PRINCIPLES OF TOXICOLOGY

518 EPIDEMIOLOGIC ISSUES IN OCCUPATIONAL AND ENVIRONMENTAL HEALTH
TABLE 21.4 Incidence Data: Rate Ratio and Rate Difference
Disease
+ –
+ A B
Exposure
– C D
Incidence of disease in exposed population = A/PY (person-year)
Incidence of disease in unexposed population = C/PY
Rate ratio = A/PY divided by C/PY
Rate difference = (A/PY – C/Py)
rate that is not due to the exposure. Therefore, standardization for age is necessary prior to making
comparisons of the cancer rates between the two different populations.
A variation on the SRR used extensively in Occupational Epidemiology is the standardized
mortality ratio (SMR) with regard to mortality or standardized morbidity ratio with regard to morbidity.
First, the person years of exposure in the study group are multiplied times the rate of death from
a particular disease in the reference or standard population (i.e., number of cases/number of
persons in the standard population/year) to generate an expected number of cases of death from
the disease. The SMR is the number of cases of death in the study population divided by the
number of cases expected from the standard population (SMR = observed/expected × 100). In the
example described above, the incidence rate of lung cancer in worker populations without asbestos
exposure is 1/1000 workers/year while the rate of lung cancer in an asbestos-exposed population
is 4/1000 workers/year with 1000 person-years. The expected number of lung cancer cases in the
exposed worker group, using the unexposed worker population as the reference or standard
population, is 1 (i.e., 1000 person-years × 1/1000 per year). Therefore, the SMR for lung cancer
with asbestos exposure would be 4/1 × 100 or 400. A SMR greater than or less than 100 indicates
increased or decreased risk of disease. SMRs and SRRs can be tested for statistical significance
using the χ 2 test and a confidence interval.
In occupational epidemiologic studies, the proportional mortality ratio (PMR) is frequently used
to measure the risk or association between exposure and disease. The PMR is used in settings when
only the population and the cause of death (or disease) are known, not the number of person-years of
exposure. In a PMR study, the expected number of deaths due to a particular disease is computed using
the proportion of these deaths in a standard population (such as the general population), rather than
the death rate as in the SMR study. The PMR study results approximate those of a SMR study when
the particular cause of death is relatively uncommon. A PMR greater than or less than 100 indicates
an increased or decreased risk of disease or death.
In case–control studies, the total number and proportions of persons with and without a disease are
defined by the study design, rather than the true proportions in the total population. As such, usually
true rates of disease cannot be derived in case–control studies. Therefore, the rate ratio is not an
appropriate measure of risk; the odds ratio is used as an approximation of the rate ratio for the risk of
disease associated with exposure (Table 21.5). An odds ratio greater than or less than 1 indicates an
TABLE 21.5 Case–Control Data: Odds Ratio
Disease
+ –
+ A B
Exposure
–
Rate ratio = odds ratio = AD/BC
C D