Smoking among Teenage Children Attending School: A Systematic Review of Observational Studies with Assessment of Quality of Studies
This systematic review describe the methods of design and analysis currently adopted in survey research published in the literature of school based observational studies for smoking, and identify the common pitfalls made by researchers.
METHODS:
This systematic review followed Cochrane style methodology in five steps: i) setting selection criteria for studies and conducting a literature search, ii) review of the abstracts, iii) review of the complete articles, iv) data extraction and quality assessment of included studies, and v) synthesis of studies.
RESULTS:
Out of the 292 abstracts retrieved, 45 articles were selected for the final review. We found inconsistencies in the definition of smoking behavior or the tool for measuring adolescent smoking was not standardized across studies which impedes generalizability. Despite the different study designs, methods of analysis and variety of covariates our results indicate that individual level factors have particular importance, but environmental level factors are also not negligible in studying the etiology of smoking.
Findings showed that parents, siblings and peers are powerful influences for adolescent smoking. The exclusion of adolescents who did not attend school on the day of the survey or who dropped out of school may produce biased results and also have an effect on generalizability. Many studies failed to measure important confounding factors and the sample size was often improperly reported. Hierarchical linear modeling, random effects modeling and structural equation modeling were employed in relatively few studies.
CONCLUSIONS:
Our results highlight concerns regarding the analysis of data from complex surveys. Many studies ignored the environmental (e.g. retailer near to school, social environment of school and implementation of smoking policies at school) effects in their studies which may produce misleading inferences. The results of multilevel analysis help us to understand school level effects and between schools variance provides more insightful information on school level covariates that influence adolescents smoking habits.