Noise From Civilian Aircraft in the Vicinity of Airports - Implications for Human Health - Noise, Stress and Cardiovascular Disease

3.3 Cardiovascular Disease in Adults (Continued)

The evidence was reviewed as to whether aircraft noise may be a risk factor for cardiovascular disease, taking into account criteria discussed in Appendix 1 for guiding the determination of a causal relationship in environmental studies. There are very few studies of environmental aircraft noise, in the vicinity of airports, dealing with cardiovascular disease. The main ones are the study by Altena, et al., (1989) (as cited in Passchier Vermeer, 1993 and Pulles, et al., (1990)) and the studies done by Knipschild (Knipschild, 1977a; Knipschild, 1977b; Knipschild, 1977c). Therefore, in an effort to assess the risk of cardiovascular disease, traffic noise studies have also been considered in this report.

Of the traffic noise studies, the Caerphilly and Speedwell study (Babisch, et al., 1993; Babisch, et al., 1999) of the effects of noise on cardiovascular disease, and its risk factors, is the most persuasive because it has a longitudinal and prospective design, with a follow-up of 10 years. The study also has reasonably well defined health outcomes and exposure levels and more controls for confounding factors than other studies.

3.3.1 Hypertension

The study by Altena, et al., (1988), (Pulles, et al., 1990), was a cross-sectional study which examined 830 persons exposed to military aircraft noise and road traffic noise. The study population was divided among six exposure intervals. Prior to adjusting for confounding factors, regression analysis showed a statistically significant increase in systolic blood pressure with aircraft noise exposure. However, there was no significant relation between noise and blood pressure level after adjustments had been made for known risk factors such as age, sex, relative body mass, etc.

Knipschild studied the consequences of aircraft exposure around Schipold airport. His investigation consisted of three parts: a prevalence study of cardiovascular disease (Knipschild, 1977a), a survey of general practitioners for attendance for cardiovascular disease (Knipschild, 1977b), and a survey of purchases of hypertensive and other cardiovascular medications by pharmacies (Knipschild, 1977c). Respectively, these studies reported increases in: (i) prevalence of hypertension; (ii) attendance at the general practitioners office for cardiovascular disease; and (iii) purchases by pharmacists of cardiovascular medication, particularly antihyper-tensives, with increased levels of aircraft noise.

Data for the study of cardiovascular disease (Knipschild, 1977a) was collected by inviting members of a community consisting of eight villages to undergo medical examination. The medical screening included the collection of medical history data, measurement of blood pressure, x-ray of the heart, and ECG. There were approximately 6000 people medically screened, at a response rate of about 40%. Respondents were separated into high and low noise exposure groups. The high noise exposure group began at about a day-night sound level (Ldn) of about 62 dBA. (The day-night sound level is the time-averaged sound level obtained by averaging the sound exposure from 0700 one day to 0700 the next with the sound level being increased by 10 dB between 2200 and 0700 hours. The Ldn value of 62 dBA was estimated from the Dutch exposure units reported in Knipschild's work using a conversion factor in section 2.1 of Passchier-Vermeer (1993)).

The percent of individuals with measured blood pressure above 175/100 was greater in the high-noise group (Relative Risk =1.8, p < 0.05). Also, the percent of participants undergoing medical treatment for hypertension was greater in the high noise group (Relative Risk = 1.5, p < 0.05). This is traditionally considered statistically significant. The influence of age and sex was taken into account in the analysis. The influence of other confounding factors was not explicitly shown in the study. However, the author indicated that smoking, body mass, and village size had been taken into account, where possible and this information was collected as part of the medical screening. The author noted that there were some indications that the socio-economic status of the high-noise group may be lower, but there is no indication that this was accounted for in the study. Lack of control for this last confounding factor has been a recurring criticism of this study in several reviews (Cohen, et al., 1986; Thompson, et al., 1989; Berglund and Lindvall, 1995).

