Portal de Periódicos da CAPES

Procalcitonin-Guided Antibiotic Prescription in Patients With COVID-19

2023; Elsevier BV; Volume: 164; Issue: 3 Linguagem: Inglês

10.1016/j.chest.2023.04.032

1931-3543

Lisa Hessels, Esther M. Speksnijder, Nienke Paternotte, Astrid van Huisstede, Willemien Thijs, Margot Scheer, Mariëlle van der Steen-Dieperink, Lieve Knarren, Joop P. van den Bergh, Kristien Winckers, Ronald M.A. Henry, Suat Şimşek, Wim Boersma, Brent Appelman, Michiel Schinkel, David Buis, Kim C.E. Sigalof, Paul Elbers, Daisy Rusch, Auke C. Reidinga, Hazra S. Moeniralam, Caroline E. Wyers, Joop P. van den Bergh, Suat Şimşek, Bastiaan van Dam, Niels C. van den Gritters, Nejma Bokhizzou, Kees Brinkman, Martijn D. de Kruif, Tom Dormans, Renée Douma, Lianne R. de Haan, Tsz Yeung Fung, Martijn Beudel,

Sepsis Diagnosis and Treatment

BackgroundDespite the low rate of bacterial coinfection, antibiotics are very commonly prescribed in hospitalized patients with COVID-19.Research QuestionDoes the use of a procalcitonin (PCT)-guided antibiotic protocol safely reduce the use of antibiotics in patients with a COVID-19 infection?Study Design and MethodsIn this multicenter cohort, three groups of patients with COVID-19 were compared in terms of antibiotic consumption, namely one group treated based on a PCT-algorithm in one hospital (n = 216) and two control groups, consisting of patients from the same hospital (n = 57) and of patients from three similar hospitals (n = 486) without PCT measurements during the same period. The primary end point was antibiotic prescription in the first week of admission.ResultsAntibiotic prescription during the first 7 days was 26.8% in the PCT group, 43.9% in the non-PCT group in the same hospital, and 44.7% in the non-PCT group in other hospitals. Patients in the PCT group had lower odds of receiving antibiotics in the first 7 days of admission (OR, 0.33; 95% CI, 0.16-0.66 compared with the same hospital; OR, 0.42; 95% CI, 0.28-0.62 compared with the other hospitals). The proportion of patients receiving antibiotic prescription during the total admission was 35.2%, 43.9%, and 54.5%, respectively. The PCT group had lower odds of receiving antibiotics during the total admission only when compared with the other hospitals (OR, 0.23; 95% CI, 0.08-0.63). There were no significant differences in other secondary end points, except for readmission in the PCT group vs the other hospitals group.InterpretationPCT-guided antibiotic prescription reduces antibiotic prescription rates in hospitalized patients with COVID-19, without major safety concerns. Despite the low rate of bacterial coinfection, antibiotics are very commonly prescribed in hospitalized patients with COVID-19. Does the use of a procalcitonin (PCT)-guided antibiotic protocol safely reduce the use of antibiotics in patients with a COVID-19 infection? In this multicenter cohort, three groups of patients with COVID-19 were compared in terms of antibiotic consumption, namely one group treated based on a PCT-algorithm in one hospital (n = 216) and two control groups, consisting of patients from the same hospital (n = 57) and of patients from three similar hospitals (n = 486) without PCT measurements during the same period. The primary end point was antibiotic prescription in the first week of admission. Antibiotic prescription during the first 7 days was 26.8% in the PCT group, 43.9% in the non-PCT group in the same hospital, and 44.7% in the non-PCT group in other hospitals. Patients in the PCT group had lower odds of receiving antibiotics in the first 7 days of admission (OR, 0.33; 95% CI, 0.16-0.66 compared with the same hospital; OR, 0.42; 95% CI, 0.28-0.62 compared with the other hospitals). The proportion of patients receiving antibiotic prescription during the total admission was 35.2%, 43.9%, and 54.5%, respectively. The PCT group had lower odds of receiving antibiotics during the total admission only when compared with the other hospitals (OR, 0.23; 95% CI, 0.08-0.63). There were no significant differences in other secondary end points, except for readmission in the PCT group vs the other hospitals group. PCT-guided antibiotic prescription reduces antibiotic prescription rates in hospitalized patients with COVID-19, without major safety concerns. Take-home PointsStudy Question: Does the use of a procalcitonin (PCT)-guided antibiotic protocol safely reduce the use of antibiotics in patients with COVID-19 infection?Results: Antibiotic use was lower when a PCT-guided protocol was used for antibiotic prescription compared with standard of care. However, there was a higher rate of readmission in the PCT-guided group, mainly because of noninfectious causes.Interpretation: PCT-guided antibiotic prescription reduces antibiotic prescription rates in hospitalized patients with COVID-19, without major safety concerns. Study Question: Does the use of a procalcitonin (PCT)-guided antibiotic protocol safely reduce the use of antibiotics in patients with COVID-19 infection? Results: Antibiotic use was lower when a PCT-guided protocol was used for antibiotic prescription compared with standard of care. However, there was a higher rate of readmission in the PCT-guided group, mainly because of noninfectious causes. Interpretation: PCT-guided antibiotic prescription reduces antibiotic prescription rates in hospitalized patients with COVID-19, without major safety concerns. In patients with COVID-19, it is challenging to distinguish the proportion of patients with a bacterial coinfection because of overlapping clinical symptoms and similar features, such as infiltrates, on radiologic imaging.1Aggarwal S. Garcia-Telles N. Aggarwal G. et al.Clinical features, laboratory characteristics, and outcomes of patients hospitalized with coronavirus disease 2019 (COVID-19): early report from the United States.Diagnosis (Berl). 