The Pediatric Infectious Disease Journal

The Pediatric Infectious Disease Journal Impact of Pneumococcal Conjugate Vaccines on Selected Head and Neck Infections in Hospitalized Israeli Children --Manuscript Draft-Manuscript Number:

PIDJ-216-595R1

Full Title:

Impact of Pneumococcal Conjugate Vaccines on Selected Head and Neck Infections in Hospitalized Israeli Children

Article Type:

Original Studies

Corresponding Author:

Tal Marom, M.D. Assaf Harofeh Medical Center Zerifin, ISRAEL

Corresponding Author Secondary Information: Corresponding Author's Institution:

Assaf Harofeh Medical Center

Corresponding Author's Secondary Institution: First Author:

Tal Marom, M.D.

First Author Secondary Information: Order of Authors:

Tal Marom, M.D. Shiran Bookstein Peretz, BSc Orna Schwartz, PhD Abraham Goldfarb, M.D. Yahav Oron, M.D. Sharon Ovnat Tamir, M.D.

Order of Authors Secondary Information: Manuscript Region of Origin:

ISRAEL

Abstract:

Introduction: Streptococcus pneumoniae is a major pathogen of pediatric head and neck infections (HNIs), e.g. acute otitis media (AOM), acute mastoiditis (AM), acute bacterial sinusitis (ABS) and meningitis. The aim of the study was to characterize the epidemiology of pneumococcal HNIs (pHNIs) before, during, and after the introduction of pneumococcal conjugate vaccines (PCVs). Methods: Children 0-16 years who were hospitalized with HNIs in the Pediatrics Department in a general hospital between 1/1/2007-12/31/2014 were retrospectively identified. Study years were categorized according to the PCV introduction timeline: 2007-2008, 'pre-PCV years', 2009-2011, 'transition years', and 2012-2014, 'post-PCV years'. pHNIs episodes were defined if pneumococcal culture or urine antigen was positive. Children who received ≥2 doses of PCV7/PCV13 were considered as "immunized". All other children were considered as "unimmunized". Results: HNIs accounted for 2.5-4.7% of the total admissions; 3-17% of them were pHNIs. 87 pHNI episodes were identified: AOM (n=42), AM (n=28) and meningitis (n=17). There was a downward trend in the overall incidence of HNIs, and particularly of pHNIs, in the post-PCV years. The average age and hospitalization duration of children with HNIs/pHNIs remained stable during the study years. In 2009-2010, pHNIs incidence sharply decreased, from 7 to 1.74/1,000 hospitalized children/year, due to ~55% reduction of pneumococcal AOM episodes. An additional decrease was observed in the post-PCV years (1.62/1,000 hospitalized children/year). Immunized children were less likely to present with pHNIs (p=0.001), but were more likely to undergo surgery (p=0.042). Conclusion: We observed a reduction in pHNIs incidence after PCV program implementation.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation

Suggested Reviewers:

Earl Harley, MD Georgetown University [email protected] Walls A, Pierce M, Krishnan N, Steehler M, Harley EH Jr. Pediatric head and neck complications of Streptococcus pneumoniae before and after PCV7 vaccination.Otolaryngol Head Neck Surg. 2015 Feb;152(2):336-41. Mathew Kronman, MD University of Washington [email protected] Pediatrics. 2014 Oct;134(4):e956-65. Bacterial prevalence and antimicrobial prescribing trends for acute respiratory tract infections. Kronman MP, Zhou C, Mangione-Smith R. David Chi, MD Children's Hospital of Philadelphia [email protected] Int J Pediatr Otorhinolaryngol. 2014 Dec;78(12):2161-4. Intra-temporal and intracranial complications of acute otitis media in a pediatric population. Mattos JL, Colman KL, Casselbrant ML, Chi DH.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation

