Introduction Streptococcus pneumoniae is a major cause of various diseases, ranging from septicemia and meningitis to pneumonia and acute otitis media (AOM). As community-acquired pneumonia (CAP) is a leading cause of childhood mortality , the World Health Organization (WHO) recommends inclusion of pneumococcal conjugate vaccines (PCVs) in childhood immunization programs . In Latin America, pneumococcal disease rates among young children are intermediate in comparison with other global areas , but the impact of PCVs in diminishing this burden has not been assessed in this region. Pneumococcal serotypes included in the ten-valent pneumococcal nontypable Haemophilus influenzae protein D conjugate vaccine (PHiD-CV) represent 70%–80% of those that cause invasive pneumococcal disease (IPD) and AOM in young children in Latin America ,. PHiD-CV was licensed for protection against IPD based on demonstration of immunological non-inferiority to the seven-valent pneumococcal CRM-conjugate vaccine (7vCRM; Prevenar/Prevnar, Pfizer) , using criteria proposed by the WHO . In contrast, for mucosal diseases–e.g., pneumonia and AOM–no such licensing criteria are defined. Furthermore, antibody levels for most of the pneumococcal serotypes contained in both vaccines, when expressed as geometric mean concentrations, tended to be higher with 7vCRM than with PHiD-CV . This has unknown implications for the magnitude of protection against mucosal diseases of importance to public health. Also, the etiology of the mucosal diseases involves many bacterial and viral pathogens ,– and can be affected by factors such as variations in pneumococcal serotype incidence –, care-seeking behavior, and antibiotic prescription practices . Since double-blind randomized controlled trials are the gold standard for establishing vaccine efficacy (VE), the Clinical Otitis Media and Pneumonia Study (COMPAS) was designed to demonstrate the efficacy of PHiD-CV against CAP and AOM, and to assess other clinical end points, such as IPD, in young Latin American children. Methods Ethics Statement The trial was sponsored by GlaxoSmithKline Biologicals. An independent data monitoring committee (IDMC), composed of seven independent experts in infectious diseases and/or statistics, provided oversight by reviewing serious adverse events (SAEs) and assessing potential treatment harm. The IDMC also made recommendations to the sponsor regarding safety measures, study design, and analysis and reporting plans. Written informed consent was obtained from children's parents/guardians, and the study was conducted in accordance with good clinical practice, all applicable regulatory requirements, and the Declaration of Helsinki. When deviations from these guidelines/regulatory requirements were detected (Text S1), corrective actions were implemented and reported to the ethics committees, IDMC, and regulatory authorities. The trial protocol (Text S2) was approved by national public health authorities and the ethical review committees for each study site (Table S1). Major amendments made to the protocol, including changes to the planned interim analysis, are listed in Text S3. Study Setting, and Socioeconomic and Public Health Indicators This study was conducted at five sites: three in Argentina (Mendoza, San Juan, and Santiago del Estero), one in Colombia (Cali), and one in Panama (Panama City). The study locations were chosen partly because local investigators and public health authorities from the three countries were experienced in collaborative epidemiological surveillance research on pneumococcal diseases, including an international surveillance study of invasive bacterial isolates (SIREVA [Sistema Regional de Vacunas] and SIREVA II [Sistema de Redes de Vigilancia de los Agentes Bacterianos Responsables de Neumonías y Meningitis]) ,. Investigators in Panama were also experienced in the conduct of AOM studies –. The study areas were mainly urban, and local climates have either clear seasonality (Argentina) or are tropical or subtropical (Panama and Colombia). Populations in these countries generally have good access to health care and drug treatments, including antibiotics (upon prescription in Argentina and Panama, but freely available in Colombia). Immunization coverage for routine childhood vaccines is high in each country (in 2011, approximately 80% to 99% for recommended vaccines) . Seasonal influenza vaccination has been recommended since 2010 in Argentina and since 2005 in Colombia and Panama for 6- to 24-mo-old children ,, for whom coverage is at least 70% in Argentina and Panama. All three countries are classified by the World Bank, based on gross national income per capita, as upper middle income economies, a classification shared by many other countries in Latin America, Eastern Europe, Asia, the Middle East, and North Africa . Gross domestic product per capita in 2011 was US$10,942 for Argentina, US$7,498 for Panama, and US$7,104 for Colombia. Adult life expectancy is similar among the countries (76 to 78 y), as is the adult literacy rate (93% to 98%) . In comparison to the global infant mortality rate (37 deaths per 1,000 live births) , infant mortality rates are low (13, 17, and 15 per 1,000 live births in Argentina, Panama, and Colombia, respectively) and similar to or lower than rates in other countries with upper middle income economies (average 16 per 1,000 live births)  but higher than the average rate in the European Union (4 per 1,000 live births) . Each study area had health care centers and at least one public pediatric hospital that were readily accessible to participants by public transport. Study Design This was a phase III double-blind randomized controlled study. The primary confirmatory objective of the study was to demonstrate the VE of PHiD-CV (three doses in children aged 18 mo or less, or four doses in children aged over 18 mo) against first episodes of likely bacterial CAP (B-CAP) occurring at least 2 wk after administration of the third dose of study vaccine in the per-protocol cohort for efficacy. Efficacy against clinically confirmed AOM (C-AOM) was evaluated as the first secondary confirmatory objective; efficacy against IPD and other CAP and AOM end points were assessed as secondary objectives (Table 1). 