Title : Multifactorial models of risk and protection for frequent substance use among health sciences students in arequipa, Peru

Introduction: The use of psychoactive drugs is a growing public health and social problem among university students worldwide.

Objective: This study aimed, firstly, to identify the factors of the criterion variable by grouping the drugs under study according to their frequency of use and, secondly, to evaluate, through three alternative prediction models, some risk and protective factors that contribute to explaining the increase or decrease in frequent use of toxic drugs.

Methods: The study employed a quantitative paradigm and a cross-sectional, explanatory-causal design. The sampling method was simple random probability sampling, and the sample consisted of 236 students. The data collection instrument demonstrated qualitative validity and reliability.

Data analysis: First, factor analysis with Varimax rotation was used to group and establish the dimensions of the criterion or dependent variable. Second, multiple linear regression analysis was used to compare three alternative models that explain and predict the variance of the criterion variable. Third, the significance of the multiple regression model was verified using Snedecor's F-statistic, and the significance of the beta coefficients was assessed using Student's t-test. The product of the standardized beta and correlation coefficients of the final model was used to determine the proportion of the total variance of the dependent variable, Factor 2, explained by each predictor or independent variable.

Results: The study revealed, first, a two-factor matrix with eigenvalues ??greater than one, which together explain 44.5% of the total variance of the criterion variable, a proportion considered moderate. Second, the three-factor multifactorial regression model is statistically significant. This implies rejecting the null hypothesis: all regression coefficients in the model are zero in the population. However, the overall coefficient of determination is not very high; only 21% of the total variance of the dependent variable, Factor 2, is explained by the 12 predictor variables included in the final model. In particular, parental education is the variable that explains the largest proportion of variance (10%), followed by sexual relations under the influence of alcohol (3%) and the student's sex (2.7%). In summary, the coefficients of determination confirmed that risk factors, such as lower parental education, male sex, and sexual relations under the influence of alcohol, predict higher rates of frequent drug use among university students. Conversely, protective factors, such as higher parental education, lower exposure to alcohol consumption, and a greater perception of the health risks associated with cannabis and cocaine use, explain lower rates of frequent psychoactive drug use among students.

Conclusion: It is essential to consider risk and protective predictors when designing a program to prevent substance use among university students.

Keywords: Frequent drug use; risk factors; protective factors; health science students; multiple regression models

To be updated shortly..