Those present:
Chairman: Professor J G Ayres ACP
Members: Mr P Adamson (HSE CRD), Prof N Bateman (NPIS), Dr C Clutterbuck, Dr J Cocker ACP, Mr R Davis (CRD HSE), Mr P Hamey (CRD HSE), Prof G Hawksworth (ACP), Dr J Graves (DH), Mr N Mole (PAN UK), Dr A Povey (ACP), Prof C Ockleford (ACP).
Secretariat: Jayne Wilder (HSE CRD), Mandy Walsh (HSE CRD).
Invited speakers Mr D Garthwaite (FERA), Mr S Langton (Defra), Dr M Toledo (Imperial College), Mr T Fanshawe (Lancaster University)
1. Apologies for absence
1.1 Apologies were received from: Mr J Battershill (HPA), Dr S Fairhurst (HSE CRD), Dr D Ray (ACP), Dr H Rees (ACP), Prof J Parry (ACP), Dr D Sen (PIAP), Dr S Waring ACP.
1.2 Two people had applied to attend the meeting as observers, but sent their apologies as they were unable to attend.
2. Preliminaries
2.1 Members introduced themselves outlining any areas of their work of particular relevance to the work of this group.
3. Minutes of the 1st meeting [PAHES 2 (2/2010)]
3.1 The minutes were agreed subject to correction of various typographical errors.
4. Matters arising
4.1 With regard to the information in paragraph 5.14, the Secretary clarified that no decision had yet been taken about the introduction of the ‘farm spray bureau’ and this was one aspect on which Defra were seeking views by means of the current public consultation.
4.2 a) Progress with draft report [ PAHES 3 & 4 (2/2010)]
4.2.1 Members noted that the tables presented in PAHES 3 (2/2010) considered the exposure measures that were currently available rather than those that could potentially be available. For example at present there is rarely an accurate biological assessment of exposure available although the possibility of taking samples could be considered. Members noted that biological sampling was more relevant to the planned 3rd meeting topic. One member promised to send some suggested amendments by email.
4.2.2 Members discussed the ‘chronology’ presented in PAHES 4 (2/2010) and suggested that it be made clear in the document which initiatives had been tried and then abandoned. Members heard that it had been suggested that the chronology be extended to include the foundation of PIAP, and agreed that some brief information would be useful to add, but detailed investigation of the records would not be necessary. Members suggested that the interdepartmental work on organo-phosphates (OPs) should also be included in the chronology because of the similarity of public concerns about OP sheep dips and medicines. The document should clarify which other departments were responsible for the relevant regulatory regimes for chemicals other than pesticides.
Action: Secretary
4.2.3 One member recalled a potentially useful paper on OPs by G Dunn that had been commissioned by this interdepartmental group and agreed to forward a copy to the secretariat.
4.2.4 Members recalled that one of the recommendations of the Royal Commission Report had been to improve training in toxicology for GPs, noting that there were likely to be several medical schools that would be very pleased to provide such continuing professional development (CPD) provision. The Chairman commented that medical schools often stated that there were many calls for additions to medical student training and that the same often applies to post-graduate training. It might therefore be the case that there was actually limited demand by GPs for this CPD in comparison to other areas of medical training.
4.3 b) Additional background papers produced [PAHES 6, 7, 8, 9 and 11 (2/2010)]
4.3.1 Members observed that PAHES 11 (2/2010) explained the process followed once a complaint has been made to PIAP. This paper also explains some of the difficulties in obtaining exposure data.
