Pan-canadian public health network council report and policy recommendations on the use of antivirals for prophylaxis during an influenza pandemic

[Previous] [Table of Contents] [Next]

Annex 3.1–Scientific Considerations on the Prophylactic Use of Antivirals in an Influenza Pandemic

Submitted by:
Patricia Huston
Chief, Emerging Infectious Disease Section,
Immunization and Respiratory Infection Division,
Public Health Agency of Canada

Intended for: Task Group on Antivirals for Prophylaxis

Context

In June 2006, the PHN Council recommended that TGAP’s deliberations include arms length systematic reviews of the literature to assess the evidence for antiviral prophylaxis for influenza. TGAP commissioned the Canadian Agency for Drugs and Technologies in Health (CADTH) to conduct reviews through their Health Technologic Inquiry Service on the efficacy and safety of Tamiflu^® (oseltamivir), Relenza^® (zanamivir) and amantadine. The CADTH report on amantadine concluded that the overall usefulness of amantadine for prophylaxis was Aquestionable” due to its side-effect profile and the potential for the emergence of resistant strains. Based on this conclusion, TGAP focused this report on the neuraminidase inhibitors. To supplement CADTH’s work, TGAP conducted additional research on adverse events, compliance and the risk of resistance specifically as it relates to Tamiflu^®, as this is what comprises 90% of the current national (F/P/T) antiviral stockpile.

1.Current indications

There are two types of prophylaxis. Post-exposure prophylaxis means taking a daily dose of medication for 10 days following close contact with an infected individual. Pre-exposure prophylaxis means taking a daily dose of medication for a number of weeks prior to expected exposure.

Tamiflu^® has been approved for post-exposure prophylaxis for adults and children over 1 year of age if started within 2 days of exposure. Relenza^® was recently approved for post-exposure prophylaxis for adults and children over 7 years of age, if started within 1.5 days of exposure.

Tamiflu^® is not approved for pre-exposure prophylaxis. Relenza^® was recently approved for “community outbreaks” for 4 weeks. In light of the fact that it has been estimated that a pandemic will come in waves lasting 6-8 weeks, taking Tamiflu^® and Relenza^® as pre-exposure prophylaxis for at least 6-8 weeks are considered “off-label use”.

2.Current practice

There is very little use of antivirals in the community by front-line physicians. The main use of Tamiflu^® at this time is to control long-term care (LTC) facility outbreaks, when it is prescribed for both treatment and post-exposure prophylaxis. This is based on a recommendation by the National Advisory Committee on Immunization (NACI).

3.Assessment of available data

There are no data available regarding the use of neuraminidase inhibitors during an influenza pandemic as this class of drugs had not been developed at the time of the last pandemic. While these drugs are currently being used in the H5N1 avian influenza outbreaks in Europe and Asia, no RCTs have been conducted in these outbreak settings. When considering the use of these drugs during a pandemic the only available data comes from studies conducted during annual influenza seasons. These data have been the focus of our analysis of efficacy, effectiveness and efficiency (see Table 2 for a definition of these terms).

Table 2: Definitions of terms used to describe how well a drug works

(Based on John Last’s Dictionary of Epidemiology)

	Efficacy	Effectiveness	Efficiency
Definition	The extent to which an intervention produces a beneficial result under ideal conditions. Can be measured as: Relative efficacy: The efficacy of an intervention relative to something else; such as a placebo or another drug. Absolute risk reduction: The observed reduction in the probability of an event in a population under study	The extent to which an intervention produces the intended beneficial effect under routine conditions	An indication of the economy or cost in resources with which an intervention of known efficacy is carried out. Number needed to treat (NNT)- number of people who must follow a regime to prevent a specific outcome.
How it is determined	Randomized controlled trials (RCTs)	Intervention studies, case series	Calculation based on RCT results
Strengths	Highest level of evidence; least prone to bias	Most like “real life”	Gives an estimate of the intensity of resources needed to get the intended effect.
Weaknesses	Artificial environment; often a “best case” scenario	Lower level of evidence compared to RCTs; more vulnerable to bias	Estimates of efficiency will change if clinical attack rate changes

