Examining the evidence for the link between lead exposure and violence, Paul Illing finds a cautionary tale

In a recent article in the Guardian newspaper, George Monbiot writes that lead poisoning could be a cause of violent crime. His article is part of a spate of media interest recently in claims that a relationship exists between tetraethyl lead in petrol and violent crime. But is it true?

The pieces lead back to an article published earlier this year titled ‘America’s real criminal element: Lead’ written by Kevin Drum and published in Mother Jones, a politically left-wing US news magazine. Drum’s article debunks the claims that former Republican mayor of New York Rudi Giuliani’s success in fighting crime was the result of policing methods and claims instead that the fall in crime was really due to the earlier removal of tetraethyl lead from petrol. This is hardly scientific literature, but good, rumbustious politics. With US-style police commissioners now being elected in the UK, the fight could spread across the Atlantic. 

However, Drum’s piece is based on articles in the scientific literature by Rick Nevin,1,2 and Howard Meilke and Sammy Zahran,3 which correlate levels of lead in petrol (or air) and violent crime (murder, manslaughter, burglary), allowing for a time lag of between 18 and 24 years between exposure and effect. These are articles published in the regular, peer-reviewed scientific literature and therefore should be treated with respect.


Nevin’s original study compared changes in children’s blood-lead levels in the US with changes in IQ. The downward shifts in IQ levels were consistent with the blood-lead to IQ relationship reported by an earlier study and population shifts in average blood-lead levels for children under age 6 between 1976 and 1991.

Have we a correlation of cause and effect or have we incidental associations?

Furthermore, long-term trends in population exposure to lead in paint and lead in air from petrol (particularly during the first year of life) were found to be remarkably consistent with subsequent changes in rates of violent crime and extramarital pregnancy, with approximately 20 year time lags. Long-term trends in lead in paint and lead in petrol exposure were also strongly associated with trends in murder rates (data going as far back as 1900). While pregnancy and IQ decrement are hardly crimes, murder certainly is. 

Nevin’s study was extended using various violent crime data from the US, UK, Canada, Australia, West Germany, France and New Zealand. The 2007 paper showed a consistent relationship between preschool blood-lead levels and crime rate trends in nine so-called advanced nations. In the 2012 summary, the data series was updated to 2011, and the correlation persisted. 

Meilke and Zahran evaluated lead emissions and latent aggravated assault behaviour in six US cities. Other things held equal, a 1% increase in tonnages of air Pb released 22 years prior corresponded with a 0.46% rise in the aggravated assault rate in the present period. Overall, they claimed that their model explained 90% of the variation in aggravated assault across the cities examined. Unless everyone implemented similar policing at the same time, it seems unlikely that policing methods are the cause of the change.

And discussion

Monbiot is a campaigner; he has written ‘My job is to tell people what they don’t want to hear’. But what if his work does point to a truth? Have we a correlation (either with lead or with the political claims) that is cause and effect or have we incidental associations? Is the reduction in crime an unintended (but welcome) benefit of the removal of lead from petrol, done for other reasons? We need to consider how the epidemiological data fits with other scientific information on lead toxicity and Bradford Hill’s postulates,4 which include strength of association, consistency, specificity, temporality, dose response and plausibility. The correlation of lead and violence appears to work across much of the developed world and the temporality is impressively consistent, but it could be argued that the work needs to be replicated by other research groups. Exclusion of other possible factors is also difficult, but the political factors associated with law enforcement regimes do appear to be an unconvincing explanation. 

It is well known that lead can affect the central, peripheral and autonomic nervous systems and induce sub-clinical and clinical encephalopathy. Increased lead levels in children are reported to have the effect of decreasing IQ. Thus there are possible neurotoxicological mechanisms that may explain these findings, even if the evidence for them is unspecific. Is this enough to establish the association as cause and effect? Many will say yes but others will be more sceptical; I think the answer is ‘just about’. If nothing else, this shows how little we know about the neurobehavioural effects of many chemicals and how difficult it is to investigate them. Serendipity and good monitoring seem to be the current detection methods. 

What to conclude from all this? Firstly, be aware of the difference between association and cause and effect; secondly, be sceptical of politicians claiming victories in the fight against crime and political journalists debunking them; and thirdly, remember that the old saw ‘more research is needed’ may actually be true. 

Paul Illing is an independent toxicologist and risk assessor and an honorary lecturer at the Centre for Occupational and Environmental Health, University of Manchester, UK