4.3.2 PAHES 6 (2/2010) was a report produced by Risk and Policy Analysts Ltd seeking to determine the usage of pesticides in the amenity sector. The report had previously been considered by the amenity action plan group. Members heard that the report provides good information on the attitudes of local authorities. It was thought that information on pesticide supply had been taken from the supply industry and then usage had been apportioned to various sectors based on the responses from local authorities. It was unclear whether there had been some ‘double counting’ arising from separate responses being provided by contractors. It was also known from previous Pesticide Usage Survey Group (PUSG) studies that the amenity sector is difficult to cover adequately and that information on ‘supply’ might be misleading due to some transfer between agricultural/horticultural sectors and the amenity sector. In addition PUSG experience had been that a much better response was obtainable by approaching each area of this sector separately when double counting could be excluded by tying in contractor applications with the specific applications made. Large parts of the amenity sector were missing from this report, and the low response rate was suggestive that the data available might be skewed towards better practice. Despite these difficulties, members heard that it had been possible for the amenity action plan to draw out some key themes.
4.3.3 Members agreed that it was difficult to draw reliable information on potential exposure from this report. The amenity sector is clearly an important area as it covers a wide range of public places (parks, open spaces, public attractions etc) where large numbers of people potentially come into close contact with pesticides. It was thought that at present there were no legal requirements to keep records in this sector although there was evidence of good record keeping practice in some areas such as golf courses and by contractors. The Sustainable Use directive will require wider record keeping by those using plant protection products. It was noted that this requirement would fall to the pesticide user and not necessarily to the authority or other land owner. It was important that the type of information required should be sufficient to ‘validate’ exposures if an incident were reported. Examples such as date, place and time of the application made and rate of application of the specific products used would help with this. Members noted that these suggestions should be incorporated into the ACP draft response to the current consultation on Sustainable Use.
Action: Secretary
4.3.4 PAHES 9 (2/2010) provided information on the confusing plethora of product names that had been requested by the ACP for consideration by this group. The Chairman commended this report. Members expressed strong concern about the potential for confusion by products carrying very similar names for formulations containing different active substances (or indeed co-formulants that might be of as much concern in poisoning incidents as the active substances). Whilst each approved formulation carries a unique MAPP number, this information would not generally be available if an incident was reported by someone other than the user. Members discussed a number of possibilities including legislation to require specific names, and concluded that it is probably an issue to be discussed by the industry to try to resolve the problem. Members noted an opportunity to simplify the naming of products was provided by the re-registration process currently underway and re-iterated their view that it was not acceptable for different formulations to carry such similar names.
Action: CRD
4.3.5 PAHES 7 (2/2010) was a WHO document ‘Guidelines on Developing a Reporting System for Health and Environmental Incidents Resulting from Exposure to Pesticides’. Members noted this recent publication was mainly aimed at the problems experienced in the developing world. As such members concluded that some of the detail might need refining for UK use (such as the definition of ‘case’ which has a specific use in epidemiology that differed from that proposed in the WHO document). Members agreed that a ‘case’ or individual person could have experienced multiple ‘exposure instances’. Members agreed that, with some amendments, the goals and objectives outlined at section 1.5 of the report could be accepted as the aims of the ‘gold standard’ model they were seeking to develop.
4.3.6 Amended extract of 1.5 from the WHO document
The primary goal and objectives of a pesticide incident reporting system
The primary goal of a pesticide incident reporting system is to alert the health and environmental authorities about the risks that pesticides may pose to human health and the environment under certain conditions and to inform the national pesticide regulatory authority about the possible need for risk mitigation measures.
Public health objectives
- reduce the incidence of pesticide related illness, injury, or adverse health and environmental effects;
- identify clusters/outbreaks of pesticide-related illness, injury, or adverse health and environmental effects;
- identify high-risk populations;
- protect relevant populations (including women, children, and unborn children).
- provide better information to the public to address their concerns
Pesticide management objectives
- identify problems and research needs;(not a primary aim but an inevitable consequence)
- identify high-risk pesticide products as well as high risk use practices;
- evaluate the effectiveness of prevention and regulatory efforts;
- collect data and information to support the development and implementation of management and control measures;
- minimize availability of spurious, outdated, deteriorated or illegal pesticide products.
