WHO Environmental Noise Guidelines for the European Region: A Systematic Review on Environmental Noise and Cardiovascular and Metabolic Effects: A Summary

1In this paper, we present the main results of a systematic review of the literature dealing with observational studies on the association between environmental noise exposure and the cardiovascular and metabolic systems. The aim was to update some of the existing exposure-response relationships, and to evaluate the overall quality of the evidence. The World Health Organisation (WHO) commissioned this systematic review. Its results form important input for the new environmental noise guidelines for the European Region. The WHO requires that new guidelines should be based on the latest scientific knowledge. The complete review can be found in the report published at the website of RIVM (the Dutch National Institute for Public Health and the Environment) via the following link: http://www.rivm.nl/en/Documents_and_publications/Scientific/Reports/2017/november/Cardiovascular_and_metabolic_effects_of_environmental_noise_Systematic_evidence_review_in_the_framework_of_the_development_of_the_WHO_environmental_noise_guidelines_for_the_European_Region.

2During the past decades, several national and international organizations have made recommendations for protecting human health from the adverse effects of environmental noise exposure. In the existing guidelines, the principal noise source of concern was transportation noise, mainly road and air traffic. The health impact of other noise sources, such as rail traffic and wind turbines, was not addressed in these guidelines. However, with the ongoing extension of railway transport facilities, and the substantial growth of wind energy facilities, the number of studies on the impact of noise from rail traffic noise and on wind turbine noise has increased.

3The existing guidelines also contain recommendations that specifically deal with the impact of noise on the cardiovascular system. The most common explanation for the effects of noise on the heart and circulatory system, is stress. The cardiovascular effects related to noise exposure may also be the consequence of a decrease in sleep quality, caused by noise exposure during the night, among other additional or interrelated mechanisms. Such reactions may also affect the metabolic system.

4The most recent environmental noise guidelines from WHO, date back to 2009, and focus on night-time exposure. Meanwhile, new evidence on the relationship between noise exposure and cardiovascular effects has accumulated. Hypertension and ischaemic heart disease have been the main outcomes of concern in observational studies on the impact of noise on the cardiovascular system. In addition, an increasing amount of studies have recently investigated the impact of noise on other cardiovascular end-points such as stroke. Furthermore, hypertension is considered as an important risk factor for other cardiovascular outcomes such as stroke and myocardial infarction. Amongst the newly published studies there were also several studies dealing with the possible effects of noise on the metabolic system, in particular with regard to outcomes such as obesity and type 2 diabetes.

5In addition, a number of the newly published studies investigated the combined effects of noise and air pollution. People living close to roads, are exposed not only to traffic noise, but also to air pollution generated by traffic. Previous studies have shown a relationship between air pollution and cardiovascular disease. Since air pollution and noise from road traffic share the same source, cardiovascular effects could be attributed to both exposure factors.

6The existing environmental noise guidelines also include recommendations that aim to reduce environmental noise exposure in settings where children spend most of their time. However, none of these recommendations takes into account the cardiovascular effects of noise on children. It is possible that people exposed to high levels of noise from an early age, might be at higher risk for cardiovascular problems later in life. Since the publication of the latest environmental noise guidelines in 1999, the number of studies investigating the impact of noise on children’s blood pressure has increased substantially.

7The first step in this systematic review was to identify and select reviews of “sufficient” quality, that described the impact of exposure to environmental noise from several sources (air, road, rail and wind turbines) on the cardiovascular or metabolic systems, in different settings (at home, at school), and populations (e.g., adults, children).

8After an extended search, we identified 37 reviews evaluating available studies into the impact of exposure to environmental noise on the cardiovascular or metabolic systems. By means of the “Measurement tool for the Assessment of Multiple SysTemAtic Reviews” (AMSTAR) we evaluated their quality, and based on the relevance for this whole systematic review, we selected 15 reviews. We carried out the evaluation in duplicate (Elise van Kempen and Maria Foraster, and then discussed the results afterwards.

