Using burden of disease or harmful effect assessments for quantifying health impacts from environmental noise? Jon-Paul Faulkner 1 University College Dublin Belfield, Dublin 4 Ireland Enda Murphy 2 University College Dublin Belfield, Dublin 4 Ireland

ABSTRACT Commission Directive (EU) 2020/367 describes how harmful effects may be calculated for ischemic heart disease (IHD), high annoyance (HA), and high sleep disturbance (HSD) for road, rail, and aircraft noise. However, traditionally, the negative health impacts of environmental noise have been assessed using the WHO's Burden of Disease methodology. In terms of evaluating how harmful effects and burden of disease methodologies compare in terms of the requirements for calculation, interpretation of results, practical application, and suitability for measuring the negative health impacts of environmental noise, our analysis suggests that the EU's harmful effects assessment may be a superior methodology. However, it must also be emphasised that the pervasive application of burden of disease assessment means that comparative analysis with non-noise-induced disease states is possible, whilst this is not possible with the harmful effect assessment approach. 1. INTRODUCTION

Environmental noise impacts on public health has advanced considerably over the past number of years. The four main sources of environmental noise in Europe are road traffic noise, railway noise, aircraft noise, and industrial noise [1]. In terms of the severity of disturbance caused, research (e.g. [2]; [3]) indicates that aircraft noise results in the most severe disturbance, followed by road traffic noise, railway noise, and industrial noise. However, due to its prevalence, as well as the relative severity of disturbance associated with it, road traffic noise is the most commonly researched environmental noise source analysed in relation to negative impacts on public health.

Following the publication of the Environmental Noise Guidelines for the European Region [4]. Annex III of Commission Directive 2002/49/EC was amended with the ratification of Commission Directive (EU) 2020/367 [5]. It describes how harmful effects may be calculated for ischemic heart disease (IHD), high annoyance (HA), and high sleep disturbance (HSD) in the context of road, rail, and aircraft noise for those EU Member States (MS) that transpose it into law. However, traditionally

1 jon.faulkner@ucd.ie 2 enda.murphy@ucd.ie

the negative health impacts of environmental noise have been predominantly assessed using the Burden of Disease methodology [6]. This paper evaluates how the different methodologies compare in terms of requirements for calculation, results, and their practical application for estimating health effects of environmental noise using the case of road traffic noise and the IHD health outcome.

2. METHOD

2.1. Estimating population exposure

In estimating national population exposure to the harmful effect and burden of disease associated with environmental noise an Irish case study was used and the new CNOSSOS-EU calculation method was applied where possible. In the context of road sources within Dublin and Cork agglomerations, the most recent 2017 round (i.e. round 3 ) of datasets, provided by local authorities (LAs), were applied in generating acoustic models and calculated using the CNOSSOS-EU method (see [7], p. 31-38). The CNOSSOS-EU calculation method was also utilised for rail sources within the Dublin agglomeration. Models for rail sources were generated for the Luas (tram rail) and Irish Rail networks (national rail network). 3 In accordance with the CNOSSOS-EU method, an assessment of noise levels at the most exposed façade for buildings with one dwelling was applied in combination with an assessment of noise level at each respective façade for residential buildings with more than one dwelling as per Directive (EU) 2015/996; L168/94-5 [8]. Similarly, calculation of inhabitants per building under CNOSSOS-EU was estimated using CASE 1B criteria:

𝑉 𝑏𝑢𝑖𝑙𝑑𝑖𝑛𝑔

𝐼𝑛ℎ 𝑏𝑢𝑖𝑙𝑑𝑖𝑛𝑔 =

𝑉 𝑡𝑜𝑡𝑎𝑙 x 𝐼𝑛ℎ 𝑡𝑜𝑡𝑎𝑙 (1)

2.2. Calculating harmful effect and burden of disease In both calculation of harmful effect and burden of disease, the fundamental quantitative basis for assessment is population risk. As such, an estimation of relative risk (RR) was applied to generate the population attributable fraction (PAF) which is applied in the estimation of noise-induced cardiovascular disease cases (e.g. the assessment of harmful effect) and in the estimation of noise- induced cardiovascular disease disability-adjusted life years (DALYs) (i.e. burden of disease). Similarly, the absolute risk (AR) related to high annoyance (HA) and high sleep disturbance (HSD) is used both in harmful effect and burden of disease calculation. In order to calculate the harmful effects of environmental noise, the methodology presented in Commission Directive (EU) 2020/367 [5] was utilized. For this, the following data is required for the year in question:

1) National Population Statistics 2) National Incidence Statistics 3) Relative Risk (RR) Statistics 4) Absolute Risk (AR) Statistics Population statistics were acquired for 2017 from the Central Statistics Office [9] of Ireland for year end April 2017. Incidence and prevalence statistics for Ireland were acquired from the Global Burden of Disease Database for 2017 [10]. RR and AR statistics were derived from the latest meta-analysis perform by the WHO [4] as previously described. In order to calculate the burden of disease as outlined in WHO [6] the following data is required for the year in question:

1) National Population Statistics

3 For further information please see Murphy et al., 2021; p. 37-38) .

2) National Incidence Statistics 3) National Mortality Statistics 4) National Life Expectancy Statistics 5) Duration of Disability Statistics 6) Disability Weight (DW) Statistics 7) Relative Risk (RR) Statistics 8) Absolute Risk (AR) Statistics Population, incidence, RR, and AR statistics were acquired as previously described. Mortality and life expectancy statistics, required in order to calculate Years of Life Lost (YLLs) as part of noise- induced burden of disease estimation, were derived from the CSO Vital Statistics for 2017 [9]. Duration of disability and disability weight (DW) statistics, required in order to calculate Years of Life Lost due to Disability (YLDs), also as part of burden of disease estimation, were acquired from WHO [7] guidance where appropriate, while DW statistics were acquired from [6,7,11] where appropriate. 3. RESULTS

Table 1 describes the population attributable fraction (PAF) and the number of attributable cases resulting from the assessment of the harmful effect of road traffic noise-induced IHD for the Irish population as of 2017; it includes estimates for agglomerations and the nation as a whole. Table 1 shows that 256 cases (i.e. 1.87% of total IHD incidence) of road traffic noise-induced IHD are estimated for the national population as of 2017 using an OSi road profile (which covers the full road network), and including an uncertainty range of between 32 and 474 cases.

T able 1: Harmful effects of road traffic noise-induced IHD for Irela nd (2017)

PAF (%) Cases Agglomerations Total 4.59 (0.60-8.29) 203 (26-367) National Total 4 1.87 (0.23-3.40) 256 (32-474) Table 2 outlines analogous results for burden of disease estimations for road traffic noise-induced IHD. It shows that 3,445 DALYs were estimated for road traffic noise-induced IHD, comprising of 3,360 YLLs (0.95% of total IHD mortality) and 85 YLD (1.85% of total IHD incidence), including an uncertainty range of between 687 and 8,827 DALYs.

Table 2: Burden of Disease from road traffic noise-induced IHD for Ireland (2017)

PAF* YLL PAF** YLD DALY National Total 0.95 3360 1.85 85 3445 Minimum 5 Maximum

0.19 2.44

676 8641

0.19 2.81

11 186

687 8827 * Mortality PAF ** Morbidity PAF 4. DISCUSSION AND CONCLUSION

Commission Directive (EU) 2020/36 introduces a new assessment methodology which has not been officially utilised by EU MS in the delivery of the END to date. Previously, assessment of environmental noise-induced disease has been predominently calculated using burden of disease methodologies. Our analysis argues that the assessment of harmful effect methodology may be preferential to burden of disease estimation for the following reasons: 1) the harmful effect methodology is more efficient, less health data is necessary, and it requires less calculation; 2) for assessment that requires the application of national incidence statistics, results can be obtained for

4 The National Total estimate accounts for major road sources outside agglomerations and is calculated for the entire population of Ireland as of 2017.

5 Minimum and maximum values represent the range of uncertainty.

spatially localised areas as opposed to overall national figures; and 3) results from harmful effect assessment may be easier to interpret than is the case for DALYs.

On the first point, for the assessment of harmful effects, once the risk statistic is calculated, the number of cases relating to the disease state is applied using the Commission Directive 202/367. For the assessment of burden of disease, on the other hand, separate calculations for years of life lost (YLLs) and years of life lost due to disability (YLDs) must be formulated and computed and finally summated in order to acquire the total DALYs for the noise-induced disease state. Additionally, for burden of disease assessment, mortality and life expectancy statistics for both genders are required for the calculation of YLLs, while incidence, disability weight (DW), and duration of disability statistics are required for the calculation of YLDs. In contrast, apart from population statistics which are required for both assessments, only incidence statistics are required for harmful effect assessment.

On the second point, because harmful effect assessment utilises incidence rate, calculation can potentially be performed at any spatial scale, depending on the scale of census data available while burden of disease estimation utilises aggregate national incidence statistics which makes spatial disaggregation impossible. This reduces the value of the approach for informing targeted health- related policymaking with respect to environmental noise mitigation.

