False alarms or missed hazards: how should regulators define “endocrine disruptor”?

November 14, 2012 at 1:50 pm | Posted in Feature Articles | 5 Comments
Tags: , , , ,

Click to download a printer-friendly PDF version of this article.

It might sound absurd to say that the definition of “endocrine disruptor” should be unscientific, but in fact there are important trade-offs entailed in how we define, for regulatory purposes, a chemical as an endocrine disruptor. These trade-offs should make us very cautious about conflating the purpose of a regulatory definition of EDC with the purpose of a scientifically correct definition of EDC, and may even show us that the process of defining EDC is a democratic matter which cannot be decided by expert committees alone.

Background to the debate on defining “endocrine disruptor”. Regulations need definitions because it has to be absolutely clear which objects and activities fall within the bounds of a particular set of laws, and which do not. How these boundaries are drawn is often fiercely debated, and agreeing on the correct regulatory definition of endocrine disrupting chemical (EDC) is no exception.

The European Commission’s DG Environment has the official lead on developing the definition, although DG for Health and Consumer Affairs has independently commissioned the European Food Safety Authority, traditionally sceptical about the risks to health posed by EDCs, to develop its own opinion about the definition of an EDC.

The concern with EDCs is that the wrong definition will result in the wrong chemicals being regulated in the wrong way: on the one hand, the UK and German authorities clearly see the definition of EDC as presenting a direct economic threat through unjustified regulation of substances for which there is too little proof of harm (DE/UK 2011, paragraph 6); on the other, environmental health groups argue that an incorrect regulatory definition will put people’s health at risk by leaving potentially hazardous compounds inadequately scrutinised for ED properties (CHEM Trust & HEAL 2012).

Concern about the hazards posed by EDCs first garnered major public attention in the mid-nineties, due in part to the publication of Our Stolen Future (Colborn et al. 1997), the first popularisation of the issues in question; however, hormonally-active substances were first used in medicine since the 1930s and the ability of chemicals to exert toxic effects via the endocrine system has been recognised since at least the 1940s (Marty et al. 2010).

Formal attempts at defining EDC go back to about 1996, with the publication of findings from a US Environmental Protection Agency workshop (Kavlock et al. 1996) precipitating the use of the term “endocrine disruptor” in the scientific literature. Although too controversial a field for any definition to yet be taken as definitive, the WHO International Programme on Chemical Safety presented in 2002 what has become a widely-referenced working definition of endocrine-disrupting chemical: “An exogenous substance or mixture that alters function(s) of the endocrine system and consequently causes adverse effects in an intact organism, or its progeny, or (sub)populations.” (WHO/IPCS 2002)

EDC regulation will set up a screening system for chemicals much like an airport scanner. We can make the scanner very sensitive, producing a lot of false alarms but also capturing a larger proportion of hazardous chemicals. Alternatively, the scanner can be calibrated to pick up only the chemicals we are certain are EDCs – but in this case we will likely miss a lot of chemicals which are EDCs but about which we have little information.

The WHO definition can be distinguished from the Kavlock et al. definition by its specific reference to “adverse effects”. Where Kavlock et al. satisfies itself with defining an EDC as any exogenous agent which interfered with hormonal processes responsible for homeostasis, the WHO/IPCS definition seeks to distinguish between effects which occur within normal homeostatic limits, and those which can be described as harmful, defining “adverse effect” thus: “A change in morphology, physiology, growth, reproduction, development or lifespan of an organism which results in impairment of functional capacity or impairment of capacity to compensate for additional stress or increased susceptibility to the harmful effects of other environmental influences.”

The WHO/IPCS definition of EDC is that which the UK and German authorities are arguing should be adopted in REACH. Even CHEM Trust, who are strongly opposed to the use of the definition in REACH, agree in principle with the veracity of a definition which describes how EDCs are toxic because they damage health via the endocrine system (CHEM Trust 2011).

So why is the problem not already solved? Quite simply, because there is a great deal of distance between what is a scientifically correct definition and what, from a regulatory perspective, is a useful definition.

