According to Drucker (1998), in the next 10 to 15 years, collecting information from external sources will be the next frontier. Following are cases of research or practical experiences in the use of multiple data sources across multiple organizations. They are subdivided into cases (I) where the different organizations are in the same sector of the economy (e.g. in business or government), and (II) where the organizations cross sectors (e.g. business and government).
Alciati and Glanz (1996) described the results of an analysis of five states' experiences using and integrating available data to develop cancer control plans for their states. These states included Georgia, Maryland, North Dakota, Vermont, and Washington State. In 1989, the National Cancer Institute funded the second round of Data-Based Intervention Research (DBIR) cooperative agreements with state health agencies to implement a four-phase planning model to establish ongoing cancer prevention and control programs. Activities focused on the identification and analysis of data relevant to the development of a state cancer control plan. Alciati and Glanz's research explores how states use different types of available data to make public health planning decisions, the levels of sufficiency of data for this planning, and the perceived costs and benefits of a data-based planning approach.
According to Alciati and Glanz, while using health data to guide public health planning efforts is not new, information on how states use existing multiple data sources for comprehensive cancer prevention and control planning is limited. What is lacking is "a clear picture of how these components fit together in a comprehensive state-level planning process, how data are used to establish cancer prevention and control priorities and to identify proven interventions for implementation, and what states perceived to be the costs and benefits of such detailed data-based planning."
Data sources and integration methods
Each state used three categories of data:
Health Data -- including mortality and morbidity (incidence); states generally relied on a small number of measures, such as the number of state deaths, age-adjusted death rates for the state, and survival.
Behavioral Data -- including health behavior, risk factors, and determinants of behavior (for example, knowledge, attitudes, and beliefs). The Behavioral Risk Factor Surveillance System (BRFSS) was the primary source of state-specific behavioral data. Behavioral data were used primarily to identify target groups for intervention.
Environmental and Health Services Data -- including environmental characteristics such as the presence of cancer control legislation and worksite policies, the availability of early detection equipment to support public health goals in cancer prevention and control, as well as information about the existence of cancer control programs and the utilization of health services. The most important sources for this information were hospital discharge datasets and state and local surveys.
For each type of data, the specific data source, the measures used, the type of subgroup analyses performed, and to the extent possible, how the data were used to establish planning priorities and identify interventions were recorded in this research. The data was then summarized, to identify the number of states using each type of data source, data measure, and subgroup analysis as well as the number of states using data to make each type of planning decision.
In these five state programs, comprehensive cancer control planning efforts used a full range of integrated data, and linked these data to decision-making for cancer control. "This research also provides a framework for public health planners to identify the type of data likely to be available for cancer prevention and control planning at the state level, various measures that can be realistically derived from these data, and how they can be linked to public health planning"(Alciati and Glanz, 1996).
Seven state health departments in Illinois, Nebraska, New Jersey, New York, North Carolina, Texas, and Wisconsin, participated in an effort to utilize a variety of state-specific cancer-related data to describe the cancer burden in their state's population. The data were then used to develop a statewide cancer plan or to supplement an existing plan to address the defined problems. The efforts in these states can serve as models for data use to prevent and control cancer and other chronic diseases. State-specific data can be used to rank needs and make a clear case that can influence resource allocation decisions. In this research, Boss and Suarez described the data sources and additional statistics that were used to provide a broad picture of the cancer burden which can assist in targeting and defining intervention needs.
According to Boss and Suarez, more data exist to describe cancer than any other disease. However, the data have been rarely systematically evaluated to target and plan public health programs in cancer prevention and control. Program planning has often been based on historical or political priorities, and therefore programs have not necessarily been located where the need or potential impact is the greatest.
Potential Solutions and Data Sources
Four major data sets were used by these states: 1) mortality data, 2) incidence data, 3) risk factor data, and 4) hospital discharge data. These data sets appear to be the most accessible and potentially useful of the examined data sources. Various additional data sets were also used, many of which are available within state government, and often within health departments. Data sources used to describe the facilities within the state included the ACOS listing of hospitals with approved cancer programs and information from the local officials of the American Cancer Society (ACS), Cancer Information Service, and the radiologic health unit of the state health department. Information on personnel resources came from state medical organizations and the state boards of medical examiners. Environmental data bases included lists of abandoned landfills and results of water and air monitoring. Limited treatment information could be obtained from ACOS patterns of care surveys, cancer centers, and public health clinics. State taxation records provided information on cigarette and smokeless tobacco sales and tax income.
Other sources were also used. Various types of insurance claims data were examined, such as Medicare, Medicaid, Blue Cross, and State Employees Insurance records. National data sources were also used primarily for comparison with local data. The national sources included the National Health and Nutrition Examination Survey (NHANES), Hispanic Health and Nutrition Examination Survey (HHANES), and Nationwide Food Consumption Survey. For example, the Texas sample from HHANES was large enough to provide state-specific data on the Hispanic population. New data variables can be derived from a variety of existing information sources. Such information is essential to plan programs that meet a particular goal.
The above two research cases focused on using different data sources in organizations in the same sector of the economy, the government sector.
| Next >