Identifying, from among various models, one set of indicators regularly available to provide reliable information to policymakers and the public about the factors key to their health prospects and their health goals--in a fashion that might help drive progress toward those goals--compels attention to certain issues: the extent to which an indicator represents a reliable harbinger of developing health prospects; the ability of an indicator to draw attention to groups at risk and problems that are amenable to change; the extent to which an indicator might be linked to an action or set of actions which might affect its course; the relevance of an indicator to a stated national objective; the anticipated availability, reliability, and periodicity of the data source; the various levels on which related information might be gathered and reported; how the measures fit together as a package that is conceptually coherent, balanced, and readily understandable; and the breadth of the audience for which an indicator might have appeal. All these are important to the selection process. For the purpose of this report, criteria are identified both for the selection of individual measures and for the selection of the set of measures.
For individual measures:
Audience Interpretability: An indicator chosen as a leading health indicator should be easily interpretable to, and understandable by, the general public and opinion leaders, as well as the health and medical communities. It should, by its nature, help to inform these groups about issues and conditions important to the national health profile and health progress.
Population Applicability: An indicator chosen as a leading health indicator should reflect an issue which impacts in important ways on the health profile of the national population.
Problem Impact: An indicator chosen as a leading health indicator should address a problem of substantial impact. Mortality, morbidity, and economic costs are all parameters that reflect the relative impact of a problem and can be calculated either for disease or injury that impact directly or for conditions that predispose to disease or injury and contribute through them to increased mortality, morbidity, or economic loss. In either case, it is anticipated that an indicator selected for inclusion in this set will be in the top tier of factors impacting on national health prospects.
Link to Objectives: An indicator chosen as a leading health indicator should be linked to one or more of the Healthy People 2010 objectives. The notion that this set of key indicators may be used to reflect in some fashion both the nature of the health objectives and the progress toward them requires that they be linked to them and reflect their content.
Reliability as an Indicator: An indicator chosen to reflect the state of the Nation (or subregion) on a particular health issue or condition should offer a generally reliable indication of the overall level and direction of issues and problems embraced by that area or condition. That is, to the extent possible, it should not reflect the problem from the perspective of only a relatively narrow aspect or population group.
Data Availability: An indicator chosen as a leading health indicator should be one for which data can be anticipated from an established data source on a regular basis. Although some exceptions may pertain, this suggests that the data should be collected on at least a biennial basis.
Multi-Level Trackability: An indicator chosen as a leading health indicator should be one for which data can be anticipated at multiple levels and for multiple groups. Specifically, data should be available at the national, state, and county levels, as well as by age, gender and ethnicity.
For the set of measures:
Audience Interpretability: The set of indicators should be easily interpretable to, and understandable by, the general public and opinion leaders, as well as the community of health and medical professionals. It should, by its nature, help to inform these groups about issues and conditions important to the national health profile and health progress.
Population Applicability: The set of indicators should reflect issues which impact in important ways on the health profile of the population as a whole and be generally reflective of the full set of factors most important to shaping that profile.
Profile Balance: The set of indicators should reflect a balance among targets that does not overemphasize any one group or condition. The set should reflect contributing factors in a manner and frequency roughly proportionate to their impact.
Relevance to Policy Action: The set of indicators identified as the Nation's leading health indicators should be useful in directing policy and operational initiatives. That is, changes reported in the status of a measure from one period to the next should offer lessons to the policy domain that are readily interpretable, if not actionable.
It should be noted that these criteria allow for the specification of social factors not customarily considered as health measures--such as income, education, and employment--which impact on health status. It is the opinion of the working group that, although these factors are not necessarily actionable by the health community, the strength of their impact on health prospects warrants their consideration as health indicators.
Each model presented in Section Three has important strengths and limitations. The mortality model fails to measure many important outcomes, such as morbidity and disability, and it fails to show a timely response to important changes in health behaviors. The social indicators model and environmental model measure important dimensions of health, but are not comprehensive enough to stand alone as models for leading health indicators. Similarly, the single parameter model is devoid of meaningful information about health determinants.
After eliminating these models, the working group turned to the remaining with three questions: 1) Which models are most relevant to the general public? 2) Which models are most related to areas in which working group members hope to see major achievements? and 3) Which models can be clearly linked to the Healthy People initiative?
Three models which represent different approaches to the nature of the indicators chosen have been selected for elaboration and are presented here as candidate sets:
Each of the three models listed above is oriented around the indicators themselves. The three models assume that results will be released on an annual basis. An alternative approach, presented below, focuses instead on release strategies (e.g., monthly, semiannual, or quarterly); the choice of indicators follows from the release strategy. These models place a premium on communication.
