 |
| EPI5.5: STANDARDIZATION: INDIRECT STANDARDISATION |
|
| EPI5.5: STANDARDIZATION: INDIRECT STANDARDISATION |
 |
 |
 |
 |
OBJECTIVE |
|
At the end of this session you have an understanding of indirect as opposed to direct standardisation, when it is indicated and how to do it. |
Mostly in Occupational health we are dealing with small populations. This means that the rates are not easy to observe and that they are not stable especially for rare conditions. Small numerators and denominators will mean that even a change of 1 or 2 will make a big difference to the rate and give rise to instability.
Hence the approach taken is called indirect standardisation.
Here the reference population supplies standard rates which are then applied to the two DIFFERENT population structures of Populations 1 and 2. This allows us to recalculate the expected number of cases that would have been observed in Populations 1 and 2 had the rates from the standard population been operating in those populations.
In Table 2 the rate in the young (0.0005, in red) is multiplied with the number of person-years in Population 1 (10 000 in red) and then again with Population 2 (1000 in red) to yield the expected numbers of cases in each population for each age stratum (in blue). These are added for Populations 1 and 2 to yield the total numbers expected and these are divided into the total numbers observed. This yields the standardised mortality (SMR) or morbidity ratio. (This is usually multiplied by 100 to give 771 and 220).| Age Group | Reference Population | Index Population 1 | Index Population 2 |
|---|---|---|---|
| Young | |||
| Cases | 50 | 5 50 | 0.5 5 |
| Person years | 100 000 | 10 000 | 1 000 |
| Rate | 0.0005 | 0.005 | 0.005 |
| Old | |||
| Cases | 400 | 2 4 | 20 40 |
| Person years | 200 000 | 1 000 | 10 000 |
| Rate | 0.002 | 0.004 | 0.004 |
| Crude rates | |||
| Cases | 450 | 7 54 | 20.5 45 |
| Person years | 300 000 | 11 000 | 11 000 |
| Rate | 0.0015 | 0.005 | 0.004 |
| Age adjusted or standardized rates | 0.0015 | 0.004 | 0.004 |
| Standardized rate ratio (direct standardization) |
0.004/0.004 = 1 | 0.004/0.004 = 1 | |
| Standardized mortality ratio (indirect standardization) |
54/7 = 7.71 | 45/20.5 = 2.2 | |
In Table 2 the rates from the age strata of the reference population (0.0005 and 0.002) (in red) are multiplied with the number of person-years in these same strata in each of the two Index Populations.
(0.0005 x 10000) + (0.002 x 1000) = 5 + 2 = 7 expected cases in Population 1 and 20.5 in Population 2.
Dividing observed (54) by expected cases (7) yields 7.71 and 45 by 20.5 = 2.2.
These are the SMRs or the standardised mortality or morbidity ratios. (This is usually multiplied by 100 to give 771 and 220).
Note now that the SMR for Population 1 is not the same as the SMR for Population 2. However, both are greater than than 1 (or 100) indicating that the mortality is higher than the reference population when comparing either Population 1 or Population 2 with that reference population. It is not however possible to make a comparison directly between Population 1 and Population 2. The reason for this is that each of these populations age structures have been used in separate analyses to directly standardise the rates in the reference population. And we know that Population 1 and Population 2 have very different age structures. This is why we can't compare the indirectly standardised measures with each other for these two populations.
