This document contains material provided by the American Academy of Ophthalmology about the IRIS Registry (Intelligent Research in Sight).

Transcription

1 Data Studio November 1, 2017 This document contains material provided by the American Academy of Ophthalmology about the IRIS Registry (Intelligent Research in Sight).

2 American Academy of Ophthalmology IRIS Registry (Intelligent Research in Sight) Analytics Teams Background The American Academy of Ophthalmology IRIS Registry (Intelligent Research In Sight) is the nation's first electronic health record-based registry devoted to eye care. The IRIS Registry advances scientific knowledge and insight through its aggregated, de-identified data on millions of patient encounters. The IRIS Registry database is comprised of de-identified data extracted from participating ophthalmology practices electronic health record systems (EHRs). There are over 13,000 clinicians in ophthalmology practices registered to electronically contribute their data to the IRIS Registry. The world s largest specialty clinical data registry, IRIS Registry, has data on over 148 million patient visits from 37 million unique patients, growing daily. IRIS Registry s database can help to inform natural history of diseases, prevalence of rare diseases, practice patterns, diffusion of technology, comparative effectiveness and more, all in the real-world setting. IRIS Registry represents a remarkable opportunity for knowledge discovery, and for collaborative studies. The IRIS Registry database is comprised of data from participating clinicians EHRs, and data elements include: patient demographics, patient medical and ocular history, clinical examination findings, diagnoses, procedures and medications. EHR data is stripped of all unique patient identifiers when it is uploaded into IRIS Registry for privacy purposes. The participating physicians own their own data. The Academy owns the de-identified, aggregated data and can analyze these data. The database has the same legal protections and safeguards from legal discoverability as do hospital-based peer review processes. The IRIS Registry is a unique resource for the evaluation of critical questions in ophthalmology. IRIS Registry contains data on the following disease conditions to provide a few specific examples: The rate of endophthalmitis after cataract surgery; the visual outcome of patients with endophthalmitis; and specific risk factors associated with developing endophthalmitis The rate of endophthalmitis after anti-vegf injections across different agents, and the comparative effectiveness of these agents based on visual acuity outcomes The prevalence of myopic choroidal neovascularization in the United States, based on both diagnostic code information and clinical data The treatment patterns for patients with myopic choroidal neovascularization, including the visual acuity outcomes for those treated and those not treated Three peer-reviewed papers have already been published from IRIS-based research within the past year, including a Jackson Lecture by Anne Coleman, PhD, MD in the American Journal of Ophthalmology and one on pathologic myopia and macular disease in Ophthalmology. Project Goals The Academy very much desires that this massive database be made available to scientific investigators to further population health in ophthalmology and blindness prevention. While the Academy does have staff and resources for IRIS Registry studies and data analytics, these are limited. Therefore, the Academy is seeking to create several IRIS Registry Analytics Teams that will be dedicated to performing

3 Data Elements Patient Birth Year Sex Race Hispanic or Latino Ethnicity Practice Zip Code Insurance Past medical history Past ocular history including diagnoses and procedures Social history including tobacco use Immunizations Medications Diagnoses - systemic and ocular by ICD code (or SNOMED code) Procedures - diagnostic and treatment by CPT code or services by HCPCS code Complications of treatment - by ICD code or if requiring another procedure, by CPT code Clinical examination findings, including visual acuity, intraocular pressure, disease severity, etc. Referral Planned next visits

