SESI 4 - Epid-Sesi - 3 - M. Hakimi

Estimating and Comparing
Rates using Dircet and

Indirect Adjustment
Siswanto Agus Wilopo

Dept. Public Health,
UGM
Copyright © 2014 Department of Public Health sawilopo@yahoo.com 1

Topics
♦ Incidence Density
♦ Incidence Rate Difference and Ratio
♦ Confidence Intervals
♦ Standardized Rates and Their Comparison

A Definition
♦ Kleinbaum, Kupper and Morgenstern, Epidemiologic

Research: Principles and Quantitative Methods (1982),
p.97:
“A true rate is a potential for change in one quantity per
unit change in another quantity, where the latter quantity is
usually time. (…) Thus, a rate is not dimensionless and
has no finite upper bound – i.e., theoretically, a rate can
approach infinity.”

Rates
♦ A well-known example of rate is velocity, i.e., change of
distance per unit of time (given, e.g., in km/h).
• In practice, it does (should?) have an upper-bound
♦ We can talk about instantaneous and average rates.

• Example instantaneous: your car velocity at a particular time-point
(can depend on the time-point, e.g., city and highway).
• Example average: your average speed after travelling a particular
distance (assumed constant across the whole trip).
♦ In epidemiology, we usually talk about average rates.

Incidence/Mortality Rate
♦ Kleinbaum, Kupper and Morgenstern, Epidemiologic
Research: Principles and Quantitative Methods (1982),
p.100:
“The incidence rate of disease occurrence is the instantaneous

potential for change in disease status (i.e., the occurrence of
new cases) per unit of time at time t, or the occurrence of
disease per unit of time, relative to the size of the candidate
(i.e., disease-free) population at time t”.
♦ We could similarly define the mortality rate.

Incidence Rate
♦ Other terms:
• an “instantaneous risk” (or probability);
• a “hazard” (especially for mortality rates);
• a “person-time incidence rate”;
• a “force of morbidity”.
♦ It is expressed in units of 1/ time.
♦ It is sometimes confused with risk.

Rates and Risks
♦ Assume that the incidence rate is constant over time (=λ),
and the same for all individuals.
♦ The risk (probability) of developing disease in time T will

then be equal to 1-e-λT.
• Risk is sometimes called a cumulative incidence.
• In a disease-free (at time 0) cohort of N individuals, you would thus
expect N(1-e-λT) new cases after time T.
• Similarly, we could talk about the risk of death.
♦ Thus, formally these are two different quantities.

Estimating Rates
♦ Rates require observations of incidence in time. Thus,
they are estimated from cohort studies.
♦ Instantaneous rates are seldom obtained. Rather, the
average rates are computed.
♦ The most basic estimator is the incidence density (ID):
I no. of new cases in the calendar period (t 0 , t1 )

ID = =
PT accrued population time
♦ PT is expressed in person-years, person-days etc.

Incidence Density
♦ A hypothetical cohort of 12 subjects.
♦ Followed for the period of 5.5 years.
♦ 3 withdrawals among non-cases
• three (7,8,12) lost to follow-up;
• two (3,4) due to death;
• two (5,10) due to study termination.
♦ PT = 2.5+3.5+…+1.5 = 26.
♦ ID=5/26=0.192 per (person-) year
or
1.92 per 10 (person-)years.

Population-Time Without Individual Data
♦ E.g., population-based registries.
♦ Person-years computed using the mid-year population.
♦ For rare events, periods of several years may be used.

• Ideally, one would like to use mid-year populations for each year.
• Alternatively, one can use information for several time-points, or the
mid-period population (these are less accurate solutions).
One may face the problem of removing those not at risk (e.g., women for
prostate cancer incidence).

Population-Time Without Individual
Data: Example

Incidence Density: Remarks
♦ It is an estimate of an average rate.

• So we will sometimes refer to it as an “incidence rate”.
♦ Any fluctuations in the instantaneous rate are obscured and

can lead to misleading conclusions. E.g.,
• 1000 persons followed for 1 year
• 100 persons followed by 10 years
produce the same number of person-years.
If the average time to disease onset is 5 years, ID in the first
cohort will be lower.

Incidence Density: Remarks
♦ If applied to the whole cohort/population, sometimes called
crude rate.
♦ However, sex, age, race etc. can have substantial influence

on the incidence of disease.
♦ Comparing crude rates for two populations, which differ

w.r.t., e.g., age, can be misleading (confounding!).
♦ Therefore, usually standardized rates are compared.

