The increase in bacterial resistance and the overuse of antibiotics have led health authorities to propose incentives for the proper use of antibiotics. Among these measures, the introduction of referring physicians for antibiotic therapy and tele-advisory devices for infectious diseases have shown positive effects on antibiotic prescriptions in hospitals. Today, an increase is observed in the consumption of antibiotics linked to ambulatory prescriptions.

The objective of the project is to deploy tele-advice devices for general practitioners and to evaluate the effects on ambulatory antibiotic dispensing.

AIRBUS-ATB is a prospective, multi-center, population-based, interrupted time-series observational study with a control group with 12 points before and 24 points after the deployment of the intervention in voluntary territories.

1. Current knowledge on the problem

   The increase in bacterial resistance and the overuse of antibiotics have led health authorities to propose incentives for the proper use of antibiotics. Among these measures are public awareness campaigns, the establishment of antibiotic reference materials and the development of indicators of appropriate use in health facilities.

   In response to these issues, the infectiology teams have developed cross-cutting activities to improve antibiotic prescribing in their healthcare institutions: drafting and implementing clinical practice recommendations; initial and continuing training of prescribers in the proper use of antibiotics; evaluation of practices for the management of infectious diseases; organization of collaborations with the CLIN, microbiology laboratories and pharmacy; monitoring and analysis of antibiotic dispensing, monitoring and analysis of bacterial resistance.

   Some teams have set up prescription support systems to meet the growing demand from non-infectious disease clinicians and to satisfy the criteria of the Composite Indicator of Antibiotic Use (ICATB 2), which relates to "access to antibiotic therapy advice". In 2000, the cross-disciplinary infectiology team at Grenoble Alpes University Hospital set up a "Mobile Infectiology Consultation (MIC)" based on a cell phone medical service available 24 hours a day, 7 days a week. This team carried out an analysis of all the informal consultations treated by this "hotline" during 2008. 7863 advice was given during the study period; 37.3% of them within the University Hospital Center to adapt or start antibiotic therapy. This activity has developed significantly and is now well established in France in the hospital environment. The Grenoble team conducted a new evaluation in 2016 and recorded nearly 16,000 calls per year (publication in progress).

   Studies of intra-hospital infectious disease teleadvisory services have shown that compliance with infectious disease advice reduces the consumption of anti-infectives and improves the prognosis of hospitalized patients. Thus Sellier et al reported the results of a prospective study carried out on 621 patients benefiting from infectious disease counseling. Adherence to the advice given was good 88.2% (548/621). When counseling was followed clinical improvement at D3 was more frequent (60.7% versus 43.9%, p=0.01) and the median duration of hospitalization was 3 days shorter (20 days versus 23, p=0.03). Adherence to telephone advice was comparable to that of formal advice given during a consultation, out of a series of 627 consecutive advice sessions reported by the same team. These results are supported by other studies that highlight the positive impact of informal consultation with infectious disease specialists on the management of inpatients, particularly in intensive care units.

   Between 2000 and 2004, the consumption of antibiotics decreased in France by 18.4%. Then it went up and down between 2005 and 2008. Since 2010 there has been a recovery trend that is confirmed every year (+5.9%). The growth in consumption appears to be much higher for outpatient prescriptions than for hospital prescriptions (90% versus 10%).

   A more recent decree (2015), specifying the role of referring physicians as well as axis 3 of the report "Saving antibiotics" recommended developing infectious disease advice for primary care physicians (general practitioners).

   Experiences concerning tele-advice to general practitioners are, however, rarer than intra-hospital experiences and have been little studied. In Grenoble, the hotline was opened from the outset to general practitioners and practitioners outside the university hospital. More than half of the advice provided (56%) by this team concerned them. Requests were more often handled by telephone when they came from general practitioners (89% versus 46% for intra-hospital requests). The reasons for calling the hotline from GPs were significantly different from those of hospital phisicians. Indeed, in this study, intra-hospital requests concerned the management of osteoarticular (14%), skin and soft tissue (9%), or pulmonary (11%) infections; whereas GPs were more often concerned about long-term fevers or unexplained inflammatory syndromes (11%).

   Marquet et al. recently described the activity of Medqual, a network of 20 infectious diseases specialists in the Pays de Loire region. During the 5 days studied, this network had received 386 telephone calls, most of which (81%) came from physicians in the same hospital. Of these, 7.7% were from other hospitals and 11.3% from private practice. 6% of the calls resulted in formal consultations and 5.5% in hospitalizations. 68.7% of responses could be given by telephone.

