Brent Harris

Developing countries like India are experiencing an increase in chronic tobacco related disease and mortality. The purpose of this study is to accurately describe any variation in smoking habits at the district level across India and to assess the suitability of using place of residence and select socioeconomic indicators in characterizing smokers. Univariate analyses, spatial autocorrelation, multiple regression, cluster and discriminant analyses are used to describe variation and differences between types of tobacco smoked in India. Cigarette and bidi smoking are found to exhibit significantly different spatial patterns and to vary by population characteristics. Place or residence, education and occupation characteristics are adequate descriptors of smoking prevalence at the district level in India. However, the results can vary based on geography. Very high and very low levels of smoking can be predicted using these variables. The appropriate allocation of assistance and prevention resources can benefit from this type of analysis.