2020; 7: 91-96Crossref PubMed Scopus (263) Google Scholar,2Yang W. Cao Q. Qin L. et al.Clinical characteristics and imaging manifestations of the 2019 novel coronavirus disease (COVID-19): a multi-center study in Wenzhou city, Zhejiang, China.J Infect. 2020; 80: 388-393Abstract Full Text Full Text PDF PubMed Scopus (643) Google Scholar It is also difficult to identify patients with a bacterial coinfection based on C-reactive protein (CRP) levels, because in both bacterial and viral infections CRP can be elevated.3Lippi G. Plebani M. Laboratory abnormalities in patients with COVID-2019 infection.Clin Chem Lab Med. 2020; 58: 1131-1134Crossref PubMed Scopus (612) Google Scholar,4Henry B.M. De Oliveira M.S. Benoit S. et al.Hematologic, biochemical and immune biomarker abnormalities associated with severe illness and mortality in coronavirus disease 2019 (COVID-19): a meta-analysis.Clin Chem Lab Med. 2020; 58: 1021-1028Crossref PubMed Scopus (1200) Google Scholar Bacterial coinfection is confirmed in only 8% of hospitalized patients with COVID-19, and in 14% of patients admitted to the ICU.5Langford B.J. So M. Raybardhan S. et al.Bacterial co-infection and secondary infection in patients with COVID-19: a living rapid review and meta-analysis.Clin Microbiol Infect. 2020; 26: 1622-1629Abstract Full Text Full Text PDF PubMed Scopus (828) Google Scholar,6Rawson T.M. Moore L.S.P. Zhu N. et al.Bacterial and fungal coinfection in individuals with coronavirus: a rapid review to support COVID-19 antimicrobial prescribing.Clin Infect Dis. 2020; 71: 2459-2468PubMed Google Scholar Because of the challenges of identifying patients with a bacterial coinfection, the rate of antibiotic prescription in patients with COVID-19 is high and is reported to be between 60% and 80% in hospitalized patients.7Docherty A.B. Harrison E.M. Green C.A. et al.Features of 20 133 UK patients in hospital with COVID-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study.BMJ. 2020; 369: m1985Crossref PubMed Scopus (1946) Google Scholar, 8Langford B.J. So M. Raybardhan S. et al.Antibiotic prescribing in patients with COVID-19: rapid review and meta-analysis.Clin Microbiol Infect. 2021; 27: 520-531Abstract Full Text Full Text PDF PubMed Scopus (398) Google Scholar, 9Beović B. Doušak M. Ferreira-Coimbra J. et al.Antibiotic use in patients with COVID-19: a 'snapshot' Infectious Diseases International Research Initiative (ID-IRI) survey.J Antimicrob Chemother. 2020; 75: 3386-3390Crossref PubMed Scopus (137) Google Scholar Overuse of antibiotic therapy during the COVID-19 pandemic may therefore contribute to antimicrobial resistance, and strategies to promote antibiotic stewardship are needed. Antibiotic therapy may also cause side effects in the individual patient, which might cause unnecessary harm. One strategy to reduce the high antibiotic prescription rates in patients with COVID-19 is by use of the biomarker procalcitonin (PCT). PCT is a 116 amino acid precursor protein of the hormone calcitonin.10Maruna P. Nedelníková K. Gürlich R. Physiology and genetics of procalcitonin.Physiol Res. 2000; 49: S57-S61PubMed Google Scholar PCT levels are normally < 0.1 ng/mL in healthy individuals.11Morgenthaler N.G. Struck J. Fischer-Schulz C. et al.Detection of procalcitonin (PCT) in healthy controls and patients with local infection by a sensitive ILMA.Clin Lab. 2002; 48: 263-270PubMed Google Scholar A bacterial infection causes broad PCT synthesis in immune cells and parenchymal cells, mediated by inflammatory cytokines (eg, tumor necrosis factor-alpha, IL-6, IL-1), resulting in higher PCT levels.12Assicot M. Gendrel D. Carsin H. et al.High serum procalcitonin concentrations in patients with sepsis and infection.Lancet. 1993; 341: 515-518Abstract PubMed Scopus (1735) Google Scholar,13Atallah N.J. Warren H.M. Roberts M.B. et al.Baseline procalcitonin as a predictor of bacterial infection and clinical outcomes in COVID-19: a case-control study.PLoS One. 2022; 17e0262342Crossref PubMed Scopus (15) Google Scholar Viral infections do not lead to higher PCT synthesis.14Muller B. Prat C. Markers of acute inflammation in assessing and managing lower respiratory tract infections: focus on procalcitonin.Clin Microbiol Infect. 2006; 12: 8-16Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar PCT is a more sensitive and specific marker for bacterial infection than CRP because CRP levels are more likely to rise in response to critical illness, trauma, surgery, and chronic disease.15Simon L. Gauvin F. Amre D.K. et al.Serum procalcitonin and C-reactive protein levels as markers of bacterial infection: a systematic review and meta-analysis.Clin Infect Dis. 2004; 39: 206-217Crossref PubMed Scopus (1341) Google Scholar, 16Stolz D. Christ-Crain M. Gencay M.M. et al.Diagnostic value of signs, symptoms and laboratory values in lower respiratory tract infection.Swiss MedWkly. 