We used a database from Medco Health Solutions, Inc. (Franklin Lakes, NJ, USA), a pharmacy benefits management company serving >50 million US members. We examined filled prescriptions for oseltamivir by members from January 2002 through May 2006. Available member data included demographic information, medication dispensed, prescriber identification, and pharmacy dispensing history. Member-level historic pharmacy dispensing data were used to assign members into chronic disease classifications (10). Prescribers were cross-referenced with an American Medical Association member database to determine specialty and years since medical school graduation. We were able to cross-reference 64% of prescribing physicians by specialty and years since medical school graduation. The Centers for Disease Control and Prevention (Atlanta, GA, USA) determined that institutional review board approval was not needed for this study because we received aggregated data that was anonymous and not identified.

To assess media coverage, we queried the LexisNexis US News database (www.lexisnexis.com) for total weekly news reports from August 1, 2003, through August 30, 2006, referring to avian influenza and oseltamivir. Weekly virologic data from the World Health Organization and National Respiratory and Enteric Virus Surveillance System collaborating laboratories were used to assess US influenza activity during 2004 and 2005 (6,7).

Oseltamivir prescription rates were calculated per 100,000 enrolled members and per 1,000 prescribing physicians. Binomial distributions were used to estimate variances for rates. Relative rate ratios (RRs) and 95% confidence intervals (CIs) were calculated for 2004 and 2005 data. P values <0.05 were considered statistically significant. Analyses were performed with SAS version 9.0 statistical software (SAS Institute, Cary, NC, USA).

Weekly rates of filled prescriptions for oseltamivir and percentage of samples positive for influenza from October 1, 2002, through June 1, 2006, were temporally associated before the 2005–06 influenza season (Figure 1). During the fall of 2005, prescriptions for oseltamivir increased without an associated increase in the percentage of samples testing positive for influenza. In contrast, during the same period, there was a temporal relationship between weekly oseltamivir prescription rates and media reports of avian influenza and oseltamivir (Figure 1).

The proportion of oseltamivir prescriptions to total anti-influenza prescriptions increased from 37.0% in 2004 to 76.9% in 2005 (Table 1). The 2005 oseltamivir prescription rate of 133/100,000 during weeks 36–44 was ≈5× the 2004 rate of 27.3/100,000 (RR 4.88, 95% CI 4.79–4.97) (Table 1).

Women were more likely to receive oseltamivir prescriptions than men in 2004 (RR 1.19, 95% CI 1.16–1.24) and 2005 (RR 1.07, 95% CI 1.05–1.09). Prescription rates increased from 2004 to 2005 for all age groups (Table 2). The highest prescription rates in 2005 were for persons 50–64 years of age (211/100,000) and those >65 years of age (168/100,000). Members <18 years of age had a >7-fold increase in prescription rates from 2004 to 2005.

Among adults, prescription rates were consistently higher in 2005 than in 2004, irrespective of chronic disease classification (Table 2). In 2005, the highest oseltamivir prescription rate was 268.5/100,000 for enrollees with pulmonary disease, and the lowest rate was 89.5/100,000 for those without chronic disease. The greatest rate increase from 2004 to 2005 occurred among those without chronic disease (RR 6.36, 95% CI 6.10–6.62). Among enrollees of all ages during weeks 36–44 in 2005, approximately one third of oseltamivir prescriptions were filled by members without chronic disease.

In 2004 and 2005, general internists had the highest average prescribing rate for oseltamivir (45.3/1,000 and 191.1/1,000, respectively) (Figure 2). Pediatricians had the lowest rates in both years (2.9/1,000 and 32.7/1,000).

For all prescribers in 2004 and 2005, oseltamivir prescription rates increased with years since the prescriber’s medical school graduation. The lowest prescription rate in 2005 (1.7/1,000) was observed in prescribers who graduated from medical school in the previous 5 years, followed by prescribers with 5–10 years (4.9/1,000) and 11–19 years (6.6/1,000) since graduation. The highest rate (10.4/1,000) was observed in prescribers with >20 years since graduation.

Rates of filled oseltamivir prescriptions during calendar weeks 36–44 increased from 2004 to 2005. These weeks in 2005 were noteworthy for increased media references to oseltamivir and avian influenza, although there was little influenza activity. Low levels of influenza-like illness and respiratory syncytial virus activity also were documented during this period (6,11).

Among Medco enrollees, the highest prescription rates were for groups with the greatest risk for influenza-associated complications: persons >50 years of age and adults with chronic diseases (2). However, oseltamivir prescription rates increased from 2004 through 2005 for each age and chronic disease category. Although absolute prescription rates were lower in persons <50 years of age, persons without chronic medical diagnoses, and children, these groups had higher rate increases from 2004 to 2005, suggesting that they and their caretakers were influenced to stockpile oseltamivir in 2005.

Physician prescribing rates for all specialties increased from 2004 to 2005. Prescription rates in 2004 were highest for general internists and family practitioners and likely reflect the primary care physician’s gatekeeper role as the entry point for those seeking medical care. Given the nature of their specialty, infectious diseases physicians may be asked to prescribe oseltamivir for personal stockpiles more than the typical provider. Because physicians need not honor all prescription requests, these increases may not be fully explained by increased patient requests because physician attitudes regarding personal stockpiling likely affected whether requests were made or honored.

Our study is subject to limitations. Although we studied a large, national population, our study population may not be nationally representative. In addition, our analyses were limited to prescriptions of oseltamivir that were filled by a pharmacy, and we do not know whether prescriptions were written with the expressed purpose of personal stockpiling for use during a pandemic.

In summary, our findings suggest that increased media reports during the fall of 2005 about the influenza (H5N1) epizootic prompted concern about the possibility of an influenza pandemic. This heightened concern led to an increase in filled oseltamivir prescriptions for personal stockpiling among a national pharmacy benefits member population. Subsequently, as influenza virus began circulating in early 2006, oseltamivir prescriptions corresponded more closely with virus activity. Efforts by federal and state governments to procure sufficient supplies of NIs to treat every patient likely to become ill during the next pandemic may quell demand in personal stockpiles. Education campaigns about appropriate use of antiviral medications that target physicians and patients during seasonal epidemics and pandemics may reduce inappropriate requests for oseltamivir and other drugs.