"Just barely statistically significant" but the MLE is a 6x reduction in hospital visits. There's huge potential here, but our ability to know has been hampered by study size.
That's not quite the right takeaway. That 6x reduction is dependent on the p-value, which in turn is dependent on the sample size. The larger the sample, the smaller the p-value, and that 6x reduction will likely shrink as well. It's extremely rare for a treatment in medicine to have a 6x benefit.
> That 6x reduction is dependent on the p-value, which in turn is dependent on the sample size
No, the 6x reduction MLE is -not- dependent on the p-value, and this shows a fundamental misunderstanding of the statistics. That said, the overall distribution of possibilities skews on the "lesser effect" side, so if you had to guess which way the estimate would move with another trial, you're more likely to get a smaller effect than a larger one.... 6x is still the most likely single value.
> It's extremely rare for a treatment in medicine to have a 6x benefit.
All kinds of drugs and treatments for acute disease have a > 6x benefit. We have plenty of conditions that have a horrible prognosis untreated and great outcomes treated.
Have a bad bacterial infection? Odds are abx will clear that right up and you'd die otherwise. Active TB has a ~60% death rate within 5 years untreated, while with normal treatment regimens only has a relapse (not death) rate of a few percent in that 5 year period. Most vaccines have high efficacies. Standard treatments can prevent the vast majority of malaria deaths.