2006; 136: 434-440PubMed Google Scholar, 17Muller B. White J.C. Nylen E.S. et al.Ubiquitous expression of the calcitonin-i gene in multiple tissues in response to sepsis.J Clin Endocrinol Metab. 2001; 86: 396-404PubMed Google Scholar, 18Uzzan B. Cohen R. Nicolas P. et al.Procalcitonin as a diagnostic test for sepsis in critically ill adults and after surgery or trauma: a systematic review and meta-analysis.Crit Care Med. 2006; 34: 1996-2003Crossref PubMed Scopus (617) Google Scholar, 19Galetto-Lacour A. Zamora S.A. Gervaix A. Bedside procalcitonin and C-reactive protein tests in children with fever without localizing signs of infection seen in a referral center.Pediatrics. 2003; 112: 1054-1060Crossref PubMed Scopus (192) Google Scholar A large Cochrane systematic review in 2017 showed that the use of PCT to guide initiation and duration of antibiotic therapy in patients with acute respiratory infection has reduced the amount of antibiotics prescribed, without compromising safety.20Schuetz P. Wirz Y. Sager R. et al.Procalcitonin to initiate or discontinue antibiotics in acute respiratory tract infections.Cochrane Database Syst Rev. 2017; 10: CD007498PubMed Google Scholar It seems plausible that the same PCT protocol can be used to guide antibiotic prescription in patients with COVID-19. A few retrospective studies indeed showed reduced antibiotic use in patients with COVID-19 who had PCT measured.21Calderon M. Li A. Bazo-Alvarez J.C. et al.Evaluation of procalcitonin-guided antimicrobial stewardship in patients admitted to hospital with COVID-19 pneumonia.JAC Antimicrob Resist. 2021; 3: dlab133Crossref PubMed Scopus (16) Google Scholar, 22Peters C. Williams K. Un E.A. et al.Use of procalcitonin for antibiotic stewardship in patients with COVID-19: a quality improvement project in a district general hospital.Clin Med (Lond). 2021; 21: 71-76Crossref PubMed Google Scholar, 23Williams E.J. Mair L. de Silva T.I. et al.Evaluation of procalcitonin as a contribution to antimicrobial stewardship in SARS-CoV-2 infection: a retrospective cohort study.J Hosp Infect. 2021; 110: 103-107Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar, 24Hughes S. Mughal N. Moore L.S.P. Procalcitonin to guide antibacterial prescribing in patients hospitalised with COVID-19.Antibiotics (Basel). 2021; 10: 1119Crossref PubMed Scopus (11) Google Scholar However, there is currently no study prospectively evaluating the implementation of a PCT-guided protocol. The aim of this study was to determine whether the use of a PCT-guided antibiotic protocol safely reduces the use of antibiotics in patients with a COVID-19 infection. The study was performed in Northwest Clinics, Alkmaar (NWZ), a large teaching hospital in The Netherlands, in collaboration with the COVIDPredict study group. COVIDPredict is a multicenter initiative of 11 hospitals in The Netherlands to collect data of hospitalized patients with COVID-19. This study is a multicenter cohort study that investigated an algorithm prescribing antibiotic therapy based on PCT levels in patients with COVID-19 disease. Three groups of patients were compared, namely a prospective group in NWZ of patients who received antibiotic therapy based on a PCT-guided protocol, a retrospective control group of patients in NWZ treated without PCT guidance, and a retrospective control group of patients from three hospitals in the COVIDPredict collaboration, also treated without PCT guidance. The study was approved by the institutional research committee of Northwest clinics. The study protocol of the COVIDPredict group was reviewed by the medical ethics committees of the Amsterdam University Medical Centre (20.131) and Maastricht University Medical Center (2020-1323). Patients participating in the prospective PCT-guided group provided informed consent for data collection. For the retrospective control groups, the need for informed consent was waived. Patients aged ≥ 18 years were eligible for inclusion if they presented at the ED with symptoms of a viral respiratory tract infection and the diagnosis COVID-19 was confirmed by means of a positive SARS-CoV-2 reverse transcriptase polymerase chain reaction on nose and/or throat swabs and/or deep respiratory samples. Patients were only included if they were admitted to the hospital. All patients were admitted between October 2020 and July 2021. For the PCT-guided group, patients were prospectively enrolled in NWZ between December 2020 and June 2021. PCT was measured within 24 h of admission. Admitted patients were treated in accordance with the current national guideline. The decision to treat patients with antibiotics was guided by PCT levels. At PCT levels < 0.25 μg/L, antibiotics were discouraged; between PCT levels of 0.25 μg/L and 0.5 μg/L, antibiotics could be considered; and at PCT levels > 0.5 μg/L, antibiotics were recommended (e-Table 1). These cutoff values were based on earlier studies.25Schuetz P. Christ-Crain M. Thomann R. et al.Effect of procalcitonin-based guidelines vs standard guidelines on antibiotic use in lower respiratory tract infections: the ProHosp Randomized controlled trial.JAMA. 2009; 302: 1059-1066Crossref PubMed Scopus (787) Google Scholar,26Bouadma L. Luyt C.E. Tubach F. et al.Use of procalcitonin to reduce patients' exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial.Lancet. 