• E.g., for cancer, age- and sex-standardized rates are used.
• They will be discussed later.
Confidence Interval for Incidence
Density
♦ By using a Poisson model, standard error of ID=I / PT can
be estimated by:
I
SE( ID) =
(PT )2
♦ Thus, an approximate 95% CI for ID is given by:
ID ± 1.96·SE(ID).
• 99% CI: ID ± 2.58·SE(ID) .

Estimating Incidence Densities: Example
♦ Postmenopausal Hormone and Coronary Heart Disease Cohort Study:
• Stampfer et al., NEJM (1985).
• Involving female nurses:
Hormone use
Yes No Total
CHD 30 60 90
Person-years 54308.7 51477.5 105786.2
• ID1 = 30/54308.7 = 0.00055; SE(ID1) = (30/54308.72)1/2 = 0.00010

• 95% CI for ID1 = 0.00055 ± 1.96·0.0001 = (0.00035, 0.00075)
• ID0 = 60/51477.5 = 0.00116; SE(ID0) = (60/51477.52)1/2 = 0.00015
• 95% CI for ID0 = 0.00116 ± 1.96·0.00015 = (0.00086, 0.00145)

Comparing Two Incidence Densities
Exposed Unexposed Total
♦ Assume data from a
Cases I1 I0 I
cohort study:
Pop.-time PT1 PT0 PT
♦ We get two estimates for non- and exposed subjects:

ID0=I0/PT0 and ID1=I1/PT1.
♦ To compare them, we can look at

• Incidence rate difference: IRD = ID1 - ID0 .
• Incidence rate ratio: IRR = ID1 / ID0 .

Comparing Two Incidence Densities:
Example
♦ Postmenopausal Hormone and Coronary Heart Disease Cohort Study:
• Stampfer et al., NEJM (1985).
• Involving female nurses:
Hormone use
Yes No Total
CHD 30 60 90
Person-years 54308.7 51477.5 105786.2
• ID1 = 30/54308.7 = 0.00055; ID0 = 60/51477.5 = 0.00116

• IRD = ID1 - ID0 = -0.00061
• IRR = ID1 / ID0 = 0.474

Poisson Model Method
♦ By using a Poisson model, Exposed Unexposed Total
standard error of IRD can be Cases I1 I0 I
estimated by:
I0 I1
SE( IRD) = + ♦ Standard error of ln IRR can be
(PT0 ) ( PT1 )2
2
estimated by:
♦ Thus, an approximate 95% CI 1 1

for IRD is given by: SE(ln IRR ) = +
I 0 I1
IRD ± 1.96·SE(IRD).
♦ Thus, an approximate 95% CI for IRR
• 99% CI: IRD ± 2.58·SE(IRD)
is given by:
exp{ ln IRR ± 1.96·SE(ln IRR) }
• 99% CI: exp{ ln IRR ± 2.58·SE(ln IRR) }
Example
Hormone use
60 30
SE( IRD) = + = 0.00018 Yes No Total
51477.52 54308.7 2
CHD 30 60 90
1 1
SE (lnIRR ) = + = 0.224 Person- 54308.7 51477.5 105786.2
60 30 years
♦ 95% CI for:
IRD: -0.00061 ± 1.96·0.00018 = (-0.00096, -0.00025)
ln IRR: ln(0.474) ± 1.96·0.22 = (-1.178, -0.315)
IRR: (e-1.178, e-0.315) = (0.308, 0.729)
♦ Both CIs allow to reject the null hypothesis of no difference.

“Test-Based” Method
♦ 95% “test-based” CI for IRD can Exposed Unexposed Total
be computed as Cases I1 I0 I
IRD ± 1.96 · SE(IRD), Pop.-time PT1 PT0 PT
where SE(IRD)= IRD / χ and
PT1 ♦ Similarly, SE(ln IRR)= (ln IRR) / χ
I1 − I ⋅
PT ♦ 95% “test-based” CI for ln IRR is
χ=
I ⋅ PT0 ⋅ PT1
ln IRR ± 1.96 · (ln IRR) / χ
PT 2
♦ Can be written as
♦ Can be re-expressed as
( 1 ± 1.96 / χ ) · ln IRR
(1 ± 1.96 / χ) · IRD
• 99% CI: (1 ± 2.58 / χ) · IRD
♦ 95% CI for IRR is thus
exp{ ( 1 ± 1.96 / χ ) · ln IRR}
Example
PT1
I1 − I ⋅ Exposed Unexposed Total
PT
χ= Cases I1 I0 I
I ⋅ PT0 ⋅ PT1
PT 2
54308.7 Hormone use
30 − 90 ⋅ Yes No Total
105786.2
χ= = 3.41 CHD 30 60 90
90 ⋅ 54308.7 ⋅ 51477.5
Person 54308.7 51477.5 105786.2
105786.2 2 -years
♦ 95% “test-based” CI for

