William A. Hyman
Professor Emeritus, Biomedical Engineering
Read other articles by this author
Note from the Editor: It is with a sad heart that I report the passing of William Hyman on July 31, 2019. William has been writing for us since 2013 and has posted over 100 articles. I had the pleasure of having lunch with William in New York City once in our years of knowing each other. We had a great relationship over the years through our emails. I will miss his family updates and our banter and stories of our past days in Texas. His daughter notified me along with an article he had written and asked them to deliver to me. Thank you William for all your insight and thoughtful and meaningful posts. Here are his final words to us. – Roberta Mullin
Predicting Intervention Effectiveness
Much of what we discuss here concerns interventions directed at some perceived problem. These types of interventions should be distinguished from new initiatives for which a problem to be solved has not been identified. The intervention might be generated by the internal discovery of a problem, vendor assertion of a problem that they can solve, or government mandate. In either case it would be nice to know if the intervention can actually be expected to accomplish anything. In this regard we can consider a range of probabilities of having a good result.
High – A high probability of a good effect is likely when the intervention directly addresses the problem, especially when the problem can be eliminated without further interventions. This kind of intervention should not rely on procedural changes or training which are generally known to be weak solutions.
Medium – A medium probability of success is likely for an intervention that requires only simple performance changes, is easy to teach and learn, and is easy to put into practice. In addition, there should be a good expectation that was is taught will actually be followed, supported by it being easy to monitor and, if necessary, enforce.
Low – The probability of success will be low if the intervention requires substantial performance changes and is challenging to teach, learn and apply. Under these circumstances it is likely that the desired change will at least sometimes not occur. Performance to the new requirements will be difficult to monitor and enforce and lack of compliance is likely to increase with time.
Poor – There will be a poor probability of success when the intervention will require significant changes in performance and it is very challenging to teach, learn and apply. In this case it is likely that new practices will not take hold, or not last very long. This will be compounded by the desired effect being very difficult to monitor and enforce.
Pointless – Interventions are likely to be pointless if they don’t actually address the problem or there is no identified problem. Additional limitations are when the intervention requires significant and challenging performance changes, and it is unlikely to be well taught and learned, unlikely to actually be deployed, and unlikely to be effectively monitored and enforced.
The best interventions have a clear goal and a clear methodology of how they will work and how they will be implemented. This usually requires effective design, training, actual adoption and monitoring. How many interventions meet these criteria?