Dr I. G. Chestnutt, author of the research summary about the paper responds: My experience in health services research over many years has taught me about the numerous practical difficulties of conducting research in the 'real world'. However, if the findings of such studies are to lead to improved patient care and contribute to the scientific evidence base, then adherence to basic scientific principles is important.

The decline in caries prevalence seen in Tristan da Cunha, does instinctively lead to the impression that children on the island benefited from the fluoride supplements. However, a study involving only 32 children, with a large age range (6–19 years), where the baseline clinical examinations were conducted some 16 years prior to the implementation of the fluoride supplementation programme, by a non-calibrated dentist and lacking a control group, contravenes some very basic principles in oral epidemiology and study design. None of the features cited in Dr Southwick's letter, however unique, overcome this problem.

In my view, this study has failed to account for confounding factors sufficiently to allow the authors to conclude that their findings 'are highly suggestive of the effectiveness of the introduction of a fluoride supplementation programme in addition to fluoride toothpaste use' ( BDJ 2003 195: 161).

As I documented in my original commentary, if claims are to be made of the efficacy of therapeutic agents, then adherence to fundamental epidemiological principles is important. An editorial in a particular issue of Evidence-Based Dentistry, reports that most systematic reviews published in dentistry have been inconclusive.1 This is frequently attributed to the poor design of the studies which contribute to these reviews.

My comments on this paper were therefore directed at policy makers and the dental research community in general, not solely on the limitations of Dr Southwick and his colleagues' work.

Far from writing in an ivory tower as Dr Southwick suggests, my practical experience in commissioning and evaluating oral health services, shows that we will in the twenty-first century, be expected to demonstrate evidence of the highest scientific standard, if we are to fulfil the expectations of bodies such as NICE (National Institute for Clinical Excellence) and SIGN (Scottish Intercollegiate Guidelines Network) and to rebuff the claims made by those opposed to fluoride and fluoridation.

In summary, I whole-heartedly agree with the need for 'real world' research. However, if we are to provide the evidence that will ultimately lead to better care for our patients and improvements in oral health, then logistical and practical difficulties should not form a barrier to science that will stand up to expert scrutiny, within dentistry and beyond. That was the point of my Commentary.