170 AJTCCM VOL. 30 NO. 4 2024
ORIGINAL RESEARCH: ARTICLES
Background. Hospital-acquired infection (HAI) in patients with COVID-19 admitted to the intensive care unit (ICU) is associated with
increased mortality. e ‘cytokine storm’ associated with COVID-19 leads to extreme elevation of inammatory biomarkers, including
C-reactive protein (CRP). Procalcitonin (PCT) has been shown to be more discriminative than CRP in distinguishing HAI from other
inammatory processes.
Objectives. To investigate the utility of PCT in detecting HAI in patients with severe COVID-19.
Methods. Clinical and laboratory data from all patients admitted to a dedicated ICU with conrmed severe COVID-19 from 1 April 2020
to 31 August 2020 were prospectively captured. HAI was conrmed by serial PCT and CRP measurements, as well as microbiological data
(positive microbiological cultures in clinical context). Data from patients who were on antibiotics on ICU admission, had a positive culture
for a presumed pathogen during the rst 48 hours of ICU admission, or already had suspected or proven HAI on admission were excluded.
Optimal cut-os with the highest sensitivity and specicity were determined. e discriminative power of PCT was assessed for each
outcome, using receiver operating characteristic (ROC) analysis describing the area under the curve. Similarly, negative predictive values
(NPVs) and positive predictive values (PPVs) were determined. e sensitivity and specicity for dierent PCT cut-o levels were calculated.
Results. Of 92 patients, 35 had conrmed HAI, which was signicantly associated with mechanical ventilation (p<0.001) and mortality
(p<0.001). ROC analysis demonstrated that a threshold PCT level of 0.22 μg/L resulted in 97% sensitivity and 40% specicity for predicting
HAI. Similarly, sensitivity and specicity for CRP were 91.4% and 38.6%, respectively, when the CRP level was 133 mg/L. In patients with a
PCT level <0.25 μg/L, the NPV was 92%, whereas for PCT levels >1.00 μg/L, the PPV was >50%. For PCT levels >40 μg/L, the PPV was 100%.
Conclusion. During HAI, PCT levels >1.00 μg/L had a moderate PPV of 52%, whereas levels <0.26 μg/L ruled out HAI with an NPV of
92%. With increased PCT values, the PPV rose to 100%, making it a better biomarker than CRP.
Keywords. Procalcitonin, SARS-CoV-2, COVID-19.
Afr J Thoracic Crit Care Med 2024;30(4):e1617. https://doi.org/10.7196/AJTCCM.2024.v30i4.1617
Study synopsis
What the study adds. During an episode of hospital-acquired infection (HAI) in patients with severe COVID-19, procalcitonin (PCT)
levels >1.00 μg/L had a moderate positive predictive value (PPV) of 52%, whereas levels <0.26 μg/L had a negative predictive value (NPV)
of 92% for proven HAI. For PCT levels >40 μg/L, the PPV was 100%.
Implications of the ndings. At levels <0.26 μg/L, PCT had an NPV >90%. is ‘rule-out’ characteristic of PCT may be especially
valuable in scenarios of diagnostic equipoise with regard to the presence of bacterial co-infection. Clinicians should take care to not
unjustiably associate elevations in PCT levels with the presence of bacterial co-infection, unless levels are extremely high, in which case
the PPV rises signicantly.
