Is it a risk factor or confounder? A discussion of selected analytic methods using education as an example

With socioeconomic status controlled, cigarette use is lower among American Indians/Alaska Natives than whites

BACKGROUND

Higher crude prevalence of cigarette use among American Indians/Alaska Natives (AI/AN) than non-Hispanic whites (NHW) has helped engender an assumption that race/ethnicity explains the difference. This study examines whether being AI/AN versus NHW predicts greater use when socioeconomic status and demographics are controlled.

METHODS

Data came from the National Survey on Drug Use and Health (2013-2017). Using logistic regressions with socioeconomic (income, education) and demographic (gender, age, marital status) controls, differences between AI/AN (n = 4,305) and NHW (n = 166,348) regarding heavier cigarette use (past month daily use, past month use of 300+ cigarettes, and nicotine dependence) and current cigarette use (past month use plus 100+ cigarettes in lifetime) were assessed. Adjusted predicted probabilities were also constructed.

RESULTS

NHW, compared to AI/AN, had greater odds of daily use: adjusted odds ratio (AOR) = 1.23 (95% CI: 1.03-1.49); predicted probabilities-15.3% and 13.0%, respectively. NHW had greater odds of using 300+ cigarettes: AOR = 1.47 (CI: 1.19-1.83); predicted probabilities-13.6% and 9.9%. NHW had greater odds of being nicotine dependent: AOR = 1.57 (CI: 1.31-1.89); predicted probabilities-10.3% and 7.1%. A difference in current use was not found. As controls, income and education were especially impactful.

CONCLUSIONS

With controls, particularly for socioeconomic status, heavier cigarette use was lower among AI/AN than NHW, and a current cigarette use difference was not indicated. This contradicts the idea that being AI/AN versus NHW independently predicts greater cigarette use, and it underscores the importance of socioeconomic status for understanding cigarette use among AI/AN.

Education and self-care activities among persons with rheumatoid arthritis

Associations between low formal education and increased morbidity and mortality have been well established among persons with rheumatoid arthritis (RA) and other conditions. This study attempted to identify a partial explanation for the association between low education and poor outcomes among persons with RA by examining self-care activities performed by persons with different levels of education. Persons with 13+ yr of education were significantly more likely to perform specific self-care activities (e.g., using a heated pool, tub, shower, OR = 2.59; using relaxation methods, OR = 3.00; using stress control methods, OR = 2.41; avoiding certain foods, OR = 1.74). The association between education and performance of self-care activities was not linear. When significant differences were noted, 13 yr of education was usually the point at which performance was significantly different than among lower education groups; individuals with 12 yr of education often exhibited lower frequencies of particular behaviors than did individuals with 9-11 yr of education. The association between higher education and performance of more self-care activities may shed light on previously described associations between education and morbidity. However, low education should not be viewed as the cause of increased morbidity and mortality, but as a proxy for a constellation of factors responsible for poor health outcomes.