Writing in 1953, Clifford pointed out that neonatal mortality rates in the United States were still far from an "irreducible minimum" and that, in the experience of many teaching medical centers, rates of 14 per 1000 live births were attainable.1
Hospital obstetrics and nursery statistics appear to offer intriguing indices of medical care. However, as Silverman et al. have pointed out, there are hazards in drawing conclusions from comparative hospital statistics without taking into account the effect of variables intrinsic to the population served.2 They suggested a method of computing death rates for premature nurseries which standardizes the factors of birth weight, length of gestation, sex, and race. Of these variables, birth weight and race are probably the most important. If the number of births in groups to be compared is large enough, sex and length of gestation are not likely to vary significantly. Social class may be