Health outcomes of young children born to mothers who received the pandemic H1N1 vaccine in 2009 during pregnancy: retrospective cohort study

Study Design, Data Sources and Population Study

We performed a retrospective cohort study based on the population of infants born to residents of Ontario, Canada, from November 2, 2009 to October 31, 2010. We used the Better Outcomes Registry & Network (BORN) birth registry. from Ontario (https://www.bornontario.ca). / en / about-born /) to identify the study population. This province-wide registry contains maternal and newborn records for all hospital births of at least 500 g or at least 20 weeks of gestational age. In addition to routine sociodemographic and clinical data, information on the receipt of the monovalent influenza vaccine pH1N1 2009 during pregnancy was collected as part of the enhanced surveillance established during the 2009 influenza pandemic. We have linked this one year birth cohort to administrative health databases to determine outcomes for child health over a five year follow-up period. All databases were linked with unique coded identifiers analyzed by ICES (https://www.ices.on.ca/). The administrative databases included the Canadian Institute for Health Information (Hospital Admissions) patient discharge database and the National Health Information System. outpatient care (emergency department visits), each containing clinical diagnoses made during medical consultations, coded with the help of the Canadian adaptation of the International Classification of Diseases. 10th Revision (ICD-10-CA), as well as the Ontario Asthma dataset, which is derived from administrative databases. We also linked to the Ontario Cancer Registry to identify pediatric cancer cases and the registered people database to determine each child's follow-up time and determine infant mortality. A more detailed description of the data sources and the coupling methodology is provided in the additional methods 1.

We excluded infants whose birth registration could not be linked to administrative databases, those born to women who were not continuously eligible for health care in Ontario during pregnancy, those whose records posed data quality issues (eg, duplicate, invalid birth date), and those not eligible for provincial government-funded health care at birth. We also excluded infants whose records lacked information on the 2009 pH1N1 influenza vaccination during pregnancy or who died on the day of their birth.

Measure of exposure and results

The exposure in question was received during pregnancy by the 2009 influenza monovalent influenza vaccine pH1N1, based on the database-specific codes in the birth registry. We classified infants born to mothers with documented pH1N1 2009 immunization during pregnancy among those exposed to vaccination and those whose mothers were not immunized against pandemic influenza 2009 during pregnancy as unexposed . The pandemic influenza vaccination campaign in Ontario began on October 26, 2009. In Canada, two pandemic vaccines have been produced (both by GlaxoSmithKline Biologicals): an unadjuvanted pH1N1 flu vaccine specifically for pregnant women , and an adduct product AS03 (Arepanrix) produced for the general population. The second was not contraindicated during pregnancy if the unadjuvanted product was not available and the risk of influenza was judged to be high5253.

A consensual list of standardized case definitions has been developed to monitor obstetric and neonatal outcomes following vaccination during pregnancy.545556 In the absence of such indications for future outcomes in pediatric health and limited research on this issue. subject, we have pre-specified three groups of infant morbidity. results for our evaluation. First, we focused primarily on immune outcomes (infectious and atopic diseases), as the developing fetal immune system is thought to be sensitive to influences such as maternal vaccination.575859 To assess safety, we included two morbidity outcomes not related to the immune system: have been used in other safety studies in pregnant women (neoplasms, sensory disturbances) 60 and in two cases of non-specific morbidity (use of health care services). emergency and hospital services, complex chronic conditions in pediatrics). Wherever possible, we used standardized clinical registries and validated algorithms to identify the results. In the absence of any established registry or algorithm available, we measured results using diagnostic codes of emergency visits and admissions to the hospital, but not from outpatient primary care visits, to limit the analysis to cases with better measurement in the available databases, as well as more serious clinical implications. Following the development of our original study protocol, we included infant mortality up to the age of 5 as an additional outcome.