10.1371/journal.pmed.1001657.t001 Table 1 Study objectives. Objective Study Cohort Primary objective To demonstrate the efficacy of PHiD-CV against B-CAP All children Secondary objectives To demonstrate the efficacy of PHiD-CV against C-AOM 7,000 children enrolled in Panama To assess the efficacy of PHiD-CV against CAP with alveolar consolidation or pleural effusion on chest X-ray All children To assess the efficacy of PHiD-CV in preventing bacteriologically confirmed AOM cases caused by: Any bacterial pathogen Vaccine, cross-reactive, and other S. pneumoniae serotypes H. influenzae Nontypable H. influenzae Other AOM pathogens (e.g., Moraxella catarrhalis, Group A streptococci, Staphylococcus aureus) 7,000 children enrolled in Panama To document the impact of PHiD-CV against: CAP cases with alveolar consolidation or pleural effusion on chest X-ray and positive respiratory viral test CAP cases with any abnormal chest X-ray with positive respiratory viral test B-CAP cases with positive respiratory viral test All children To document the impact of PHiD-CV against: Suspected CAP cases CAP cases with any abnormal chest X-ray Suspected CAP cases with CRP ≥ 40/80/120 µg/ml, regardless of chest X-ray reading CAP cases with either alveolar consolidation/pleural effusion on chest X-ray or with non-alveolar infiltrates and CRP ≥ 80/120 µg/ml All children To document the impact of PHiD-CV against: Bacteriologically culture-confirmed IPD cases caused by any of the ten pneumococcal VTs VT IPD identified through positive culture or nonculture pneumococcal diagnosis with additional nonculture VT serotyping Invasive disease caused by cross-reactive pneumococcal serotypes, other pneumococcal serotypes, and H. influenzae All children To document the impact of PHiD-CV on reducing nasopharyngeal carriage of S. pneumoniae (VTs and others) and H. influenzae Subset of 2,000 children in Panama To document the impact of PHiD-CV on antibiotic prescriptions Subset of 2,000 children in Panama (same as carriage subset) To assess the immune response to PHiD-CV Subset of 1,000 children in Argentina and Panama To assess the reactogenicity of PHiD-CV in terms of solicited general and local symptoms Subset of 1,000 children in Argentina and Panama (same as immunogenicity subset) To assess the safety of PHiD-CV in terms of unsolicited adverse events 7,000 children enrolled in Panama To assess the safety of PHiD-CV in terms of SAEs occurring during the entire study period All children AOM, acute otitis media; B-CAP, likely bacterial community-acquired pneumonia; C-AOM, clinically confirmed acute otitis media; CAP, community-acquired pneumonia; CRP, C-reactive protein; IPD, invasive pneumococcal disease; PHiD-CV, pneumococcal nontypable Haemophilus influenzae protein D conjugate vaccine; SAE, serious adverse event; VT, vaccine serotype. Healthy infants aged 6–16 wk were enrolled by pediatricians during well-baby clinic visits. Enrollment took place from 28 June 2007 until 30 December 2008. The inclusion and exclusion criteria for participation in the trial are listed in Table 2. Infants were randomized (1∶1 ratio) to receive either PHiD-CV (Synflorix) and diphtheria–tetanus–acellular pertussis–hepatitis B–inactivated poliovirus–Haemophilus influenzae type b vaccine (Infanrix hexa) (PHiD-CV group) or hepatitis B vaccine (Engerix-B) and diphtheria–tetanus–acellular pertussis–inactivated poliovirus–Haemophilus influenzae type b vaccine (DTPa-IPV/Hib; Infanrix-IPV/Hib) (control group) at approximately 2, 4, and 6 mo of age. This was followed by one dose of PHiD-CV or hepatitis A vaccine (Havrix), respectively, at 15–18 mo of age, both coadministered with DTPa-IPV/Hib (Figure 1). 10.1371/journal.pmed.1001657.g001 Figure 1 Vaccination schedule. The following vaccines were used: PHiD-CV, Synflorix; diphtheria–tetanus–acellular pertussis–hepatitis B–inactivated poliovirus–Haemophilus influenzae type b vaccine (DTPa-HBV-IPV/Hib), Infanrix hexa; DTPa-IPV/Hib, Infanrix-IPV/Hib; hepatitis B, Engerix-B; hepatitis A, Havrix (all by GlaxoSmithKline Vaccines). In addition to these blinded study vaccines, the following vaccines were administered or were recommended: measles–mumps–rubella vaccine at 12 mo of age, hepatitis B vaccination at birth, and hepatitis A vaccination at 12 and 18–21 mo of age, with the second dose given at least 28 days after the study vaccine booster dose. In Argentina, Neisseria meningitidis group C conjugate vaccine (NeisVac-C, Baxter International) was offered at 12 mo of age; in Colombia and Panama, varicella vaccine (Varilrix, GlaxoSmithKline Vaccines) was offered at 12 mo of age; in Colombia, two doses of oral rotavirus vaccine (Rotarix, GlaxoSmithKline Vaccines) were offered within the first 6 mo of life. 10.1371/journal.pmed.1001657.t002 Table 2 Inclusion and exclusion criteria. Inclusion Criteria Exclusion Criteria A male or female between, and including, 6 and 16 wk of age (between 42 and 118 d) at the time of the first vaccination. Infants born preterm (after a gestation period of 10 polymorphonuclear cells/mm3, >100 leukocytes/mm3, or 10–99 leukocytes/mm3 and either glucose 100 mg/dl), confirmed meningitis due to S. pneumoniae or H. influenzae (presence of S. pneumoniae or H. influenzae in CSF culture or purulent meningitis with blood culture positive for S. pneumoniae or H. influenzae), and probable bacterial meningitis due to S. pneumoniae or H. influenzae (purulent meningitis with negative result for CSF and blood culture but with antigen testing [Latex or BinaxNOW] positive for S. pneumoniae or H. influenzae in CSF, or meningitis with blood culture positive for S. pneumoniae or H. influenzae and with negative result for CSF culture but antigen testing [Latex or BinaxNOW] positive for S. pneumoniae or H. influenzae in CSF). Other clinical syndromes of focal invasive disease that were recorded included the following: bacteremic pneumonia (pneumonia confirmed by radiological evidence of any lung infiltrate and the presence of S. pneumoniae or H. influenzae identified by blood culture), empyema (isolation of S. pneumoniae or H. influenzae from pleural fluid obtained by thoracentesis, or pleural fluid with polymorphonuclear cells and isolation of S. pneumoniae or H. influenzae from blood culture), peritonitis (isolation of S. pneumoniae or H. influenzae from peritoneal fluid obtained by laparocentesis, paracentesis, or laparotomy), osteomyelitis (isolation of S. pneumoniae or H. influenzae from bone aspirate, or clinical symptoms and imaging compatible with osteomyelitis and isolation of S. pneumoniae or H. influenzae