The Recorder's Office in Clark County "ensures the timely recording, microfilming, and permanent retention of documents; keeps an accurate and accessible indexing system for the research and retrieval of recorded documents; and ensures fiscal responsibility for the funds and fees collected while providing a cost-effective and efficient service to customers."

The Recorder's Office is responsible for maintaining the public record of Clark County, Nevada. The Recorder has maintained all real estate transactions, financing documents, maps, mining claims, military papers, declarations of homestead, mechanic's liens, marriage certificates, and real property transfer tax. These data sources are integrated from different organizations across government and business sectors.

Over the years, the method of recording documents has evolved from a system of manual transcription to microfilming and computerized indexing of documents. In order to respond to the critical information needs of customers, the business process reengineering (BPR) team implemented the reengineering of processes along the development of an integrated system to improve efficiency and accuracy in the recording process, and to provide increased quality of customer service. The driving force behind the BPR project is customer satisfaction. The results of the project can be useful for the county to develop standards for technological improvements on customer service.

At Software Engineering Institute (SEI) of Carnegie Mellon University, its MassCHIP (the Massachusetts Community Health Information Profile) links varied databases for the Massachusetts Department of Public Health. Statistical public health data is collected by a number of public and private organizations within the Massachusetts Department of Public Health (DPH). The DPH is responsible for health service planning and relies heavily on these diverse sources for accurate information.

Prior to working with SEI, the focus and structure of the data in DPH were not coordinated, limiting the benefit of gathering the great expanse of information. The problems were: multiple agencies sharing information manually; 37 separate databases existed within the DPH, unconnected and user unfriendly; data existing on many platforms and in many incompatible formats; data accessing was difficult, time-consuming, and consequently expensive; inability to query data from separate databases; a single means of extracting and storing data was needed to streamline the transfer of information between agencies. Technically, it was necessary for the database structure to accommodate a variety of data types. In addition, a query navigator was required so that inexperienced users could build queries. A mechanism for presenting selections of data in a variety of ways to a variety of viewers was also needed.

DPH worked with SEI to address the above problems. SEI developed a classic data warehouse for the health and statistical information provided by many state agencies and some private concerns. The system is called MassCHIP (Massachusetts Community Health Information Profile). It is an infrastructure that allows data sharing between departments. The user can view the data by using tabular formats, a variety of charts, and geographical maps, and data may also be exported into many formats for uses with analysis tools of the user's choice. The various sources of data use different coding schemes to store information. Platform independence and the ability to query across multiple databases have been realized. Centralized directory services are supported and data become available to anyone with Internet access. In addition, a data transformation utility was developed to translate multiple sources of data into the MassCHIP structure.

This is an example of a case of where government agency collaborated with business and technical professionals to support health service planning.

A number of data sources routinely available to state health departments can be analyzed as part of a state health department cancer control planning effort. In 1986, the National Cancer Institute (NCI) initiated the Data-based Intervention Research (DBIR) Program, a program of grants and cooperative agreements awarded to state health departments to build ongoing cancer control programs to ensure the translation of cancer prevention and treatment science into practice. The New York State Department of Health (DOH) was one of the first six states funded under this Program in 1987.

New York State DOH efforts involved the following steps:

• Identifying Data -- a total of 27 different sources of data were identified for evaluation. In general, these were population-based and included information on all New York State residents.
• Defining Data Characteristics
• Assessing Data Usefulness and Quality
• Analyzing Data
• Defining the Cancer Burden
• Prioritizing the Cancer Burden

Data sources of population-and non-population-based data for New York State include:

• New York State Department of Health: Cancer registry, Statewide Planning and Research Cooperative Data System, Family planning data system, Heavy metals registry, Vital records mortality data, and Behavioral Risk Factor Surveillance System
• New York State Department of Taxation and Finance: Cigarette tax information
• New York State Department of Environmental Conservation: Industrial chemical survey
• Roswell Park Cancer Institute(RPCI): tumor registry of RPCI patients, patient epidemology data system, patterns of care data base
• U.S. Department of Commerce: U.S. census data

Assessment of the magnitude of the problem was based on five additional factors:

• Impact of the cancer on the population as a whole
• The impact of the cancer on specific sub-populations
• Impact of the cancer on the medical care system
• Time trends in incidence
• Risk factor prevalence

The experience of the New York State DOH suggests that "state and local health departments have access to data sources that are useful in cancer control planning and the establishment of priorities for public health action. Combined with a systematic approach to planning, these data provide a solid foundation for ensuring that limited resources are directed to areas of greatest need and support efforts with the highest probabilities of success"(Lillquist, et al., 1994).

