however, that almost any access problem can be characterized on all three of these dimensions. (For example, there are substance abusers who require drug treatment services for their addictive disorders.) Moreover, as Aday (in press) has pointed out, many of the vulnerable populations whose access concerns us as a society have crosscutting needs. For instance, the broad group of alcohol and substance abusers can include high-risk mothers with fetal alcohol syndrome, intravenous drug users with AIDS, mentally ill substance abusers, drug users who attempt suicide, addictive families suffering domestic abuse, homeless people with substance abuse problems, and substance-abusing refugees.

Once a topic has been selected, the major challenge is to conceptualize an identifiable personal health service to serve as a utilization indicator and as an outcome measure that can be related to use or lack of use. The best utilization indicators are those for which there is a fairly well-recognized service intervention with clear guidelines regarding who should receive the service. Good outcome measures must be more than prevalence or incidence rates; they must reveal something about access. The best example is the incidence of a vaccine-preventable disease such as measles. In contrast, the incidence of colds would be difficult to relate to problems in access.

Once past the conceptual stage, it is necessary to identify a source of routine data. Potential data bases must be explored and the quality of the information assessed. As with the indicators in the initial monitoring set, a variety of problems involving data collection frequency, availability, and disaggregation must be confronted.

To illustrate the kind of analysis required, the committee commissioned papers on four of the topics: AIDS, substance abuse, and migrants and the homeless (Appendixes A-C). In fact, these papers go beyond illustration to provide the first strategic steps toward developing access monitoring indicators in these areas of interest.