Biasing factors in the measurement of trends in heroin use.

A number of articles concerned with the epidemiology of heroin use which have appeared recently report that there has been a major "epidemic" of heroin use in the United States in the past decade and that the epidemic has been subsiding since about 1970. An examination of a number of biases inherent in the indirect measures of heroin use with which these conclusions have been reached and a growing body of direct, population-wide survey data indicates that these conclusions are not wholly warranted, and that if there was any epidemic of heroin use at all it was probably rather modest in intensity.