The application of this planning process and framework for setting intervention priorities in New York State also revealed several other important facts (Lillquist, et al., 1994): 1) data were unavailable for a number of cancer control areas that may otherwise have been chosen for intervention. "Work on this project enhanced recognition of the lack of information in some priority areas and stimulated developments to collect it"; 2) the data that were available were most useful in assessing the impact of various forms of cancer on the population and in identifying sub- populations with unusually high rates of disease or exposures to known risk factors. These data provide the foundation for targeting intervention efforts; 3) assessment of information specific to local communities and target groups was important for several reasons, such as providing the foundation for evaluating intervention outcomes; and 4) the planning process and framework used by New York might be useful for similar efforts in other states.

The purpose of the Child Care Policy Research Consortium is "to examine child care as an essential support to low-income families in achieving economic self-sufficiency while balancing the competing demands of work and family life". Research partnerships funded by the Child Care Bureau include state child care agencies, university research teams, national, state and local child care resource and referral networks, providers and parents, professional organizations, and businesses.

The Administration for Children and Families' (ACF) child care programs are administered by the Child Care Bureau within the Administration on Children, Youth and Families in the U.S. Department of Health and Human Services (DHHS). ACF's child care programs help low-income families to get child care and other supportive services so they can work or participate in an approved education and training program, then achieve economic self-sufficiency.

The Child Care Bureau has created a Child Care Automation Resource Center to provide better service delivery and support for child care programs; and to ensure more reliable data collection on all child care recipients and providers and timely, accurate child care data reporting.

In addition to the practice in child care at the Child Care Bureau, issues such as the value and shortcomings of state and local administrative databases as a source of data on child care were also addressed (Approaches to Data Collection, Chapter 5, 1997). The promising opportunities to forge partnerships between academic researchers and federal and state child care agencies, resource and referral agencies, and others who manage local databases has been noted. The need for efforts aimed at improving the comparability of data across local and state databases was also addressed.

The Quality Measurement Advisory Service (QMAS) report introduces "the range of dimensions of hospital quality that can be measured, reviews some of the diverse tools that are available today, and presents illustrations of how they are being applied in the ‘real world' by various organizations and associations around the country." The report is based on the proceedings of a workshop sponsored by the QMAS in the spring of 1997.

The QMAS was established in Washington State in early 1996 to assist state and local health care coalitions, purchasing groups, and health information organizations in their efforts to measure health care quality. It is a not-for-profit collaborative initiative of the Foundation for Health Care Quality, the Institute for Health Policy Solutions, and the National Business Coalition on Health.

Measuring the dimensions of hospital quality is clearly a complex and challenging task. There are three basic sources of data used in QMAS hospital assessment: 1) administrative files (e.g., claims or bills), 2) medical records, and 3) patient survey results. To determine whether a data source will be feasible and adequate to the task, it is critical to determine the information it contains, its accuracy and reliability, which patients are included, the cost, whether the data are computer- readable, and the currency of the data.

Administrative data refer to information generated as a by-product of administering care and services in the hospital, primarily from billing for reimbursement or from efforts to meet regulatory requirements. The data typically contain information such as patient demographics, diagnostic codes and procedures performed, and the charges billed to payers. Hospital administrative data are typically drawn from the following three sources: 1) payers, 2) state health data organizations, and 3) hospital associations. Administrative data are computerized and easy to analyze. They are also typically inexpensive and can be used to assess quality, efficiency, and other performance issues.

While administrative data offer information on care from the hospital's perspective, the patients' perspectives can also play an important role in the measurement of health care quality.

Patient surveys are important in assessing quality because they are useful for determining the patients' viewpoints about the care that they received; they assess the quality of interpersonal communications, and the patients' physical and psychosocial functioning outcomes. Surveys can identify problems and actions depending upon the nature of the questions asked. The rich data provided by surveys can pinpoint specific problems in the delivery process, and directly identify actions to improve care.

Clinical data refer to the clinical attributes of patients, and represent factors that health care professionals use for patients, such as symptoms (e.g. chest pain), vital signs (e.g. blood pressure), and laboratory test results. They are the types of observations written down by health care providers in the medical record, and are the data used to diagnose patients and determine treatment plans. Clinical data are derived primarily from medical records, which contain detailed information about each patient. Clinical data capture extensive dimensions of quality. They address a broader array of quality dimensions than can be addressed with administrative data. To be useful for quality measurement purposes, clinical data must be made computer-readable, which can be costly and time-consuming.

Clinical data also have some drawbacks, such as: 1) expensive cost of data collection. Electronic medical records that may reduce the cost of data collection are not yet used widely; and 2) lack of integration with outpatient and preadmission information. Crucial information remains in the patients' outpatient records in their doctors' offices, and hospital and physician patient records are usually kept separate.

The following three practices illustrate some of the organizational strategies that have been adopted in different geographic areas.

