SMJ 2003 48(4): 102-104
Dr Helene Irvine
Consultant in Public Health Medicine
(Communicable Disease and Environmental
Health)
Public Health Protection Unit
Department of Public Health
Greater Glasgow NHS Board
350 St Vincent Street
Glasgow, G3 8YU
email
(w): helene.irvine@gghb.scot.nhs.uk
Now the same unit is publishing the Scottish findings, subsumed within the much larger original study, based on just 61 special waste landfill sites in Scotland, as a paper in its own right. Paradoxically, these show a non-statistically significant deficiency of congenital anomalies around Scottish special waste landfill sites (relative risk of 0.96) and no statistically significant results for any other point estimates, which sit randomly on both sides of a relative risk of unity. The superficial conclusion one is tempted to make is that living near English and Welsh special waste landfill sites might be bad for fetuses but that living near Scottish special waste landfill sites is probably safe. Because the relative risk was also elevated for all British landfill sites combined (special and non-special), the authors query whether Scottish special waste landfill sites are better regulated than non-special waste sites in England and Wales and whether non-special waste sites south of the Border illegally receive hazardous waste. However, the authors also admit that residual confounding and data artifacts are alternative explanations for the positive findings in the British study.
The likely explanation for the discrepant results is that the Scottish study of special waste landfill sites, that shows no statistical association, reflects the true picture - that there is no causal association between living beside landfill sites (special or otherwise) and giving birth to children with congenital anomalies and the larger British study misleads the reader by virtue of a number of epidemiological pitfalls as described below.
The first is the quality of the congenital anomaly data (the dubiety of which probably applies to both studies, although possibly unequally) using national registers and termination data. Although routinely collected health data is generally thought to be much better quality in Scotland than in England, the quality of national congenital anomaly data is thought to be poor and patchy throughout Great Britain. Certainly, in Scotland, the Scottish Congenital Anomaly Register (SCAR) has commanded little faith on the part of epidemiologists, who tend to use EUROCAT data for those selected sites where it is collected (in Scotland this means only Glasgow). One can only assume the equivalent database in England, where the sheer scale of data collection makes it more cumbersome and incomplete, is even worse. This weakens both analyses and cannot be overcome by the size of the British study.
Another obvious pitfall is the blurring of the division between aetiological significance (relative risk more than 2 or 3) and statistical significance (P< 5% or 1%). Overall, there was a one percent excess in congenital anomalies of all kinds (relative risk 1.01) in the population living within 2 km of a landfill site (which remarkably amounted to 80% of the British population, which means most of us are 'exposed'!) and this was just statistically significant thanks to the large power of the study (99% CI 1.005 to 1.023). These are tiny excesses, which just scrape by in terms of statistical significance.
In addition, the relative risks for a few of the larger sub-groups of anomalies were also statistically significantly, although not aetiologically, elevated in the British study including those for neural tube defects, hypospadias / epispadias and abdominal wall defects. From their data, these sub-groups made up only about 14% of all the anomalies, however, suggesting that the other 86% were under-represented around the landfill sites, on average, in order for the overall relative risk to average out at just 1.01. This is an example of selective reporting of statistically significant positive findings. Several dozens of categories of anomalies will have been under-represented around the landfill sites. The authors have not tried to suggest (wisely) that living near landfill sites would have conferred a protective effect for these types of anomalies.
The other important weakness, particularly relevant to the British study is that offered by Bonferroni's principle which suggests that one is bound to find 1% of one's analyses to be statistically significant by chance (when testing at 99% confidence levels) even when there is no real association. In a study of this size with so many categories of anomalies, clinical interventions and undesirable birth outcomes, different types of landfill sites, different degrees of adjustment for confounders, etc. there are bound to be hundreds of results, and therefore at least two or three of which would be spuriously statistically significant.
In addition to showing no aetiologically significant relative risks (i.e. only the weakest of statistical associations were demonstrated in the British study) they have yet to demonstrate most of the other criteria of causality (e.g. dose response, consistency of findings, biological plausibility, etc.). For instance, no common biological mechanism links any chemicals or metals in the emissions from landfill sites (in air, soil or leachate) with neural tube defects, cardiovascular defects, epispadias/hypospadias and abdominal wall defects.