The three candidate sets of leading health indicators described below demonstrate alternative approaches to developing indicator sets. They are intended to provide a vehicle for illustrating various characteristics of the models that are valued by HHS staff and viewed as most supportive of the Healthy People initiative, not to limit or restrict the activities of the expert panel to be convened or of the Secretary of HHS in making the final choice. The indicators have been selected as illustrative examples, based on Healthy People 2000 objectives. The working group encourages that consideration be directed beyond the Healthy People 2000 objectives to the draft Healthy People 2010 objectives (available April 30, 1998) and other indicator sets.
Health Status Model
The health status model would establish as leading health indicators the set
identified by the Committee for Objective 22.1. Included in this set are
measures that have been either established for regular collection at the
national, State, and local levels, supplemented by several previously
identified as priority data needs under objective 22.1 and which are
anticipated to be available for regular tracking at the State and national
levels for Healthy People 2010. In this scenario, some 29 indicators
would be designated to reflect health status and progress with respect to
incidences for leading causes of morbidity and mortality, prevalence of factors
increasing risk for those conditions, and use of services to reduce their
impact. Presented here are those indicators, with the numbers for the relevant Healthy
People 2000 objectives noted in parentheses (table
1). The working group has substituted the proposed Healthy People 2010
categories (increase years of healthy life, promote healthy behaviors, protect
health, assure access to quality health care, and strengthen community
prevention) for the original categories developed by the Committee for
Objective 22.1.
|
Increase Years of
Healthy Life
Promote Healthy
Behaviors
Protect Health
|
Assure Access to
Quality Health Care
Strengthen Community
Prevention
|
|
*
Parentheses indicate related Healthy People 2000 objective number. |
|
Discussion: During the past decade, the National Center for Health Statistics and the Committee for Objective 22.1 have used a consensus process to develop health status indicators that are accepted and used widely at the state and local levels. With use of the Health Status Model, three important criteria can be met: first, the set of indicators is interpretable to health and medical professionals and, in a number of States, and counties, is being regularly communicated to the public; second, data to track the indicators generally are available at the national, State and local levels, as well as by age, gender, race, and ethnicity; and third, the indicators are known to be reliable. Although titled Health Status Model, the indicators cover a broad range of health determinants, including traditional factors such as health outcomes, risk factors and behaviors, environmental factors, delivery of health services, and access to care, as well as more recently recognized factors such as poverty and insurance status. The breadth of indicators makes it likely that the set reflects the factors most important to shaping the health of diverse communities across the Nation.
In this candidate set of leading health indicators, the HHS working group proposes several modifications to the Committee for Objective 22.1's work. The candidate set includes 17 of the health status indicators and 12 of the priority data needs, for a total of 29 measures (see Table 1). Only one of the original health status indicators, total deaths, is not reflected in this set. In reviewing the priority data needs, the working group found that although the 16 measures are important from a public health perspective, only 12 have reliable sources of data at the State level. These 12 measures are included in the candidate set. Data for racial and ethnic subgroups of the population are available for most of the indicators.
The principal drawback to the Health Status Model relates to the number of indicators in the set. Having too many indicators presents challenges both for meaningful translation to the public and for priority setting for policy.
Health Disparities Model
The health disparities model would focus on the overarching goal of Healthy People 2010 of "eliminating health disparities," perhaps the most vital health challenge to the Nation at this point. HHS staff who work on Healthy People have identified health disparities among racial and ethnic groups, age groups, economic groups, disability groups, gender groups, and geographic locations as a primary challenge for the next decade.
The health disparities model would identify a set of leading health indicators which, according to current data, represent for each focus area of Healthy People significant disparities in baselines between the general population and some particular special population group, defined by age, race, ethnicity, gender, disability, socioeconomic status, or geographic location (such as rural-urban); in developing an initial approach to this set (see table 2), the working group identified Healthy People 2000 objectives from the 1995 Midcourse Review for which data indicated differences of 25 percent or greater in the baselines between the general population and a special population group. The final set of objectives could be selected from recommendations emanating from the focus area working groups or could be selected based on criteria related to the magnitude of the disparities.
|
Table 2 Increase Years of
Healthy Life Promote Healthy
Behaviors Protect Health Assure Access to
Quality Health Care Strengthen Community
Prevention |
* High/low gap: difference between the special
population group defined by age, race, ethnicity, gender, disability,
socioeconomic status, or geographic location with the best health experience
and that with the worst health experience for the indicator.
