764 74horax 1995;50:764-768 Effect of ambient levels of smoke and sulphur dioxide on the health of a national sample of Thorax: first published as 10.1136/thx.50.7.764 on 1 July 1995. Downloaded from 23 year old subjects in 1981 Jane F Scarlett, Joanne M Griffiths, David P Strachan, H Ross Anderson Abstract episodes of air pollution could have important Background - There is concern that, des- effects on mortality and led to the Clean Air pite the fall in air pollution levels since the Act of 1956. The Clean Air Act dramatically 1950s, there may still be adverse effects at reduced emissions and ambient levels ofsmoke current levels. A study was carried out and SO2 from coal combustion.2 The as- to investigate the association between air sociation between air pollution and mortality3 pollution and respiratory symptoms in 23 and morbidity from bronchitis4 persisted in the year old subjects in 1981. 1960s, but studies from the 1970s found no Methods - Data on cough, phlegm, and association.'9 It was generally believed that air wheeze were available on 11552 members pollution had ceased to be an important public ofthe 1958 national birth cohort. Counties health problem. in the UK were ranked by annual average There has been a resurgence of concern level of black smoke and sulphur dioxide about the possible effects of pollution, espe- (SO2), and then divided into five groups. cially that from motor vehicles, which now The subject's county of residence de- produce as much particulate pollution as all termined their categorisation of pollution other sources and account for the majority of exposure. The association between air pol- oxides of nitrogen.2 Using more sophisticated lution exposure and respiratory symptoms statistical techniques for analysing time series was examined by logistic regression, ad- data, a number of studies - mainly from the justing for social class, sex, and smoking. USA but also from Europe - have dem- Results - The ranges of the air pollution onstrated effects on daily mortality, admissions, groups were 2*0-13*0, 13-1-18*7, 19*6-208, symptoms, and lung function at ambient levels http://thorax.bmj.com/ 21-0-25X8, and 26*1-551 igIm' for black below current national and international stand- smoke, and 7-0-36-4, 36-7-42*7, 430-50*5, ards.'0 In the one reported study from the UK 52*0-59*3, and 609-87*7 pg/m3 for SO2. The associations between hospital admissions and overall prevalences of cough, phlegm, smoke and S02 levels were reported within wheezing since age 16, and wheezing in current guideline levels." Concerns about the the past year were 13-3%, 10-3%, 9-4%, health effects of air pollution have also been and 44%/o, respectively. Phlegm symptoms fuelled by concern about an increase in asthma increased with increasing smoke levels and other atopic diseases. on September 24, 2021 by guest. Protected copyright. with evidence of a plateau. Cough and Most evidence relates to short term effects wheeze were not associated with black of air pollution, but there is also evidence from smoke; no symptom was associated with the USA that chronic effects may occur.'2-15 SO2. In the subgroup with wheeze at ages An opportunity to study the chronic effects of 16-23 there was no effect ofsmoke level on air pollution in the UK using a cross sectional phlegm. analysis is provided by respiratory data from Mid Surrey Health Authority, West Park Conclusions - Low ambient levels ofblack the 1958 national birth cohort who were sur- Hospital, Epsom, smoke were associated with decreased pre- veyed in 1981 at the age of 23. Data about the Surrey KTl9 8PB valence ofphlegm symptoms in young ad- potential confounding factors of smoking and J F Scarlett ults in the UK in 1981. The effect was social class are also available. Although average London School of evident below the current EC guideline of levels of smoke and SO2 were higher in 1981 Hygiene and Tropical 34-51 igIm' annual black smoke. In 1991 than at present, the range of exposures at Medicine, London the annual mean WC1E 7HT smoke level for each county level overlaps with the current range J M Griffiths county ranged from 3-4 to 26 5 igIm3, span- and falls within EC guideline levels. Thus, the ning all but the last exposure group used results will be of relevance to the present day Department of Public here. This is consistent with the existence situation. Health Sciences, St George's Hospital of adverse and possibly chronic effects at Medical School, current levels. London SW17 ORE (Thorax 1995;50:764-768) D P Strachan Methods H R Anderson Keywords: phlegm, air pollution, smoke. STUDY SUBJECTS AND HEALTH DATA Reprints will not be The study used information from 11 552 sub- available. jects aged 23 years who were interviewed in Received 2 June 1994 1981 as part of the National