The Health Effects Institute
"A Partnership of the U.S. Environmental Protection Agency and Industry"

Joint WHO/HEI Meeting on Particulate Air Pollution
Brussels, Belgium
6 and 7 March 2001

Recap of progress made and outstanding needs

Comments from discussants
Closing commentsI

Robert Maynard, Department of Health, United Kingdom

1. It’s no longer worth arguing about whether the time-series studies represent causal associations or not. This point was not raised before this meeting. I believe there is still room for some doubt, but we have to move forward and it is a blessing that we haven’t got into that argument at this meeting. Of course the time-series studies are model dependent, and as Dr. Nicolich pointed out there are the difficulties of predictive vs. descriptive analysis.
   
2. The HEI reanalysis of the effects of long-term exposure has gone a long way to settling doubts about those studies. As the number of studies developed in the time-series framework have increased, people were beginning to become increasingly sure that they were saying something; but in the cohort studies we have only two - three if you would count the seventh day Adventist study - and a few studies just beginning in Europe. So there was some doubt about those studies which has been laid to rest by the outstanding analysis done by Dr. Krewski et al. The comments on the reanalysis in the back of the report by the original authors Dockery and Pope are generous in the whole, but there is a point of criticism that over-analysis is a danger and may produce results that aren’t real.
   
3. There are different levels of complexity. The French general physiologist Claude Bernard (1850s) stated the following: "When investigating biological phenomena – when you have come to understand a level, you discover another level just below it, which is more complex than the one you’ve just understood. It is important to reach a level of sophistication at which you could usefully interfere with the system." We need to reach a level of understanding the effects of air pollutants on health where we can do something about the effects of air pollutants on health. As somebody said, in a few years we may be worrying whether it was causal because we didn’t know which gene was being affected. To me that is going beyond the level at which we can usefully intervene – we could usefully intervene now.
   
4. Demonstrating whether our policies have had a useful effect on health – is about knowing what the public health impact actually will be. We seem to be lacking a cross-bearing, at least in the short term period. In the short term we rely on the time-series studies. In trying to discover whether we’ve had an effect or not, we use the same studies that were used to predict the effect that we would have had by making a given change in the level of air pollution, so I think it’s circular.
   
5. How should we calculate the effects of reducing levels of particles. There are time-series and cohort studies. There is a sense that cohort studies represent the whole of the picture and time-series studies only a fairly small part of the picture. I’m wondering if we need to do calculations in time-series of the number of deaths at all, or just stick to cohort studies and look at the effects of long term exposure to air pollutants? This would relegate time-series to providing collateral or supporting evidence for cohort studies, but still time-series studies may shed important light on mechanisms. So we can’t ignore them, and they may be instrumental in getting some links across (eg. the study on the Utah steel mill closing). Regarding time-series for morbidity data – which coefficient are used in which countries? There clearly is heterogeneity in the data. How are adjustments made? Should we use meta-analyses? Do a meta-analysis of western Europe, vs. eastern Europe or a combination? Which coefficient is chosen in terms of the lags, 1, 2 or 5 days, or distributed lags?
   
6. Vehicle generated pollution is a toxicologically important source. Should we try to vary composition and size distribution of emissions from vehicles - not just reduce emissions in general but focus on most toxic part? Here, funding from automotive and oil industry is needed as much as from regulators.
   
7. Is it conceivable that in some cities, depending on background concentration of particles, temperature profile for the city, and the wealth of the city, that particles may have an insignificant effect on health? Particles may not be important, in the sense that time-series studies may be capable of demonstrating their importance. If this is true – what implications would it have for the control of air pollutants in cities which fit into that group (if that groups exists)?
   
8. It is often said that the assumption about a threshold is critical to cost-benefit analysis. I’m not sure that this is true. In the UK there has been considerable enthusiasm for working out the whole size of the impact (Anderson presented a Department of Health study which calculated 8,200 deaths from exposure to PM10). Is it helpful to work out what benefits are gained from controlling air pollutants in general? It is helpful to show politicians and the public that air pollution has a large effect on health. We’re really interested though in how much benefit we’ll get from reducing levels of air pollutants by a feasible amount. In our lifetimes we can predict that air pollutants are going to come down by a certain amount (I guess), and that’s the part of the reduction that we should be interested in calculating benefits for. The threshold assumption has importance in the model used to construct the relationship. By putting in an (arbitrary) threshold to lock in the bottom of the model, you must surely affect the slope of the line. I’m worried that dispensing with the idea of a threshold may change the slope of the models we actually use (but I’m not sure it’s true).
   
9. If the new European Union framework focuses on long-term exposure and susceptible groups, then that is good news. Susceptible groups may be a large amount of the population (basically, as Dr Jantunen put it, everybody except non-smoking young men). We know nothing or little about women or the elderly. Some part of that very large group of the population is actually susceptible. In fact, the large group embraces the susceptible individual, but does not define this individual. Defining the susceptible individual remains the real challenge.

