Effectiveness of 13-pneumococcal conjugate vaccine (PCV13) against invasive pneumococcal disease in children in the Dominican Republic

The Dominican Republic is a middle-income country in the Caribbean with a population of 10.3 million, and a birth rate of 21 births per 1000 people (2013) [13]. The study was conducted at the national reference pediatric hospital in the Dominican Republic, a pediatric referral tertiary care hospital located in the capital city Santo Domingo. IPD surveillance began at the hospital in 2000 with a focus on pediatric meningitis and pneumonia with effusion.

For each case-patient, we aimed to enroll at least four controls matched on age and neighborhood of residence. Children were eligible to be enrolled as controls if they were a resident of the case-patients’s neighborhood for at least one month prior to the matched case-patient’s hospital admission date, age-eligible to have received at least one dose of PCV13 by the time of hospital admission for the case-patient, not hospitalized during the prior month for pneumonia, meningitis or sepsis, and not previously enrolled in the study. Using the case-patient’s residence as a starting point, controls were sought house-to-house using a random walk approach commonly used for vaccine coverage surveys [14]. Up to three attempts were made to reach households targeted for enrollment. A registry was kept of all households approached.

Prior to enrollment, written informed consent was obtained from the parent or guardian of all potential study participants. Socio-demographic, risk factor, and clinical data were collected from study participants by interviews with the parents or guardians using a standardized questionnaire. Medical records were also reviewed to abstract case-patient data, including anthropometric data. For control anthropometric data, study staff measured the height of the children and recorded the weight based on records supplied by the parents or guardians. Written evidence of immunization history was actively sought and examined. The primary source of vaccine history was the child immunization card; if that was not available, parents’ verbal vaccine information was confirmed through vaccination records at immunization clinics. Those with no written documentation of any vaccine receipt (e.g. no immunization card available and vaccine history could not be confirmed in vaccination records at immunization clinics) were excluded from the analysis unless the parent or guardian reported that their child had not received any vaccinations beyond doses given at birth (such children were included and considered to have received zero doses of PCV13).

Children were included as controls if the age-matching criteria were confirmed and their birthdate was within 51.4 days or 81.8 days of the matched case-patients’ birthdate for case-patients aged less than and greater than or equal to six months, respectively. Criteria were originally set at 1 month (30.4 days) and 2 months (60.8 days), respectively, but were revised to add 21 days to for each group to address challenges faced by field research staff with finding controls meeting these criteria, especially in neighborhoods with low population density. We defined a reference date for the controls that matched the time of hospital admission for the corresponding case-patient.

Specimens were collected from patients for diagnosis as per routine medical care. Invasive specimens were plated on 5% sheep blood agar and incubated overnight in CO2 at 37 degrees Celsius. Suspect colonies were identified as S. pneumoniae by susceptibility to optochin and bile solubility tests. Suspect alpha-hemolytic colonies were identified as S. pneumoniae by susceptibility to optochin and/or bile solubility tests. Cerebrospinal fluid (CSF), pleural, and joint fluids were tested for presence of pneumococcal antigen by immune-chromatographic test BinaxNow (i.e. Binax) and for presence of the pneumococcal lytA gene by quantitative PCR [15]. Pneumococcal isolates were serotyped by Quellung, while lytA positive specimens that were culture-negative were serotyped by quantitative multiplex PCR targeting 37 serotypes (including VT serotypes) [16, 17]. To obtain deoxyribonucleic acid (DNA) extracts for polymerase chain reaction (PCR) reactions, 200–400 μL of the clinical specimens were transferred into 1.5 ml cryotubes containing 300 μL of buffer#4 (Roche isolation kit III) and transferred to MagNA Pure Compact instrument using the external lysis protocol according to manufacture instructions. When samples were viscous or containing clots or fragments, 200–300 μl of ATL Buffer (Qiagem) and 30 min incubation at 56 °C treatment was added to dissolve the fragments. DNA extracts were eluted to 100 μL and stored at -20 °C until PCR testing was performed using Quanta Biosciences PerfeCTa® qPCR ToughMix®, Low ROX™, for the lytA assay and PerfeCTa Multiplex qPCR ToughMix®, Low ROX™, for the multiplex serotyping assays [15].

Vaccine doses were considered to be valid only if given on or after 6 weeks of age, at least 21 days after the previous PCV dose, and at least 14 days before hospital admission for case-patients (or the reference date for controls). An up-to-date PCV13 variable was generated if the number of valid doses was at least one dose for children aged ≤4 months and 13 days, at least two doses for children aged between 4 months plus 14 days and ≤ 12 months plus 13 days, and at least three doses for children ≥12 months plus 14 days at the time of hospital admission or reference day. Serotyping by PCR does not distinguish between the serotypes within the following serogroups: 6A/6B, 7F/7A, and 18C/18F/18B/18A; thus, we classified these according to their serogroup in the analysis. Low weight-for-age Z score was defined as <− 2 using World Health Organization (WHO) growth curves [18]. We used multivariable conditional logistic regression to calculate the matched odds ratio of PCV13 vaccination. Matched pairs with at least one discordant set (i.e. different vaccination status between case-patients and controls) contributed to the models. Vaccine effectiveness (VE) was calculated as (1- adjusted matched odds ratio) X 100%. We assessed confounding by adding variables to the model one-by-one. Any variables which changed the VE estimate by more than 10% were included. Serotype-specific VE models were explored where there was sufficient power. A sensitivity analysis was also performed by excluding case-patients that were positive by PCR for serogroups that included a non-pneumococcal vaccine-type serotype. Data were analyzed using SAS software (version 9.3; SAS Institute).

The protocol was approved by the ethics committees of Hospital Infantil Robert Reid Cabral and the Pan American Health Organization (PAHO). The Centers for Disease Control and Prevention (CDC) determined the protocol to be public health non-research.