Those are great questions Tony.  I've attached one of the better recent articles on the subject.

Brief personal answers are:

1) Observational tends to be more adequate when indications for treatment are trivial, data are high quality, and patient outcomes are of high precision and are objective.

2) Carefully think through the variables to collect on everyone, collect them with accuracy, make sure they include all indications for surgery, and follow 100% of patients long enough.



On 01/15/2015 12:15 PM, Tony Asher wrote:
>
> Hi Hui and Frank and Ted. I had a quick question.
>
>  
>
> We are meeting with NINDS in February to discuss ways to advance scientific discovery in neurosurgery. Among other topics, we will likely discuss RCT versus observational analyses.
>
>  
>
> I was wondering if either of you had any references that I could suggest to the group regarding the following:
>
>  
>
> 1)      In what experimental situations are RCTS most strongly indicated? What questions lend themselves to observational analyses? What questions might reasonably be addressed by either (depending on the design)?
>
> 2)      What methods can be used to improve the validity and reliability of observational analyses (both  with respect to study design, and also data collection/analysis)? I’m particularly interested in methods that can be applied to registry settings.
>
>  
>
> Thanks in advance for your thoughts!
>
>  
>
> Best
>
>  
>
> Tony
>
>  
>
> Anthony L. Asher, MD, FAANS, FACS
>
> Co-Medical Director, Neuroscience Institute, Carolinas Healthcare System
>
> Director, American Board of Neurological Surgery
>
> Carolina Neurosurgery and Spine Associates