Childhood De- Returned to authors 11 August 1994 The London smog of 1952, associated with velopment Survey (NCDS, a cohort of people Revised version received very high levels of sulphur dioxide (SO2) and born between 3 and 9 March 1958). Subjects 27 February 1995 Accepted for publication black smoke, is known to have caused around were asked two questions on asthma/wheezy 11 April 1995 4000 deaths.' This showed convincingly that bronchitis: (1) have you suffered from asthma Air pollution effects on health of 23 year olds 765 or wheezy bronchitis since your 16th birthday? monitors within the county. Counties were then (2) have you suffered from asthma or wheezy ranked by their mean pollution levels and ar- bronchitis in the past 12 months?; two ques- ranged into five exposure groups with as near tions on cough: (1) do you usually cough first equal numbers of subjects in each group as Thorax: first published as 10.1136/thx.50.7.764 on 1 July 1995. Downloaded from thing in the morning in winter? (2) do you possible. usually cough during the day or night during Analyses of the effect of county pollution the winter?; two questions on phlegm: (1) do levels on county sympLom prevalences could you usually bring up phlegm first thing in the not be carried out for several reasons. Firstly, morning in winter? (2) do you usually bring allocating individual pollution measures to each up phlegm during the day or night during subject leads to an analysis ofthe effect ofliving winter?; and three questions on smoking: (1) in each county rather than an analysis of the have you ever smoked a cigarette, a cigar or a effect of exposure to ambient pollution levels. pipe? (2) do you smoke cigarettes at all now- By using exposure groups people from different adays? (3) have you ever smoked cigarettes counties are in the same group, and thus the regularly? effects of potential confounding factors such as Subjects were classed as having cough symp- climate or geographical location are reduced. toms ifthey answered yes to either ofthe cough Secondly, allocating the mean county value as questions and as having phlegm symptoms if a measure of pollution exposure implies a de- they answered yes to either of the phlegm gree ofprecision in the exposure estimate which questions. Smokers were classified as "cur- does not exist. It is more certain that, within rent", "ex"l, "experimental", or "never". Ex- exposure group 1, despite individual variation smokers were those who used to smoke regu- in exposure, individuals have greater exposure larly whilst experimental smokers had smoked to pollution than those within group 2. Thirdly, but not regularly. Subjects were grouped into simple correlations between the prevalences of classes I and II, IIINM, IIIM, IV and V using symptoms by county and the mean pollution the Registrar General's classification of social levels were inappropriate because some coun- class. The most detailed geographical in- ties had few subjects (<10), and thus symptom formation available was subjects' counties of prevalences were either zero or large. Grouping residence. Only subjects with complete data subjects into five groups meant all available on all relevant factors were included in each information could be included in the analysis. analysis which has led to different de- Smoke and SO2 were ranked and grouped nominators for each analysis. separately, hence subjects may be in a different exposure group for each of the two pollutants. A combined smoke and S02 level for the 50 POLLUTION DATA counties with both measurements was cal- http://thorax.bmj.com/ Pollution data were obtained from the Air Pol- culated by ranking counties by their smoke lution Division of Warren Spring Laboratory levels and SO2 levels, and adding the ranks. and consisted ofthe annual mean, median, and The odds ratio ofeach ofthe four respiratory 98th percentile measurements for smoke and symptoms was computed using logistic re- SO2 for all monitoring sites active in 1981. gression models within which the potential con- Black smoke was measured using the British founding factors of smoking, social class, and Standard reflectance method and SO2 using sex were included. In the first instance a test peroxide samplers.2 Monitoring stations were for homogeneity across the pollution groups on September 24, 2021 by guest. Protected copyright. excluded if values were missing for more than was performed, and this was followed by a test 14 daily measurements in the year, or for seven for trend where there was some evidence of a consecutive days. In addition, monitoring sta- trend on inspection. The trend analyses used tions at purely industrial sites were excluded the median value of pollution in each exposure as these may not reflect the air pollution ex- group as a covariate.