from blood culture), soft tissue infection (positive blood culture for S. pneumoniae or H. influenzae and clinical signs of cellulitis or abscesses, or isolation of S. pneumoniae or H. influenzae from abscess-aspirated material), septic arthritis (clinical and radiological presentation compatible with acute arthritis and positive blood culture for S. pneumoniae or H. influenzae, or isolation of S. pneumoniae or H. influenzae from material obtained by arthocentesis or arthrotomy), pericarditis (isolation of S. pneumoniae or H. influenzae from fluid obtained by pericardiocentesis or pericardiotomy, or presence of pleural friction rubs, muted heart sounds, or alterations indicative of a pericardial effusion in thoracic radiology, echocardiogram, or electrocardiogram and isolation of S. pneumoniae or H. influenzae from blood culture). Invasive disease without focal infection, such as bacteremia (positive blood culture for S. pneumoniae or H. influenzae indicating viable bacteria in blood without an obvious focus of infection), was also reported. In addition, generalized infections were reported as sepsis (defined as the presence of a systemic inflammatory response syndrome as a result of suspected or proven infection ), severe sepsis (sepsis plus cardiovascular organ dysfunction or acute respiratory distress syndrome or two or more other organ dysfunctions), or septic shock (sepsis and cardiovascular organ dysfunction). Identification of cases of invasive disease was based on clinical suspicion of the treating physician followed by a positive culture. Blood culture was recommended in all cases of febrile illness (axillary fever≥39.0°C or rectal temperature≥39.5°C) without overt focus of infection or with suspicion of meningitis, pneumonia, empyema, septic arthritis, peritonitis, osteomyelitis, pericarditis, or bacterial infection of soft tissue. In addition, a blood culture was to be seriously considered in children with a recent (within 24 h) history of febrile illness, especially if antipyretic treatment could have interfered with the temperature measurement at the time of physical examination of the child. In Panama, where dengue was endemic, it was recommended to perform dengue testing, according to local practices. In addition, study personnel at each site reviewed the records of the microbiology laboratories to identify study participants for whom S. pneumoniae or H. influenzae was isolated from blood cultures or other normally sterile body fluids and/or CSF, and in which testing for S. pneumoniae or H. influenzae antigens (BinaxNOW or Latex) was positive. Immunogenicity Assessment Immune responses against PHiD-CV antigens were planned to be assessed in approximately 1,000 infants (first 500 enrolled in selected centers in Argentina and Panama) and are reported from blood samples taken 1 mo after dose 3, just before booster, and 1 mo after booster vaccination. Serum anti-pneumococcal, serotype-specific IgG antibodies were measured using GlaxoSmithKline's 22F-inhibition ELISA with a cutoff value of 0.05 µg/ml, as described before ,. An antibody concentration of 0.2 µg/ml measured by this 22F-inhibition ELISA correspond to a concentration of 0.35 µg/ml in the WHO reference laboratory non-inhibition assay . Opsonophagocytic activity (OPA) was measured using a pneumococcal killing assay with HL60 phagocytes, as described previously , with a cutoff titer of eight . Antibody concentrations and OPA titers were also determined for cross-reactive serotypes 6A and 19A. Antibodies against nontypable H. influenzae protein D were measured by ELISA with an assay cutoff of 100 ELISA units (ELU)/ml. Serious Adverse Events SAEs are reported in this paper. Investigators were requested to report any SAE, defined as any medical event that resulted in death, was life-threatening, caused disability, or required hospitalization or prolongation of hospitalization, within 24 h of identification. Laboratory Analyses Laboratory procedures are described in Text S4. Roles of Investigators and Sponsor The study was sponsored by GlaxoSmithKline Biologicals, the vaccine developer and manufacturer. The data generated in the trial are subject to a confidentiality agreement between the investigators and sponsor that allowed the investigators full access to the study data at the end of the study and included an obligation for GlaxoSmithKline Biologicals to permit publication without excessive delay. Data Collection and Management The contract research organizations I3 and Progenitor (currently Encorium) were contracted to perform monitoring, Eurofins Medinet to perform biological sample processing, and S-Clinica to perform blinded statistical analyses. At each study center, data were remotely entered on electronic case report forms and transferred to GlaxoSmithKline for data management. All data cleaning processes were blinded to study group, and analyses before the end of the study were conducted by an external statistician (from S-Clinica, Belgium) who performed the analyses using SAS and the StatXact 8.0 procedure in SAS based on cleaned datasets and quality control programs provided by GlaxoSmithKline. Statistical Analysis The per-protocol cohort for efficacy comprised participants who had complied with vaccination and did not fulfill elimination criteria. In the per-protocol analysis, follow-up was censored at whatever came first: data lock point, time of last contact (in case of withdrawal), time of unblinding, time of booster vaccination (if the booster dose was not given correctly), or 18 mo of age (if the booster dose was not given by this age). The intent-to-treat analysis assessed efficacy against disease episodes from administration of the first vaccine dose. The incidence (percentage) of first disease episodes was calculated as the number of first episodes during the respective follow-up period divided by n, multiplied by 100. VE was estimated as one minus the hazard ratio and obtained, with its 95% CI, from a Cox regression model based on time to first disease episode. The Cox model was applied for VE and associated 95% CI and p-value computations when at least one event was observed in each group. Where zero cases were observed in one or more groups, the Poisson model, conditional to the number of cases, was applied without specific correction. Primary outcome The primary confirmatory objective was to demonstrate VE of PHiD-CV vaccination (three