IRD: (1 ± 1.96/3.41) · (-0.00061) = (-0.001, -0.0002)
Close to the one based on the Poisson approximation (not in general).
ln IRR: (1 ± 1.96/3.41) · ln(0.474) = (-1.176, -0.317)
IRR: (e-1.176, e-0.317) = (0.309, 0.728)
“Exact” Confidence Interval for IRR
♦ The presented CIs for ln IRR (and IRD) assume that the
estimates of ln IRR vary according to the normal distribution.
• Hence their form, e.g., ln(IRR) ± 1.96 · SE(ln IRR).
♦ The use of the normal distribution is an approximation.

• Can be problematic, especially in small samples.
♦ It is possible to construct a CI for ln IRR using the “exact”

distribution (i.e., without approximating it by the normal).
• The CI is valid in all samples; in large samples, it is close to the
approximate CIs.
• Computation is a bit more difficult (but easily handled by computers).
Standardized Rates
♦ We will introduce the standardization w.r.t. age.
♦ We will assume that our population is stratified by age (i.e.,

subdivided into age-groups).
• One needs to define age-groups (e.g., 0-4, 5-9,…).
♦ One needs to compute age-specific rates (ID).

• Population-time and no. of cases for each age-group are required.
♦ There are two methods of standardization:

• Direct;
• Indirect.
Standardization
♦ Direct method
• Age-specific rates of the study population are applied to the age-
distribution of the standard population
(rates study → age standard)
• Theoretical rate that would have occurred if the rates
observed in the study population applied to the standard
population.
♦ Indirect method
• Age-specific rates from the standard population are applied to the
age-distribution of the study population.
(rates standard → age study)
Direct Standardization
Age Study Population Standard Population
Group (e.g., USA 1990)
Observed Person-years Rate Observed Population Rate
<40 I1 PT1 I1/ PT1 B1 N1 B1/N1
40-64 I2 PT 2 I 2/ PT 2 B2 N2 B2/N2
65+ I3 PT 3 I 3/ PT 3 B3 N3 B3/N3
Total It PT t I t/ PT t Bt Nt Bt/Nt
♦ Crude Rate in study population = It / PTt .
♦ Directly Standardized Rate (DSR):

DSR = { (I1/PT1)N1 + (I2/PT2)N2 + (I3/PT3)N3 } / Nt
= (I1/PT1)(N1/Nt) + (I2/PT2)(N2/Nt) + (I3/PT3)(N3/Nt).
Make sure units are consistent!!!

Direct Standardization
♦ If there is no confounding, crude rate is adequate.
♦ DSR by itself is not meaningful – it makes sense only when

comparing two or more populations.
• If possible, compare age-specific rates.
• The rates should exhibit more or less similar trends (also in the
standard).
♦ DSR depends on the choice of the standard population.

• The age-distribution of the latter should not be radically different from
the compared populations.
• There are several standard populations (e.g., for the world,
continents etc.).
Indirect Standardization
♦ Direct standardization requires age-specific rates for all

compared populations.
♦ If these are not available, or they are imprecise, the

indirect method is preferred.
♦ Both should lead to similar conclusions; if they do not, the

reason should be investigated.

Indirect Standardization
Age Study Population Standard Population Expected
Group (e.g., USA 1990)
Obs Person- Rate Obs Population Rate
years
<40 I1 PT1 I1 / PT1 B1 N1 B1 /N1 E1= PT 1* (B1 /N1)
40-64 I2 PT2 I2 / PT 2 B2 N2 B2 /N2 E2= PT 2* (B2 /N2)
65+ I3 PT3 I3 / PT 3 B3 N3 B3 /N3 E3= PT3* (B3 /N3)
Total It E1+ E2+ E3=∑∑Ej
♦ Standardized (Incidence or Mortality) Ratio (SIR or SMR):

SIR = It / ∑ Ej = Observed / Expected .
♦ Take Indirectly Standardized Rate (ISR) as:

ISR = SIR ·(crude rate for the standard population).
Make sure units are consistent!!!
Standardization of Rates: Example
♦ Infant deaths (for children less than 1 year of age) in Colorado and
Louisiana in 1987.
•Colorado: 527 deaths out of 53808 life births; crude rate = 9.8 per 1000.
•Louisiana: 872 deaths out of 73967 life births; crude rate = 11.8 per 1000.
♦ Crude infant mortality rate for Colorado is lower than for Louisiana.
♦ In the US, infant mortality depends on race.
Race USA, 1987

