Objective To investigate the accuracy of procalcitonin (PCT) as a diagnostic marker of nosocomial sepsis (NS) and define the most accurate cut-off to distinguish infected from uninfected neonates.
Setting Six neonatal intensive care units (NICUs).
Patients 762 neonates admitted to six NICUs during a 28-month observational study for whom at least one serum sample was taken on admission.
Main outcome measures Positive and negative predictive values at different PCT cut-off levels.
Results The overall probability of an NS was doubled or more if PCT was >0.5 ng/ml. In very-low-birth-weight (VLBW) infants, a cut-off of >2.4 ng/ml gave a positive predictive value of NS near to 50% with a probability of a false-positive diagnosis of NS in about 10% of the patients.
Conclusions In VLBW neonates, a serum PCT value >2.4 ng/ml prompts early empirical antibiotic therapy, while in normal-birth-weight infants, a PCT value ≤2.4 ng/ml carries a low risk of missing an NS.
Statistics from Altmetric.com
Successful treatment of nosocomial sepsis (NS, late-onset sepsis) in neonates depends crucially on early diagnosis and prompt beginning of targeted antibiotic therapy. NS is nevertheless difficult to detect because initial clinical manifestations are usually not specific. Without reliable infection markers, neonatologists often start antibiotics for suspected sepsis, exposing many patients to unnecessary treatment. Some authors have proposed procalcitonin (PCT), a calcitonin precursor, as a marker for NS in neonates because tissue PCT release increases during infections.1,–,6 A reliable diagnostic marker enabling early detection of NS could reduce the unnecessary recourse to antibiotic therapy, which is increasingly responsible for the selection of resistant micro-organisms.
We designed this study to investigate the accuracy of serum PCT as a diagnostic marker for detecting NS in day-to-day clinical practice in neonatal intensive care units (NICUs). To define the most reliable PCT cut-off to distinguish between infected and uninfected neonates, we measured PCT levels on admission to the NICU in 762 neonates of normal and very low birth weight (VLBW, less than 1501 g). In those neonates who subsequently developed NS, we measured PCT levels during the infection.
What is already known on this topic
Early diagnosis of neonatal sepsis remains a relevant problem in neonatal intensive care units.
Fast procalcitonin (PCT) assay has been investigated as a marker of infection.
What this study adds
The usefulness of the PCT assay varies according to different birthweight groups of newborns.
About 50% of neonates with a birth weight of <1501 g and PCT of >2.4 ng/ml have a sepsis, while a lower PCT virtually excludes a sepsis, irrespective of birth weight.
In a 28-month prospective, multicentre observational study on NS in six NICUs, PCT levels were measured on blood samples routinely collected on admission from 762 enrolled patients (including 205 VLBW, <1501 g). Neonates with early-onset (maternally transmitted) sepsis were excluded from the study. In neonates developing signs of NS at least 48 h after admission, PCT levels were measured within 24 and 48 h after the onset of the NS. The highest PCT value measured during the sepsis was considered for analysis of accuracy.
The criterion for NS diagnosis was defined according to clinical, biological and microbiological parameters: postnatal signs of sepsis, and a positive central or peripheral blood culture, looking at the recommendations of the Center for Disease Control,7 as modified and applied in our study on nosocomial infections. C reactive protein value was included among the biological signs of sepsis.8 NS was defined as confirmed (culture-positive) or suspected (culture-negative) sepsis.
PCT levels were assessed using a quantitative immunoluminometric method (Lumitest PCT-Q; BRAHMS, Hennigsdorf, Germany). The lower detection limit was 0.1 ng/ml. All PCT samples were assayed by the same laboratory, without knowing the results from the microbiology laboratory.
To assess the accuracy of PCT for the detection of neonatal NS, the sensitivity, specificity, and positive and negative likelihood ratios (LR+=sensitivity/1−specificity and LR−=specificity/1−sensitivity) at different PCT cut-off levels were calculated. These factors express how much the test increases the probability of a positive (or negative) diagnosis, and were analysed with receiver operating characteristic (ROC) curves, in VLBW neonates and neonates with a birth weight of >1500 g. The probability of NS after the result of PCT (positive predictive value) was calculated, multiplying LR+ and LR− by the probability of NS incidence, and was expressed as percentage of patients with NS. Inferences on medians were tested with the Mann–Whitney test; p values <0.05 were considered to indicate statistical significance. Statistical analysis was performed using STATA version 10.0 software.
The institutional ethical boards of all centres approved the study.
Over the 28-month study period, 762 newborn infants were prospectively enrolled: 557 had a birth weight of >1500 g, and 205 were VLBW. The proportion of VLBW neonates in the six NICUs participating in the study ranged from 20.8% to 40.2%. The mean age of neonates at first PCT blood sampling was 1.97±5.23 days (median 0).
Sixty-five of the 762 neonates had NS, giving an overall incidence of 8.53% (95% CI 6.64% to 10.74%). In these 65 infected neonates, 78 NS events occurred: in 33 neonates (22 VLBW) there were 41 suspected NS events, and in 32 neonates (21 VLBW) there were 37 confirmed events, giving an incidence density of 0.44% (95% CI 0.35% to 0.55%). The incidence of NS was higher in VLBW than in heavier-birthweight neonates (19.02%, 95% CI 9.68% to 19.68% vs 4.67%, 95% CI 3.07% to 6.76%, respectively).
The most frequently detected bacterium was Klebsiella pneumonia (16), followed by Staphylococcus epidermidis (5), Pseudomonas aeruginosa (5), Enterococcus and Enterobacter cloacae (two each), Escherichia coli (1), Staphylococcus aureus (1) and fungi (5). They were all detected in blood.
The median value of the highest PCT level during NS was significantly higher in neonates with sepsis than in those without (3.58 vs 0.49 ng/ml; p<0.001) and in culture-positive than in culture-negative sepsis (10.83 vs 2.27 ng/ml; p<0.001). In the ROC curve analysis of all patients, the area under the curve (AUC) was 0.80 (95% CI 0.75 to 0.85) and varied from 0.79 (95% CI 0.71 to 0.86) for neonates with a birth weight of >1500 g to 0.82 (95% CI 0.74 to 0.89) for VLBW neonates (figure 1).
At all three predefined PCT cut-off levels, approximately two patients testing positive were required for one correct diagnosis of NS (table 1). Given the higher pretest probability among VLBW infants (ie, the actual incidence of NS), the positive predictive value was higher in VLBW than in heavier neonates. A cut-off of >2.4 ng/ml gave a positive predictive value of NS of approximately 50% in VLBW infants, with a probability of incorrectly diagnosing NS in about 10% of the patients (ie, a negative predictive value of 90%) (table 1). Conversely, a low cut-off level, <0.5 ng/ml, excluded the risk of missing a NS in fewer than 5% of the patients (ie, a negative predictive value higher than 95%).
When data were analysed for the subset of 29 neonates with NS caused by Gram-negative bacteria, the best cut-off value was ≥0.9 ng/ml (AUC 0.80), correctly identifying 86.21% of the patients, with an LR+ of 3.18 and LR− of 0.13.
Our findings suggest that serum PCT is a useful diagnostic marker for NS in day-to-day clinical practice in NICUs. We found that serum PCT increases the probability of detecting NS by at least doubling the probability expressed by the incidence of NS if the test is positive and is especially useful in VLBW neonates. In these patients, the high incidence of NS is sufficient to predict a positive test and to justify starting empirical antibiotic treatment while waiting for the culture results and beginning more precise therapeutic targeting. Culture results are usually obtained long after the onset of symptoms, and positive cultures do not always coincide with the clinical symptoms of NS. Thus, in clinical practice, cultures serve mainly to guide specific antibiotic therapy, and second to provide a definitive diagnosis of neonatal NS. In neonates with a birth weight of >1500 g who carry a low probability of NS, negative PCT tests provide more reliable diagnostic information than positive tests.
When we sought to define the most reliable PCT cut-off to distinguish between NS infected and uninfected neonates at the onset of infection, the calculated AUC was in the same range as that described by others (0.74–0.93).3,–,6 In a study using PCT to discern late-onset sepsis in preterm neonates, Vazzalwar et al2 reported sensitivities comparable with ours (97% for a cut-off of >0.5 ng/ml and 72% for a cut-off of >1.0 ng/ml) but higher specificities (80% and 100%). In a study with a high incidence of infections, Jacquot et al reported a negative predictive value of 100% and suggested that a cut-off of ≤0.6 ng/ml rules out NS.6 Unexpectedly, we did not confirm either the high LRs with a 0.99 ng/ml cut-off (LR+ 8.17 and LR− 0.03),5 or those reported, from 4.18 to 3.17 ng/ml, with cut-off points in the range 0.59–1.34.3 Conversely, our study suggests that a PCT value below 2.4 ng/ml may help to exclude NS, irrespective of birth weight. Interleukin 10 and nCD64 have been evaluated as neonatal sepsis markers, with a sensitivity of 95% in combination.9 Their complexity and high cost limit their suitability for routine use.
In conclusion, serum PCT can help to detect NS in neonates in NICU. In VLBW infants admitted to a NICU, a serum PCT value of >2.4 ng/ml can increase diagnostic accuracy before blood culture confirms NS onset. In neonates with a birth weight of >1500 g, a serum PCT value of ≤2.4 ng/ml suggests that NS is highly unlikely.
Funding The study was funded by the Italian Ministry of Health research grant number 99010661.
Competing interests None.
Ethics approval Ethics approval was provided by the institutional ethical board at each centre approved the study.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.