e utility of procalcitonin as a biomarker of hospital-acquired
infection in severe COVID-19
C Schmidt,1 MB ChB ; P S Nyasulu,2,3 MSc (Epidemiology), PGDip (Epidemiology), PhD ; I Fwemba,4 PhD (Biostatistics) ; U Lalla,1
MB ChB, FCP (SA), FCRP, Cert Critical Care (SA), Cert Pulmonology (SA) ; B W Allwood,1 MB ChB, FCP (SA), Cert Pulmonology
(SA), PhD ; A Parker,5 MB ChB, FCP (SA), Cert ID (SA), PhD ; J J Taljaard,4 MB ChB, MMed (Int) ; L N Sigwadhi,6 MSc (Bio-
statistics) ; J L Tamuzi,2 MD, MSc, MPH ; A E Zemlin,6 MB ChB, FC Path (SA) Chem, MMed (Chem), PhD ; R T Erasmus,6 MBBS,
FMCPath, FACB, DABCC, DHSM ; C F N Koegelenberg,1 MB ChB, FCP (SA), FCRP, Cert Pulmonology (SA), PhD
1 Division of Pulmonology, Department of Medicine, Faculty of Medicine and Health Sciences, Stellenbosch University and Tygerberg Hospital, Cape Town, South Africa
2 Division of Epidemiology and Biostatistics, Department of Global Health, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa
3 Division of Epidemiology and Biostatistics, School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa
4 School of Public Health, University of Zambia, Lusaka, Zambia
5
Division of Infectious Diseases, Department of Medicine, Faculty of Medicine and Health Sciences, Stellenbosch University and Tygerberg Hospital, Cape Town, South Africa
6 Division of Chemical Pathology, Department of Pathology, Faculty of Medicine and Health Sciences, Stellenbosch University and National Health Laboratory Service –
Tygerberg Hospital, Cape Town, South Africa
Corresponding author: C Schmidt (charle21.cs@gmail.com)
AJTCCM VOL. 30 NO. 4 2024 171
ORIGINAL RESEARCH: ARTICLES
Reports from many international and local studies in the early days
of the COVID-19 pandemic showed that bacterial co-infection was
relatively uncommon at initial presentation.[1,2] Hospital-acquired
infection (HAI) in patients admitted to an intensive care unit (ICU)
was also clearly shown to be associated with increased mortality.[1]
In any patient admitted to an ICU, several clinical, radiological and
laboratory markers may indicate the presence of an HAI. A major
challenge with SARS-CoV-2-infected patients is that the now well-
recognised ‘cytokine storm’ leads to, among other manifestations,
fever and extreme elevation of inammatory biomarkers, including
C-reactive protein (CRP).[3]
Procalcitonin (PCT) is a glycoprotein, the pro-peptide of calcitonin
devoid of hormonal activity. Under normal circumstances, it is
produced in the C-cells of the thyroid gland. In healthy humans,
serum PCT levels are undetectable (<0.1 µg/L).[4] Prior to the
COVID-19 pandemic, several studies demonstrated that PCT levels
were more discriminative than the white blood cell count and CRP
in distinguishing serious bacterial and fungal infection from other
inammatory processes.[4,5]
At the start of the present study, it was not yet clear whether bacterial
co-infection would play a major role early or late in severe COVID-19.
We therefore aimed to investigate the utility of PCT in detecting HAI
in patients with severe COVID-19 admitted to an ICU.
Methods
Setting and study design
Clinical and laboratory data from all patients with conrmed severe
SARS-CoV-2 pneumonia admitted to the dedicated COVID-19 ICU
at Tygerberg Hospital, Cape Town, South Africa, from 1 April 2020
to 31 August 2020 (the rst local wave) were prospectively captured
as part of a multidisciplinary study collaboration. Tygerberg Hospital
is a 1 380-bed tertiary referral centre that serves a population of ~3
million. e collection of data was approved by the Health Research
Ethics Committee of Stellenbosch University (ref. no. N20/04/002_
COVID-19). The investigators and authors had no access to
information that could identify individual participants during or aer
data collection.
Clinical and microbiological data
Patients who had no evidence of bacterial superinfection and who
were not on antibiotics on admission to the ICU were identied. Data
extracted included patient demographics, comorbidities, laboratory
data, serial PCT and CRP measurements, outcome, invasive
ventilation, microbiological data (blood cultures, tracheal aspirate
cultures, stool cultures and urine cultures) conrming infection, and
date of antibiotic initiation on the basis of conrmed or suspected
HAI. Data from patients who were on antibiotics on ICU admission,
had a positive culture for a presumed pathogen during the first
48 hours of ICU admission, or had a suspected or known HAI on
admission were excluded. e data on the rst proven HAI episode
were used for analysis (should more than one episode have occurred).
