A comparison of standardized and proportional mortality ratios

Abstract

Proportional mortality analyses are traditionally considered to be unreliable because they lack information on persons at risk. Standardized mortality ratios (SMRs) are often used in preference to proportional mortality ratios ( PMRs ) even when the denominator or numerator of rates is known to be biased. Examination of data from 30 randomly selected occupational units described by the U.K. Office of Population Censuses and Surveys ( OPCS ) revealed, however, that age-standardized cause-specific SMRs and PMRs have an almost constant relationship: the ratio of the cause-specific PMR closely approximating the all-cause SMR of the group under consideration. Hence, a PMR above 100 almost always indicates that the corresponding cause-specific SMR is greater than the all-cause SMR (and vice versa). Furthermore, approximately 70 per cent of conditions with significantly high PMRs above 200 have corresponding SMRs which are also significantly high. When cautiously interpreted, the PRM may, therefore, be a useful indicator of an increased frequency of disease in a particular occupational or other group.