2010; 375: 463-474Abstract Full Text Full Text PDF PubMed Scopus (891) Google Scholar The protocol was communicated multiple times to the involved physicians and was underscored by means of posters at the ED. The treating physicians were able to deviate from the PCT protocol when necessary. In those cases, they were asked to supply a motivation for their decision. Patients of the first control group were admitted to NWZ. PCT levels were not measured because measurements were not standardized for every ED admission. Admitted patients were treated in accordance with the national guideline, to which in June 2020 a supplement was added regarding antibiotic use in patients with COVID-19.27Sieswerda E. de Boer M.G.J. Bonten M.M.J. et al.Recommendations for antibacterial therapy in adults with COVID-19 - an evidence based guideline.Clin Microbiol Infect. 2021; 27: 61-66Abstract Full Text Full Text PDF PubMed Scopus (121) Google Scholar,28Medicamenteuze behandeling voor patiënten met COVID-19Federatie Medisch Specialisten website.https://richtlijnendatabase.nl/richtlijn/covid-19/behandeling/medicamenteuze_behandeling_voor_patienten_met_covid-19.htmlDate accessed: March 12, 2023Google Scholar The decision to treat patients with antibiotics was left to the treating physician, based on available clinical parameters, including radiology and laboratory parameters. The second control group consisted of patients from three hospitals of the COVIDPredict Study Group. We retrospectively collected data from patients from one academic hospital, namely Maastricht University Medical Center, Maastricht, and two teaching hospitals, namely VieCuri Medical Centre, Venlo, and Martini Hospital, Groningen. In these hospitals, PCT levels were not available on admission for any of the patients with COVID-19. Patients were treated in accordance with the national guideline. The decision to treat patients with antibiotics was left to the treating physician, based on available clinical parameters. PCT was measured with Atellica IM BRAHMS Procalcitonin reagent on an Atellica IM analyzer (Siemens Healthcare Diagnostics Inc). PCT reference values were established as < 0.05 μg/L. Within-laboratory precision is determined as < 7% coefficient of variation for samples from 0.05 to 2.0 μg/L and < 5% coefficient of variation for samples > 2.0 μg/L. The primary outcome was the proportion of antibiotic prescription during the first 7 days of admission. Secondary outcomes were the proportion of antibiotic prescription during the total admission, length of hospital stay, admission to the ICU, mechanical ventilation, noninvasive ventilation, 30-day all-cause mortality, 90-day all-cause mortality, and readmission within 30 days. Microbiological results, detected by blood cultures, sputum culture, urine culture, and/or other sites, were also noted. Adherence to protocol and protocol failure was registered. Protocol failure was defined as starting antibiotics for lower respiratory tract infection within 7 days after initially withholding antibiotics based on PCT results. The following data were collected for all included patients: demographics, comorbidities, duration of symptoms and symptoms prior to admission, vital signs, radiologic findings on admission, laboratory results on admission, results of bacterial cultures during admission, and antibiotic therapy prescribed during admission. Outcome data were collected as previously described. The total follow-up duration was 90 days. Data collection was conducted in compliance with Good Clinical Practice guidelines. Data were pseudonymized. Source documents for data collection were the medical records of the patients. Data were entered into an electronic case report form (Castor EDC 2021.5.0; Castor) by each hospital independently. For statistical analysis, IBM SPSS statistics 28 software (IBM) was used. Missing data were accepted after maximal efforts to retrieve data. Standard descriptive statistics were used to describe the continuous or categorical values. Variables were compared between study groups using the Kruskal-Wallis test for continuous variables and the χ2 or Fisher exact test for categorical variables, as applicable. Antibiotic use was compared between groups using a logistic regression model and corrected for confounding variables. Ninety-day survival was estimated using the Kaplan-Meier method and adjusted for confounding variables using a Cox proportional hazards model. Cox proportional hazard assumptions were assessed for each variable by plotting scaled Schoenfeld residuals. Other secondary outcomes were compared using a logistic regression model. The variables for the adjusted models were chosen based on a priori knowledge for being relevant patient characteristics and known risk factors for a worse outcome in patients with COVID-19.29Parohan M. Yaghoubi S. Seraji A. et al.Risk factors for mortality in patients with Coronavirus disease 2019 (COVID-19) infection: a systematic review and meta-analysis of observational studies.Aging Male. 2020; 23: 1416-1424Crossref PubMed Scopus (224) Google Scholar,30Grasselli G. Greco M. Zanella A. et al.Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy, Italy.JAMA Intern Med. 2020; 180: 1345-1355Crossref PubMed Scopus (972) Google Scholar The assumptions for both logistic regression models were assessed by using collinearity statistics, the Box-Tidwell transformation, and Cook’s distance. P < .05 was considered significant. In NWZ, 1,335 patients were admitted with COVID-19 between October 2020 and July 2021. Of these patients, 216 were included in the PCT-guided group. Fifty-seven patients were retrospectively included in the non-PCT-guided group. Based on the inclusion criteria, 486 of the 6,165 patients in the COVIDPredict database were included in the control group of other hospitals (e-Fig 1). A post hoc analysis of the excluded patients is provided in e-Table 2. The main characteristics of the study population are summarized in Table 1. The median age of all patients was 68 years (interquartile range [IQR], 57-78 years), and 475 participants (62.6%) were male. Significant differences in the baseline characteristics of the study groups were seen in median age, smoking status, COPD as comorbidity, diabetes as comorbidity, CRP levels at admission, WBC count at admission, and CURB-65 score at admission. In the PCT-guided group, the median PCT was 0.13 μg/L, and 103 participants (75.5%) had a PCT < 0.25 μg/L, 30 participants (13.9%) had a PCT between 0.25 and 0.50 μg/L, and 23 participants (10.6%) had a PCT > 0.50 μg/L.Table 1Patient CharacteristicsCharacteristicPCT-Guided Hospital 1 (n = 216)Non-PCT-Guided Hospital 1 (n = 57)Non-PCT-Guided Other Hospitals (n = 486)P ValueAge at admission, y64 (53-73)73 (59-80)70 (59-79)< .001aKruskal-Wallis test.,bStatistically significant difference (P < .05).Sex.341cχ2 test. Male138 (63.9)32 (56.1)305 (62.8) Female78 (36.1)25 (43.9)181 (37.2)Smoking.018bStatistically significant difference (P < .05).,cχ2 test. Tobacco use or former tobacco use99 (45.8)18 (31.6)191 (39.3) No tobacco use103 (47.7)35 (61.4)206 (42.4)Comorbidities COPD23 (10.6)5 (8.8)105 (21.7)< .001bStatistically significant difference (P < .05).,dFisher-Freeman-Halton exact test. Asthma26 (12.0)3 (5.3)51 (10.6).348dFisher-Freeman-Halton exact test. Diabetes35 (16.2)15 (26.3)149 (31.2)< .001bStatistically significant difference (P < .05).,dFisher-Freeman-Halton exact test. Congestive heart failure16 (7.4)5 (8.8)27 (5.6)< .001bStatistically significant difference (P < .05).,dFisher-Freeman-Halton exact test. Malignancy18 (8.3)10 (17.5)52 (10.9).346dFisher-Freeman-Halton exact test.BMI, kg/m2.580dFisher-Freeman-Halton exact test. < 202 (1.3)0 (0.0)11 (3.8) 20-2448 (31.2)14 (34.1)92 (31.8) 25-2980 (51.9)20 (48.8)147 (50.9) ≥ 3024 (15.6)7 (17.1)39 (13.5)Time from onset of symptoms to admission, d9 (7-11)10 (5-14)8 (5-12).129aKruskal-Wallis test.CRP levels on admission, mg/L91 (52-140)57 (36-123)76 (33-136).024aKruskal-Wallis test.,bStatistically significant difference (P < .05).WBC levels on admission, ×109/L5.7 (4.6-7.9)6.1 (4.7-8.3)7.3 (5.1-9.8)< .001aKruskal-Wallis test.,bStatistically significant difference (P < .05).CURB-65 score at admission< .001bStatistically significant difference (P < .05).,dFisher-Freeman-Halton exact test. 081 (37.5)9 (15.8)90 (19.2) 159 (27.3)15 (26.3)155 (33.0) 254 (25.0)16 (28.1)157 (33.50) 317 (7.9)13 (22.8)56 (11.9) 43 (1.4)4 (7.0)11 (2.3) 52 (0.9)0 (0.0)0 (0.0)Treatments received during admission Corticosteroids199 (92.1)46 (80.7)528 (70.1)< .001bStatistically significant difference (P < .05).,dFisher-Freeman-Halton exact test.Values are median (interquartile range), No. (%), or as otherwise indicated. CRP = C-reactive protein; PCT = procalcitonin.a Kruskal-Wallis test.b Statistically significant difference (P < .05).c χ2 test.d Fisher-Freeman-Halton exact test. Open table in a new tab Values are median (interquartile range), No. (%), or as otherwise indicated. CRP = C-reactive protein; PCT = procalcitonin. Data on number and proportion of patients receiving antibiotic prescriptions are presented in Figure 1 and Table 2.Table 2Antibiotic Prescription During the First 7 Days of Admission and During Total AdmissionCompared GroupsNo. of PatientsObserved, No. (%)Unadjusted ORaχ2 test. (95% CI)P ValueAdjusted ORbLogistic regression model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, and malignancy. (95% CI)P ValueAntibiotics prescribed during first 7 d of admission PCT-guided vs non-PCT-guided hospital 1.002cPCT-guided21658 (26.8)0.44 (0.24-0.81).008cStatistically significant (P < .05).0.33 (0.16-0.66)Non-PCT-guided hospital 15725 (43.9)......... PCT-guided vs non-PCT-guided other hospitals< .001cPCT-guided21658 (26.8)0.37 (0.26-0.53)< .001cStatistically significant (P < .05).0.42 (0.28-0.62)Non-PCT-guided other hospitals486217 (44.7).........Antibiotics prescribed during total admission PCT-guided vs non-PCT-guided hospital 1.132PCT-guided21676 (35.2)0.70 (0.38-1.23).2290.60 (0.31-1.17)Non-PCT-guided hospital 15725 (43.9)......... PCT-guided vs non-PCT-guided other hospitals< .001cPCT-guided21676 (35.2)0.45 (0.32-0.63)< .001cStatistically significant (P < .05).0.53 (0.36-0.77)Non-PCT-guided other hospitals486265 (54.5).........PCT = procalcitonin.a χ2 test.b Logistic regression model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, and malignancy.c Statistically significant (P < .05). Open table in a new tab PCT = procalcitonin. During the first 7 days of admission, antibiotics were prescribed in 58 patients (26.8%) in the PCT-guided group, 25 patients (43.9%) in the non-PCT-guided group in hospital 1, and 217 patients (44.7%) in the non-PCT-guided group in the other hospitals. Antibiotics most prescribed were