Dale Lehman, PhD, Director of the Center for Business Analytics, Loras College
This paper was submitted in the recent SPRINT Data Analysis Challenge held by the New England Journal of Medicine. The challenge was an international competition to investigate the benefits of opening the data from clinical trials – a well-publicized trial examining the benefits and side effects of more aggressively treating people with high blood pressure. There were 143 submissions, and this was the only submission that identified that participants who missed scheduled visits failed to realize the benefits of more intensive treatment of blood pressure. I attribute the identification of this factor to my use of JMP – other software does not easily reveal the fact that nearly 25 percent of the SPRINT participants missed some scheduled visits, and that this group showed no significant benefit from intensive treatment. This group also had increased risk of serious adverse events. Conversely, patients who attended all scheduled visits had even better outcomes than the published effectiveness of intensive treatment in SPRINT, but had increased risk of serious adverse events. JMP also permitted a visual representation of the uncertainties associated with both the primary outcome (heart attacks) and the severe adverse events (mostly resulting from more intensive use of drugs to treat high blood pressure). The importance of missed visits and the graphical representation of uncertainty both demonstrate features of JMP that facilitate better data analysis.
Note: I will present the results in the attached paper, along with a live demonstration of their derivation in JMP. I cannot provide the JMP files, since I am not allowed to share the data from the competition.