which such a determination can be based are limited. The last systematic attempt to estimate the prevalence of tuberculin reactivity in the United States was that of the National Health Survey of 1971–72, which concluded that in American adults aged 25 to 74 years the prevalence of tuberculin reactivity was 21.5 percent (12). A reasonable estimate for the current date for Americans of all ages might be 5 to 10 percent, and perhaps 10 percent in adults.
Tuberculous infection is not uniformly distributed among Americans, and it is important to stratify any assessment of the general population risk so that infection in health care workers can be compared with that in the appropriate reference community. For example, tuberculosis is much more common in urban areas than in rural areas, and even in the cities of middle America it is not as frequent as in this country’s coastal cities. Many demographic factors correlate with tuberculosis incidence in America. Health care workers are employed, while nearly 60 percent of tuberculosis in the United States occurs among the unemployed (1). Importantly for this consideration, a substantial number of health care workers are foreign-born, one-third of them coming from the Philippines; among other employed persons, one-quarter come from Mexico (A. Curtis, personal communication of material presented to a workshop held in December 1999). The importance of this difference rests with the difference in tuberculosis in the countries of origin for these groups. Based on recent World Health Organization estimates, the current incidence of sputum smear-positive tuberculosis in the Philippines is 260/100,000, and that in Mexico is 58/100,000 (13).
There are relatively few tuberculin surveys available from which one can calculate the ARI in American populations and none in recent years. Moreover, those surveys that have been conducted have often been flawed by the use of poorly standardized tuberculin testing techniques and by poor characterization of the populations studied, especially with respect to demographic characteristics. The use of tuberculins other than purified protein derivative (PPD) at 5 tuberculin units may lead to an overestimation of the actual prevalence of tuberculin reactivity. Data from 12 selected surveys conducted in the United States during the middle half of the 1990s are presented in Table C-1. These studies rarely involved serial testing or testing of more than one age group. In that situation, the calculation of ARI for Table C-1 was done from birth, assuming a reactor rate of zero at birth, a maneuver admittedly flawed because it assumes the risk to be uniform throughout life. The error thus introduced has the potential for underestimating the adult risk relevant to health care workers.