The results of this study confirm that public demand for anthrax-related antimicrobial agents was substantial in Minnesota and Wisconsin, since one fourth of primary care clinicians received requests for these drugs. We found that physicians received these requests more often than nonphysicians. Despite requests from patients and family members, relatively few antimicrobial agents were dispensed for anthrax prophylaxis. Extrapolation of survey responses to all primary care clinicians in Minnesota and Wisconsin suggests that ≈500 anthrax-related antimicrobial drug courses were dispensed during the last quarter of 2001. Even if nearly all of these were fluoroquinolones, the volume of anthrax-related fluoroquinolone use represented only a small fraction of total use in Minnesota and Wisconsin during that period.
The prescribing and retail distribution data showed surges in fluoroquinolone use during October 2001, but total use of antimicrobials also increased for unknown reasons during that period. Fluoroquinolone prescriptions as a proportion of all antimicrobial prescriptions were not elevated, which was consistent with the findings from the clinician survey. Although tetracycline/doxycycline prescriptions as a proportion of all outpatient antimicrobial prescriptions (excluding fluoroquinolones) were elevated in October and November 2001, the clinician survey indicated that this increase was unrelated to anthrax prescribing. None of the survey respondents reported using tetracycline/doxycycline for this purpose, and ciprofloxacin was the anthrax drug that received most media attention in late 2001.
Two other studies have addressed national use of antimicrobial agents following the anthrax cases in 2001. In one study that used a national pharmacy claims database, the rate of ciprofloxacin use increased 9.8% in October 2001 relative to October 2000 (15
). As expected, the greatest increase in use was observed in New York, the mid-Atlantic states, and Florida. Ciprofloxacin prescribing rates were not reported for Minnesota or Wisconsin. In this study, the denominator was defined as the number of covered persons who filled a prescription for any drug or eligible health product during that month. As a result, the observed rate differences may have been influenced by both changes in the numerator (number of ciprofloxacin prescriptions) and changes in the denominator (number of persons filling any prescription).
A similar study used IMS Health National Prescription Audit Plus7 data to compare national ciprofloxacin use from July to December of 2001 and 2000 (16
). Comparison drugs included oral azithromycin and cefuroxime, which are commonly used in outpatient practice but not recommended for anthrax prophylaxis. Ciprofloxacin prescriptions increased by 42%; cefuroxime prescriptions declined by 3%. The results were not reported by region, and they included prescriptions in New York, Florida, and other affected regions. The authors did not assess monthly ciprofloxacin prescriptions as a percentage of all antimicrobial prescriptions. The results of our study suggest that short-term variations in single drug prescribing should be interpreted with caution when the specific diagnoses or prescribing indications are not known. We found that a short-term increase in fluoroquinolone use in Minnesota and Wisconsin was accompanied by an overall increase in antimicrobial drug use. Thus, factors unrelated to anthrax may have also contributed to the observed increase in fluoroquinolone use during October 2001, especially in unaffected regions of the United States.
Whether patterns of antimicrobial use in Minnesota and Wisconsin are generalizable to other unaffected regions of the United States is not known. For example, total ciprofloxacin prescriptions in October 2001 appeared to increase >25% in some unaffected states, including Nevada, California, and New Mexico (15
). No information is available regarding the clinical indications for these prescriptions, and how much of this increase can be attributed to anthrax-related prescribing is unclear. Other factors may also contribute to regional differences in prescribing, since physicians in the northeastern and southern United States are more likely to prescribe broad-spectrum antimicrobials than those in the midwestern or western regions (17
The survey results in Minnesota and Wisconsin may have underestimated actual anthrax-related prescribing, since clinicians who dispensed antimicrobial agents may have been reluctant to return the survey. However, the cover letter and survey questions were neutral regarding the appropriateness of antimicrobial drug use, and the survey was anonymous. Poor recall is another potential source of error, since the survey was conducted approximately 1 year after the first cases of intentional anthrax occurred. Because we were asking about unusual events that were outside the scope of normal clinical practice, we assumed that clinicians would still recall any anthrax-related prescribing. Finally, the survey results did not allow us to determine if patients consumed these agents for anthrax prophylaxis, or if they were stockpiled for future use.
The human anthrax cases in 2001 and the related events illustrate how quickly demand for a critical drug can escalate as a result of heightened public anxiety and media attention. Most physicians in Minnesota and Wisconsin managed public and patient expectations without dispensing antimicrobial agents. However, social factors clearly influence prescribing decisions (18
), and effective public and physician communication will be essential to promote rational behavior if similar or more extreme situations arise in the future. A communications strategy should be developed in advance that includes identifying key experts at the state and national level for news media communications and devising a plan for coordination and consistency of messages from different agencies.