Patients categorised as having ‘suspected’ sepsis never (at any point)
had proven HAI.
Apart from demographic data and risk factors for severe COVID-19,
the highest form of respiratory/ventilatory support (up to intubation
and mechanical ventilation) and the daily clinical suspicion of HAI
(according to the treating physician) were documented. ‘Conrmed’
HAI was supported by confirmation of positive microbiological
data (positive blood, tracheal aspirate and urine cultures) excluding
culture contaminants, whereas ‘suspected’ could not be proven by
microbiological means.
All cultures were submitted to the on-site National Health
Laboratory Service microbiology laboratory and processed using
standard procedures. Identification and antibiotic susceptibility
testing of cultured isolates involved use of the automated VITEK
2 system (bioMérieux, France), and was supplemented where
necessary with the ETEST (bioMérieux, France) to conrm the
minimum inhibitory concentration. Positive culture results were
deduplicated based on site of sample collection, with a positive result
showing the same pathogen with the same susceptibility prole
within a 5-day period being considered a single episode. Organisms
such as coagulase-negative staphylococci and Bacillus cereus were
considered contaminants if they were only cultured once, or as
pathogens if they were cultured more than once in the same patient
in an appropriate setting (e.g. central line-associated bloodstream
infection) where the attending physician deemed these cultures to
be clinically signicant
.
PCT and CRP measurements
PCT was determined by Elecsys BRAHMS PCT (Roche Diagnostics,
Germany), an electrochemiluminescence immunoassay, measured on
the cobas e 601 (Roche Diagnostics, Germany). CRP was measured by
means of CRP4, an immunoturbidimetric method, on the cobas c 501
(Roche Diagnostics, Germany).
Outcome measures
e primary outcome measure was proven sepsis in patients with
severe COVID-19 pneumonia admitted to the COVID-19 ICU. e
secondary outcome measure was suspected sepsis. Covariates such as
positive culture, length of stay, age and discharge were all considered
as exposure factors.
Statistical analysis
Data were analysed using R Studio 4.2.3 (R Core Team, USA).
Statistical signicance was set at p<0.05 with corresponding 95%
condence intervals (CIs). Continuous variables were expressed as
means with standard deviations for normally distributed data and as
medians with interquartile ranges for non-normal data. Categorical
variables were expressed using frequencies and percentages. Pearson’s
χ2 test of independence was used to identify associations between
categorical variables and the outcomes of interest. e t-test was
used to compare the means of continuous data where the data had a
normal distribution, and the Mann-Whitney U-test where data did
not have a normal distribution. e PCT and CRP values on the day
of onset of either proven or suspected HAI were used for analyses.
We used a generalised model with binomial family to t the model
on proven and suspected PCT and CRP predictors. Optimal PCT
and CRP cut-os were determined using PRO in R Studio. Optimal
cut-os were determined to be the highest value at which sensitivity
and specicity were highest. ese were determined for COVID-19
172 AJTCCM VOL. 30 NO. 4 2024
ORIGINAL RESEARCH: ARTICLES
patients with both suspected and proven sepsis. e discriminative
power of PCT and CRP was assessed for each outcome, using
receiver operating characteristic (ROC) analysis describing the area
under the curve (AUC). Similarly, negative predictive values (NPVs)
and positive predictive values (PPVs) were also determined. e
sensitivity, specicity and positive likelihood ratio for dierent PCT
cut-o levels were calculated. To determine the prognostic accuracy of
PCT and CRP at both time points, ROC curves were constructed and
the AUCs were calculated together with the corresponding 95% CIs.
Ap-value <0.05 was considered statistically signicant in all analyses.
Results
In total, the data on 92 patients were included in the study (Table 1).
Of these, 35 patients had proven HAI conrmed by blood, tracheal
aspirate or urine cultures, 25 had suspected HAI, and 32 showed
no evidence of HAI. Notably, intubation, mechanical ventilation
and mortality were signicantly associated with both proven and
suspected HAI (p<0.001).