4

5 Additional details about a few elements

6 Specific Elements: definitions, limitations, issues Date of Service: for privacy reasons, AAO is providing Date of Service by year and week, not by actual service date. i.e. the date of 1/2/2015 will be reported as Week:1, Year: Dates of service can refer to an encounter date or a procedure date and are found in different tables: The EncounterDate, found in the Tblencounter table, refers to the date of an encounter or visit. The EffectiveDate^, found in the PatientProcedure table, refers to the date of a procedure. The StartDate^, found in the PatientMedications table, refers to the date at which a medication was prescribed. The StopDate^, found in the PatientMedications table, refers to the date at which a medication was to be stopped. I.e. if a medication was prescribed with a 30-day supply, the StopDate should be 30 days after the StartDate. (Dates related to diagnoses are explained in the Diagnosis section below.) Diagnosis: The IRIS Registry contains two types of diagnoses: ophthalmic diagnoses and systemic diagnoses. Ophthalmic diagnoses are more reliable and available since the registry is comprised of ophthalmic practices geared toward capturing ophthalmic data. Diagnoses can be found in the PatientProblem table under the variable name PracticeCode*. Diagnoses may be captured with ICD-9 or ICD-10 codes. At times, a patient with one diagnosis may be coded with both ICD-9 and the ICD-10 codes for that particular diagnosis. Therefore, it is often necessary to convert all diagnoses to either ICD-9 or ICD-10 before counting unique patients/unique eyes etc. with particular diagnoses. The onset date, or date at which a diagnosis first occurred, is found in the PatientProblemHistory tables as the Diag_Date^ variable. Diag_date, or onset date, is often not populated in IRIS Registry data, in which case one can use DocumentationDate as a proxy. DocumentationDate^ refers to the date at which the diagnosis was documented by the practice. This variable is more populated than the onset date variable (diag_date). However, note that the onset of a condition may have occurred before it was documented, meaning that using the documentationdate does not always accurately capture whether a condition existed by a certain date. Again, it can be used as a proxy.

7 Additionally, note that the diagnostic PracticeCode is located in the PatientProblem table while the diag_date and documentationdate variables are located in the PatientProblemHistory table. Therefore, in order to obtain both a patient s diagnosis and the date of onset or documentation, one must link these two tables using the PatientProblemUID code. IOP: IOP measures are found in the Tblencounter tables, listed separately for right and left eyes with the following variable labels: IOPLt (left eye IOP measure) and IOPRt (right eye IOP measure). Patients with both eyes are listed separately with a right and left measure; there is no both eye variable in this table. Practices enter IOP data manually in many EMR systems, so recorded IOP values often contain letters or symbols that are extraneous and cannot be used in conjunction with some SAS functions. Therefore, the IOP values must be cleaned up and transformed in SAS into numeric values to be of use to research questions. IOP values of -1 must be excluded because they are not actual values; -1 signifies that our vendor was not able to map the value from the patient s medical record. Note that some practices have unusually high proportions of patients with IOP values of 1, suggesting that IOP data from these practices may not be reliable. Laterality: Eye laterality is captured in several places, in several different ways. Diag_eye in the PatientProblem table identifies the diagnosed eye and can have the value of 1 (right), 2 (left), 3 (both left and right), or 4 (unspecified eye). Unfortunately, a large proportion of diagnosis data has unspecified eyes. In some cases, the researcher can make assumptions based on other information in the patient record to identify the diagnosed eye. In other cases, the unspecified eye cannot be determined. Proc_eye in the PatientProcedure table identifies the procedure eye and can have the value of 1 (right), 2 (left), 3 (both left and right), or 4 (unspecified eye). Unfortunately, a large proportion of procedure data has unspecified eyes. In some cases, the researcher can make assumptions based on other information in the patient record to identify the procedure eye. In other cases, the unspecified eye cannot be determined. For both diag_eye and proc_eye, note that for most analyses, the value of 3 (both eyes) should be transformed to reflect a right eye (1) and a left eye (2) in order to accurately count the number of eyes. It is also important to transform from both to left and right eyes to provide consistency among sample patients; some EMR records may not use the both (3) value and patients with diagnoses in both eyes diagnosed/with a procedure etc. will have been recorded separately as a right (1) and left (2). Note that laterality is built into some ICD-10 codes (from PracticeCode in the Problem tables), usually as the 7 th position of a code, but sometimes other positions of the ICD-10 codes. - For example, a patient with Diabetes Mellitus due to underlying condition with mild diabetic retinopathy with macular edema is identified by the E ICD-10 code. E refers to that diagnosis in the right eye; E refers to that diagnosis in the left eye, E refers to a bilateral diagnosis. 2