One criterion of causality they have examined (temporality, i.e. that exposure to the hazard preceded the outcome in time) is refuted by the 522 landfill sites (all waste types) that demonstrated statistically significantly raised hospital admissions for treating abdominal defects before opening of the landfill site. This counterintuitive finding was also observed by the Welsh group, who studied the anomaly rate at Nant-Y-Gwyddon landfill site before and after the landfill site became functional (Fielder et al 2000).
Finally, their methodology is not the ideal way to examine the potential for ill health, which will be subtle if it exists, caused by living near a landfill site (whatever the kind). By combining hundreds of landfill sites, most of which are completely benign in terms of their emissions, they are diluting any effect that might arise from the minority of more sinister sites. Ecological studies of this kind are relatively easy to do and what they produce is anyone's guess - too many hazardous outcomes, or too few and usually of borderline statistical significance. Table II in the Scottish study demonstrates exactly that - a hodgepodge of relative risks that are all over the place and with wide confidence intervals that include unity, reflecting the much smaller sample size of this study compared to the Great Britain equivalent. They would be better finding a 'mother of a landfill site', known for its foul smell, dodgy consignment data and suspicious borehole monitoring records that is closely surrounded by dense housing, and study the health effects on that local population bearing in mind the direction and nature of the prevailing winds. Researchers would be advised to liaise closely with the Scottish Environment Protection Agency/Environment Agency officers who know which sites to concentrate on.
In response to local concerns, Greater Glasgow NHS Board conducted a detailed and comprehensive study of routinely collected health data around such a site: Paterson's of Greenoakhill, one of the largest special waste landfill sites in Scotland. It found no evidence of ill health associated with living near the landfill site, and in fact, found the conception and birth rate 50% higher than expected and twice as many children aged 0-4 living around the site than expected. This is interpreted as suggesting that parents intending to have large families are choosing to live near these sites, presumably because the mortgages are less of a financial burden. A recent study by DEFRA confirmed the huge disamenity effect on the value of Scottish property within .25 km of a landfill site. These findings are not in keeping with a toxic effect on reproduction.
Another feature is Elliot et al's admission in the original paper that the area within 2 km of these special waste landfill sites tended to be more urban and deprived than that beyond 2 km and had higher percentages of births born to the most deprived tertile of women, women under 20 years of age, black women and women of ethnic origin. It is quite possible that it is the inability to adequately adjust for these key confounders that explains why similar studies might show small excesses of undesirable outcomes for those living within 2 km of such landfill sites. In other words, unadjusted social deprivation may account for any residual excesses in birth outcome statistics around landfill sites and not any toxic effect of the landfill site itself.
In fact, Elliot et al's explanation of the contradiction between the finding of the current Scottish and original British study of 2001 is inadequately brief. They suggest that it might be due to:
- chance
- sociodemography of areas surrounding landfill sites.
- some aspect of the type or management of waste in Scottish special waste landfills compared with the rest.
Let's explore these three factors in more detail.
- chance - This is the most likely explanation. By conducting such a large study of British landfill sites, involving hundreds of separate analyses they found a small number of statistically significant though aetiologically non-significant findings. This is now being contrasted with the mixed results of non-statistically significant findings from a very small study associated with what is probably a similar hazard north of the border.
- sociodemography of areas surrounding landfills. What little evidence exists suggests that lower income families might be preferentially concentrated around landfill sites in Scotland. This is supported by the recent DEFRA study of property values around landfill and how these are disproportionately reduced around Scottish landfill sites (40%) because of the disamenity factor compared to English sites (7% on average). This would result in an excess of anomalies around landfill sites due to ingestion of drugs/alcohol during pregnancy or intrauterine infection that would need to be adjusted for. However, this reasoning fails to explain why relative risks were higher for some sub-groups of anomalies born to women living around British versus Scottish landfill sites when the poverty would be more concentrated around the Scottish sites. Adjustment was carried out using the same methodology in both studies. However, the Carstairs index was developed in Scotland and may be a more effective adjustment tool in Scotland.
- some aspect of the type or management of waste in Scottish special waste landfills compared with the
rest. England's sites will be larger on the whole and have larger and denser populations around them. It also follows that English sites receiving far greater quantities of waste than most Scottish sites will receive greater quantities of potentially polluting wastes mixed through municipal wastes whether the site was classified as special or not. Historically, in the interests of cost-effectiveness, much of the most potentially polluting Scottish special waste was moved to England where it was cheaper to process in bulk.
