of managed care organizations, and representatives of the prenatal care providers should also be involved.

Surveillance systems are needed to support policy development and program evaluation regarding perinatal HIV transmission. Chapter 3 of this report illustrates how epidemiologic surveillance data can help to focus attention on critical dimensions of a public health problem. Such analyses are hampered, however, by the lack of national HIV prevalence data and the discontinuation, in 1994, of the Survey of Childbearing Women. Chapter 6 shows how data from provider and patient surveys, clinical and health plan records, and other sources such as birth certificates can be used to monitor the performance of providers and identify bottlenecks in prevention activities. Data of this sort, however, are not universally available, and often are defined differently from one population to another. Thus, in order to support the previous recommendation about performance measures, and to generally guide prevention efforts,

The universal testing approach that the committee recommends, as well as the call for health plan performance measures, should facilitate the development of appropriate public health surveillance systems.

The Ryan White CARE Act Amendments of 1996 could make it difficult to maintain the recommended surveillance system. The Act states that continued federal funding to the states could be contingent upon (see Chapter 1):

1. a 50% reduction (or a comparable measure for states with less than ten cases) in the rate of new AIDS cases resulting from perinatal transmission, comparing the most recent data to 1993 data; and
2. ensuring that at least 95% of women who have received at least two prenatal visits prior to 34 weeks of gestation have been tested for HIV.

3. The first of these measures is imprecise. Does the "rate of new AIDS cases" refer to the number per year, the proportion of all newborns with AIDS, the proportion of children born to HIV-infected mothers who have AIDS, or some other concept? Children born with HIV infection may not progress to AIDS for years, so monitoring new AIDS cases per se reflects prevention efforts far in the past. How should it be determined whether any specific case was the result of