Direct and Indirect standardisation. Flashcards
If a rate is described as ‘crude’ what has not happened to it?
It has not been adjusted.
What makes it difficult to compare crude rates of mortality between populations?
Factors related to mortality may be distributed differently between the two populations.
Apart from standardisation how can you compare two crude rates that have different distributions of underlining factors?
Comparison of stratum specific rates.
What are the steps of direct standardisation (3 steps) ?
- Select a standard population.
- Obtain the proportion of person-time (or people) in each stratrum of the selected standard population.
- Compute a weighted average of the stratum–specific measures in each population in order to combine stratum specific rates into a single measure.
In order to carry out direct standardisation what do you need to know about the standard and studies population?
- The distribution of the variable in the standard population.
- Stratum specific rates of the study populations.
The sum of what produces the incidence rate (in direct standardisation)?
(Rate in stratum l in population P)(Proportion of person time in stratum l in the standard population S). Sum for all stratums.
The sum of what produces the incidence proportion/risk (in direct standardisation)?
(Risk in stratum l in population P)(Proportion of person time in stratum l in the standard population S). Sum for all stratums.
Why do you not need to state a confidence interval in a standardised rate calculation?
As all the data has been included.
Although in theory any population can be used in direct standardisation, what population(s) should you use?
Standard populations that include all study populations.
Where can you obtain artificially created populations to use for direct standardisation?
WHO.
What is the standard mortality ratio?
SMR= Observed events/ Expected events (if the study population had the same rates as the standard)
What do you need to know to carry out indirect standardisation?
- Known stratum specific rates in the standard population.
2. Known distribution of the variable in the study populations.
What does a SMR of 1 mean?
There is no difference in observed events/expected events between the standard population and study populations.
What does a SMR of >1 mean?
Less events in the study population than expected.
What does a SMR<1 mean?
More events in the study population than expected.
When would you use indirect standardisation over direct standardisation?
When the number of events are low.
Does direct or indirect standardisation apply the stratum specific rates to the study population from the standard population?
Indirect.
Does direct or indirect standardisation apply the stratum specific rates to the standard population from the study population?
Direct.
What can you not investigate with standardisation?
The factor you have standardised for.
Standard rates are summary measures. This can mask what in your data?
Trends/ differences in stratum specific rates which could be interesting.
What type of standardisation allows for multiple comparisons to be made?
Direct.
Do you need to know the overall crude rate of the entire population in direct standardisation?
No. This is an advantage of direct standardisation.
What are two disadvantages of direct standardisation?
- Changing the standard population might change the conclusion made.
- High numbers of events required for stable age-specific rates.
You do not need to know the overall crude rate in direct standardisation, however you do in indirect standardisation. What do you not need to know in indirect standardisation?
Stratum specific counts in standard population.
Can indirect or direct standardisation be estimated with poisson regression?
Indirect.
Why is indirect standardisation preferable when populations are small?
Greater precision than direct standardisation. Lower counts can result in unstable estimates.
Is it easier to compare SMR’s with direct or indirect standardisation?
Direct.
When are standardised rates useful?
When you want to make comparisons between populations or across time periods.
Why can using the same standard populations consecutively over many years be problematic?
The standard populations may change over time which would result in the different rates.
Standard populations should represent all study populations equally. True or False?
False. It can represent them differently as long as it makes sense.
