Methods

Substances Included

For human medicine, use is assessed for the antibiotics relevant to the surveillance of antibiotic use and resistance. These are defined the following Anatomical Therapeutic Chemical (ATC) groups according to the WHO:

• A07AA – Intestinal anti-infectives
• J01 – Antibacterials for systemic use
• J04A – Antimycobacterials
• P01AB – Nitroimidazole derivatives

Antibacterials for systemic use (J01) form the core of the analysis, while the remaining groups capture additional agents used to treat specific infections.
 

AWaRe Classification

The WHO has developed the AWaRe classification to guide the appropriate use of antibiotics. It sorts antibiotics into three main groups:

• Access antibiotics are recommended as first or second choice treatment options for common infections. They are effective against the most frequent bacteria and carry a lower risk of driving resistance. The WHO recommends that they make up the largest share of antibiotic use.
• Watch antibiotics have a higher risk of developing resistance. They should be used more carefully and prioritized as key targets of stewardship programs and monitoring.
• Reserve antibiotics should be reserved for last-resort treatment of confirmed or suspected infections due to multi-drug-resistant organisms. They are used only to treat severe infections caused by bacteria that no longer respond to other antibiotics. They should also prioritized as key targets of stewardship programs to preserve their effectiveness.

In addition to these three groups, the WHO lists a Not recommended category, which includes fixed-dose combinations of several antibiotics that are not advised for use. Some substances are also left Unclassified, meaning the WHO has not assigned them to any AWaRe category. Each antibiotic is assigned to one of these groups based on its ATC code.

Antibiotic Classes

Antibiotics can also be grouped by their chemical structure and the way they act, again on the basis of their ATC codes. Common classes include penicillins, cephalosporins, macrolides, and quinolones.
 

Looking at use by class shows which types of antibiotic are prescribed most often, while the AWaRe classification shows how appropriate that use is especially regarding development of bacterial resistance. Together, the two views give a fuller picture of antibiotic use.

To allow comparison over time, across regions, and between healthcare institutions of different sizes, raw consumption data (number of packages) are converted into standardised metrics. This process follows the Anatomical Therapeutic Chemical (ATC) classification and the defined daily dose (DDD) methodology established by the WHO Collaborating Centre for Drug Statistics Methodology.

• Conversion of packages to DDD: The first step standardises the absolute volume of antimicrobials into DDD, the assumed average maintenance dose per day for a drug used for its primary clinical indication in adults.
  1. Calculation of the active substance in grams: The mass of the active substance is calculated by multiplying the number of packages by the number of units per package (e.g. tablets, vials) and the dose strength per unit in grams. Doses expressed in international units are converted to grams. For antibiotic combinations, enzyme inhibitors (e.g. clavulanic acid) are not counted towards the dose unit. The dose for combined antibiotics (e.g. trimethoprim/sulfamethoxazole) is specified by the WHO.
  2. Conversion to DDD: The mass of the active substance is divided by the official WHO-assigned DDD value for the relevant ATC code and route of administration, yielding the number of DDD consumed during the observation period.
• Calculation of standardised metrics: The absolute number of DDD is divided by a population or hospital-activity denominator to produce standardised consumption metrics.

1. DDD per 1000 inhabitants per day (DID)

   This metric is mainly used for national sales data (IQVIA™) and health insurance claims data (santéservices ag). It estimates the share of the population treated daily with a specific antibiotic.

   Methodology: The number of DDD is divided by the respective population size (from the Swiss Federal Statistical Office, FSO) and by the number of days in the observation period (e.g. 365 days for an annual rate). The result is multiplied by 1000. The permanent resident population on 31 December of the corresponding year is used. For the most recent year, the population relies on an extrapolation based on the growth of previous years. This value may be adjusted slightly once the official FSO datasets are published.

2. Number of prescriptions per 1000 consultations or 100,000 inhabitants

   This metric is mainly used for data published by the Sentinella network in the outpatient setting.

