Recent eLetters

Peter Aaby and colleagues present again highly controversial data in the Archives, namely that vaccines have non-specific effects on infant survival depending on the timing and sequence of vaccines and on sex.[1] The same research group has already published many articles on this topic which have generated much debate [2-4] without reaching a final conclusion whether their claims are correct or not (for a full publication list of the research group see http://www.bandim.org/index.php?publications and http://www.ncbi.nlm.nih.gov/pubmed?term=aaby%20p). Despite the authors' own conclusions, this current paper may not be as convincing as supposed.

My first argument relates to the small numbers of deaths in the different categories on which their conclusions are based (see table 3).[1] In 3 sub-groups, only 2 or 3 deaths occurred during the 4 months of study. If just 1 death less or more had occurred in the opposite sub- groups, leaving total numbers of deaths and sex-related deaths the same, then BCG at birth followed by DTP would increase the deaths rate ratios (DRR) to 2.04 for girls and boys (not only for girls as claimed by Aaby and colleagues), whereas in the late BCG group there would be an advantage for boys (DRR 0.40), but no effect at all on girls (DRR 1.20) (for these hypothetical recalculations the table can be supplied by email as it was not possible to include it here for technical reasons).

My second argument relates to the high attrition rates and hence incomplete reporting. When looking at figure 1, out of the initial cohort of 2320 neonates 490 infants were not included in the follow-up after 2 months because either they had died, moved, travelled, refused to participate or had no vaccination card.[1] This equals an attrition rate of 21% at this point of the trial. Out of the remaining 1830 infants, only 1501 (including the 50 deaths) could be assessed in the final analysis at 6 months, a loss of another 14% of the original cohort. The authors do not provide further data about this group of infants lost to follow-up whether any additional deaths occurred and if so in which sub-group. But in 2011, the same group published results of the same trial with exactly the same cohort covering the first 12 months of life in another journal where they examined the non-specific effect of early versus late BCG vaccination on survival.[5] In that paper, they reported 8 more deaths between the age of 2 and 6 months, 3 in the early and 5 in the late BCG vaccination sub- group. Thus there must be some more information on the outcome of infants between the age of 2 and 6 months which is not reported in this publication (for details compare the trial profiles in this (figure 1) and the earlier publication (figure 1)).[1,5] It would be most welcome if the authors could explain why they did not report on these 8 additional deaths here and what the results would be with these additional deaths.

These observations demonstrate that the question of beneficial or harmful non-specific effects of vaccines on infants has not been settled yet. As the expanded program on immunization is so central to the health and survival of infants all over the world, I fully agree with the authors that randomised controlled trials are needed to answer this extremely important, but still controversial issue.

References

1. Aaby P, Ravn H, Roth A, Rodrigues A, Lisse IM, Diness BR, Lausch KR, Lund N, Rasmussen J, Biering-Sorensen S, Whittle H, Benn CS. Early diphtheria-tetanus-pertussis vaccination associated with higher female mortality and no difference in male mortality in a cohort of low birthweight children: an observational study within a randomised trial. Arch Dis Child. 2012 Feb 13. [Epub ahead of print]

2. Fine PE, Smith PG, Evans SJ. Non-specific effects of BCG? J Infect Dis. 2012 Feb;205(3):515.

3. Aaby P, Roth A, Biering-Sorensen S, Ravn H, Rodrigues A, Whittle H, Benn CS. No evidence of bias in trial showing BCG reduces neonatal mortality. J Infect Dis. 2012 Feb;205(3):515-7.

4. Shann F. The nonspecific effects of vaccines and the expanded program on immunization. J Infect Dis. 2011 Jul 15;204(2):182-4.

5. Aaby P, Roth A, Ravn H, Napirna BM, Rodrigues A, Lisse IM, Stensballe L, Diness BR, Lausch KR, Lund N, Biering-Sorensen S, Whittle H, Benn CS. Randomized trial of BCG vaccination at birth to low-birth-weight children: beneficial nonspecific effects in the neonatal period? J Infect Dis. 2011 Jul 15;204(2):245-52.

Conflict of Interest:

None declared

I read with interest the article by Roderick et al. Authors seem to recommend post exposure prophylaxis with aciclovir from day 7 onwards for 7 days for immunocompromised patients. Whilst there is good evidence for such use in immunocompetent subjects to enable development of effective immunity and suppress the clinical disease, there is no good evidence for its use in immunocompromised patients. In fact your references 17 and 18 clearly acknowledges the lack of evidence. The reference provided Fisher JP et al(17) cross references aciclovir post exposure prophylaxis to an article by Ishida Y, Tauchi H, Higaki A, et al. Postexposure prophylaxis of varicella in children with leukemia by oral acyclovir. Pediatrics 1996;97:150-1. Ishida et al used post exposure aciclovir in 3 patients within 48 hours and 2 out of them has mild symptoms. They also used normal human immunoglobin for 2 out of the three patients which may have modified the out come. Ishida et al recommended early use to prevent primary viremia and severe disease.

