|
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||
|
 |
|
||||||||
|
FOR MEMBERS' USE ONLY COMEAP/2003/1 COMMITTEE ON THE MEDICAL EFFECTS OF AIR POLLUTANTS Minutes of the meeting held on Friday 21st February 2003 in Room 125A, Department of Health, Skipton House, 80 London Road, London SE1 6LH. Present: Chairman: Professor J G Ayres Members: Mr B Armstrong Secretariat: Dr H Walton Assessors: Dr M Meadows (DEFRA) In Attendance: Dr Ian Mudway (King's College, London - Item 4.1D)
1. Apologies for absence were received from Professor Ross Anderson, Professor Philip Poole-Wilson, Professor Anthony Frew, Dr Virginia Murray, Dr Paul Harrison (Assessor), Mr Martin Williams (Assessor), Dr Bob Maynard (Secretariat), Ms Julia Cumberlidge (Secretariat). 2. The Chairman announced that Professor Stephen Holgate had been made Chairman of EPAQS and informed Members that he proposed that the Chairman of EPAQS should automatically have a position on COMEAP. 3. The Chairman welcomed Dr Ian Mudway from King's College, London who had come to present his and Professor Frank Kelly's paper (COMEAP/2003/1D). 4. The Chairman also welcomed Ms Shahneela Baig who works in the Air Pollution Unit at the Department of Health and who was attending as an observer. 5. The Chairman reminded Members to declare any interests they may have at the beginning of each agenda item. 6. The Chairman reminded Members to submit their expenses claims forms. Blank forms had been tabled.
7. Changes to the minutes were requested on page 6 paragraph 34, first sentence. The Committee felt that the construction of the sentence could infer that the Committee did not wish to pursue quantification of changes in lung function. 8. The Committee decided that to alleviate any potential misunderstanding over what had been agreed at the meeting in November, the sentence should be re-written to show that mean changes in lung function were quantifiable, although difficult to interpret. 9. It was agreed that formal wording to replace the first sentence would be decided through correspondence between the Chairman and the Secretariat. [Action: JA / Secretariat] 10 [The following wording was agreed: The Committee agreed that, although there was sufficient evidence to quantify an effect of ozone on lung function, the clinical significance of the quantified effects would be difficult to interpret.]
Annual Reports 10. The Chairman informed members that the Annual Report for 2001 could be found on the COMEAP website and that the Annual Report for 1999/2000 would be on the website shortly. Indoor Air 11. The Chairman informed members that this would be discussed at the
next meeting of COMEAP. 12. The Chairman explained that the need for an extra meeting of COMEAP in 2003 had arisen from discussion at the recent Cardiovascular sub-group meeting. It had been decided that the delays in the production of this sub-group's report (due to the urgent nature of the asthma sub-group report and the Secretariat's involvement with work on anti-terrorism) had necessitated this extra meeting. 13. The Committee accepted the need for the extra meeting and agreed on Friday 19th September as the date for the extra meeting and agreed to postponing the 24th October meeting until 14th November. 14. The Secretariat would inform members who were not present at this meeting of the new meeting dates. [Action: Secretariat] Annual Air Pollution Meeting 1st-2nd April 2003 - IEH, Leicester 15. The Chairman reminded members that registration forms were to be returned to the organisers at IEH by the 7th March 2003 and that expenses for the meeting (including travel, accommodation and attendance fees) should be claimed. Claim forms would be provided within members' welcome packs. Appointments 16. The Chairman notified members that an advertisement would soon be published for the appointment of new members for COMEAP, with interviews ideally taking place in May. Website COMEAP/2003/8 17. Members studied paper COMEAP/2003/8 which detailed the number of hits the COMEAP website had received. The tables showed the 50 most popular COMEAP webpages accessed/visited during the period 1st October 2002 - 31st January 2003. The tables were broken down to show how many different computers had accessed the pages and also how many times the same computer had accessed the pages on a number of occasions. 18. The Chairman drew members' attention to the fact that in four months, the website had been accessed by over 5,000 different computers and in total there had been over 18,000 hits. The Asthma and Air Pollution Report of 1995 was the 14th most visited page and the second most visited report (although it is thought that the most visited report may have been such because of its positioning on the webpage). 19. It was suggested that the Secretariat update members on a yearly basis, and compare the results with other relevant Advisory Committee websites where possible. [Action: Secretariat] Reports 20. The Chairman informed the Committee that the report '50 years on - the struggle for air quality in London since the great smog of December 1952' would be sent to members shortly.
