Tobacco and income level: A systematic review and meta-analysis

CIAPPONI, AGUSTÍN; BARDACH, ARIEL; GLUJOVSKY, D; ARUJ, PATRICIA; MAZZONI, AGUSTINA; COMANDE, D; GIBBONS, LUZ; LINETZKY, B; CASETTA, B

Conferencia; ISPOR 2nd Latin America Conference; 2009

OBJECTIVES The association between income level (IL) and tobacco consumption, and its consequences, varies over time and between regions of the world. The aim of this study was to assess this association. METHODS We included observational studies, published in the last 20 years, with direct assessment of IL. We searched in the main literature databases, conferences index, tobacco control agencies, and contacted experts. Two independent researchers screened titles and abstracts. The full-text of selected studies and its risk of bias (using a STROBE-based checklist) were assessed by two researches. One reviewer extracted data, and another checked it. Disagreements were solved by consensus. We performed a random effects meta-analysis of adjusted ORs using Stata 9.0. We performed pre-planned subgroup and sensitivity analysis to evaluate heterogeneity (I2>50%) by calendar decade, continent, WHO region, country standardized mortality rate, risk of bias, gender, and age. RESULTS Out of 9575 references we included 137 articles (86% cross-sectional studies). A total of 125 papers (N=31,146,096 subjects) reported higher smoking prevalence among people with Low-IL than High-IL (OR 1.48, 95%CI 1.38-1.59), especially after the ?90s decade -except for the Eastern Mediterranean Region. Considering only studies with the three IL categories, a gradient was observed (for both genders): Low-IL vs. High-IL OR 1.54 (1.39-1.72), and Middle-IL vs. High-IL 1.25 (1.16-1.33). Twenty studies (N=987,885) reported adjusted data for tobacco attributable diseases (coronary heart disease; pulmonary disease; low birth weight, cardiovascular and all cause death). All studies, except three with null effect, exhibited statistically significant higher ORs of tobacco outcomes as household income decreased. The median proportion of tobacco/total household expenditures was 10.7%, 3.7%, and 1.8% in Low-IL, Medium-IL and High-IL respectively. CONCLUSIONS This is the first meta-analysis confirming and quantifying an inverse relationship between IL and smoking prevalence and consequences.