HCLS/ClinicalObservationsInteroperability/IRBApprovalQuestions.html

Questions related to Getting De-identified Patient Data

1. What kind of patient data are we looking for?

At present, we are focused on patient diseases (e.g., hypertension and diabetes). So, we are interested in identifying a cohort which will include variables of interest such as demographics (e.g., age, gender), the medical condition itself, any symptoms and observations, any procedure performed and/or medication administered to the patient, historic events etc. The idea is that the variables of interest should help us understand why the patient is suffering from a particular disease in the first place. For example, co-morbidity is a good indicator for cases related to hypertension. Additionally, for our initial study, patient consent is not mandatory as long as the data is not published and de-identified.

2. What are we going to do with the patient data?

The objective of this study is to enable interoperability between applications for patient recruitment and electronic medical records. Typically, the patient data items can be represented using detailed clinical models (DCM)and clinical protocols and queries using study data tabulation model (SDTM). Consequently, there is need to specify a set of mappings between the elements of these models (possibly supplemented with existing ontologies and vocabularies such as HL7 RIM) that would enable protocol queries to be translated and evaluated against the patient data. Such a functionality can be achieved using existing semantic Web standards (e.g., RDF, OWL, SPARQL) and tools and technologies (e.g., Protégé, Jena, OpenLink Virtuoso).

3. What will be outcome of our investigation?

The scope of this initial study is quite limited. We are only focusing on a few patient diseases and enabling interoperability between a handful of DCM and SDTM models. Additionally, we are concentrating on a limited number of clinical trial protocols. However, we believe that this exercise will provide us with insights, experience and research questions that will have to be addressed by the COI community. Furthermore, we believe such a study will help us understand and develop a set of best practices/guidelines that are the key to enabling multi-domain interoperability and further demonstrate the applicability of semantic Web techniques in the healthcare domain.

4. What is the study population?

Describe the study population in terms of age, gender, diagnosis and other variables of interest.

5. What is the data to be collected?

Some data items listed on the application form are:

1. Administrative Data
   1. Billing data
   2. Coded encounter data (diagnoses, procedures, dates)
   3. Demographic data (age, gender, vital status)
2. Health/Medical
   1. Allergies
   2. Discharge Summaries
   3. Doctors Orders
   4. History/Physical
   5. Immunizations
   6. Medications List
   7. Office/Clinic notes
   8. Operative/Procedure notes (e.g., endoscopy)
   9. Pharmacy
   10. Problem List
3. Health Medical Reports/Results
   1. Blood Bank
   2. Laboratory
   3. Pathology
   4. Radiology
4. Sensitive/Personal Information
   1. HIV Status
   2. Mental Health
   3. Reproductive History (e.g., abortions)
   4. Sexual Behavior/Sexually Transmitted Diseases
   5. Substance Abuse (Drug or Alcohol Abuse)