The Impact of Air Pollution on the Consumption of Antimicrobials in the General Population

The Impact of Air Pollution on the Consumption of Antimicrobials in the General Population (the ONAIR Project)

Objectives: To test the hypothesis that certain levels of air pollution may lead to increases in 1) antimicrobial consumption in the general population and 2) the presence of antibiotic resistance genes.

Methods: The project will consist of two sub-studies:

1. A time series analysis of the association between daily antimicrobial consumption and air pollution in the 11 largest cities in Catalonia (2006-2022). Outcome variable: daily consumption of antimicrobials. Independent variables: air pollution and confounding variables. The analysis will use generalized additive models: through a random effects meta-analysis, the concentration-response estimates for each city will be added together to calculate the overall estimates.
2. Descriptive observational study of the relationship between the concentration of resistance genes and antimicrobial consumption in the Barcelona area (2021-2023).

Outcome variable: concentration of antibiotic resistance genes. Independent variables: defined daily dose of antimicrobials per 1000 inhabitants-day. Descriptive analysis of the concentration and diversity of antibiotic resistance genes according to air quality monitoring station (urban vs. hospital environment), month, and correlation with antimicrobial consumption.

The ONAIR project will consist of the two sub-studies described below:

SUBSTUDY 1

1.1 Design: A two-part time series analysis of the association between the concentration of the main air pollutants (PM2.5, PM10, NO2, O3) between 2006 and 2022 and the daily antimicrobial consumption by the general population.

1.2 Scope of the study and authorized sources of information: The study will be carried out in the 11 cities in Catalonia with more than 100,000 inhabitants, distributed in the four provinces of the region. These cities are home to 3,283,501 people. Daily data on air pollution will be obtained from the Atmospheric Pollution Monitoring and Forecasting Network (XVPCA). Daily consumption of antimicrobials will be estimated from the official prescription data recorded by the Council of the College of Pharmacists of Catalonia (CCFC) authorized by the Department of Health in Catalonia. The secondary clinical variables under study will be obtained through the Agency for Quality and Health Assessment of Catalonia.

1.3 Study subjects: Adults and children living in the 11 cities under study.

1.4 Definitions: Air pollution. This is defined as the detection of any substance in the air that can damage the health of humans, animals or plants, or cause material damage. Air pollution comes from different sources; it varies according to its composition which, in turn, varies according to the temperature and humidity of the environment. For the purposes of this study, since not all air quality monitoring stations record data on all the compounds, PM10, PM2.5, NO2 and 03 will be analysed. DDD (defined daily dose). This is the technical unit of measurement and comparison equivalent to the mean daily maintenance dose of a drug when used in its main indication via a specified route of administration in adults, expressed in terms of the amount of the active principle. It is defined and updated by the World Health Organization (WHO) (https://www.whocc.no/ddd/definition\_and\_general\_considera/#General). DDD/TID (defined daily dose per thousand inhabitants per day). This parameter is designed to compare the consumption of drugs expressed in numbers of DDD for different geographical areas and/or time periods. It is calculated by multiplying the number of DDD consumed per thousand inhabitants and dividing by the number of days in the period considered and by the number of inhabitants. DDD/TID formula = (nºP x NºF x AF x 1000) / (DDD x nº of inhabitants x nº of days) (nºP = number of packages dispensed daily. NºF = number of pharmaceutical forms (i.e., pills, capsules etc.) per package. AP = amount of active principle per pharmaceutical form). Population denominator: to calculate the number of inhabitants, the population (users with individual health cards and assigned to a care centre) under the care of each practitioner will be used, calculated from the data of the primary care information system, also including the nursing home population.

1.5 Description of the variables: Main outcome variable: defined daily dose of antimicrobials per 1000 inhabitants-day. Main independent variables: concentrations of PM10, PM2.5, NO2 and 03.

Confounding variables: Year, day of the week, temperature and relative humidity.

Secondary outcome variables: Defined daily costs of antimicrobials per 1000inhabitants-day, mortality due to natural causes, mortality due to respiratory causes, hospital admissions due to respiratory causes and primary care consultations due to respiratory symptoms.

1.6 Data collection: Pollution data: Data on pollution in the 11 cities under study from 2006 to 2020 will be obtained from the XVPCA for time series analysis. In each air quality zone, distinct areas can be identified according to land use (level 1: urban, suburban or rural areas) and the type of sources of emission (level 2: traffic, industrial or background).

The time series of the mean daily temperature and relative humidity from 2006 to 2020, measured by the Catalan Meteorological Service, will be recorded. Daily data on pollution, temperature and humidity in these areas in 2021 and 2022 will be collected prospectively. The day of the week and the time series of daily influenza cases will also be considered as confounding variables. The associations of concentrations of the main pollutants, the daily consumption of antimicrobials (DDD/TID) and the secondary variables described will be evaluated in separate analyses using a standard time series approach, a two-stage analytical protocol that has been widely applied in previous time series studies in many different cities. Population and antimicrobial consumption data. Data will be collected by age (> 65 years), sex and region.

For the calculation of the DDD/TID and dispensed cost/TID the following daily data will be collected by postal code: the medication (combination of active principle + pharmaceutical form + dose + number of pharmaceutical forms per package), ATC code, number of packages dispensed, date of dispensing and official price of each package. The number of users with individual health card and an assigned care centre will also be recorded.

1.7 Statistical analysis: In the first stage, associations between the daily dose of antibiotic therapy and the air pollutants evaluated will be estimated in each city using quasi-Poisson generalized linear models. As in previous studies, the following covariates will be included in the main model: the year (smoothed by cubic splines) to control the trend and temporal seasonality of the indicators; a variable indicating the day of the week to take into account weekly short-term variations; and temperature (smoothed by splines) and relative humidity (smoothed by splines) to control potential nonlinear confounding effects of weather in areas in which this information is available. Since the possible effect of a pollutant on the daily antibiotic dose may be projected over the following days, a range of time lags will be explored in the estimated models. The same procedure will be applied to the temperature. In the second stage, a random effects model will be used to meta-analyse estimates of the association between the pollutant and the daily antimicrobial consumption obtained in each city. Thus, it will be obtained the aggregate estimates and their 95% confidence intervals of the percentage change in daily antimicrobial consumption for each increase of 10μg/m3. The I² test will be used to evaluate the heterogeneity between cities. Similarly, models will be estimated to assess associations between secondary outcome variables and air pollutants. The statistical analysis will be carried out using the R statistical package (version 3.6.1 or higher (R Foundation for Statistical Computing) using mgcv for the main models and rmeta for the random effects model. Statistical significance will be considered for p <0.05.

1.8 Limitations: A possible limitation may be the difficulty in distinguishing between the impact of pollution on antibiotic consumption as a direct cause-effect relationship or as a confounding factor in the cause-effect relationship between air pollution and respiratory exacerbations of previous pathology.

SUBSTUDY 2

2.1 Design: Descriptive observational study of the concentration of antibiotic resistance genes (ARG) in the area of Barcelona during the period 2021-2023. The differences between the concentration of these genes at an air quality monitoring station close to a hospital and at another station elsewhere in the city, will be evaluated, as well as the interseasonal and inter-monthly differences. The presence of certain resistance genes will be assessed in relation to the consumption of certain groups of antimicrobials in the study population.

2.2 Scope of the study and sources of information: The study will be carried out in the city of Barcelona. The air pollution samples will be obtained for analysis from three of the XVPCA's air quality monitoring stations: one of them in the city centre where there is a significant volume of traffic, and the other one located near a third level hospital.

2.3 Study subjects: Adults and children living in the city of Barcelona.

2.4 Definitions: ARG (antibiotic resistance gene): the fragment of the genome of a microorganism that encodes antimicrobial resistance mechanisms. Microorganisms can acquire ARGs either via inheritance or through horizontal transfer of genes from other microorganisms. Air pollution: see the definition in subproject 1, section 1.4.

2.5 Study variables: Main outcome variable: Concentration of antibiotic resistance genes/month. Main independent variable: defined daily dose of antimicrobials per 1000 inhabitants-day.