GAMS Update 22. The US EPA had requested the re-analysis of 20 studies which used the program 'S plus' and this was nearing completion. Publication of the results with comments from the panel would be provided in a HEI report due in April 2003. 23. The Committee was told that, in general, reanalysis of the other studies had resulted in smaller changes in the coefficient relating particles and mortality than was found in the NMMAPS study. 24. There were two issues with the standard software. Firstly, a correction was needed to the convergence criteria for fitting medium to long-term variations in mortality. Correction for this was straightforward. Secondly, the standard software typically underestimated the standard error. Correction for this was less straightforward, although a 'direct fix' had recently become available. The standard error problem could be avoided by using macrocubic splines instead. 25. Dr Armstrong informed the Committee that one question which had arisen out of discussion was how to choose how much smoothing should be carried out and how much to control for temperature. 26. Dr Armstrong also informed the Committee that studies looking at Particulate Matter had been reanalysed. If these studies included ozone as well, some re-analysed coefficients for ozone might become available. 27. The Committee were informed that the confidence intervals of the mean across cities did not change that much with the re-analysis and between city variation changed less than within city variation.
28. The Committee was informed that a separate document containing only the plots which were embedded within paper COMEAP/2003/1A had been tabled for ease of use. 29. Dr Walton informed Members that the studies analysed in the production of this paper had been taken at face value and did not include any changes in light of GAM re-analysis. 30. Dr Walton reminded Members that at the June 2002 meeting of COMEAP, the Committee had considered the time-series study evidence on mortality and hospital admissions and 8-hour ozone. The paper they were now considering, provided the evidence from studies examining associations with 1-hour and 24-hour average ozone concentrations, which the committee had requested in an attempt to ascertain whether the studies of 1 hour or 24-hour average ozone concentrations gave general support to the conclusions based on 8 hour average ozone concentrations. (i) What is the evidence across all averaging times for associations with particular health outcomes? All cause mortality 31. It was agreed that there was clear evidence for an association of ozone with all cause mortality. However, this did not necessarily mean that there was a causative association. 32. Several members raised questions about causality that needed discussion. These included: - whether the associations had been adequately adjusted for the effect
of weather on all cause mortality 33. It was acknowledged that it was difficult, when using death data, to find a group of subjects who wouldn't be affected by respiratory compromise, so it was conceivable that ozone could affect other causes of death in addition to respiratory mortality. On the other hand, if the people affected were already particularly ill, it was questioned whether they would actually be outside and therefore exposed to a significant level of ozone. 34. It was proposed that there should be a workshop to discuss these issues and to allow the Committee to come to a firm view over causality. However, other members considered that the Committee should just accept that it was the shape of associations that was being discussed and that causality should be considered at another time. Discussion proceeded in terms of associations. The Chairman proposed coming back to the issue of causality later (see paragraph 42). 35. It was agreed that the most appropriate averaging time was 8 hours. For interpreting the meaning of studies using different averaging times the actual correlation with 8 hour averages needed to be known for each dataset. For 1 hour averaging times it would be expected that the correlation would be high. However, for 24 hour averages the correlation was less close because high ozone in the daytime can be preceded by particularly low ozone levels overnight. It was agreed to give the 24 hour average studies less weight but not to ignore them completely. 36. Members also discussed how much weight to put on locally relevant studies such as that in Edinburgh (where no association was found) or on a broader group of studies. However, the study in Edinburgh was small and it was unlikely that the large number of positive associations which occurred elsewhere had occurred by chance. 37. The Committee noted that NMMAPS results on ozone had not yet been included and agreed that they would be useful, as would other results that had been reanalysed with corrected GAM software. Respiratory Mortality 38. The Committee agreed that ozone was associated with respiratory mortality. Cardiovascular Mortality 39. It was agreed that Figure 17 (1 hour average) showed evidence of an association, although the studies were not entirely independent of each other (data from some studies were partially included in others). 40. When questioned on publication bias, the Secretariat confirmed that there was no evidence of publication bias in the studies on 8hr average ozone. 41. The Committee agreed that, considering the 1hr and 8hr data together, an association between cardiovascular mortality and ozone was a little more likely than on the basis of the 8 hour average data alone. Respiratory Admissions 42. It was agreed that Figures 21 and 28 showed a clear association with the 1hr and 24 hr data, as nearly all results were statistically significant. 43. The Committee therefore concluded that there was sufficient evidence to show an association between respiratory admissions and ozone. Asthma Admissions 44. The Committee did not believe that there was no evidence for an association
with asthma admissions but rather that the evidence was unconvincing.