Both post-exposure and pre-exposure prophylaxis were reviewed. To assess their potential for use during a pandemic, efficacy, effectiveness, and efficiency were considered. TGAP determined that the typical summary of efficacy of neuraminidase inhibitors as being A60-90% effective@ B was potentially misleading. People might infer that by taking antiviral drugs they could decrease their risk of getting influenza by 60-90% B which is not correct. For example, in the Peters et al study, 0.4% of people in the prophylaxis group had symptomatic lab confirmed influenza compared with 4.4% of people in the placebo group. The relative efficacy of this is (4.4–0.4) ÷ 4.4 or 91%. The absolute risk reduction, however, is 4.4–0.4 or 4%. The large difference between these two figures is due to the fact that symptomatic lab confirmed influenza was relatively rare in both the placebo and Tamiflu^® groups. TGAP concluded it was important to present the evidence in terms of both relative efficacy and absolute risk reduction to fully understand the differences in the way data can be presented. It also wanted to see if there were differences in the efficiencies of the two types of prophylaxis in terms of number of people needing to take prophylaxis to prevent one case.

Methods

1.Relative efficacy, absolute risk reduction and efficiency

CADTH conducted an independent systematic review of the evidence for Tamiflu^®, Relenza^® and amantadine. After the literature search was completed, the list of identified studies was circulated to the Pandemic Influenza Committee’s Antiviral Working Group to ensure completeness. CADTH was asked to assess the quality of the studies, so that higher quality studies could be given greater weight than lower quality studies. TGAP had one opportunity to identify any errors and omissions in the draft report. CADTH also undertook an independent peer review of their reports; it takes full responsibility for the analysis of the findings⁷.

TGAP focused mostly on the RCT data from the CADTH reports. It summarized the measures of relative efficacy based on the primary outcome measure -- symptomatic, laboratory confirmed influenza. TGAP calculated the absolute risk reduction. It also determined the efficiency of each intervention by calculating the number of people who would need to take a course of prophylaxis to prevent one case (typically called the “number needed to treat” (NNT). Findings from effectiveness data (case series on LTC facility outbreaks) were summarized. More details are available in the CADTH report.

2.Safety, compliance and the risk of resistance

CADTH summarized the safety data from the studies reviewed; we supplemented this with data from the product monographs. To assess the likelihood of compliance, TGAP examined data from the RCTs on neuraminidase inhibitors, and conducted MEDLINE searches looking at the literature on compliance in general and from two case scenarios with similarities to the pandemic situation: the 2003 anthrax exposure in the United States postal workers, and the use of antiviral prophylaxis during avian influenza outbreaks. To assess the risk of resistance, TGAP conducted a MEDLINE search, focusing on articles published in the last 5 years and representing multiple perspectives.

Results

Overall, 23 studies of neuraminidase inhibitors were identified: 13 randomized controlled trials (RCTs), 1 pilot RCT, 8 case series of LTC facility outbreaks, and 1 descriptive study of the use of antivirals during the avian influenza outbreak in the Netherlands. On further examination 1 RCT was a duplicate publication, so was excluded.⁸ Of the 12 RCTs, 5 were on oseltamivir and 7 were on zanamivir. All underwent a quality assessment, using a widely accepted AJadad score@ that varied from 0 (lowest quality) to 5 (highest quality). TGAP focused on the RCT data as it is the least prone to bias and other threats to validity. Effectiveness data from case series is most useful once efficacy has been established under RCT conditions.

1.Relative efficacy and absolute risk reduction

Table 3 offers a summary of the 12 RCTs, including data on relative efficacy, absolute risk reductions and the number needed to treat (in this case, the number needed to complete a course of prophylaxis) to prevent one case of influenza.