The survey of general practitioners took place in three villages around the airport over a one-week period in which 19 general practitioners recorded the age, sex, address, reason for visit (diagnosis) and medication used for all of their patients. The contact rate for cardiovascular diseases was reported as being greatest in the village with the highest noise exposure. Similarly, the usage of antihypertensive medication was higher, especially in women. As in the previous study, the effects of socio-economic status were not taken into account. The author noted that the control population, with the lowest noise exposure, had a higher-socioeconomic status and a greater proportion of white-collar workers.

The drug survey (Knipschild, 1977c) was conducted in 2 villages around Schipold airport, the high exposure and control villages of the general practitioner survey (Knipschild, 1977b). Over the period 1967-1974, drug purchases by the village pharmacies, per adult per year were used as an indicator of the consumption of medications in the subject populations. The village designated as the high noise exposure area experienced a change over time in the noise exposure levels; from little prior to 1969, to high noise exposure from 1969-1973 and only daytime noise during 1973-74. By contrast, the noise level in the control area remained constant. In the exposed area, a gradual increase was reported for purchases of cardiovascular drugs by the village pharmacies. The final value was up to two times the initial rate, the largest contribution being for antihypertensive drugs. This was not affected by the reduction in nighttime noise level in 1973. In the control area there was no change over time.

These studies suggest that there may be an association between hypertension and aircraft noise at Ldn values greater than about 62 dBA. However, the evidence for this association is not convincing because of the lack of controls for socioeconomic status in the first two studies and the lack of a statistical analysis in the third.

As there are so few studies on aircraft noise and cardiovascular effects in adults, it was necessary to broaden the review to include traffic noise studies. Babisch has recently completed a comprehensive review of this subject (Babisch, 2000). In this paper, Babisch notes that dose assessments in most of the traffic noise studies were crude, usually based on noise maps of the region. Studies usually only had two exposure groups - low and high. Also, subjective estimates of exposure were sometimes used. For example, Herbold (1989) based his estimates on self-reporting of the type of road adjacent to the participant's home. The noise levels for these types of roads were then simply grouped as "low" and "high," depending on the type of road. Neus, et al., (1983a, 1983b) based their estimates of noise exposure on traffic volume.

From the review by Babisch (2000), it is also clear that, using 95% confidence intervals, associations were not consistently found in independent studies. Only 4 of 10 s tudies reviewed by Babisch yielded associations between traffic noise and hypertension. Of these 4 studies, Babisch noted that 2 would meet modern standards of control for confounding factors. These were both cross-sectional studies. Babisch (2000) concluded that there was little epidemiological evidence of an increased risk of hypertension in subjects exposed to traffic noise.

3.3.1.1 Hypertension - Conclusions

The review of studies that investigated the potential link between hypertension and either aircraft or traffic noise exposure, indicated that the available evidence does not appear to convincingly demonstrate an association between aircraft noise and hypertension.

3.3.2 Ischemic Heart Disease

Ischemic heart disease is characterized by insufficient perfusion of oxygen to the heart muscle. For the study by Altena, et al., (Pulles, et al., 1990; Altena, et al., 1988), ischemic heart disease was assessed by clinical symptoms of angina pectoris (chest pain), myocardial infarction (heart muscle damage), or electrocardiogram (ECG) abnormalities as defined by criteria of the World Health Organization (WHO). This study, described in the previous section, did not show any increase in the prevalence of ischemic heart disease with increasing exposure to aircraft or traffic noise. There is a possibility that the negative finding may have been due to selection bias since those with hypertension were excluded from the study and this condition is a known risk factor for ischemic heart disease.