doses in children aged ≤18 mo or four doses if aged >18 mo) against first episodes of B-CAP occurring at least 2 wk after administration of the third dose of PHiD-CV in the per-protocol cohort for efficacy. The study was designed to detect a VE of 20% at study end, but allowed an earlier efficacy conclusion if a higher VE was achieved at the time of the interim analysis. As defined in the study protocol, the interim analysis was planned to occur at least 18 mo after study start, once 535 first B-CAP episodes were identified. A sample size of 24,000 healthy infants was planned to allow 21,600 evaluable children for the per-protocol CAP efficacy analysis. Power for B-CAP efficacy at the interim analysis and at the final analysis was calculated according to various VE assumptions and numbers of first B-CAP cases. With 535 first B-CAP cases and assuming a true VE of at least 25%, the study had 87.6% power to be conclusive at the interim analysis considering an adjusted one-sided alpha level of 1.75%. At the interim analysis, the primary objective was considered to be conclusive and final if the one-sided nominal p-value for the null hypothesis (B-CAP VE ≤ 0%) was lower than the adjusted (Pocock adjustment) one-sided significance level of 1.75% (p 0% with 80% and 90% power (non-adjusted one-sided test, nominal type I error of 2.5%) was calculated according to various true VE assumptions. If VE was 20%, 662 and 880 first C-AOM episodes were required to demonstrate positive VE with 80% and 90% power, respectively. If VE was 15%, 1,226 and 1,637 first C-AOM episodes were required for 80% and 90% power, respectively. Since the objective of the trial was to demonstrate VE in terms of a lower incidence of B-CAP or C-AOM in the vaccine group compared to the control group, only comparison in one direction was deemed necessary, hence the use of a one-sided statistical hypothesis linked to VE (with associated one-sided alpha level). The first secondary objective (to demonstrate VE of PHiD-CV against first C-AOM episodes) was considered significant if the primary objective was significant and the one-sided p-value for VE against C-AOM was <0.025, allowing for control of the primary confirmatory B-CAP objective and the sequential secondary confirmatory C-AOM objective. Other secondary outcomes and safety Other analyses, including VE against first IPD episodes, were descriptive, and therefore there was no adjustment for multiplicity for the associated informative p-values. Interpretations of nonoverlapping 95% CI boundaries should therefore be made with caution, as no predefined criteria were fixed, with no formal control of the alpha level. A two-sided Schoenfeld residual test for the null hypothesis that the VE is homogeneous over the age/time range was performed in order to assess the proportionality assumption associated with the Cox model. An analysis of VE against consolidated CAP was conducted among children who received the booster vaccine dose; VE was calculated for cases reported before the booster dose, given at 15–18 mo of age, and for cases reported after the booster dose. In another subanalysis, AOM severity was categorized according to the five symptoms on the Dagan scale. Cases were analyzed for which the presence or absence of each of the five symptoms had been recorded. In addition, a second complementary analysis of severity was conducted, considering cases that had at least one symptom score recorded, with the assumption that unrecorded symptoms were not observed. Immunogenicity analyses are reported for the per-protocol cohort for immunogenicity, defined as vaccinated children who met all eligibility criteria, complied with protocol-defined procedures, and had at least one antibody assay result available. An incorrect version of the informed consent form was used to obtain consent for children in the immunogenicity group in Panama (see Text S1). When the error was detected, in agreement with the independent ethics committee, 501 parents/guardians were recontacted to confirm their agreement to the use of the immunogenicity data of their child/ward. For 262 children, parents/guardians could not be contacted to provide consent or did not agree to the use of immunogenicity data. In addition, two children were excluded because the original informed consent forms were lost during the re-monitoring activities. Therefore, 264 children were excluded from the intent-to-treat cohort for immunogenicity (Figure S1). Seropositivity rates, ELISA geometric mean antibody concentrations, and geometric mean OPA titers were calculated with 95% CIs. Safety analyses (including mortality) were performed on the intent-to-treat cohort. Results Trial Profile The numbers of children enrolled, randomized, and eligible for primary end point analysis are shown in Figure 3, and end-of-study cohort numbers are provided in Figures 4 and S2. Demographic characteristics of the groups were well balanced (Table 3). As shown in the study timeline (Figure S3), the data lock point for the interim CAP efficacy analysis was August 31, 2010, and the study ended on July 28, 2011 (last child, last contact). 10.1371/journal.pmed.1001657.g003 Figure 3 Trial profile for children included in the analysis of the primary study end point. Elimination criteria shown for one reason only, although more than one reason for elimination could apply per child. aForbidden underlying medical conditions included, but were not limited to, major congenital defects, serious chronic illness, or confirmed or suspected immunosuppressive or immunodeficient conditions. 10.1371/journal.pmed.1001657.g004 Figure 4 Trial profile for children included in the end-of-study analysis of acute otitis media. Elimination criteria shown for one reason only, although more than one reason for elimination could apply per child. aForbidden underlying medical conditions included, but were not limited to, major congenital defects, serious chronic illness, or confirmed or suspected immunosuppressive or immunodeficient conditions. 