Send request to the Database Manager through the Database Request Portal

 

I am going to be abstracting data from charts, when should I reach out?

The best time to talk about a new project is prior to data collection! A well thought-out and structured database makes analyses and management much easier. Dr. Paula Strassle is available to help construct and review your database during her office hours or a meeting can be scheduled. It may also be possible to generate patient lists and abstract data from the EMR through NC TraCS.

Note: If your database is in an unanalyzable format, it will have to be fixed before your project will be added to the queue.

When are Dr. Strassle’s office hours?

You can drop-in to ask questions and get help every Wednesday from 10-11AM and every Thursday from 12-1PM. If you are unable to make either of these times, email Dr. Strassle to set up a meeting.

What national databases are available?

The database manager will maintain publicly available databases (e.g. NIS, SEER) and be available to pull data for analysis purposes. Descriptions of national databases may be found below. If you are interested in using a database not currently listed, please contact the database manager to discuss options for obtaining it.


List of National Databases


I am using a discharge/claims database for my research (e.g. HCUP, Medicare). How do I go about creating code lists to identify my diagnoses and procedures?

Generating code lists are one of the most important things for studies using discharge/claims data. The first thing to do is to determine which types of codes you will need. For example, HCUP captures procedures using ICD codes, but Medicare uses primarily CPT codes. Additionally, if your study crosses October 2015, you will need both ICD-9-CM and ICD-10-CM code lists for your diagnoses and procedures (if applicable).

Once you know what types of codes you will need, look at what previously published research exists. Has another study created a code list for the same variable? Has a code list been validated? (e.g., Charlson comorbidity index components have code lists already) There are also several online resources available to review codes, including a UNC managed resource available through the Cancer Information and Population Health Resource (CIPHR).

If you need both ICD-9-CM and ICD-10-CM codes, first identify all of the ICD-9-CM codes you need, then use the General Equivalence Mappings (GEMs) created by CMS to identify your ICD-10-CM codes. Mappings exist in both forward (match ICD-9-CM codes to ICD-10-CM codes) and backwards (match ICD-10-CM codes to ICD-9-CM codes) directions, and you need to use BOTH code lists to identify the ICD-10 codes you need, as they will give you different results (it doesn’t make sense, I know!) Not including both lists may cause you to underestimate the prevalence of your diagnoses and procedures, and bias your results!

If you have never created a code list for research before, you must attend one of Dr. Paula Strassle’s office hours or set up a meeting to discuss.


I can’t find a database that has all the variables I need. What should I do?

One option is to drop the variable that cannot be obtained in the database you are interested in. It may also be possible to link two (or more) databases together. For example, SEER does not have extensive information on treatment plans and timing and claims data does not have information on laboratory variables and cancer stage, so researchers linked the two (SEER-Medicare linked database). The Carolina Data Warehouse for Health (CDW-H) is also in the process of linking EMR data to insurance claims. It may also be possible link data in-house.


After I submit a request, how long will it take to get my results?

Once your project is approved, how long a request will take depends on the current backlog and how long your specific request will take. New requests are reviewed on the first Monday of every month and will be approved, sent back for revisions (revise and resubmit) or declined. All approved projects (including those currently in the queue) will then be scored using the system below to determine priority. Projects will be worked on in order from highest to lowest total point value.

Score Reason
+2 Analysis will take <5 hours of analyst time to complete
+3 Analysis is in support of an NIH or CDC grant application
+1 Analysis is in support of a pharmaceutical contract (or similar)
+1 Analysis is in support of a QA project with direct benefit to surgery
0 Analysis could be reasonably completed by requester with supervision
-1 Requester did not discuss project with analysts before submitting (e.g. office hours)
0-2 Request has a hard deadline (score varies based on reason and date)
0-2 Request is considered a ‘hot topic’ or novel research question
+1 Request has been waiting in the queue for 6 weeks (+2 for 12 weeks, etc.)

You will be informed of your place in the queue when your project is accepted. Additionally, once an analyst begins working on your request, they will provide an estimated timeframe for how long it will take to get you preliminary results. You can also check the status of your request by emailing your analyst (if you’ve been assigned one) or surgerydataman@med.unc.edu.


My project has been approved, what should I do while I wait for my results?

There are several things you can (and should) do while you wait! Conduct a literature review to pull articles that you may need to write the introduction and discussion of your paper. Create an outline of your manuscript, including dummy tables. Write the introduction. Identify potential target journals and review their submission and formatting requirements.


I need a statistician for a grant proposal, can I use a Dr. Strassle?

Yes, Paula can work on funded research projects, although all effort must be budgeted for in the grant.