Cover Letter

June 25, 2016 George H. McCracken, Jr., MD Pediatric Infectious Disease Journal, Editor-in-Chief Dear Dr. McCracken: Enclosed please find the manuscript entitled "Impact of Pneumococcal Conjugate Vaccines on Head and Neck Infections in Hospitalized Israeli Children" for consideration as an original article for publication. In this retrospective study, we identified children who were hospitalized with acute head and neck infections, such as acute otitis media, acute mastoiditis, acute bacterial sinusitis and meningitis, during the introduction of the pneumococcal conjugate vaccines into the National Immunization Program in Israel. We analyzed the incidence of infections over the study period, and specifically focused on culture/urine antigen positive pneumococcal episodes. There is no financial support to disclose. There is no conflict of interests for all authors. The current version of this manuscript has not been submitted or accepted elsewhere. All authors have seen and approved the manuscript, contributed significantly to the work. The manuscript has not been previously published nor is not being considered for publication elsewhere. There are no color figures in this submission. We thank you for your kind consideration. Sincerely, Tal Marom, MD Corresponding Author Department of Otolaryngology-Head and Neck Surgery Assaf Harofeh Medical Center, Faculty of Medicine, Tel Aviv University 70300 Zerifin Israel E-mail: [email protected] Cell phone: +972-58-7941822

Responses to Editors/Reviewers

1

July 27, 2016 George H. McCracken, MD The Pediatric Infectious Disease Journal, Chief Editor Re: MS PIDJ-216-595, "Impact of Pneumococcal Conjugate Vaccines on Selected Head and Neck Infections in Hospitalized Israeli Children" Dear Dr. McCracken, Thank you very much for your consideration and thorough review of our manuscript. Please find the specific reviewers’ comments (in italics) and our responses below. We created a new Figure 1 (CONSORT diagram); therefore Figures 1A and 1B in the previous version are now Figures 2A and 2B, respectively. Reviewer #1: 1. The authors review cases of what they label as head and neck infections - pre and post PCV era to determine the impact of these vaccines. The infections that that the authors are including are not really "Neck" infections per-se, and does not include head and neck infections - in addition, meningitis is also included - and it's not really in the category of head/neck - it's more a CNS infection. We agree with this comment, but wish to keep the original name. The term 'head and neck infections' pools all the infectious processes in this region, and we could not find a better name to accurately describe the infections that we have studied. In fact, we even looked for Bezold's abscess in the neck, which is a known complication of acute otitis media (but yet there were no cases identified in our study). Other terms, such as 'upper respiratory tract infections', classically do not include meningitis, and 'head infections' or 'cranial infections' are neither descriptive enough nor in common use in the literature. We also used the word 'selected' in the title to imply that only a few infections were studied. We defined in the Introduction and Methods sections the specific infections which were studied, and we believe that this is sufficient. Thank you for understanding. 2. For table 2, would be more specific and outline the specific infections over time. It would be more interesting to see which infections changed over time. Do the same for Table 3 and 4. As requested in the author's instruction, we did not want to present the data twice. The requested data is already shown in Figure 1A (overall HNIs, corresponds to Table 2) and Figure 1B (pHNIs, corresponds to Table 3) in a clear manner. If you wish, we can create a supplemental table which provides all the numbers (as an on-line material only). As for Table 4, it compares unimmunized with PCV-immunized children who presented with pHNIs, regardless of the site of infection. We clarified that in the text. 3. Looking at figures 1 and 2 - it seems that HNIs were on already declining - and that there was a modest decline after PCVs. What's the explanation for this? .

