We abstracted cases that met these inclusion criteria and entered them into a database (Epi-Info 6.01, CDC, Atlanta, Georgia). We used Stata software, version 7 (Stata Corp., College Station, Texas), for all statistical analyses. We also used several definitions. We defined time to seeking care as the number of days between symptom onset and first visit to a medical provider (coded as “unknown” if treatment-seeking details were not available in the case summary), time to diagnosis as the number of days between the first visit to a medical provider and the diagnosis of malaria (coded as “unknown” if these dates were not specified), time to treatment as the number of hours between a diagnosis of malaria and the initiation of antimalarial treatment, appropriateness of chemoprophylaxis regimen as the extent to which CDC recommendations published at the time of travel were followed, and appropriateness of treatment as the determination based on the most recent recommendation from The Medical Letter for that year. We coded adherence to chemoprophylaxis as “adherent” or “nonadherent” if this was specifically mentioned; if not, we coded this as “unknown.” We defined a preventable death as one in which the person 1) took no chemoprophylaxis, 2) took (or was prescribed) inappropriate chemoprophylaxis, 3) took the correct chemoprophylaxis but did not completely adhere to the prescribed regimen, 4) delayed seeking medical care for more than 2 days after the onset of symptoms, 5) sought medical care but did not receive a diagnosis on the day of initial presentation with malaria, 6) was given a diagnosis of malaria but treatment began more than 1 day after diagnosis, or 7) was treated with an antimalarial drug that was inappropriate for the infecting species and region of acquisition.