Road traffic noise exposure was not associated longitudinally with IHD morbidity and mortality in these cohort analyses. There was also no consistent association with psychological ill-health, although there have been associations with depressive symptoms and insurance claims for depressive illness in other studies [8, 33]. Earlier analyses in this cohort did find a longitudinal association between road traffic noise and symptoms of anxiety . As hypothesised the interaction of road traffic noise, noise sensitivity and IHD outcomes was not statistically significant. By contrast, in accordance with the hypothesis, road traffic noise and noise sensitivity did show significant interactions with psychological ill-health. Independently, noise sensitivity was associated with lower rates of mortality and, in contrast, predicted higher levels of psychological ill-health, the latter in keeping with our initial hypothesis.
Earlier analyses in this cohort, at 10 year follow up, showed an increased relative risk of IHD in relation to road traffic noise especially among those 15 years or more in the same place . Exposure misclassification and the lengthening of the interval between baseline traffic noise assessment and ascertainment of health outcomes may explain the lack of association in these recent analyses. There may have been self-selection out of the noisiest areas or not into noisy areas, even at baseline, for the most noise sensitive, or this could have occurred during follow up. However, mobility of noise sensitive persons out of noisy areas has not been found in other studies .
Previous studies have also found that noise sensitivity did not moderate the association of road traffic noise exposure and IHD events . Similarly, noise sensitivity was not a predictor of cardiovascular outcomes in an earlier study of civil servants except for participants in the lower employment grades where it predicted angina pectoris . This is in contrast to a study of Finnish twins where noise sensitivity was a predictor of cardiovascular mortality in women but not in men . Such a gender difference is in keeping with the results from the Caerphilly Study in men but not with the results in civil servants both men and women (Whitehall II Study), although in the latter study men and women were combined for analysis. A further explanation might lie in the outcomes chosen. In the Finnish Study noise sensitivity was a significant predictor of cardiovascular mortality (ICD codes 390–459,100–199) but not coronary heart disease mortality (ICD codes 410–414, 120–125). In the Caerphilly Study we only included ischaemic heart disease (also known as coronary heart disease ICD codes 410–414). Although this outcome measure issue would not explain the lack of results in the Whitehall Study which did include cardiovascular outcomes such as stroke morbidity and mortality.
If noise sensitivity were an independent predictor of physical ill-health it would be expected that it should be associated increased mortality rates. The fact that we found it was associated with lower mortality rates was not what we expected, although in keeping with findings from a previous study . Lee et al  found that highly anxious young people had lower accident mortality up to the age of 25 years because they tended to avoid putting themselves in high risk situations which could have high mortality risk attached. Conversely, in older people they found high levels of anxiety were associated with higher non-accident mortality rates, perhaps as a result of prolonged physiological hyper-reactivity associated with chronic anxiety. Our cohort was middle-aged and older men, not strictly comparable with the population in Lee et al’s study, nevertheless, it may be that noise sensitive people are more cautious and less likely to take risks that could increase mortality. Noise sensitivity has been associated with phobic disorders in a sample of women  and fearfulness and avoidance which are part of phobic disorders might be characteristic of some people with noise sensitivity and could be associated with health-protective behaviours. Earlier analyses in this cohort found an association between noise level and noise sensitivity with less highly sensitive men living in the highest noise exposure areas so it may be that more sensitive men tend to choose to live in less noisy areas where that choice is possible . It does not seem that this effect on mortality is mediated through health behaviours as our results were adjusted for smoking, leisure-time physical activity, BMI and alcohol use. Lower mortality rates were also found when noise sensitivity was replaced by trait anxiety in the models supporting an essential role for long-term anxiety in this association.
Noise sensitivity may moderate the effects of road traffic noise on psychological ill-health, although there was some inconsistency between phases 3 and 4 and the confidence intervals were wide so that our analyses may have been underpowered. Independently of road or aircraft noise exposure noise sensitivity has been shown to be strongly associated with a range of common mental disorders [21, 23, 38, 39, 40]. Most of these earlier studies have been cross sectional, this study as well as another study  have confirmed that noise sensitivity is associated with psychological ill-health longitudinally.
Noise sensitivity has also been associated with neuroticism [21, 23, 38, 41, 42]. Neuroticism is a construct similar to trait anxiety and has links to negative affectivity, a tendency to report life experiences and perceptions negatively. Noise sensitivity does not seem to be just trait anxiety or neuroticism and Shepherd et al.  have found higher correlations with introversion/extraversion than with neuroticism . A question often asked in this literature is how specific is noise sensitivity to noise or much is it part of a wider range of responses to environmental stimuli driven largely by chronic anxiety ? Although, inevitably, much of the predictive power of noise sensitivity for common mental disorders derives from its association with trait anxiety  we did find that these associations still remained after adjusting for Spielberger trait anxiety. It is possible that there was still some residual confounding in the analyses of noise sensitivity and psychological ill-health adjusting for trait anxiety.
Noise sensitivity has been linked to sensitivity to other aspects of the environment such as sensitivity to chemicals, electromagnetic fields, light and odours [40, 45, 46]. Sensitivity to chemicals is often defined under the rubric of Multiple Chemical Sensitivity (MCS). MCS is a condition purported to be related to exposure to low levels of environmental chemicals which for most people would not result in health effects. In one study seventy three per cent of MCS also were noise intolerant . Another name for this condition that suggests it covers a broader spectrum of exposures than just chemicals is ‘Idiopathic Environmental Intolerance’ (IEI). This is defined by three criteria: ‘1. It is an acquired disorder with multiple recurrent symptoms. 2. It is associated with diverse environmental factors tolerated by the majority of people. 3. It is not explained by any known medical or psychiatric/psychological disorder’ . A strong overlap has been found between Idiopathic Environmental Intolerance and Somatoform Disorders. In a study comparing Somatoform Disorders with Idiopathic Environmental Intolerance more than half of the latter group could be classified as fulfilling the criteria for Somatoform Disorders . Both groups also had higher levels of trait anxiety and somatic symptom attributions than the control group. More subjects with IEI reported allergies than the control group although these were not supported by objective changes in IGE levels. There was longitudinal stability of these conditions over a year and baseline negative affectivity and somatosensory amplification (a tendency to focus and amplify symptoms) predicted these conditions at one year follow up . A strong association has been found between Idiopathic Environmental Intolerance and mood, anxiety and somatoform disorders across the lifecourse  and equally between MCS and major depressive disorder, and generalised anxiety disorder and severe psychological distress .
It has been proposed that this environmental sensitivity may relate to a hyper-responsive central nervous system with increased reactivity of the limbic system in the brain although no physiological evidence has been found to support this . However, noise sensitivity is not always accompanied by other environmental sensitivities. Baliatsas et al,  found noise sensitivity associated with environmental sensitivity in 9–50% of highly noise sensitive people in their general practice community sample. Thus although noise sensitivity may be a symptom of IEI or even somatisation disorder in some cases it is not necessarily associated with other environmental sensitivities in all cases [35, 53]. To that extent noise sensitivity is not a single reified entity but may be a non-specific indicator of sensitivity to sounds alone or part of a wider IEI or psychiatric syndrome. Thus noise sensitivity might have a multiple origins .
Noise sensitivity has been associated with uncomfortable loudness levels in laboratory studies but has not been associated with especially sensitive hearing thresholds [54, 55]. Thus it does not seem to be related abnormalities in the peripheral auditory system. The associations with sympathetic nervous system activity may reflect associations with state or trait anxiety rather than being specific to noise sensitivity [11, 56]. A study using electro- and magnetoencephalography measuring mismatch negativity found that noise sensitivity categorised with the Weinstein scale was associated with altered sensory processing in the auditory cortex implying a central cortical origin for noise sensitivity . This very interesting study requires replication; it is a type of neurophysiological validation of a self-report noise sensitivity scale but it does not directly link these auditory processing characteristics to vulnerability to ill-health as might be expected if noise sensitivity is related to increased susceptibility to ill-health. Intriguing exploratory EEG studies suggest that there may be a deficit in sensory gating in noise sensitive subjects leading to sensory ‘overload’ . Further research in these disciplines may well be productive.
It may be difficult to generalise too far from these results as the population, although representative of the local area, was confined to middle-aged and older men living in a very specific geographical area. A strength of the study was the careful ascertainment of cardiac outcomes, the high response rate and follow up response longitudinally. The psychological ill-health outcomes would have been stronger had we had a standardised psychiatric interview instead of a questionnaire. Missing data for psychological ill-health outcomes in phase 3 and 4 was a limitation. Adjustment for room orientation could be over-adjustment as it may remove some of the variance due to noise exposure, although analyses not including adjustment for room orientation were little changed.