There was only modest evidence of a moderate effect of ozone on asthma
admissions. 45. The Committee did not agree that the 24hour data provided good evidence for an association between ozone and COPD admissions. Although the associations were positive and statistically significant, the associations were small (the y axis in Figure 28 was on a wider scale than for some of the other figures). However, members were reminded that the results from the APHEA I study using 1 hour average ozone concentration and the 8 hour average results did provide evidence to support an association between ozone and COPD admissions. 46. The Committee agreed overall, that there was data to support COPD admissions being associated with ozone. Pneumonia Admissions 47. Members noted that there was clear evidence, at least at high concentrations in rats, that ozone diminished microbiocidal activity of macrophages, which made it plausible that ozone affected pneumonia admissions. On the other hand, associations at lag 0 were regarded as implausible as there would not be time for an infection to take hold. 48. The Committee agreed that the data did support an association of ozone with pneumonia admissions. Cardiovascular Admissions 49. The Committee agreed that the data for cardiovascular admissions was unconvincing and did not show an association with ozone. Ischaemic Heart Disease Admissions 50. Members did not believe that there was convincing evidence of an association. Heart Failure Admissions 51. Members did not believe that there was convincing evidence of an association. 52. Members noted that the fact that ozone was not clearly associated with cardio-vascular admissions suggested that the other findings could not simply be due to confounding by temperature as increased temperatures were associated with both cardiovascular and respiratory effects. (ii) Seasonal Differences 53. The Committee discussed various possible reasons for the seasonal differences. It was considered that the way adjustments were made for temperature / weather may be more important than acknowledged within the paper. It was agreed that it would be useful to have more multipollutant model studies when looking at seasonal differences. 54. This led to a more general discussion about multi-pollutant models. It was noted that particle numbers were not mentioned or studied in all of the data analysed and it was thought that particle numbers should be included in multipollutant models. However, this point was argued, as particle numbers are not necessarily correlated with ozone. On the other hand, high secondary particles do track with ozone. Ozone may also be negatively correlated with nitrogen dioxide in urban areas. It was suggested that studies in rural areas where ozone is higher and less subject to confounding by urban pollutants, might be helpful. However, studies in rural populations would have to cover a larger area in order to produce data from a large enough population. It was also noted that people from rural areas might be admitted to hospital in urban areas, especially if seriously ill. (iii) Thresholds 55. The Committee were reminded that they were being asked to consider whether the rise in ozone at low concentrations in urban areas was of concern. For quantification of the health effects it was important to know whether there was a linear response between daily changes in ozone concentrations and daily changes in health outcomes, as this would affect the way the calculations were done. 56. The point was made that thresholds must be specific to certain effects and within the studies presented, there were endpoints where ozone was shown not to have an effect. 57. The relationship between personal vulnerability and population thresholds was discussed. It was noted that people would be interested in what their own vulnerability was and a population study would be irrelevant to them. It was noted that adaptation could mean people had different thresholds at different times. There were biological reasons why there might be thresholds. However, it was considered difficult to find a threshold in population based data. 58. It was considered that, rather than use the word threshold, it should be discussed whether there was a linear response within the ozone concentration range of interest i.e. the range across which ozone concentrations had been rising in urban areas of the UK. 59. The point was made that the study of all cause mortality in Rotterdam found a near-linear increase (Figure 11) and the slope only decreased slightly until all days above 15 ppb 24 hour average had been removed. Another study of all cause mortality found a continuous trend across quintiles (Table 12). Those studies that did show non-linear associations had problems such as a lack of control for confounders. However, there was not enough evidence overall on a linear response. Overall Summary 60. It was agreed that where associations had been found, the evidence against a linear response was not convincing. However, the evidence for a linear response was not sufficiently strong to rule out other shapes for the association. (iv) Scope of report and causality 61. The Committee considered how the information on ozone considered by members should be taken forward. The Chairman provided 3 options: 1. Produce a report and accept causation, but do not discuss it, rather
examine the evidence for a threshold when looking at associations with
