Table 6 summarizes key data gaps by type that stakeholders perceive in today’s market. Note that a “data gap” is not synonymous with an “information need”. In practice, information needs are met by merging multiple kinds of data to answer a question. Thus, an absent core data element (like transaction and service based information on encounters with appropriate identifiers to link it back to people or health plans can limit the ability to develop information to meet a number of needs. Thus, our focus here on data gaps is to highlight core data elements or kinds of data that are needed to answer many of the kinds of questions stakeholders raise. The first sets are substantive concerns dealing with aspects of the health care system, while the last relate to attributes of useful data.
In terms of structural information, key gaps in information exist with specific components of the system, though the more serious concerns arise from the rapidly evolving and shifting linkages between these components, which influences the process of care. The absence of a comprehensive census of health insurance plans or arrangements or set of identifiers with information or plan characteristics is a key gap since it means there is not effective frame for data collection. Two other important gaps involve limitations in data on ambulatory care and nonphysician providers; and information on community-based alternatives to long-term care.
With respect to linkages in the health system, the key gaps appear to be related to the limited information on ownership of provider practice and aggregation of providers as consolidation proceeds. The gaps are also related to the limited information on the arrangements between managed care plans and providers. Information about how components relate across the continuum of care also are limited. Our interviews suggest that gaps are a major concern for private sector stakeholders, as well as others. These concerns also are relatively recent, reflecting changes in the delivery of health care with consolidated and managed care.
In terms of the process of care, the concerns are generally that the available structural information and encounter systems may not provide the kind of understanding needed about how medical practice functions and is determined. This makes it hard to develop good estimates of resource needs and further limits the information available on the efficiency of the are process. Most basically, these needs reflect the broader question of how to obtain process information on how care is delivered and also to better link it with structural information (i.e., on individual providers and systems, including their characteristics). There is a related need for information that could be used to determine how to develop the ability to do standard comparisons based on patient-level information. This effort might involve comparing how care is delivered across in managed care, and fee-for-service sectors. This gap exists for two reasons: (1) there is typically no universal set of transaction or service level information upon which to base analysis especially for the under 65 population and with the growth of managed care; and (2) there are considerable barriers in linking these data to structural data (also absent) about providing plans or other entities.
Like information on the process of care, information on outcomes of care is limited by the absence of both clinically relevant data on performance and the ability to link it to individuals, providers, and health plans. Expenditure data are limited by the categories in which such spending can be assessed and by the absence of meaningful information on both out-of-pocket spending and spending for distinct subpopulations or to achieve given outcomes.
For all kinds of information, there are serious concerns about the absence of data below the national level, that is, for states, markets, and politically relevant localities (such as inner cities). There is also the concern that data is not captured in a way that provides flexibility for analysis that can be adjusted to residence or that allows the data to be aggregated so that it meets diverse needs for information at the provider, patient, and population levels. There is also a concern to establish consistent trends in measurement while remaining flexible and timely to account for change.
SUMMARY OF DATA GAPS IDENTIFIED FROM THE PERSPECTIVE OF
STAKEHOLDERS OUTSIDE THE FEDERAL GOVERNMENT
Component Parts of the Health Care System
- No comprehensive census of health insurance plans or arrangements exists
- Limited information available on community based alternatives to long term care, intermediatemodels
- Data on ambulatory care and non-physician providers is limited, incomplete, and inconsistent.
Linkages among Parts of the System
- Little information on ownership and/or aggregation of physicians and/or hospitals
- Integration and arrangements (including transfer of risk) between health plans and provider entities and their constituent individual providers not identifiable in data.
- Linkages across the continuum of care not identifiable in data.
- Duplication exists across diverse data collection efforts
PROCESS AND OUTCOMES
- Transaction level information on services which can be linked to individuals, providers, or health plans are often absent
- Clinically relevant data are limited
- Need the ability to link expenditure data to functional role rather than setting
- More meaningfully categorized expenditure data are needed including out of pocket spending and spending for given outcomes or people
CROSS-CUTTING CONCERNS: GEOGRAPHICAL RELEVANCE AND STRATEGIC FLEXIBILITY
- State specific data are often lacking for all or some states. Data for policy relevant localities (e.g. inner city) is even more limited
- Data need to be adjustable for residence versus service location (especially when areas cross states)
- Data need to support consistent trend analysis but data captured need to be flexibly defined and timely to account for change in the health system
- Data should permit flexible aggregation at different levels (e.g. service level; provider level; patient level; or population level)
- Data need to be capable of being linked to relevant population-based units to support targeted analysis (e.g., individual with a given health problem).
SOURCE: MPR Analysis