Another important point of discussion in relation to the comparison between these two assessment methodologies regards the application of both incidence and prevalence statistics in burden of disease calculation. In the Burden of Disease from Environmental Noise Quantification of Healthy Life Years Lost in Europe [6], while it is stated that incidence statistics should be used in calculation [6, p.7], when discussing national health data to be applied in calculation, the terms incidence and prevalence are used interchangeably throughout the document (see [6])) For example, it is stated that the exposure-based approach used in environmental burden of disease assessment should apply, “a population-based estimate of the incidence or prevalence of the outcome from surveys or routinely reported statistics”, [6]. Indeed, along with providing mortality RR values for a range of cardiovascular disease, the WHO [12] report both incidence and prevalence RR statistics. However, the application of prevalence in the estimation of risk is incorrect. This is because, contrary to prevalence, risk is the percentage of an at-risk population that presents with a particular disease-state within a particular time-period [13]. Incidence statistics should always be applied rather prevalence statistics where both are available if the understanding of risk calculation is to be interpreted correctly. Finally, it should be noted that the pervasive application of burden of disease assessment means that comparative analysis with non-noise-induced disease states is much easier, relative of course, to a new methodology. This is a primary and important advantage burden of disease assessment has over harmful effect assessment. Hence, burden of disease results attained from this analysis can be compared with non-noise-induced disease states, while harmful effect results cannot. This comparison with non-noise-induced disease states is crucial for persuading policy and planning authorities to take environmental noise more seriously as a priority public health concern. 4. ACKNOWLEDGEMENTS

This research was funded under the Environmental Protection Agency (EPA) Ireland Research Programme 2014–2020, grant number 2017-HW-MS-10. 6. REFERENCES

1. European Environment Agency (EEA)., 2020. Noise in Europe 2020. EEA Report, No 22,

Copenhagen, Denmark.

2. Seidler, A., Hegewald, J., Seidler, A.L., Schubert, M., Wagner, M., Dröge, P., Haufe, E., Schmitt,

J., Swart, E. and Zeeb, H., 2017. Association between aircraft, road and railway traffic noise and

depression in a large case-control study based on secondary data. Environmental research , 152 , pp.

263-271.

3. Beutel, M.E., Jünger, C., Klein, E.M., Wild, P., Lackner, K., Blettner, M., Binder, H., Michal, M.,

Wiltink, J., Brähler, E. and Münzel, T., 2016. Noise annoyance is associated with depression and

anxiety in the general population-the contribution of aircraft noise. Plos one , 11 (5), p.e0155357.

4. World Health Organization, 2018. Environmental noise guidelines for the European region. World

Health Organization, Copenhagen, Denmark.

5. European Commission. Commission Directive (EU) 2020/367 of 4 March 2020 amending Annex

III to Directive 2002/49/EC of the European Parliament and of the Council as regards the

establishment of assessment methods for harmful effects of environmental noise. Official Journal of

the European Union, L 67/132.

6. World Health Organization, 2011. Burden of disease from environmental noise: Quantification of

healthy life years lost in Europe. World Health Organization, Copenhagen, Denmark.

7. Murphy, E., Faulkner, J.P., Rice, H. and Kennedy, J., 2021. Transitioning to Strategic Noise

Mapping under CNOSSOS-EU (Noise-Adapt) (2017-HW-MS-9). EPA Research Report.

Environmental Protection Agency, Wexford, Ireland.

8. European Commission. 2015. Commission Directive (EU) 2015/996 of 19 May 2015 establishing

common noise assessment methods according to Directive 2002/49/EC of the European Parliament

and of the Council. Official Journal of the European Union, L168/1-128.

9. Central Statistics Office (CSO)., 2017. Vital Statistics Annual Report 2017. CSO, Dublin, Ireland

10. Global Burden of Disease Database (The Institute for Health Metrics and Evaluation). Data

retrieved on 17.09.20 via: http://ghdx.healthdata.org/gbd-results-tool.

11. Concha-Barrientos, M., Steenland, K., Prüss-Üstün, A., Campbell-Lendrum, D.H., Corvalán,

C.F., Woodward, A. and World Health Organization, 2004. Occupational noise: assessing the burden

of disease from work-related hearing impairment at national and local levels . World Health

Organization.

12. Van Kempen, E., Casas, M., Pershagen, G. and Foraster, M. 2018. WHO environmental noise

guidelines for the European region: a systematic review on environmental noise and cardiovascular

and metabolic effects: a summary. International journal of environmental research and public

health, 15(2), p.379.

13. Alexander, L., Lopes, B., Ricchetti-Masterson, K. and Yeatts, K., 2017. Incident vs prevalent

cases and measures of occurrence. Retrieved on 18.09.20 via

https://sph.unc.edu/files/2015/07/nciph_ERIC1.pdf .