5. Outline of work of the Small Area Health Statistics Unit Imperial College
5.1 Dr Toledano gave a short presentation of the work of the Small Area Health Statistics Unit (SAHSU) within the DH/MRC centre for Environmental Health. Using some examples of the work of the Unit, Dr Toledano explained that they had developed a small area spatio-temporal methodology using a hierarchical Bayesian framework. She explained that in areas of sparse information standard approaches would result in uncertain or unstable estimates of potential impact of eg point source pollution. The small area statistical methodology essentially enabled the analysis to ‘borrow’ information from neighbouring areas with stronger data. This methodology enabled increased certainty for those areas of sparse data.
5.2 The SAHSU offered a ‘rapid response’ service enabling investigation of possible ‘clusters’ of disease or disease mapping. The rapid response service draws upon the existing databases such as the post coded or ‘address-level’ health data for GB and various socio-economic databases derived eg from the National census. Dr Toledano noted that investigation of clusters is particularly difficult to do reactively, and SAHSU therefore prefer to do proactive ‘a priori’ testing of hypotheses.
5.3 Dr Toledano commented that one of the difficulties faced by SAHSU is that in small areas there is a smaller population and therefore apparent ‘higher risk’ areas will occur as a result of random ‘noise’ in the data. The hierarchical methods they have developed can be used to stabilise the data and essentially derive a type of ‘weighted average’ to inform the risk estimate. The methodology is also suitable to consider seasonal effects.
5.4 Members asked whether SAHSU had undertaken any work on pesticides. Dr Toledano confirmed that as yet they were at early stages. So far SAHSU had modelled possible exposures to groups of pesticides drawing upon the PUSG data and the June agricultural returns. At present this modelling is still at early stages and there had not yet been any attempt to put it together with health information.
5.5 Members asked how account was taken of population movement which might affect exposure to a point source pollutant, either on a daily basis e.g. as people travelled to work, or over the course of a lifetime. Dr Toledano explained that this was dependent on the type of study. For example in considering birth outcomes it was less likely that migration would be included over the critical period (1st trimester for birth defects 3rd trimester for low birth weight). However in considering birth cohorts account is taken of migration using for example information on employment and whether it is outside the home ‘zone’. Although SAHSU does not have individual data it is assumed that there is quite an even ‘spread’ at a population scale.
5.6 Members asked whether SAHSU planned a longitudinal study on low birth weight. Dr Toledano confirmed this was planned using the Bradford birth cohort. There would also be ‘in life-course’ work.
5.7 Dr Toledano agreed to email a copy of her presentation to accompany these minutes
6. Pesticide Usage Survey Statistics
6.1 Mr Garthwaite made a short presentation [link] outlining the programme of pesticide usage surveys undertaken by the Pesticides Usage Survey Group (PUSG). He explained that the data were collected in accordance with a stratified random sample taken from the Defra June Census to give representative results. Trained surveyors make around 800 visits to individual farms each winter and collect information on their pesticide use in the course of a personal interview lasting about an hour. Although participation in the survey is voluntary, PUSG usually find about 90% uptake from the arable sample and about 95% uptake from the various horticultural surveys. About 5% total arable area grown is sampled on alternate years and about 20% of the area grown is sampled for horticultural crops on a 4 year rolling programme.
6.2 Data on products used, rates, dates, crop growth stage, reasons for use, tank mixing and volumes of water used are collected on individual fields (or groups of fields treated identically) on the holding. The information obtained is anonymised then raised to give an estimate of regional and National use
6.3 Mr Langton explained something of the wider context of agricultural data. If the PUSG data are applied to the statistics on the crop grown there is a fairly consistent pattern of use within the same crop. Information on where crops are grown is often drawn from the June survey, but those data are gathered on a holding basis. Whilst for agricultural holdings about 60% of fields are within about 1Km of the holding address, some of the remaining 40% can be located considerable distances from the holding location. Defra uses these data and produces a local area statistic which is adequate at a Government Office/Regional level. EDINA (based in Edinburgh) uses Parish based data but again there is the problem that not all of the holding will necessarily be within the same Parish. In terms of more accurate local information he suggested that the ‘best data’ were probably those held in the ‘Support’ databases. From 1993-2004 these databases provide field level data on where specific crops were grown. These data accounted for rotation, but did not specify whether crops were winter or spring, which could affect the pesticide programmes used.