Cleveland's quality measurement program was initiated in the early 1990's by business leaders associated with the Health Action Council of Northeast Ohio, an alliance of 140 large employers. With help from Cleveland Health Quality Choice (CHQC), an organization established for this purpose, the group examined clinical quality and patient satisfaction with Cleveland-area hospitals. CHQC has published biannual reports that hospitals can use in quality improvement programs since 1993. Some Council members also use the data in their purchasing decisions.

The project to measure hospital quality in Cleveland started with two strong assumptions: 1) employers would reward hospitals that demonstrate good value, based on the CHQC data; and 2) the providers would use the data to improve service delivery. Based on their experiences using an integrated data set to assess quality and decision making, they offered the following advice (http://www.qmas.org/tools/guide-assessing/33total.htm):

• Disregard disclaimers about the data quality because the collected data is always useful in some fashion
• Solicit hospital participation directly. Hospital associations may oppose measures that might jeopardize an individual facility's status
• Carefully consider which parties should be involved at the beginning
• Consensus between providers and purchasers is very important to a quality-driven health care marketplace
• Coalitions have important roles in consensus building and their role will become very political in nature
• Experienced vendors should be selected for quality measurement initiatives
• Create a technical advisory board, including members from the business community, information systems professionals, statisticians, and physicians
• Share data with the public, but this can change the dynamics of the relationship between purchasers and providers, and limit meaningful dialogue on less rigorously obtained data
• Realize that the data will be used in ways that were not anticipated
• Data can become stale if not used to drill deeper into quality performance issues

The Greater St. Louis Health Care Alliance is a voluntary organization of 22 hospitals, 7 managed care organizations, and 30 business members. Hospitals are evaluated by the measures of risk- adjusted clinical outcomes and patient satisfaction. The Alliance chose to gather information on these measures from four vendors (http://www.qmas.org/tools/guide-assessing/33total.htm):

• The Picker Institute -- in St. Louis, Picker collects patient satisfaction survey data from 12,000 telephone interviews each year
• Aspen Systems Corporation -- Aspen abstracts clinical data from medical records
• Michael Pine & Associates -- they report clinical data that have been adjusted according to physician-reviewed models that use statistically significant risk factors
• Cleveland Health Quality Choice -- CHQC analyzes the data and produces the comprehensive and summary reports

Mr. Sutter, the program director, offers the following lessons for those who are considering an integrated approach to quality measurement (http://www.qmas.org/tools/guide-assessing/33total. htm):

Managerial lessons:

• Garner genuine collaboration and commitment up front
• Specifically define goals at the beginning of the project
• Provide extensive education for users
• Select experienced, qualified vendors and draft explicit contracts
• Conduct a timely, effective quality review process
• Use project management techniques to manage schedules, deadlines, and logistics

Measurement lessons:

• Do not measure hospital costs. It is not worth the time, money, or effort to measure them
• Determine how data for transferred patients and patients discharged to skilled nursing facilities will be handled
• Quantify the appropriate sample size to ensure statistical validity

Provide a combined performance measurement, reflecting both clinical outcomes and patient satisfaction results

The Employer Health Care Alliance Cooperative, known as the Alliance, was founded in 1990. It is an employer-owned, nonprofit cooperative based in Madison, Wisconsin. In 1997, the Alliance represented nearly 120 large to mid-size employers with approximately 25,000 employees, as well as over 700 small employers with about 7,000 employees. The Alliance's interest in quality measurement data is twofold: 1) to improve the health delivery services and the health status of its members' employees; 2) to support the vision of accessible, affordable and effective care, three integrated strategies that drive quality improvements are employed: collaboration, informed self- interest, and contractual incentives. The three strategies are described below (http://www.qmas. org/tools/guide-assessing/33total.htm):

Collaboration

Collaboration, defined as two or more organizations working together to achieve a common goal, is appropriate for situations where all parties agree that a project is worthwhile, yet beyond the scope of any one organization. While collaboration can be useful in such situations, for fostering learning and relationship-building as well as serving as a motivating force to attract others' involvement, it can also be very time consuming and resource intensive.

Informed Self-interest

Informed self-interest is a strategy "where comparative data is shared privately or publicly with providers and there is a commitment to monitor the data on a periodic basis." This strategy is useful in situations where variability exists among organizations. This strategy can create expedient results and foster competition, but it can also engender defensiveness from hospitals. In addition, it may undermine the measurement if the project focuses intensely on one or two problem areas.

Contractual Incentives

Contractual incentives involve the use of performance targets in determining hospital reimbursement. Structuring reimbursement to reflect performance is useful when other approaches are not working. It is important that purchasers are willing to pay for it, and results in purchasers having access to valid and reliable data to substantiate hospitals' performance.