10.1371/journal.pmed.1001657.t003 Table 3 Characteristics of the trial participants included in the analysis of the primary study end point (CAP efficacy cohort; interim analysis), CAP and IPD end-of-study analysis (CAP/IPD efficacy cohort), and AOM end-of-study analysis (AOM efficacy cohort). Efficacy Analysis Population Characteristic Category Intent-to-Treat Cohort Per-Protocol Cohort PHiD-CV Group Control Group PHiD-CV Group Control Group CAP interim n = 11,875 n = 11,863 n = 10,295 n = 10,201 Mean age ± SD At dose 1 (weeks) 9.2±1.9 9.2±1.9 9.2±1.9 9.2±1.9 At booster dose (months) 16.1±1.6 16.1±1.6 16.1±1.6 16.1±1.6 Sex, n (percent) Female 5,826 (49.1) 5,801 (48.9) 5,072 (49.3) 4,987 (48.9) Male 6,049 (50.9) 6,062 (51.1) 5,223 (50.7) 5,214 (51.1) Race a , n (percent) White 6,757 (56.9) 6,751 (56.9) 5,958 (57.9) 5,918 (58.0) Other or mixed race 5,118 (43.1) 5,112 (43.1) 4,337 (42.1) 4,283 (42.0) Follow-up time b Sum of time to first B-CAP (years) 25,516 25,329 19,513 19,260 CAP/IPD end of study n = 11,798 n = 11,799 n = 10,211 n = 10,140 Mean age ± SD At dose 1 (weeks) 9.2±1.9 9.2±1.9 9.2±1.9 9.2±1.9 At booster dose (months) 16.1±1.6 16.1±1.6 16.1±1.6 16.1±1.6 Sex, n (percent) Female 5,796 (49.1) 5,767 (48.9) 5,040 (49.4) 4,947 (48.8) Male 6,002 (50.9) 6,032 (51.1) 5,171 (50.6) 5,193 (51.2) Race a , n (percent) White 6,756 (57.3) 6,751 (57.2) 5,950 (58.3) 5,909 (58.3) Other or mixed race 5,042 (42.7) 5,048 (42.8) 4,261 (41.7) 4,231 (41.7) Follow-up time b Sum of time to first B-CAP (years) 31,480 31,265 24,821 24,545 Sum of time to first IPD (years) 32,117 32,023 25,244 25,043 AOM end of study c n = 3,602 n = 3,612 n = 3,010 n = 2,979 Mean age ± SD At dose 1 (weeks) 9.0±1.3 9.0±1.3 9.0±1.3 9.0±1.3 At booster dose (months) 15.8±1.8 15.8±2.0 15.7±1.7 15.7±1.8 Sex, n (percent) Female 1,762 (48.9) 1,775 (49.1) 1,478 (49.1) 1,464 (49.1) Male 1,840 (51.1) 1,837 (50.9) 1,532 (50.9) 1,515 (50.9) Race, n (percent) Other or mixed race 3,588 (99.6) 3,597 (99.6) 2,998 (99.6) 2,966 (99.6) White 14 (0.4) 15 (0.4) 12 (0.4) 13 (0.4) Follow-up time b Sum of time to first C-AOM (years) 9,018 8,835 6,720 6,605 a 59% of participants were recruited in Argentina (race predominantly white or with European heritage), and the remaining were recruited in Colombia and Panama (participants predominantly mixed race). b Follow-up time calculated as sum of follow-up periods of each child, expressed in years, censored at the first occurrence of a respective end point event. c Recruited in Panama only. SD, standard deviation. The sum of the follow-up period, censored at the first occurrence of a respective end point event in each group, is provided in Table 3. For the intent-to-treat analysis, mean duration of follow-up was 26 mo for the interim CAP analysis, 33 mo for the end-of-study CAP/IPD analysis, and 31 mo for the end-of-study AOM analysis. For the per-protocol analysis, mean duration of follow-up was 23, 30, and 28 mo, respectively. Disease Reporting and Vaccine Efficacy Primary outcome and other CAP outcomes Pneumonia rates for the two groups are presented in Table 4. Fewer than 1.5% of X-rays in the study were categorized as non-interpretable. At least one suspected CAP episode was reported in approximately 20% of all children. Around one-third of suspected CAP cases were radiologically confirmed (including perihilar infiltrates and other abnormal findings), of which approximately 30% were WHO-defined consolidated CAP and 40% were B-CAP (WHO-defined consolidated CAP, or nonconsolidated CAP with CRP ≥ 40 µg/ml). 10.1371/journal.pmed.1001657.t004 Table 4 Efficacy of PHiD-CV against first community-acquired pneumonia and invasive pneumococcal disease episodes. Cohort/Case Definition Intent-to-Treat Analysis Per-Protocol Analysis PHiD-CV Group Control Group VE, Percent (95% CI) PHiD-CV Group Control Group VE, Percent (95% CI) Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a CAP interim n = 11,875 n = 11,863 n = 10,295 n = 10,201 Consolidated CAP 223 1.9 (1.6, 2.1) 289 2.4 (2.2, 2.7) 23.4 (8.8, 35.7) 155 1.5 (1.3, 1.8) 206 2.0 (1.8, 2.3) 25.7 (8.4, 39.6) B-CAP 341 2.9 (2.6, 3.2) 414 3.5 (3.2, 3.8) 18.2 (5.5, 29.1) 240 2.3 (2.0, 2.6) 304 3.0 (2.7, 3.3) 22.0 (7.7, 34.2)b Radiologically confirmed CAP 854 7.2 (6.7, 7.7) 947 8.0 (7.5, 8.5) 10.5 (1.8, 18.4) 625 6.1 (5.6, 6.5) 711 7.0 (6.5, 7.5) 13.3 (3.4, 22.1) Suspected CAP 2,455 20.7 (19.9, 21.4) 2,616 22.1 (21.3, 22.8) 7.3 (2.1, 12.3) 1,916 18.6 (17.9, 19.4) 2,019 19.8 (19.0, 20.6) 6.7 (0.7, 12.3) CAP/IPD end of study n = 11,798 n = 11,799 n = 10,211 n = 10,140 CAP Consolidated CAP 251 2.1 (1.9, 2.4) 319 2.7 (2.4, 3.0) 21.8 (7.7, 33.7) 181 1.8 (1.5, 2.0) 231 2.3 (2.0, 2.6) 22.4 (5.7, 36.1) B-CAP 377 3.2 (2.9, 3.5) 450 3.8 (3.5, 4.2) 16.7 (4.5, 27.4) 275 2.7 (2.4, 3.0) 333 3.3 (2.9, 3.6) 18.2 (4.1, 30.3) Radiologically confirmed CAP 919 7.8 (7.3, 8.3) 1,015 8.6 (8.1, 9.1) 10.0 (1.7, 17.7) 681 6.7 (6.2, 7.2) 764 7.5 (7.0, 8.1) 11.9 (2.3, 20.5) Suspected CAP 2,667 22.6 (21.9, 23.4) 2,880 24.4 (23.6, 25.2) 8.7 (3.8, 13.4) 2,108 20.6 (19.9, 21.4) 2,237 22.1 (21.3, 22.9) 7.3 (1.6, 12.6) IPD c Culture-confirmed 7 0.1 (0.0, 0.1) 21 0.2 (0.1, 0.3) 66.7 (21.8, 85.9) 6 0.1 (0.0, 0.1) 17 0.2 (0.1, 0.3) 65.0 (11.1, 86.2) Vaccine serotypes 0 0.0 (0.0, 0.0) 18 0.1 (0.1, 0.2) 100 (77.3, 100) 0 0.0 (0.0, 0.0) 16 0.2 (0.1, 0.3) 100 (74.3, 100) Cross-reactive serotypesd 2 0.0 (0.0, 0.1) 1 0.0 (0.0, 0.0) −99.5 (−2,100.2, 81.9) 2 0.0 (0.0, 0.1) 1 0.0 (0.0, 0.1) −98.6 (−2,089.5, 82.0) Other serotypes 4 0.0 (0.0, 0.1) 2 0.0 (0.0, 0.1) −99.5 (−989.2, 63.5) 3 0.0 (0.0, 0.1) 0 0.0 (0.0, 0.0) NC Not serotyped 1 0.0 (0.0, 0.0) 0 0.0 (0.0, 0.0) NC 1 0.0 (0.0, 0.1) 0 0.0 (0.0, 0.0) NC VE estimated as one minus the hazard ratio and obtained, with its 95% CI, from a Cox regression model based on time to first episode when at least one event was observed in each group and conditional on number of cases when no case in at least one group (i.e., VE equal to zero or −infinite (VE, 1 − [x/0]). a Number of first episodes during the respective follow-up period divided by n, multiplied by 100. b Significant p-value (p = 0.002); one-sided p-value from Cox regression model to test null hypothesis VE ≤ 0% with one-sided alpha of 1.75%. c S. pneumoniae was isolated from all culture-confirmed invasive disease cases. d Pneumococcal serotype 6A, 9N, or 19A. NC, not calculable. VE against first B-CAP episodes at interim analysis was 22.0% (95% CI: 7.7%, 34.2%) and met prespecified criteria for conclusive results of the primary objective (p = 0.002). VE was also observed against the secondary CAP end points, with consistent results between the intent-to-treat and per-protocol analyses for the interim and end-of-study time points (Table 4). VE against WHO-defined consolidated CAP at end of study was 21.8% (95% CI: 7.7%, 33.7%) and 22.4% (95% CI: 