** Parentheses indicate the Healthy People 2000 objective number for the indicator.
Discussion: The candidate set shown in table 2 includes 32 objectives. The final list might be selected to include at least one indicator for each focus area of Healthy People 2010 and to include examples of compelling areas of disparity for each of the special populations. This would have the advantage of displaying both the scope of the Healthy People 2010 initiative and the nature of the special population groups requiring special attention. Tracking this set of health indicators would facilitate mobilizing action toward the elimination of the disparities, as well as serving as an indicator of extent to which specific special populations were benefiting from the Healthy People 2010 intervention strategies. The Health Disparities Model has some of the same advantages as the Health Status Model--interpretable to health professionals, uses available data sources, blends a variety of approaches to health improvement, and is linked directly to the objective process. Like the Health Status Model, the relatively large number of indicators presents a challenge to translation for the public. However, the impressive differences between population groups would provide "newsworthy items" which facilitate keeping the issues in the forefront of public interest. An important barrier to its local applicability is the limited availability of data at the subnational level to track progress for special populations in each of these dimensions. A set of leading health indicators oriented primarily around the problems of those special populations with the greatest disparities in health status should maintain the focus of the public upon those problem areas which should have the highest priority in the allocation of available resources. A drawback of this model is that impressive health improvements in the general population could be overlooked if little or no progress is made on disparities.
Summary Measures/Leading Contributors Approach
The summary measures/leading contributors approach would establish as leading health indicators a combined set focused both on key contributors to health and on aggregate measures that reflect the results for Americans in terms of health disparities and of quality and length of life. Analyses indicate that over half of all deaths that occurred in 1990 could be attributed to nine factors which might be considered as important root determinants of death and disability--tobacco, diet and activity patterns, alcohol, microbial agents, toxic agents, firearms, sexual behavior, motor vehicles, and illicit use of drugs. Because of the importance of focusing public and policy attention on issues to which interventions can best be directed, the candidate set of leading health indicators shown in table 3 is based in part on the leading contributors notion and in part on summary indicators linked to the major goals of Healthy People 2010: increase the years of healthy life, and eliminate the disparities in health.
Discussion: Combining these two approaches builds directly upon important purposes of the Healthy People process: achieving the goals of improved quality of life and reduced health disparities, and focusing attention on actions and policies which can alter future health. With respect to the summary measures, one of the overall goals of Healthy People 2000 is to increase the years of healthy life. To measure progress toward the goal, the National Center for Health Statistics developed a quality-adjusted life years metric (i.e., years of healthy life) that combines mortality (quantity of life) and morbidity and disability (quality of life) into a single measure.
* Parentheses indicate related Healthy People
2000 objective or chapter.
|
Table 3 |
|
|
Summary Measures
|
Leading Contributors 1.
Cigarette
smoking prevalence (3.4)* 2.
Overweight
prevalence (2.3) 3.
Sedentary
pattern prevalence (1.5) 4.
Alcohol
misuse (4.1) 5.
Immunization
rates (20.11) 6.
Air
quality exposure (11.5) 7.
Firearm
deaths (7.3) 8.
Condom
use rates (19.10) 9.
Motor
vehicle deaths (9.3) 10.
Illicit
drug use (chp. 4) 11.
Depression
(6.15) 12.
Population
uninsured 13.
High
school graduation rates (8.2) 14. Children living in poverty |
The summary measures approach presented in table 3 suggests using, as convenient and currently available measures, data on years of potential life lost, hospital days per 100,000 population, and reported disability levels--in each case for both the general population and ethnic minorities--but there are numerous alternatives:
Each of the options will require considerable developmental work and methodological research to determine the robustness and sensitivity of the candidate measures. NCHS will begin developmental work on a summary measure for Healthy People 2010 in mid-1998, when it convenes a conference of experts to resolve the data collection and methodology issues involve
in constructing summary measures of population health. The conference will build on the Institute of Medicine's workshop on summary measures of population health status, held in December 1997, and should provide direction for further development of summary measures for Healthy People 2010.
With respect to the leading contributors component of table 3, the purpose of these indicators is to show changes in health prospects. The set addresses health concerns which are subjects of intervention programs mounted by public and private sectors, as well as individual efforts. Data for a number of the indicators are available on an annual basis and at the State level (through the Behavioral Risk Factor Surveillance System). Similar data are available through the Youth Behavioral Risk Factor Surveillance for many States. The highest quality data sources for some of the indicators may reside in agencies outside the health sector at the national, State, and local levels (e.g., justice, education, transportation, environment, and housing).
The obvious advantage of a summary measure is the ability to provide a single summary figure of population health that can easily be compared to other groups or geographic areas. The disadvantage is they often do not provide enough specificity as to what is influencing changes in the health of a population and therefore can be difficult to interpret. Even when single summary measures are used, they usually need to be accompanied by a range of measures that address specific components of the health of the population--hence, the addition of the leading contributors in the approach described here. While the combination of these two models offers a means of tracking both determinants and their results for health status, left out are certain dimensions of substantial interest to various parts of the population--for example, rates of death from problems like heart disease, cancer, stroke, and HIV infection or service delivery issues like mammography, cholesterol screening, or prenatal care. There may be substantial challenge to keeping people's attention in the absence of indicators such as these.
