Doses of carbamazepine and valproate in bipolar affective disorder

Sir: Taylor & Duncan (Psychiatric Bulletin. April 1997. 21. 221-223) make clinical recommenda tions for dosing regimen and plasma levels of carbamazepine and sodium valproate in bipolar affective disorders based on a Mediine search. We believe that there are three areas in which this paper is inadequate. First, the authors did not detail key words, search strategy or years of search. Whether or not the search was limited to English was not denned. This clearly detracts from the validity of the paper. The lack of a detailed methodology means that the literature search that they have carried out could not be easily reproduced by an independent researcher. Second, although heterogeneity of treatment response is mentioned in the discussion, no attempt is made to explore the reasons for the variability in results between studies or validate the comparison of treatment responses between sub types of bipolar disorder such as acute mania, rapid-cycling bipolar and cyclothymia. Study de sign is not discussed and short-term open studies appear to be given equal weighting to double-blind studies. For example, equal weight is given to the information gleaned from the five double-blind studies concerned with the use of carbamazepine and the presumably open studies of valproate. Third, there are intrinsic difficulties associated with Mediine searches (Adams et al, 1994). To provide a complete systematic review of current research findings a Mediine search would need to be supplemented by extensive hand searching. Relying solely on Mediine searches will miss a large body of published literature including conference abstracts, 'grey' literature and dis

Sir: Taylor & Duncan (Psychiatric Bulletin. April 1997. 21. 221-223) make clinical recommenda tions for dosing regimen and plasma levels of carbamazepine and sodium valproate in bipolar affective disorders based on a Mediine search. We believe that there are three areas in which this paper is inadequate.
First, the authors did not detail key words, search strategy or years of search. Whether or not the search was limited to English was not denned. This clearly detracts from the validity of the paper. The lack of a detailed methodology means that the literature search that they have carried out could not be easily reproduced by an independent researcher.
Second, although heterogeneity of treatment response is mentioned in the discussion, no attempt is made to explore the reasons for the variability in results between studies or validate the comparison of treatment responses between sub types of bipolar disorder such as acute mania, rapid-cycling bipolar and cyclothymia. Study de sign is not discussed and short-term open studies appear to be given equal weighting to double-blind studies. For example, equal weight is given to the information gleaned from the five double-blind studies concerned with the use of carbamazepine and the presumably open studies of valproate.
Third, there are intrinsic difficulties associated with Mediine searches (Adams et al, 1994). To provide a complete systematic review of current research findings a Mediine search would need to be supplemented by extensive hand searching. Relying solely on Mediine searches will miss a large body of published literature including conference abstracts, 'grey' literature and dis sertations. Unpublished data, bearing in mind a bias against 'negative' studies, will not be included in the review. There is no indication that any hand searching or supplementary information gathering was used in this study.
While this paper addresses an interesting and important area, we advise that its conclusions must only be regarded as preliminary in light of these methodological shortcomings. Biological Psychiatry) and collect details of posters and abstracts. We examine the reference sections of all papers on file and then retrieve other relevant reports. As a consequence, our Mediine searches usually only reveal that we have on file all Medline-referenced papers and many more be sides.
Our search could not be reproduced by an independent researcher. This is partly because we did not specify search terms, but also because searches are time-dependent.
In addition, our hand search was, as already described, extensive and was, by definition, not systematic or reproducible.
We reviewed studies for which full trial details were available. Our intention was to question normal prescribing practice for which there is virtually no support in the literature. The conclusions we drew are evidence-based, but preliminary, given the restricted nature of our article. We contend, however, that our conclu sions are more cogent than current, unsup ported dose recommendations.

Financial implications of Caiman changes in psychiatry
Sir: I am writing to highlight the discrepancy in the pay structure for trainees in psychiatry as a result of the Caiman changes in postgraduate training. Presently there is a dichotomy among 784 Psychiatric Bulletin (1997). 21. 784-789