Facilitating Early Treatment of Influenza in Hospitals: Empiric Antivirals or Empiric Diagnostics?

(See the Major Article by Katzen et al on pages 52–8.) Current clinical guidelines recommend early administration of antivirals to hospitalized patients with laboratory-confirmed or suspected influenza. However, compliance with these guidelines is often limited in clinical practice, hence, many influenza patients admitted to hospitals remain left untreated and potentially at risk for complications and death. Reasons for this lack of compliance may include doubts and misconceptions of clinicians about the clinical efficacy of oseltamivir in hospitalized patients. Placebo-controlled, randomized clinical trials, the cornerstone of evidence-based medicine, have only been conducted in previously healthy individuals with uncomplicated influenza and have led to the licensure of oseltamivir for acute, uncomplicated influenza only. In view of current treatment recommendations for the (off-license) use of oseltamivir to treat severely ill patients, such placebo-controlled studies are generally deemed not feasible in hospitalized influenza patients for ethical reasons. One could perhaps argue the ethics of not performing such studies if these were to result in improved compliance to evidence-based recommendations and improved outcomes for future generations of patients. In the mean time, there are only observational studies at hand to inform guidelines and convince clinicians about the benefits of oseltamivir. An important additional piece of such observational evidence is published in this issue of Clinical Infectious Diseases [1]. In a large, retrospective, cohort study of hospitalized patients with laboratory-confirmed influenza, Katzen et al show that the administration of antivirals within 6 hours after hospital admission was associated with a shorter length of hospital stay when compared to treatment initiation between 6 and 24 hours or more than 24 hours after admission. In a separate analysis, they also show that empiric antiviral treatment, defined for the purpose of this study as administration of oseltamivir before a positive influenza test was received or within 6 hours after admission, was associated with a reduced risk of in-hospital mortality in addition to a shorter duration of hospitalization. The difference in timing of treatment between the empiric and nonempiric groups was substantial (median 4.3 vs 33.5 hours after admission). These findings confirm indications of benefit of timely oseltamivir treatment from previous observational studies [2, 3], which were mostly restricted to patients with influenza A/H1N1pdm09 during the 2009–2010 season and extend these to 4 post-pandemic years when other seasonal influenza viruses were also circulating. An additional strength is the fact that Katzen et al analyzed timing of treatment as defined by time since hospital admission rather than since illness onset, as used in previous studies. The definition used in this study is more relevant to clinical practice and may help to avoid the misconception of many clinicians that oseltamivir loses its benefit if not started within 48 hours after illness onset. The latter treatment window stems from studies in healthy individuals with uncomplicated influenza; however, observational studies have clearly suggested longer windows of opportunity for severely ill hospitalized patients. The current study adds to this evidence as most patients were admitted over 72 hours after illness onset (median 3.6 days). Katzen et al also investigated patient-level factors associated with early institution of antiviral treatment and found that pregnant patients and patients with diabetes were more likely to receive early treatment, while immunocompromised patients were less likely to receive early treatment. The latter observation is concerning since there is an obvious need for timely and adequate treatment in immunocompromised patients. Possible reasons for the observed treatment delay in this patient group are not discussed in their report, but one could hypothesize a delay by physicians in considering and diagnosing influenza in these patients, perhaps due to initially less typical symptomatology in the immunocompromised. In this light, it is interesting to note that presenting symptoms of fever and myalgia, both typical of influenza, were associated with early administration of oseltamivir. It should be noted that this retrospective, cohort only included patients with laboratory-confirmed influenza, implying that influenza was suspected and subsequently tested by attending physicians at least at some point during admission. Similar to the range in time-to-treatment initiation, diagnostic results were reported within 24 hours after admission in nearly one half of patients, between 1 and 2 days in approximately one third, and still later in one fifth. Assuming little variation in time from laboratory testing to reporting, these observations from clinical practice might imply that delays of treatment initiation are secondary to delays in suspecting influenza and subsequent testing by physicians in a substantial proportion of patients. Following this line of thought, rather than poor adherence to recommendations for early treatment of suspected influenza per se, as concluded in this study, poor and delayed clinical suspicion of influenza in hospitalized patients may be the issue. Empiric initiation of treatment, as argued for by Katzen et al, may well lead to earlier treatment in at least a proportion of patients and is worthy of formal evaluation in clinical trials. However, as discussed above, the effectiveness of empiric treatment approaches will depend on an accurate clinical suspicion of influenza (in addition to influenza activity, which varies per season). For this reason, rather than or complementary to empiric treatment, the use of “empiric diagnostics,” that is, not restricting the use of influenza diagnostics to those in whom influenza is clinically suspected by the physician but to test all those presenting with acute respiratory symptoms and/or fever during influenza seasons, may be more effective in facilitating early initiation of antiviral treatment. Indeed, 2 recent studies, 1 observational and 1 interventional, have shown that timely influenza diagnostics in patients who present with respiratory illnesses improves approriateness of antiviral therapy, including the initiation of treatment in those who were not empirically treated [4, 5]. Further evaluation of such empiric diagnostic approaches in randomized clinical trials is merited. In summary, the study by Katzen et al represents another piece of evidence suggesting benefits of early initiation of oseltamivir in hospitalized influenza patients. While observational studies have obvious inherent limitations, the strength of such studies is that these truly reflect clinical practice, thereby providing real-world evidence for possible shortcomings in clinical care and inspiring potential solutions for improvement. Also, in this respect, this study provides important food for thought. Potential conflicts of interest. M. D. dJ. reports travel and other fees from Janssen, Shionogi, GlaxoSmithKline, and Vertex outside the submitted work. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.