Whereas relative efficacy showed a similar range for the different types of prophylaxis studies, absolute risk reductions showed differences between the different types of prophylaxis. When risk of exposure was high (in post-exposure prophylaxis studies) the absolute risk reduction (ARR) was 10-15%. When risk of exposure was low (in pre-exposure prophylaxis studies) the ARR was 1-4%.

a) Five Tamiflu^® (oseltamivir) trials

Of the 5 Tamiflu^® trials, there were 2 excellent quality trials and one trial with a low quality score on pre-exposure prophylaxis, and 2 good quality trials on post-exposure prophylaxis. There were no trials on LTC facility outbreaks.

Of the 2 high quality trials of pre-exposure prophylaxis, one RCT showed a relative efficacy of 74% with healthy people in the community (Hayden, 1999) the other showed a relative efficacy of 91% in elderly in retirement homes (Peters, 2001). Both had a similar absolute risk reduction of 3.6% and 4% respectively. The number needed to prevent one case varied from 25-28 people. One study received a low quality score as only an abstract was available (Kashiwagi, 2000), but their results were consistent with the trials that had the higher quality scores.

Of the 2 good quality trials on post-exposure prophylaxis, one showed a relative efficacy of 68% (Hayden, 2004) and the other showed a relative efficacy of 89% (Welliver, 2001) with an absolute risk reduction of 15.1% and 11.2% respectively. The number needed to treat to prevent one case varied from 6-9 people.

Seven Relenza^® (zanamivir) trials

Of the seven Relenza^® trials, there were two excellent quality trials on pre-exposure prophylaxis, two very good quality trials on LTC facility outbreaks, and two excellent, and one fair quality trials on post-exposure prophylaxis.

Of the 2 trials for pre-exposure prophylaxis, relative efficacy ranged from 67% (Monto, 1999) to 84% (Campbell) and absolute risk reduction was 4% and 1.1% respectively. Of note, in the Campbell trial only 1.3% of people in the placebo group were diagnosed with influenza vs. <1% in the zanamivir group. Although this was statistically significant it was very inefficient; the number needed to prophylax for 6 weeks to prevent one case of influenza was 91 people.

Of the 2 very good quality trials for LTC outbreaks, one showed a statistically significant difference (Gravenstein, 2005); the other did not (Ambrozaitis,2005). The Gravenstein trial found a 61% relative efficacy of Relenza^® compared to rimantadine and and an absolute risk reduction of 5%. In the Ambrozaitis trial, although there were fewer people in the zanamivir group that got symptomatic laboratory confirmed influenza than in the placebo group, the difference was not statistically significant.⁹

Of the 3 trials on post-exposure prophylaxis, the two excellent quality trials showed a relative efficacy of 81% (Monto, 2002) and 79% (Hayden, 2000), and an absolute risk reduction of 15% and 10% respectively. The one fair quality trial (Kaiser, 2000) showed no significant difference between the placebo and Relenza^® groups. This trial was identified as being “underpowered” (not enough people to show a significant difference) and only used post-exposure prophylaxis for 5 days. The number needed to treat to prevent one case of influenza was similar to oseltamivir: 6.6 B 10 people.

2.Effectiveness

Randomized controlled trials are done on a relatively small number of people under ideal conditions, and so are often followed up with “real-world” effectiveness studies. CADTH reported that 5 intervention studies were done on the neuraminidase inhibitors; it turned out one was a duplicate publication -- there were 3 on Tamiflu^®; one on Relenza^®. All these studies were done in long-term care facilities and described a combined approach of treatment of cases and prophylaxis of residents.

Of the three Tamiflu^® studies, only 2 reported on effectiveness. Parker et al reported on a single facility that used Tamiflu^® for outbreak control. It resulted in an overall attack rate of 10%. This was compared with a similar nearby facility that had an outbreak at approximately the same time and did not use Tamiflu^®; the attack rate was 19%. Bowles et al reported on outbreaks in 10 LTC facilities. Five used Tamiflu for prophylaxis, two did not meet inclusion criteria for outbreak evaluation and of the 3 that remained, all met the criteria for outbreak control that were established a priori. The drug was generally well-tolerated.

