Study aim
The objective of this trial is to establish whether the PCT-guided strategy is superior to standard of care based on existing experience, expertise and implemented guidelines. The primary efficacy endpoint is that PCT-guided strategy is superior in terms of antibiotic use, expressed as the Defined Daily Dosage (DDD), and duration of antibiotic therapy expressed in days of therapy. As this trial is designed to shorten antibiotics safely, the primary safety endpoint will be overall mortality measured at 28 days and at 1 year.
Secondary endpoints are reinfection rate, length of stay in the ICU and cost effectiveness.
Study design and setting
This is a nationwide multicenter prospective, randomised, controlled, open intervention trial performed in 16 Dutch surgical and medical ICU’s.
The Ethics Committee of the VU University Medical Center Amsterdam, Netherlands, approved the study protocol which is in full compliance with the Helsinki declaration. The study is coordinated by a steering committee (the authors), consisting of intensivists from participating ICU’s, who all contributed to the design and execution of this trial. Furthermore, a statistician, an epidemiologist and a pharmacist support the steering committee.
The trial is supervised by an independent safety monitoring board that is not involved in the design and conduct of the trial, or in the recruitment of patients. The board consists of a pulmonologist, an intensivist and a statistician.
Stratification, randomisation and blinding
Patients will be stratified by diagnosis and centre. Stratification will be performed according to the diagnosis of sepsis, severe sepsis and septic shock. Allocation of patients to either treatment group is concealed by using a centralised randomisation procedure with a computer generated list produced by an independent research organisation, the Julius Center for Human Research, Utrecht, the Netherlands.
Patients
Included will be patients with an age >18 years, and for whom antibiotics were initiated for an assumed or proven infection on admission or during ICU admission. Current guidelines, like the Surviving Sepsis guidelines, advocate swift initiation of antibiotics whenever infection/sepsis is considered [5]. This study will not interfere with this decision of the physician to start antibiotics. The study, therefore, is the ultimate combination of early-goal-directed therapy and reduction of antibiotic duration. Patients can be included within 24 hours after receiving their first dose of antibiotics. Informed consent has to be obtained in writing from the patient or his or her legal representatives prior to inclusion.
Exclusion criteria are: (1) the inability to acquire a written informed consent, or (2) when prolonged antibiotic therapy is indicated (>3 weeks, e.g. endocarditis, cerebral/hepatic abscess), (3) patients with severe infections due to viruses, parasites or tuberculosis, (4) patients entering the ICU for merely short-term post-operative observation, (5) patients with an estimated length of stay less than 24 hrs, (6) patients suffering from cystic fibrosis, (7) severely immunocompromised patients such as patients with HIV and a CD4 count of less than 200 cells/mm, neutropenic patients (<500 neutrophils/mL), (8) patients with solid organ transplantation, or (9) moribund patients.
The use of corticosteroids does not interfere with PCT measurements, while effects on CRP levels have been demonstrated [20]. Since the purpose of this study is to demonstrate the efficacy and safety of PCT-guided therapy in a real-life setting patients on corticosteroid therapy will not be excluded from the study. Use of systemic corticosteroids will be monitored and recorded throughout the trial.
Treatment and intervention
The decision to start antibiotic treatment will in no way be affected by the trial. In all patients antibiotics will be started based on a clinical suspicion of infection or microbiological evidence of infection. This decision is fully at the discretion of the treating intensivist. Once antibiotics are administered for newly suspected or proven bacterial infection, patients or their legal representatives will be asked for informed consent. If informed consent is obtained, the patient will be randomised to either the standard therapy arm (control group) or the PCT arm (intervention group). Randomisation will be stratified for diagnostic group and study centre. When a patient is randomised for the control group, no PCT measurements will be performed. Other laboratory tests for infection such as CRP are allowed. When a patient is randomised for the PCT group, PCT will be measured at base-line called the T0-serum sample (T0: as close to initiation of antibiotics as possible, at least within 24 hrs). On the following days the ICU team will be provided with daily PCT values until ICU-discharge or until the third day after all systemic antibiotics have been discontinued. Along with daily PCT values a non-binding advice will be generated to consider stopping the prescribed antibiotics if PCT has decreased to <20% of its peak value (relative stopping threshold) or has reached a value of below 0.5 ng/ml (absolute stopping threshold). PCT levels will not be used to initiate antibiotic therapy. In case of two consecutive PCT levels below 0.5 ng/ml a stopping advice will be generated.
As for the control group, the physician will receive routine daily laboratory values as requested and no additional advice. Thus, treatment for both patient groups will be based on existing experience, expertise and local protocols, however, in the intervention group doctors are additionally provided with daily PCT levels, and a non-binding advice on continuation or discontinuation of antibiotic therapy by email, electronic patient data management system feedback, or research nurses. In both groups daily multidisciplinary reviewing will be performed on antibiotic therapy and duration, based on the clinical course, microbiological results and treatment guidelines. If doctors do not adhere to the stopping rules, reasons for non-adherence will be recorded.