   Bal et al questioned 284 general practitioners using an infectiology hotline. 97.9% of them were satisfied with the hotline. They said they used this service to improve patient care (96.5%) and appreciated the speed of access (86.3%). Their questions mainly concerned antibiotic treatments (66.2%) but also diagnostic aids (46.5%) and requests for consultations or hospitalizations (29.6%). The latter requests are particularly important to optimize the patient's journey.

   Teleconsulting in infectious diseases therefore seems to meet a need by allowing rapid recourse to a physician specialized in infectious diseases.

   The objective of our project is to extend the teleconsulting activities for general practitioners to new territories by teams of volunteer infectious disease specialists. The investigators wish to evaluate the effects of this activity on the evolution of antibiotic dispensing and on the prescribing practices of general practitioners.

   Outpatient dispensing of antibiotics will be analyzed from the dispensing of drugs registered by the Assurance Maladie, through the study of the general sample of beneficiaries (EGB) at the 100th of the protected population.

   Advice for general medicine will be recorded on an ad hoc database, the AIRBUS database. This system, which is independent of hospital information systems, should enable referring infectious diseases specialists to record in real time their out-of-hospital tele-advice activity.

   The originality of the project lies first of all in the target of the intervention. Up to now, the devices to assist in the prescription of antibiotics have been intended for hospital services within the framework of referring physicians in antibiotic therapy and mobile infectiology consultations. The deployment of tele-advisory devices in infectious diseases to primary care medicine is an innovation whose usefulness has yet to be demonstrated.

   The other element of originality lies in the choice of an objective judgment criterion: the consumption of antibiotics on an outpatient basis. The investigators are aware that the number of treatments for which specialized advice will be provided will be minimal compared to the mass of antibiotic prescriptions in ambulatory medicine. However, the investigators believe that the advice given by infectious diseases specialists will have an educational effect likely to induce changes in practices and changes in dispensing by halo effect. This is why the investigators suggest to collect the judgement criteria over a long period (3 years) and to collect it in the same way in territories not covered by a tele-advice system.

   This work will also make it possible to better understand the needs of general practitioners in terms of assistance in the management of infectious pathologies and to adapt the offer of continuing education in this specialty.

   In addition, the implementation of teleconsulting devices by hospital specialists constitutes a new model of relationship between professionals that could contribute to improving collaboration between primary care medicine and hospital specialist medicine.

2. Research hypotheses

   It can be hypothesized that an infectious advice hotline for general practitioners could have the same type of effects on the consumption of anti-infectives and the clinical evolution of patients as intra-hospital advice. This is suggested, for example, by a recent study by Meeker's team which describes the importance of behavioural interventions on the quality of antibiotic prescriptions in general medicine. In addition to the positive impact on patient care through real-time response and thus immediate action on therapeutic decisions, this advice would allow additional 'hands-on' training of the doctors involved in city infectiology and the correct use of antibiotics, through advice given in concrete situations (halo effect). Its impact would also be measurable on care trajectories (e.g. saving on medical consultations, avoiding emergency visits).

   If our hypotheses are confirmed, the infectious diseases tele-advice system could be deployed in other French healthcare territories. In the long term, it is hoped that this system will have a beneficial impact on the consumption of antibiotics and on the development of bacterial resistance.

3. Research objectives and endpoints

   3.1. Main objective

   To determine the effects of the deployment of an infectious disease tele-advice system for general practitioners on the overall and class dispensing of antibiotics in ambulatory medicine compared to the absence of specific intervention.

   3.2 Secondary objectives

   • To qualify the needs of general practitioners in infectiology for advice.
   • To study the responses provided by the infectious disease referents (type of response, time spent).
   • To study the compliance of general practitioners with this advice.
   • To study the impact of advice on the patient's care pathway.
   • To study the opinion of requesting GPs on the device.

4. Study description

   4.1 Study design

   This is a prospective, multi-center, population-based, interrupted time-series observational study with a control group with 12 points before and 24 points after the deployment of the intervention in voluntary territories.

   4.2 Intervention

   The intervention consists of setting up in each voluntary center an infectious disease tele-advice system available to general practitioners via a dedicated cell phone line. GPs will receive information when the telecounseling system is set up in their practice health territory.

   4.3 Population

   The primary study population will consist of health insurance-registered systemic antibiotic dispenses for patients present in the EGB for 36 months: 12 months prior to deployment of the telecounseling devices and 24 months post-deployment.

   • The experimental group will be made up of all metropolitan departments covered by an infectious disease tele-advice system open to general practitioners; 13 metropolitan departments will be concerned.
   • The control group will be made up of all the metropolitan departments not covered by an infectious disease tele-advice system open to general practitioners; 82 departments will be involved. Few centers have set up a response system accessible to general practitioners. For those centers that are available, the number of notices is very low.