   Methodology: The number of prescriptions is divided by the number of consultations (in the practice and at home). The result is multiplied by 1000. Alternatively, the number of prescriptions can be divided by a Sentinella population based on the population receiving primary care from the physicians participating in the Sentinella network.

3. DDD per 100 bed-days

   This metric standardises hospital antibiotic consumption based on institutional activity. DDD per 100 bed-days accounts for the total duration of hospital stays.

   Methodology: The total number of DDD is divided by the total bed-days for the observed hospital or unit and time period. The result is then multiplied by 100.

Due to rounding, values may differ slightly between figures or percentages may not sum exactly to 100%.

ANRESIS – Hospital Antibiotic Use Data

Inpatient antibiotic consumption is monitored using data from the network of acute care public hospitals and private clinics participating voluntarily in the ANRESIS surveillance system.

• Collected Data: Depending on the hospital, antibiotic use is recorded either as the number of packages dispensed from hospital pharmacies to wards, or as patient-level administration data extracted directly from clinical information systems. Data include inpatient care from entire hospital sites and separate adult ICUs.
• Exclusions: To prevent structural bias, data explicitly exclude ambulatory, rehabilitation, geriatric long-term care, and long-term psychiatric care units.
• Network Scope: The surveillance covers acute somatic care hospitals, comprising around 75 participating hospital sites, representing approximately 55–60% of all acute care facilities and 70–80% of all inpatient bed-days over the years across Switzerland. Adult intensive care unit (ICU) coverage represents about two-thirds of Swiss hospitals equipped with ICU beds.

IQVIA™ – National Sales Data

National monitoring of antibiotic consumption uses comprehensive pharmaceutical industry sales data (Sell-In) compiled by IQVIA™.

• Collected Data: This dataset quantifies the total number of packages of antibiotics sold directly by the pharmaceutical industry to distribution networks. The data include three channels: data from public pharmacies (APO), self-dispensing physicians (SD), and hospitals (SPI). Unlike sentinel hospital networks, the IQVIA™ dataset is not restricted to acute care. It explicitly includes comprehensive distribution data spanning rehabilitation facilities, geriatric centres, psychiatric clinics, and certain nursing homes.
• Network Scope: The database provides an overview, allowing for the measurement of antibiotic consumption at the national level and across distinct linguistic regions (German-speaking, including Liechtenstein; French-speaking; and Italian-speaking Switzerland).

santéservices ag – Health Insurance Claims Data

Outpatient prescription tracking uses mandatory health insurance claims data based on the Tarifpool database. This captures data for services where an invoice was submitted to health insurers or billed directly to the insurers by the service provider.

• Collected Data: The system quantifies the number of packages of antibiotics dispensed by prescription in public pharmacies, by self-dispensing physicians, or in outpatient departments of hospitals, and billed to health insurers. The Tarifpool covered 85% of the mandatory-insured population in 2015, expanding to over 97% after 2019.
• Uncaptured Data: Bills that are retained by the insured person (e.g., due to a high deductible) and unsubmitted paper-based invoices are not captured in the database.
• Data Cleaning: Missing data are not filled in or estimated. When a discrepancy between cost and package counts is detected (such as reporting anomalies where package counts were mixed up with units within a package), the number of packages is explicitly corrected using the weighted median cost per package.
• Extrapolation: Raw data are extrapolated in a two-step process: first, to correct for the fact that the Tarifpool, which provides product-level data, does not cover all insurers and providers; and second, to represent the entire Swiss population with mandatory health insurance.
• Standardisation: To account for structural demographic variations among cantonal populations, DID rates are directly standardised according to age and sex.

Sentinella – Outpatient Prescription Surveillance Data

Prescription indications are derived from the Swiss Sentinel Surveillance Network ("Sentinella"), a collaborative project between general practitioners, the Federal Office of Public Health (FOPH), and university institutes for family medicine.

