Effects of Local Variation, Specialty, and Beliefs on Antiviral Prescribing for Influenza

Although influenza strikes 5%–20% of the population each year [1], little is known about physicians' approaches to this common illness. Randomized trials have demonstrated the efficacy of anti-influenza drugs in shortening the course of influenza infection and in preventing secondary infections that require antibiotics, but the effects of antiviral therapy on the rates of hospitalization and mortality are less–well documented [2]. In addition, there are limited data on children, high-risk patients, and elderly patients. In October 2004, during a vaccine shortage, the Centers for Disease Control and Prevention [3] issued interim antiviral guidelines, which recommended that high-risk patients receive amantadine or rimantadine for chemoprophylaxis and oseltamivir or zanamivir for treatment. The recommendations for other patients are vague, and diagnosis is not addressed.

Diagnosis of influenza can be difficult, because many viruses produce symptoms indistinguishable from those of influenza. Rapid testing for influenza allows for timely diagnosis [4], but it is not known how many physicians use rapid testing or whether physicians who test are more likely to treat. Although rapid testing has been shown to affect decision-making in the emergency department [5], the cost-effectiveness of rapid testing has been questioned [6, 7].

The influenza epidemic of 2003–2004 was particularly severe, and it included a large number of pediatric deaths [8]. Because there were no official treatment guidelines at the time, we expected to find substantial variation in practice by location and medical specialty. Texas was hit particularly hard by the epidemic, and we anticipated that awareness of influenza testing and treatment would be higher there than in Massachusetts, which experienced average influenza rates.

Methods. Baystate Medical Center is an acute-care hospital providing primary and tertiary care to residents of western Massachusetts. Physicians affiliated with the medical center practice in 1 of the hospital' 3 neighborhood health centers or in private practices in the area. Scott & White Hospital and Clinic is located in Temple, Texas. It comprises a 550-physician, multiple-specialty group practice; a 503-bed tertiary care hospital; and a self-contained health plan.

We designed a cross-sectional survey of physician knowledge, attitudes, and practices surrounding rapid diagnosis of and antiviral therapy for influenza. During March and April of 2004, we e-mailed 579 physicians in the Departments of Medicine, Pediatrics, Family Practice, and Emergency Medicine at Baystate Medical Center (277 physicians) and at Scott & White Hospital (302 physicians) and invited them to participate in an online influenza survey. The e-mailed invitations, which contained an embedded hyperlink to the survey, were followed up with 2 reminder notices sent 1 week apart. Physicians were excluded if they did not have a valid e-mail address, were not involved in clinical care, or had a subspecialty practice in which they did not see patients with influenza.

Data were collected using a self-administered, 41-question, Web-based survey (available on request) devised by the investigators. The survey contained questions about demographic information, antiviral prescribing, use of rapid testing, and the physician' beliefs about influenza. The survey was pilot-tested to assure clarity and coherence, and it was subsequently modified in response to the pilot test results.

The initial e-mail explained that the purpose of the study was to examine the physicians' knowledge, attitudes, and practices regarding influenza testing and treatment. Anonymity was assured, and completion of the survey implied informed consent. The study was closed on 15 June 2004. The study was approved by the institutional review boards of both Baystate Medical Center and Scott & White Hospital.

The data were entered automatically in an electronic database by the Internet survey company SurveyMonkey.com, then downloaded with SAS software version 8.2 (SAS Institute). Patient demographic variables were compared using the χ² test or Fisher' exact test, as appropriate. The association of main study variables with prescription of antivirals were assessed with logistic regression models and backward variable selection technique.

Results. Three hundred and thirty-six physicians (176 from Massachusetts and 160 from Texas) completed the survey (response rate, 58%). There were significant demographic and clinical differences between physicians in Massachusetts and those in Texas ( table 1).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Reasons given by survey participants from Scott & White Hospital in Temple, Texas, and Baystate Medical Center in Springfield, Massachusetts, for not prescribing antiviral therapy for influenza patients.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Demographic information and responses of survey participants, by location.

Sixty-one percent of the respondents reported that they had prescribed antiviral therapy during the past year, and 32% had written >5 prescriptions. Factors associated with the decision to prescribe or not prescribe antiviral drugs appear in table 2. Seventy-six percent of respondents in Texas reported prescribing antiviral therapy at least once in the past year, and 50% had written >5 prescriptions. In Massachusetts, however, only 48% had ever prescribed antiviral therapy (P <.0001). Internists comprised the lowest proportion of prescribers (48%), and family practitioners, the highest (84%).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2

Demographic and clinical variables associated with prescribing antiviral therapy.