ozone. 62. Several members did not want to put out a report without something on causality. On the other hand, members did not want to pretend that there had been a thorough discussion of causality when there had not. The Secretariat considered that it would be difficult to make practical use of the Committee's views on the linearity of the response if doubts had been raised about causality but the issue had not been resolved one way or the other. 63. The Chairman suggested that a distillation of all the working papers was needed in any case. There was some further information to come - the reanalysis of the studies which used the 'S plus' programme due from HEI in April and there was information from the APHEA II study on a dose response for ozone which was yet to be published. It was noted that there was also the symptom data to be analysed. In addition, it was considered that a one page statement for the Committee's own use on thresholds and linearity would be helpful. [Action: Secretariat]
64. Dr Walton introduced this paper, explaining that it suggested it was appropriate to use a non-linear model to control for temperature and that ozone associations were greater in studies which use non-linear models for temperature. 65. Members considered that the way studies were controlled for temperature should depend on the shape of the relationship between temperature and mortality in the location concerned. It was noted that non-linear models for temperature were already usual practice in more recent studies. What was more of an issue was what lag was appropriate - for effects of cold, longer lags than those usually used may be important. This was however, less relevant for ozone. It was also noted that temperature might be an effect modifier as well as a potential confounder. 66. Members were less convinced about the criteria the authors used for
model selection. It was difficult to follow which factors were visible
as confounders and which were not. The main criterion for defining confounders
should be a judgement on whether the factors were likely to be causal.
It was also noted that, although both approaches were legitimate, it was
more usual now to put all variables in and allow them to 'compete' for
the effect rather than to use two stages (all other variables and then
pollutants). 68. In conclusion, the Committee agreed that it was important to use non-linear models for the effect of temperature on health outcomes, although the Committee were not convinced of the criteria proposed by the authors for overall model selection.
69. Dr Walton introduced this paper and reminded the Committee that they had at the last meeting discussed plots of coefficients for 8-hour average ozone concentrations and all cause mortality against mean or maximum ozone concentrations. From these, they had concluded that there was an absence of evidence for a threshold based on data presented and it had been suggested that a test for the statistical significance of trend be carried out and that a similar analysis be performed for other outcomes. However, simulations of the expected results of this approach, performed by St Georges', suggested that there were difficulties of interpretation if there was a threshold. 70. The Committee agreed that the trend tests supported the earlier conclusion of the Committee - that there is an absence of evidence for a threshold in the data on 8 hour average ozone concentrations and all cause mortality. 71. The Committee also agreed that the approach had a limited capacity to give information on the numerical value of the threshold, although it could give an indication whether there was a threshold or not. 72. The simulations suggested that the study coefficients were influenced by the range around the mean (where thresholds exist) and where the threshold falls in relation to this. The Committee noted that if the coefficient was statistically significant, it was likely that a reasonable proportion of the data was above the threshold. 73. It was agreed not to put too much emphasis on these plots although they could be useful in certain circumstances.
74. Dr Mudway introduced this paper which considered whether there was a threshold for ozone-induced inflammation . It was explained to members that this paper was commissioned by the Secretariat from Professor Frank Kelly's group at King's, College London. Members were also informed that the paper was not in its final draft stage, but was being presented for comment on the findings, which would be incorporated into the review prior to publication. Members were also invited to consider the implications from the epidemiological findings. 75. Dr Mudway informed Members that the analysis was based on the comparison of the percentage of neutrophils recovered by bronchoscopy based lavage in response to a set dose of ozone. 76. The Devlin (1996) study was considered the most influential of the 21 peer reviewed studies considered. It was also acknowledged that the amount of exercise which took place varied considerably between the studies. This had been taken into account by describing the dose of ozone as the product of concentration, ventilation rate and duration of exposure (CVT)i. The data suggested a threshold for ozone-induced inflammation at a CVT of 500-620i . 77. Members queried the method of determining the point at which there was no effect. Taking the point where the line crossed the 95% confidence interval of the control group was not quite right as this was the 95% confidence interval of the mean rather than the confidence interval around individual values.
