Dr Rowlands raises an important question, triggered by homicide data in my recent paper on the effect of September 11 on suicide and homicide in England and Wales. He argues that when tackling violence in our society, the current Government may plan services on the basis of information that is misleading and flawed.

The data used in my paper - in excess of 130 000 unnatural deaths (E950–959 and E980–989, excluding E988.8) - were obtained from the ONS in 2002 then updated in 2003; 7400 of these deaths were classed by the ONS as manslaughter and unlawful killing (homicide; ICD–9 E969).

It was clearly pointed out in my paper that routinely collected data was a major limitation of the study, but I had to accept the nationally collected data from ONS as reliable and as complete as possible. It should be pointed out that before 1993, ONS data were based on year of registration of death but the data that were actually used in the analysis relating to September 11 related to the year when suicide and homicide occurred.

The paper made no reference whatsoever, implicitly or explicitly, to homicide trends in England and Wales since 1979. The only comment about trends in homicide was made in relation to seasonal variations to show that the reduction in homicide noted after August was not related to the events of September 11 but merely represented some seasonal pattern. The higher homicide figures that Dr Rowlands quoted may have been, as he rightly pointed out, the result of notification of deaths that actually occurred in earlier years.

Dr Rowlands has used the paper to make a political point about ‘a Government currently determined to medicalise violence’. I fail to see the relevance of his otherwise valid comment to this paper, the first and so far the only available literature on the effect of September 11 on suicide and homicide in countries other than the USA.