8 - On the other hand, Adherent Leukoma is identified by the ICD-10 code: H17.0 and laterality is identified in the 5 th position. The code H17.01 refers to adherent leukoma in the right eye; H17.02 refers to adherent leukoma in the left eye; H17.03 refers to bilateral adherent leukoma. Medications: The medication table contains information on both non-prescription products (or over the counter drugs), medications prescribed by an ophthalmologist in his/her practice, and possibly medications prescribed by other providers outside of the ophthalmology practice (medications patients have or are taking). If Rx_Flag is equal to 1, then we can assume the medication record is a drug prescribed by a physician in an ophthalmology practice. The prescribing provider NPI & corresponding Practice ID are also known for these records. We have three main medication source vocabularies: RxNorm, NDDF, and GPI. Generic Product Identifier, GPI, is a lesser used vocabulary representing a small sample of our data and mainly utilized by one EMR system (SRS). We extract most useful information from RxNorm and NDDF/NDC source codes. Two crosswalks enable us to convert a numeric medication code to its corresponding brand name or active ingredient (and sometimes therapeutic class). We also have SIG, route, product form, brand name, generic name, and dose dispensing quantity information. However, this type of data is not as well populated because not all EMR systems consistently provide this textual information. Therefore, it is best to apply external crosswalks (provided from the FDA or NLM) to classify that data. Procedures: The PatientProcedure tables capture procedures with the PracticeCode** variable. Procedures are identified by CPT or HCPC codes. For intravitreal injections, one should generally proceed in two steps: first pull the CPT code 67028, which identifies that there was an injection procedure, and second, separately pull the HCPCs codes which identify the drug injected (e.g. anti-vegf or injected steroids etc.) Visual Acuity (VA): Visual acuity measures include VA type and VA method and are found in the VisualAcuity tables. VA Type can include Best Corrected or Corrected (meaning that glasses or contacts were used), Not corrected, and Unspecified. The VisualAcuity tables contains separate variables for each of these types, and by eye (left or right; both is not an option in this table). For example, the table contains VisualAcuity_Left_Value_BC to refer to the Best Corrected value reported for a patient s left eye. VA Method can include 14 variations of Distance and Near methods. Distance is the more commonly used method. VA Values are mostly reported in Snellen Feet (20/20, 20/40, etc.) which generally should be converted into a numeric LogMAR value (-0.30 through 3.0) for use in analyses. NOTE: There are a small percentage of records that are documented using Jeager Standard which should only be measure with a 3

9 Near Method. However, AAO does not currently know how to accurately report these records and therefore cannot convert these to LogMAR. The VisualAcuity tables also include the date when the VA measure was taken. Dates are reported by eye (left or right; both is not an option in this table) and by type (best corrected, uncorrected or unspecified). For example, the variable VA_Left_Date_Not_Correct refers to the date at which a patient s VA was measured in the left eye using the Not Correct type ^ Because these dates are currently provided by FIGMD in character fields, they must be converted to a date field in order to execute some SAS functions such as identifying/finding procedures that occurred during a certain period. * PracticeCode in the Problem table should not be confused with the PracticeCode variable in the Procedure table which captures procedure generally CPT or HCPCS- codes. The variable names are the same, but the contents of the variables differ. ** PracticeCode in the Procedure table should not be confused with the PracticeCode variable in the Problem table which captures diagnosis generally ICD-9 or ICD-10 codes. The variable names are the same, but the contents of the variables differ. 4

10 American Academy of Ophthalmology IRIS Registry Data Dictionary Patient Demographics Element ID: 2050 Name: Year of Birth Coding instructions: Indicates the patient's year of birth. Short name: YOB Format: Date Format Value: yyyy Element ID: 2060 Name: Sex Coding instructions: Indicates the patient s sex at birth. Short name: Sex Element ID: 2070 Name: Race-White Coding instructions: Indicates if the patient is White. Short name: RaceWhite Target Value: The value on arrival at this facility Selections: Code Selection Text Definition 0 1 White (Race) Date: 1 Version: 1.0

11 [Type here] American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 2071 Name: Black/African American Coding instructions: Indicates if the patient is Black/African American. Patient Demographics Short name: RaceBlack Selections: Code Selection Text Definition 0 1 Black/African American (Race) Element ID: 2072 Name: Asian Coding instructions: Indicates if the patient is Asian. Short name: RaceAsian Selections: Code Selection Text Definition 0 1 Asian (Race) Date: 2 Version: 1.0