Sensitivity and specificity in the classification of EDCs. When it comes to identifying EDCs we can expect to make mistakes. In fact, there are two ways of being right and two of being wrong when it comes to identifying a chemical as an EDC, as follows:

  • We identify a chemical as an EDC and in reality it is an EDC (a true positive)
  • We identify a chemical as not being an EDC and in reality it is not an EDC (a true negative)
  • We identify a chemical as an EDC when in reality it is not an EDC (a false positive)
  • We identify a chemical as not being an EDC when in reality it is an EDC (a false negative)

It is the false positives and false negatives which are the problems. From a scientific point of view, when the objective is to correctly identify the set of chemicals which are endocrine disruptors, these are obviously to be avoided, as they represent errors in scientific understanding. The useful thing about scientific inquiry, when it comes to not making mistakes, is that you are not obliged to fully commit yourself one way or the other – if you are not sure if a chemical is an EDC or not, you can always sit on the fence until you have enough data to decide.

The difference in regulation, however, is that you cannot sit on the fence. Regulatory definitions set up a binary classification system where something either qualifies to be regulated under the law, or does not. This is like an airport scanner, which either goes off or does not, with no room for don’t-knows; in the case of EDCs, false positives are effectively false alarms, while false negatives represent undetected hazards. Because there is no middle ground (the alarm either goes off, or does not) all we can do is calibrate the system to strike a balance between what we catch and what we miss, and how many false alarms you are willing to tolerate.

If we responded to every single chemical for which showed some EDC properties as if it was an actual, full-blown EDC, we could be sure we would be catching in our regulatory net nearly all the chemicals which are EDCs (in the terminology, it is sensitive). Because we are responding to indicators of harm rather than proof, we can reasonably expect that a proportion of the chemicals in the net would not in fact be EDCs (see Figure 3).

We can prevent false positives by placing a high burden of proof on classifying something as an EDC, or by saying that only the most potent EDCs with the most obvious, observable adverse health effects should be classified in regulation as EDCs (this is almost the same thing: potently toxic chemicals have obvious effects and are readily observable in relatively straightforward screening tests; the health effects of less potent chemicals can be much more difficult to discern in screening tests, with the tendency then for screening to produce less powerful evidence of harm).

A high burden of proof makes it much less likely that a chemical which is not an EDC will end up being regulated as if it is one (in the terminology, this is a specific regulatory system). However, it does make it more likely that we will generate more false negatives, which are the chemicals which are regulated as if they are not EDCs when in fact they are EDCs. These are the harmful substances which slip through the regulatory net.

Sensitivity, specificity, and false positives and negatives. Click to enlarge.

It is possible to have detection systems which are both sensitive and specific (are perfect predictors), but there is usually a trade-off between sensitivity and specificity. It is worth noting that, if your current screening system is both sensitive and specific enough then the use of a scientific definition of EDC is a natural choice, because you are not threatened by either false positives or false negatives. For the purposes of this discussion, however, we are going to assume that our understanding of how chemicals may harm health through interactions with the hormone system is not advanced enough for our EDC detection systems to be both specific and sensitive.

Where should we set the balance between sensitivity and specificity? We can choose a regulatory definition which is either specific or sensitive, but what are the trade-offs and implications?

By demanding “consequently causes an adverse effect” (DE/UK 2011, paragraph 40), the UK and German authorities are advocating a very specific definition of EDC: not only does a chemical have to show clear adverse effects, the effects have to be shown to be mediated by perturbation of the endocrine system. This high-potency, high burden-of-proof definition ensures that a minimum number of chemicals are erroneously identified as EDCs.

From an economic point of view, at least from the point-of-view of the immediate concerns of chemical manufacturers and users, pretty much any positive classification (true or false) can be seen as costly because the use of any substance which tests positive chemical is restricted, at worst costing money in product redesign and at best depriving businesses of an opportunity to make money. (Whether there is a net economic gain or loss to the state from restricting the use of these compounds is a separate issue, as that weighs long-term public health implications against short-term business interests.)

As we have seen, economic concerns are prominent in the UK/DE position paper (DE/UK 2011, paragraph 6). Nobody can reasonably argue that a screening system should avoid positives altogether; however, a very high burden of proof will certainly reduce the number of positive tests overall, and if high enough will virtually eliminate false positives altogether.