2.6 Collection: Sample collection: Samples from the filters used to measure PM concentrations will be collected at three stations in XVPCA zone 1, corresponding to the city of Barcelona. Samples will be collected at two stations representative of urban traffic-related pollution (Plaça Universitat and Poble Nou) and at another station representative of a hospital area (Parc de Vall D'Hebron). The sampling will take place on the first Tuesday of each month. In this way, three monthly measurements will be made over a total of 30 months, so a total of 90 samples will be collected. If there is an episode of high contamination during the study, extra sampling will be assessed. Meteorological data corresponding to each measurement day will also be recorded.

Sample processing: Samples will be collected through Pallflex high volume (30m3) quartz fiber filters (15cm in diameter). Before use, the filters will be sterilized at 205°C for 5 hours and conditioned for 48 hours at 20°C and 50% relative humidity. Each filter will be weighed three times every 24 hours in a microbalance with 1g sensitivity. After sampling, the filters will be taken to the microbiology laboratory where samples of PM2.5 will be extracted, as follows. First, the filter will be cut with a sterilized cutter. Then, each filter fragment will be added to 2mL of sterile purified water with 0.05% Tween 20 solution in a sterile centrifuge tub; then, a 20 minute sonication will be performed followed by 40 minutes of vortex agitation at 2800rpm. The PM2.5 samples will be stored at -20ºC until they are definitively processed. DNA extraction and shotgun sequencing of microbial DNA: From the dilution obtained, 1mg/mL will be extracted. DNA extraction will be performed automatically using KingFisher (ThermoFisher) equipment with the MagMaxTM Microbiome Ultra Nucleic Acid Isolation (ThermoFisher) kit. The libraries will be constructed using the Illumina Nextera-XT® kit.

Sequencing will be done 2x150bp on a NextSeq550® by Illumina. The quality of the sequences will be analysed with the FastQC1 program and the corresponding filtering will be performed with the Trimmomatic2 program. Human contamination will be eliminated by mapping the sequences filtered against the human genome with the Bowtie 23 program. Metaphlan v24 software will be used to infer the microbial composition. To study the resistome, the acquired antibiotic resistance genes and the chromosomal point mutations will be identified with ResFinder 3.05. Population data and antimicrobial consumption data: These data will be collected in the same way as described for subproject 1 (see section), but in the area of Barcelona, and specifically in the postal codes corresponding to the areas where metagenomic data are collected.

2.7 Statistical analysis: Heatmaps will be created for the relative abundance of ARG based on the differences of 2-ΔCT. A network analysis will be performed based on Spearman's rank correlations to visualize the coincidence of ARG subtypes in the air and in bacteria communities. The ARG concentrations in different months and seasons will be compared using boxplots. In addition, a Pearson correlation will be made to relate the concentration of ARG to the consumption of antimicrobials in the areas under study. To assess the richness and diversity of the bacterial community, Alpha diversity will be calculated using Mothur 1.30.1 in accordance with the literature (Schloss et al, 2009) for a confidence level of 95%. The statistical analysis will be carried out using the R statistical package version 3.6.1 or higher (R Foundation for Statistical Computing).

2.8 Limitations: The main limitation of this substudy is its descriptive nature. The period of time under study does not allow the establishment of direct temporal relationships between the degree of air pollution and the concentration of ARG in a particular day. In addition, due to the complexity of the technique, only a limited number of ARG determinations can be performed; therefore, the analysis of their relationship with the consumption of antimicrobials by the population will have to be descriptive.

3. ETHICAL CONSIDERATIONS: This biomedical research project has been evaluated by the Research Ethics Committee of the Bellvitge University Hospital. It will be carried out under the precepts of the Declaration of Helsinki (Fortaleza, 2013). Aggregate clinical data from the database of the Department of Health of Catalonia will be used. For this reason, the informed consent requirement has been waived by the Ethics Committeee. Nonetheless, the data will be treated in accordance with Spanish (Organic Law 3/2018) and European (Regulation [EU) 2016/679) regulations.

Population based study.