---

Comments from discussants

Klea Katsouyanni, University of Athens, Greece

• Sensitive subgroups
• Linking science to policy (APHEIS and others)
• Separating key components of exposure (NO₂ vs. PM) and what does each one stand for
• Incorporating heterogeneity into impact assessment
• Long term effects – more work, especially in Europe, imaginative study designs (not just cohorts)
• European – US collaboration (APHEA/NMMAPS, Boston/London ICD studies
• More dialogue between interested parties (industry, NGOs / consumers)

Bert Brunekreef, University of Utrecht, the Netherlands

• Trying to find a threshold (not very important: if there is one, it’s lower than the levels we’ve been studying; actual benefits would be greater than if the curve would go through the origin)
• Harvesting (NMMAPS and APHEA 2 will give the information)
• Personal exposure in relation to ambient exposure – quite some studies done: link studies to validate GIS – better spatial resolution
• Responsible pollutants – look at source contributions (current, not past)
• Long-term effects: few studies, difficult (use existing cohorts)
• Case-control studies infrequently used: interesting model, combined with GIS approach: develop further to get to answers on long-term air pollution more quickly
• Morbidity endpoints: try to figure out what the long term effects of current and or recent exposures are. Small children, prenatal events, have promise: not just death
• Changes in exposure which have occurred already – difficult to interpret, look at context of other health public health actions
• Heterogeneity issue: use to identify factors that explain the heterogeneity, not just detect but focus on finding out why

Anthony Seaton, University of Aberdeen, United Kingdom

Addressing long-term effects: Struggling try to fit in what he hears with plausible clinicopathological model. Heart and lung are very different organs. The heart has great power of recovery. Constant low doses of particles delivered to the lung are unlikely to do long-term harm. It is only when you add very high levels (eg by smoking cigarettes) when you start to get the damage which can then be affected by additional incremental particle exposure. On the other hand, if the heart is damaged (as it is in most cases) by atheroma occurring in the coronary arteries, it is a completely different issue. Multiple or long-term small exposures – if they are atherogenic, which it is suggested they may be – for instance by raising fibrinogen slightly and therefore changing the distribution of fibrinogen in the whole population – would quite plausibly cause long-term cardiac effects. That may be the explanation, rather than the lack of power that we haven’t seen long-term effects. You have to kill off an awful lot of lung to get death from lung disease. You don’t have to kill off more than one or two heart muscles to get death from heart disease.

Kenneth Donaldson, Napier University, United Kingdom

There weren’t many toxicology presentations at the meeting. The steel mill in Utah is a model study; there may be anomalous cities where the levels of PM suggest there should be an amount of effects but they’re not there – if particles could be collected for use in toxicology studies this may be very helpful. This gap needs to be filled. Regarding the link between heart and lungs: now there are plausible hypotheses. For instance, in UK 2/3 of deaths from cardiovascular causes are due to plaque rupture. Now it has been established that plaque formation and destabilization are inflammatory processes. So there is a clear link between the likelihood for particles to cause inflammation and the likelihood for particles to impact on this very important disease which causes the cardiovascular deaths in most cases. There is clear role for oxidative stress in this condition also. This is another area to be targeted. In terms of magic/silver bullets – in toxicology we should still focus on metals, ultrafines, and endotoxins. The principal hypothesis remains oxidative stress. What’s toxic in PM? Some components are more likely to be more toxic (eg. metals, ultrafines). Turn around the question: what’s more likely to cause inflammation? Which are components that are most likely to cause oxidative stress? Still metals and ultrafines are the main culprits. Recent ultrafine studies are mixed news: composition is still important since some ultrafines are toxic and others aren’t. Also, they form aggregates and may not be counted as ultrafines any more, but they could still act as ultrafines in rat lung.

Karola Taschner, European Environmental Bureau, Brussels, Belgium

Much evidence has been presented at this meeting about the fact that particles are harmful to health, whether ultrafine, fine or coarse. Any of these particles had effects, even if only on susceptible groups. This means that action could be taken, especially since were now finally in a position where the technology is available. The question is now whether to continue and fine-tune where the silver bullets are concerning particulate. It may be the mixture, but all other pollutants are going down (due to regulatory measures) and particles are left to be regulated. Now, the precautionary principle has to be addressed. Should we look for ever more evidence, or take the opportunity, and look at the suffering of asthmatics, small children, and seriously ill people. The European Union needs to turn indicative limit values into mandatory values to meet the quality objectives they have set. Especially for increasing diesel particles due to increased traffic action needs to be taken, on the vehicle side and on the fuel side.