   Number of planned observations to be recruited:

   A study of ambulatory dispensing of systemic antibiotics recorded in the SNIIRAM database in 2015 in the Rhône-Alpes region found a consumption of 64,165,140 Defined Daily Doses (DDD) for 6,510,561 inhabitants, i.e. 27 DDD/1000 inhabitants per day.

   By extrapolating these data to the Simplified General Sample of Beneficiaries (1/97th of the SNIIRAM database) for metropolitan France, it can be estimated that the analysis will cover 70 to 80 million DJD per year, i.e. 210 to 240 million DJD over 3 years. The unit of analysis will be the consumption per month and per territory, i.e. 96x36 = 3456 units month-department.

   The secondary population of the study will be constituted by the opinions requested by a general practitioner from an infectiological tele-advice system. The opinions requested during 12 months, from September 2018 to August 2019 will be recorded in the AIRBUS database.

   Number of planned observations to be recruited

   Based on a study carried out in 2016 in 5 departments of the Rhône Alpes region, it can be estimated that each center should give about 30 pieces of advice per month to the general practitioners of their department. Therapeutic advice represents 40% of the advice given to GPs. The investigators can estimate the total number of counsels expected for the study at 4100 in one year, of which about 1500 will be followed up on D7.

   4.4 Data collection

   Infectiological warnings

   All requests for infectiological opinions made by general practitioners will be recorded in the AIRBUS database, a computer application developed on the Voozanoo platform (Epiconcept) by the SPILF. The platform is accessible via the Internet and is authorized to host medical data.

   Each participating antibiotic referent will have an identifier and a personal code allowing him/her to record his/her opinions and to access his/her own database. For each request, the referring physician will record, with the agreement of the requesting physician :

   • the name of the requesting general practitioner and his or her telephone number,
   • the reason for the appeal (advice for a patient, general question) ;
   • the characteristics of the patient (age, sex, diagnosis)
   • the date and time of the call ;
   • the identity of the sponsoring physician;
   • the modalities of the response (consultation, telephone, e-mail, other);
   • the notion of hospitalization following the notification (yes, no, in infectiology, emergency, in another department);
   • the notion of carrying a multi-resistant bacterium (BMR);
   • the time spent for counseling (in minutes, including application filling);
   • the response given by the infectiologist (starting an anti-infectious treatment, no modification of the current treatment, optimization of the treatment by oral relay or de-escalation or shortening the duration or changing the dose, stopping the antibiotic treatment, no treatment necessary, diagnostic assistance).

   The recording of opinions may be carried out by the participating physician or by a clinical research assistant. The application will also be available on a tablet or smartphone.

   Follow-up of notifications at D7

   For opinions involving therapeutic advice (initiation or modification of antibiotic therapy) an ARC will contact the requesting GP 7 days (± 2 days) after the opinion. During a telephone interview, he will collect the following information:

   • the prescriptions made by the physician, in order to assess compliance with the advice.
   • the evolution of the patient if the doctor has this information. This evolution will be coded according to previously defined criteria (cure, decrease in fever, decrease in the intensity of symptoms, decrease in a possible inflammatory syndrome) ;
   • basic information on the doctor's mode of activity
   • the GP's opinion on the device (usefulness, satisfaction, influence on practices, proposal(s) for improving the system) Responses will be collected on a 4-way Likert scale. These follow-up data will be recorded in a computer application developed for this purpose on the Voozanoo platform.

   Antibiotic dispensing

   Antibiotic dispenses will be tracked using monthly extractions from the SNIIRAM's simplified EGB database (EGBs) provided by the National Institute of Health Data (INDS). The extraction will concern 1/97th of all systemic antibiotic dispensations (oral or injectable) registered by the health insurance (EGBs database) for 3 years. For each dispensation the International Nonproprietary Name (INN) of the antibiotic, the quantity dispensed, the date and place of dispensing and the identity of the prescriber will be collected. These dispensations will be classified by month, department and therapeutic class. The dispensations will be expressed in defined daily doses (DDD) in relation to the number of inhabitants per department present in the EGB.

   Follow-up of the dispensations will begin in the year preceding the intervention and will continue during the study year and one year after, for a total of 36 months.

   4.5 Statistical analysis

   Analysis plan

   A Statistical Analysis Plan (SAP) will be developed before the database is frozen, reviewed by the coordinating investigator and an independent statistician. This plan will outline the principles guiding the analysis, including :

   • The analysis sample and any subgroups ;
   • the frequentist approach to the analysis
   • The statistical significance level and the level of confidence intervals ;
   • The possible adjustment of the p-value to take into account the multiplicity of tests;
   • The descriptive statistics used for the qualitative and quantitative variables; and
   • The possible transformations of the variables ;
   • Variables created or recoded;
   • The association measures used;
   • The univariate and multivariate analysis strategy. The SAP will include an explicit detailed description of the univariate and multivariate analyses presented in the final statistical report (FSR). Unplanned analyses in the SAP, either exploratory or confirmatory, are possible. They will appear as such in the FSR and manuscripts submitted for publication.