2

We assume that the reviewer referred to Figure 1A (all-cause HNIs) and Figure 1B (pHNIs), there was no Figure 2. We observed a reduction in the HNIs incidence, but not in pHNIs incidence, in the pre-PCV years (2007-2008), due to decrease in AOM hospitalization rates. Since our study population was comprised of hospitalized children (and not in the outpatient setting), and data from earlier years (before 2007) were not collected, we cannot draw concrete conclusions. We may attribute this to the publication and implementation of local AOM diagnosis and treatment guidelines published in 2006, which resulted in a decrease in AOM over-diagnosis. We commented on that in the Discussion section. Reviewer #2: 1. The paper submitted by Marom and colleagues is a retrospective observational study to evaluate the impact of Pneumococcal Conjugate Vaccines on Head and Neck Infections in Hospitalized Israeli Children. The data is interesting and their hypothesis important. Thank you for the positive comment. The manuscript needs substantial improvements in its statistical analysis, results presentation and discussion as it will be presented here: 2. Abstract: While the conclusion assumes a direct association between PCV implementation and reduction of pHNIs the study is not designed to claim that assertion. The authors could conclude that they observed a reduction after PCV program implementation. Agree, revised as suggested. 3. Page 6, line 33: the use of the word "calendrical" is better when one refers to pertaining or used by a calendar system, therefore it is more customary to use the word calendar. Consider modification. Agree, revised as suggested. 4. Methods (Definitions of HNI): The authors insinuate that they used the described definitions to ascertain cases fitting the diagnosis, yet later is clear that they retrieved record using the ICD-9 coding. Therefore, unless these children were part of a prospective study were specific data regarding the outlined definition were used, then they should explain how the definitions were applied to the retrieved cases. For example, if a child is admitted with mastoiditis as per ICD-9 code 383.0 then how many of the signs and symptoms as stated in the definition that patient presented with. This is a crucial point to clarify and specify. The paper will benefit from presenting a CONSORT diagram where the authors show how many records with ICD-9 codes were initially screened, how many excluded and how many fit the case definitions. As defined in our Methods section, this was a retrospective study. In order to ascertain the quality of the collected data, we carefully read all the eligible medical

3

records which appeared in the retrieved list we came up with from the archives. Only children who met the criteria for each of the selected infections were eventually included. Methods section: "We cross-matched and verified that only children who met the eligibility criteria for HNIs were included. For each verified HNI episode, we retrieved the patient's age, sex and duration of hospitalization. For each verified pHNI episode, we retrieved the patient's age and sex; history of current disease; culture type and results; duration of hospitalization, the need for imaging studies and surgery, as well as complications". We also created a CONSORT flowchart diagram, as requested, where we show how many records with ICD-9 codes were initially screened, how many excluded and how many eventually were included (now new Figure 1). 5. Methods (Page 8, lines 47-55): The authors do not describe how immunization status was ascertained for each case (immunization card, electronic registry, etc.). Did they have data on immunization status for each case? In addition the statement that "the study population is not considerably different from the general Israeli population; therefore, the general vaccination rates were projected to the study population" is not clear. Did they use those proportions to impute the immunization status of the cases? What projections were done? Immunization data is stored in Mother and Child Health Clinics across Israel, which are all affiliated to the Israeli Health Ministry. Due to the retrospective nature of this study, we could not access these data from the patient's immunization card or other electronic databases. Yet, as we have commented, our population is similar in many characteristics to the one in Southern Israel, where Ben Shimol reported a high vaccination rate as mentioned. Furthermore, vaccination data were routinely collected by physicians from caregivers upon hospitalization at the Pediatrics Department in each case. Caregivers were asked if their child's vaccination status was up-to-date. According to these data, children >4 months of age who received ≥2 doses of PCV7/PCV13 were considered as "immunized". All other children were considered as "unimmunized" In addition, in our previous paper where we did have access to the immunization data, we showed a very high uptake of PCVs in children in our area (but was not added to the text, in order to minimize confusion with other previous cohorts): Changing trends of acute otitis media bacteriology in central Israel in the pneumococcal conjugate vaccines era. Tamir SO, Roth Y, Dalal I, Goldfarb A, Grotto I, Marom T. Pediatr Infect Dis J. 2015 Feb;34(2):195-9. doi: 10.1097/INF.0000000000000536. We elaborated that in the Methods section, where we justified our vaccination categorization. 6. Methods (Page 9, lines 11-14): When the authors state that "statistical analysis was performed per H&N infection episode, and not per patient. Sequential episodes were considered to be separate events if they were ">30 days apart"; they need to