Life %Life Infant Rate
Births Births Deaths (x1000)
Black 641567 16.8 11461 17.9
White 2992488 78.6 25810 8.6
Other 175339 4.6 1137 6.5
Total 3809394 100 38408 10.1

♦ The distribution of race of new- Race Colorado Louisiana
born children is different in the Life Birth % Life Births %
two states. Black 3166 5.9 29670 40.1
White 48805 90.7 42749 57.8
Other 1837 3.4 1548 2.1
♦ Infant mortality rates depend on Total 53808 100 73967 100
race.
• Race is a confounder.
Race Colorado Louisiana
Rate Rate
♦ Compare race-specific infant (x1000) (x1000)
mortality rates. Black 16.4 17.7
• Unclear (differences in various White 9.6 8.0
directions). Other 3.3 1.9
Total 9.8 11.8
♦ Direct standardization: apply state- and race-specific rates to the
standard race distribution (US, 1987).
Race US, 1987 Colorado Louisiana

Ni Ni / Nt Rate Rate*Ni Rate*Ni /Nt Rate Rate*Ni Rate*Ni /Nt
(Births) (x1000) (x 1000) (x1000) (x1000)
Black 641567 0.168 16.4 10521.7 2.76 17.7 11355.7 2.98
White 2992488 0.786 9.6 28727.9 7.54 8.0 23939.9 6.28
Other 175339 0.046 3.3 578.6 0.15 1.9 333.1 0.09
Total 3809394 1 39828.2 10.45 35628.7 9.35
♦ DSR for Colorado: 10.45 (per 1000 life births; crude: 9.8).
♦ DSR for Louisiana: 9.35 (per 1000 life births; crude: 11.8).

♦ Indirect standardization: apply race-specific rates of a standard
population (US, 1987) to the race-distribution of the states.
Race US Colorado Louisiana

Rate Life Births Deaths Rate*PTi Life Births Deaths Rate*PTi
(x1000) (PTi) (Obs.) (Exp. Deaths) (PTi) (Obs.) (Exp. Deaths)
Black 17.9 3166 52 56.7 29670 525 531.1
White 8.6 48805 469 419.7 42749 344 367.6
Other 6.5 1837 6 11.9 1548 3 10.1
Total 10.1 53808 527 488.3 73967 872 908.8
♦ SMR for Colorado: 527/488.3 = 1.08 (8% higher than the US).
• ISR = SMR x 10.1 = 10.9 (race-adjusted infant mortality-rate).
♦ SMR for Louisiana: 872/908.8 = 0.96 (4% lower than the US).
• ISR = SMR x 10.1 = 9.7 (race-adjusted infant mortality-rate).
♦ Is it reasonable to use the
adjusted rates?
•The plot of race-specific rates
shows similar trend
(black>white>other).
•The distribution of race in the US
is similar to the two states
(white>black>other).
•Results for both standardization
methods are similar.

Comparison of Directly Standardized
Rates
♦ If we have two standardized rates, we may want to compare them.
♦ For the direct method, assume we have DSR1 and DSR2.
♦ 95% CI can then be obtained using the normal approximation:

(DSR1 - DSR2) ± 1.96 · SE(DSR1 - DSR2) .
• 99% CI: (DSR1 - DSR2) ± 2.58 · SE(DSR1 - DSR2) .
♦ The standard error is given by
2
 Nk 
SE (DSR1 − DSR2 ) = ∑  ⋅ SE( IRDk ) 
 Nt 
where IRDk is the stratum-specific intensity rate difference.
Rates
♦Alternatively, we might look at the standardized rate ratio:
SRR=DSR1/DSR2.
♦95% CI for SRR can be written as: SRR 1 ± (1.96 / Z), where
DSR1 − DSR2
Z=
SE( DSR1 − DSR2 )
• 99% CI can be written as: SRR 1 ± (2.58 / Z ).

Rates: Example
♦ DSR1 (Colorado): 0.01045 (10.45 per 1000 life births).
♦ DSR2 (Louisiana): 0.00935 (9.35 per 1000 life births).
Race US Colorado Louisiana

%Births Births Deaths Rate Births Deaths Rate IRDi SEi
(Ni/Nt) (PTi) (Ii) (IDi) (PTi) (Ii) (IDi) (x1000) (x1000)
Black 16.8 3166 52 16.4 29670 525 17.7 -1.3 2.4
White 78.6 48805 469 9.6 42749 344 8.0 1.6 0.6
Other 4.6 1837 6 3.3 1548 3 1.9 1.4 1.7
Total 100 53808 527 9.8 73967 872 11.8
SE (DSR1 − DSR2 ) = (0.168 ⋅ 0.0024 )2 + (0.786 ⋅ 0.0006 )2 + (0.046 ⋅ 0.0017 )2 = 0.0006

Rates: Example
SE (DSR1 − DSR2 ) = (0.168 ⋅ 0.0024 )2 + (0.786 ⋅ 0.0006 )2 + (0.046 ⋅ 0.0017 )2 = 0.0006
♦ DSR1 = 0.01045; DSR2 = 0.00935.
♦ DSR1 - DSR2 = 0.0011.