5.7%, 36.1%) in the intent-to-treat and per-protocol analyses, respectively (Table 4). In the sensitivity analysis of the primary objective (see Methods), 144 children were excluded out of 20,496 children. The results of this analysis showed 239 first B-CAP cases in the PHiD-CV group and 304 in the control group. VE against first B-CAP episodes was 22.3% (95% CI: 7.9%, 34.4%; p = 0.002), which was consistent with the final per-protocol CAP efficacy analysis result. Schoenfeld residual analysis did not indicate heterogeneity of VE over the entire follow-up period. In line with this test, in the intent-to-treat analysis, VE against consolidated CAP was 21.6% (95% CI: −1.4%, 39.5%) among children aged less than 12 mo, 26.7% (95% CI: 2.3%, 45.0%) in children aged between 12 and 24 mo, and 21.3% (95% CI: −12.9%, 45.1%) in children aged between 24 and 36 mo. In the per-protocol analysis these percentages were 15.1% (95% CI: −22.5%, 41.2%), 31.8% (95% CI: 7.2%, 49.9%), and 20.7% (95% CI: −15.9%, 45.8%), respectively. VE against consolidated CAP was 21.7% (95% CI: −0.5%, 38.9%) prior to the booster dose and 26.3% (95% CI: 5.9%, 42.3%) after booster vaccination in the intent-to-treat analysis. In the per-protocol analysis, VE was 15.1% (95% CI: −15.6%, 37.6%) and 26.3% (95% CI: 4.4%, 43.2%) pre- and post-booster, respectively. First secondary outcome and other AOM outcomes At least one C-AOM episode was reported in approximately 7% of children (Table 5). VE against C-AOM was 16.1% (95% CI: −1.1%, 30.4%) in the per-protocol analysis (p = 0.032) and 19.0% (95% CI: 4.4%, 31.4%) in the intent-to-treat analysis (p = 0.007). VE against pneumococcal AOM and vaccine serotype AOM was 56.1% (95% CI: 13.4%, 77.8%) and 67.1% (95% CI: 17.0%, 86.9%), respectively, in the per-protocol analysis, with consistent results in the intent-to-treat analysis (Table 5). Among cases for whom the presence or absence of all five symptoms on the Dagan scale was recorded (approximately 80% of all C-AOM cases), 46% were categorized as mild, with few (9%) severe cases in both the intent-to-treat and per-protocol analyses. Among all C-AOM cases that had at least one symptom recorded as present or absent on the Dagan scale, 51% and 52% were mild in the intent-to-treat and per-protocol analyses, respectively, and 8% and 7%, respectively, were severe. Serotype 19F was the most common vaccine serotype isolated from middle ear fluid (Tables S3 and S4). Among the cases of AOM with H. influenzae, all but one were nontypable H. influenzae (Table 5). 10.1371/journal.pmed.1001657.t005 Table 5 Efficacy of PHiD-CV against first acute otitis media episodes (end-of-study analysis). Case Definition Intent-to-Treat Analysis Per-Protocol Analysis PHiD-CV Group (n = 3,602) Control Group (n = 3,612) VE, Percent (95% CI) PHiD-CV Group (n = 3,010) Control Group (n = 2,979) VE, Percent (95% CI) Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a Number of First Episodes Incidence, Percent (95% CI)a C-AOM 254 7.1 (6.2, 7.9) 308 8.5 (7.6, 9.5) 19.0 (4.4, 31.4) 204 6.8 (5.9, 7.7) 239 8.0 (7.1, 9.1) 16.1 (−1.1, 30.4) Culture-confirmed C-AOM 45 1.3 (0.9, 1.7) 67 1.9 (1.4, 2.4) 33.6 (3.2, 54.5) 32 1.1 (0.7, 1.5) 45 1.5 (1.1, 2.0) 29.9 (−10.4, 55.4) Pneumococcal C-AOM 17 0.5 (0.3, 0.8) 38 1.1 (0.8, 1.4) 55.7 (21.5, 75.0) 12 0.4 (0.2, 0.7) 27 0.9 (0.6, 1.3) 56.1 (13.4, 77.8) Vaccine serotype C-AOM 7 0.2 (0.1, 0.4) 23 0.6 (0.4, 1.0) 69.9 (29.8, 87.1) 6 0.2 (0.1, 0.4) 18 0.6 (0.4, 1.0) 67.1 (17.0, 86.9) Cross-reactive serotype AOMb 5 0.1 (0.1, 0.3) 7 0.2 (0.1, 0.4) 29.0 (−123.7, 77.5) 3 0.1 (0.0, 0.3) 4 0.1 (0.0, 0.3) 25.7 (−232.2, 83.4) Other serotype C-AOM 6 0.2 (0.1, 0.4) 7 0.2 (0.1, 0.4) 14.8 (−153.7, 71.4) 3 0.1 (0.0, 0.3) 4 0.1 (0.0, 0.3) 25.7 (−231.9, 83.4) H. influenzae C-AOM 20 0.6 (0.3, 0.9) 24 0.7 (0.4, 1.0) 17.3 (−49.8, 54.3) 12 0.4 (0.2, 0.7) 14 0.5 (0.3, 0.8) 15.0 (−83.8, 60.7) Nontypable H. influenzae C-AOM 19 0.5 (0.3, 0.8) 24 0.7 (0.4, 1.0) 21.5 (−43.4, 57.0) 12 0.4 (0.2, 0.7) 14 0.5 (0.3, 0.8) 15.0 (−83.8, 60.7) VE estimated as one minus the hazard ratio and obtained, with its 95% CI, from a Cox regression model based on time to first episode. a Number of first episodes during the respective follow-up period divided by n, multiplied by 100. b Pneumococcal serotype 6A, 18B, 19A, or 23A. IPD outcomes Most of the culture-confirmed IPD cases were caused by vaccine serotypes (Table 4). In the per-protocol analysis, VE was 100% (95% CI: 74.3%, 100%) against IPD caused by vaccine serotypes and 65.0% (95% CI: 11.1%, 86.2%) against any IPD. Pneumococcal serotype 14 was the most commonly isolated IPD serotype (Tables S5 and S6). Immunogenicity The per-protocol immunogenicity cohort included 334 and 331 children in the PHiD-CV and control groups for primary vaccination, respectively, and 232 and 215 children, respectively, for booster vaccination. As shown in Figure S1, the difference in the number of children between the primary and booster vaccination time points was mainly due to noncompliance with booster vaccination or blood sampling schedules (visits outside of scheduled intervals in the protocol, predominantly as a result of social disruptions caused by the H1N1 influenza pandemic). In the per-protocol immunogenicity cohort, for each of the ten vaccine pneumococcal serotypes, the percentage of children with antibody concentration ≥ 0.2 µg/ml after PHiD-CV primary vaccination was at least 93.1%, and the percentage with OPA titer ≥ 8 was at least 90.8% (Table S7). Robust immune responses were observed after booster vaccination (Table S7). For the cross-reactive serotypes 6A and 19A, at least 85.4% of children had an antibody concentration ≥ 0.2 µg/ml, and at least 80.5% had an OPA titer ≥ 8 after booster vaccination. Anti–protein D antibody geometric mean antibody concentrations following PHiD-CV vaccination were 2,455 ELU/ml (95% CI: 2,248, 2,681) after primary vaccination and 2,787 ELU/ml (95% CI: 2,436, 3,189) after booster vaccination, compared to 101 ELU/ml (95% CI: 91, 112) and 92 ELU/ml (95% CI: 83, 103), respectively, in the control group. The immunogenicity results of the intent-to-treat cohort for immunogenicity were consistent with those of the primary and booster per-protocol cohorts for immunogenicity (data not shown). Serious Adverse Events SAEs were reported for 21.5% (95% CI: 20.7%, 22.2%) of PHiD-CV recipients and 22.6% (95% CI: 21.9%, 23.4%) of children in the control group (Tables 6 and S8). One SAE, an apparent life-threatening event in the control group, was considered by the investigator to be causally related to vaccination. It occurred on the day of the second hepatitis B vaccine and DTPa-IPV/Hib vaccine dose administration, and resolved without sequelae. 