Table 1

Estimated percentage-point changes in mean cigarette consumption per day as a function of e-cigarette use and NRT use among smokers from November 2006 to March 2016, based on ARIMAX models

All users of nicotine replacementOnly daily users of nicotine replacement
Percentage change per 1 % change in the exposure (95%  CI) P values
Any current use of e- cigarettes
(immediate impact)
−0.011 (−0.025 to 0.002)
0.097
−0.012 (−0.026 to 0.002)
0.091
−0.010 (−0.024 to 0.004)
0.149
−0.011 (−0.026 to 0.003) 0.130
NRT use for harm reduction
(immediate impact)
0.012 (−0.028 to 0.053)
0.546
0.015 (−0.026 to 0.055)
0.475
0.003 (−0.019 to 0.025) 0.7940.005 (−0.017 to 0.027) 0.672
Mass media expenditure
(immediate impact)
<0.001 (−0.001 to 0.001)
0.984
<0.001 (−0.001 to 0.001) 0.880
Total percentage change due to the exposure (95% CI) P values
Smoking ban
(pulse effect)
0.015 (−0.070 to 0.101)
0.724
0.013 (−0.072 to 0.099) 0.756
Increase in age-of-sale
(pulse effect)
−0.041 (−0.126 to 0.044)
0.342
−0.043 (−0.128 to 0.042) 0.324
Move to local authority control
(pulse effect)
−0.019 (−0.105 to 0.067)
0.662
−0.027 (−0.112 to 0.058) 0.533
Licensing for NRT for harm reduction
(pulse effect)
0.021 (−0.067 to 0.110)
0.639
0.020 (−0.069 to 0.109) 0.661
NICE guidance on harm reduction
(pulse effect)
−0.024 (−0.109 to 0.061)
0.578
−0.028 (−0.114 to 0.057) 0.512
Best fitting modelARIMAX(0,1,1)(0,0,0)12 ARIMAX(0,1,1)(0,0,0)12 ARIMAX(0,1,1)(0,0,0)12 ARIMAX(0,1,1)(0,0,0)12 ARIMAX(0,1,1)(0,0,0)12 ARIMAX(0,1,1)(0,0,0)12
Non-seasonal AR p valueNANANANANANA
Non-seasonal MA p value<0.001<0.001<0.001<0.001<0.001<0.001
Seasonal AR p valueNANANANANANA
Seasonal MA p valueNANANANANANA
R2 0.650.650.660.650.640.66
Bayes factor e-cigarette (0.009 (0.1))2.44 (0.46)2.68 (0.55)1.95 (0.35)2.12 (0.41)
Bayes factor NRT (0.009 (0.1))0.77 (0.14)0.74 (0.13)0.69 (0.09)0.63 (0.08)
  • An AR(1) means that the value of a series at one point in time is the sum of a fraction of the value of the series at the immediately preceding point in time and an error component; an MA(1) means that the value of a series at one point in time is a function of a fraction of the error component of the series at the immediately preceding point in time and an error component at the current point in time.

  • AR, autoregressive; ARIMAX, Autoregressive Integrated Moving Average with Exogeneous Input; MA, moving average; NA, not applicable; NICE, National Institute for Health and Care Excellence; NRT, nicotine replacement therapy.