9It appeared that most of the studies covered by the selected reviews, reported on the impacts of road and aircraft noise exposure among adults. Nine reviews included one or more meta-analyses, resulting in more than 13 exposure-response relationships. For most available exposure-response relations, the reviewers were not able to provide a quality judgement of the individual studies. For a number of (new) health end-points (e.g., obesity) and/or noise sources (e.g., rail traffic, no reviews or exposure-response relationships were available.

10Following the results of the evaluation of existing reviews, we decided to carry out a new systematic review on the impact of noise on the cardiovascular and metabolic system in order to update some of the existing exposure-response relationships, and to assess the quality of the existing evidence.

11We identified observational studies on the impact of noise from air, road, and rail traffic and wind turbines on the cardiovascular or metabolic systems published from 2000 until October 2014 in several literature databases (Medline/PubMed, SCOPUS, EMBASE and SCISCEARCH (see Appendix A for the applied search profiles). To ensure that most of the studies could be identified, we manually scanned reports and proceedings in the fields of epidemiology, and noise and health. We supplemented the results of this search with studies that were already identified by means of the 15 reviews, which we evaluated during the first step of this systematic review (see Section 2.1). Overall, we identified more than 600 publications which were screened in duplicate (Elise van Kempen and Maria Foraster) using predefined criteria. We selected 61 studies for data-extraction, where detailed quantitative information was available on exposure and health outcomes, enabling estimation of exposure-response relationships. However, conducting a systematic review often takes a lot of time. While working on this review, new results became publically available. In order to keep our results more up to date, it was decided to extend our study material with more recent results beyond the studies that had we already identified for the period 2000–October 2014. However, only updated and new results of studies published between November 2014 and August 2015, were included and processed. Consequently, we were able to include the latest results published between November 2014 and August 2015 of several selected studies: DEBATS, REGICOR, SDPP, HUBRO and DCH. In- and exclusion criteria were extensively described in the complete systematic review.

12From the selected 61 studies (described in 113 records), we extracted the following data via a structured data extraction form:Data on general study characteristics (e.g., study design, study period, study location);Population characteristics (sampling of the study population, number of participants, response- and attrition rate, gender, age;Exposure assessment and health outcome assessment, and;The results of the study.

13Data on general study characteristics (e.g., study design, study period, study location);

14Population characteristics (sampling of the study population, number of participants, response- and attrition rate, gender, age;

15Exposure assessment and health outcome assessment, and;

16The results of the study.

17We carried out the data extraction in duplicate (Elise van Kempen, Maribel Casas and/or Göran Pershagen) and then discussed the results, with the exception of studies on the impact of wind turbine noise (n = 3) and studies on the impact of noise on children’s blood pressure. For those studies data, extraction was carried out by one person only (Elise van Kempen).

18In the selected studies we evaluated the risk of bias by means of a checklist developed by the WHO: (i) information bias due to exposure assessment; (ii) bias due to confounding; (iii) bias due to selection of participants; (iv) information bias due to non-objective health outcome assessment, and (v) information bias due to non-blinded health outcome assessment. A protocol of how the studies were scored on each of these five items can be found in Section 3.3 of the complete evidence review available via the link specified in Section 1.1. For each study, the evaluation was carried out independently by two or three reviewers (Elise van Kempen, Maribel Casas and Göran Pershagen). From the scores on the different items, we calculated a total risk of bias score (see also Appendix B for an overview of the risk of bias scores per study).

19The main effects under investigation were hypertension, IHD, stroke, type 2 diabetes, change in body mass index (BMI), change in waist circumference, and change in mean blood pressure in children. In order to make a comparison between the studies, we expressed their results in a uniform way and calculated the following outcome variables:For studies on the impact of noise on hypertension, IHD, stroke or type 2 diabetes, we calculated the natural logarithm of the Relative Risk (RR) and its variance per 10 dB(A);For studies on the impact of noise on children’s blood pressure, we calculated the blood pressure change (mmHg for a noise level increase of 10 dB(A) and its variance for both systolic and diastolic blood pressure; andFor studies on the impact of noise on obesity markers BMI and waist circumference, we calculated the change in BMI (kg/m2) per noise level increase of 10 dB(A) and its variance, and the change in waist circumference (cm) per noise level increase of 10 dB(A) and its variance.

20For studies on the impact of noise on hypertension, IHD, stroke or type 2 diabetes, we calculated the natural logarithm of the Relative Risk (RR) and its variance per 10 dB(A);

21For studies on the impact of noise on children’s blood pressure, we calculated the blood pressure change (mmHg for a noise level increase of 10 dB(A) and its variance for both systolic and diastolic blood pressure; and

22For studies on the impact of noise on obesity markers BMI and waist circumference, we calculated the change in BMI (kg/m2) per noise level increase of 10 dB(A) and its variance, and the change in waist circumference (cm) per noise level increase of 10 dB(A) and its variance.

23To retain the link with the European Noise Directive (END, we expressed noise exposure in LDEN. However, most studies did not report an RR per 10 dB (LDEN). Where noise exposure was expressed by means of another noise indicator than LDEN (e.g., LAeq,16hr or LAeq,24hr), a conversion to LDEN was needed. Appendix II of the complete review gives an overview of the conversion rules that we applied.

24For data-aggregation, we included only estimates from studies that were well matched, adjusted, or stratified for at least age and sex. If more than one risk estimate was available for a study, we used the estimates for men and women separately and for separate age-categories, where possible. After selecting the study estimates, we calculated a pooled estimate using the STATA-command METAN to fit a random-effects model. To test consistency of the effect estimates across studies, we used Cochran’s Q-test. We calculated the I2-statistic to reflect the percentage of between-study heterogeneity. For some outcomes, we were able to investigate how the summary estimates were affected by sources of heterogeneity. To this end, we carried out a meta-regression analysis using the STATA-command METAREG. Where meta-regression analyses were not possible, we carried out sub-group analyses.

25When enough study estimates were available, we attempted to give insight in the extent of publication-bias by means of funnel plots: scatter plots of the studies’ effect estimates (RR per 10 dB) against the inverse of the standard error. Also we applied Egger’s test of publication bias using the STATA-commands METAFUNNEL and METABIAS.

26The WHO required us to assess the quality of the evidence that has been retrieved in this review. In other words, we had to assess to what extent we were confident that an estimate of an association between noise and an outcome is likely or unlikely to be changed by further research.

27To this end, we applied a modified version of the GRADE considerations: a systematic and explicit approach to making judgements about quality of evidence. In summary, for every outcome, we had to assess the quality of evidence according to several criteria (e.g., study design, study quality, consistency and precision of the results, directness of the evidence, publication bias, whether an exposure-response gradient was present, the magnitude of the effect found, and possible confounding. The scores for the different GRADE criteria are presented in Appendix C to Appendix H as well in Appendices III–VIII of the complete systematic review. How we adapted GRADE for this systematic review is extensively described in Chapter 10 of the complete evidence review. The main divergence from GRADE was that the initial level of certainty was rated “high” for cohort and case-control studies, “low” for cross-sectional studies and “very low” for ecological studies. Furthermore, we upgraded the evidence if the relative risk was 1.5 or higher, but downgraded if based on only one study. GRADE has four levels for the quality of evidence, ranging from “very low” to “high” (see Table 1). The level of the quality of evidence will be linked with the guideline values and recommendations that WHO will include in their environmental noise guidelines.

28In this section, for each outcome the main findings of the review and the conclusions of the weighing of the evidence are presented. The report with the complete findings including the systematic evaluation of the included studies, and the reasoning behind the weighing of the evidence, can be found in the complete systematic review.

29A note for the reader: since we carried out the literature search for this systematic review, new studies have been published that investigate the associations between transportation noise exposure and metabolic and cardiovascular disease. Unfortunately, owing to time constraints, we were not able to carry out a structured and extensive additional search for new studies published in the period November 2014–March 2017. However, in order to identify at least some of the new studies we were missing, we carried out a search on SCOPUS in March 2017. For this, we applied the same SCOPUS-search profile as was used to identify studies for the current review. In an “ideal” systematic review, we should have included the results of these newly identified studies in the results of the current review, and where necessary updated our results. However, due to time constraints, we have not yet been able to systematically evaluate the newly identified studies. Nevertheless, we have decided to present their results in a narrative way, and attempted to assess how they affect the results of the current review. The differences in results with these recent studies and earlier reviews are described in detail for each outcome in the complete systematic review.