The Relenza trial was done in a LTC facility and the conclusion was that its use was “temporally associated with termination of the outbreak.” Of the 140 people who were offered prophylaxis, 129 accepted and 100 were able to take it, suggesting that in this population with high co-morbidity, slightly over 30% were either unwilling or unable to take the inhaler.

3.Efficiency

To assist in assessing the population-based implications of the use of antivirals for prophylaxis, TGAP calculated the “Number needed to treat” (or in this case, prophylax) to prevent one case. This number is often used in evidence-based decision making to assess the intensity of resources required to obtain a desired effect. It is calculated by simply assessing the inverse of the absolute risk reduction (see Table 3).

When applying the data from seasonal influenza, this calculation reveals a qualitative difference between pre-exposure and post-exposure prophylaxis. For seasonal influenza it shows that if one instituted a pre-exposure prophylaxis regime, one would expect to have to give 26.5 people a 6-week course of Tamiflu^® to prevent one case. If one instituted a post-exposure prophylaxis regime, one would expect to have to give 9 people a 10-day course to prevent one case.

Based on this, the number of doses of Tamiflu^® (or Relenza^®) to prevent one case can then be assessed. This would be:

<100 doses for post-exposure prophylaxis: (9 people for 10 days =90)

>1000 doses for pre-exposure prophylaxis: (26.5 people for 42 days or 6
weeks = 1113 doses)

It should be noted that the number of doses need to prevent one case of pandemic influenza would likely be less than seasonal influenza for both types of prophylaxis, as the clinical attack rate would be higher. However, it is also likely that post-exposure prophylaxis would continue to be more “efficient” than pre-exposure prophylaxis as post-exposure prophylaxis specifically targets those at high risk of developing infection.

4.Summary of efficacy, effectiveness and efficiency

In summary, efficacy data based on RCTs during seasonal influenza are adequate to support the use of Tamiflu^® and Relenza^® for both pre-exposure and post-exposure prophylaxis. Relative efficacy ranged from 60-90% for both types; absolute risk reduction was 1-4% for pre-exposure and 10-15% for post-exposure prophylaxis. There was no direct RCT evidence to support the current practice of using Tamiflu^® for LTC outbreaks. This practice is based on expert opinion drawing on both treatment and prophylactic trials done in other settings as well as effectiveness data from two case series that have reported a beneficial effect. Efficiency data, calculated by the number needed to prophylax to prevent one case of influenza, shows that for seasonal influenza, post-exposure prophylaxis is much more efficient than pre-exposure prophylaxis.

5.Safety

TGAP summarized the safety data for neuraminidase inhibitors, and safety concerns arising from its use in populations known to be vulnerable to influenza and its complications.

a) Adverse events

Neuraminidase inhibitors are generally described as a well-tolerated, and there is complete consensus that they have a better adverse event profile than amantadine. Nonetheless, these drugs are not devoid of adverse events or serious adverse reactions.

Table 4 identifies the adverse events reported in the product monographs and in the
published RCTs. Of note, is the under-reporting of adverse events in RCTs, which is now
well-documented.¹⁰ This is in part due to a lack of standardization of adverse event reporting.

The most frequent adverse event following the use of Tamiflu^® was nausea, occurring in about 7% of people taking a daily prophylactic dose (vs. 3.9% for placebo). Some studies note that this tends to occur initially and may subside. Diarrhea occurred in about 3% of people, vomiting and abdominal pain in about 2%—none of which were significantly higher than placebo.

During clinical trials serious adverse events are recorded in all arms of the trial and do not necessarily imply causality. The most common serious adverse events recorded were allergic and serious skin reactions such as anaphylaxis, toxic epidermal necrolysis, Stevens-Johnson Syndrome and erythema multiforme. Other serious adverse events reported include unstable angina, pseudomembranous colitis, pneumonia and peritonsillar abscess. The product monograph also notes:

AThere have been postmarketing reports (mostly from Japan) of self-injury and delirium with the use of Tamiflu^® in patients with influenza. The reports were primarily among pediatric patients. The relative contribution of the drug to these events is not known. Patients with influenza should be closely monitored for signs of abnormal behaviour throughout the treatment period.@

Relenza^® has a similar adverse event profile, with the exception that nausea is less frequent at 3% (about the same as placebo). However, there have been reports of people who have experienced bronchospasm and decline in respiratory function with Relenza^®, so it is not recommended for patients with severe underlying airways disease.