ROC curve analysis of PCT and CRP yielded AUCs of 0.690
and 0.572, respectively (Figs 1 and 2). A PCT level of 0.22 μg/L
demonstrated sensitivity of 97% and specicity of 40% (Table 2). For
CRP, a level of 133 μg/L yielded sensitivity and specicity values of
91.4% and 38.6%, respectively (Table 3).
Table 2. Procalcitonin thresholds for dierent indices in
patients with proven hospital-acquired infection
reshold (μg/L) Sensitivity Specicity PPV NPV
0.24 0.94 0.40 0.48 0.92
0.26 0.94 0.42 0.49 0.92
0.93 0.73 0.58 0.51 0.78
1.01 0.73 0.60 0.52 0.79
1.27 0.61 0.62 0.49 0.72
40.63 0.03 1.00 1.00 0.63
PPV = positive predictive value; NPV = negative predictive value.
Table 3. C-reactive protein thresholds for dierent indices in
patients with proven hospital-acquired infection
reshold (mg/L) Sensitivity Specicity PPV NPV
47.5 1.00 0.05 0.39 1.00
56.5 1.00 0.07 0.39 1.00
207.0 0.52 0.53 0.40 0.64
437.5 0.03 0.98 0.50 0.63
PPV = positive predictive value; NPV = negative predictive value.
Table 1. Baseline (admission) characteristics and bivariate analysis of all patients
Characteristic All (N=92), n (%)*
Proven HAI
(n=35), n (%)*
Suspected HAI
(n=25), n (%)*
No sepsis (n=32), n
(%)* p-value†p-value‡
Age (years) 53.5 (11.1) 56.2 (10.4) 52.3 (10.8) 51.4 (11.8) 0.09 0.77
Sex (male) 48 (52.1) 19 (54.2) 13 (52) 16 (50) 0.81 1
Diabetes 48 (52.1) 19 (54.2) 14 (56.0) 15 (46.8) 0.63 1
Hypertension 53 (57.6) 24 (68.6) 13 (52.0) 16 (50.0) 0.14 1
Obesity 62 (67.39) 23 (65.7) 15 (60.0) 24 (75.0) 0.44 0.26
HIV 14 (15.21) 4 (11.4) 5 (20.0) 5 (15.6) 0.73 0.74
TB 4 (4.34) 0 1 (4.0) 3 (9.3) 0.10 0.62
PCT (μg/L),
mean (SD)
1.25 (3.63) 1.06 (1.89) 2.52 (6.19) 0.39 (0.60) 0.08 0.78
CRP (mg/L),
mean (SD)
204.4 (114.7) 211.8 (127.1) 222.4 (90.9) 181.5 (113.6) 0.23 0.11
Pro-BNP (μg/L),
mean (SD)
1 335.3 (3 867.2) 874.2 (1 387.2) 2 416.8 (6 350.4) 948.9 (2 642.7) 0.89 0.27
Trop T (ng/dL),
mean (SD)
40.0 (74.4) 51.2 (100.1) 45.7 (63.7) 21.3 (30.4) 0.15 0.09
HbA1c (%),
mean (SD)
8.5 (3.0) 8.4 (3.1) 9.5 (3.2) 7.7 (2.6) 0.38 0.06
Mechanical
ventilation
57 (62.0) 27 (68.5) 21 (84.0) 9 (28.1) <0.001 <0.001
Mortality 63 (68.5) 29 (82.9) 22 (88.0) 12 (37.5) <0.001 <0.001
HAI = hospital-acquired infection; TB = tuberculosis; PCT = procalcitonin; SD = standard deviation; CRP = C-reactive protein; Pro-BNP = pro B-type natriuretic peptide; Trop T = troponin T;
HbA1c = glycated haemoglobin.
*Except where otherwise indicated.
†Proven v. none.
‡Suspected v. none.
AJTCCM VOL. 30 NO. 4 2024 173
ORIGINAL RESEARCH: ARTICLES
A sub-analysis was conducted by varying
the cut-o values. In patients with PCT
levels <0.25 μg/L, the NPV was 92%,
whereas PCT levels >1.00 μg/L had a
PPV >50%. At PCT levels >1.25 μg/L,
sensitivity was lower than specicity, and
levels >40 μg/L resulted in a PPV of 100%.