30We evaluated 40 studies that investigated the impact of noise from air, road, and rail traffic and wind turbines on the risk of hypertension. Appendix B presents the separate risk of bias tables. Appendix C presents the different GRADE tables (summarized in Table 2).

31There were positive associations between noise from air, road, or rail traffic and hypertension in the cross-sectional studies, which formed the largest part (n = 38) of the available evidence (Table 2). After aggregating the results of 26 studies (comprising 154,398 individuals, including 18,957 cases), we derived an RR of 1.05 (95% CI: 1.02–1.08) per 10 dB (LDEN) for the association between road traffic noise and the prevalence of hypertension. The studies were carried out within the range of approximately 20–80 dB (LDEN). For aircraft noise (nine studies), we estimated an RR of 1.05 (95% CI 0.95–1.17) per 10 dB (LDEN) (comprising 60,121 residents, including 9487). For rail traffic noise (five studies), we derived an RR of 1.05 (95% CI: 0.88–1.26) per 10 dB (LDEN) (comprising of 15,850 individuals, including 2059 cases of hypertension). Although there was evidence for moderate to high heterogeneity among studies, the meta-regression analyses could not reveal clear sources for this observed heterogeneity.

32Despite the fact that most studies were able to adjust for important confounders, and were able to ascertain individual exposure levels, we rated the quality of the evidence from the cross-sectional studies mainly as “very low”. This is, among other reasons, because the response rate in many of the studies was lower than 60%. Furthermore, most studies ascertained hypertension by means of self-report only.

33In the two evaluated cohort studies that investigated the impact of traffic noise on hypertension, no increased risks were found of hypertension related to traffic noise exposure. This is confirmed by a recent meta-analysis, including individual data from six cohort studies on the association between road traffic noise and the incidence of hypertension. The reason for this apparent discrepancy in the findings between the cross-sectional and cohort studies is unclear.

34Overall, we consider the quality of the evidence supporting an association between traffic noise exposure and hypertension as “very low”, indicating that any estimate of effect is very uncertain.

35We evaluated 22 studies that investigated the association between exposure to noise from air, road, and rail traffic and IHD. Appendix B presents the separate risk of bias tables, and Appendix D presents the different GRADE tables (summarized in Table 3). The majority (n = 11) were of cross-sectional design.

36The studies that investigated the impact of air traffic noise found indications of an increased risk of IHD. Exposure to aircraft noise was associated with the prevalence of IHD, the incidence of IHD, and mortality due to IHD. Only the association between aircraft noise and the incidence of IHD was statistically significant. We estimated an RR of 1.09 (95% CI: 1.04–1.15) per 10 dB (LDEN) after aggregating the results of two studies comprising of 9,619,082 participants, including 158,977 incident cases of IHD. Since most studies on the impact of aircraft noise were of ecological and cross-sectional design (see Table 3), the quality of the evidence from these studies was mostly rated as “very low”. However, the results of the current review are consistent with the results of new longitudinal studies, which reported positive associations between aircraft noise and mortality due to IHD.

37Overall, we rate the quality of the evidence supporting an association between air traffic noise and IHD as “low”, indicating that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

38We found evidence that noise from road traffic is associated with an increased risk of IHD. An increase in road traffic noise was associated with significant increases in the prevalence of IHD, and the incidence of IHD. The evidence for a relationship between noise from road traffic and the incidence of IHD was the most robust. After combining the results of three cohort studies and four case-control studies (comprising 67,224 participants, including 7033 incident cases of IHD, we found an RR of 1.08 (95% CI: 1.01–1.15) per 10 dB (LDEN) for the association between road traffic noise and the incidence of IHD within the range of approximately 40–80 dB LDEN. This means that if road traffic noise levels increase from 40 to 80 dB (LDEN), the RR = 1.36. We rated the quality of the evidence that comes from these studies to be “high”. Supporting evidence came from studies on the association between road traffic noise and the prevalence of IHD. We rated the quality of evidence from these studies as low. The results of the current review are strengthened by the results of several recently published longitudinal studies.

