Welcome to Population Health, Population Health IT and Data Systems. This is Lecture e. This component, population health, discusses the application of informatics and informatics methods in population health management. This unit, population health IT and data systems, explains the challenges and opportunities of using different data types, data sources, and data systems for population health IT. The objectives for this lecture are to describe the advantages and disadvantages in using various data sources for population health management and analysis, explain the added value of nontraditional data sources for population health IT. This lecture discusses traditional and nontraditional data sources that can be use for population health analytics. Traditional data sources of population health include insurance claims data, such as hospital claims, professional claims and pharmacy claims, and clinical data such as electronic health records, EHR and registries. Insurance claims data have inherent advantages and disadvantages when used for population health analytics. The advantages of claims data include. Insurance claims usually cover a broad scope of data across various healthcare providers. This will give population health analysts a more complete picture insured events that have occurred to a patient across various entities in the larger healthcare ecosystem. For the same reason, the diagnosis and medication records that are extracted from claims often have a good reliability. In addition, because of the use of unified forms and electronic claim submissions, insurance claims often have a higher consistency across different claims data sources. This is critical for merging various claims data sources and developing a large enough population wide data set to predict a given outcome in the underlying population. And finally, population wide claims datasets, already exist and can be acquired by population health analyst to develop the desired models to predict the trends of healthcare utilization in the underlying population of interest. Now, there are also disadvantages in using claims data such as Insurance claims include a limited number of data types that can be used for population health analytics. Indeed, the primary source of data types extracted from claims and used for population health analytic are demographics, diagnosis, medication, if available, and utilization. Therefore, other potentially valuable data types such as lab values, vital signs, problem list, and family history are missing. Claims only exist for the insured patients thus, any models developed based on claims can only be used for the insured population. This means that claims based predictive models cannot predict healthcare utilization in the uninsured population with a fair accuracy. And finally, claims data only include events, procedures, and medications that are covered by the payer, depending on the individual member's eligibility. Known covered procedures and medication such as over the counter medications are fully missing. Hospital insurance claims data include both in patient and out patient services that are provided and are tied to a facility such as a hospital. The hospital claims are often referred to as the facility claims. The inpatient portion of the hospital claims are based on the events and procedures that occurred after admission to the hospital facility. On the other hand, the outpatient facility claim are generated for services such as emergency room visits, ambulatory surgeries, and other services provided in an institutional setting, where there is a facility charge. Facility claims are usually billed using the Uniform Billing or UB form. More specifically, the UB04, also known as CMS1450, previously UB92, which is copyrighted by the National Uniform Billing Committee NUBC is the main form used for hospital claims. Most insurers use the same unified form, however, submissions are increasingly accomplished electronically. For example, the centers for Medicare and Medicaid services, CMS, uses the ANSI ASC X12N 8371 standard to receive hospital claims. This image shows the UB04, also known as CMS1450 form, as noted on the image. The form includes various sections such as patient information and demographics, provider information, payer information, discharge information, list of procedures, list of diagnosis and a section for additional remarks in information, if needed. The procedure codes are usually encoded using Current Procedural Terminology, CPT, and healthcare common procedure coding system, HCPCS, National Drug Codes, NDCs, are used for medications. Sometimes internal classification of diseases. ICD procedure codes are found on the header of UB04 facility claims, but they are generally less specific than CPT and HCPCS procedure codes. The diagnoses codes are coded in ICD9 or ICD10. This table shows the data elements of the UB04 form. As shown, there are a variety of data elements, such as patient information and demographics, provider information, payer information, discharge information, procedure and revenue codes and diagnosis codes. A number of these data elements are commonly used for population health analytic. Such as demographics, procedures and diagnoses. Note that not all fields are required and the requirements vary between inpatient and outpatient settings. Physician or professional claims include procedures performed by clinicians, respiratory therapists or physical therapists, regardless of the place of service. Professional claims are billed using the CMS 1500 form. Because of the fact that professionals use a different claims form, hospital admissions may produce two claims. The facility services, such as room and board, ancillary services, and in-hospital drugs. And a professional services, such as physicians, surgeons, nurse practitioners, and so on. Most CMS 1500 forms are electronically submitted. CMS uses the ANSI ASC X21N 837P standard to receive professional claims. This image shows the CMS 1500 form as noted on the image the form includes various sections such as patient information and demographics, provider information, payer information, list of procedures and list of diagnosis. The procedure codes are usually encoded using CPT and HCPCS. Note that