Martine Meyer, Renault, France

Struggled with good, sound science – very difficult to go back and tell industry what they have to reduce. Car manufacturers have already made many efforts in technology, fuel comp, after treatment systems. What to do on composition of complex mixture? Need more advise.

Answer to Ms Meyer from Erich Wichmann – Anthony Seaton – Nino Kunzli

Wichmann: We now know that different types of particles may be relevant. If we are able to reduce all of them, that would be the easiest solution. As we heard at this meeting, technology (filters) is available to reduce particles to a very low level.

Seaton: On the one hand, we are trying to dissect out of PM₁₀ what is important, while on the other hand we talk about setting standards for combinations of pollutants. PM₁₀ is in fact the classical combination of pollutants, and any efforts to reduce PM₁₀ are helpful.

Kunzli: We have struggled at this meeting to find out what the "silver bullets" are. In my opinion, the silver bullet is burning fossil fuels, which adds a story about climate change. We should be more efficient about our strategies and push for a strong move into other energy system, away from fossil fuel and into renewable energy. No emission or low emission cars are needed, which would be beneficial to the purposes of this meeting and of other meetings on climate change as well.

Jonathan Samet, Johns Hopkins University, USA

The National Research Council has been publishing reports on particulate matter research. NRC report 1 is on priorities for particulate matter research and report 2 on the implementation of research in US. Report 1 has defined 10 key uncertainties. Report 3 (recently released) summarizes a lot of research in progress. Report 4 should summarize what we know and what we need to do with it. At present, a tremendous amount of information has been generated and is in the process of being generated, but we are having a hard time pulling it together. How much have we advanced? The real need is for good ways and tools to evaluate the research in hand, before we decide what is needed in terms of further research efforts. Progress has clearly been made compared to about 6 ago, but we need to develop better yardsticks.

---

Closing comments

Lynne Edwards, European Commission DG Research

Further decisions are needed in European Union in 2003 and 2004. Following are 3 areas which are going to be hot political items.

1) The question whether vehicle emissions are important. People have other reasons to be concerned about traffic. When we talk about vehicle emissions, do we mean just combustion emissions? Do resuspended particles from traffic matter? What about other combustion sources? What about natural sources and fugitive emissions from stockpiles? These issues are important because of different implications for the European member states.

2) The question of how much secondary particles matter. This is a transboundary problem. One could take the view that it doesn’t matter – because acidification, eutrofication, & ozone will have a transboundary strategy to reduce all the precursors anyway. In our calculating what every member state needs to do, when we set targets for secondary particulates – we will come to a different answer on burden sharing. You get a different answer if you chase particles than if you chase just acidification, or ozone, or eutrofication.

3) The question of what are the impacts. We’ve been into enough details in this meeting as to exactly what benefits from each particular control measure people are going to get. I do want to say that Health Impact Assessment is very much a developing art, and it gets grossly misused. I’d like to be in a position to be able to misuse it less. Personal view (not commission): I would rather talk about life expectancy instead of numbers of bodies, which would be much less prone to misuse.

Finally, the more things are politically important, and the more uncertainties there are, the more the different parties try to go back to the science and appeal for certainty in the science. Most politicians do not have much science background, and we often don’t know what we’re talking about; so there is great opportunity to misuse the science - as well as economics. I’d like you to be thinking about how you could help us not to misuse the science.

Michal Krzyzanowski, World Health Organization

One of the future responsibilities which WHO will have towards member states will include evaluation of the developing science. We are aware that it is not finished, but it is moving forward and will be useful for advising the countries. The new EU program CAFÉ calls for sound science, and calls for transparency. This meeting here is an example that we are trying to involve sound science. It’s also a demonstration that at this stage there is a great deal of transparency in the discussion. Scientists try to listen to the demands of policy process. Compared to 10 years ago, we are now better prepared for the new revision of guidelines. There are two important points for this discussion: (1) we need to complete the process of revision in a short time (2 years). Many studies are ongoing, which will not be available for the review. A good point is that we have a more systematic approach, in the form of databases such as maintained by HEI and member states. There is more consistency in the organization and design of studies, and there is value in collaboration within regions but also across regions (transatlantic). There is more cross linking where studies are designed to respond to questions raised by other studies. WHO together with the Commission is thinking about how to structure the process, how to make it objective, sound, and transparent. You will be informed and also involved in this process. (2) There is a need for persistence in implementing the science. It would be an empty discussion if we wouldn’t be innovative in looking for new topics and hypotheses, ideas and new ways of implementation. It is up to the science to respond to this, and it is up to the funders (Commission, HEI, European national funding) to increase their efforts.

Back to agenda

---

Home | About HEI | What's New? | Newsletter | HEI International | Publications
Research | Funding | Meetings | Contact HEI | Links | Search | Site Map
Copyright © 2004 Health Effects Institute. Please send comments to webmaster@healtheffects.org