In addition, English sites will generally receive less rainfall to damp down any air-borne dust generated by dumping.
Historically, groundwater resources (which may have been more prone to contamination in England given the greater density and scale of landfill sites) have been utilised in England for potable supplies to a far greater extent than in Scotland where surface water abstraction has been more commonly utilized. However, it is unlikely that a population around a landfill site would end up drinking the polluted ground water generated by that same landfill site.
All of these circumstantial factors would suggest that, historically, English landfill sites would be more 'toxic' than their Scottish equivalents, if not toxic (in terms of contaminated drinking water) to the population immediately around the landfill site in question.
In fact, the quality of regulation of English landfill sites has historically been better than Scottish regulation due to the different regulatory structures in place.
In particular, after 1994 when the regulatory framework changed across the UK, English sites were more likely to be engineered-lined sites, while most Scottish sites continued to use the less precautious 'dilute and dispersal model'. Regulation of Scottish landfill sites improved markedly after 1996 when SEPA took over the role of regulation from the local authorities who previously acted as both the disposal and regulatory authorities Over the next 5 years, regulation will continue to improve throughout Britain due to the ever tighter standards required by European Directives.
In summary, there is circumstantial evidence to suggest that some English landfill sites might have been 'rogue sites' until 1994. In theory this provides a possible explanation for any excess of disease or illness around English landfill sites observed between 1982 and 1997. However, the argument presented above about how any hazardous effect of a minority of rogue sites will be diluted by combining them with the majority of benign sites applies once again as do all the other epidemiological arguments. Furthermore, the evidence showing that anomalies were raised at many of these sites before the operation of the landfill site argues against the landfill providing a toxic effect. Also, the historical sub-optimal regulation of Scottish sites was not reflected in higher risks of anomalies between 1982 and 1997 as ruled out by the new Scottish study.
Nevertheless, this Scottish paper is important and needs wider publicity. A large number of Scots live close to these landfill sites in urban, relatively socially deprived areas and their concerns about the effects of their health have been fueled by papers published by Dolk et al (1998); Elliot et al (2001) and Vrijheid et al (2002). A huge amount of concern about health was created in the populations around Paterson's tip and Greengairs landfill site in Scotland by Friends of the Earth Scotland and local pressure groups who need to be able to distinguish between the potential to contaminate the groundwater (which is real) and a threat to the health of local populations (which is far less likely). There has been relatively little published about this subject in Great Britain and almost none in Scotland.
Given how few studies have been published on this subject it is puzzling that they have omitted the paper by Vrijheid et al (2000). However, even more disappointing is the fact that they have omitted the more recent important paper by Vrijheid et ali. In this most recent publication, they found there was no correlation between congenital anomaly rate and special landfill site when the latter were ranked on the basis of their degree of hazard as deemed by regulators. This important recent work has seriously knocked back the hypothesis that special waste landfill sites cause harm to the unborn child and should be included in their review of the literature.
Most landfill sites will be producing lots of odourless methane tinged with sulphur-smelling contaminants such as mercaptans and hydrogen sulphide. However, the majority of these will not be causing any harm to the local population other than providing a major public nuisance that should be addressed on those grounds. Unfortunately, it is difficult to get anything done about foul-smelling landfill sites even when their license stipulates that they are not allowed to smell off the site. This single issues highlights, better than most, the limitations of any regulatory body when attempting to attribute foul smells even when its officers are confident they know its source.
In addition, landfill sites have been generally unattractive and a source of vermin, wind-blown rubbish and sea gulls. They are made even more unpleasant to live beside by the large amount of lorry traffic they attract. In addition, it is considered undesirable to live near these landfill sites so the value of one's property tends to suffer if one lives near a landfill site.
We need to reduce our production of waste at a both individual and corporate level, improve our ability to segregate and recycle waste, improve our ability to safely incinerate waste, educate the public about all these issues so they can see how their own habits contribute to the problem, etc.
In conclusion, it seems fitting that this editorial would conclude that 'exposure to most landfill sites probably does not harm the fetus and that more of the right kind of evidence is needed. In the meantime, let's look more closely at the evidence that does exist.'
Reference
1. Morris, Thomson Jarup et al. No excess risk of adverse birth outcomes in populations living near special waste landfill sites in Scotland. Scot Med J 2003 48(4): 105-107