Details of the review of the Knipschild studies are provided above. The prevalence of cardiovascular disease was determined by a questionnaire and clinical examination (Knipschild, 1977a). The following data were recorded for each participant: clinical symptoms of angina pectoris (according to a standard WHO questionnaire), medical treatment for heart trouble and hypertension, usage of cardiovascular drugs, ECG abnormalities, heart shape and blood pressure measurements. The results indicated that, in the noise exposed population, there was a statistically significant increase in all but two of these endpoints, compared to the control group. The Relative Risks and p values were: (i) 1.4, p < 0.05 for medical treatment for heart trouble; (ii) 1.4, p < 0.01 for usage of cardiovascular drugs; and (iii) 1.6, p < 0.05 for pathological heart shape. No statistically significant differences were found for angina pectoris and ECG abnormalities, which are important indicators of ischemic heart disease, respectively.

The general practice survey showed an increase in contacts with the physician for cardiovascular disease in the noise-exposed area. The drug survey also showed an increase in cardiovascular drug use over time in a noise-exposed area. However, neither of these studies provides sufficient information to adequately assess their relevance to ischemic heart disease.

The shortcomings of the Knipschild studies have been described in the discussion above on hypertension. These shortcomings also apply to the ischemic heart disease endpoints. Furthermore, given the lack of statistically significant associations between noise level and two important indicators of ischemic heart disease in these studies, the available aircraft noise studies do not provide convincing evidence of an association between ischemic heart disease and environmental aircraft noise exposure.

The remaining studies on environmental noise and ischemic heart disease are traffic noise studies. These include a retrospective study of myocardial infarction in the city of Erfurt (Babisch, 2000) and 2 prospective case control studies of myocardial infarction in Berlin, Germany (Babisch, et al, . 1994). In addition, prospective 10 year longitudinal studies of cardiovascular risk were done in the cities of Caerphilly and Speedwell in Wales and England, respectively (Babisch, et al., 1993; Babisch, et al., 1999). In Babisch's review of these studies (Babisch 2000), noise levels were reported as outdoor time-averaged traffic noise levels (06:00-22:00).

As reported by Babisch (2000), a high and significant proportional morbidity ratio in the Erfurt study was determined for areas with noise levels between 71-75 dBA compared to areas with noise levels of 61-65 dBA. However, methodological issues about the validity of the results have been raised by Babisch (2000). (This study is only available in German and has not been reviewed by the authors).

In the Berlin pre-study and main study (Babisch, et al, . 1994), increase in incidence of myocardial infarction was assessed relative to populations living in areas with noise levels less than 60 dBA. Increases were observed but they were not statistically significant at the 95% confidence level. The lower limit of the 95% confidence interval was less than 1.0 for all odds ratios determined in these studies. The reported values of the odds ratios in the pre-study were 1.5 and 1.2 in the 61-65 dBA and 66-70 dBA noise level range, respectively. The 95% confidence intervals were 0.6-3.9 and 0.5-2.9, respectively. The corresponding odds ratios in the main study were 1.2 and 0.9 in the 61-65 dBA and 66-70 dBA noise level range, respectively. The corresponding 95% confidence intervals were 0.8-1.7 and 0.6-1.4. Even the reported mean values of the odds ratios (relative risks) showed no consistent trend with noise level under 70 dBA. The small sample size of the pre-study led to only one case being available in the highest range of noise levels, precluding any further conclusions from being drawn from that study. In the main study, above 70 dBA, the mean value of the relative risk increased from 1.1 to 1.5 as the range of noise levels increased from 71-75 dBA to 76-80 dBA. The 95% confidence intervals were 0.7-1.7 and 0.8-2.8, respectively.

Further analysis of the Berlin study data was carried out by Babisch for the two highest ranges of noise levels. First, to ensure that the subjects had undergone sufficient exposure to the noise, analysis was restricted to subjects who had lived in the study areas for more than 15 years. Also, to improve the statistical power, the data was grouped into a single high noise exposure level of 71-80 dBA. The relative risk was then found to be 1.3 with a 95% confidence interval of 0.9-2.0. Babisch considered this result to be borderline significant (p < 0.10). The results of this study are not sufficient to demonstrate an association of traffic noise level with incidence of myocardial infarction. However, they suggest that further research is needed in areas with high populations exposed to high noise levels.

