Title: Effect of heptavalent pneumococcal conjugate vaccination on invasive pneumococcal disease in preterm born infants

Transcription

1 Author's response to reviews Title: Effect of heptavalent pneumococcal conjugate vaccination on invasive pneumococcal disease in preterm born infants Authors: Simon Rueckinger Mark van der Linden Rüdiger von Kries Version: 2 Date: 26 November 2009 Author's response to reviews: see over

2 Dipl.-Stat. Simon Rückinger To BMC Infectious Diseases Editorial Office Heiglhofstr München Telefon +49 (0) Telefax +49 (0) München, Effect of 7-valent pneumococcal conjugate vaccination on invasive pneumococcal disease in preterm born children Dear Ms Titmus, Dear Ms Pafitis, Thank you for considering our article for publication. We have revised the article according to the reviewer s comments. Furthermore, we have included context information to the background section and provide information on ethics, as requested by the editor. Please find attached a revised version of the manuscript and a point to point response to the reviewer comments. We would be happy to see the paper published in BMC Infectious Diseases. Reviewer: Thilde Nordmann Winther In my opinion the background is sketchy. I think it would gain, by adding some more facts. (discretionary) We have now added information on context of the study to the background section. General vaccination of all infants was recommended in July 2006 in Germany I recommend the addition with PCV7 : General vaccination with PCV7 of all infants was recommended in July 2006 in Germany. (discretionary) We have changed the text accordingly. No description of materials? I find this essential. Please describe the criteria for inclusion/exclusion etc.. (major) Thank you for pointing to this lack of details. We have now added references where more details of the study are described. We also have added a more precise case definition with inclusion and exclusion criteria. The case definition of IPD was isolation of Streptococcus pneumoniae from at least one culture of blood, cerebrospinal fluid, or a sample from any other normally sterile body site. Cases were excluded if they did not match the case definition, were older than 15 years, or double reports. Physicians filled comprehensive questionnaires on disease manifestation, outcome, vaccinations, and underlying conditions. There was a check box specifically asking for preterm birth and week of gestation of preterm births.

3 Universität München Institut für Soziale Pädiatrie und Jugendmedizin Seite 2 Details of the study have been described elsewhere [13, 21, 22]. The study is part of the International Network of Paediatric Surveillance Units (INoPSU) [23]. Section one: Hospitals were contacted on a monthly basis via postcards. I do not understand this. Why did you contact the hospitals via postcards? What was the purpose? Did you exchange patient data this way, did you attempt to remind your colleagues of the study or? (major) Every month all hospitals are contacted via postcard to ask if they had a case of IPD within the last month. If they report that there was a case, they also fill a questionnaire for it. We have rewritten the phrase as follows: Hospitals were contacted on a monthly basis via postcards to ask if a case of invasive pneumococcal disease had been observed the last month. Section two: You assume, that about 7% of children in Germany are born preterm. Please explain the basis of this assumption. (major) Some seven percent prematures in western population is text book knowledge (see for example Schaffer s Diseases of the Newborn). We have double checked this proportion in the Bavarian perinatal survey confirming that this holds for a sample covering almost 20% of births in Germany. We have also discussed potential changes in the proportion of preterm born children over time. Section two: Why did you choose to compare IPD in children born exactly in the years 2000 and 2007? I guess you choose 2007 because PCV7 was introduced in the routine vaccination program in Germany in 2006, but why 2000? Is this year representative? (major) The hospital surveillance system started in 1997 and preterm birth has been included in the questionnaire in In 2001, a recommendation to vaccinate preterm born children was issued. The hospital surveillance was confined to vaccinated cases in July 2003 and was again extended to all cases of IPD in January So 2000 is the only (preterm) birth cohort that can be tracked for two years before vaccine introduction and 2007 is the only birth cohort that can be tracked for two years after vaccine introduction. Showing data from other birth cohorts would be very interesting but we simply don t have the data. Section two: We also compared IPD notifications from the general German birth cohort the years 2000 and 2007 using the same approach. Please clarify this. Is IPD a notifiable disease in Germany? (major) Thank you for pointing to this apparently confusing presentation. Both for the general birth cohorts and the preterm birth cohorts our analysis is based on notifications from the hospital surveillance and we have now clarified this in the text (IPD is not notifiable in Germany): Also based on the hospital surveillance we compared IPD notifications for all children (irrespective whether preterm born or not) for the birth cohorts of the years 2000 and 2007 using the same approach. How did you conduct the genotyping? (major) We have not performed genotyping. We have added a description of the methodology of serotyping to the manuscript. Microbiologic laboratories were invited to send pneumococcal strains isolated from a normally sterile body site to the German National Reference Center for Streptococci, where the species identification was confirmed and serotyping, using Neufeld s Quellung method [23], was performed.