39A visualization of the shape of the association between road traffic noise and the incidence of IHD, indicated that the risk of IHD increases continuously for road traffic noise levels from about 50 dB (LDEN). This is consistent with the findings of another recent meta-analysis on the association between road traffic noise and IHD. The WHO guidelines of 1999 stated the following: “epidemiological studies show that cardiovascular effects occur after long-term exposure to noise with LAeq,24hr values of 65–70 dB”. In the WHO Night-noise guidelines, published in 2009, a general threshold of 55 dB (LNight) was recommended for protection of cardiovascular disease.

40Overall, taking into account all available evidence on the association between road traffic noise on IHD, we rate the quality of the evidence supporting an association between road traffic noise and IHD to be “moderate”, indicating that further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. However, for road traffic noise and the incidence of IHD, the quality of the evidence was rated as high.

41Compared with noise from road and air traffic, we found only a few studies that investigated the impact of noise from rail traffic. These had a cross-sectional design. After aggregating the results of the studies on the association between rail traffic noise and the prevalence of IHD, we found a non-significant RR of 1.18 per 10 dB (LDEN).

42Overall, we rate the quality of the evidence supporting an association between exposure to noise from rail traffic and IHD to be “very low”, indicating that any estimate of effect is very uncertain.

43Compared with the number of studies on the impact of noise on hypertension and IHD, relatively few studies were available that investigated the impact on stroke (n = 9). Appendix B presents the separate risk of bias tables, and Appendix E presents the different GRADE tables (summarized in Table 4).

44According to the results of the ecological and cross-sectional studies an increase in aircraft noise was associated with an increase in the prevalence and the incidence of stroke. None of these associations was statistically significant (see Table 4). The observations found for the prevalence and incidence of stroke were supported by the ecological studies on the association between air traffic noise and mortality due to stroke.

45No association between air traffic noise exposure and mortality due to stroke was observed in the evaluated cohort study. This is consistent with the results of new longitudinal studies, which showed no clear indications of an association between aircraft noise and mortality due to stroke.

46The results of the studies that investigated the impact of road traffic were not consistent. Only for the association between road traffic noise and the incidence of stroke, there was a statistically significant RR of 1.14 (95% CI 1.03–1.25) per 10 dB (LDEN). This result was based on one cohort study, comprising 51,485 participants, including 1881 incident cases of stroke.

47In the evaluated cross-sectional and ecological studies on the association between road traffic noise and the prevalence of stroke or mortality due to stroke, no increased risks of stroke due to road traffic noise were observed. This was not consistent with the results of recently published longitudinal studies, which showed that an increase in road traffic noise was statistically significantly associated with an increase in mortality due to stroke. As part of the current review, only one cross-sectional study was evaluated, which investigated the association between rail traffic noise and the prevalence of stroke.

48Overall, we rate the quality of the evidence supporting an association between traffic noise and stroke to be “low”. This indicates that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

49For the current review, we were able to evaluate seven studies that investigated the association between environmental noise and the risk of diabetes. Four studies investigated the possible impact of transportation (air, road, rail traffic noise. Appendix B presents the separate risk of bias tables, and Appendix F presents the different GRADE tables (summarized in Table 5).

50We found two studies that investigated the impact of air traffic noise on the occurrence of diabetes. In a cross-sectional study on the association between air traffic noise and the prevalence of diabetes, a non-significant RR of 1.01 per 10 dB (LDEN) was found. In the evaluated cohort study on the association between air traffic noise and the incidence of diabetes, no increased risk of diabetes due to air traffic noise was observed (see Table 5).

51We found indications that noise from road traffic increases the risk of diabetes. The two evaluated cross-sectional studies showed an increasing but non-significant trend of the prevalence of diabetes with road traffic noise exposure. In the evaluated cohort study, an increase in road traffic noise was statistically significantly associated with an increase in the incidence of diabetes. An RR of 1.08 (95% CI: 1.02–1.14) per 10 dB (LDEN) across a noise range of approximately 50–70 dB (LDEN) was estimated.

52Remarkably, an increase in rail traffic noise was associated with a decrease in the risk of diabetes in one cross-sectional study while a cohort study found no statistically significant association.

53Overall, we rate the quality of the evidence supporting an association between traffic noise and diabetes to be “low”. This indicates that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

54The number of evaluated studies that investigated the impact of noise on markers of obesity was limited to four: one cohort study and three cross-sectional studies. Appendix B presents the separate risk of bias tables, and Appendix G presents the different GRADE tables (summarized in Table 6). All the studies showed that an increase in traffic noise was associated with an increase in obesity markers, although, according to one study, this was present only in certain subgroups. In the cohort study, an increase in aircraft noise of 10 dB (LDEN) was associated with a significant increase in waist circumference of 3.46 (95% CI: 2.13–4.77) cm during 8 to 10 years of follow-up (see Table 6). The evidence of traffic noise affecting obesity markers is strengthened by the results of two recent longitudinal studies.

55Overall, we rate the quality of the evidence supporting an association between traffic noise and markers of obesity, respectively, as “low”. This indicates that further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

56We evaluated eight studies investigating the impact of noise on children’s blood pressure. Appendix B presents the separate risk of bias tables, and Appendix H presents the different GRADE tables (summarized in Table 7). Seven studies were cross-sectional; one study reported both the results of cross-sectional and longitudinal analyses. With the exception of the association between road traffic noise at school and systolic blood pressure, we observed positive but non-significant associations between exposure to road traffic noise and blood pressure (see Table 7). No combined exposure-response estimate could be computed from the studies on the impact of aircraft noise, since no quantitative results were provided in one of the studies.

57Overall, we rate the quality of the evidence supporting an association between traffic noise and blood pressure in children, as “very low”, indicating that any estimate of effect is very uncertain.

58Overall, we evaluated only three cross-sectional studies that investigated the impact of noise from wind turbines on the cardiovascular and metabolic systems. Important limitations of these studies were the low response rates (two studies had response rates of less than 60%) and, the fact that in all studies the cardiovascular or metabolic endpoint was ascertained by questionnaire or interview. In these studies, we observed that an increase in wind turbine noise was associated with non-significant increases in self-reported hypertension and non-significant decreases in self-reported cardiovascular disease. For self-reported diabetes, the results appeared inconsistent.

59Overall, we rate the quality of the studies supporting an association between exposure from wind turbine noise and adverse effects in the cardiovascular or metabolic system as “very low”, indicating that any estimate of effect is very uncertain.

60The current review shows that a large number of studies have investigated the impact of noise on the cardiovascular system, but applying the GRADE, the quality of the evidence is often rated as relatively low. This does not mean that exposure to noise has no effect on the cardiovascular system, but encourages further research to improve the quality of the evidence. After all, there is a strong biological plausibility that noise affects human health. Furthermore, in many of the evaluated studies, we observed statistically significant associations between noise and cardiovascular endpoints. The most robust were the effects of road traffic noise in relation to IHD. Combining the results of 7 longitudinal studies, revealed an RR of 1.08 (95% CI: 1.01–1.15) per 10 dB (LDEN) for the association between road traffic noise and the incidence of IHD. We rated the quality of the evidence from these longitudinal studies as high. Supporting evidence came from studies on the association between road traffic noise and the prevalence of IHD.

61Several recent reviews have been published on cardiovascular effects of environmental noise exposure, which are described in detail in the full systematic review. The quantitative results regarding exposure-response relationships following meta-analyses agree well with our review. However, most earlier reviews did not include a detailed quality assessment of individual studies.

62This review also addressed the possible impact of noise on the metabolic system. In comparison with the studies on the impact of noise on the cardiovascular system, the number of available studies was rather limited. The results of these studies were not always consistent. In addition, the quality of the evidence was rather low. It is therefore, at this moment too early to draw definite conclusions with regard to the impact of noise on the metabolic system.

63The results of the current review shows that at this moment, not enough studies of good quality are available that investigated the impact of noise on the cardiovascular and metabolic system. The plausibility of an association calls for further efforts with improved research. In order to improve the quality of the existing evidence, more studies with a cohort or case-control design are needed.

64In order to improve the quality of the existing evidence, we also recommend that more well designed studies on health effects in relation to exposure to wind turbines and rail traffic noise are set up and carried out.