ceftriaxone, cefuroxime, and amoxicillin (e-Table 3). Patients in the PCT-guided group had lower odds of receiving antibiotics in the first 7 days of admission (adjusted OR, 0.33; 95% CI, 0.16-0.66) when compared with the non-PCT-guided group in the same hospital. When compared with the non-PCT-guided group in the other hospitals, the odds of antibiotic prescription during the first 7 days were also lower in the PCT-guided group (adjusted OR, 0.42; 95% CI, 0.28-0.62). The median CRP level of patients who did receive antibiotics during the first 7 days of admission was 75 mg/L (IQR, 37-126 mg/L) compared with 91 mg/L (IQR, 43-150 mg/L) in patients who did not receive antibiotics during the first 7 days. This was not significantly different (P = .60). During the total duration of admission, antibiotics were prescribed in 76 patients (35.2%) in the PCT-guided group, 25 patients (43.9%) in the non-PCT-guided group in hospital 1, and 265 patients (54.5%) in the non-PCT-guided group in the other hospitals. Patients in the PCT-guided group had lower odds of receiving antibiotics during the total duration of the admission (adjusted OR, 0.53; 95% CI, 0.36-0.77) when compared with the non-PCT-guided group in the other hospitals. When comparing the PCT-guided group with the non-PCT-guided group in the same hospital, there was no significant difference in the odds of receiving antibiotics during the total admission (adjusted OR, 0.60; 95% CI, 0.31-1.17) (Table 2). Adherence to the PCT-guided protocol was 94% in patients with PCT < 0.25 μg/L and 100% in patients with PCT > 0.50 μg/L, suggesting good protocol compliance. The main reason for starting antibiotics in the PCT < 0.25 μg/L and the PCT 0.25 to 0.50 μg/L group was a high suspicion of bacterial infection based on clinical symptoms. Other reasons are provided in e-Table 4. Protocol failure, defined as starting antibiotics within 7 days of admission, after initially withholding based on PCT level, was 8.3%. The median length of stay was 6.0 days in the PCT-guided group (IQR, 3.0-12.0 days), and did not differ significantly from the non-PCT-guided group of hospital 1 (median, 7.0 days; IQR, 4.0-10.5 days) nor from the non-PCT-guided group of other hospitals (median, 6.0 days; IQR, 3.0-11.0 days). Other secondary outcome measurements are shown in Tables 3 and 4.Table 3Summary of Secondary Outcome Measurements and Hazard RatioSecondary Outcome MeasurementsNo. of EventsUnadjusted Hazard RatioaUnivariate cox regression analysis. (95% CI)P ValueAdjusted Hazard RatiobCox proportional hazards model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, malignancy, and treatment with corticosteroids. None of the variables, nor the model as a whole, violated the Cox proportional hazard assumptions. (95% CI)P Value30-d overall survival PCT-guided hospital 1 (n = 216)211.0 (reference)...1.0 (reference)... Non-PCT-guided hospital 1 (n = 57)40.72 (0.25-2.09).5400.50 (0.16-1.53).223 Non-PCT-guided other hospitals (n = 486)651.4 (0.86-2.31).1691.08 (0.59-1.98).80490-d overall survival PCT-guided (n = 216)231.0 (reference)...1.0 (reference)... Non-PCT-guided hospital 1 (n = 57)71.14 (0.49-2.66).7570.85 (0.35-2.10).732 Non-PCT-guided other hospitals (n = 486)691.37 (0.85-2.20).1911.04 (0.59-1.83).906PCT = procalcitonin.a Univariate cox regression analysis.b Cox proportional hazards model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, malignancy, and treatment with corticosteroids. None of the variables, nor the model as a whole, violated the Cox proportional hazard assumptions. Open table in a new tab Table 4Summary of Secondary Outcome Measurements and ORSecondary Outcome MeasurementsObserved, No. (%)Unadjusted ORaUnivariate logistic regression model. (95% CI)P ValueAdjusted ORbLogistic regression model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, malignancy, and treatment with corticosteroids. (95% CI)P ValueReadmission within 30 d PCT-guided (n = 216)15 (6.9)1.00 (reference)...1.00 (reference)... Non-PCT-guided hospital 1 (n = 57)3 (5.3)0.74 (0.21-2.67).6500.52 (0.11-2.45).410 Non-PCT-guided other hospitals (n = 486)7 (1.4)0.20 (0.08-0.49)< .001cStatistically significant (P < .05).0.23 (0.08-0.63).005cStatistically significant (P < .05).ICU admission PCT-guided (n = 216)27 (12.5)1.00 (reference)...1.00 (reference)... Non-PCT-guided hospital 1 (n = 57)6 (10.5)0.82 (0.32-2.10).6851.04 (0.39-2.77).932 Non-PCT-guided other hospitals (n = 486)40 (8.2)0.63 (0.37-1.05).0780.67 (0.36-1.24).204PCT = procalcitonin.a Univariate logistic regression model.b Logistic regression model with the following covariates: age, sex, BMI, asthma, COPD, diabetes, congestive heart failure, malignancy, and treatment with corticosteroids.c Statistically significant (P < .05). Open table in a new tab PCT = procalcitonin. PCT = procalcitonin. Analysis of 30-day overall survival and 90-day overall survival showed no significant differences between the three groups. After adjusting for potential confounders, these findings remained consistent. Survival curves of 90-day survival are shown in e-Figures 2 and 3. Patients in the non-PCT-guided group of other hospitals had a lower probability of readmission (adjusted OR, 0.23; 95% CI, 0.08-0.63). There were no statistically significant differences in the odds of ICU admission or the odds of mechanical ventilation between the three