Susan Harris and colleagues have described a clinical quality improvement (CQI) initiative to reduce the rates of inappropriate induction of labour.1 They claim that their initiative was associated with a sustained reduction in induction rates and recommend that “similar projects be undertaken at other institutions.” We are unconvinced that their data support these conclusions. Specifically, the authors provide only descriptive data without statistical testing. We reanalysed the data using time-series regression models, which allow assessment of and adjustment for preintervention time trends.2,3 Although our reanalysis has limited statistical power owing to the small number of data points, we found a decreasing trend in induction rates before the intervention (0.45% decrease per 6 months, p = 0.10) and no evidence of a continuing trend after the intervention (0.11% decrease, p = 0.64). However, there was evidence of an overall shift in pre- to post-intervention rates (absolute reduction of 2.6% in the 6 months following the intervention, p = 0.06). This could be due to a small intervention effect, although we are uncertain of its clinical significance. We invite the authors to consider conducting a more powerful time-series analysis by disaggregating their data into shorter intervals that still allow stable point estimates of performance.
The authors state that their CQI initiative was “very time-consuming,” representing “a significant cost to the institution.” Hospitals have limited resources to spend on quality improvement. There are substantial opportunity costs if hospitals adopt unproven methods. If we are to generate a robust evidence base for quality improvement activities, we should demand that quality improvement strategies be evaluated with the same scientific standards that are used to evaluate any clinical intervention. This paper fails to provide compelling evidence that CQI works or provides good value for money. Further evaluation is required before widespread adoption of CQI can be recommended.
References
- 1.
- 2.
- 3.