ABSTRACT
Background: Randomized
controlled trials of drug use, and
violent behavior are difficult to
conduct and are unethical. We used
propensity score matching (PSM)
methodology to test whether there is
an association between using drugs,
and violent behaviour. We built a
model to predict the probability that a
student will perpetrate violence
Methods: A cross sectional study
involving 2000 adolescents in
secondary schools was conducted.
The outcome variable was violent
behaviour, exposure variable was
drug use while other covariates were
alcohol use, gender, age, family
type , father 's and mother’s
education, person who student lives
with, person who brought student
up , missed classes without
permission, and parental marital
status. We used Propensity Scores
to form a comparison or control
group (students who do not use
drug) that is similar to the exposure
group (students who use drugs) in all
observed predictors of violence. We
estimated the propensity score by
logistic regression.
Results: Our findings showed that
drug use was more common among
students whose parents had no
formal education, those who were
brought up by their grandparents,
those who live with their fathers only,
those whose parents were divorced
and those who used alcohol and
missed classes. Our results further
showed a significant association
between drug use and violent
behavior
Conclusion: This analysis
strengthens the evidence of the
association between drug use and
violent behaviour. It also showed
that PSM methodology, by
compensating for selection bias, is a
useful tool for estimating the
relationship between drug use and
violence in cross-sectional studies.
Key words: Keywords: Violence, Drug use,
Propensity scores, Adolescents
|