• Collected Data: For each patient to whom they have prescribed an antibiotic, participants complete a weekly questionnaire. They record the number of consultations (including face-to-face consultations and home visits), as well as their choice of antibiotic and the reason for prescribing it, as assessed by the physician.
• Exclusions: Observations from members who reported irregularly (i.e., fewer than 39 weeks per year and fewer than one prescription per week) were excluded.
• Network Scope: The surveillance infrastructure is sustained by a network of approximately 140 general practitioners and 25 paediatricians across Switzerland. Even if it is a voluntary system that is likely to select physicians with a particular interest in infectious diseases, it is considered representative of the Swiss population of primary care physicians.

IS ABV – Veterinary Sales Data

Marketing Authorisation Holders (MAHs) regularly report the sales figures for their products to IS ABV.

• Data Capture: Each product is entered in IS ABV with a unique identification number, brand name, ATCvet code, target species, and information on the approved method of application. Pharmaceutical premixes are listed separately. The entry also includes the number of "basic units" sold (e.g. vials [incl. volume], tablets, injectors, tubes, or bags [incl. weight]).
• Calculation: Total quantities are calculated by multiplying the volume of active ingredient in each basic unit by the number of basic units sold. Combinable filters (year, ATCvet code, administration route) allow specific queries. For antibiotics expressed in international units, conversion factors are applied according to the template provided by the European Surveillance of Veterinary Antimicrobial Consumption project (ESVAC) of the European Medicines Agency [2]. Each MAH reviews and approves its data, which are summarised by preparation and year. Finally, the data are checked by Swissmedic before publication.
• Inclusions & Exclusions: Products licensed solely for export are excluded, as they are not used in Switzerland and therefore do not contribute to resistance development in the country. The methods of application follow those used in comparable reports abroad (France, AFSSA; United Kingdom, VMD): oral, parenteral, intra-mammary, and topical/external. The only distinction possible is between "livestock", "companion animals", and "mixed", according to the authorisations. Specific animal species or age groups can be covered only if clearly stated in the marketing authorisation (e.g. intra-mammary injectors for cows or products for treating piglets).
   

IS ABV – Veterinary Prescription Data

The information system for antibiotics in veterinary medicine (IS ABV) records antibiotic prescriptions for animals.

• Network Scope: Since October 2019, veterinarians have been required to report all antibiotic prescriptions and sales for all animal species. The database allows the treatment intensity of farm animals and companion animals to be assessed and takes into account the different types of production (e.g. piglet rearing or dairy farming). It also enables regional, national, and international comparisons of antibiotic consumption and treatment intensity.
• Data Capture: Mandatory reporting can be done via the practice software or a local IS ABV app; group therapies can be reported only via the app. Reporting through the practice software has the advantage that prescriptions need to be recorded only once, in the veterinary practice or clinic. For the evaluation, however, it means that two reporting channels must be taken into account, which is a potential source of error. Most veterinary practices and clinics use the practice-software channel.
• Data Quality: Reporting veterinarians can check their prescription reports stored on the IS ABV server. Regular feedback of submitted data has been in place since May 2021, with monthly feedback to veterinarians and continuous access by farmers to their personal consumption data. Only veterinarians can update their own data, and improvements in error rates have been observed since this feedback was introduced. Ultimately, responsibility for correct reporting lies with the veterinarian.
• Data Cleaning: A three-step process identifies and excludes outliers. The first and second exclusion criteria are based on the median declared quantity per day and animal per antimicrobial class, as well as on the preparation and the production group: prescriptions above 15 times the median and/or above the 99th percentile were excluded. Finally, all prescriptions were checked manually using the "four-eyes" principle to remove any obvious errors. The data cleaning concerned only penicillins, tetracyclines, and sulfonamides. In total, 6557 prescriptions (0.8% of all prescriptions) were excluded from the active ingredient quantity analyses. This procedure was not possible for the prescription type "issued from stock", as neither the use category nor the number of animals treated had to be recorded.