professional claims do not contain ICD procedure codes. The diagnosis codes are coded in ICD9 or ICD10. This table shows the data elements of the CMS 1500 form. As shown, there is a variety of data elements such as patient information and demographics, provider information, payer information, and procedure and diagnosis codes. A number of these data elements are commonly used for population health analytic, such as demographics, procedures, and diagnoses. Most medication claims are submitted electronically and through transactional systems. Due to the semi-automated process and the fact that coding is not influenced by subjective bias, medication claims data are often adjudicated more quickly. Medications can be claimed through CMS 1500 or other forms by commercial insurers, although most of the transactions occur electronically. Outpatient drugs are usually assigned in NDC code. Medication claims may also contain additional descriptive information about the drug itself. Note that medications that are prescribed by a physician, but are never filled, will not generate claims In contrast to medical claims, most medication claims do not include the underlying diagnosis. As shown in this slide, commercial payers use different forms for medication claims. Note that most medication claims are submitted electronically. Most medication claims data are managed by the Pharmacy Benefit Management or PBM intermediaries. PBMs facilitate the receipt and processing of medication claims between providers and payers. Some large PBMs with national coverage, which is an advantage for population health analytics, include Express Scripts CVS Health, UHC/OptumRx Catamaran, and Prime Therapeutics. Other nonclaim sources of medication data mainly focus on prescription events. These sources include EHRs, which contain the prescription data about a medication but miss the information about whether or not a medication was filled. Surescripts, which operates the nation's largest commercial e-prescribing network. Prescription Drug Monitoring Program, PMDP, which refers to the state-run programs that collect and distribute data about the prescription and dispensation of federally controlled substances and other potentially addictive drugs. EHR data have advantages and disadvantages when used for population health analytics. The advantages of using EHR data include EHR's often contained unique data types such as lab values, vital signs, social data, problem list and other clinical information. EHR's often contain procedures and medications that are not covered by insurance, such as over the counter medications. And EHR's also include both insured and uninsured patients. Now, there are also disadvantages in using EHR data, such as EHR's usually include data collected by one provider entity and do not include data collected elsewhere. EHR's often have a lower consistency across different data providers, especially considering the free text. And finally, population-wide EHR data sets are uncommon and very costly to develop. The meaningful use EHR incentives along with the certification criteria for EHRs have helped to shape a minimal required set of data types that EHRs should collect. Such as patient demographics, past medical history, problem list, diagnosis, procedures, allergies and medications. Despite the EHR certification criteria, the variation of data types and data quality of EHR data collected across different providers is still hindering the common development and use of population wide EHR data warehouses for population health analytics. Also, it should be noted that EHR data are usually limited to data collected from patients within a given provider's network. And lack information collected elsewhere. Potential EHR data types that could be useful for population analytics include demographics, diagnoses and problem lists, procedures, medications, family history, social history, vital signs, immunization records, surveys and patient-reported outcomes, and additional meta-data and free-text notes and reports. According to the Arizona Department of Health Service, a patient registry is an organized system that uses observational study methods to collect uniform data. Such as clinical data to evaluate specified outcomes for a population defined by a particular disease, condition or exposure and that serves a predetermined scientific, clinical or policy purpose. The registry database is the files derived from the registry. Some types of registries include, Local EHR-based registries, such a hospital level registry to identify patients with a certain condition, need or risk. Clinical trial and condition-based registries, which are often funded federally on profit associations. Public health registries that are often limited to a certain geographical boundary. And national reporting registries such as CDC, FDA or NIH registries that have a national coverage. Registries also provide valuable data types for population health analytics such as demographics. The diagnosis, medications, family history, social history, immunization records, surveys and patient reported outcomes. There are also a number of non-traditional data sources that can be used for population health analytic. These potential data sources include patient-provided and patient-generated data sources, public health and vital records data sources, social services data sources, environmental and geographical data sources. Resource availability data sources, consumer and non-medical data sources, health information exchange, HIE data sources. Other medication data sources such as sure scripts and PDMP and other potential data sources yet to be collected on a population level. This concludes Lecture e of Population Health IT and Data Systems. The current and potential data sources for population health analytic were discussed in this lecture. The traditional data sources reviewed in this lecture include insurance claims, medication data, EHR data, registries, the non-traditional data sources listed in this lecture include patient-provided and patient-generated data sources. Public health and vital records data sources, social services data sources, environmental and geographical data sources, HIE data sources, and other potential data sources.
