age adjusted rates r.m. pandey additional professor department of biostatistics all india institute...
DESCRIPTION
Employment statusPopulation Impairment Rate per 1,000 Currently employed 98, Currently unemployed 7, Not in the labor force 56, Total163, Example: Hearing Impairment Due to Injury in 17+ persons Conclusion: compare the crude rates between the groups ? Is the conclusion valid ?TRANSCRIPT
Age Adjusted Rates
R.M. PandeyAdditional Professor
Department of BiostatisticsAll India Institute of Medical Sciences
New Delhi
Crude Rates:
• a single number computed as a summary measure for an entire group
• disregards differences caused by age, sex and other characteristics
Adjusted Rates:
• a single number obtained by adjusting for imbalance in age, sex and other characteristics while comparing a rate for two or more groups
Employment status Population Impairment Rate per 1,000
Currently employed 98,917 552 5.58
Currently unemployed 7,462 27 3.62
Not in the labor force 56,778 368 6.48
Total 163,157 947 5.80
Example: Hearing Impairment Due to Injury in 17+ persons
Conclusion: compare the crude rates between the groups ?Is the conclusion valid ?
The groups may differ substantially with respect to age, sex and other important characteristics
e.g.: two populations from different geographical areas:
– (A) composed of entirely males; (B) entirely females
- Can never be sure whether the difference in mortality is due to location or due to gender.
Such a situation, gender is referred as a confounder, because it associated with with both location and death rate.
It obscures the relationship between location and death rate.
Age
Currently Employed Not in labor Force
Population % Population %
17-44 67,987 68.7 20,760 36.6
45-64 27,592 27.9 15,108 26.6
65+ 3,338 3.4 20,910 36.8
Total 98,917 100.0 56,778 100.0
Compare age structure of the two populations:
Age Population Impairments Rate per 1,000
17-44 94,930 441 4.65
45-64 43,857 308 7.02
65+ 24,370 198 8.12
Total 163,157 947 5.80
Age-specific impairment rates:
5.80 is the weighted average of the age specific rates
Age is a confounder in the relationship between hearing impairment and employment status
AgeCurrently Employed Not in Labor Force
Popn. Impaire-ments
Rate/1000
Popn. Impaire-ments
17-44 67,987 346 5.09 20,760 80
45-64 27,592 179 6.49 15,108 117
65+ 3,338 27 8.09 20,910 171
Total 98,917 552 5.58 56,778 368
Rate/1000
3.85
7.74
8.18
6.48
More accurate comparison: age-specific comparisons
AgeCurrently Employed Not in Labor Force
Popn. Impaire-ments
Rate/1000
Popn. Impair-Ements
17-44 67,987 346 5.09 20,760 80
45-64 27,592 179 6.49 15,108 117
65+ 3,338 27 8.09 20,910 171
Total 98,917 552 5.58 56,778 368
Rate/1000
3.85
7.74
8.18
6.48
More accurate comparison: age-specific comparisons
Crude rate provided an incomplete picture of the true situation
Subgroup specific rates:
• More accurate comparison than crude rates
• We would end up with overwhelming number of rates to compare
More Convenient:
Summarize an entire situation with a single number calculated for each subpopulation, a number that adjusts for difference in composition
Two Ways:
1. Direct method of standardization
2. Indirect method of standardization
Direct Method of Standardization:
Step 1: Select the standard population.
Step 2: compute the expected events that would result if , instead of having different age distributions, all
populations were to have same standard age structure
Step 3:Compute the adjusted rate as total expected events in the group divided by the total standard
population
Total Currently Employed Not in Labor Force
Age (1) (2) (3) (4) (5)StandardPopn.
Rate/1000 Expected Impairm.
Rate/1000 ExpectedImpairm.
17-44 94,930 5.09 483.2 3.85 365.5
45-64 43,857 6.49 284.6 7.74 339.5
65+ 24,370 8.09 197.2 8.18 199.3
Total 163,157 965.0 904.3
Age adjusted impairment rate for:
Currently employed = 965.0/163,157 = 5.91 per 1,000
Not in the labor force = 904.3/163,157 = 5.54 per 1,000
Indirect Method of Standardization:
Step 1: use a set of standard age-specific rates along with the actual age composition of each population
Step 2: compute the number of events that would have occurred in the two groups if each took on the age specific rates of the standard population while retaining
its own age distribution
Step 3: compute standardized event ratio as observed/expected events for each group
The indirect method often concludes with this ratio.
Step 4: actual age adjusted rates for each group
= event rate in the standard population x standardized event ratio of the group
Total Currently Employed Not in Labor Force
Age (1)
(2) (3) (4) (5)
Rate/1,000
Population Expected Impairm.
Population ExpectedImpairm.
17-44 4.65 67,987 316.1 20,760 96.545-64 7.02 27,592 193.7 15,108 106.165+ 8.12 3,338 27.1 20,910 169.8
Total 5.80 98,917 536.9 56,778 372.4
Standardized event rate
Currently employed = 552/536.9 = 1.03 = 103%
Not in the labor force = 368/372.4 = 0.99 = 99%
Age adjusted rates
5.08 x 1.03 = 5.97
5.08 x 0.99 = 5.74
Uses of Standardised Rates
Smoking Group
Death rate per 1,000 person years
A B C
Non smokers 20.2 11.3 13.5
Cigarettes 20.5 14.1 13.5
Cigar/pipes 35.5 20.7 17.4
Smoking Group
Mean age(years)
A B C
Non smokers 54.9 49.1 57.0
Cigarettes 50.5 49.8 53.2
Cigar/pipes 65.9 55.7 59.7
Smoking Group
Death rate per 1,000 person years
A B C
Non smokers 20.2 11.3 13.5
Cigarettes 28.3 12.8 17.7
Cigar/pipes 21.2 12.0 14.2
Taking age group of non-smokers as standard
• Must know when to use an adjusted rate rather than crude rate
• If no confounders: the crude rate is adequate
• If confounders present: subgroup specific rates are sufficient
• Adjusted rates should be considered if they are meaningful
• If distribution of standard population is radically different than the populations being compared, standardization is inappropriate
• Also, when direct standardization is applied, subgroup specific rates should have same general trends in all the groups being compared as well as in the standard population
•Direct method of standardization is used more frequently than indirect method
• Direct method requires subgroup specific rates for all popns.
• Application of either method should lead to same conclusion