Skip to main content

Measure Specification

Define the Data Source(s)

Measure specifications should include data sources and methods of data collection that are acceptable and preferably digital. When contemplating the source(s) of data, the measure developer must consider the feasibility and methods of collecting data from that source. The data source(s) used to calculate a measure will determine the reliability, validity, feasibility, and usability of the measure. If the measure may use more than one data source (e.g., registry and electronic clinical quality measure), the measure developer should generate detailed specifications for each data source. The measure developer should collect evidence so results calculated from the different data sources are comparable. Measure developers should avoid paper patient records as much as possible, as use of paper records is labor intensive and therefore, increases costs.

Examples of Data Sources

Data Source

Description

Strengths

Limitations

Administrative Data

Administrative data include multiple types of information originally collected for administrative purposes. Examples of administrative data sources are birth registries and tax records. 

Other types of administrative data include individual-level demographics obtained from eligibility or enrollment information, crime reports, and census information. Payroll data and other databases containing information about providers/clinicians can also be a source for some types of measures.

Administrative data can provide information not usually found in a clinical database. Use of administrative data is less burdensome than manual abstraction to measured entities for data collection.

The primary purpose of these data is not for the purpose of quality measurement, measured entities collect these data for other purposes, e.g., admissions, discharges, and transfers.

Claims Data

The source of claims data is health care reimbursement or payment information. This information can come from submitted and adjudicated claims or from the provider’s billing system. Claims include admission and discharge dates, diagnoses, procedures, and source of care.

Professionally coded using standardized data and drawn from large populations (i.e., more representative of the populations of interest).

Use for quality measurement is less burdensome than manual abstraction to measured entities for data collection.

The primary purpose of these data is not for the purpose of quality measurement. Measured entities collect these data for other purposes, e.g., billing. Therefore, they can have varying degrees of clinical detail and often limited in content, completeness, timeliness, and accuracy.

Paper Patient Medical Records Paper patient medical records are a traditional source of clinical data for measures. These records may include data from the clinical laboratory, imaging services, personal health records, and pharmacy.  Paper patient medical records have detailed clinical data with a rich description of care. Identifying test sites that can serve as data sources can be difficult; abstraction of paper records is time-intensive; requires expert staff (cost and time) to interpret each record and input data findings into a format suitable for analysis; abstraction can be open to subjectivity and interpretation or lack of consistency in the abstraction of the data.
Electronic Patient Medical Record Electronic patient medical records are a move from the traditional source of clinical data, paper patient medical records, to digital sources for measures. These records may include data from the clinical laboratory, imaging services, personal health records, and pharmacy.  Electronic patient medical records have detailed clinical data with a rich description of care with a reduced cost of accessing clinical information as compared with paper patient medical records. Identifying test sites that can serve as data sources can be difficult; inconsistent adoption of electronic health record (EHR) systems, especially across settings; extracting the data requires expertise, time, and money; hurdles related to continuing use of paper notes for point-of-care documentation; use of drop downs and structured fields can reduce the richness of the clinical data and descriptions of care; structured data not always using or mapped to standard terminologies; and potential negative impact on clinical workflow.
Electronic Clinical Record Electronic clinical data consist of individual-level information amenable for extraction or pushed in an electronic format usable by a measure, for example, data from bedside vital sign monitoring devices and personal health devices. Bedside vital sign data can be directly pushed to the EHR and personal health device data may be uploaded to the EHR.  Electronic clinical data have a reduced cost of accessing clinical information from the individual's medical record or personal health device (e.g., home blood glucose monitor). Identifying test sites that can serve as data sources can be difficult; extracting the data requires expertise, time, and money; hurdles related to continuing use of paper notes for point-of-care documentation; device data may be external to the medical record; and still only partially implemented in some settings.
Registry

A registry is a collection of clinical data for assessing clinical performance quality of care. 

Registries may be part of a larger regional or national system that may operate across multiple clinicians and institutions. Examples of national registries include the Chest Pain – MI Registry™ (from the American College of Cardiology and American Heart Association), the Society of Thoracic Surgeons™ National Database, and the Paul Coverdell National Acute Stroke Registry.

Registries collect data, usually standardized, from multiple sources and across care settings; often available as an electronic upload, with the registry directly and electronically submitting data.  It is unknown how registry requirements impact workflow; feasibility of data collection is determined by the data requirements imposed by the registry. Registries may impose fees so there may not be representation from all relevant providers and some selection bias for those who choose to participate.
Standardized Patient Assessment

CMS uses data items or elements from validated health assessment instruments and question sets to provide the requisite data properties to develop and calculate quality measures. Examples of these types of data include the Long-Term Care Facility Resident Assessment Instrument, the Outcome and Assessment Information Set, and Inpatient Rehabilitation Facility Patient Assessment Instrument.

Well validated and tested There is a potential for bias as some have mixed use for determining reimbursement, meeting conditions of participation, and assessing quality; may be proprietary, therefore no available non-proprietary reliable or valid tool.
Patient-Reported Data and Survey Individuals may provide data directly in the form of a survey, questionnaire, or assessment. Surveys (e.g., Consumer Assessment of Healthcare Providers and Systems® surveys that collect information on beneficiaries' experiences of care) are advantageous because they ask about concepts such as individuals’ experiences and feelings. Patient-reported outcomes, such as pain assessments and quality of life indices, provide the person’s perspective on their health, quality of life, or functional status. Unique source of data available only from the individual or individual’s family/significant other; direct way to collect person's experience Validated/reliable assessment tools are needed (these may be proprietary); some self-reported data are not often used in the delivery of care, so likely small number of responses (n); not always reliably or consistently collected; may be costly and time-intensive for data collection.
Last Updated: