Substance Use
Effect of the survey mode on drug use reports: results of a mixed method for data collection during the COVID-19 pandemic in Chile Nicolas Rodriguez* Nicolas Rodriguez Jose Miguel Cabezas Mariel Mateo Pinones Francisco Cumsille Esteban Pizarro
Background: The COVID-19 pandemic presented unprecedented challenges to data collection worldwide, prompting the need for innovative solutions to ensure the continuity of surveys. This study tested the survey mode effect on drug use reports during the COVID-19 pandemic in Chile. Methods: We use data from the National Study of Drugs in the General Population in Chile, a nationally representative cross-sectional survey implemented since 1994. In 2020, along with the traditional CAPI (Computer-Assisted Personal Interviewing) mode, the CATI mode (Computer-Assisted Telephone Interviewing) was used if sanitary conditions did not allow face-to-face application. At the national level, 89% of the interviews were conducted face-to-face (CAPI) and 11% by telephone. Propensity Score Weights were used to balance respondent characteristics in comparing face-to-face and telephone respondents. Results: Results indicate that telephone respondents were more likely to report marijuana, alcohol, and cigarette smoking than in-person respondents. The survey mode effect was not significant in the lifetime report but was significant in the past year and past 30 days. In the past year, their 95% confidence intervals (CI) did not include the value of one for reporting intervals of the smoking cigarettes (OR = 1.52; 95% CI= 1.20–1.92), alcohol (OR = 1.31; 95% CI= 1.07–1.60), and marihuana (OR = 1.64; 95% CI= 1.10–2.45). In the past 30 days, confidence intervals did not include one for smoking cigarettes (OR = 1.53; 95% CI= 1.19–1.96), alcohol (OR = 1.28; 95% CI= 1.00–1.64), and marihuana (OR = 2.16; 95% CI= 1.28–3.65). We found consistent results using other matching techniques. Conclusions: Future research is needed to evaluate the mixed-mode surveys and their potential effect on the trend and data quality in household‐based drug survey administration.