\*\*\*Patients were not recruited for nor enrolled in this study. This study is a retrospective observational study. Data from medical records or insurance claims databases are anonymised and used to develop a patient cohort. All diagnoses and treatment are recorded in the course of routine medical practice.\*\*\*
This analysis will use the i3 Ingenix proprietary National Health Information (NHI) database, containing claims and enrollment data back to 1993, with the opportunity to link patient and physician survey data to pharmacy and medical claims, and clinical laboratory results. These data are use for a wide range of pharmacoepidemiologic, healthcare utilization, and economic analyses. In addition to data derived from affiliated health plans, data from large, national employer groups are also stored in the database. The data undergo regular audits and quality control procedures. The accessible information includes demographics, pharmacy use, and all medical and facilities claims, which provide data on services, procedures, and their accompanying diagnoses. Underlying information is geographically diverse across the US, and is updated frequently. The insured population from which the database draws the data comprises approximately 4 percent of the US population. The data used for analysis is de-identified.
Study Cohorts From October 2006 through April 2009, we will count patients receiving prophylactic Relenza, defined as a dispensing of Relenza to a person who meets two criteria: (1) not having had a diagnosis of influenza associated with medical care on the day of dispensing or in the preceding three days, and (2) a household member has had a diagnosis of influenza associated with medical care on the day of the index dispensing or within the preceding three days. "Household members" will be defined operationally as persons within the RDM sharing a common family identifier and address.
We will assemble the study cohorts for the 3 influenza seasons running from October 2006 through April 2009. Cohort membership is based on the treatment an individual receives and the nature of household exposure.
Exclusion criteria include:
* Sex unknown
* Year of birth is missing or where the date of dispensing or service date preceded the year of birth
* Age at index date is less than 5 years
* Not enrolled in the health plan for 6 months prior to the study entry date
Observation begins for the prophylaxis and treated patients on the day following Relenza dispensing and for the non-prophylaxis and untreated patients on the third day following their medical visit, and continues for a total of 30 days.
Covariates For each individual we will identify age, sex, medical conditions requiring care in the six months preceding cohort entry from insurance claims for health services. Medical conditions requiring care will be defined at the 3-digit ICD9 diagnostic level.
Outcomes We will identify medical visits and hospitalizations with influenza-like illness and respiratory disease in the 30 days following cohort entry.
Analysis
Within categories of age of 5-14, 15-44, 45-64 and 65+ years, we will tabulate:
* the number of individuals in each of four cohorts (cohorts defined below)
* the number of medical visits for influenza-like illness and respiratory illness in the 30-day follow-up period
For each outcome we will estimate the following antiviral efficacy measures detailed in Halloran's analysis of clinical trials data \[Halloran, 2007\]. The first two effects below will be standardization to the age distribution of all persons receiving prophylactic Relenza;
1. Direct Effect of Relenza Prophylaxis: Risk in Cohort 1 is less than Risk in Cohort 3 Prophylaxis patient with untreated index case vs patient without prophylaxis with untreated index
2. Total Effect of Relenza Prophylaxis and Treatment: Risk in Cohort 2 is less than Risk in Cohort 3 Prophylaxis patient with treated index case vs. patient without prophylaxis with untreated index
We will calculate the following effect estimate with standardization to the age distribution of untreated persons.
3. Protective Effect of Relenza Prophylaxis on Susceptible: Risk in Cohort 2 is less than Risk in Cohort 4 Prophylaxis patient with treated index case vs. patient without prophylaxis with treated index
Using binomial regression and a linear risk model, we will obtain adjusted estimates for Direct Effect, Total Effect, and Protective Effect, defined as above, adjusted for age group, sex, and any diagnosis with a prevalence of at least 10 percent in the cohorts combined.
