In m-allik/SocEpi: Health Inequality Analysis

Motivation

Research on health inequalities

Crating small-area deprivation measures
Calculation of standardized rates
Measuring socio-economic inequalities in health

\fontsize{6pt}{7.2}\selectfont

Brown, D., Allik, M. , Dundas, R. and Leyland, A. H. (2014) Carstairs Scores for Scottish Postcode Sectors, Datazones and Output Areas from the 2011 Census. Technical Report. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow.

Allik, M., Brown, D., Dundas, R. and Leyland, A. H. (2016) Developing a new small-area measure of deprivation using 2001 and 2011 census data from Scotland. Health and Place, 39, pp. 122-130.

Allik, M. Brown, D., Dundas, R. and Leyland, A. H. (under review) Differences in Ill Health and Health Inequalities among Ethnic Groups: Population Study using 2011 Scottish Census

Other packages

Roughly 150 packages related to "public health", "health inequalities", "social epidemiology", "epidemiology", "demography", "mortality rate", "mortality".

Only four more closely related to this work: popEpi, epiR, Epi, epitools

Overlap: calculation of standardized rates, rate ratios and SE/CI for these
Differences: scope/flexibility of these functions, methods for SE/CI
Gap: not many tools specific to health inequalities research

Calculating RII/SII - formula

rii(data, health, population, ses, age, ...)

Standardized rates (ESP 2013 default)
Linear regression: $y = \alpha + \beta x$
- SII is $\beta$ coefficient
RII is SII divided by mean health
Weighted regression an option

Calculating RII/SII - population mid-points

x <- c(0.1, 0.3, 0.5, 0.7, 0.9)

z <- c(0.16, 0.18, 0.2, 0.22, 0.24)
#sum(z)
x2 <-c(0, cumsum(as.numeric(z))[-length(z)]) + z/2

y <- c(150, 200, 270, 310, 400)

plot(x, y, xlim = c(0, 1), ylim = c(0, 500), bty="n", col="blue", pch = 20, xlab="Population mid-points", ylab = "Mortality rate")
abline(lm(y~x), col="blue")
points(x2, y, col="red", pch=20)
abline(lm(y~x2), col="red")

text(0, 500, paste("RII = ", round(lm(y~x2)$coef[2], 1), " (population distribution: 16, 18, 20, 22, 24)",
                   sep = ""), col="red", pos = 4)
text(0, 450, paste("RII = ", round(lm(y~x)$coef[2], 1), " (population distribution: 20%)", 
                   sep = ""), col="blue", pos = 4)

Confidence intervals

Usually either not shown or based on S.E. from regression
Based on simulation
- Lumme et al (2015) A Monte Carlo method to estimate the confidence intervals for the concentration index using aggregated population register data. Health Services & Outcomes Research Methodology, 15(2), 82-98.
- Wider CIs