12 American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 2073 Name: Race-American Indian/Alaskan Native Patient Demographics Coding instructions: Indicates if the patient is American Indian or Alaskan Native. Short name: RaceAmIndian Target Value: The value on arrival at this facility Selections: Code Selection Text Definition 0 1 Race-American Indian/Alaskan Native (Race) Element ID: 2074 Name: Race-Native Hawaiian/Pacific Islander Coding instructions: Indicates if the patient is Native Hawaiian/Other Pacific Islander. Short name: RaceNatHaw Target Value: The value on arrival at this facility Selections: Code Selection Text Definition 0 1 Race-Native Hawaiian/Pacific Islander (Race) Date: 3 Version: 1.0

13 [Type here] American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 2075 Name: Race-Other Patient Demographics Coding instructions: Indicates if the patient having another race apart from above mentioned. Short name: RaceOther Selections: Code Selection Text Definition 0 1 Other (Race) Element ID: 2076 Name: Hispanic or Latino Ethnicity Coding instructions: Indicates if the patient is Hispanic or Latino ethnicity as determined by the patient/family. Short name: HispOrg Selections: Code Selection Text Definition 0 1 Hispanic or Latino Ethnicity Date: 4 Version: 1.0

14 American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 3020 Name: Insurance- Private Health Insurance Coding instructions: Indicates if the patient has private health insurance. Patient Demographics Short name: InsPrivate Element ID: 3022 Name: Insurance-Medicaid Coding instructions: Indicates if the patient is insured by Medicaid. Short name: InsMedicaid Element ID: 3023 Name: Insurance-Military Healthcare/Department of Defense/Tricare Coding instructions: Indicates if the patient has military health care. Short name: InsMilitary Date: 5 Version: 1.0

15 [Type here] American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 3028 Name: Insurance-Medicare (Fee for service) Patient Demographics Coding instructions: Indicates if the patient is insured by Medicare (fee for service). Short name: InsMedicare_FeeforSer Element ID: 3029 Name: Insurance-Medicare (Advantage care) Coding instructions: Indicates if the patient is insured by Medicare (managed care/hmo) Short name: InsMedicare_MngdADGCare Element ID: 3030 Name: Insurance-Blue Cross/Blue Shield Coding instructions: Indicates if the patient is insured by Blue Cross/Blue Shield. Short name: InsBlueCrossBlueShield Date: 6 Version: 1.0

16 American Academy of Ophthalmology IRIS Registry Data Dictionary Patient Demographics Element ID: 3031 Name: Insurance- Other Government / Indian Health Service/ State Local government Coding instructions: Indicates if the patient is insured by other government Short name: InsOtherGov Element ID: 3032 Name: Insurance-Department of Corrections Coding instructions: Indicates if the patient is insured by Department of Corrections Short name: InsDeptOfCorr Element ID: 3033 Name: Insurance-Managed Care Unspecified Coding instructions: Indicates if the patient is insured by Managed Care Unspecified. Short name: InsUnSpecified_MngdCare Date: 7 Version: 1.0

17 [Type here] American Academy of Ophthalmology IRIS Registry Data Dictionary Element ID: 3034 Name: Insurance- No Payment Listed Coding instructions: Indicates if the patient has no payment listed Patient Demographics Short name: InsNoPaymentListed Element ID: 3035 Name: Insurance- Miscellaneous/Other/Foreign National, Worker s Comp, Auto Insurance Coding instructions: Indicates if the patient is insured by Miscellaneous/Other. Short name: InsMiscOther Element ID: 3100 Name: Payer ID Coding instructions: Indicates the Payer ID of the patient's primary insurance payer. Payer ID is a national numbering system that identifies healthcare payers authorized by CMS for healthcare claims processing and other electronic data interchange transactions. Short name: PayerID Format: text Format Value: None Date: 8 Version: 1.0

18 American Academy of Ophthalmology IRIS Registry Data Dictionary Date: 9 Version: 1.0