Thanks for posting your source.
The difference between the finasteride and control group reporting these symptoms is only 4 people; a 0.5% difference. Generally from what I've seen, side effects reported at less than 1% are discarded as not statistically significant, but this varies depending on the group size.. Also, "Nervous system or psychiatric adverse event" includes extremely commonly reported symptoms such as dizziness or vertigo.
I did not read this entire report and I have never seen it before. Does it specifically indicate which symptoms were experienced by these patients? A cursory glance over it didn't uncover details, but if these specifics are outlined in the study I will read it (you should know since you did read it)
This is why I was reluctant to post a link. That you would draw your own conclusions based on your misunderstandings. I'll try to clear things up as best I can.
"The difference between the finasteride and control group reporting these symptoms is only 4 people; a 0.5% difference. Generally from what I've seen, side effects reported at less than 1% are discarded as not statistically significant, but this varies depending on the group size."
Aside from the wrong statistic used, when
p is less than 1% the data are NOT discarded. How to get
p is beyond the scope of this discussion. Please do not attempt to come up with
p. Just try to understand what
p means.
P is short for probability. When you see
p, think:
probability that the results occurred by chance. When
p < 0.01 the interaction has reached
statistical significance
At least intuitively you were thinking along the right lines. There is a threshold for what data get accepted and which data get discarded. It's important to note that the threshold is arbitrary. In social sciences, it's usually
p < 0.05. In medicine it's usually
p < 0.01. Critical safety information can get thrown out, even if
p = 0.01. For example, if a drug group is having more Cardiovascular Adverse Events than a Control group, and
p = 0.02, the results are not statistically significant, and there will be no Cardiovascular AE warning. Wouldn’t you, as a consumer, want to know if more people had heart attacks in the drug group than the control group, and there was a 98% probability the results did not occur by chance? Unless it’s a greater than 99% (
p < 0.01), you won’t get a warning. The statistical methods were developed in the 1920’s, are considered outdated, and have always been controversial. Read The Cult of Statistical Significance for more information.
'Also, "Nervous system or psychiatric adverse event" includes extremely commonly reported symptoms such as dizziness or vertigo.’ Yes. This is a problem with ALL adverse events categories, especially “mental” adverse events. They get lumped into one group. “Depression” goes into the same category as memory impairment and disorientation. Very problematic. Wouldn’t you want to know if your pilot or Air Traffic Controller was taking a drug that could cause disorientation?
“I did not read this entire report and I have never seen it before. Does it specifically indicate which symptoms were experienced by these patients?”
No it does not, nor does it indicate the severity of each group’s symptoms. The Finasteride group could be having severe “Nervous System and Psychiatric Adverse Events” including impairment of memory and suicidal ideation and the control group could be reporting normal depressive moods and behaviors, but we’ll never know. If the data are recorded, it’s in Merck’s database. I have no idea if it is. But they are not required to submit it to the FDA.