We can anticipate that this approach will generate more false negatives (see Figure 2). From an environmental health point of view false negatives are a problem because they are harmful substances which fall outside the regulatory net. More protective of health would be to have a sensitive regulatory definition which will capture more of the actual EDCs – but this will likely come at the immediate economic cost of an increased number of false positives.

There are other reasons which could motivate one to support a specific over a sensitive definition, or vice versa. Confidence in the accuracy and exhaustiveness of current test regimes would lead one toward using the scientific definition in regulation.

Alternatively, one might be confident that EDC testing will accelerate so quickly in the next 10 years or so that there is no need for a sensitive definition, as the chemicals which are missed by current standards will very soon be discovered and incorporated into the regulatory framework; on the other hand, you might think highly predictive screening is some way off, and prefer to compensate with a sensitive regulatory definition.

One can argue the explicit regulatory intent of REACH is precautionary, so we therefore have no choice but favour sensitivity and a preponderance of false positives in the regulatory definition of EDC.

There is an interesting and related observation about how EDC regulation threatens to turn toxicological testing “inside-out” (Marty et al. 2010). Where toxicology has traditionally been an outside-in discipline, where an observation of harm is followed up with study into how the substance causes harm, it contrasts with “inside-out” toxicology which is based on finding mechanistic flags for concern followed by having to prove the substance is safe before approval for use.

This observation helps us understand how a sensitive regulatory definition, in line with the precautionary principle, threatens to tip on its head the use of toxicological information in risk assessment: if precaution demands sensitivity, and sensitivity necessitates a mechanistic approach flagging potential hazard rather than a specific approach emphasising proof of harm, then toxicology becomes the discipline of proving a chemical is harmless enough to bring to market, rather than harmful enough that it should be taken off.

Balancing the implications of requiring a high or low burden of proof in identifying EDCs.

Where groups advocating a precautionary approach to chemicals regulation may not trust an outside-in approach, more traditional practitioners of toxicology may be leery of an inside-out approach. So here more than anywhere else it is important to understand the trade-offs implicit in outside-in toxicology, as by its nature it is specific rather than sensitive. The key question therefore is:  should traditional, specific approaches to toxicological testing be maintained in the face of multiplying indicators of harm and the precautionary intent of modern chemicals regulation?

Although we can’t answer that question here, the above analysis shows that a decision to go for a sensitive or specific definition is not a purely scientific one which can be settled by technocrats, but instead is one driven by a range interests including economics, public health protection, immediate financial concerns, confidence in the ability to detect EDCs, the importance of avoiding hazards or managing risks, and so forth.

Since many of these drivers are based in an individual’s values rather than society’s knowledge, or are at least a judgement of what we think we know and how we ought to react, no individual can reasonably determine on anyone else’s behalf how EDC should be defined. This brings us to perhaps our most surprising conclusion, given that we started with specificity and sensitivity: deciding on a definition of EDC has to be a democratic, open process.

Setting up an airport scanner is much more (though not entirely) a purely technical process where you decide how sensitive or specific you want it to be; deciding on a definition of EDC, with the value-based decisions it entails, is not.


RSS feed for comments on this post. TrackBack URI

  1. […] of a risk-based approach to the regulatory definition of “endocrine disruptor” (DE/UK 2011, see this H&E article for a full discussion) are both testament to […]

  2. […] On the one hand, stakeholders favouring environmental interests have been arguing that a science-based approach should give weight to novel hazard data as a way of flagging for regulatory scrutiny chemicals which might be harmful (H&E #54). […]

  3. […] False alarms or missed hazards: how should regulators define “endocrine disruptor”? The trade-offs entailed in how we define, for regulatory purposes, a chemical as an endocrine disruptor should make us very cautious about conflating the purpose of a regulatory definition of EDC with the purpose of a scientifically correct definition of EDC, and may even show us that the process of defining EDC is a democratic matter which cannot be decided by expert committees alone. (November 2012) […]

  4. […] argue for much less (for more detail on how to interpret requirements on evidence before action, see this article in H&E). The compromise on where that threshold for existing chemicals should be set still needs to be […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Blog at WordPress.com.
Entries and comments feeds.

%d bloggers like this: