S.-bound conveyance (aircraft, ship, or vehicle), or (2) within 72 hours after arriving in the United States, or (3) at any time after arriving in the United States from an illness possibly acquired during

international travel. We extracted the following data from QARS reports: demographics (age and sex), mode of transportation (aircraft, ship, land vehicle, or pedestrian), location of death, travel dates, traveler type (ie, passenger or crew member), citizenship, presence of chronic medical conditions, and cause of death. When data were missing from QARS death reports, we ABT-888 contacted CDC quarantine stations, medical examiners’ offices, and hospitals to complete case reports. Data were entered into a Microsoft Excel® database. Causes of death were categorized as cancer, cardiovascular, infectious disease, unintentional injury, intentional injury, and other (Table 1). Death rates for passengers on international

commercial conveyances were calculated for each year, by conveyance type. To present full, continuous yearly data (ie, four quarters) and to adjust for seasonality, we defined year 1 as July 1, 2005 to June 30, 2006; year 2 as July 1, 2006 to June 30, 2007; and year 3 as July 1, 2007 GSK458 clinical trial to June 30, 2008. We defined quarter 1 as January to March, quarter 2 as April to June, quarter 3 as July to September, and quarter 4 as October to December. To calculate mortality rates for cruise ship passengers, we divided the total number of reported cruise ship passenger deaths that met the case definition by the number of cruise passenger-nights traveled. We calculated the denominator by using data from the many U.S. Maritime Administration (MARAD) for cruises with an international itinerary and a port of arrival in the United States during years 1, 2, and 3.30 To determine

mortality rates for commercial aircraft passengers, we divided the total number of reported commercial aircraft passenger deaths that met the case definition by the number of airline passengers arriving in the United States from foreign ports. Denominator data were obtained from the U.S. Bureau of Transportation Statistics (BTS).31 Since MARAD and BTS do not collect data for crew members from cruise lines or airlines, respectively, we were unable to calculate crew mortality rates. We conducted bivariate analysis by using likelihood ratio chi-square tests, both asymptotic and exact, to evaluate associations between sex and cause of death. We analyzed monthly, quarterly, and yearly death rates among commercial aircraft and cruise ship passengers from July 2005 through June 2008 by using a general linear regression model in SAS (SAS 9.2, SAS Institute, Inc., Cary, NC, USA) to assess seasonality and trends in death rates over time.