6.4 The horticultural sector was not subsidised and thus there were no data held in the support databases. However the June data were likely to be more reliable for these holdings as they were generally less scattered. Support data belonged to the Rural Payments Agency, but Defra held copies. It was likely these would require some further processing before they could be released outside Defra.
6.5 CRD commented that they had used the SEISMIC database where cropping could be associated with particular soil types, but Mr Langton considered that the data held in the support databases would be more useful if access could be arranged.
6.6 Members noted that the PUSG trends data suggested a large reduction in the volume of pesticide used, but overall usage had increased. Mr Garthwaite explained that a number of factors underlay these changes. In the case of herbicides for example, much of the reduction in volume related to a change from active substances applied at high doses to compounds like the sulphonyl-ureas that are very active and hence only required at low doses of grammes/Ha. For fungicides there had been a trend for farmers to use reduced rates but mix with other products also at reduced rates.
6.7 The Chairman noted that the pesticide usage data presented large changes over time. He asked if there were specific causes for this. It was clarified that weather conditions had a significant impact on pesticide use. With this in mind the Chairman confirmed that data on weather was also likely to be required year on year.
6.8 Members asked whether there had been any attempts to link the usage data to sales data. Mr Garthwaite responded that although he had not had any personal experience of this he understood that it had been attempted in the past without a lot of success. He understood that it had been difficult to obtain sales data, and that the difficulties in comparing the sales and usage statistics were related to lack of information about which sector the product had been used in. A tendency to purchase products whilst prices were low and to store them in case of future need also caused problems in estimating usage from sales data. The surveyors had noted a reduction in such stockpiling recently, speculated to have been a result of the increased rate of changing approval status following the EU review programme.
6.9 The Chairman asked that Dr Toledano and Mr Langton discuss whether it would be feasible to obtain the support data for a quick ‘look’ to see whether they might be useful in modelling local exposure for rural residents.
7. Combining Health Information, Computing and Statistics (CHICAS) at Lancaster University
7.1 Dr Fanshawe provided a short explanation [link] of the work undertaken at CHICAS considering both environmental and spatial epidemiology. They had expertise in using spatial and spatio-temporal models for example in considering impacts of air pollution. They had used these techniques to model the relationship between exposures and health outcomes. They also had experience of mapping these relationships, and had considered long term health effects of both long and short term exposures.
7.2 Dr Fanshawe pointed out that an important aspect to bear in mind from these techniques is that the level of a pollutant in the air is not necessarily related to the amount that gets into the body. The Chairman commented that individual information was required to address this aspect.
7.3 In response to a question, Dr Fanshawe confirmed that CHICAS work tends to be project specific and reliant upon data specifically collected for the purpose. They did not have routine access to the national health databases.
8. Discussion of statistical possibilities and limitations to improving information on general exposure to pesticides.
8.1 The Chairman opened a more general discussion. He suggested the group needed to consider the ‘utopian ideal’ in terms of exposure information and then to consider perhaps how it might be possible to approach this. He felt that there was a need for good background usage data without discontinuity and ensuring methods remained constant over time. This would ensure that data on trends were useful. Personal measures of exposure to all known pesticides would provide the ‘Utopian’ ideal exposure information. He accepted that this was not possible; both in terms of the inconvenience it would cause individuals and the extremely high cost of such an approach. He therefore asked whether it was possible to use a ‘co-efficient’ of exposure derived for example from ‘workspace’ information. In terms of local information, the presentations during this meeting suggested that information from the PUSG surveys together with cropping data from either the June surveys or the Rural Payments Agency database might be suitable to provide an estimate of local pesticide use. This could then be mapped against local populations to derive a measure of exposure. In terms of a hierarchy of usefulness, such local information was more use than regional information and National information.