4 Universität München Institut für Soziale Pädiatrie und Jugendmedizin Seite 3 I find the result section very insufficient. It would be interesting to know the age-distribution of children with IPD. Did you find any differences between the preterm and the full-term children? You identified 14 and 8 children with IPD born preterm in 2000 respectively How long time did the surveillance last? Any differences between the two groups? How many had a GA<32 weeks respectively GA<37 weeks? Any underlying diseases/conditions that could cause at greater risk of IPD? (major) We confined our analyses to children born in 2000 and For these birth cohorts we analysed surveillance data for and respectively. Therefore, the age for both birth cohorts is confined to 0-24 months. Given the relatively small numbers and short time interval considered we do not feel a comparison of the age at infection would be meaningful. Among the preterm born children from 2007, 3 (37.5%) out of 8 had a GA of <32 weeks and one child had partial trisomy 15. Among the preterm born children from 2000, 6 (42.9%) out of 14 had a GA <32 weeks and one child had bronchopulmonary dysplasia. We have added these informations to the results section. Information about vaccination of the full-term children would be of great value (major). Information on vaccination status among IPD case notifications for this surveillance is presented in reference 13. Fully vaccinated cases with a serotype included in PCV7 are very rare. Within the context of this study, we do not feel a presentation for the subgroup of children born in 2007 would be of major interest. I think a description of the serotype distribution is necessary in a study evaluating the effect of PCV7. (major) We have now also provided information on serotype distribution for the 14 preterm born infants born in Since the focus is on the preterm children we do not feel that providing serotype information for the full term children is necessary. Reference Nr. 13 shows that the reductions among the serotypes included in PCV7 were responsible for the general decrease in IPD cases in Germany. According to your result section, you identified PCV7 serotypes from two patients, which had not received PCV7 prior to disease onset. However, according to table 2, you identified PCV7 serotypes from two patients of whom one had already received the first dose of PCV7 one month before disease onset there must be an error? (major) There were two patients with PCV7 serotypes (ID 2 with serotype 23F and ID 4 with serotype 14). Both had no pneumococcal vaccinations. We double checked this and can not see an error in table 2. These data suggest, that PCV7 vaccination is similarly effective in protecting preterm born infants from IPD compared to full term children. I am not sure if you can draw this conclusion on basis of the data. You identified 8 preterm born children with IPD only. One of them received one dose of PCV7 before disease onset (?) and one child were fully vaccinated prior to disease onset but you do not know the serotype. Furthermore, it is known, that there are natural fluctuations in the pneumococcal disease rates over time. Therefore, the reduction in IPD cases identified in this study is not necessarily due to protection from the PCV7. Some of the decline in the IPD rate could be explained by herd immunity. According to a paper by Haber et al. (Herd immunity and pneumococcal conjugate vaccine: A quantitative model, Vaccine, 2005, 25: ) conjugate vaccines may be able to induce herd effects even in situations where vaccine coverage is far from complete.(major)

5 Universität München Institut für Soziale Pädiatrie und Jugendmedizin Seite 4 Thank you for pointing to three cirtical issues: 1. Limitation of subgroup analysis: Subgroup analysis is an issue if data driven. In our study the subgroup analysis was based on the a priori hypothesis of probably similar effects in premature and suggested by a very small previous subgroup analysis. It is reassuring that these results are confirmed in our data and this is why we feel that the data are worth publishing. 2. Small numbers: We acknowledge that small numbers might be likely to account for the failure detect a significant decrease in the rate of IPD in preterm born children. The similar reduction in incidence among the preterm born children compared to the general birth cohort is rather indirect evidence but given the lack of published data we still feel that it has some relevance. 3. Potential impact of herd immunity: This is an important issue. There was no fully vaccinated child among the preterm born cases. We have discussed the potential of herd immunity. Herd immunity surely does not need 100% vaccine uptake. But with the general vaccination program just started and vaccine uptake among newborns of less than 80% do you really expect herd effects among the preterm born? Reviewer: Eugene Millar The manuscript presents the very relevant issue of invasive pneumococcal disease (IPD) and its prevention among pre-term infants. Pre-term birth is a known risk factor for a number of infectious diseases, thus the evaluation and identification of effective prevention strategies in this sub-group is warranted. [1] The limitations of this paper are quite clear. The author s reference to ongoing surveillance activities in all German pediatric hospitals is not substantiated with requisite information. How many hospitals and approximately how many providers are included? What is the approximate size of the population served? In 2008, there were 473 providers from pediatric hospitals and pediatric surgical wards. The methods section states that all pediatric hospitals in Germany are included in the study. Therefore, the entire German pediatric population is an appropriate denominator, even given the fact that reporting is not complete. Furthermore, we have now included references in the methods section that describe the surveillance in more detail. [2] They mention that the pediatricians surveyed were contacted by postcards, but there is no mention of the total number of providers surveyed or the response rate among this population. This is critical to the interpretation of the case numbers and the incidence rates that stem from those numbers. This is an interesting question but maybe not of major relevance in the scope of this study: 1. This is not a sentinel surveillance but active surveillance in all pediatric hospitals with response rates of above 95%. 2. As outlined from capture recapture analysis, active surveillance could not achieve 100% reporting rates. However, suboptimal reporting rates would only be relevant in the scope of the study if e.g. premature were more or less likely to be reported and/or if this had changed over time. There is no reason why this could be assumed. [3] The method of estimation of the pre-term born infant population (seven percent of all births) is very crude. I am not convinced this is an appropriate or reliable measure of the

