FIGURE 5.2 Age of onset of DIS/DSM-III Alcohol Abuse and Dependence. Population estimates are from 125 new cases (with 20th and 50th percentiles marked). Source: NIMH Epidemiologic Catchment Area Study, Four Sites, 1978 to 1983.

on the occurrence of a handful of new cases among many thousands of persons at risk for developing the disorder over the one-year follow-up. The volatility is best demonstrated by the thin black bars, which represent age-specific rates at each site. The open rectangle is the age-specific rate for all four sites collapsed, and it shows more stability due to pooling of data. The curve itself involves the maximal smoothing of the data, using only two degrees of freedom. The cumulative form of presentation obscures this statistical volatility. The cumulative distributions presented for the five disorders below are weighted statistically to represent a population with age, gender, and race characteristics identical to those of the U.S. population (as in Robins and Regier, 1991).

The cumulative form of presentation, along with the statistical weighting, links the data to the concept of attributable risk. The cumulative form allows one to estimate the proportion of cases in the United States that would be prevented if a 100 percent effective preventive intervention were applied at or before a given age. The attributable risk proportion, which is the percentage effectiveness of the intervention in eliminating the risk factor, and the cumulative percentage drawn from the figures below, can be chain multiplied to estimate the effectiveness of a given intervention program and to compare various intervention strategies. Data on the cost of modifying risk factors could also be incorporated in such calculations in designing a preventive intervention research program.