groups. In patients with COVID-19, the overall prevalence of bacterial coinfection at admission confirmed with cultures was 0.9%. Microbiology results for each group are summarized in e-Tables 5 and 6. There was no significant difference in the rate of confirmed infections between the groups. Higher PCT levels were not significantly associated with a higher percentage of confirmed coinfections. One patient with a PCT level < 0.25 μg/L developed bacteremia shortly after admission. The aim of this study was to investigate the use of a PCT-guided algorithm in patients with COVID-19 infection to reduce the antibiotic prescription rate without safety issues. Our multicenter cohort study demonstrates that a PCT-guided antibiotic protocol reduces the proportion of patients treated with antibiotic therapy by 17% (same hospital) and 18% (other hospitals) for the first 7 days of admission and by 9% and 19%, respectively, for the total duration of the admission. However, when comparing the PCT-guided group with the non-PCT-guided group in hospital 1, antibiotic prescription did not differ significantly for the total duration of admission. This may have partly been caused by the low numbers in the non-PCT-guided group in hospital 1. There were no significant differences between the three groups in 30-day and 90-day overall mortality. Also, no significant differences in the proportion of patients admitted to the ICU or who received mechanical ventilation were found. The length of stay was similar in all three groups. We expect that the duration of hospital stay in patients with COVID-19 is mainly determined by the need for oxygen therapy, independent of bacterial coinfection, or antibiotic use. The rate of readmission within 30 days, although the numbers were low, was higher in the PCT-guided group compared with the non-PCT-guided group in other hospitals. Only two of the 15 readmitted patients in the PCT-guided group were admitted with hospital-acquired pneumonia as readmission reason, of which one was with bacteriemia; therefore, most readmissions were related to noninfectious causes. Median time between admission and readmission was 6.0 days in both groups of hospital 1 and 14 days in the other hospitals, indicating that in hospital 1 there might be a tendency to dismiss patients more quickly. Overall, we have no clear evidence of higher complication rates in the PCT-guided group. Our findings are in line with previous studies investigating the use of PCT in guiding antibiotic therapy in patients with COVID-19. Williams et al23Williams E.J. Mair L. de Silva T.I. et al.Evaluation of procalcitonin as a contribution to antimicrobial stewardship in SARS-CoV-2 infection: a retrospective cohort study.J Hosp Infect. 2021; 110: 103-107Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar retrospectively evaluated the effectiveness of a hospital guideline that recommends withholding antibiotics in patients with PCT ≤ 0.25 μg/L. They found a threefold decrease in antibiotic use in patients with PCT ≤ 0.25 μg/L, with no increase in mortality. Calderon et al21Calderon M. Li A. Bazo-Alvarez J.C. et al.Evaluation of procalcitonin-guided antimicrobial stewardship in patients admitted to hospital with COVID-19 pneumonia.JAC Antimicrob Resist. 2021; 3: dlab133Crossref PubMed Scopus (16) Google Scholar conducted a retrospective single-site cohort study to compare antibiotic prescription rates in patients with and without PCT measured. Antibiotic use was lower in patients who had PCT measured (mean antimicrobial duration, 4.4 vs 5.4 days, respectively), with similar composite outcomes, comprising death, admission to ICU, and readmission. Hughes et al24Hughes S. Mughal N. Moore L.S.P. Procalcitonin to guide antibacterial prescribing in patients hospitalised with COVID-19.Antibiotics (Basel). 2021; 10: 1119Crossref PubMed Scopus (11) Google Scholar retrospectively investigated the correlation of PCT on admission and antibacterial agents, and found that a negative PCT on admission correlated with shorter antibacterial therapy. The strength of our study is that we are the first, to our knowledge, to implement a prospective PCT-guided protocol in the treatment of patients with COVID-19. The protocol adherence in our study was very high (94%-100%) in the PCT-guided group. This demonstrates that using a PCT-guided antibiotic protocol is practically feasible. Previous studies in patients with COVID-19 used a PCT threshold of ≥ 0.25 μg/L.21Calderon M. Li A. Bazo-Alvarez J.C. et al.Evaluation of procalcitonin-guided antimicrobial stewardship in patients admitted to hospital with COVID-19 pneumonia.JAC Antimicrob Resist. 2021; 3: dlab133Crossref PubMed Scopus (16) Google Scholar,23Williams E.J. Mair L. de Silva T.I. et al.Evaluation of procalcitonin as a contribution to antimicrobial stewardship in SARS-CoV-2 infection: a retrospective cohort study.J Hosp Infect. 2021; 110: 103-107Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar We adopted the PCT protocol from the ProHOSP and PRORATA trials, therefore using a higher threshold.25Schuetz P. Christ-Crain M. Thomann R. et al.Effect of procalcitonin-based guidelines vs standard guidelines on antibiotic use in lower respiratory tract infections: the ProHosp Randomized controlled trial.JAMA. 2009; 302: 1059-1066Crossref PubMed Scopus (787) Google Scholar,26Bouadma L. Luyt C.E. Tubach F. et al.Use of procalcitonin to reduce patients' exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial.Lancet. 