Until we have good evidence, in similar cases when VZIG is not available or practical, it is recommended to use aciclovir at the earliest date post exposure rather than wait for 7 days.

Conflict of Interest:

None declared

We'd like to thank Dr Litt for informing us that he was the first to comment on this treatment method (1) and apologise for having missed this in our review (2). We would, however, add that the function of a short evidence-based review like Archimedes is to weigh the evidence for the treatment, and in particular look at when it has been subjected to trials. Harsh as it might seem, this sadly rarely includes the initial discovery of a therapy. We can see that with hindsight it might have been polite if we had included a very brief mention of the original observation in our opening line of our conclusion, and we would certainly endeavour to to this if we publish further on this subject.

References 1. Litt JZ. Abstract Dont excise--exorcise. Treatment for subungual and periungual warts. Cutis. 1978 Dec; 22(6): 673-6. 2. Stubbings A, Wacogne I. Question 3. What is the efficacy of duct tape as a treatment for verrucas vulgaris? Arch Dis Child. 2011; 96: 897-9.

Conflict of Interest:

None declared

Dear Editor,

We read with interest the recent paper by Lim et al. reporting the prevalence of specific polysaccharide antibody deficiency in a cohort of 96 children with chronic wet cough.[1] Specific polysaccharide antibody deficiency was defined in terms of inadequate pneumococcal serotype- specific antibody levels following immunisation with Prevenar and/or Pneumovax II. Results of other immunological investigations were not provided other than they had normal total immunoglobulin G levels.

We consider that it is important to investigate children with apparently inadequate responses to pneumococcal immunisation more thoroughly because poor antibody responses may be a feature of other specific immunological conditions.[2] We would also argue that the label of "chronic wet cough" is an imprecise term. Lim et al. do not specify the proportion of children who underwent bronchoscopy and bronchoalveolar lavage in their cohort. If a child has sufficient chronic symptoms sufficient to warrant further investigations including immunological assessment then flexible fibreoptic bronchoscopy and bronchoalveolar lavage (BAL) should be considered routinely. BAL culture is the 'gold standard' to diagnose bacterial lower respiratory tract infection in this situation. Sputum is rarely obtained from younger children and cough swabs may not reflect lower airway microbiology accurately.[3]

Zgherea et al. recently reported bronchoscopic findings in 197 children with chronic wet cough.[4] In this study 46% of BAL samples were positive for respiratory bacterial pathogens and there was also a high incidence of structural airway abnormalities, including laryngomalacia and/or tracheomalacia in 30% of children aged 3 years.

We therefore believe that careful clinical assessment of children with "chronic wet cough" is indicated. This should include both immunological investigations and bronchoscopy in order to confirm bacterial infection and to identify any structural airway abnormalities.

References

1. Lim MT, Jeyarajah K, Jones P, et al. Specific antibody deficiency in children with chronic wet cough. Arch Dis Child. Online First: 30 January 2012. doi:10.1136/archdischild-2011-300691

2. Driessen G, van der Burg M. Educational paper: primary antibody deficiencies. Eur J Pediatr. 2011;170:693-702

3. Equi AC, Pike SE, Davies J, et al. Use of cough swabs in a cystic fibrosis clinic. Arch Dis Child. 2001;85:438-9

4. Zgherea D, Pagala S, Mendiratta M, et al. Bronchoscopic findings in children with chronic wet cough. Pediatrics. 2012;129:e364-9

Conflict of Interest:

None declared

Dear Sir,

We read with interest the observations made by Hughes et al but it is important for readers to be aware of the limitations of their means of data analysis.

Hughes et al have combined all the pre-CSII HbA1c values and calculated a mean HBA1c (Referred to as Pre-CSII HbA1c) and then compared this with subsequent post CSII HBA1c measurements at different time points. The problem with this method is that glycaemic control fluctuates in all young people with diabetes and simply using t-tests to calculate significance is over simplistic. It does not allow assessment of longitudinal trends which may not be detectable between two time points but which may be significant across a series of points. We would advocate using methods that explicitly account for repeated measures such as mixed effects model to determine the influence of both random effects (unmeasured variation in patients, which are inherent in each patient over the period of study) and fixed effects (the effects of CSII) to determine the true impact of CSII initiation. One therefore needs to assess longitudinal trends as well as un-measured patient effects.

Moreover, as the authors point out, the number of subjects fall to less than one third of the initial value (n=67 pre CSII and n=18 at 18 months of CSII usage) which has the potential to introduce substantial bias. For example, there may be differences between patients who have been on CSII for a longer period of time and those who have been on CSII for a shorter duration.

We recognise that CSII may have advantages in the pre-school child but teasing out the advantages and disadvantages of this means of insulin delivery requires more sophisticated studies than that reported by Hughes et al. Kind regards,

AS,MT,SR and TC

Conflict of Interest:

None declared