Table 1

Effects of India’s National Tobacco Control Programme (NTCP) on bidi consumption (Consumer Expenditure Survey 1999–2000; 2004–2005; 2011–2012 pooled data)

Part 1: Prevalence of households reporting bidi consumption, adjusted OR (95% CI)
Model 1Model 2Model 3
n=341 975n=270 265n=107 291
Constant0.49 (0.15 to 1.55)0.39* (0.31 to 0.50)0.59* (0.20 to 1.74)
NTCP indicator0.99 (0.80 to 1.23)1.00 (0.81 to 1.24)0.44 (0.16 to 1.21)
Time indicator, t2 0.70* (0.65 to 0.74)0.71* (0.67 to 0.76)0.55* (0.49 to 0.63)
Time indicator, t3 0.74* (0.69 to 0.79)0.76* (0.70 to 0.81)0.62* (0.53 to 0.73)
Interaction term, DID10.99 (0.77 to 1.26)0.96 (0.75 to 1.23)1.21 (0.93 to 1.58)
Interaction term, DID21.03 (0.78 to 1.35)1.01 (0.76 to 1.33)1.18 (0.88 to 1.57)
Effects attributable to NTCP1.03 (0.84 to 1.28)1.04 (0.84 to 1.30)0.97 (0.75 to 1.24)
Part 2: Log-transformed monthly consumption of bidi sticks per person, conditional on the households reporting bidi consumption, adjusted coefficient (95% CI)
Model 1Model 2Model 3
n=86 818 n=69 366 n=25 767
Constant4.24* (4.09 to 4.39)3.24* (3.10 to 3.38)4.17* (3.96 to 4.38)
NTCP indicator0.03 (-0.10 to 0.16)0.03 (-0.10 to 0.16)−0.24* (−0.46 to -0.01)
Time indicator, t2 −0.11* (−0.16 to -0.07)−0.11* (−0.16 to -0.06)−0.15* (−0.24 to -0.05)
Time indicator, t3 −0.37* (−0.44 to -0.31)−0.36* (−0.43 to -0.29)−0.51* (−0.62 to -0.40)
Interaction term, DID1−0.04 (-0.23 to 0.13)−0.05 (-0.23 to 0.13)0.002 (-0.19 to 0.19)
Interaction term, DID20.02 (-0.13 to 0.19)0.02 (-0.14 to 0.19)0.10 (-0.09 to 0.30)
Effects attributable to NTCP0.07 (-0.13 to 0.28)0.07 (-0.13 to 0.28)0.10 (-0.14 to 0.34)
  • Explanation of variables:

  • NTCP indicator equals to 1 for households residing in an NTCP district, 0 otherwise.

  • t2 equals to 1 for survey year 2004–2005, 0 otherwise.

  • t3 equals to 1 for survey year 2011–2012, 0 otherwise.

  • DID1 was the interaction between NTCP indicator and t2.

  • DID2 was the interaction between NTCP indicator and t3.

  • Effects attributable to NTCP was calculated as the difference of DID2–DID1.

  • These three different types of control groups were modelled against households residing in an NTCP district.

  • Model 1: control group included households residing in a non-NTCP district from any state.

  • Model 2: control group included households residing in a non-NTCP district situated in a state where some districts were NTCP-implemented sites.

  • Model 3: control group included households residing in a non-NTCP district located in a state with no NTCP activities in any of its districts.

  • All regression models were adjusted for demographic and socioeconomic status of the households including sector (rural/urban); size (≤5/ >5 members); proportion of members in each age-group (0–4, 5–14, 15–29, 30–59, ≥60); proportion of females/males; proportion of members in each educational level (illiterate, primary, middle, secondary, higher secondary, graduate and above); religion (Hindu/Muslim/Christian/others); caste (scheduled tribe/scheduled caste/other backward class/others); employment type (self-employed/regular labour/casual labour/others); and wealth quintile (poorest/poor/middle/rich/richest); and state level fixed effects.

  • *P<0.05.

  • DID, difference-in-differences.