Skip Navigation
AHRQ--Agency for Healthcare Research and Quality: Advancing Excellence in Health Care

Topic Suggestion Description

View Topic Suggestion Disposition (PDF) 178 kB

Date submitted: April 09, 2010

Briefly describe a specific question, or set of related questions, about a health care test or treatment that this program should consider.
Invention and coding of data analysis methods generally useful for making valid claims from large observational data sets.
Does your question include a comparison of different health care approaches? (If no, your topic will still be considered.)
If yes, explain the specific technologies, devices, drugs, or interventions you would like to see compared:
The methods and software would be designed so that valid comparisons between different methods could be made for subgroups of patients.
What patients or group(s) of patients does your question apply to? (Please include specific details such as age range, gender, coexisting diagnoses, and indications for therapy.)
The methods would allow for comparisons to be made within subgroups of patients for all medical conditions.
Are there subgroups of patients that your question might apply to? (For example, an ethnic group, stage or severity of a disease.)
Subgroups would be identified dynamically as the subgroups that differ most in reaction to a specific treatment or medical process. Specific subgroups could be analyzed or the data could be used to point to specific subgroups as responding differently to treatments.
Describe the health-related benefits you are interested in. (For example, improvements in patient symptoms or problems from treatment or diagnosis.)
I'm interested in building tools that researchers can use to better find and understand patterns in observational data bases.
Describe any health-related risks, side effects, or harms that you are concerned about.
Treatment efficacy and side effects can be identified within subgroups of patients quickly and efficiently.

Appropriateness for EHC Program

Does your question include a health care drug, intervention, device, or technology available (or likely to be available) in the U.S.?
Which priority area(s) and population(s) does this topic apply to? (check all that apply)
EHC Priority Conditions (updated in 2008)
AHRQ Priority Populations
Federal Health Care Program
  • Medicaid
  • Medicare
  • State Children's Health Insurance Program (SCHIP)


Describe why this topic is important.
Large complex observational data sets will be used to identify patterns useful for patients, physicians and other medical decision makers. It is important that these patterns be valid (not just chance and not just due to bias) so that the best decisions are made on patient care.
What specifically motivated you to ask this question? (For example, you are developing a clinical guideline, working with a policy with large uncertainty about the appropriate approach, costly intervention, new research you have read, items in the media you may have seen, a clinical practice dilemma you know of, etc.)
I am an expert statistician and I would like the best technology to be available to people doing the work of pattern identification. I know all too well that it is easy to falsely identify patterns.
Does your question represent uncertainty for clinicians and/or policy-makers? (For example, variations in clinical care, controversy in what constitutes appropriate clinical care, or a policy decision.)
If yes, please explain:
Chance variation can lead to false claims. For example, two studies did not support the claim that Type A personality leads to heart attacks. The original claim appears to be a false positive claim. Sometimes very sick patients are "channeled" to a particular treatment so that treatment may appear to be dangerous. Clinicians and policy makers should be acting on the best data analysis possible.

Potential Impact

How will an answer to your research question be used or help inform decisions for you or your group?
Use of fast and valid data analysis methods will broadly impact all medical decision making.
Describe the timeframe in which an answer to your question is needed.
Data analysis methods should be selected/invented and prototyped in computer code and tested on large observational data bases. Analysis systems should be alpha and beta tested. The process should take 18 months. It could be compressed to a one-year project.
Describe any health disparities, inequities, or impact on vulnerable populations your question applies to.
This project would apply to all populations. It would be specifically aimed at subgroup analysis.

Nominator Information

Other Information About You: (optional)
Please choose a description that best describes your role or perspective: (you may select more than one category if appropriate)
  • Researcher
Are you making a suggestion as an individual or on behalf of an organization?
Please tell us how you heard about the Effective Health Care Program
Reading about Comparative Effectiveness Research