Fun with statistics; why you don't take trial results at face value

[annapurna]

Rich just posted a new article about a comparison of cervical ADR to ACDF. I glanced through the summary and found yet another glaring example of the fun games that can be played with statistics and why it's important to not take these summary results at face value.

The trial statistics from the article are below:

• Artificial cervical disc received by 621 patients
• ACDF received by 592 patients
• Mean ages in three FDA-IDE trials: 42-44
• Gender: 51 percent male, 44 percent female
• Full 2-year follow-up completed by 94 percent of the arthroplasty group, 87 percent of the ACDF group.

Given that this is a human trial and that humans have two genders that you'd hope a skilled surgeon could easily tell apart, you'd expect that the sum of the male and female patients would reach 100%. What happened to the other 5%? Obviously, there was some hiccup where 5% probably didn't have their gender reported or recorded but it does make you question the rest of the statistics and how careful the rest of the study was. The common engineering joke here is "In God we trust; all others bring data."

[Cirobi]

With that kind of statistical silliness, I can't help but wonder if they did actually have some hermaphrodites or cross-gendered folks participate. I'd hope they'd try not to let that happen for the sake of consistency in the study but still.... statistics make me giggle.