6 Universität München Institut für Soziale Pädiatrie und Jugendmedizin Seite 5 population at-risk of the primary endpoint of IPD. When comparing a reduction in annual number of cases from 14 (2000) to 8 (2007), it is not difficult to imagine how an uncertain denominator would distort the results and interpretation even more. This is an important issue. However, some seven percent prematures in western populations is text book knowledge (see for example Schaffer s Diseases of the Newborn). We have double checked this proportion in the Bavarian perinatal survey confirming that this holds for a sample covering almost 20% of births in Germany. We acknowledge that there may be subtle changes in the rates of premature births over time but these if present are minor certainly below 1%. This does by no means affect our results in a way that could change the conclusions. [4] Given the bounds of uncertainty around the rate estimates (26.1 per in 2000 versus 16.7 per in 2007), one cannot argue that rates of IPD among pre-term born infants have decreased. Furthermore, if the suggested decline is to be attributed to 7-valent pneumococcal conjugate vaccine at all, some indication of serotype distribution among the 14 cases in 2000 would have been helpful. There is no suggestion of how serotype information was obtained nor whether a systematic method of serotype identification is already in place. We have added information on how the serotyping was performed to the methods section: Microbiologic laboratories were invited to send pneumococcal strains isolated from a normally sterile body site to the German National Reference Center for Streptococci, where the species identification was confirmed and serotyping, using Neufeld s Quellung method [11], was performed. Furthermore, we have added the information on serotype distribution among the 14 cases among preterm born infants from 2000 to the results: Among the 14 cases of IPD among preterm born children, 9 had information on serotype and for 7 of these the serotype was included in PCV7 (14 (3 cases), 18C (2 cases), 19A, 23F, 3, 9V). This contrasts with the serotype distribution for cases born in 2007 where 2 out of 5 cases had a serotype included in PCV7. Both cases with a serotype included in PCV7 were not vaccinated. The rate among preterm born infants was decreased. Given the small numbers we are not able to demonstrate a significant decrease. Therefore, our data are far from giving a final answer to the research question but they surely add something to an area where research is sparse and absolute numbers will be small in any study. [5] Lastly, in a manuscript which struggles to argue for declining rates of IPD among this select risk group, the absence of trend data (or, at the very least, the absence of an explanation for the non-use of data) is the greatest deficiency. One would assume that surveillance activities, while imperfect by nature, are ongoing. The inclusion of data from would have been very useful in interpreting these trends. Thank you very much for this valuable suggestion. We agree that the argument would be further strengthened by trend data for further birth cohorts. However, there is only one vaccinated birth cohort that could be followed up for at least two years was the first birth cohort for which prematurity had been consistently assessed in cases. Since in 2001 a recommendation had been given to vaccinate children with risk factors (including prematurity) this cohort could not be analysed. Additionally hospital surveillance for IPD was not maintained between 2003 and We now discuss this as a limitation and point to the lack of data for the birth cohorts from 2001 to 2006.

7 Universität München Institut für Soziale Pädiatrie und Jugendmedizin Seite 6 I am in full agreement with the authors that IPD is a serious and preventable disease among pre-term born infants and that the routine use of PCV7 will contribute significantly to prevention efforts. Evidence for the role of PCV7in reducing disease among vaccinated as well as unvaccinated populations continues to mount. However, those conclusions are drawn from other population-based studies, not the one presented here.. We fully agree that there are a number of publications demonstrating impressive beneficial effects of PCV7. Unfortunately none of these population based studies has explicitly addressed potential effects in prematures. This is what this study adds. The conclusions of our article do not claim any findings that come from other studies. Our study adds to published evidence from one clinical trial and highlights delayed vaccinations among preterm born children. Major revisions and/or supplements to the existing data are required before I would consider this for publication Thank you very much for your valuable comments which have all been considered in the revised version of the manuscript. Thank you once again for considering the article for publication. We look forward to hear from you. Yours sincerely, Simon Rückinger (on behalf of all authors)