This sub-analysis aimed to maximise
specicity and minimise sensitivity of the
PCT cut-o values.
In patients with suspected biomarkers,
the maximum sensitivity and specicity
values were 92.0% and 46.0%, respectively.
Using a criterion to maximise sensitivity
and specificity yielded a maximum
sensitivity, but resulted in a lower NPV
and PPV of 59.6% and 5.6%, respectively,
with an AUC of 0.69 (Fig. 3). Similarly,
for CRP, maximum sensitivity and
specicity values were 72.0% and 44.6%,
respectively, with a resulting lower PPV
and NPV of 18.9% and 67.3%, and
an AUC of 0.527 (Fig. 4). In patients
suspected of having HAI, PCT levels
<0.25 μg/L were associated with an NPV
>96%, while levels >1.00 μg/L had an
NPV >80% and a PPV >39%. At levels
>1.47 μg/L, sensitivity was lower than
specicity. For PCT levels >40 μg/L, the
PPV was 71% (Table 4).
Discussion
We found that patients with COVID-19
pneumonia in an ICU, who were admitted
without any evidence of secondary sepsis,
had high initial CRP levels and low PCT
levels. More importantly, our data show
that, during an episode of HAI, PCT
levels >1.00 μg/L resulted in a moderate
PPV of 52%, whereas levels <0.26 μg/L
showed a good NPV of 92% for proven
HAI. Furthermore, with increased PCT
levels, the PPV rose to 100%, making
it a better biomarker than CRP, with
higher predictive value for predicting
HAI in patients with severe COVID-19
pneumonia. CRP values were noted to
be consistently elevated and had a poor
NPV for HAI.
Van Berkel et al.[6] concluded that
PCT was a useful biomarker for
bacterial sepsis in severe COVID-19.
ese investigators observed a 93% PPV
for PCT levels >1 μg/L and an 81% NPV
for levels <0.25 μg/L. ey also noted that
CRP levels were consistently elevated
Sensivity
1.0
0.8
0.6
0.4
0.2
0.0
Specicity
1.0 0.8 0.6 0.4 0.2 0.0
AUC: 0.697
Fig. 1. Receiver operating characteristic curve for procalcitonin in patients with proven hospital-
acquired infection. (AUC = area under the curve.)
AUC: 0.572
1.0
0.8
0.6
0.4
0.2
0.0
Sensivity
1.0 0.8 0.6 0.4 0.2 0.0
Specicity
Fig. 2. Receiver operating characteristic curve for C-reactive protein in patients with proven hospital-
acquired infection. (AUC = area under the curve.)
Table 4. Procalcitonin thresholds for dierent indices in patients with suspected
hospital-acquired infection
reshold (μg/L) Sensitivity Specicity PPV NPV
0.24 0.96 0.37 0.38 0.96
0.255 0.96 0.38 0.38 0.96
0.93 0.72 0.54 0.38 0.83
1.005 0.72 0.56 0.39 0.83
1.475 0.64 0.65 0.42 0.82
33.02 0.04 0.98 0.50 0.72
PPV = positive predictive value; NPV = negative predictive value.
174 AJTCCM VOL. 30 NO. 4 2024
ORIGINAL RESEARCH: ARTICLES
irrespective of bacterial co-infection, and
therefore did not support CRP as a biomarker
to detect bacterial infection in COVID-19.
is conclusion is in keeping with our study
ndings that CRP serves as a poor biomarker
for the detection of bacterial infection. We
observed consistently elevated CRP levels
irrespective of bacterial co-infections.
The findings of our study concur with
the exclusionary utility of PCT observed by
VanBerkel et al.[6] ese investigators also
reported that low PCT levels had an NPV
of 92% for proven bacterial co-infection.
Pink et al.[7] observed a similar result in their
2021 study, demonstrating that a PCT level
<0.55 μg/L had an NPV of 93% to rule out
HAI. In contrast to Van Berkel et al.,[6] our
ndings for the PPV of PCT demonstrated
poor clinical utility, with levels between 0.27
μg/L and 1.27 μg/L only able to predict for
proven HAI in ~50% of cases of COVID-19.
