Changes in Prescribing of Antiviral Medications for Influenza Associated With New Treatment Guidelines

Abstract

In 2006, the Centers for Disease Control and Prevention recommended discontinuing the use of adamantanes (amantadine and rimantadine) to treat influenza because of high levels of resistance to this class of antivirals. We examined changes in prescribing practices resulting from this recommendation and found that prescribing of adamantanes declined nationwide, with these drugs accounting for approximately 40% of the antivirals prescribed for influenza from 2000 to 2005 and only 2% in 2006. This finding provides evidence of a rapid change in clinical practice associated with the dissemination of treatment guidelines. Evaluating the effectiveness with which public health recommendations are translated into practice is important given the ongoing emergence of resistance to antiviral drugs and a novel H1N1 influenza virus.

Dissemination of epidemiological information to physicians is crucial for controlling emerging infections and ensuring appropriate prescription of antimicrobial medications. Resistance to antimicrobials can emerge rapidly, requiring prompt practice changes; however, research on the ways in which physicians respond to new information about antimicrobial-resistant infections is limited.

In January 2006, high rates of resistance to adamantanes (amantadine and rimantadine) led to a Centers for Disease Control and Prevention (CDC) Health Alert advising clinicians to discontinue use of these drugs and prescribe only neuraminidase inhibitors (oseltamivir and zanamivir) for influenza treatment and prophylaxis.¹ Further changes in resistance during the 2008–2009 influenza season² and the pandemic caused by a novel influenza A(H1N1) virus³ highlight the importance of rapid information dissemination to clinicians.

Our objective was to evaluate changes in antiviral prescribing practices stemming from the revised 2006 treatment recommendations. Such information could provide insight into the effectiveness of current methods of disseminating treatment recommendations for influenza and other infections.

METHODS

We examined data from the National Ambulatory Medical Care Survey and the National Hospital Ambulatory Care Survey from 2000 to 2006, focusing on changes occurring before and after January 2006. These surveys, conducted by the National Center for Health Statistics (NCHS), incorporate a random sample of visits to physicians' offices and emergency departments, enabling nationally representative estimates of annual visits.⁴ Survey data include reasons for visits, based on codes from the International Classification of Diseases, Ninth Revision, Clinical Modification,⁵ and medications prescribed. NCHS staff conduct training for participants and perform periodic validations to ensure accurate recording.

We identified visits during which physicians prescribed either an adamantane or a neuraminidase inhibitor to calculate estimates of numbers of antiviral prescriptions overall and within each of these 2 classes. For years preceding 2006, we combined and annualized data in 2-year groups to address sample size limitations (in accordance with NCHS recommendations). Because adamantanes are also prescribed for neurological conditions, we manually reviewed and excluded visits relating to neurological conditions that lacked a diagnosis indicating an influenza-like illness. We used the χ² trend test to evaluate changes in prescribing practices.

RESULTS

The rate at which antiviral medications were used in the treatment of influenza changed substantially in 2006 relative to the preceding study years; adamantanes accounted for 37% to 43% of antiviral prescriptions from 2000 to 2005, declining to 2% in 2006 (Figure 1; P = .03). This drop corresponded to the release of the 2006 CDC Health Alert recommending discontinuation of adamantanes in treating influenza.

FIGURE 1.

Numbers (bars) of antivirals prescribed for influenza and percentages of adamantane prescriptions in relation to total antiviral prescriptions: United States, 2000–2006.

Note. Data for 2000–2001, 2002–2003, and 2004–2005 are annualized.

In absolute terms, the estimated number of visits during which an adamantane was prescribed for influenza (annualized) ranged from 183 024 to 328 017 in the years leading up to 2006. During 2006, the estimated number of visits during which an adamantane was prescribed declined to 20 444 (95% confidence interval [CI] = 0, 51 508) (Figure 1). Correspondingly, neuraminidase inhibitors accounted for 57% to 63% of antiviral prescriptions between 2000 and 2005, increasing to 98% in 2006 (Figure 1).

DISCUSSION

We found a rapid decline in the prescription of adamantanes to treat influenza in 2006 that was temporally associated with new guidelines advising clinicians to discontinue use of this class of antivirals. These guidelines initiated a cascade of subsequent publications in both the lay and medical press.^6,7 Physicians have apparently learned about these new guidelines through professional as well as lay channels and changed their prescribing practices accordingly.

Diffusion of innovation theory regarding physician adherence to guidelines provides insight into factors that may have contributed to these rapid changes in practice. Messages that are simple (e.g., discontinue adamantanes), are consistent with individual beliefs, offer an alternative consistent with current practice (e.g., prescribe neuraminidase inhibitors), and are associated with minimal disagreement are most likely to be rapidly adopted.^8,9 Although internal and external barriers may delay adoption of new guidelines in the case of certain diseases,⁸ dissemination of epidemiological information about antimicrobial-resistant infections may facilitate rapid adoption because of the unequivocal nature of surveillance data.

Because this was not a controlled study, several limitations and alternative explanations must be considered. Without directly surveying physicians about the specific reasons they prescribed neuraminidase inhibitors as opposed to adamantanes in 2006, we cannot exclude the possibility that coincident factors unrelated to emerging resistance to adamantanes (e.g., pharmaceutical detailing or marketing or changes in drug supply) contributed to the changes in practice observed. Notably, however, there were no changes in the use of adamantanes to treat neurological conditions during the study period.

Sample size limitations required us to aggregate our data into 2-year intervals, which may have masked a more gradual change in prescribing practices. In addition, our data set captured only outpatient visits, and thus our results are not representative of prescription of antivirals in other settings such as hospitals and nursing homes.

Many factors may have contributed to the rapid changes in influenza prescribing practices observed in this study, including dissemination of the 2006 CDC guidelines via multiple networks and the compatibility of the message with physicians' beliefs regarding resistance to antimicrobials. As mentioned, new resistance patterns and a novel H1N1 virus emerged during the 2008–2009 influenza season, requiring further changes in selection of antiviral drugs. This situation highlights the importance of continuously evaluating the effectiveness with which public health recommendations are translated into practice and identifying interventions that facilitate this process.