Clinical Research without background in Statistics?

[Chris Knight]

---

Members don't see this ad.

How can one best overcome lacking knowledge of statistics in order to be effective in clinical research? any recommendations for websites/brief books with that can give someone a sturdy base of knowledge?

Despite extensive research in developmental genetics in undergrad and a molecular biology major, I managed to make it all the way to med school without a drop of knowledge in statistics (well maybe a drop, I know enough to understand what most articles are talking about, though admittedly in the dark as to how some come to their conclusions, and surely I can't do such analysis myself). Somewhere along the line I was misdirected and ended up going rather far in my math coursework, but never touched statistics. One of my biggest regrets of undergrad.

I will be taking intro to epidemiology and intro to biostatistics courses come spring, but I'd like to get involved in clinical research beforehand. So much clinical work involves statistical analysis, I really am scared at this point to get involved with a project with how basic my knowledge is. Maybe I'm needlessly worrying and theres more than enough for me to do before I ever start crunching numbers, but I'd like the piece of mind. Advice?

 

[bubbachuck]

the textbook we used in MS1 is Rational Medical Decision Making by Goutham Rao for our biostatistics class. It seems decent and it's definitely clinically oriented. I hear if you're in AMSA you can get a 5% discount on Amazon.com.

 

[angel80]

You don't need to understand statistics completely to do research. All hospitals/research centers have a biostats department. You submit your data to them, and they crunch the numbers. You just need basic fundamental knowlege, which doesn't take too long to learn.

 

[fantasty]

You don't need to understand statistics completely to do research. All hospitals/research centers have a biostats department. You submit your data to them, and they crunch the numbers. You just need basic fundamental knowlege, which doesn't take too long to learn.

Caveat: Involving them (the biostatisticians) *before* the study will make you (and them) a lot happier. If poor data is collected, the statisticians cannot magically make it better.

 

[RxnMan]

Caveat: Involving them (the biostatisticians) *before* the study will make you (and them) a lot happier. If poor data is collected, the statisticians cannot magically make it better.

Not just that, but the study design will tell you what stats should be calculated. That is, when you sit down to formalize your hypothesis, you should know what comparisons you want to do. Those comparisons fit certain criteria which will determine the most appropriate test to apply. (Sometimes something new will occur to you as you go through the study, but this won't change the fundamental study question.)

Sometimes folks will run a series of haphazard experiments and then turn their results in to the statistician to see if there's something recoverable.

I've heard this referred to as the difference between fishing and going hunting.

With respect to the OP - It might be easier to get involved on someone else's clinical project for right now. Most folks will jump in on a resident's project and collect a bunch of data for them (caveat: this level of involvement generally does not get your name on the by line). That'll give you a good idea of what's going on and you'll have time to pick up the stats that they'll be planning on using. By the time the next project comes along, you'll be up to speed with the stats and the research methods they use.