Disability weights: Difference between revisions

Latest revision as of 10:08, 26 August 2013

This page is a knowledge crystal of subtype method. The page identifier is Op_en2307
Moderator:Virpi Kollanus (see all)
This page has been peer reviewed. It has gained 2 onors. See review. Citation of this page: Jouni T. Tuomisto, Otto Hänninen, Anne Knol, Virpi Kollanus, Olli Leino: Disability weights. Severity weights of diseases for DALY calculations.. Opasnet 2010. [1]. Accessed 19 Nov 2024. This page has also been published elsewhere: Main material was originally published by WHO in http://www.who.int/entity/healthinfo/bodreferencedisabilityweights.xls.
Upload data {{#opasnet_base_link:Op_en2307}}

Disability weights describe the weights that are given to different diseases in the calculation of disability-adjusted life years (DALY).

Question

What are disability weights for different disases, and how should they be used in calculating burden of disease?

Answer

This code can be used on this page to calculate YLD (years of life under disease). The code can also be copied to another page in Opasnet. It runs on R, so it is also possible to copy and use it outside the Opasnet workspace.

+ Show code - Hide code

library(OpasnetUtils)
library(xtable)

YLD <- new("ovariable",
	name = "YLD",
	dependencies = data.frame(Name = "morbidity.cases"),
	formula = function(dependencies, intermediates = FALSE, ...) {
	
		ComputeDependencies(dependencies, ...)
	
		disability.weight <- EvalOutput(new("ovariable",
			name = "disability.weight",
			data = tidy(opbase.data("Op_en2307"), objname = "disability.weight") # [[Disability weights]]
		))
		if(!exists("duration")) {
#			duration <- tidy(op_baseGetData("opasnet_base", "Op_enXXXX"), objname = "duration")
			duration <- EvalOutput(new("ovariable", # This is a placeholder because actual durations are not known.
				name = "duration",
				data = data.frame(durationResult = 1)
			))
		}

		out <- morbidity.cases * duration * disability.weight
	
		return(out)
	}
)

#objects.put(YLD)
#cat("Ovariable YLD has been created. Save the run key for further use.\n")

morbidity.cases <- new("ovariable", 
	name = "morbidity.cases", 
	data = interpret("500-1000") # data.frame(morbidity.casesResult = 1)
)

YLD <- EvalOutput(YLD, N = 10)

summary(YLD)

NOTE! Old official WHO disability weights were used in a previous version of this page. The current version uses disability weights from Salomon (Salomon et al., Lancet 2012-2013

^[1]).

D↷

Rationale

Data

Disability weights are obtained by posing two different Person Trade-Off (PTO) questions to expert panels

Question 1 compares life extensions for disabled and healthy people
Question 2 compares cures for illness with extension of life

Ideally, these values should be derived from empirical data among representative populations. Combining the answers to these questions results in an average disability weight. Usually, these weights are presented as a fixed parameter, without a confidence interval or other type of distribution.

Arguments defending this approach include:

DALY weights are an intrinsically subjective value for which no "true estimate" exists. They should be treated as a normative value, which does not need further uncertainty intervals.
Uncertainty in very small DALY weights (for less severe effects) affects overall outcomes of DALY calculations much more than uncertainty in large disability weights. This is a "flaw" of the method: if health would be defined as 1 and death as 0 (as is being done with QALYs), the more severe outcomes would have more overall uncertainty. The uncertainty is greatest for the smallest disability weights and narrows as more severe weights are attained (figure 5.16

^[2] ).This provides a skewed image. In order to avoid this, distributions should not be used. Another solution might be to normalize the weights. D↷

Arguments opposing this approach include:

Estimates of DALY weights differ between experts, so this variation should be presented in the overall distribution of these weights.

Including uncertainty intervals of disability weights will probably not affect the ranking of overall disease burden estimates very significantly (WHO GBD study). However, it will affect the confidence interval of individual estimates.

List of data files or sources: