Table 3.
Bidi smoking and related risk factors among adults aged 15 years or older by using multilevel logistic regression analysis, pooled GATS India 2009–2010 and 2016–2017a,b
| Characteristics | OR | 95% CI |
|---|---|---|
| Sex | ||
| Male | 17.1* | (11.3–25.9) |
| Female | Ref. | − |
| Residence | ||
| Rural | 1.5* | (1.3–1.8) |
| Urban | Ref. | − |
| Age (years) | ||
| 15–24 | Ref. | − |
| 25–44 | 3.0* | (2.3–4.0) |
| 45–64 | 5.5* | (3.8–8.0) |
| ≥65 | 4.5* | (3.1–6.5) |
| Education level | ||
| No formal education | 6.0* | (4.9–7.3) |
| Primary | 4.3* | (3.6–5.0) |
| Secondary | 2.4* | (2.1–2.7) |
| Higher than secondary | Ref. | − |
| Occupation | ||
| Government or non-government | 3.3* | (1.9–5.5) |
| Daily wages/casual laborer or selfemployed | 3.5* | (2.1–5.9) |
| Retired or unemployed | 3.1* | (2.0–4.8) |
| Homemaker | 2.7* | (1.7–4.3) |
| Student | Ref. | − |
| Wealth index (quintiles) | ||
| Lowest | 2.5* | (2.1–3.2) |
| Low | 2.3* | (1.9–2.9) |
| Middle | 2.0* | (1.7–2.5) |
| High | 1.8* | (1.6–2.1) |
| Highest | Ref. | − |
| Tobacco users, non-bidi only | ||
| Yes | 1.6* | (1.2–2.1) |
| No | Ref. | − |
| Survey year | ||
| 2009–2010 | Ref. | − |
| 2016–2017 | 0.8* | (0.7–1.0) |
| Knowledge scale | ||
| Knowd | 0.9* | (0.9–0.9) |
| Variance within clusters | VPCc | |
| Primary sampling unit | 2.2* (1.7–2.7) | 15.9% |
| State | 2.2 (1.8–2.6) | 31.8% |
a
Included three hierarchical levels: individual level, primary sampling unit (psu) level, and state level.
b
All estimates are weighted, scaled on the basis of Asparouhov's method to reduce bias in variance estimates.
c
Variance partition coefficient (VPC) measures the proportion of total variance, which lies at the cluster level. VPCs indicate that the variability in bidi smoking in both waves of GATS were associated with between-state and between-PSU differences, respectively.
d
Know: knowledge about dangers of smoking. OR: odds ratio.
*
p<0.05.