Markedly elevated PCT levels (>40 μg/L)
were highly specic for bacterial co-infection
in severe COVID-19 pneumonia.
Our ndings with regard to the poor PPV
of mildly raised PCT for bacterial co-infection
have been corroborated in several other
studies. Roy et al.[8] argued that elevated PCT
levels are a marker of disease severity rather
than a superadded bacterial infection. is
conclusion was supported by their finding
that in 101 patients with severe COVID-19
pneumonia, levels >0.25 μg/L were associated
with only one blood culture-conrmed case
of bacterial co-infection. Similarly, Heer
et al.[9] demonstrated that increased PCT
concentrations were not positively correlated
with the prevalence of microbiologically
proven sepsis. Elevated PCT levels were,
however, associated with the requirement for
invasive mechanical ventilation, a marker of
disease severity. Garrido et al.[10] found that
serial PCT values in critically ill COVID-19
patients were not beneficial in detecting
HAI and rather served as markers of organ
failure. These authors noted an inverse
correlation between PCT concentrations and
the glomerular ltration rate, postulating that
the elevations were due to decreased renal
clearance. It therefore appears from their
ndings that, like CRP, elevated PCT serves
more as a marker of disease severity than a
biomarker for bacterial co-infection.However,
we note that a potential limitation of the studies
by Heer et al.[9] and Garrido et al.[10] is that
many of their patients received antimicrobial
therapy on their ICU admission, which may
have added to the false-negative rate of the
microbiological data.
We observed from our results that there
is clinical utility for PCT in the context of
severe COVID-19 pneumonia. At levels
<0.26 μg/L, there was a >90% NPV. is ‘rule-
out’ characteristic of PCT may be especially
valuable in scenarios of diagnostic equipoise
with regard to the presence of bacterial co-
AUC: 0.690
1.0
0.8
0.6
0.4
0.2
0.0
Sensivity
1.0 0.8 0.6 0.4 0.2 0.0
Specicity
Fig. 3. Receiver operating characteristic curve for procalcitonin in patients with suspected hospital-
acquired infection. (AUC = area under the curve.)
Fig. 4. Receiver operating characteristic curve for C-reactive protein in patients with suspected
hospital-acquired infection. (AUC = area under the curve.)
AUC: 0.527
1.0
0.8
0.6
0.4
0.2
0.0
Sensivity
1.0 0.8 0.6 0.4 0.2 0.0
Specicity
AJTCCM VOL. 30 NO. 4 2024 175
ORIGINAL RESEARCH: ARTICLES
infection. Clinicians should take care not to unjustiably associate
elevations in PCT levels with the presence of bacterial co-infection,
unless levels are extremely high, in which case the PPV rises signicantly
(100% PPV with levels >40 μg/L). In the absence of a reliable biomarker
to predict for bacterial co-infection in severe COVID-19 pneumonia,
clinical acumen remains the most valuable tool in the timely treatment
of secondary infection and judicious use of antimicrobials.
Our study has certain strengths, including the fact that we were able
to access serial PCT and CRP data, which were captured prospectively.
A potential weakness is that HAI could only be proven in 35 of the 60
patients suspected of having it. Blood cultures, while highly specic,
only have a sensitivity of ~41%, and the false-negative rate may
therefore have inuenced the results of this study.[11] Furthermore,
patients were admitted during various times of the day to the very
high-turnover ICU, which may have inuenced the comparisons.
Moreover, the eect of changes in creatinine clearance on PCT levels
was not corrected for in the analysis, and may have impacted on the
poor PPV of even signicantly raised observed PCT values.
It should be emphasised that the decision to initiate antibiotics
was taken by the treating physician(s), and most oen not based on a
single parameter. We merely report what the value of the PCT could
be as an adjunct in the decision to treat (or not treat) a suspected HAI.
