Infectious Diseases Experts as Part of the Antibiotic Stewardship Team in Primary Care

Overview

A cluster-randomised multicentre blinded clinical trial will be performed in six primary care centres located in the southern metropolitan area of Barcelona (Spain). The objective is to assess whether including experts on infectious diseases (ID) within the antimicrobial stewardship (AMS) team of primary care achieves higher reductions on overall antibiotic consumption and increases the quality of prescription in diagnosed upper respiratory and urinary tract infections. Centres will be randomly assigned to receive a standard-AMS or an advanced-AMS (intervention). Advanced-AMS includes all standard-AMS strategies plus general practitioner chance to discuss clinical cases by telephone to ID expert on working days (8:00 am to 8:00 pm), and by biweekly meetings.

SCOPE OF THE STUDY: A multicentre, cluster-randomised trial evaluating whether the direct integration of infectious disease specialists into primary care teams reduces total antibiotic consumption and enhances prescribing quality without compromising patient safety. TIMELINE: From June to September 2021 standard-AMS will be promoted to increase understanding and acceptance of AMS by GP. Baseline data collection will begin on October 2021 and will last six months. After these six months, on 1 April 2022, centres will be randomly assigned to receive an advanced-AMS, or to continue the standard-AMS. The intervention will be spanned 12 months, until 31 March 2023. STUDY GROUPS. The 6 centres will be assigned to receive an advanced AMS or a standard AMS, 3 in one and 3 in another. Take a look on study design and arms and interventions section. RANDOMISATION. Take a look on the study design section. BLINDING. The two secondary endpoints, unnecessary antibiotic therapy and adequacy of therapy, will be evaluated by blinded investigators. Discrepant decisions among them will be discussed with a third blinded antibiotic expert to reach a consensus. Due to the nature of the intervention, it is not possible to blind the primary care centres or the GPs. OUTCOMES. Outcomes will be measured in each primary centre with a monthly periodicity during 12 consecutive months. These variables are: * Main objective. To assess the impact of advanced-AMS on overall antibiotic consumption measured by defined daily dose of antibiotic per 1,000 inhabitants per day (DHD). * Secondary objectives: * To assess the impact of the advanced-AMS on unnecessary antibiotic prescriptions in diagnosed upper URTI and UTI. * To evaluate the impact of the advanced-AMS on the adequacy of antibiotic prescriptions. * To evaluate the impact of the advanced-AMS on the total number of re-attendance to GP or emergency department for any reason within 30 days after the initial GP evaluation. * To evaluate the impact of the advanced-AMS on the number of hospital admissions for any reason within 30 days after the initial GP visit. DEFINITIONS. * Consumption is measured in DHD, following the Anatomical Therapeutic Chemical (ATC) classification and the defined daily dose (DDD) methodology established by the WHO (https://www.who.int/tools/atc-ddd-toolkit/about-ddd) * Unnecessary prescription will be considered as follows: * An antibiotic prescribed in URTI, which disagrees with the local antibiotic guidelines. * An antibiotic is prescribed in asymptomatic bacteriuria, which is defined as having a positive urine culture with ≥ 10EXP5 CFU/ml of an uropathogen in patients without signs or symptoms of urinary tract infection (dysuria, urinary frequency or urgency, suprapubic pain, fevers, or flank pain). * Adequacy of antibiotic therapy will be considered when prescription fulfils the local guidelines: type of antibiotic + dose, and length of therapy. DATA COLLECTION AND MONITORING. * Primary Outcome Data (Cluster Level), antibiotic DHD, will be monthly recollected from two sources: • Catalan Health System (SCS) Electronic Prescribing Registry: To provide data on all reimbursed and dispensed systemic antibacterials (ATC Group J01) use in population aged \> 14 years attended in each participating centres. * Central Insured Persons Registry (RCA): To provide demographic denominators (insured population aged \>14 years) aggregated by sex and age for each participating centre. * Secondary and Safety Outcome Data (Patient Level) will be collected through monthly point prevalence surveys conducted on the last working day of each month by a cross-sectional analysis. On these selected patients, information from electronic medical records (SAP logon® and eCAP) will be obtained. Investigator will obtain the following variables: demographics, Charlson score, type of infection, antibiotic prescribed, doses and length of antibiotic therapy, re-attendance to the GP, attendance to the emergency department, and hospital admission. All the included patients will be followed up 30 days after the initial primary care physician visit. To ensure completeness of data, a telephone call will be allowed if any crucial data is missing in the electronic medical record. Information regarding the burden of the intervention such as the number of telephone calls received by the ID expert per month and the number of participants to biweekly meetings with the ID will be also collected. All data will be recorded by the study monitor on a standardized data abstraction form by using RedCap (Research Electronic Data Capture) hosted at IDIBELL, a secure web application for building and managing online databases. WITHDRAWAL / FOLLOW-UP Given the pragmatic design based on electronic health records and point-prevalence surveys, participant withdrawal is expected to be minimal. Patients who cannot be reached by telephone to provide missing information will still be included in the intention-to-treat analysis for all outcomes for which electronic data are available. SAMPLE SIZE. The study will include six primary care centres, providing coverage for 147,644 inhabitants (approximately 37% of the Delta del Llobregat reference population as of 2020). The selection of these six clusters represents a pragmatic sampling approach driven by logistical and organisational