Critique of review of deaths after chiropractic, 1

Abstract

To the Editor: It is widely acknowledged that the methodology and reporting of systematic reviews of adverse effects need to improve (1). Part of the problem is that data about harm are rarely found in randomised clinical trials (RCTs) so that identifying important harms of treatment often require a broad range of data sources, including cohort studies, case–control studies, case series and case reports. As all such studies are subject to bias to a greater extent than RCTs, their findings must be interpreted with caution (2). Given the above, we read with interest the systematic review titled, ‘Deaths after chiropractic: a review of published cases’ (3), by Professor Edzard Ernst. Ernst states in the introduction to his review that, ‘The aim of this study was to summarise all cases in which chiropractic spinal manipulation was followed by death.’ We reviewed Professor Ernst’s article and went to both primary and secondary sources in the peer-reviewed literature to double check the professional identity of the care providers in the 26 cases that he reported finding. We found that in three separate cases, Ernst incorrectly assigned the care associated with death to a chiropractor. The following is a summary of our findings: ‘Regarding the case documented by Schmitt (Reference 11 in Ernst’s review), Terrett (4) states, ‘The original paper reports a heilpraktiker (health practitioner)…’’ ‘Regarding the case documented by Mas et al. (Reference 19 in Ernst’s review), Terrett (4) states, ‘Personal correspondence with Professor Mas (Letter, 1st June 1993) reveals that he is not sure of the professional background of the practitioner involved’.’ ‘Ernst states that one death caused by care delivered by a chiropractor was documented by Dziewas et al. (Reference 27 in Ernst’s review). However, Dziewas in an exchange of letters to the editor (5,6) stated,’ ‘We acknowledge that we do not know whether the patients in question had received prior treatment by a ‘duly licensed chiropractor’ or by a ‘briefly trained chiropractitioner’… Chiropractors as in the USA represent a small minority in Germany, the influence of which can be neglected in our cohort … it would indeed have been wise to use the unspecified term ‘manipulation of the cervical spine’ instead of ‘chiropractic manipulation’.’ In the conclusion to his review Ernst states that he believes, ‘There are reasons to suspect that under-reporting (of death due to manipulation by chiropractors) is substantial…’ Ernst, however, by incorrectly labelling care providers as chiropractors in three separate cases, has added to the over-reporting of deaths associated with chiropractic care. It should be noted that in all three cases the fact that the care providers were not chiropractors is easily confirmed by checking primary and/or secondary data sources readily accessible from the peer-reviewed literature. It is interesting that all three of the errors that Ernst made, in assigning professional identity, were errors that increased the apparent association between chiropractors and death. In general, the quality of systematic reviews of adverse effects might be improved by including a systematic means by which to deal with the biases inherent in many of the non-RCT data sources often included in such reviews. A logical place to start in developing a systematic means by which to deal with such biases would be to search for and review secondary data sources in the peer-reviewed literature. None. AW and MB both contributed to the conception and content of the letter. Both authors read and reviewed the related systematic review. AW acquired data from primary and secondary sources. AW and MB contributed to drafts of the letter and both authors revised, suggested modifications and approved the final version to be submitted for publication.