In the Caerphilly and Speedwell studies, the increase in risk in noise-exposed areas was assessed relative to populations in areas where the noise levels were less than 55 dBA. The Caerphilly and Speedwell studies are a series of investigations in which two cohort studies were done on the effects of traffic noise. They were part of a larger study to examine the predictive power of known and new risk factors for ischemic heart disease. These cohorts were studied over a ten year period. Combined analysis is available for a 6 year period.

This Caerphilly and Speedwell study has advantages over many others in that it is prospective in design. Exposure assessment is based on noise level measurement. Disease outcome is determined by hospital records according to well defined criteria. More confounders have been taken into account in this analysis than in any other study.

In the Caerphilly 10 year follow-up there was a slightly higher relative risk of ischemic heart disease in the 56-60 dBA and the 66-70 dBA subgroups, but this was only marginal and non-significant. In the Speedwell 10 year follow-up there was no increase in ischemic heart disease in any of the groups. However, Babisch, et al., (1999) also provided an analysis of the data, pooling the populations in a 6 year follow-up. For this pooled data, in the highest noise-exposed group, 66-70 dBA, the adjusted odds ratio increased from 1.07 to 1.59 as further refinements were made to the exposure classification of the subjects. The 95% confidence interval also varied with these changes from 0.70-1.65 to 0.85-2.97. This included examining a subsample in residence not less than 15 years and taking into account window orientation, and window-opening practices.

Babisch, et al., (1999) also analyzed the data using an alternative model in which the noise exposure was set equal to th e product of noise level with years of residence. Using this analysis, there is an increase in the odds ratio per year in residence from 1.007 to 1.017 in the highest noise category as the exposure assessment is refined accounting for window orientation and window opening practices. The 95% confidence intervals ranged from 0.992-1.023 to 0.998-1.036, respectively. The odds ratio of 1.017 was considered by Babisch, et al., (1999) to be borderline significant, being greater than unity at a p value < 0.10.

The results of the Caerphilly and Speedwell study appear to be equivocal. The trend of increasing odds ratios, with improvements to the exposure assessment and when time of residence was considered, suggests there may be a slight increase in ischemic heart disease among those exposed to chronic high levels (> 66 dBA) of environmental noise. Nevertheless, there is considerable overlap of the 95% confidence intervals of these odds ratios suggesting that the increase may only have been due to chance. Also, the odds ratio of 1.017 per year in residence was only considered to be borderline significant by Babisch, et al., (1999), again suggesting that the change in odds ratio with increasing time of residence may also have been due to chance.

3.3.2.1 Ischemic Heart Disease - Conclusions

There is no convincing evidence for a causal relationship between environmental noise and ischemic heart disease. At traditional 95% confidence levels used to assess statistical significance, dose response relationships have not been demonstrated. Also, potential trends with improved exposure assessment procedures and increasing years in residence may have been due to chance. Furthermore, the strength of the associations is typically relatively weak, with observed relative risk ratios or odds ratios ranging from 1.3 to 1.6, at most in the Berlin and the Caerphilly and Speedwell studies. In these studies, important confounding factors were taken into account and efforts had been made to reduce bias, including the effort of determining exposures by measurement.

However, the available studies provide some evidence to suggest that there may be a slight increase in the risk of ischemic heart disease in people residing in areas with daily averaged traffic noise levels greater than 65 dBA. This indicates that more research on this subject is needed. Also, there needs to be continued assessment of future research on the potential for cardiovascular risks from aircraft noise. This follows from the relative consistency of elevated risk among the exposure groups with daily averaged sound levels greater than 65 dBA. It also follows from the temporal effect suggested by the increasing odds ratios with years of residence in the Caerphilly and Speedwell study. The need for more research in this area is also consistent with the suggested trend of increasing odds ratios with improved exposure assessment.