Road traffic noise and psychiatric disorder: prospective findings from the Caerphilly study

BMJ 1996; 313 doi: (Published 03 August 1996) Cite this as: BMJ 1996;313:266
  1. Stephen Stansfeld, senior lecturera,
  2. John Gallacher, research psychologistb,
  3. Wolfgang Babisch, research scientistc,
  4. Martin Shipley, senior lecturerd
  1. a Unit for Social and Community Psychiatry, Department of Psychiatry and Behavioural Sciences, University College Medical School, London W1N 8AA
  2. b MRC Epidemiology Unit (South Wales), Llandough Hospital, Penarth, Glamorgan
  3. c Institut fur Wasser- und Lufthygiene des Bundesgesundheitsamtes, Berlin, Germany
  4. d Department of Epidemiology and Public Health, University College London Medical School, London WC1E 6BT
  1. Correspondence to: Dr Stansfeld.
  • Accepted 16 May 1996

Although environmental noise causes annoyance, there is little evidence from studies of psychological symptoms, psychotropic drug use, mental hospital admissions, and community studies that it causes psychiatric disorder.1 Exposure to aircraft noise was not associated with psychiatric disorder in a cross sectional survey in west London,2 but the population exposed to noise may have been biased by prolonged noise exposure and may represent “survivors” of noise, the most vulnerable to noise having moved away or never having moved into the noisy area. The relation between traffic noise at baseline and psychiatric disorder at follow up is explored in the Caerphilly collaborative heart disease study, a population unlikely to have been selected by noise exposure.

Subjects, methods, and results

In the second phase of the Caerphilly study all men aged 50-64 years living in Caerphilly, south Wales, were invited to attend a screening clinic.3 Follow up was carried out five years later. Street measurement of A-weighted sound pressure level (dB(A)) was used to derive traffic noise maps, and subjects were grouped into five-decibel categories of traffic noise emission level, in terms of the average sound pressure (Leq) from 6 am to 10 pm.4

The 30 item general health questionnaire was used to establish the presence of psychiatric disorder; it was validated against psychiatric interview in a subsample and a case threshold of 4/5 was established.5 Depression and anxiety subscales were extracted from the general health questionnaire.

The association between noise exposure and subsequent psychiatric disorder was investigated using multiple regression. We adjusted for the possible confounding effect of baseline morbidity by including this as a covariate in the regression model. Least squares means have been presented both unadjusted and adjusted for baseline psychiatric morbidity so that the magnitude of this confounding can be examined. This sample size has a power of 90% to detect differences of 0.22 in anxiety scores, 0.15 in depression scores, and 0.45 in general health questionnaire scores between high and low noise levels.

A total of 2398 men were present in the phase 2 sample of the study, the baseline for these analyses. The initial response rate to the survey was 89%. At five year follow up, 162 men had died, 31 had moved, 337 refused to attend the clinic, and 143 either refused the questionnaire or it was omitted. Prospective data were available for 1725 men. Non-response at follow up was not associated with either noise exposure or baseline psychiatric caseness.

If traffic noise causes psychiatric disorder, a dose-response relation between traffic noise level and psychiatric disorder might be expected. We found no overall association (or linear trend) between noise level at baseline and psychiatric disorder measured prospectively (table 1), even after adjusting for confounding factors such as social class, employment status, marital status, physical illness, and baseline morbidity. However, the subjects living at the lowest noise level had the lowest level of psychiatric disorder. Adjusting for room orientation to the street and hearing threshold made little difference to these results.

Table 1

Association between road traffic noise level and psychiatric disorder. Values are mean (SE) score on general health questionnaire (GHQ)unless otherwise stated

View this table:

Noise could be more specifically related to anxiety or depression than to the broad span of morbidity covered by the general health questionnaire. Subscales for anxiety and depression extracted from the questionnaire showed similar patterns as above, although there was a non-linear relation between noise exposure and anxiety, adjusting for baseline anxiety score (table 1).


Although there was little association between road traffic noise level at baseline and overall minor psychiatric disorder at follow up, there was some evidence for differences in anxiety scores. The results of this prospective study confirm the results of previous cross sectional studies and suggest that environmental noise is not an important cause of overall psychiatric disorder but nevertheless may contribute to anxiety. We cannot rule out the possibility that effect modification by unmeasured variables or response bias in the measurement of morbidity may be masking an association between noise and psychiatric disorder. The traffic noise levels in this sample are fairly typical of those in Britain but do not include the highest levels of traffic noise exposure and therefore do not preclude an association at higher levels of noise exposure. It is also possible that environmental noise might have a pathogenic effect on mental health only in concert with other stressors which have not been assessed in this study.

We thank Dr Peter Elwood for his continuing support and encouragement, Peter Sweetnam and Dan Sharp for helpful statistical advice, David Poor for data management, and Louise Price for secretarial support.


  • Funding The study was supported by the Medical Research Council. Stephen Stansfeld was a Wellcome Trust training fellow during the initial period of the study.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract