Reason for withdrawal from publication
The Editor withdrew this review as of Issue 1, 2008. This review has been updated and replaced following the publication of a new review: Jacquerioz FA, Croft AM Jacquerioz FA, Croft AM. Drugs for preventing malaria in travellers. Cochrane Database of Systematic Reviews 2009 , Issue 4 . Art. No.: CD006491. DOI: 10.1002/14651858.CD006491.pub2 .
The editorial group responsible for this previously published document have withdrawn it from publication.
Keywords: Adult; Female; Humans; Male; Antimalarials; Antimalarials/adverse effects; Malaria; Malaria/immunology; Malaria/prevention & control; Malaria, Falciparum; Malaria, Falciparum/immunology; Malaria, Falciparum/prevention & control; Mefloquine; Mefloquine/adverse effects; Randomized Controlled Trials as Topic; Travel; Treatment Outcome
Feedback
Herxheimer
Summary
Methods
1. The four outcomes you list are not evidence of efficacy and tolerability, but evidence of INefficacy and INtolerability.
Reply: This comment is strictly speaking quite correct (though it doesn't alter our results and conclusions). We have included this as a passing observation in our Methods section.
2. You analysed only the 5 most commonly reported symptoms in the neuropsych and g.i. categories. Does this mean you ignored the "other" symptoms reported within these categories? I think they should be analysed as "other neurops" and "other g.i." symptoms ‐ otherwise you may get potentially serious distortion, esp. since several different terms are likely to have been used for essentially similar symptoms and what is "most commonly reported" will be partly determined by the terms used and how they are categorised/lumped.
Reply: This is a helpful suggestion, although the analysis would not be at all likely to alter our conclusions. We will endeavour to carry out this analysis in a later update of the review, which will also include an analysis of all the "adverse effects" case reports (which in the latest update is simply presented as an annotated bibliography).
4. Half the authors responded to your request for unreported outcomes, but you don't say which of those 5 actually sent you unreported data, and which replied "sorry, ain't got none". In the acknowledgments you collectively thank those who sent you unpublished data, but I think you should name them individually.
Reply: We are somewhat reluctant to "name and shame" those who never replied to us, as there might be perfectly good reasons for this (eg, death of the investigator, or change of post), and we wouldn't want to be seen to be witch‐hunting! However the idea of thanking those who did respond with unpublished data is a good one, and we have included their names in our Acknowledgments section.
Characteristics of included studies
Did you come across any possible conflict of interest? The institutional investment in many of these trials was very important ‐ whether Roche, WHO TMP, armed forces, and should be remembered in evaluating these studies. If no report acknowledged funding or conflict of interest, that is also worth saying.
Reply: Very helpful suggestion. We have included a short comment where appropriate in the Notes column of the "Characteristics of Included Studies" table.
Results
1. The Ohrt trial comparing mefloquine against placebo raises a serious question about the ethics of using a placebo in such a high‐risk setting. Was this approved by an ethics committee? What information were the participants given about the trial? Did they volunteer, or just get orders? The ethical aspects of the other placebo‐controlled trials also deserve discussion.
Reply: The authors of the Ohrt trial reported that "Our study was conducted in accordance with U.S. Army, U.S. Navy and republic of Indonesia regulations governing the protection of human participants." They also state that "All volunteers gave written, informed consent." There is no mention of an ethics committee, as such. We have included a note about the somewhat worrying ethics of this trial, in the "Characteristics of Included Studies" table.
2. I am uncomfortable about your lumping together all the alternative regimens without any discussion or explicit justification.
Reply: We are uncomfortable about this also, although it does help to show, in a clear visual format, that there is a consistent tendency for mefloquine to be less well tolerated than all the standard alternatives. We may drop this lumping in a future update of the review, and separate out all the interventions being compared. No action taken at the moment.
3a. You report "slightly more withdrawals from the study in the mefloquine group than with the alternative regimens but this is not significant". You should say "not statistically significant".
Reply: We have changed the text, as suggested.
3b. Were the reasons for withdrawal and non‐compliance not tabulated in any of the trials? That is surely important evidence. Worth saying that too in Para 5 of Implications for research, where you merely make the general accounting point without saying that we need to know the reasons why.
Reply: We have modified the text at several points, to reflect these suggestions.
4. The review skirts around the issue of ADR profiles. It is important and potentially revealing to compare the detailed ADR profiles for mefloquine (and other drugs, not necessarily in this review) derived from observational studies, from your high‐grade RCTs, and from the excluded trials. Where they all point the same way they will strengthen each other, where they don't they will point us towards questions that need sharper investigation.
Reply: We have now "enriched" the review with much more data about the adverse effects ‐ initially as an annotated bibliography of case reports. Future updates of the review will carry some detailed analysis of these case reports.
5. The 10 trials differed interestingly in that five began with a loading dose and five did not. I would expect a higher incidence of adverse reactions, and a greater intensity of those reactions in the groups given loading doses, and also that the rate of ADRs would then after say 4 to 8 weeks gradually drop back to the level of the groups that got no loading dose. There might even be a difference in ADR profile, though I hesitate to predict that.
