Health Impact Assessment - Stage 5

01 Develop a plan that addresses populations of concern, accounts for alternative scenarios, and estimates the economic value of impacts

Draw upon your experience in Stage 4 to consider: the extent to which the HIA addressed the questions most important to the decision-maker; whether the decision-maker appears to have acted upon the information you provided; when you lacked the data, methods or tools to quantify certain impacts; what key sources of uncertainty remain unaddressed, either qualitatively or quantitatively. In particular, identify what populations of concern were either considered incompletely, or not considered at all, in the Stage 4 analysis because you lacked concentration-response estimates or background incidence rates for these populations. Decide whether you want to account probabilistically for time-activity patterns and the influence of indoor/outdoor time on population exposure.  

Determine whether data exist to consider a full probabilistic uncertainty analysis in which each input parameter is expressed as a distribution that can be sampled (Coffman et al., 2020). Decide whether, and how, to incorporate alternative future scenarios when quantifying impacts over multi-decade periods (Ebi et al., 2005). Determine whether, and how, you want to quantify the economic value of impacts, ensuring that your approach is equitable. Finally, reflecting upon your experience in Stage 4, decide whether your team is comprised of subject matter experts that can guide each stage of a Stage 5 analysis.  

02 Build upon stage 4 to increase the size of the team, now including an exposure scientist, economist and clinician

As you expand the scope of the analysis to both identify the limitations uncovered in Stage 4, and include new endpoints and approaches for quantifying effects, you may want to include an exposure scientist, health economist and clinician on your team.  The exposure scientist can help characterise the influence of housing stock, commuting patterns and other factors on individual exposure, including populations of concern (Özkaynak et al., 2014). The health economist can assist in developing, applying and interpreting cost of illness and willingness to pay measures needed to quantify the economic value of air pollution effects (Maniloff et al., 2024). And, the clinician can help characterise the biological mechanisms that help explain the impacts observed in the epidemiological studies.

As with Stage 4, decide at this point whether you have the resources sufficient to perform an HIA that adequately addresses the policy question; if not, confer with the decision-maker to re-scope the analysis (Bhatia et al., 2014). Confirm with team members that they each understand the type and format of data needed for the analysis, so that they are collecting or modelling the right input parameters.  

03 Characterising exposure: collect monitoring data

Build on the Stage 4 analysis to identify gaps in the air quality monitoring network—particularly those in highly populated locations and areas in which there is a disproportionate share of populations of concern. You may wish to extend the reference monitor network to these areas or deploy low-cost sensors and use these as a complement to the reference monitors (WMO, 2025).  

An intensive monitoring campaign deploying larger numbers of low-cost sensors, conducted over weeks or months, can identify areas of elevated concentrations that might be missed by reference monitors. Such a monitoring campaign could focus on populations of concern to better characterise exposures among these groups; these concentrations can in turn be used in an HIA to quantify impacts at the local scale. Likewise, these monitoring observations could inform an exposure model, simulating the variability in personal exposure among and between population groups.  

04 Characterising exposure: collect or perform air quality modelling simulations

Building upon Stage 4, your team will be experienced in collecting or simulating air quality modelling predictions. There should be a feedback loop from the HIA team back to the air quality modelling team, describing what new averaging metrics, pollutants, or model domains are needed to conduct the analysis. The intensive monitoring campaign performed above could be used to bias-correct the air quality modelling simulations performed in this stage (Baker et al., 2009).

Your team may also consider developing, or applying, “reduced-form” methods for quantifying and monetising air pollution effects (Fann et al., 2009, 2012; Simon et al., 2023). These techniques allow you to more easily estimate air pollution impacts by applying a series of simplifying assumptions regarding the emissions-to-air quality relationship. Such an approach could be developed to screen policy options before conducting full-scale photochemical modelling.  

05 Characterising exposure: collect detailed population data 

Build upon Stage 4 by accounting for uncertainty in the projected population counts. Consider projecting the number, age-structure, and distribution of future populations using a scenario analysis (e.g. SSPs) (Gurney et al., 2022). Scenarios are an educated guess about future states of the world—including economic growth, the fertility rate, migration, and other factors—which in turn influence the size and distribution of the population.  

You may also want to relax a key assumption regarding population changes often found in traditional HIAs. Population counts are taken to be a fixed input, where the size, distribution, and growth of the population to be independent of air quality changes. For example, if an HIA estimates that a policy results in 1,000 fewer people dying prematurely in one year, the total population is unaffected in the following year. However, the Life Table approach accounts for the relationship between changes in air pollution-related deaths in one year and the number of people alive in the next year (Roman et al., 2022). If the decision-maker needs to know how air pollution will affect the population over a multi-year period, a Life Table may be preferable to a static HIA.

06  Characterising exposure: collect background health data 

In this Stage, you will have identified background incidence and prevalence rates for a full suite of mortality and morbidity outcomes, taking care to ensure that these outcomes are non-overlapping. As an example, you will have identified prevalence rates for common chronic illnesses—including asthma, cerebrovascular disease, Chronic Obstructive Pulmonary Disorder (COPD), and diabetes mellitus—such that you can quantify both incident cases of these diseases as well as associated consequence (e.g. exacerbated cases of asthma, hospital visits due to COPD, etc) (Künzli et al., 2008).  

If your analysis is quantifying future impacts, you will be projecting rates of death (and possibly also disease) using life tables from the census. The assumptions you make regarding future death rates will be harmonised with the assumptions used to project the population above (Anenberg et al., 2016). 

07 Select concentration-response parameters 

You now have access to a broader library of region- or country-specific epidemiological studies than Stage 4, thanks in part to the research agenda you set in the above stages, where you identified gaps in the epidemiological literature. Whether selecting local, regional or global epidemiology studies, take care to document your approach for selecting among studies including the criteria you considered (US EPA, 2024a).

You may also incorporate concentration-response relationships from studies that account for effect modification. For example, some epidemiological studies consider the role that stressors (e.g. temperature) may play in altering the association between pollution and adverse effects (Jhun et al., 2014). These concentration-response relationships may in turn be used in a HIA (Fann et al., 2021). These and similar analyses find that temperature tends to increase the risk of ozone-attributable mortality, all else equal—indicating the importance of accounting for temperature as a stressor when addressing traditional air pollutants like ozone. One advantage to this approach, as compared to a traditional HIA, is that you can consider two or more exposures or an exposure and a socioeconomic attribute like air conditioning prevalence or access to greenspace.  

08 Select economic unit values

The economic value of air pollution attributable health effects is substantial, and the value of avoided morbidity effects alone can equal or exceed the cost of controlling air pollution and mitigating climate change (UNEP/CCAC, 2026). To the extent that you are quantifying the economic value of effects, identify ways to monetise the full suite of adverse effects associated with the pollutant exposure. For example, if you are quantifying incident cases of a chronic endpoint like Alzheimer’s, carefully consider all of the “downstream” costs associated with that event, including cost of care (both direct medical costs as well as the value of the lost time of care takers), hospital admissions, doctor’s visits, and medications.  

If you are monetising the value of avoided premature deaths, and using a Value of Statistical Life (VSL), determine whether there is an appropriate value to use from your country or region—or if instead you need to transfer a value from another region (see here for updated estimates from OECD). If choosing the latter approach, ensure that you are accounting correctly for differences in income between the country supplying the VSL and the country to which you are applying it (Robinson et al., 2019).