In the old stratum IRR I 1 I 00040002 2 Comparing Population B to the non plant

In the old stratum irr i 1 i 00040002 2 comparing

This preview shows page 20 - 24 out of 29 pages.

In the old stratum, IRR = I 1 /I 0 = 0.004/0.002 = 2 Comparing Population B to the non-plant workers In the young stratum, IRR = I 1 /I 0 = 0.005/0.0005 = 10 In the old stratum, IRR = I 1 /I 0 = 0.004/0.002 = 2 Plant A Plant B Non-Plant Workers Person- Years Cases Rate Weight Person -Years Cases Rate Weight Person- Years Cases Rate Weight Young 10,000 50 0.005 0.91 1,000 5 0.005 0.09 100,000 50 0.0005 0.33 Old 1,000 4 0.004 0.09 10,000 40 0.004 0.91 200,000 400 0.002 0.67 11,000 11,000 300,000 Rothman, K. J. 1986. Modern Epidemiology. Boston: Little, Brown. p.42-49
Image of page 20
Harvard TH Chan School of Public Health EPI202 – Epidemiologic Methods II Fall 2015 21 Aluminum Plant Exposure and Cancer Mortality Direct Standardization (SRR and SRD) Compare Plant A workers to Non-Plant Workers SRR A = 2.89 SRD A =0.0028 cases/person-year Compare Plant B workers to Non-Plant Workers SRR B = 2.89 SRD B =0.0028 cases/person-year Plant A Plant B Non-Plant Workers Person- Years Cases Rate Weight Person -Years Cases Rate Weight Person- Years Cases Rate Weight Young 10,000 50 0.005 0.91 1,000 5 0.005 0.09 100,000 50 0.0005 0.33 Old 1,000 4 0.004 0.09 10,000 40 0.004 0.91 200,000 400 0.002 0.67 11,000 11,000 300,000 Aluminum Plant Exposure and Cancer Mortality Indirect Standardization (SMR and SMD) Compare Population A to Non-Plant Workers SMR A =7.71 SMD A =0.0043 cases/person-year Compare Population B to Non-Plant Workers SMR B =2.20 SMD B =0.0022 cases/person-year Plant A Plant B Non-Plant Workers Person- Years Cases Rate Weight Person -Years Cases Rate Weight Person- Years Cases Rate Weight Young 10,000 50 0.005 0.91 1,000 5 0.005 0.09 100,000 50 0.0005 0.33 Old 1,000 4 0.004 0.09 10,000 40 0.004 0.91 200,000 400 0.002 0.67 11,000 11,000 300,000
Image of page 21
Harvard TH Chan School of Public Health EPI202 – Epidemiologic Methods II Fall 2015 22 Question… Aluminum Plant Exposure and Cancer Mortality Comparing Populations The SRRs used the common standard of the non-plant workers, so the rate ratios can validly be compared. When the age-specific incidence rates for the two cohorts are the same, the two SRRs will be the same no matter what the weights of the common standard. On the other hand, even if two populations have identical stratum-specific rates and therefore their directly standardized rates are identical, their indirectly standardized rates can be quite different. The two SMRs here were computed using different standards. In the presence of effect measure modification by the standardization factor(s), SMRs are not comparable unless the distribution of the stratification factor(s) among the exposed (i.e., the weights) is the same in the two populations. If there were no effect modification, however, any weighting scheme chosen, including the weighting scheme employed in indirect standardization, would yield estimates which would be comparable across groups.
Image of page 22
Harvard TH Chan School of Public Health EPI202 – Epidemiologic Methods II Fall 2015 23 Direct vs. Indirect Standardization Advantages and Disadvantages Direct (SRR) Requires subgroup-specific cumulative incidence/rate Problematic when stratum-specific rate estimates are imprecise or unknown (e.g. company records only include total cases and person- time distribution) Allows for comparison across study populations standardized to the same standard population Indirect (SMR) Often used in occupational epidemiology Particularly useful when stratum-specific cumulative incidences/rates are missing in one of the groups under comparison
Image of page 23
Image of page 24

You've reached the end of your free preview.

Want to read all 29 pages?

  • Fall '16
  • MURRAY
  • Epidemiology, TH Chan School of Public Health, Harvard TH Chan, TH Chan School

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture