Burden of disease

Because mortality and prevalence, taken separately, each leave out half the picture: ranking by number of deaths ignores the diseases that disable for years without killing, and ranking by prevalence ignores how severe the condition is. This applies when you have to prioritise between several health problems, each with a different profile of lethality and chronicity. A disease that kills early and one that disables for decades cannot be compared by looking at a single dimension. A burden of disease study combines both in a common metric, the DALY, which adds together the years of life lost to premature mortality and the years lived with disability. That makes it possible to rank health problems with very different profiles in the same unit and to set priorities on a basis a decision-maker can defend, rather than on a single partial indicator.

The DALY (disability-adjusted life year) is a metric that expresses the burden of a disease as the sum of two components: years of life lost to premature mortality (YLLs) and years lived with disability (YLDs). It serves as a common unit for any comparison of population impact between diseases. YLLs are estimated from deaths and the age at which they occur, set against a reference life expectancy; YLDs, from the prevalence of each condition and a disability weight that reflects the severity of living with it. Added together, they give a single burden figure that captures both dying early and living with health loss or disability. It is worth understanding that a DALY is an estimate built on assumptions, such as the reference life expectancy, the disability weights and the comorbidity adjustment, not a direct measurement observed in each individual: its value lies in the comparability it provides, as long as those assumptions are applied consistently.

You need cause-specific mortality data and prevalence or incidence data for the conditions to be assessed, with a reasonable correspondence to the cause categories of the reference framework. This applies to any burden study, where the quality of the sources determines the quality of the estimates. Disability weights are not set arbitrarily by ADI or by the client: they are taken from those used in the Global Burden of Disease framework or other recognised references, depending on the study context, so that they reflect the relative severity of living with each condition and make the estimates comparable with national or global figures. The client provides the mortality and morbidity data for their population; the service reconciles them with the framework's categories, selects the disability weights and the reference life table, and estimates the burden. The sources do not always need to arrive cleaned, but there must be a body of mortality and prevalence data that can be mapped onto the reference framework in a defensible way.

Total burden measures how much a condition weighs on a population; avoidable burden estimates how much of it could be reduced by acting on modifiable factors. To distinguish them, we break the burden down by attributable risk factors such as tobacco, alcohol or obesity, estimate the population-attributable fraction for each from the available evidence and, on that basis, how much would be avoidable through prevention. This is a population-level estimate, not a causal explanation of each individual case. The distinction matters for priority setting: the condition with the greatest burden is not necessarily the one with the greatest scope for prevention. Avoidable burden does not make the decision on its own, either: a condition with limited scope for modification may still deserve priority because of its severity, equity considerations or the care needs it creates. The analysis provides the potential reduction, and the decision weighs it alongside those other criteria.

These are two different questions that should not be confused. A burden of disease study prioritises between many health problems: it ranks which diseases impose the greatest population burden and how much of it is avoidable, to show where to look. Economic evaluation of health interventions (HEOR) assesses the cost-effectiveness of a specific intervention: whether it offers a defensible cost-effectiveness ratio compared with its alternative. Both use the DALY as a metric, but in different ways: here, the cost per DALY averted, when cost data are available, is one more criterion in a prioritisation matrix; in HEOR it is the result of a full cost-effectiveness analysis, with its models, its probabilistic sensitivity analysis and its acceptability curves. The natural sequence links them: the burden study helps prioritise where to act, and the HEOR evaluation then analyses whether the intervention chosen for that problem has an acceptable cost-effectiveness ratio under given assumptions. If the question is "which problems should I spread my effort across?", it is burden of disease; if it is "is this specific intervention cost-effective?", it is HEOR.

It does not replace the decision: the burden of disease study ranks the health problems and quantifies the avoidable impact, but resource allocation belongs to the health manager, who weighs feasibility, equity, budget, social acceptability and institutional context, which the analysis alone does not cover. It applies as a clarification of scope, because a burden ranking can be misread as an investment verdict. What the service delivers is the quantitative input: the burden ranking, the avoidable fraction and, where cost data are available, the cost per DALY averted as an additional criterion, brought together in a prioritisation matrix. That matrix supports the decision; it does not replace it. The burden ranking is one of the dimensions the manager has to weigh, not the only one. The separation protects the client: prioritisation guided only by the size of the burden can be technically incomplete if it does not take account of feasibility, equity and real implementation constraints. The service quantifies and ranks; the manager decides.

A burden of disease study estimates, before or independently of any action, how much burden could potentially be avoided, by attributing it to modifiable risk factors; ex post evaluation of interventions estimates the attributable effect of a specific intervention that has already been implemented, using observed data. One answers "how much burden could be reduced under given assumptions about modifying the risk factor?"; the other, "what attributable effect is estimated for an intervention already rolled out?". They are not the same: avoidable burden is a potential estimate based on attributable fractions and existing evidence; ex post evaluation estimates the effect actually observed against a counterfactual. They complement each other in the decision sequence, since avoidable burden helps prioritise where it is worth intervening and ex post evaluation later estimates whether the intervention that was rolled out worked. But one works on potential and the other on observed effect.

At the end of a burden of disease study you receive a prioritisation matrix with YLLs, YLDs and DALYs by cause, adjusted for comorbidity, a ranking of conditions by burden, and the breakdown by risk factor with the estimated avoidable fraction. Where suitable data are available, we also include time trends and cost per DALY averted as an additional prioritisation criterion. The report makes the sources and assumptions visible (including disability weights and reference life expectancy) so that the figures can be reviewed and checked against reference estimates, and explains where data quality or mapping to GBD causes limits the level of detail or comparability. The client provides the data and decides how resources are allocated; we build the quantitative basis to support that decision before a committee or funder with population impact, not just prevalence or mortality.