Conclusion
During an episode of HAI, PCT levels >1.00 μg/L had a moderate
PPV of 52%, whereas levels <0.26 μg/L had an NPV of 92% for proven
HAI. Furthermore, with increased PCT values, the PPV rose to 100%,
making it a better biomarker than CRP with a higher PPV for HAI.
Data availability. e datasets generated and analysed during the present
study are available from the corresponding author (CS) on reasonable
request. Any restrictions or additional information regarding data access
can be discussed with the corresponding author.
Declaration. CFNK is a member of the editorial board. The research for
this study was done in partial fulfilment of the requirements for CS’s
MMed (Int) degree at Stellenbosch University.
Acknowledgements. We thank Dr Saadiq Moolla and Dr Caitlin Lyer for
assistance with data collection.
Author contributions. e study was conceptualised by CS and CFNK.
All authors contributed to data collection. e data were analysed by IF
and PSN. e manuscript was written by CS, PSN and CFNK and critically
reviewed by all co-authors.
Funding.None.
Conicts of interest.None.
1. Lansbury L, Lim B, Baskaran V, Lim WS. Co-infections in people with COVID-19:
A systematic review and meta-analysis. J Infect 2020;81(2):266-275. https://doi.
org/10.1016/j.jinf.2020.05.046
2. Moolla MS, Reddy K, Fwemba I, et al. Bacterial infections, antibiotic use and
COVID-19: Lessons from the intensive care unit. S Afr Med J 2021;111(6):575-
581. https://doi.org/10.7196/SAMJ.2021.v111i6.15590
3. Liu F, Li L, Xu M, et al. Prognostic value of interleukin-6, C-reactive protein, and
procalcitonin in patients with COVID-19. J Clin Virol 2020;127:104370. https://
doi.org/10.1016/j.jcv.2020.104370
4. Cleland DA, Eranki AP. Procalcitonin. Updated 23 April 2023. In: StatPearls
(Internet). Treasure Island, Fla.: StatPearls Publishing, January 2024. https://www.
ncbi.nlm.nih.gov/books/NBK539794/ (accessed 28 July 2021).
5. Hoeboer SH, van der Geest PJ, Nieboer D, Groeneveld ABJ. The diagnostic
accuracy of procalcitonin for bacteraemia: A systematic review and meta-
analysis. Clin Microbiol Infect 2015;21(5):474-481. https://doi.org/10.1016/j.
cmi.2014.12.026
6. Van Berkel M, Kox M, Frenzel T, Schouten J; RCI-COVID-19 study group.
Biomarkers for antimicrobial stewardship: A reappraisal in COVID-19 times? Crit
Care 2020;24(1):600. https://doi.org/10.1186/s13054-020-03291-w
7. Pink I, Raupach D, Fuge J, et al. C-reactive protein and procalcitonin for
antimicrobial stewardship in COVID-19. Infection 2021;49(5):935-943. https://
doi.org/10.1007/s15010-021-01615-8
8. Roy A, Powers HR, Craver EC, et al. Antibiotic stewardship: Early discontinuation
of antibiotics based on procalcitonin level in COVID-19 pneumonia. J Clin Pharm
er 2022;47(2):243-247. https://doi.org/10.1111/jcpt.13554
9. Heer RS, Mandal AKJ, Kho J, et al. Elevated procalcitonin concentrations in
severe Covid-19 may not reflect bacterial co-infection. Ann Clin Biochem
2021;58(5):520-527. https://doi.org/10.1177/00045632211022380
10. Garrido P, Cueto P, Rovira C, et al. Clinical value of procalcitonin in critically ill
patients infected by SARS-CoV-2. Am J Emerg Med 2020;46:525-531. https://doi.
org/10.1016/j.ajem.2020.11.011
11. Li Y, Guo J, Yang H, Li H, Shen Y, Zhang D. Comparison of culture-negative and
culture-positive sepsis or septic shock: A systematic review and meta-analysis. Crit
Care 2021;25(1):167. https://doi.org/10.1186/s13054-021-03592-8
Received 17 October 2023. Accepted 17 October 2024. Published 10 December 2024.