10.1371/journal.pmed.1001657.t006 Table 6 Serious adverse events reported from study start and administration of the first vaccine dose up to study end in at least 1.0% of children (intent-to-treat cohort: all children). SAEs Reported in ≥1.0% of Children, n (Percent) PHiD-CV Group (n = 11,798) Control Group (n = 11,799) Gastroenteritis 553 (4.7) 497 (4.2) Pneumonia 478 (4.1) 557 (4.7) Bronchiolitis 473 (4.0) 518 (4.4) Dehydration 463 (3.9) 438 (3.7) Asthmatic crisis 192 (1.6) 210 (1.8) Bronchial obstruction 127 (1.1) 141 (1.2) Bronchitis 124 (1.1) 129 (1.1) Febrile convulsion 95 (0.8) 135 (1.1) Any SAE(s) 2,534 (21.5) 2,668 (22.6) A full listing of SAEs is provided in Table S8. Mortality Nineteen deaths among 11,798 children (0.16%; 95% CI: 0.10%, 0.25%) were reported in the PHiD-CV group, and 26 deaths among 11,799 children (0.22%; 95% CI: 0.14%, 0.32%) were reported in the control group. This suggested that PHiD-CV vaccination reduced all-cause mortality by 27.0% (95% CI: −31.8%, 59.6%), although this difference was not statistically significant. The majority of the reported fatal cases (nine in the PHiD-CV group, 16 in the control group) involved children aged less than 1 y. None of the deaths were considered by the investigator to be causally related to vaccination. Discussion Efficacy of PHiD-CV was shown against a variety of CAP, AOM, and IPD end points, with consistent results between the per-protocol and intent-to-treat analyses. In the interim analysis, VE against first episodes of B-CAP was 22.0% and therefore (as predefined in the protocol) was considered conclusive for the primary objective. In this analysis, efficacy against WHO-defined consolidated CAP was 25.7%. In the end-of-study per-protocol analyses, VE was 16.1% against C-AOM, 67.1% against vaccine pneumococcal serotype C-AOM, 100% against vaccine serotype IPD, and 65.0% against IPD of any serotype. This study thereby provides a comprehensive assessment of clinical protection of PHiD-CV against invasive and mucosal infections in Latin American children. In addition, despite the lower immunogenicity of PHiD-CV compared to 7vCRM, the antibody levels achieved were sufficient to provide a level of protection against WHO-defined CAP with alveolar consolidation (22–26%) that was consistent with the 20–37% efficacy reported in previous studies with other PCVs, including 7vCRM –. It should be noted that these studies were conducted with vaccine candidates of different serotype valencies and carrier proteins (seven- and nine-valent CRM197-conjugated vaccines, and 11-valent vaccine containing tetanus toxoid and diphtheria toxoid carrier proteins), and took place in a variety of geographic and socioeconomic settings. In our study, VE against WHO-defined consolidated CAP among children aged 12–24 mo and 24–36 mo remained above 20%, which was consistent with results obtained with 9vCRM in the Gambia  and South Africa  but differed from studies of 7vCRM in the US  and an 11-valent vaccine in the Philippines . In the latter two studies there was evidence of diminished efficacy already in the second year of life. The reasons for waning VE in some studies but not others are unknown but might be associated with inconsistent use of booster vaccination between studies, differences in strength of natural boosting between populations, or proportional increases in pneumonia caused by pathogens other than S. pneumoniae or by non-vaccine serotypes . It should be noted that, with the descriptive analyses of VE per age subgroup and pre-/post-booster dose, it is difficult to disentangle the effect of age from the effect of the booster dose. The number of AOM cases identified in Panama was lower than expected, and various factors might have contributed to this outcome. Case detection was especially low during the first 2 y of the study, and by the time it was enhanced, the age period of expected peak in AOM incidence  had likely passed for most participants. The requirement for evaluation by a pediatrician followed by further assessment by the ENT specialist and exclusion of children at high risk for pneumococcal infection could also have contributed. Some parents might have been reluctant to seek medical advice because of perceptions of AOM as a usually mild disease or might have obtained care from nearby pharmacies or clinics not involved in the study. Regardless of the factors responsible, the relatively low incidence of AOM may help with interpretation of the C-AOM protection results since, in the prespecified primary outcome (the per-protocol analysis), the lower limit of the 95% CI around the 16.1% point estimate was slightly below zero (95% CI: −1.1%, 30.4%). Analysis of the intent-to-treat cohort gave a similar point estimate (19.0%) but with a 95% CI lower limit above zero (95% CI: 4.4%, 31.4%). This suggests that the lack of statistical significance in the per-protocol analysis was due to a lack of cases rather than to true lack of efficacy. Nevertheless, we consider these results as highly relevant to other settings, as the AOM cases had both pediatrician and ENT confirmation, and around half of the C-AOM episodes were mild, with the remainder being moderate or severe, thus representing the broad spectrum of clinical AOM. In any case, differences in case definitions contribute little to differences in PCV efficacy estimates against AOM . Notably, VE against AOM episodes caused by vaccine serotypes in our study was similar to that observed in two European studies, the Finnish Otitis Media (FinOM) trial of two seven-valent PCVs (one investigational) and the Pneumococcal Otitis Efficacy Trial (POET) of the 11-valent predecessor to PHiD-CV conducted in the Czech Republic and Slovakia ,. We note that the most common serotypes causing AOM in both Latin America and Europe are members of serogroups 6, 14, 19, and 23 ,. Similar to the Pneumococcal Otitis Efficacy Trial and in contrast to the Finnish Otitis Media trial, VE against AOM caused by nontypable H. influenzae in COMPAS suggested a beneficial effect, although the result was not significant (95% CI included zero). It should be noted that the COMPAS