A lack of standardized data among various sources is one of the most serious barriers to sound health care needs identification and decision making. In Kansas, health care occupations information is maintained within numerous agencies, and a series of discussions were held with the Kansas health care occupations credentialing boards to discuss a minimum data set for health occupations in Kansas. A minimum dataset was recommended based upon recommendations from the National Center for Health Statistics (NCHS). The Health Care Data Governing Board developed policies for collection, security analysis of health data, and dissemination of information from the Kansas health care database. In 1995, the Governing Board implemented its first data collection initiative, a health system inventory. Subsequently, centralized and evaluated data was collected on health-care occupations from eight credentialing boards encompassing 23 occupations. The mechanisms bring data issues to the forefront, and data products are available for decision making.

The accomplishments of Kansas' health care database developers and partners include (http:// www.ink.org/public/hcdgb/khcd95report.html):

• Publication of the Kansas Health Data Resources Directory which provides a concise guide to health data collected within the state
• Development of centralized computer and analytic resources within Kansas that standardize, house, analyze, and disseminate health data
• Implementation of the health system inventory and publication of the first health care provider standard reports which are tabulations of health care occupations within Kansas
• Development and approval of a model data collection instrument for health care occupations to serve as a guide for data collection on these occupations in Kansas
• Acquisition of hospital discharge summary data for analysis and public distribution
• Development and consensus on health status indicators to be collected for the database using sentinel measurements to monitor the health status of Kansans
• Revision of policy questions to increase the priority for evaluating the quality of health care in Kansas

The Kansas Health Data Resource Directory catalogs the health data resources maintained in Kansas state government, universities, and private agencies. It serves as a pointer to locate health information collected in Kansas. It also provides information about who may be contacted about the data. The Resource Directory serves as a reference about the kinds of data available, helps to identify duplication in data collection, and ultimately may facilitate data sharing between agencies. It will also be useful to anyone collecting or using health care data.

A lot of policy questions were raised by the Health Care Data Governing Board to ensure that the data collected and maintained in the health care database are relevant. These potential policy questions can be used to guide the database design. The questions relate to the availability and distribution of health care services, utilization, expenditures, health status, and outcomes. To answer these questions, the following need to be identified and determined with immediate priority (http://www.ink.org/public/hcdgb/khcd95report.html):

• the providers available in the state and the services they are providing
• the sources and applications of funding for health services by provider and payer type
• the demographic characteristics of the uninsured and underinsured
• the distribution and access problems for health services in Kansas
• the utilization of health services in Kansas
• the characteristics of the population utilizing public health services
• the health status of Kansans
• the cost to insure the uninsured and underinsured population
• the differences of utilization patterns and resulting outcomes across Kansas
• the services provided by the primary care providers in Kansas
• the portion of the health care dollar spent on preventive medicine
• the services provided by public health departments
• the full costs of medical litigation
• the effectiveness of operating service networks developed under health care reform
• the effects of Kansas risk adjustment factors on community rating
• the effects of insurance mandates on premium costs
• the results of cost and outcome comparison between and among types of primary care professionals
• the utilization rates and costs of common procedures for individual hospitals, clinics, ambulatory centers and community health centers
• the impact of health care reform on quality of care

An integrated data system of health care providers is being developed. It will allow users to analyze data across professions and facilities from multiple data sources. The data will be available to customers via: 1) Internet access and electronic data transfer; 2) information formats generated through standard reports and special requests; and 3) publications and media articles (http:// www.ink.org/public/hcdgb/khcd95report.html).

The Kentucky KIDS COUNT is "a unique consortium of researchers and children's activists who have significant expertise in the aggregation, interpretation, and use of data to affect public policy." The consortium's work includes producing a series of reports on children and families to publicize the needs of children, influencing budget and programs decisions, and monitoring state and local performance for children.

Multiple data sources are used for the 1996 Kentucky KIDS COUNT profiles of the status of children and their families in the state and its 120 counties and for the profiles of education-related data in the 176 local school districts in the state. For example, data on children's enrollment rates, attendance rates, dropout rates, free and reduced lunches, and per pupil spending were provided by the Kentucky Department of Education. Data on child abuse and neglect were derived from annual data compiled by the Department of Social Services, and the Cabinet for Families and Children. Data on food stamps were provided by the Department of Social Insurance, and the Cabinet for Families and Children. Data on Medicaid were derived from data compiled by the Department for Medicaid, the Cabinet for Families and Children, and the Social Security Administration. Population data were obtained from the Urban Studies Institute, the University of Louisville, and population forecasts. Data on pupil/teacher ratio, retention rates, and revenue by source were provided by the Kentucky Department of Education, Division of Finance. Data on supplemental security income (SSI) were obtained from data compiled by the U.S. Social Security Administration.