Safety Study of GBS Following Menactra Meningococcal Vaccination

Through a partnership that uses the data and other resources of three national and two regional health plans and insurers, we are conducting a multi-site retrospective study of the relationship between immunization with tetravalent meningococcal conjugate vaccine (MCV4) and Guillain-Barré syndrome (GBS) in adolescents (ages 11-21 years of age) over the 42-month period of March 1, 2005 to August 31, 2008. This collaboration has several features of a distributed data network, and so provides an instructive example of the feasibility of large scale population-based post-market evidence development.

Participants include five research organizations that are either part of, or closely affiliated with, a health plan or insurer: HealthCore Inc. (affiliated with WellPoint), Highmark Blue Cross Blue Shield of Pennsylvania, i3 Drug Safety, Aetna, Inc., and Kaiser Permanente Center for Health Research of Hawaii. There is a coordinating center at the Harvard Medical School Department of Ambulatory Care and Prevention at Harvard Pilgrim Health Care.

The organizations provide coverage for approximately 80 million individuals (all ages), of whom the information for approximately 50 million will be available for analysis. Each organization provides scientific, informatics, and management expertise regarding its own population and data, and the overall design and conduct of the study.

This retrospective study uses a hybrid design with three analytic components: 1) characterization of the entire cohort with regard to demographics, eligible time under observation, immunization status, incidence rates of GBS during various categories of person time (never immunized, before immunization, during the 42 days after immunization, etc.), and comorbidities and other covariates, 2) a nested case-control study assessing risk associated with immunization, and 3) a self-controlled case series assessing immunization associated risk.

Cohort analysis

Sites will use programs distributed by the coordinating center to compute incidence rates of GBS according to age, sex, season, and MCV4 vaccination status among the full base population (health plan enrollees who meet age and cohort eligibility criteria). Additionally, vaccination rates for MCV4 according to age, sex, and calendar year will be computed among all cohort members. Incidence rates of GBS provided by these analyses will be used to calculate the attributable risk of GBS following MCV4 vaccination should a significant relationship be identified.

Nested case control analysis

A nested case-control analysis utilizing pooled analytic datasets from all sites will be used to estimate the risk of GBS during the 42 days after MCV4 vaccination relative to the risk of GBS among vaccines and non-vaccinees at other times (the study's primary objective), adjusted for age, sex, and other vaccinations among 11-<19 year olds.

All potential cases of GBS among the cohort members will be identified, and a sample of 100 non-cases from the same risk set as each case will be randomly selected and matched on age, sex, and state of residence. Their vaccination histories and other covariates will be included in analytic datasets to be transferred to the coordinating center for pooled multivariate analysis. Covariates will be estimated as of an index date assigned within each case-control set which is the date of GBS onset for the associated case.

The relative risk of GBS associated with MCV4 vaccination will be estimated using conditional logistic regression analysis in SAS.

Variables will include:

• Dependent variable: GBS case status
• Main independent variable: Analysis A: MCV4 vaccination in 1-42 days prior to index date; Analysis B: MCV4 vaccination in 1-7, 8-14, 15-28, 29-42 days prior to index date
• Matching factors: Age; sex; study site; state of residence
• Other covariates: Exposure to Tdap, Td, Hepatitis B, HPV, tetanus, and influenza vaccine during the 42 days prior to the GBS onset date; exposure to MPSV4 vaccine during the 42 days prior to the GBS onset date(separate model from model with MCV4 exposure)

We also will assess the potential effect modification between the different vaccines, and according to age, season, and geographic region, through stratified analysis and the testing of interaction terms within the conditional logistic regression model.

Self-Control Analysis

An additional case-series analysis (to be referred to as "Analysis C") will be performed among the subset of adjudicated GBS cases with prior MCV4 vaccination. Date of MCV4 vaccination will serve as the index date for each patient. The relative incidence rate of GBS during a "control" period of 43 or more days following vaccination will be compared to the incidence during the "exposed" risk window of 1-42 days following vaccination using log linear modeling with terms for the exposed and control periods.

Variables will include:

• Dependent variable: Time to GBS
• Main independent variable: Analysis C: Time period 1-42 days post vaccination compared with 43 or more days post vaccination
• Other covariates (to be included as sample size permits): Time periods corresponding to 1-42 days following each of: Tdap, Td, Hepatitis B, HPV, tetanus, and influenza vaccinations

We also will assess the potential effect modification between the different vaccines, and according to age, season, geographic region, and race/ethnicity, through stratified analysis and the testing of interaction terms within the log linear model, as sample size permits.

Analysis C will be performed once at the end of the study period.

GBS Case Descriptive Analysis

Several additional secondary objectives will be addressed from evaluation of GBS cases based on review of abstracted information from their medical charts, and administrative claims data.

Cases of GBS identified among cohort members 19-<22 years old will be identified from administrative claims data, and adjudicated through review of abstracted medical record information in a manner equivalent to that for cohort members 11-<19 years old. These cases will be quantified, and their clinical presentation described and compared with that of cases among 11-<19 year olds, with regard to at least the following factors: proportion of cases classified as definite GBS cases, severity of illness as measured by mortality, hospitalization for GBS, total duration of hospitalizations in days, mechanical ventilation required, total number of days of mechanical ventilation, intravenous immunoglobulin therapy, plasmapheresis, and immunosuppressive therapy. Similar comparisons will be made regarding the clinical presentation of GBS among MCV4 vaccinees compared to non-vaccinees.

The number of GBS cases with history of each of the other vaccines included in the study in the 42 days prior to GBS onset will also be described.

A line list will be created for all potential cases of GBS. For adjudicated cases of GBS, the listing will indicate the final case status, elements of the case definition that were and were not met, immunization status (MCV4, MPSV4, Tdap, Td, HepB, HPV, tetanus, and influenza) and indicators of infection or febrile illness in the prior 42 days, gender, age, month and year of GBS onset, and a brief description of disease severity. Additional line lists will be created for censored cases (adjudicated non-cases) that include the reason for censoring, and for potential cases for which charts could not be obtained or insufficient information was obtained from the chart to adjudicate case status.