4

recognize that if a case was counted twice in the numerator; it should be counted twice in the denominator. In addition, the risk of those having a recurrent infection is different than those that had only one during the observation period, and that is why most of the time only first cases are counted (to avoid the bias introduced by inadvertently selecting a special population - unknown immunodeficiency). The authors need to justify their inclusion of recurrent cases and describe how many were counted more than once. Also see our response to this reviewer, Q15. We did exactly what the reviewer said. In order to reduce the bias and have "safe margins", we defined an episode if at least 30 days elapsed from the previous HNI episode. We clarified this in the Methods section. 7. Methods (Page 9, lines 16-23): The statistical analysis is not well described and does not match what is presented in the results and tables. Who or what are the unmatched samples? For example, it is unclear how the authors arrived to the p 2 doses of pHNI cases (31%) to the expected 92% in the Israeli children they should state so, but it is not clear what the other variables (gender, duration of hospitalization, imaging, surgery, etc.) where compared against. One possibility is to use the non-pHNI cases as controls to gain insight into how these groups differ or not in PCV immunization and other variables, but the authors do not describe the methods appropriately. We clarified that in the text. In table 4, the unimmunized group served as the control group, and the interventional group was comprised of the children who were vaccinated (as defined in the Methods section). This is true for all the variables which are listed there. We footnoted that. We did not collect all the data shown in Table IV for the entire all-cause HNIs cases, but just for the pHNIs cases, which were our interest. We specifically mentioned that in the text: "For each verified HNI episode, we retrieved the patient's age, sex and duration of hospitalization. For each verified pHNI episode, we retrieved the patient's age and sex; history of current disease; culture type and results; duration of hospitalization, the need for imaging studies and surgery, as well as complications". We hope it is clearer now. 8. Methods: Please provide information of how the pneumococcal cases were ascertained. How many by culture, how many by urine antigen, etc. It now appears in Figure 1 and in the Results section, pHNIs paragraph. 9. Results (Page 10, lines 13-16): The authors need to show the data on the proportion of unimmunized children if they want to pursue this argument, especially as it is not clear that they ascertained immunization status of every case. Tables III and IV now provide the requested data.

5

10. Discussion (page 11, lines 8-11): The authors cannot claim evidence for vaccine effectiveness from a case series only analysis unless they define a proper control group. If they made the inference from the proportion of vaccinated population in Israel it is important to mention it including the potential selection and ascertainment biases from hospitalized cases only. We toned down the theme, and mentioned your limitations. We mentioned twice that our results are in hospitalized children, and so does the title of this article. 11. Discussion (page 11, lines 18-23): Even when the figure shows a sharp decrease in HNI and AOM in the Figure and the authors' state that a substantial decrease in these infections was observed, they should make an effort to run the proper statistical comparison of proportions to determine how significant this finding is. Thank you. We eventually performed some statistical tests and provided the requested data in the footnotes of Figure 2B: "When comparing 2009 (PCV7 introduction year) with 2008 (pre-PCV7 year), the overall pHNIs, AOM, and AM hospitalization rates dropped in 73%, 85% and 12% , p=0.04, p=0.02, p=0.23, respectively. Meningitis hospitalization rate remained unchanged. When comparing 2007 (first pre-PCV year studied) with 2014 (last post-PCV year studied), the overall pHNIs, AOM, AM and meningitis rates dropped in 89%, 79%, 81% and 74%, p=0.02, p=0.02, p=0.2, p-0.03, respectively". 12. Discussion (page 12, lines 11-13): There is nothing in the results or tables that backs up the statement that a significant reduction in AM and meningitis was observed. Please add the data and statistical comparisons to the results. A table comparing the incidence in the pre-PCV to the post-PCV periods and the difference with confidence intervals and a statistical test would be beneficial to the paper. The sentence now reads: "This was also followed by a significant reduction in AOM rates and to a much lesser extent, in AM and meningitis rates". As shown in Figures 2A and 2B, the data is provided according to the time-line of the study, and we marked when PCV7 and PCV13 were introduced .We also added: "There were no appreciable changes in the incidence of all-cause AM, ABS and meningitis", and " Since there are already now 2 figures, 4 tables, and requests for 2 additional tables (one from Reviewer #1 and now this request) - please advise if we can exceed the journal's limit for this manuscript. 13. Discussion (page 12, lines 33-35): The authors describe the "meticulous gathering of data regarding enrolled children" but this is a retrospective study of cases and unless these children were participating in a prospective study with standardize data entry they are subject to the biases and confounding of retrospective reviews. In addition, the "matching in an exact manner the child's immunization status" is not described in the methods as pointed out before. How this data was ascertained (immunization card?). We rephrased that sentence, and it reads now as: "Our main strengths are the meticulous collection of data regarding enrolled children, matching in an exact

