Scenario | Possible interpretation of the result | Estimated change* |
Scenario 1, only data after launch in the intervention area | The outcome of interest appears to have decreased over the postlaunch time period | Not possible to estimate a change in the outcome |
Scenario 2A, first and last time point of the intervention period | There appears to have been an increase in the outcome from the prelaunch to postlaunch period. Extrapolating the observed values over the entire 15 months of intervention suggests that the new intervention had increased the outcome by 37.6 units ((44.9–42.4)x15) | 37.6 |
Scenario 2B, disaggregated change from starting period | The outcome of interest appears to have decreased over time from the prelaunch time period, with an estimated change of −120.1 units over the period ((34.4–42.4)x15) | −120.1 |
Scenario 3A, simple average of historical intervention area data | There appears to have been little change from the prelaunch to postlaunch periods in the outcome, with the average value going from 35.1 to 35.4 ((35.4–35.1)x15) | 4.9 |
Scenario 3B, matched preintervention and postintervention | There appears to have been little change from the prelaunch to postlaunch periods in the outcome, with the average value going from 35.1 to 35.4. However, it appears from the data that there was an increasing trend in the outcome before the intervention and a decreasing trend afterwards ((35.4–35.1)x15) | 4.9 |
Scenario 4A, comparison of averages postintervention in control and intervention areas | Compared with the control area the intervention area had a lower average level of the outcome after the launch of the intervention | −146.0 |
Scenario 4B, matched postintervention control and intervention area | Compared with the control area, the intervention area had a lower average level of the outcome after the launch of the intervention. The control area appeared to have a flat trend in the outcome over the postlaunch period compared with a decreasing trend in the intervention area ((35.4–45.1)x15) | −146.0 |
Scenario 5, ITS analysis of intervention area | Compared with the prelaunch intervention area the postlaunch saw a decrease in the trend over time in the outcome, from positive to negative, which was statistically significant. See the online supplementary appendix for regression | −258.8 |
Scenario 6, ITS analysis of control and intervention area | Both control and intervention areas saw a shallowing of the trend over time. The intervention area saw a greater decrease in the trend, being negative compared with the relatively flat trend in the control. This difference was statistically significant. The control area was found to be a match to the intervention area in the prelaunch period (the regressions lines are aligned). See the Supplementary Appendix for regression | −146.0 |
*Negative values indicate that the new service reduced the outcome.
ITS, interrupted time series.