constraints to ensure the implementation of the advanced AMS. Power calculation: The power calculation is based on the primary objective (reduction in overall antibiotic consumption). While the clinical target is a 10% reduction from the 2017 baseline median (8.94 DHD), the study specifically powered to detect a mean difference of 2 DHD between the intervention and control groups. Statistical Assumptions: * Intracluster Correlation Coefficient (ICC): 0.2 * Intracluster Variance: 4 * Statistical Power (1 - β): 80%. * Type I Error (α): 0.05 (two-sided) STATISTICAL ANALYSIS. All analyses identified in the protocol and in this SAP will be conducted after the last subject completes the follow-up period and the database is locked. Statistical significance will be set at a two-sided p-value \<0.05. All relevant effect estimates will be presented with 95% confidence intervals (CIs). Full details of additional analyses will be presented in the clinical study reports, and any such analyses will be clearly identified as post-hoc. Analyses will be performed using R software (version 4.0.1 or later) \*Data description and baseline characteristics Categorical variables will be summarised using frequencies and percentages. Continuous variables will be described using the mean and standard deviation (SD) or median and interquartile range (IQR), depending on the data distribution. Demographic and baseline data will be presented using descriptive statistics only; no formal testing will be performed. \*Primary Outcome Analysis (DHD) The primary outcome is total antibiotic consumption. A descriptive analysis will be performed on the monthly consumption of antibiotics by the study group (standard and advanced AMS), time and cluster. The monthly consumption of antibiotic evolution will be represented on a graph by months, and a smoothed curve will be estimated using locally weighted smoothing. Antibiotic consumption will be compared according to the study group using a time series model. The dependent variable will be the monthly consumption of antibiotics expressed in DHD and the independent variable the study arm. In the estimation of the model, the seasonality, trend, and first- and second-order lags of the series will be analysed. In addition, the use of hierarchical models owing to cluster design will be considered. \*Secondary Outcome Analysis (adequacy and safety) A McNemar test will be used to compare the groups and categorical secondary outcomes: unnecessary prescription, adequacy of prescription, re-attendances to GP or the emergency room and hospitalisations. Additionally, log-binomial regression models will be used to estimate the effect of the intervention in terms of risk on the outcomes adjusted by patients' clinical profiles. To estimate models, patients nested in primary care centres will be accounted; therefore, a correction of the standard errors will be conducted using the variance-covariance matrix. Finally, inter-rater reliability between blinded investigators regarding the necessity and adequacy of prescriptions will be assessed using the Cohen's Kappa statistic. \*Changes from the original planned statistical analyses The following adjustments were made to the original analysis plan to improve statistical robustness. All modifications will be fully documented in the clinical study reports: * Unpaired data: Chi-square tests were used instead of the originally considered McNemar test, as the study compare independent groups. * Model selection: Mixed-effects logistic regression was selected over log-binomial models to address convergence issues; results are presented as Odds Ratios (OR). * Primary outcome modelling: The initially planned time-series model was replaced by a mixed-effects Poisson regression. This approach better handles the cluster-level data distribution and accounts for random effects by centre. * Exploratory analysis: An exploratory subgroup analysis by antibiotic type will be conducted for the primary outcome. This analysis was not pre-specified in the original SAP version 1. LIMITATIONS. The study does not contemplate randomisation by prescribing physician, but by conglomerates to avoid possible contamination effects at the prescribing physician and patients in the same centre. As it is a cluster analysis, there could be differences in the degree of prescription of doctors within the same centre and between the different centres. In order to reduce this problem, the variability of prescribing intra-cluster and between clusters has been calculated during the period 2017 in a group of primary care centres in the intervention area, differences that have been taken into account for the calculation of the sample. Design of the present study does not include a third group without any AMS intervention due to department policies during 2019. ETHICAL ASPECTS. This research will be carried out under the standards of good clinical practices following the principles of the Declaration of Helsinki (Fortaleza, 2013), as well as the current legislation applicable to this type of study. This study has been approved by the Ethics Committee of Hospital Universitari de Bellvitge and the Foundation University Institute for Primary Health Care Research Jordi Gol i Gurina (IDIAPJGol) who waived the need to obtain informed consent since the intervention is mainly aimed at health-care professionals (GP) who will be previously informed of the objective and other relevant aspects of the study (REF. PR277/19 - REF. 4R20/26). Clinical data treatment of the included patients in the research project as well as that of associated professionals (primary care physicians) will be carried out following the established by the Organic Law 3/2018, of December 5, and by Regulation (EU) 2016/679 of the European Parliament and of the Council of April 27, 2016 on Data Protection (RGPD). Based on these regulations, the sponsor will guarantee the protection of the confidentiality of the participants in the program, both patients and doctors. The data collection sheets will be identified with a code. The name of the doctor or patient will not appear in any document, or any publication or communication of the study results.