6

manner the child's immunization status, as previously described, and the current disease". We commented on the immunization status earlier. 14. Table and Figure titles need to stand on its own with better description of the data and naming the location of the results (Tel Aviv, Israel). The title of our paper clearly mentions that we studied Israeli children. We described in the Methods section the location of our medical center and size of referral population. Since we are not the only medical center in the greater Tel Aviv area (6 medical centers serving some 1.5 million people), we cannot comment that our findings can be inferred to other places, although we believe that there are no substantial differences in the characteristics of the pediatric population. 15. Table 2 and Methods: Were any patients with recurrent admissions or >1 admission included in the denominator or numerator? Please provide the number of cases who had > 1 infection as stated in the Methods. As we defined, we counted the number of HNIs/pHNIs episodes, not the number of children, since patients could have been admitted more than once during the 8 years of the study. Those children were counted both in the denominator and the numerator in our calculations. We had 4 children who had >1 pHNI admissions: 2 with AOM, 1 with AM and 1 with meningitis. We added that to the Results section. 16. Table 2 and 3: There is a major discrepancy in the number (n=) of HNIs from both tables. Unless there is a reason that some subjects were dropped off (no culture or testing done for Spn), those numbers should be equal. If there is a reason for exclusion they should be specified in the Tables footnotes. You are right, and we apologize for the inconsistency. The numbers of HNIs per each calendar year were corrected. Reviewer #3: 1. Can the author mention the immunization coverage for the PCV7 / 13 among Israeli children in the area involved in the study? Please see our response to Reviewer #2, Q5. There are no official vaccination data for the Tel Aviv area. 2. In Table III, the percentage of immunized children among those who developed pHNIs from 2007 - 2014 could be included and described. Per your request, we added this information in Table III. 3. The younger age for pHNIs developing among immunized patients as shown in Table IV may be not significant (NS), but may reflect on non-vaccine serotypes being more likely to occur in this younger age group just as the increased likelihood for surgery among the immunized group could indicate more severe serotypes affecting those who were immunized. Some references can be mentioned.