Infectious findings included upper and lower respiratory tract infections, gastrointestinal infections, otitis media and a composite of these four categories of infections. Sensory disorders included loss of vision and hearing. We searched for ICD-10-CA diagnostic codes for each of these results in the primary or secondary diagnostic code field positions of the hospital admission databases and medical services. ;emergency. Asthma diagnoses were derived from the Ontario Asthma Data Set, which uses a validated algorithm (sensitivity 89%, specificity 72%) to identify cases of asthma. Asthma from administrative health databases.61 Children in our cohort who were in the Ontario asthma dataset but were coded for an asthma diagnosis before the start of the study. 6-month-old age was not classified as an asthma case unless an asthma diagnosis code could be found in the admissions data bases. hospital or emergency department later during follow-up.62 Confirmed pediatric cancer diagnoses came from the Ontario Cancer Registry. . We have modified an existing algorithm to identify children with a complex chronic disease (expected to last more than 12 months and require specialized care, possibly including hospitalization in a tertiary care facility) .63 We Verified infant mortality from the database of registered persons. See Supplementary Tables A and B for a list of diagnostic codes used to identify all morbidity outcomes.

statistical analyzes

Characteristics of the study population were described using frequencies for categorical variables and medians (interquartile intervals) for continuous variables. We used standardized differences to evaluate the baseline covariate equilibrium between the two groups of exposure, an absolute standardized difference of less than 0.10 being considered as an indication of a well-balanced covariate64.

We used weightings derived from propensity scores to adjust the confounders in our study. We developed a logistic regression model to compute a propensity score for each infant, representing the predicted probability of vaccination against pandemic H1N1 influenza in 2009 during pregnancy. Before running the propensity score models, we used multiple imputation to correct missing values ​​for covariates: 6.9% of the records contained missing information for one or more covariates that we had Intended to include in the propensity score. the percentage of missing data for each of the individual variables was below 1% for most (rural residence, public health unit area, parity, fifth of neighborhood income) and was highest for maternal smoking during pregnancy ( 3.9%). We included the following preselected birth register covariates in propensity score models, which were developed using each of the 10 sets of multiple imputation data: maternal age, parity, maternal smoking , design season, antenatal care provider, pre-existing for maternal medical comorbidity, obstetric complications, antenatal steroid use, multifetal pregnancy, fifth of neighborhood income, rural residence and area of ​​public health units. Subsequently, we developed an inverse likelihood of treatment weights (IPTW), in which the value for each infant exposed to the vaccine was the inverse of the propensity score and the value for each unexposed infant was the highest. inverse of 1 minus the propensity score65. weighted, we normalized the entire population under study by multiplying it by the marginal propensity score.

Follow-up began at the date of birth and continued until the child became ineligible for health care in Ontario (due to emigration or death) or has reached the age of 5 years. However, for the results at the end of the term (asthma, neoplasms, sensory disturbances, mortality), the end of the follow-up corresponds to the date of the event for the persons presenting the result. We used Cox proportional hazards models (asthma, neoplasms, sensory disturbances, mortality), negative binomial models (infectious disease outcomes, use of urgent and inpatient health services) and binomial log models (conditions complex chronic in children) to estimate unadjusted and adjusted risk ratios, incidence rates and risk ratios, respectively, with 95% confidence intervals. To generate adjusted results, we ran weighted models for each result using the stabilized IPTWs generated from each of the 10 multiple imputation datasets. We then statistically combined the resulting β parameters and standard errors to produce a single adjusted point estimate and a 95% confidence interval. For Cox models, we found that the proportional hazards assumption for the exposure variable was met based on the examination of Schoenfeld residual plots and Wald tests for the ## EQU1 ## 39, interaction between exposure status and time.

In the sensitivity analyzes, we examined the results for which the 95% confidence interval around the adjusted point excluded the null value. First, we have limited the study population to infants with at least two good baby visits or primary vaccination visits during the first year of life to make sure that the child had access to the health care system (see Additional Methods 2). We also repeated our main analyzes by further adjusting the number of ambulatory maternal consultations in the six months preceding the indexed pregnancy and the number of non-obstetric hospital admissions in the two years preceding the start of the indexed pregnancy in order to take into account account the different needs of health care. or access. With regard to asthma, we also adjusted the status of the mother's asthma. We recalculated the confidence intervals after the application of a Bonferroni correction to account for several comparisons (n ​​= 10 predefined morbidity results) and examined a negative control result (baseline rate). all-cause injuries) to assess any rival explanations66. additional analyzes to characterize the effect of excluding records with missing information on the pH1N1 influenza vaccination. We used SAS version 9.4 for all analyzes.

Patient participation

This research was conducted without the participation of the patient. Patients were not consulted to develop the research question, nor were they involved in identifying the study design or results. We did not invite any patients to participate in the interpretation of the results, nor in the drafting or revision of this document. It is not intended to directly involve patients in the dissemination of these research results.