Sir, as part of a recent CPD study session focussing on critical appraisal of evidence and the implications for dental public health; a group of us reviewed the paper The prevention of 'dry socket' with topical metronidazole in general dental practice (BDJ 2006; 200: 210–213) by H. Devlin et al. using a standard 'CASP' critical appraisal tool developed for assessing randomised controlled trials.

The Critical Appraisal Skills Programme (CASP) is a well known and widely used programme within Learning & Development at the Public Health Resource Unit. http://www.phru.nhs.uk/casp. Since 1993, the programme has helped to develop an evidence-based approach in health and social care, working with local, national and international groups.

We are really pleased to see research carried out in general practice and recognise that the study was done with great commitment from the authors. It is crucial that general dental practitioners are encouraged to organise, run and publish clinical trials such as this that enable clinical developments to take place for the benefit of patients. However, as a result of the structured critical appraisal exercise, a number of key points and suggestions emerged that we felt should be shared with a wider audience.

Firstly, we felt that a more closely targeted research question could have been identified, to better clarify what question the study set out to answer. It was not possible to determine whether either of the interventions described (intra alveolar application of metronidazole gel or a placebo) could reduce the incidence of alveolar osteitis compared with what we believe to be the routine clinical practice in the UK of no agent being inserted into a socket post operatively. We suggest that either the control should have been no intervention or there should have been a third arm of the trial with no intervention for comparison. We recognise this would require a change in methodology to ensure blindness of the observer.

Secondly, it is usual for well conducted randomised control trials to follow up all participants in order to compare the effects of an intervention, whereas this study only followed up individuals who returned with symptoms, thereby reducing the validity of this particular study. We noted however that the authors did refer to this and other limitations of the study in their discussion.

The BDJ adopted the CONSORT guidelines for reporting on Randomised Controlled Trials in 1999. These include the production and publication of a flowchart following the participants through the trial. A flowchart is a great help to busy readers to enable them to appraise the trial. http://www.nature.com/bdj/about/consort.htm

Thirdly, although smoking was identified as a possible causal agent in the development of dry socket, and patients were advised not to smoke during the healing period, no assessment of smoking behaviour was reported at follow up for those who did return with symptoms. Given the recognised public health implications of the effects of smoking, this is an unfortunate omission.

We found the 'CASP' tool very helpful in structuring our assessment of the paper and would like to propose that consideration be given to this or a similar framework being adopted by both authors and reviewers to assist with the writing and assessment of scientific papers. We also suggest that CONSORT guidelines for randomised control trials be routinely followed by BDJ authors and reviewers both to assist the reader and to promote the highest standards of published research.

Several of us have been authors of papers and referees for peer reviewed journals in the past (including the BDJ) and would have welcomed such guidance.

Dr H. Devlin responds: Thank you for your interest in our work and for the opportunity to respond to the points raised by our colleagues.

We believe that our study was quite clearly focused or targeted in terms of the population studied, the intervention given and the outcomes that were considered. The study arose from discussions amongst general dental practitioners who wanted to find out whether topical metronidazole gel might reduce the incidence of dry socket in their practices. It was a considerable undertaking as they completed pilot studies and obtained funding and ethics committee approval. A manufacturer of metronidazole and a suitable placebo gel was found, the mode of gel delivery was designed and the blinding procedures put in place. Patient recruitment took a further three years. Using a non-intervention control would have been impossible to conduct as a double blind study as both operator and patient would realise that nothing was being placed in their socket. From our preliminary studies, the incidence of dry socket following non-intervention was already known, and incorporating such a group would have considerably lengthened the study.

Despite the additional inconvenience and increased difficulty in recruiting patients, let us assume that we had asked all patients to return for follow-up after a routine extraction. Would you classify those patients who failed to attend as symptomless or eliminate their large numbers from analysis? There can be variability in diagnosis of dry socket by different observers as not all patients present with the classical signs and symptoms. How would we have ensured consistency of diagnosis amongst the different observers? Due to these methodological considerations, we used a similar line of attack to that of other researchers in this field. The CONSORT guidelines and flow-chart are designed to highlight inappropriate patient exclusions, numbers of patients withdrawn, high loss-to-follow-up and other potential problems. Our intervention (extraction and gel application) occurred once, therefore presentation of numbers of patients who failed to complete the drug trial or numbers of patients who withdrew are unhelpful.

The public health implications of smoking are well known, and patients were advised not to smoke during the healing period. I agree with the many studies that have shown that dentists and hygienists have an important role to play in smoking cessation intervention. High quality training and adequate remuneration of dental personnel is essential for this to work effectively in general dental practice. However a recent article1 in the BDJ concluded that 'the majority of dentists have received no training in tobacco cessation strategies'.

Although we did not use the 'CASP' tool, we are also trying to achieve and promote the highest standards of published research.