8.2 Members commented that this approach provided an estimate of background exposure. However for acute exposures it might be possible (if ethical agreement were given) to obtain information on individual exposures. This would require adequate analytical methodology and biological samples to be obtained quickly. Such samples would perhaps enable the development of a GCMS library to help with qualitative identification of some pesticides and enable a database of biological monitoring results to be developed. It was noted that most pesticides used today are metabolised fairly rapidly and did not tend to bioaccumulate. Whilst there is good information on the routes and rates on metabolism of pesticides in rodents, there was little information in humans. It was difficult to calculate human exposure accurately by comparison of human findings in samples with information on the rate of metabolism in rodents. There was some knowledge of route and rate of metabolism in humans for some organo-phosphates and some synthetic pyrethroids and there had been studies measuring these in biological samples. If biological samples were taken in cases seen in hospitals, analysis might be able to confirm exposure, but it was noted that there would need to be a method of analysis available for a specific metabolite of each pesticide. The advantage of this type of sampling was that it was more likely that the patients would agree to samples being taken where ethical agreement had been given in a hospital context, and multiple samples might enable a ‘back extrapolation’ to determine dose.
8.3 Members noted that the available health data that had been discussed was available at ‘postcode’ level, whereas the exposure data proposed would be at ‘field’ level. Members sought confirmation as to whether these would be sufficiently compatible to be ‘joined up’ in the analyses suggested. This was confirmed, but it was noted that postcodes tend to be where people lived and agricultural data where people did not live.
8.4 The Chairman summarised that it might be possible to obtain measured personal exposures following acute exposure. In terms of considering longer term exposures any measured levels should be representative of background levels. The working group heard that there is some published data on background levels of pesticides in the USA for about a dozen substances. Analysis for the low level of background exposure was difficult. The US NHANES study as a whole had been very expensive to conduct. There was a theoretical possibility that chronic exposure could affect gene expression, and it was suggested that blood samples (if taken) could be stored for future studies.
8.5 Members agreed that it was more realistic to consider local use rather than biomonitoring as an indicator of potential exposure. In this context it was noted that although the RPA data became less specific after 2004, the information available from GIS systems was better after that date for geographic/topographical information. There was some cropping information available but it was limited. In principle it should be possible to note from GIS information that a specified field was used to grow arable crops and then consult the June survey to get an idea of the range of arable crops grown on that particular farm. It should also be possible to identify the organic farms and thus in theory to derive an estimate of likely pesticide use at a field level. Members heard that a feasibility study for this type of project was likely to cost in the region of £10,000 with nearer £60,000 being likely to undertake a fuller project.
8.6 Members asked how frequently the geographic information was updated and heard that in principle it was refreshed as changes occur. It was being reviewed this year. The Land Cover Map from the Countryside Survey is updated about every 8 years. Identifying crops from this information was difficult, but it was thought to be possible to achieve about 90% identification for satellite approaches in general.
8.7 As the aim of PUSG was to obtain regional data members noted that the variation within the data would be an important factor in determining whether they would be useful in producing exposure estimates at a local level e.g. by interpolation of results. It was noted that the spatial links between adjacent farms would be difficult as there is no clear ‘geographical’ relationship as there is for air pollution. It was likely therefore that additional validation would be required in determining whether this would be possible.
8.8 Members heard that there is a commercial company that had provided data to water companies. It was understood that the method they used involved taking the crop information, dragging a polygon across the relevant area and estimating likely pesticide use within that polygon based on survey data.
8.9 The Chairman drew the discussion to a close by asking if those involved in the various statistical exercises that had been described during the meeting could consider what might be feasible in terms of estimating local exposure to pesticides and report back to the Secretary. He thanked everyone for their participation in what had been a very interesting discussion.
9. Date of the next meeting
9.1 The next meeting will be held on 15th July in York.
J G Ayres
2010