7

We agree, but didn't want to get into this issue, since we don't have serotype data in our study. We added the following sentence in Discussion section: "PCV immunized children presenting with pHNIs were younger than unimmunized children (not statistically significant). This observation may reflect that non-vaccine serotypes were more likely to occur in the younger age group, and similarly, the increased likelihood for surgery among PCV immunized children could indicate more severe serotypes affecting those who were immunized. Support for these theories can be in other papers (and added 3 references)". 4. Fig 1A and B could be presented more clearly and using consistent legends or markers especially for AOM and AM. Since we are limited to black and white figures only, we did our best to emphasize AOM and AM. Maybe we can work on this later with the production team, if the paper is accepted. 5. Conclusion can only mention the reduction in the over-all HNI and pHNIs incidence among hospitalized children after PCV immunization but not about reducing the nasopharyngeal carriage since this is not part of the study. Agree, we dropped that section. 6. Page 13 Lines 35-37 need to be revised as the statement is unclear. These sentences read now as: "We demonstrated the effect of both vaccines (PCV7/PCV13) on HNIs, and more specifically on pHNIs epidemiology. Yet, due to the short interval between PCV7 introduction and its replacement by PCV13, within 16 months, we could not show the effect of each vaccine separately. In our view, our results show the bigger net effect of PCV13". We believe we have satisfactorily revised the manuscript as suggested. If accepted, Dr. Ovnat Tamir will be the corresponding author. Thank you very much for your kind consideration. Sincerely, Tal Marom, M.D. Corresponding Author Department of Otolaryngology - Head & Neck Surgery Assaf Harofeh Medical Center 70300 Zerifin Israel Phone: (+972) 58-7941822 E-mail: [email protected]

Title Page

Impact of Pneumococcal Conjugate Vaccines on Selected Head and Neck 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Infections in Hospitalized Israeli Children Tal Marom, MD1, ±, Shiran Bookstein Peretz, BSc2, Orna Schwartz, PhD3, Abraham Goldfarb, MD4, Yahav Oron, MD4, Sharon Ovnat Tamir, MD4 Authors and Affiliations: 1

Department of Otolaryngology-Head and Neck Surgery, Assaf Harofeh Medical

Center, Tel Aviv University Sackler School of Medicine, Zerifin, Israel 2

Faculty of Medicine, Tel Aviv University Sackler School of Medicine, Tel Aviv,

Israel 3

Microbiology Laboratory, Edith Wolfson Medical Center, Tel Aviv University

Sackler School of Medicine, Holon, Israel 4

Department of Otolaryngology-Head and Neck Surgery, Edith Wolfson Medical

Center, Tel Aviv University Sackler School of Medicine, Holon, Israel ±

Address correspondence to:

Tal Marom, MD Department of Otolaryngology-Head and Neck Surgery Assaf Harofeh Medical Center Tel Aviv University Sackler School of Medicine 70300 Zerifin Israel Phone: +972-58-7941822 Fax: +972-8-9779421 Email: [email protected] Running head: Pediatric H&N Infections in the PCV Era

Abstract

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

ABSTRACT Introduction: Streptococcus pneumoniae is a major pathogen of pediatric head and neck infections (HNIs), e.g. acute otitis media (AOM), acute mastoiditis (AM), acute bacterial sinusitis (ABS) and meningitis. The aim of the study was to characterize the epidemiology of pneumococcal HNIs (pHNIs) before, during, and after the introduction of pneumococcal conjugate vaccines (PCVs). Methods: Children 0-16 years who were hospitalized with HNIs in the Pediatrics Department in a general hospital between 1/1/2007-12/31/2014 were retrospectively identified. Study years were categorized according to the PCV introduction timeline: 2007-2008, 'pre-PCV years', 2009-2011, 'transition years', and 2012-2014, 'post-PCV years'. pHNIs episodes were defined if pneumococcal culture or urine antigen was positive. Children who received ≥2 doses of PCV7/PCV13 were considered as "immunized". All other children were considered as "unimmunized". Results: HNIs accounted for 2.5-4.7% of the total admissions; 3-17% of them were pHNIs. 87 pHNI episodes were identified: AOM (n=42), AM (n=28) and meningitis (n=17). There was a downward trend in the overall incidence of HNIs, and particularly of pHNIs, in the post-PCV years. The average age and hospitalization duration of children with HNIs/pHNIs remained stable during the study years. In 2009-2010, pHNIs incidence sharply decreased, from 7 to 1.74/1,000 hospitalized children/year, due to ~55% reduction of pneumococcal AOM episodes. An additional decrease was observed in the post-PCV years (1.62/1,000 hospitalized children/year). Immunized children were less likely to present with pHNIs (p=0.001), but were more likely to undergo surgery (p=0.042). Conclusion: We observed a reduction in pHNIs incidence after PCV program implementation.

