Study design

This is a cross-sectional mixed-methods study carried out in Madrid, Spain, in three phases (Fig. 1). In phase 1, we conducted a systematic social observation in 500-meters street network buffers around secondary schools to collect contextual data on alcohol and tobacco accessibility and promotion. In phase 2, we surveyed a sample of secondary school students between 14 and 18-years about sociodemographic data and alcohol and tobacco consumption, and determinants of its consumption. In phase 3, we performed semi-structured interviews and one discussion group with students in these centers who have ever consumed alcohol or tobacco to complement the information gathered by the questionnaire in phase 2. Transversally, we will geocode all data obtained from these three phases to explore spatial patterns of the presence of alcohol and tobacco elements in the environment and consumption among adolescents by using Geographic Information Systems (GIS).

Figure 1. Chart illustrating the interrelationship of the methodologies used. 

Phase 1: systematic social observation around selected secondary schools

All schools that taught secondary education in the city of Madrid (n = 565) were classified into nine strata defined as the combination of the tertiles of neighborhood socioeconomic status (SES) and tertiles of alcohol and tobacco outlet density (i.e., SES High-High Density; SES High-Medium Density; etc.). To account for deprivation, we used a validated predefined index, which includes indicators based on unemployment, precarious employment, occupational class, educational level, and property value.¹⁵ We obtained alcohol and tobacco outlet locations from two public databases.^16,17 We defined outlet density as the number of outlets within a street network buffer of 500 meters around schools, using ArcGIS 10.6 (Esri, Redlands, CA, USA). We defined a “strata-balanced” strategy to contact and invite schools for participation to ensure a minimum of six centers within each strata. This may guarantee a minimum geographic variability of cases within each strata by including centers from different neighborhoods throughout the city (and avoid geographic-location bias). Centers within each strata were ranked by their alcohol and tobacco outlet density value. In strata representing high neighborhood outlet density, the contact of centers with the highest densities were prioritized. In contrast, the contact of centers with the lowest densities were prioritized in low neighborhood outlet density strata. Centers in medium neighborhood outlet density strata were randomly ranked for contact prioritization. The response ratio was 12.42%.

We used systematic social observation to gather data on alcohol and tobacco points of sale, promotion elements/banners and signs of consumption within a predefined street network buffer of 500 meters around schools that accepted to participate in the study. Data were collected between August and October 2021, excluding school holiday weeks. An updated version of the OHCITIES questionnaire^18,19 was used to record the alcohol items. We modified an existing questionnaire used elsewhere to collect tobacco exposure.^20-22

We provided theoretical-practical training to fieldworkers for data collection of alcohol and tobacco environments. We prepared maps of the buffer observation area around each secondary school (Fig. 2).

Figure 2. Field collection map in a buffer of 500 meters by street network around the educational center. 

Phase 2: individual data collection

We conducted a survey among students between 14-18 years, enrolled in secondary schools in the city of Madrid between February and April 2022. An initial sample size of 1,089 students was calculated, 363 participants for each socioeconomic level stratum (assuming power of 80% and losses of 20%). This would allow us to detect changes in alcohol and tobacco prevalence of 7% between different socioeconomic strata based on the results obtained in a previous study.²³ Forty nine of the 55 centers involved in this study agreed to participate in the collection of individual consumption data, obtaining a total sample of 2287 students (with a loss of six centers in this phase, of which four were from strata representing high SES and high or medium alcohol and tobacco outlet density areas).

The questionnaire included information related to sociodemographic data, alcohol and tobacco consumption patterns and determinants of their consumption. Alcohol and tobacco related questions were based on the Survey on Drug Use in Secondary Education in Spain (ESTUDES).²⁴ The questionnaires were provided and filled in paper format, except for seven centers that claimed for an online format. In these cases, we used a free online survey tool (https://www.surveymonkey.com).

Phase 3: Qualitative data collection

A theorical, intentional, non-probabilistic purposive sampling was carried out.²⁵ Inclusion criteria were being a student between 14 and 18 years of age and having ever consumed alcohol or tobacco. We recruited informants through the dissemination of a flyer in schools. We performed twenty semi-structured interviews (six from schools located in high and medium SES neighborhood, respectively, and eight from low SES neighborhood schools) and a discussion group (organized with five students from a medium SES neighborhood school) between April and September 2021. The topics covered in the interviews and discussion group tried to complement the information collected by the questionnaire in relation to alcohol and tobacco consumption patterns and factors influencing its consumption. Moreover, a section to explore health promotion strategies to prevent alcohol and tobacco consumption was included.

Statistical and geographic analyses

We will obtain prevalence ratios of consumption and consumption patterns. We will carry out qualitative and thematic descriptive analysis using ATLAS-ti software²⁶ and we will use sequential explanatory method²⁷ to relate quantitative and qualitative results. Transversally, we will use Geographic Information Systems to geocode exposure and consumption data. Geographic Information Systems will allow us to create measurements to characterize alcohol and tobacco environment (i.e., outlet proximity and density, density and visibility of promotion elements and signs of consumption) and analyze spatial inequalities in the alcohol and tobacco accessibility, promotion and consumption among different geographies in the city of Madrid. We will use Poisson regression models with multilevel robust variance, with the individual at level 1, the educational center at level 2 and the neighborhood at level 3 to assess associations between ecological data and individual characteristics.

Ethical considerations

This study was conducted according to the principles promulgated by the Declaration of Helsinki and the Ethics Committee of the University of Alcalá (CEI/HU/2020/48, CEIP2021/5/128) granted ethical approval. We informed participants and their legal representatives about the study and we requested a written informed consent from both parties, including also the permission to be contacted in the future. Confidentiality and anonymity of all data is guaranteed.