study was not powered to assess the nontypable H. influenzae end point. Also similar to the Pneumococcal Otitis Efficacy Trial, there was no evidence for replacement of vaccine serotypes with non-vaccine pneumococcal serotypes or other otopathogens during 31 mo of intent-to-treat follow-up for AOM analysis; in contrast, evidence of replacement disease was found early in the follow-up of the Finnish Otitis Media trial with 7vCRM, as of 2 mo of age . Overall, as AOM is a common disease , the 16%–19% VE observed against C-AOM represents substantial reductions in AOM case numbers, from both public health and health care cost perspectives. In previous double-blind randomized trials of PCVs (all seven-valent) administered in infancy, VE estimates against clinical AOM were 0%–7%  and appeared unaffected by population variability . With the 11-valent protein D conjugate vaccine, VE against clinical AOM was 34% . However, comparisons of C-AOM results among trials are complicated by differences in care-seeking and case detection methods, as well as disease burden. All invasive disease cases were first episodes and pneumococcal, and, as anticipated from the regional epidemiology of pneumococcal serotypes , most were caused by vaccine serotypes. The results presented here are similar to those obtained in a recently published cluster-randomized double-blind trial conducted in a different socioeconomic setting, Finland. There, PHiD-CV was highly effective against IPD when given either in a 3+1 or 2+1 vaccination schedule to infants or as a two-dose catch-up to children in the second year of life . As in COMPAS, following the 3+1 PHiD-CV schedule, VE against IPD caused by any vaccine serotype was 100% in the Finnish study. This point estimate is similar to or slightly higher than those observed in studies of other PCVs against IPD ,,,. One limitation of this study, as noted earlier, lies in the low number of reported AOM cases, despite efforts to improve surveillance during the study. Another limitation was the study setting being in mostly urban areas, potentially affecting its generalizability to rural populations. This setting was chosen to aid recruitment and to improve the likelihood of capturing all pneumonia cases, as that might have been difficult in rural areas without easy access to health care centers and pediatric hospitals. An additional study limitation was that approximately 14% of participants were excluded from the per-protocol efficacy analyses, mainly because of early withdrawal and noncompliance with the primary vaccination schedule in both groups. This was predominantly as a result of adverse media coverage of the study in 2007/2008 linked to unfounded rumors of a causal relationship between PHiD-CV vaccination and infant mortality . Subsequent investigations of the blinded safety results were conducted by the IDMC and local regulatory authorities. Although these concluded that the study could continue as planned, many parents became unwilling to allow their children to continue to participate. Nonetheless, these exclusions do not appear to have affected the robustness of the study, as VE results were consistent between the intent-to-treat and per-protocol analyses. Importantly, a 27% reduction (not statistically significant) in overall mortality in PHiD-CV recipients was observed, which was most evident in children in the first year of life, when the risk of severe pneumococcal disease is highest . Reductions in all-cause mortality reported in other pneumococcal conjugate VE studies in young children, both with a nine-valent vaccine, were 14%–16% in the Gambia  and 5% in South Africa . In conclusion, COMPAS is, to our knowledge, the first double-blind randomized controlled trial to assess pneumococcal conjugate VE in Latin America, a region with an intermediate pneumococcal disease burden. The results of this study demonstrate the efficacy of PHiD-CV and document the magnitude of its impact against CAP and AOM, which are mucosal diseases of public health importance, commonly encountered in young children in clinical practice. Supporting Information Figure S1 Number of children in the primary or booster vaccination intent-to-treat and per-protocol cohorts for immunogenicity. (DOCX) Click here for additional data file. Figure S2 Trial profile for children included in the end-of-study analyses of community-acquired pneumonia and invasive pneumococcal disease. (DOCX) Click here for additional data file. Figure S3 Clinical Otitis Media and Pneumonia Study timeline. (DOCX) Click here for additional data file. Table S1 National public health authorities and ethical review committees. (DOCX) Click here for additional data file. Table S2 Children at high risk of invasive pneumococcal infection. (DOCX) Click here for additional data file. Table S3 Occurrence of first clinically confirmed acute otitis media episodes (per-protocol cohort for acute otitis media vaccine efficacy analysis). (DOCX) Click here for additional data file. Table S4 Occurrence of first clinically confirmed acute otitis media episodes (intent-to-treat cohort for acute otitis media vaccine efficacy analysis). (DOCX) Click here for additional data file. Table S5 Occurrence of invasive pneumococcal infection episodes (per-protocol cohort for community-acquired pneumonia/invasive pneumococcal infection vaccine efficacy analysis). (DOCX) Click here for additional data file. Table S6 Occurrence of invasive pneumococcal infection episodes (intent-to-treat cohort for community-acquired pneumonia/invasive pneumococcal infection vaccine efficacy analysis). (DOCX) Click here for additional data file. Table S7 Pneumococcal antibody concentration and opsonophagocytic activity in PHiD-CV group (per-protocol immunogenicity cohort for primary or booster vaccination). (DOCX) Click here for additional data file. Table S8 Serious adverse events reported from study start and administration of the first vaccine dose up to study end (intent-to-treat cohort: all children). (DOCX) Click here for additional data file. Text S1 Ethical considerations and informed consent. (DOCX) Click here for additional data file. Text S2 Trial protocol. (PDF) Click here for additional data file. Text S3 Major protocol changes. (DOCX) Click here for additional data file. Text S4 Laboratory analyses. (DOCX) Click here for additional data file. Text S5 CONSORT checklist. (DOCX) Click here for additional data file.