonen et al. demonstrated that PCT was able to significantly discriminate patients with positive blood cultures from those with negative blood cultures, while CRP levels remained similar in both groups. ROC curve analysis showed that PCT was higher than CRP in differentiating patients with blood culture proven bloodstream infection from those without [25]. Also Arai et al. showed that in a group of adult infected patients (aged more than 50 years) blood culture results were not affected by WBC and CRP, but they were affected by PCT (and platelets), and they suggested that these results might help in explaining the mechanisms of sepsis [26]. For APACHE score there was a statistical difference in each quartile evaluated with a progressive increase of the score for the higher quartiles, probably because this score represents the severity of organ conditions. It could also be different due to the age of patients, i.e., comorbidities mostly affect the score in the elderly. In a recent paper by Innocenti et al. it was shown that APACHE score, together with other prognostic scores, was significantly higher in septic patients with a bad prognosis, and could be considered an accurate tool for risk stratification of sepsis in ED [27]. To our knowledge there are no data available in literature regarding the subdivision of quartiles of age and on the evaluation of PCT or CRP and WBC in infectious or sepsis conditions in ED. When the correlation between PCT and WBC or between PCT and CRP was evaluated, we found that a direct significant correlation between PCT and WBC mean values was found in the II, III and IV quartiles (the highest correlation, r = 0.34, p < 0.0003), but not in the first (Table 3), while no significant correlation was shown between PCT and CRP. The correlation between PCT and WBC was not really strong but there was a rapid response of PCT and WBC to the therapy which could be explained by their similar kinetics. The kinetics of CRP is slower and its response more non-specific to antibiotic therapy. Literature data agree with our findings, PCT has been shown to

detect infections and sepsis more rapidly than CRP, also in an ED setting [28–30]. When ROC analysis was performed PCT was shown to remain the best significant predictor for diagnosis of sepsis (AUC 0.79, p < 0.0001) and for prognosis (AUC 0.72, p < 0.0001), together with CRP for diagnosis of sepsis (AUC 0.72, p < 0.001) but not for prognosis, while WBC did not show significant results (Figure 1). These data confirm our previous studies on the diagnostic role of PCT and CRP in patients affected with local infections and sepsis in ED [30]. The explanation of the different behavior of these three biomarkers in the diagnosis of sepsis could be found in their different kinetics, and in the influence of age and sex on WBC, but also for their daily biological fluctuations as recently demonstrated by Tang et al. [31]. It is well known that in elderly people the WBC are not always increased, even in severe sepsis, because of the progressive aging of the bone marrow that results in a decrease of WBC population. In fact, one of the four SIRS findings is to have WBC > 12,000/mm3 or < 4000/ mm3 [11, 32, 33]. When we combined the three biomarkers together for diagnosis of sepsis, there was an increase in the predictive diagnostic value (AUC 0.80, p < 0.0001), even greater than the combination of PCT+WBC (AUC 0.79, p < 0.0001), and of PCT+CRP (AUC 0.71, p < 0.001) (Figure 2). Jaimes et al. demonstrated in a large study population that the repeated measurements of PCT, D-Dimer, and CRP were not able to clearly discriminate septic from non-septic patients, but PCT was the only biomarker to identify a subgroup of severely ill patients in ED. The diagnostic accuracy of the three biomarkers, according to latent class analysis (LCA) gold standard for diagnosis of sepsis, was demonstrated by high ROC curves (CRP 0.71; PCT 0.95; D-Dimer 0.73). They did not use a combination of the three biomarkers, but one was considered singularly [15]. Instead, the innovative and adjunctive information of our study is that a combination of the three studied biomarkers could improve diagnostic accuracy, even greater than a combination of only two of them. For the prognosis, in our population, a combination of the three parameters appears to be less useful in comparison to the biomarkers when evaluated singularly, in fact while AUC were slightly high (WBC+PCT 0.65–WBC+CRP 0.64–WBC+CRP+PCT 0.65) (Figure 2), they did not reach statistical significance. We can speculate that biomarkers used singularly could be more useful in predicting outcomes. This is probably because the wide inter-individual variability of values for WBC or for the other two biomarkers causes an extreme dispersion of significant values. The results of a multiple regression analysis to identify variables associated with sepsis or mortality in total population show that



PCT, WBC and age subdivided in quartiles were strongly independently associated with the presence of sepsis (p