Each of the three approaches described above focuses on the model for selecting indicators. The three approaches assume that results will be released on an annual basis. An alternative approach, presented below, focuses instead on release strategies (e.g., monthly, semiannual, or quarterly); the choice of indicators follows from the release strategy. These models place a premium on communication.
Monthly Report Approach
The monthly report approach would establish as leading health indicators a small set of 12, to be reported in seriatim at a set time each month, chosen for their ability--individually and collectively--to capture the blended issues of centrality to the health prospects of the citizenry, capacity and performance of the health system, and concern and interest to the public. The items indicated as examples reflect a mixture of outcomes (e.g., infant mortality, cardiovascular diseases, cancer) and determinants (e.g., obesity, tobacco, alcohol and other drug use), as well as acute (e.g. homicides, injuries, emerging infectious diseases) and chronic (e.g., HIV, mental health issues, depression, dementia) conditions. In addition, the approach includes annual reporting on a set of summary measures related to quality-adjusted life years and disparities among population groups (table 4).
Discussion: The principal assumptions for this approach are that 1) a smaller set will provide an easier focus for keeping the population's attention; 2) the repetition of regular, monthly foci over a span of years will help educate the public about issues of importance and help drive policy attention and follow-up; and 3) the fact that the scheduled roll-out date for the release of national (and State and local) data on a given topic is set so far in advance will help the national media prepare their stories well in advance, thus improving both the quantity and the quality of the press coverage, as well as forcing the consideration of policy initiatives in the area. Selecting such a short list from among the many candidates will leave out a number of issues of both interest and importance, such as motor vehicle and other unintentional injuries, stroke deaths, diabetes prevalence, asthma, environmental health, the share of the population uninsured, and mammography rates. Although elements of these issues are captured in the list presented, and there would be many opportunities to highlight them both in the monthly sessions and in individual releases, not including them in the annual schedule would raise questions about their priority. Reporting data on a monthly basis presents a substantial challenge. Monthly estimates of mortality and natality are published, but very little detail is provided about the events, and there are important methodologic issues that need to be considered (e.g., seasonality of deaths and certain causes of death).
|
January |
Infant mortality (14.1)* |
|
* Parentheses indicate related Healthy People 2000 objective or chapter. |
|
Quarterly/Semiannual Report Approach
The quarterly/semiannual report approach would establish as leading health indicators a number of measures which could be collected regularly enough to be provided as a quarterly (or semiannual) package to provide some insights into the experience of the previous 3 months with respect to key health issues (table 5). These measures could reflect experience with respect to key changes in health status (e.g., infant deaths, premature heart disease deaths, cancers, work-related deaths, homicides, suicides), in health risk (e.g., cigarette sales, alcohol- and/or drug-related emergency room visits, births to mothers under age 18, new HIV cases, food- and water-borne illnesses, population exposed to poor air quality), and in health services (e.g., hospital admissions, births without first trimester care, hospitalizations for asthma, new measles cases, new tuberculosis cases, population without health insurance, persons losing health insurance during the quarter). As quarterly or semiannual releases, they could be compared with the experience of both the previous period, much like corporate performance indices are reported.
NCHS may be able to provide preliminary data on mortality and natality as well as some results from the National Health Interview Survey on a semiannual basis.
|
Table 5 |
|
Quarterly/Semiannual
Indicators of Health Status Quarterly/Semiannual
Indicators of Health Risk Quarterly/Semiannual
Indicators of Health Services |
|
* Parentheses indicate related Healthy People 2000 objective or chapter. |
Discussion: The notion of releasing quarterly the results for a common set of indicators has advantages related to periodicity, pacing, and consistency which can enhance educational value over time, as well as potentially prompt policy action. The example set of semiannual indicators presents sufficiently few indicators that the issues should be relatively accessible to both the public and the media trying to interpret the quarterly results. Disadvantages reside in the fact that with a small number of measures, certain important issues might be overlooked (e.g., overweight and physical activity patterns); the fact that, given the need for data reflecting subannual intervals, some important issues (e.g., tobacco use) depend upon surrogate measures for reporting (cigarette sales); the necessity to ensure that the data sources are indeed available at the intervals required; the potential for seasonal variability to complicate interpretation; and the reliability on a quarterly basis of data pulled out of databases designed for annual interpretation and use and, hence, are more susceptible to greater fluctuations than are meaningful or understandable as actual trends.