rest on all who have responsibility to act. For there to be true accountability, performance must be monitored.
The workshop discussions suggested that performance indicators are used for multiple purposes: to identify problems and generate hypotheses, as political tools for mobilization in the community, to suggest ideas for improvement, and in fact, to monitor the performance of specific sectors of the community. Among the characteristics of useful performance indicators is a focus on populations and rates, rather than on absolute numbers of contacts involved in the interventions. In other words, the denominator is as important as the numerator.
Indicators were described as most useful when they focus on areas where improvement is possible. Global health status indicators often have little practical use for guiding health improvement strategies. More useful are indicators that incorporate a "theory of improvement"—that is, they suggest a clear means of moving from measurement to action. Indicators that have been shown to change in intervention studies should be preferred over those that may be more conceptually elegant but may not be able to capture the impact of an intervention. "Responsive" indicators of this sort allow real change to be distinguished from random variation.
The value of standard epidemiologic health needs assessments was questioned. Often, needs assessments merely document problems that are already well known. On the other hand, needs assessments focused on factors in the community that influence program implementation—politics, resources, barriers, key players—may be very useful. The speaker also emphasized that although coalitions are an essential component of community-based health improvement projects, they can consume substantial resources. Participants may, for example, spend an average of 3–4 hours a month conducting coalition-related work. It has also proved difficult to document a relationship between the characteristics of coalition operations and health outcomes achieved. The contributions that coalitions make to health improvement activities need to be better understood.
The committee was urged to articulate a model (or models) of health improvement that specifies use of performance indicators and holds social and nonclinical improvement strategies to the same evidence base as clinical strategies. Such a model should