79. The differences between the left hand and right hand graphs in the paper were discussed. Analysing the percentage increase in neutrophils per unit dose was not considered as helpful as it did not provide an absolute level. The right hand graph helped to spot outliers. 80. The Committee agreed that neutrophila of the airways was not the best endpoint as it could be regarded as a normal response. An inflammatory response was not necessarily related to symptoms. Albumin in lavage fluid might be a better endpoint as it indicated leakage across the epithelium. Dr Mudway noted that it was harder to compare studies using this endpoint due to dilutional issues. 81. It was also noted that studies of asthmatics were not included. 82. Ozone exposure at the air quality standard of 50 ppb averaged over 8 hours did result in inflammation at a ventilation rate of 20-30L/minute/m2. However, it was unlikely that this ventilation rate would be maintained for a full 8 hours. Dr Mudway commented that, in terms of the neutrophil response in the lung, the air quality standard was more than adequate for healthy subjects provided that they were not performing heavy exercise for long durations outdoors. This suggested that, for healthy subjects, emphasis on advice to limit exercise outdoors during high ozone days could be more important than reductions in concentrations.
Public Availability: It was further noted that paragraphs 84 and 85 should not be made publicly available at the present time since these paragraphs referred to work that was currently in progress and not in the public domain. 86. Subject to these changes, the Committee agreed the statement. 87. The Chairman noted that a broad discussion document on how to deal with these types of issues would be helpful. It could be very time consuming for COMEAP to deal with each specific circumstance separately rather than giving generic advice on the pollutants which could then be applied to different situations.
88. The Chairman provided Members with an update, informing them that there had been a meeting of the CV sub-group at the end of January 2003. The focus of group discussion had addressed the results of the meta-analysis data produced by Ross Anderson and his team at St Georges' Hospital. 89. The Chairman informed Members that the group had decided to 'play in' the cardiovascular fraternity by organising a workshop with key players. This would take place before the September meeting of COMEAP. 90. Quantification would not be carried out in this report due to the start of a second quantification report next year. 91. Submissions for the CV report were requested by the end of March 2003 and it was agreed that the first draft of the report would be brought to COMEAP for the September meeting. Members of the CV sub-group who were due to retire from COMEAP after the June meeting, would be invited back to attend the September meeting when the report would be discussed.
92. The report was provided to give Members an idea of current thoughts on whether air pollution caused asthma and the Chairman explained that the Discussion Notes within the report had been left in to invite ideas and comment. 93. Members agreed that the report was good and made only the following comments: · The influence of genetics on predisposition, especially in children was missing from the text on page 4. · It would be useful at the beginning of every chapter to have a summary (as was done with the EPAQS reports). · Within the introduction, it was important that it was made clear that air pollution had decreased and the incidence of asthma increased. · Urban and rural differences should be introduced earlier. · Changes in trends should also be considered. However, it was thought unnecessary to go into great depth. 94. A more mature draft of the report would be brought to COMEAP again at the June meeting.
95. The Chairman handed over to Martin Meadows of DEFRA, who provided an update. 96. The four key messages were as follows: · The last National Air Quality Strategy was published in January 2000. · The 100 Local Air Quality Management Areas were mainly designated due to NO2 with a few due to particles, (which predominantly related to traffic emissions). · The scope of the review of the strategy had not been decided, but it was likely that a review of specific measures to reduce air pollution would be included. The objectives for each pollutant would not automatically be reviewed but could be if necessary. · Paragraph 6&7 referred to tighter objectives for particles and benzene, which had been agreed in 2002. There were different particle objectives for London, the rest of England and Wales and Scotland. 97. The first meeting of EPAQS would be on the 20th March and would provide advice on the releases of industrial pollutants.
98. Clean Air Act for Europe - priorities were particulate matter, ground level ozone and ecological effects. 99. The limit values for particles, sulphur dioxide, nitrogen dioxide and lead were being reviewed, taking into account how member states were progressing in achieving them and WHO's views on any new health evidence.
100. It was decided that this would be discussed at the next meeting as by then Members would be in possession of the final report.
111. This paper was for interest only and any comments would be taken at the next meeting.
112. The Secretariat informed members that a meeting was to be held on the 24th February re. the COMEAP five year review. Members were told that as COMEAP was an advisory rather than an executive non-departmental public body the 5 year review would take the form of a 'light touch' review.
113. It was confirmed that the next meeting of COMEAP would be held on the 20th June 2003 at Skipton House. Secretariat
|
|
|
|
||||||
You
are here: 