• 95% CI: 0.0011 ± 1.96·0.0006 = (-0.0002, 0.002).
• CI includes 0 - we cannot reject H0 of no difference.
♦ SRR = DSR1 / DSR2 = 1.12.

• Z = (DSR1 - DSR2) / SE = 1.83.
• 95% CI: 1.12 1 ± (1.96 / 1.83) = (0.99, 1.26).
• CI includes 1 - we cannot reject H0 of no difference.
Comparison of Indirectly Standardized
Rates
♦ In directly standardized rates, stratum specific-rates for different study
populations are combined using the same weights (relative stratum-
sizes in the standard population).
♦ In indirectly standardized rates, the weights (PTi / “expected Ii”) differ.
♦ Thus, technically speaking, ISRs (SIRs) should not be compared.
♦ On the other hand, it is valid to ask whether SIR (or SMR) is different
from 1.
♦ To do that, one can construct a 95% CI, e.g., as follows:

SIR ± 1.96·(√observed events)/(expected events).

Standardization of Rates
♦ Standardization is a simple way to remove effect of
confounding.
♦ It can be extended to more than one confounder.
♦ Similar techniques can be used for differences or ratios of

rates.
♦ An alternative is a stratified analysis (later).

SESI 4 - Epid-Sesi - 3 - M. Hakimi

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SESI 4 - Epid-Sesi - 3 - M. Hakimi

Uploaded by

Copyright:

Available Formats

Estimating and Comparing

Rates using Dircet and

Siswanto Agus Wilopo

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 1

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 2

♦ Kleinbaum, Kupper and Morgenstern, Epidemiologic

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 3

♦ We can talk about instantaneous and average rates.

♦ In epidemiology, we usually talk about average rates.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 4

“The incidence rate of disease occurrence is the instantaneous

♦ We could similarly define the mortality rate.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 5

♦ It is expressed in units of 1/ time.

♦ It is sometimes confused with risk.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 6

♦ The risk (probability) of developing disease in time T will

♦ Thus, formally these are two different quantities.

I no. of new cases in the calendar period (t 0 , t1 )

♦ PT is expressed in person-years, person-days etc.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 8

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 9

♦ E.g., population-based registries.

♦ Person-years computed using the mid-year population.

♦ For rare events, periods of several years may be used.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 10

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 11

♦ It is an estimate of an average rate.

♦ Any fluctuations in the instantaneous rate are obscured and

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 12

♦ However, sex, age, race etc. can have substantial influence

♦ Comparing crude rates for two populations, which differ

♦ Therefore, usually standardized rates are compared.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 14

• ID1 = 30/54308.7 = 0.00055; SE(ID1) = (30/54308.72)1/2 = 0.00010

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 15

♦ We get two estimates for non- and exposed subjects:

♦ To compare them, we can look at

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 16

• ID1 = 30/54308.7 = 0.00055; ID0 = 60/51477.5 = 0.00116

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 17

♦ Thus, an approximate 95% CI 1 1

♦ Both CIs allow to reject the null hypothesis of no difference.

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 19

♦ 95% “test-based” CI for

♦ The use of the normal distribution is an approximation.

♦ It is possible to construct a CI for ln IRR using the “exact”

♦ We will assume that our population is stratified by age (i.e.,

♦ One needs to compute age-specific rates (ID).

♦ There are two methods of standardization:

♦ Crude Rate in study population = It / PTt .

♦ Directly Standardized Rate (DSR):

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 25

♦ DSR by itself is not meaningful – it makes sense only when

♦ DSR depends on the choice of the standard population.

♦ Direct standardization requires age-specific rates for all

♦ If these are not available, or they are imprecise, the

♦ Both should lead to similar conclusions; if they do not, the

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 27

♦ Standardized (Incidence or Mortality) Ratio (SIR or SMR):

♦ Take Indirectly Standardized Rate (ISR) as:

Race USA, 1987

Copyright © 2014 Department of Public Health sawilopo@yahoo.com 29