Data collection and management
Data management will be performed by the investigators or research nurses of the participating ICU. The investigators must ensure that the patient’s anonymity is maintained. The subjects will be identified by a trial identification number. The list containing the subjects name and allocation numbers are kept in strict confidence by the principal investigator. The database used for electronic data transfer of any clinical research file (CRF) or subject related data will be protected by a password. Data base integrity and data safety as well as privacy are warranted by the contracted research organisation, the Julius Center for Human Research, Utrecht, the Netherlands.
Procalcitonin assays
PCT will be measured with various validated assays: the automated Kryptor platform (Thermo Fisher Scientific, Hennigsdorf, Germany), the Roche Elecsys Thermo Fisher Scientific PCT assay, the Siemens Centaur Thermo Fisher Scientific PCT assay or using BioMerieux’s Vidas Thermo Fisher Scientific PCT assay. The Thermo Fisher Scientific Kryptor sensitive PCT will be applied on this platform using Time Resolved Amplified Cryptate Emission (TRACE) technology and is based on a polyclonal antibody against calcitonin and a monoclonal antibody against katacalcin, which binds to the calcitonin and katacalcin sequence of the calcitonin pro-hormone. The test is considered a homogeneous immunoassay (sandwich principle) and is validated on serum and plasma (EDTA and heparin) matrix. The direct measuring range of the assay is from 0.02-50 ng/ml, with automated dilution extending the upper range to 1000 ng/ml. The Functional Assay Sensitivity (FAS) is 0.06 ng/ml.
The Roche Elecsys Thermo Fisher Scientific PCT assay also uses an immunoassay based on a sandwich principle based on a polyclonal antibody against calcitonin and a monoclonal antibody against katacalcin, which binds to the calcitonin and katacalcin sequence of the calcitonin prohormone. The direct measuring range of the assay is from 0.02-100 ng/ml. The Functional Assay Sensitivity (FAS) is 0.06 ng/ml. Procedure time of the assay is 18 minutes. The assay is validated on serum and plasma (EDTA and heparin) matrix.
The BioMerieux Vidas Thermo Fisher Scientific PCT assay is an Enzyme-Linked Fluorescent Assay (ELFA) involving a one-step immunoassay sandwich method using a Solid Phase Receptacle (SPR). The system uses a ready to use test strip into which a sample volume of 200 micro litres of serum or lithium heparin plasma is pipetted. The direct measuring range is 0.05-200 ng/ml. The FAS is determined to be 0.09 ng/ml.
Procedure time for all these assays is less than 30 minutes. Each participating centre will have access to a Kryptor machine, a suitable Vidas or Roche immunoanalyser to expedite the determinations and its adjunctive advice.
Whenever results exceed the maximum measuring range of any of the above described machines, manual dilution will be used, in accordance with the manufacturer’s specifications to achieve the true quantified value of a patient result. All patient results will be given in two decimals for the entire measuring range.
Sample size and statistical analysis
Antibiotic duration is the main primary outcome of this on-going study. The expected standard deviation (SD) of antibiotic treatment, based on previous intervention trials [15–20] will be 6 days in both groups. Assuming a mean baseline antibiotic duration of 8 days, with a significance α-level of 5% and a power of 90%, 757 patients are needed in both groups. Patients that are discharged from the ICU before a stopping advice could have been issued and who are still being treated with antibiotics are considered “dropped out”. Assuming a drop-out rate of 20%, we need to include 908 patients in each group, so for the superiority margin in total 1816 patients will need to be included.
As the PCT intervention arm should be non-inferior, in terms of safety (mortality), in comparison to the standard treatment, the non-inferiority margin for PCT guided antibiotic management regarding 28-day-mortality is set on 8%. Based on previous trials, like the PRORATA-study [17], we assumed a 28% mortality rate in each group. With a one-sided significance α-level of 2.5% and a power of 90%, 714 patients are needed in both groups. Again with a drop-out of 20%, 1714 patients will need to be included. With such a sample size (and alpha 5% and beta of 20%) we would have been able to detect a 15% relative increase in mortality. Such conventional boundaries are very acceptable in most trials investigating the efficacy of new drugs. Here, we aim to do better, with an alpha of 2.5% and beta of 10%.
Losses to follow-up on both arms will be registered including the reason for loss to follow-up. Furthermore, the primary endpoint will be explored for association with potential prognostic factors in a survival analysis. The factors that will be considered are age, sex, APACHE IV-score, SOFA-score and usage of corticosteroids or dialysis.
The primary analysis population will consist of all randomized patients following the intention-to-treat principle. For the primary analysis, losses to follow-up (in both arms) will be included in the 28-day analysis of mortality and antibiotic consumption. The primary analysis will be repeated with the subset of patients who complied with the stopping threshold, as so-called “per protocol analysis” of mortality and antibiotic consumption on day 28.
Interim analysis
The first interim analysis was performed after enrolment of 750 patients. Preceding this analysis the non-inferiority margin was set at 8% and a p-value of 0.0294 was used, corresponding with the performance of one interim-analysis in accordance with the Pocock method [21]. The DSMB examined the data and the trial would have been stopped immediately if the above predefined margins were reached. Furthermore, the DSMB reviewed the trial’s progress and adverse events according to treatment assignment.
Duration
With the current inclusion rate of 50 patients per month, the total duration time is estimated at 36 months and the last patient is expected to be included around May 2013