2010; 375: 463-474Abstract Full Text Full Text PDF PubMed Scopus (891) Google Scholar This allows further reduction of antibiotic use. The rate of bacterial coinfections reported in the literature is < 10%. The rate of bacterial coinfection in the group of patients with a PCT level > 0.50 μg/L was 10.6% in our study, but the antibiotic prescription rate was still 26.8%. Because we also now know that in hospitalized patients with COVID-19 there is often a high inflammatory reaction in the body, possibly a higher cutoff value of ≥ 0.50 μg/L should be used. The Dutch Working Party on Antibiotic Policy released a recommendation on antibacterial therapy in patients with COVID-19, with the consensus to be restrictive in antibacterial use.27Sieswerda E. de Boer M.G.J. Bonten M.M.J. et al.Recommendations for antibacterial therapy in adults with COVID-19 - an evidence based guideline.Clin Microbiol Infect. 2021; 27: 61-66Abstract Full Text Full Text PDF PubMed Scopus (121) Google Scholar This guideline probably led to a reduction in antibiotic use in Dutch hospitals shortly before the start of our study. The effect of a PCT-guided antibiotic protocol might be more pronounced in countries less restrictive with antibiotic use. There are also some limitations to our study. In patients with COVID-19, confirming a true bacterial coinfection remains very difficult. The presence of a bacterial infection confirmed by cultures at admission was very low in our study (0.9%). We did not have results of all microbiological cultures drawn in the other hospitals group during the total admission duration for all patients. In the literature, it is described that in 8% of patients with COVID-19, a bacterial coinfection is confirmed by cultures during the total admission.5Langford B.J. So M. Raybardhan S. et al.Bacterial co-infection and secondary infection in patients with COVID-19: a living rapid review and meta-analysis.Clin Microbiol Infect. 2020; 26: 1622-1629Abstract Full Text Full Text PDF PubMed Scopus (828) Google Scholar,6Rawson T.M. Moore L.S.P. Zhu N. et al.Bacterial and fungal coinfection in individuals with coronavirus: a rapid review to support COVID-19 antimicrobial prescribing.Clin Infect Dis. 2020; 71: 2459-2468PubMed Google Scholar This indicates that even with a PCT-guided antibiotic protocol, the antibiotic prescription rate might still be too high. Furthermore, the partly retrospective design may have introduced biases. The exact duration and dose of antibiotic therapy was not always available, especially in the patients from the COVIDPredict database. This may have led to an overcalculation or undercalculation of antibiotic therapy in the other hospital group, causing information bias. There might also be selection bias in the PCT-guided group because informed consent was not always possible in critically ill patients. This may have led to an underrepresentation of critically ill patients in the PCT-guided group, as shown by the CURB-65 score at admission; however, the percentage of ICU admissions is similar in all groups. A proportion of eligible patients admitted with COVID-19 between October 2020 and July 2021 was not considered for enrollment in the PCT-guided group. This was mainly because of many staff changes and multiple departments involved during the COVID-19 crisis. We think this to be at random, therefore not leading to selection bias; however, this cannot be known with certainty. To further investigate this bias, we did a post hoc analysis of the patients not considered for enrollment, but with a PCT available. The antibiotic prescription rate of these patients was comparable with the patients without a PCT available, which highlights the importance of a well-defined protocol for the use of PCT. Finally, all patients were treated in accordance with the national guideline. However, during the pandemic, the standard of care continued to evolve. The biggest change in care was the introduction of treatment with corticosteroids, which influenced patient outcomes. In our study, the study and control groups were admitted during the same time period; therefore, the general treatment strategy was comparable. During the pandemic, different variants of SARS-CoV-2 started to emerge, with possible different patterns of secondary infections. This may have implications for the effectiveness of a PCT-guided protocol in the future. This is the first study prospectively evaluating a PCT-guided antibiotic protocol in patients with COVID-19. Our study supports that a PCT-guided antibiotic prescription protocol can be safely implemented in clinical practice. The pragmatic decision using only a few exclusion criteria makes the implementation feasible for all patients admitted with COVID-19. To further support the evidence that a PCT-guided protocol reduces the use of antibiotic therapy, a randomized controlled trial should be conducted comparing standard care vs a PCT algorithm. In this study, a PCT-guided approach for prescription of antibiotics in patients with COVID-19 in the first 24 h after hospital admission resulted in a significantly lower antibiotic prescription rate, without any major safety concerns. The authors have reported to CHEST that no funding was received for this study.