In a multivariate analysis, the following factors were associated with prescribing antivirals: location (Texas vs. Massachusetts; OR, 3.5; 95% CI, 2.2–5.9), higher patient volume (OR, 2.0; 95% CI, 1.5–2.6), use of rapid testing (OR, 1.8; 95% CI, 1.1–3.0), and the beliefs that antiviral therapy shortens illness (OR, 4.9; 95% CI, 1.8–13.2) and decreases mortality rates (OR, 3.0; 95% CI, 1.7–5.5).

Respondents who indicated they had not prescribed any antiviral therapy offered several reasons for not prescribing ( figure 1). The reasons did not differ significantly by location. No one cited concern about inducing antiviral resistance.

Discussion. Although effective antiviral therapy for influenza has been available for almost 40 years, widespread prevention strategies have focused on vaccination, and little is known about physicians' use of antiviral medications to treat influenza. This is the first study to assess physicians' reported prescribing of anti-influenza therapy and the possible determinants of prescribing in the absence of any published guidelines or specific recommendations. In a multivariate analysis, we found that location, practice size, the performance of rapid testing, and the beliefs that antiviral therapy shortens the course of illness and decreases mortality all predicted the prescription of antivirals.

We assessed physician beliefs about antiviral effectiveness because we thought that belief in the effectiveness of antivirals would lead to prescription of antivirals. Although 87% of respondents in our study correctly identified that the medications shorten the course of illness, only 28% knew that they prevent bacterial complications. On the other hand, 29% believed that antiviral therapy decreases mortality, despite a lack of evidence. Not surprisingly, the belief that antiviral medications prevented serious complications was linked to antiviral prescribing. However, divergent beliefs about effectiveness explained only a small part of the variation in practice. Physicians in Massachusetts and Texas held similar beliefs about antiviral efficacy and gave the same reasons for not prescribing antiviral therapy, yet practice varied widely between the 2 locations. We hypothesized that antiviral prescribing would be more common in Texas due to the severity of the 2003–2004 influenza epidemic in that state, and our hypothesis was borne out in both univariate and multivariate analyses.

Most physicians who did not prescribe antiviral therapy reported that patients presented too late for treatment. If that is the case, it may be that physicians have failed to properly educate patients about influenza. Because many patients feel they should not seek medical attention until they have had several days of fever, they miss the opportunity to be treated with specific therapy, which reinforces the behavior of not seeking medical attention for influenza symptoms. It may be time, especially for high-risk patients, to reconsider this behavior. More troubling is the opinion, held by 40% of respondents, that influenza is self-limited and does not require treatment. Among high-risk patients, even those who have been vaccinated, the rate of secondary bacterial pneumonia after influenza infection is high. The appropriate use of antiviral therapy in this group could decrease the number of both lower respiratory tract complications and hospitalizations [9]. Finally, physicians expressed concern about cost. Several analyses have addressed the cost-effectiveness of antiviral therapy in different populations and have come to the conclusion that in most circumstances it is cost-effective [2, 6, 10–12].

Our study had some limitations. The retrospective nature of the study may have introduced recall bias. The anonymity of the survey precluded the analysis of nonresponders' beliefs and practices, and although our response rate was high, it is possible that inclusion of data from those who chose not to respond could have altered our results. Finally, our study population was limited to 2 medical centers that may not be representative either of their regions or of the nation as a whole. Nevertheless, we demonstrate wide variation in practice that cannot be ascribed to chance.

Anti-influenza therapy could decrease the substantial morbidity associated with annual influenza epidemics. Despite this, a large proportion of physicians fail to prescribe these agents. On the basis of the responses of nonprescribers, it appears that better education of providers about influenza and antiviral therapy, coupled with education of patients that would encourage presentation within the 48-h treatment “window,” could increase the rate of influenza treatment in the outpatient setting. The Centers for Disease Control and Prevention' interim guidelines [3] may ameliorate this situation, but they may not be renewed this season. In addition, these guidelines are vague about the use of antivirals for average-risk patients, do not address diagnosis, and do not provide documentation to support the recommendations. Given the prevalence of and severity of influenza and the wide range of beliefs and practices regarding treatment, detailed and well-documented practice guidelines are needed.

• Received June 16, 2005.
• Accepted September 2, 2005.

• © 2005 by the Infectious Diseases Society of America