epiR (version 2.0.85)

epi.indirectadj: Indirectly adjusted incidence risk estimates

Description

Compute indirectly adjusted incidence risks and standardised mortality (incidence) ratios.

Usage

epi.indirectadj(obs, pop, std, units, conf.level = 0.95)

Value

A list containing the following:

crude.strata

the crude incidence risk estimates for each stratum.

adj.strata

the indirectly adjusted incidence risk estimates for each stratum.

smr

the standardised mortality (incidence) ratios for each stratum.

Arguments

obs

a one column matrix representing the number of observed number of events in each strata. The dimensions of obs must be named (see the examples, below).

pop

a matrix representing population size. Rows represent strata (e.g., region); columns represent the levels of the explanatory variable to be adjusted for (e.g., age class, gender). The sum of each row will equal the total population size within each stratum. If there are no covariates pop will be a one column matrix. The dimensions of the pop matrix must be named (see the examples, below).

std

a one row matrix specifying the standard (or `reference') incidence risks to be applied to each level of the covariate to be adjusted for. The length of std should be one plus the number of covariates to be adjusted for (the additional value represents the incidence risk in the entire population). If there are no explanatory variables to adjust-for std is a single number representing the incidence risk in the entire population.

units

multiplier for the incidence risk estimates.

conf.level

magnitude of the returned confidence interval. Must be a single number between 0 and 1.

Author

Thanks to Dr. Telmo Nunes (UISEE/DETSA, Faculdade de Medicina Veterinaria - UTL, Rua Prof. Cid dos Santos, 1300-477 Lisboa Portugal) for details and code for the confidence interval calculations.

Details

Indirect standardisation can be performed when stratum-specific incidence risk estimates are either unknown or unreliable. If the stratum-specific incidence risk estimates are known, direct standardisation is preferred.

Confidence intervals for the standardised mortality ratio estimates are based on the Poisson distribution (see Breslow and Day 1987, p 69 - 71 for details).

References

Breslow NE, Day NE (1987). Statistical Methods in Cancer Reasearch: Volume II - The Design and Analysis of Cohort Studies. Lyon: International Agency for Cancer Research.

Dohoo I, Martin W, Stryhn H (2009). Veterinary Epidemiologic Research. AVC Inc, Charlottetown, Prince Edward Island, Canada, pp. 85 - 89.

Lash TL, VanderWeele TJ, Haneuse S, Rothman KJ (2021). Modern Epidemiology. Lippincott - Raven Philadelphia, USA, pp. 75.

Sahai H, Khurshid A (1993). Confidence intervals for the mean of a Poisson distribution: A review. Biometrical Journal 35: 857 - 867.

Sahai H, Khurshid A (1996). Statistics in Epidemiology. Methods, Techniques and Applications. CRC Press, Baton Roca.

See Also

epi.directadj

Examples

## EXAMPLE 1 (without covariates):
## Adapted from Dohoo, Martin and Stryhn (2009). Data have been collected on 
## the incidence of tuberculosis (TB) )positive herds in two areas ("A" and 
## "B"). Provided are the counts of incident TB positive herds and counts 
## of the herd population at risk. The herd level TB incidence risk for this 
## population is 0.060 (6 TB positive herds per 100 herds at risk).

## In their text Dohoo et al. present this example as incidence rate data 
## (the number of TB positive herds per herd-year at risk). For this example 
## the frequency of tuberculosis is expressed as an incidence risk 
## (i.e., the number of TB positive herds divided by the size of the herd 
## population at risk). 

obs.m01 <- matrix(data = c(58,130), nrow = 2, byrow = TRUE,
   dimnames = list(c("A","B"), ""))
pop.m01 <- matrix(data = c(1000,2000), nrow = 2, byrow = TRUE,
   dimnames = list(c("A","B"), ""))
std.m01 <- 0.060

epi.indirectadj(obs = obs.m01, pop = pop.m01, std = std.m01, units = 100,
   conf.level = 0.95)
   
## In area A the observed number of TB positive herds was 0.97 (95% CI 
## 0.73 to 1.22) times the number expected. In area B the observed number
## of TB positive herds was 1.1 (95% CI 0.90 to 1.3) times the number expected. 

## EXAMPLE 2 (with covariates):
## We now have, for each area, the data stratified by herd type (dairy, beef).
## The incidence risks of TB for beef herds, dairy herds and the 
## total population are 0.025, 0.085, and 0.060 (2.5, 8.5 and 6.0 TB positive
## herds per 100 herds at risk), respectively:

obs.m02 <- matrix(data = c(58,130), nrow = 2, byrow = TRUE,
   dimnames = list(c("A","B"), ""))
pop.m02 <- matrix(data = c(550,450,500,1500), nrow = 2, byrow = TRUE,
   dimnames = list(c("A","B"), c("Beef","Dairy")))
std.m02 <- matrix(data = c(0.025,0.085,0.060), nrow = 1, byrow = TRUE,
   dimnames = list("", c("Beef","Dairy","Total")))

epi.indirectadj(obs = obs.m02, pop = pop.m02, std = std.m02, units = 100,
   conf.level = 0.95)

## > $crude.strata
## >   est    lower    upper
## > A 5.8 4.404183 7.497845
## > B 6.5 5.430733 7.718222

## > $adj.strata
## >        est    lower    upper
## > A 6.692308 5.076923 8.423077
## > B 5.571429 4.628571 6.557143

## > $smr.strata
## >   obs exp       est     lower    upper
## > A  58  52 1.1153846 0.8461538 1.403846
## > B 130 140 0.9285714 0.7714286 1.092857

## The crude herd level incidence risk of TB in area A was 5.8 (95% CI 4.0 
## to 7.5) per 100 herds at risk. The crude herd level incidence risk 
## of TB in area B was 6.5 (95% CI 5.4 to 7.7) per 100 herds at risk.

## The indirectly adjusted herd level incidence risk of TB in area A was 6.7
## (95% CI 5.1 to 8.4) per 100 herds at risk. The indirectly adjusted 
## herd level incidence risk of TB in area B was 5.6 (95% CI 4.6 to 6.6) 
## per 100 herds at risk.