Manuscript

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

INTRODUCTION Streptococcus pneumoniae is a leading causative pathogen of common pediatric head and neck infections (HNIs), such as: acute otitis media (AOM), meningitis, acute mastoiditis (AM( and acute bacterial sinusitis (ABS). The latter could be accompanied by related complications, e.g., orbital cellulitis and subperiosteal abscess 1-4. Prior to the introduction of pneumococcal conjugate vaccines (PCVs), S. pneumoniae was the cause of ~11% of all deaths in children aged 1-59 months 5. In the last two decades, PCVs were integrated worldwide into National Immunization Programs (NIPs), and were found to be effective, safe and immunogenic in infants 6, 7 Numerous papers reported on the changes in the epidemiology, microbiology and clinical presentation of pediatric HNIs upon vaccination with PCV7 4, 8-11. Of them, AOM and AM were the 2 most commonly investigated infections, due to their relative high prevalence among hospitalized children. Only one U.S. study pooled various pediatric HNIs among PCV immunized children aged 1-4 years from the Kids National Inpatient Database (1997-2009), however it focused mainly on the economic aspects of these infections (hospital stay and healthcare costs) 1. To our knowledge, there is limited data on the epidemiology of pediatric HNIs in the post-PCV13 era. Most pediatric HNIs are managed in an outpatient setting, and since obtaining cultures is routinely non-feasible, empiric antibiotics treatment is administered. Only few HNIs require hospitalization, when there are suspected complications, or if oral antibiotic treatment has failed 12-15. The causative agents of HNIs in non-hospitalized and hospitalized children differ. A recent study demonstrated that children hospitalized with AOM have bacteriological cultures which are similar to children with AM, rather than to non-hospitalized children with AOM 16. In this study, we analyzed the

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

incidence and epidemiology of pneumococcal HNIs (pHNIs) in children hospitalized with HNIs in the years prior, during and after the introduction of PCVs in the Israeli NIP. PATIENTS AND METHODS Study Design and Population

The study was approved by the local Institutional Review Board. We retrospectively identified children 0-16 years, who were hospitalized between 1/1/2007-12/31/2014 in the Pediatrics Department in a secondary hospital (which serves some 150,000 children). There were no appreciable changes in the population size during the study years. Children presenting to the pediatric emergency room with HNIs, who were not hospitalized, but eventually had positive pneumococcal cultures as reported by the laboratory a few days following their visit were excluded, since an analysis on all outcomes could not be performed. Children with congenital malformations of the ear, nasopharynx and palate were excluded, as were children with immunological disorders. For each calendar year, we retrieved the number of admitted patients to the Pediatrics Department. Rates were calculated per 1,000 hospitalized children.

Definition of HNIs

AOM was diagnosed by otoscopy when children presented an acute onset of symptoms (10 days without clinical improvement; or 3) when URI symptoms had worsened after an initial improvement.

Meningitis was diagnosed in children presenting an acute onset of headaches, nuchal rigidity, high temperature (≥100.70F (38.20C)), and altered mental status that underwent lumbar puncture for cerebrospinal fluid (CSF) testing.

Definitions of pHNIs

We comprised a patient log of children with pHNIs by retrieving records of children who had 1) relevant International Classification of Diseases-9 codes in their discharge letter (as shown in Table I), and 2) pneumococcal culture from the middle ear, blood, abscess, CSF, or had a positive urine pneumococcal antigen. Serotyping for S. pneumoniae was performed only in children presenting suspected meningitis, since it is considered as an 'invasive pneumococcal disease' (IPD), which mandates further characterization. No serotyping was available for the other pHNIs, as they are currently not considered as IPDs, and thus do not mandate it. Nevertheless, due to the small number of pneumococcal meningitis cases, serotype data were not included, since any conclusions concerning

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

serotype changes secondary to PCV implementation could not be drawn. We cross-matched and verified that only children who met the eligibility criteria for HNIs were included. For each verified HNI episode, we retrieved the patient's age, sex and duration of hospitalization. For each verified pHNI episode, we retrieved the patient's age and sex; history of current disease; culture type and results; duration of hospitalization, the need for imaging studies and surgery, as well as complications. When children were coded for both AOM and AM, they were counted as one episode of AM, which is the more severe form of disease.