   Analysis of the main judgment criterion.

   Antibiotic consumptions extracted from the SNIIRAM database will be analyzed with the SAS software. They will be expressed in Defined Daily Doses (DDD) per 1000 inhabitants per day.

   Antibiotic dispensations extracted from the EGBs database will be analyzed with SAS software. They will be expressed in Defined Daily Doses (DDD) per 1000 inhabitants per day. The data will be analyzed using interrupted time series. The secular trend of dispensing will be modeled before the intervention and then the modifications of this trend will be compared between territories covered and not covered by the intervention by a test of the first-order interaction between time and type of territory. Seasonality will also be modelled using a 12-order moving average and the autocorrelation of residuals using the ARIMA model. The investigators will use the Box jenkins approach and in particular the autocorrelation and partial autocorrelation functions to develop the ARIMA model. The analysis will cover all antibiotic dispensing and dispensing of the major classes of antibiotics.

   Analysis of notices

   The statistical unit is the advice given by the infectiologist. The test sample will consist of all the opinions included. A flow chart in accordance with the CONSORT/STROBE transparency recommendations will present the number of test specimens.

   The characteristics of the opinions will be summarized by the usual descriptive statistics: number and percentage for qualitative variables and mean and standard deviation (or median and 25th-75th percentiles, in case of asymmetric distribution) for continuous quantitative variables.

   In univariate analysis, the characteristics of the opinions associated with compliance will be analyzed using the Chi-square test, replaced by Fisher's exact probability in case of expected numbers below 5, for the qualitative variables, and the Student test, replaced by the non-parametric Wilcoxon test, in case of deviation from normality, for the continuous quantitative variables. The investigators will identify characteristics independently associated with compliance using a multivariate logistic regression model. To prevent over-fitting of the model to the data, the covariates introduced in the multivariate model will be selected a priori based on the literature and the team's previous publications in the field. Because of the non-independence of opinions from the same infectiologist, the investigators will use a two-level randomized model (opinion/infectiologist). In a further analysis, the investigators will develop a three-level model (opinion/infectiologist/institution). The adherence analysis will include the complete observations for the dependent variable and the independent covariates (casewise analysis). To ensure the robustness of the results, the investigators will repeat this analysis using a multiple imputation method for missing data.

   The analysis of the other criteria will be exploratory. As such, no adjustments to the degree of statistical significance will be made. Statistical comparisons will be performed using the Student t test or the non-parametric Wilcoxon test for continuous quantitative variables and using the Chi-square test or the Fischer's exact probability for qualitative variables.

   Point estimates will be framed by a 95% confidence interval. The statistical significance level will be set at 5%, in bilateral situations. The analyses will be performed with Stata SE software (version 15.0 or later, StataCorp, College Station, TX, USA).

5. Ethical considerations

   The study does not involve direct action on patients; no nominative information will be collected. The study involves prospective collection of data from healthcare professionals: infectious disease physicians and general practitioners who are applicants. It therefore complies with type 3 of the Jardé law: "non-interventional research that does not involve any risk or constraint, in which all acts are performed and products are used in a usual manner".

   An information sheet will be distributed to each physician who is a member of a participating infectiology team and the physician's consent will be obtained in writing at the opening of the center. Information will be given by telephone to each requesting physician and his/her consent will be collected orally for the recording of the notification data and for the follow-up of the notification at D7.

   The protocol and the information documents of the doctors will be submitted for opinion to a committee for the protection of persons drawn by lot.

6. Auditing

Monitoring of study processes and documents will be conducted by personnel designated by the Direction de la Recherche Clinique et de l'Innovation (DRCI) at Grenoble university hospital.

Inclusion Criteria:

• Every patients included in the general sample of beneficiaries database (secondary use of an existing cohort of anonymized patients)

Exclusion Criteria:

• No exclusion criteria

The main study population consists of the dispensations of systemic antibiotics recorded by the health insurance in the base of the general sample of beneficiaries (1 / 100th of patients protected) for 36 months: 12 months before the deployment of telecommunication devices and 24 months after the deployment.

The secondary population of the study will be constituted by the opinions requested by a general practitioner from an infectiological tele-advice system. The opinions requested during 12 months, from September 2018 to August 2019 will be recorded in the AIRBUS database.