Clinical Research Help

[soccerusa]

For all the clinical researchers out there,

I am starting to work on some clinical research with a professor and I am going to be responsible for "building a database" as the professor put it. For people who have done this kind of work before, does this mean just an excel file with patient info or does he mean a full blown database necessitating computer code? The retrospective project will likely involve about 100 patients.

Also, I am meeting with him to go over the project this week. Are there any specific questions you can think of to go over in the meeting? He is a busy guy, so I feel like our face to face time will be limited.

Were there any specific problems you came across that would be beneficial for me to discuss with him upfront?

Thanks for the help!

---

Members don't see this ad.

 

[Amba]

Most likely it will be either collecting patient data in Excel or in a secure online repository like Redcap.

Questions I would ask or discuss:

* Timeframe and goals for the project. Will multiple papers result from it? Authorship order? Opportunity for creative input and to write? Or will you just be collecting data for a long time without a specific research question or hypothesis in mind (which imo is a huge waste of time on your part).

* Expected time to complete the database. Collecting 5 data points per patient is completely different than collecting 40.

* If you know any of the residents or fellows in this person's department, they can often provide more specific advice and tips for working with specific faculty members.

 

[ridethecliche]

You might want to take a few modules of the R programming on coursera. It'll likely pay dividends as you start to organize and analyze data.

 

[tachycore]

Most clinical databases are created with some idea of a study design.

Yes, I'm referring to that 4-week course (or however long medical schools spend nowadays) of epidemiology that you will do or did. And yes, the fancy biostatistics with R programming, STATA, SAS, SPSS, etc. like data organized in certain ways (spreadsheets are easiest but less flexible for data entry / form creation). And they will need you and your mentor to know hypotheses to test beforehand.

If you know your design, then you can look up examples of final products (aka, publications) in the biomedical literature. From there, you can figure out what minimum variables are necessary for your clinical field. Knowing all these issues beforehand will save you considerable trouble when you need to do crunch the number and spit out results.

This is coming from someone who has been involved in several clinical databases. And still works on various forms for research to this day.

 

[qw098]

You might want to take a few modules of the R programming on coursera. It'll likely pay dividends as you start to organize and analyze data.

https://www.coursera.org/course/rprog

Seems like more of a programming course?

• Week 3: Loop functions, debugging tools
  
  
• Week 4: Simulation, code profiling
  

 

[NotAProgDirector]

For 100 patients, Excel would be fine.

 

[Dr.Tao]

First and foremost, get some one that knows something about clinical data on your team STAT!!

Excel is absolutely NOT the option you want to go with for a building a database for clinical research for reasons too numerous to name. At the very least, you should create a database in ACCESS (again, get someone who knows this or you'll be stuck working it out forever).

Third, R while a great statistical programming language for carrying out statistical analysis, it is NOT used for building databases. Build your database in ACCESS then use R to do your statistical analysis. Or if you can find someone with expertise to help you, get them to import your excel spreadsheet into MySQL (database) and go from there.

Signed a Health Data Analyst/Future Clinical Informaticist

 

[ridethecliche]

First and foremost, get some one that knows something about clinical data on your team STAT!!

Excel is absolutely NOT the option you want to go with for a building a database for clinical research for reasons too numerous to name. At the very least, you should create a database in ACCESS (again, get someone who knows this or you'll be stuck working it out forever).

Third, R while a great statistical programming language for carrying out statistical analysis, it is NOT used for building databases. Build your database in ACCESS then use R to do your statistical analysis. Or if you can find someone with expertise to help you, get them to import your excel spreadsheet into MySQL (database) and go from there.

Signed a Health Data Analyst/Future Clinical Informaticist

I meant importing excel databases. We just used to get dumps of data from redcap as excel files and toss them into STATA/R.

Honestly, sigmaplot or graphpad (I think that's what it's called) are great intro stats packages. Sigmaplot produces really good publishable graphs and figures, has a good user interface, and the analysis interface is tops. Highly recommended if you're just looking to analyze some data to present. It does present a few issues when you're trying to set up limitations for certain tests which was an issue for one of the pubs I had and required me to fire up STATA and figure out how to use it well enough to run a few scenarios.

 

[Dr.Tao]

We just used to get dumps of data"

Actually the phrase is "data dumps", but nice try at "flexing" your stats/programming "muscles", LOL!!!

PS-Holla' back when you learn C++ and Hadoop, ROTFL!!!

 

[mcloaf]

Actually the phrase is "data dumps", but nice try at "flexing" your stats/programming "muscles", LOL!!!

PS-Holla' back when you learn C++ and Hadoop, ROTFL!!!

People like you are the reason people like me just end up putting the database in excel and figuring out the stats themselves.

 

[phonyreal98]

While I agree with what TheTao is saying, that doesn't excuse him (or her) for being a bit hubristic with his (or her) posts in this thread.

 

[soccerusa]

Thanks for the help. I had the meeting this week and we are just working in excel. I actually have a very large programming background and I think the level of data collection and analysis we are doing will be easily done in excel.

 

[ridethecliche]

Actually the phrase is "data dumps", but nice try at "flexing" your stats/programming "muscles", LOL!!!

PS-Holla' back when you learn C++ and Hadoop, ROTFL!!!

Sorry. I was tired after dissecting in lab for hours in the evening and posting to try to be helpful instead of stroking my programming prowess on the allopathic section of a medical forum.

 

[Dr.Tao]

While I agree with what TheTao is saying, that doesn't excuse him (or her) for being a bit hubristic with his (or her) posts in this thread.

ROTFL usually contes humor in a post. Unless humor follows a hierarchy around here, Attendings > Fellows > Residents ect. when it comes to adding humor/being humorous?

To the OP, I just spent 3+ months making an excel "database" useful for additional clinical research studies. Sure, excel maybe the "go to" route due to time constraints. But one day you or someone else (around abstract submitting time perhaps?) is going to wish you had put the work in on the front end. Doing so could save you TONS of time in the future, not to mention how easily it is to manipulate the data once you have it in a real database.

Good luck!

 

[ridethecliche]

Especially when redcap is pretty much in use everywhere and is stupid easy to learn.

 

[Dr.Tao]

Especially when redcap is pretty much in use everywhere and is stupid easy to learn.

I'm only aware of a few investigators that used RedCap at my old institution (MD Anderson) and none at my current community academic hospital. But thanks for the info, this is yet another feather to add to my database knowledge cap!

SDN ROCKS for being a source good info!!🙂

 

[ridethecliche]

I'm only aware of a few investigators that used RedCap at my old institution (MD Anderson) and none at my current community academic hospital. But thanks for the info, this is yet another feather to add to my database knowledge cap!

SDN ROCKS for being a source good info!!🙂

REDCap is pretty much the service/tool of choice to hand over to a 'clueless' investigator. The databases are incredibly easy to build and the output is robust. You can also edit just about everything even after the project is live so you can catch a mistake or make a change while still keeping data integrity.