Table 2

Change in the monthly PPI dispensing counts, rate of switching to a lower strength PPIs and rates of treatment discontinuation at each intervention point, estimated using ARIMA models adjusted for seasonality

Seasonal ARIMA model specification*Mean monthly dispensings†Level shift‡ from April 2015Level shift from May 2016
% (95% CI)% (95% CI)
Dispensing counts
 PPIs(2,1,0) (0,1,0)12164 3610.6 (−0.7 to 2.0)−1.7 (−2.9 to −0.4)
 High strength(1,1,0) (0,1,0)1228 728−0.2 (−3.1 to 2.7)0.3 (−2.6 to 3.3)
 Standard strength(0,1,1) (0,1,0)12128 1561.2 (0.1 to 2.3)−1.7 (−2.7 to −0.7)
 Low strength(0,1,1) (0,1,0)127481−0.1 (−2.6 to 2.5)−0.8 (−2.9 to 1.4)
 Statins (control)(0,1,1) (0,1,1)12187 4071.0 (−1.1 to 3.3)−1.6 (−3.7 to 0.5)
Switching to lower strength PPI
(0,1,1) (1,0,0)123.0 (−1.6 to 7.7)0.2 (−3.8 to 4.4)
Treatment discontinuation
 PPIs(2,1,0 1,1,0)12−0.1 (−0.4 to 0.2)−0.1 (−0.4 to 0.2)
 Statins (control)(1,0,0) (0,1,0)120 (−0.3 to 0.4)0.3 (0 to 0.6)
  • *ARIMA (p,d,q) x (P,D,Q)12 model where 12 indicates seasonal differencing at 12-month lag.

  • †Mean monthly dispensings in the year leading up to NPS MedicineWise programme in April 2015.

  • ‡A sudden, sustained change for the remainder of the study period.

  • ARIMA, autoregressive integrated moving average;PPI, proton pump inhibitor.