Estimating Vaccine Effectiveness Against Hospitalized Influenza During Pregnancy: Multicountry Protocol for a Retrospective Cohort Study

Background Although pregnant women are believed to have elevated risks of severe influenza infection and are targeted for influenza vaccination, no study to date has examined influenza vaccine effectiveness (IVE) against laboratory-confirmed influenza-associated hospitalizations during pregnancy, primarily because this outcome poses many methodological challenges. Objective The Pregnancy Influenza Vaccine Effectiveness Network (PREVENT) was formed in 2016 as an international collaboration with the Centers for Disease Control and Prevention; Abt Associates; and study sites in Australia, Canada, Israel, and the United States. The primary goal of this collaboration is to estimate IVE in preventing acute respiratory or febrile illness (ARFI) hospitalizations associated with laboratory-confirmed influenza virus infection during pregnancy. Secondary aims include (1) describing the incidence, clinical course, and severity of influenza-associated ARFI hospitalization during pregnancy; (2) comparing the characteristics of ARFI-hospitalized pregnant women who were tested for influenza with those who were not tested; (3) describing influenza vaccination coverage in pregnant women; and (4) comparing birth outcomes among women with laboratory-confirmed influenza-associated hospitalization versus other noninfluenza ARFI hospitalizations. Methods For an initial assessment of IVE, sites identified a retrospective cohort of pregnant women aged from 18 to 50 years whose pregnancies overlapped with local influenza seasons from 2010 to 2016. Pregnancies were defined as those that ended in a live birth or stillbirth of at least 20 weeks gestation. The analytic sample for the primary IVE analysis was restricted to pregnant women who were hospitalized for ARFI during site-specific influenza seasons and clinically tested for influenza virus infection using real-time reverse transcription polymerase chain reaction. Results We identified approximately 2 million women whose pregnancies overlapped with influenza seasons; 550,344 had at least one hospitalization during this time. After restricting to women who were hospitalized for ARFI and tested for influenza, the IVE analytic sample included 1005 women. Conclusions In addition to addressing the primary question about the effectiveness of influenza vaccination, PREVENT data will address other important knowledge gaps including understanding the incidence, clinical course, and severity of influenza-related hospitalizations during pregnancy. The data infrastructure and international partnerships created for these analyses may be useful and informative for future influenza studies. International Registered Report Identifier (IRRID) DERR1-10.2196/11333


I.
Specification of records included in dataset II.
Setting and Season Information III.
Acute Respiratory or Febrile Illness (ARI/F) Summary Variables from Discharge Codes (DC) IV.
Acute Respiratory or Febrile Illness (ARI/F) Specific Diagnosis Codes V.
Maternal Characteristics VI.
High Risk Medical Conditions in Discharge Codes VII.
Pregnancy History and at Admission VIII.
Hospital Event Description IX.
Illness Onset X.
Virus Testing XI.
Birth Outcomes XIII.
Analysis Specification

I. Specification of Records Included in Dataset
One record per hospitalization. Each record meets all of the following criteria i : a. Female age 18-50 at time of inpatient admission. Exclude any women that were less than 18 years or over 50 years at any time during the pre-defined flu season. b. Pregnant at time of admission. c. Date of admission during flu season 2010/11 thru 2015/16 d. Eventual pregnancy outcome of live birth or stillbirth at >= 20 weeks gestational age. e. Discharge diagnosis of ARI/Febrile illness OR any influenza testing performed during hospitalization or in 3 days prior to admission.
Data may include more than one hospitalization per woman across seasons; more than one hospitalization within a season will be addressed as follows: ▌pg. 1  A second hospitalization (with an ARI/F code) admission that occurs ≥15 days from discharge date of a previous hospitalization with ARI/F will be considered a new hospitalization, adding to the denominator of observed hospital events for that season.  A second hospitalization (with an ARI/F code) will be considered part of the same hospital event if the second admission occurs ≤14 days of the prior hospitalization discharge date. How information from both hospitalizations will be utilized may vary by manuscript project. For the VE project: o Any influenza testing that occurred during either hospitalization or within 3 days prior to either admission will be examined. RT-PCR influenza positive result from any of these tests will indicated an influenza positive hospitalization. o Reference both sets of discharge codes when creating derived variables. For example, ICD-10 code of J10 listed in the second set of discharge codes but not the first would still result in a "1" for ARIF_SUM_PI.
Different manuscript projects will utilize these events differently. For example, the vaccine effectiveness (VE) paper will exclude hospitalizations if they do not involve an ARI/F code even if influenza testing was performed.
The variables in Section XIII: Vaccination Patterns apply to the following population: One record per pregnancy per influenza season (Unique PID-SEASON_YEAR combo). Each record meets all of the following criteria: a) Female age 18-50 years as of pregnancy start date. Exclude any women that were less than 18 years or over 50 years as of pregnancy start date. b) Pregnancy overlapped at least one of the pre-defined, site-specific influenza seasons. c) Pregnancy reached >=20 weeks gestation Notes: If a pregnancy overlaps two flu seasons, two records are generated, one for each season. If a woman has two pregnancies in a single influenza season, use the first pregnancy.