Reply: It seems that the use of a mefloquine loading dose was being promoted by the manufacturer, but has not been approved anywhere for use by travellers, other than military. The reason seems to be that yes, it is less well‐tolerated than the once‐weekly dose. If the mefloquine loading dose starts to come into general use, we will investigate it in detail and report on it further. At this stage, we have included simply a brief observation about the loading dose, in our Discussion section.
Discussion
The observational findings that women report mefloquine side effects more commonly than men. An obvious difference is in mean weight, so they receive higher mean dose/kg, and the accumulation due to the long half‐life of mefloquine means that the sex differential is likely to widen with duration of treatment. This view could be tested by looking at time‐to‐ADR‐report data for the two sexes.
Reply: Given that 98% of the participants in the included studies are male, we cannot draw any robust conclusions at this stage on what might or might not be the explanation for this gender difference in the tolerability of mefloquine. Where appropriate, however, we have included a comment on the gender of dropouts in the Notes column of the "Characteristics of Included Studies" table.
Conclusions
In implications for research, Para 3, you propose at least one RCT in travellers etc, but then say dogmatically that they should not be native‐English or native‐Dutch speakers! The reason you give seems to me far from cogent. The publicity would make such a trial harder to do but no less necessary. The central issue is decent reliable balanced information for the public and a fortiori for trial participants. Much more effort has to go into composing and providing that.
Reply: This is a very valid comment, and we have now modified the pertinent section of our Conclusions section.
Reply
See replies to statements in the Summary section.
Contributors
Contributors Andrew Herxheimer Jan 27 1998
Erny
Summary
Results
1. Higher withdrawal rates with mefloquine were only shown in comparison with placebo but not when mefloquine was compared with alternative chemoprophylaxis such as doxycycline or chloroquine and proguanil.
Reply: This is a reasonable comment, and we have included this observation in our Discussion section.
2. The use of withdrawal rates is a problematic measure of tolerability because the reasons for withdrawal from controlled trials are frequently unrelated to the study drug… It is misleading to use withdrawal rates as a direct measure of tolerability without considering the reasons that led participants to abandon the study.
Reply: Systematic reviews of clinical interventions commonly use withdrawal rates as a measure of the tolerability or otherwise of the study interventions. In randomised controlled trials participants are allocated at random between the arms, and if the tolerability of each arm is the same, the number of withdrawals from each arm should also be equal. This effect is not shown with mefloquine prophylaxis. No action taken on this comment.
Discussion
1. All the drugs available today for malaria prophylaxis have adverse effects ‐ the seriousness of malaria warrants tolerating certain temporary adverse effects.
Reply: We accept the first half of this statement. However the "enhanced" systematic review we have now published (which includes data about adverse effects from non‐randomised studies) shows that some of the adverse effects that have been associated with mefloquine prophylaxis have lasted months or years ‐ and hence have not been "temporary". This is cause for concern.
2. If more severe adverse effects occur, travelers need to seek medical advice and if justified need another drug prescribed, so that they don't remain without a protection for the rest of their journey.
Reply: We accept this statement, and have incorporated it into the Discussion section of our review.
3. Alarmist reports on adverse effects that are based on weak grounds must be challenged by evidence based recommendations. Otherwise travellers' lives will be put at risk unnecessarily, through under‐utilisation of efficacious prophylactic drugs.
Reply: We agree that recommendations on anti‐malarial drug use should be based on good evidence (which is why we prepared this review in the first place!). We do not accept that drawing attention to drug adverse effects is "alarmist". By studying and discussing adverse effects honestly, clinicians can be helped to prescribe drugs more safely. No action taken on this comment.
Conclusion
Croft and Garner's meta‐analysis is inconclusive. The interpretation of this inconclusive body of evidence is, however, clearly biased.
Reply: We accept that in the absence of appropriate research, our review is inconclusive (and we have said so in our Conclusions section). We do not believe that it is "clearly biased".
Reply
See replies to statements in the Summary section.
Contributors
Samuel Erny September 25 1998
What's new
| Date | Event | Description |
|---|---|---|
| 9 November 2008 | Amended | Converted to new review format. |
History
Protocol first published: Issue 1, 1996 Review first published: Issue 2, 1997
| Date | Event | Description |
|---|---|---|
| 12 November 2007 | Amended | Review withdrawn as of Issue 1, 2008 |
| 15 May 2005 | New search has been performed | New studies sought but none found |
| 28 August 2000 | New citation required and conclusions have changed | Substantive amendment |
| 27 August 2000 | New search has been performed | Substantive amendment |
Sources of support
Internal sources
Liverpool School of Tropical Medicine, UK.
Defence Secondary Care Agency, UK.
Army Medical Services Research Executive, UK.
External sources
Department for International Development, UK.
Withdrawn from publication for reasons stated in the review