PCV Status In Israel, childhood vaccinations against S. pneumoniae are given to all infants at Mother and Child Health Clinics at 2, 4 and 12 months of age. Following its introduction in the Israeli NIP (2009), PCV7 was administered to infants who were born from January 2008. In November 2010, PCV13 replaced PCV7 in the routine NIP. Therefore, 2007-2008 were considered as the 'pre-PCV years', 20092011 were considered as the 'transition years', in which both PCV7 and PCV13 were implemented in the NIP (2011 was considered as the "catch-up" year for PCV13), and 2012-2014 were considered as the 'post-PCV years'. Due to the short interval between PCV7 and PCV13 introduction, analyzing the changes occurring in the post-PCV7 years was not available; therefore we focused only on the postPCV13 years. Due to the retrospective nature of this study, we could not access immunization data from the patient's immunization card or other external electronic databases. Yet, our population is quite similar to the one in Southern Israel (situated 60 miles from us), where Ben Shimol recently reported a high vaccination rate for PCVs. By June 2011 and December 2012, ~80% and ~90% of 7-11 months old

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Israeli children living in Southern Israel received ≥2 PCV7 and/or PCV13 doses, respectively 17. By June 2014, ~95% of them received ≥2 PCV13 doses. Furthermore, vaccination data were routinely collected by physicians from caregivers upon hospitalization at the Pediatrics Department in each case. Caregivers were asked if their child's vaccination status was up-to-date. According to these data, children >4 months of age who received ≥2 doses of PCV7/PCV13 were considered as "immunized". All other children were considered as "unimmunized". Statistical Analysis Data were recorded using Microsoft Excel office software. Subjects included were those that met eligibility criteria, who complied with the study design protocol and for whom all data were collected. The statistical analysis was performed per H&N infection episode, and not per patient. Sequential episodes were considered to be separate events if they were >30 days apart, in order to avoid the bias of inadvertently including partially-treated episodes or selecting a special population, for example with an unknown or undiagnosed immunodeficiency. In recurrent cases (>1 admission), episodes were counted twice in the incidence calculation, both in the numerator and the denominator. Statistical analysis was performed using SPSS 17.0 software. Contingency table analysis for comparing rates between unmatched samples was performed using the Chi-square test or Fisher’s exact test, as appropriate. The Student independent samples t-test was used to compare continuous variables. All tests were considered significant if p-values were ≤0.05. RESULTS HNIs HNIs accounted for 2.5-4.7% of admissions to the Pediatric Department in the study years, and there was a downward trend in the incidence in the post-PCV years compared to previous years (Table

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

II). Boys were more likely to be hospitalized with these infections than girls, 58% (477/820) vs. 42% (343/820), p=0.02. There were no appreciable changes in the average age of hospitalized children and in the average hospitalization duration across the study years. pHNIs Figure 1 shows a CONSORT diagram how children were initially identified and eventually included. As shown in Table III, 87 children were identified with pHNIs: 83 had positive culture, and 4 had positive pneumococcal urine antigen (all meningitis patients). AOM (42), followed by AM (28) and meningitis (17) were the most common pHNIs. There were no cases of pneumococcal ABS or pneumococcal head and neck abscesses (i.e., Bezold's abscess). There were 4 children who had >1 admission: 2 with AOM, 1 with AM and 1 with meningitis. Boys were more likely to have pHNIs than girls, but the difference was statistically insignificant. In the pre-PCV years, there were ~20% positive pneumococcal culture rates, which sharply declined to