Neuraminidase inhibitors and vulnerable populations

It is important to keep in mind that scientific evidence of efficacy is always linked to a specific type of patient or population, thus, we cannot simply say that prophylaxis works, and assume it works for everyone. Most studies of prophylaxis have been done on healthy adults.

Table 5 lists populations identified in the Canadian Pandemic Influenza Plan for the Health Sector as being at risk of increased morbidity and mortality from influenza¹¹, and then identifies what is known about the use of neuraminidase inhibitors in these populations. Unfortunately, there are very little data available on the use of neuraminidase inhibitors in vulnerable populations, such as the very young, the elderly, immunocompromised individuals and pregnant women.

6.Compliance

Compliance is the extent to which people actually follow the instructions given for a therapeutic intervention. In a trial setting this could include withdrawing from the study or simply missing some doses of medication. A review of the reports on withdrawals from the neuraminidase inhibitor trials showed that there was premature discontinuation of the antivirals in 3% (Hayden 1999) to 6.5% (Peters 2001) of people in the prophylaxis group. Monto et al (2002) noted that over 10% missed at least one day of medications (60/553). Again, there was lack of standardization of reporting, with some not including any compliance data (Kaiser 2000). However, compliance in trials tends to be better than compliance in the real-world setting.

a) Compliance literature

To assess how compliance for prophylaxis might be during a pandemic, TGAP went to the literature on compliance for therapeutic regimes in general and then looked at some similar case scenarios to pandemic influenza in particular.

There is a large body of evidence looking at compliance. One summary concluded that compliance is a problem with all medications, with little relation to age, sex, race, or education and that adherence decreases as adverse events increase and as the duration of the regime increases. Unfortunately there is not convincing evidence that interventions to improve compliance are effective. McDonald et al did a scientific review of RCTs of interventions to enhance patient compliance to medication and found fewer than half the interventions tested were associated with any statistically significant difference, and even the most effective interventions did not lead to large improvements.¹² Intuitively, one would expect compliance would increase with increased perceived risk to self and to one’s close contacts. To get some indication anticipated compliance with antivirals during a pandemic, TGAP examined compliance rates for antibiotic prophylaxis of postal workers in the context of an anthrax exposure and antiviral prophylaxis of cullers during an avian influenza outbreak.

Antibiotic prophylaxis for anthrax exposure

In October 2001, there was an intentional release of anthrax through the postal system in the United States and 10,000 people in 6 US cities were potentially affected. The recommendation was to take at least 60 days of post-exposure prophylaxis with ciprofloxacin (an antibiotic). Interventions were carried out to promote compliance, and a number of follow-up studies were done.

In a study of the six US cities that experienced anthrax exposures, overall adherence to antibiotic prophylaxis was poor (44%) ranging from 21% at a postal facility in New York City to 64% of persons exposed at the Brentwood postal facility in Washington DC. Adherence was highest among participants on an investigational new drug protocol, confirming previous observations that compliance is better under the monitored conditions of a study.¹³

Antiviral prophylaxis during avian influenza outbreaks

Antivirals, specifically Tamiflu^®, have been used in recent years to prevent illness due to avian influenza outbreaks caused by H7N7, H7N3 and H5N1 strains. Anecdotally, it has been reported that compliance among health care workers caring for human H5N1 cases in Turkey was not as high as expected. This was attributed to cultural differences and to unfounded fear of adverse events.¹⁴

During the H7N7 outbreak of avian influenza in the Netherlands in 2003, it was estimated that at least one thousand persons were infected with the virus. In a report describing the outbreak, the authors concluded that neither poultry farmers nor those combating the epidemic complied satisfactorily with preventive measures.¹⁵ It was noted that oseltamivir prophylaxis was taken by 48% of the 185 poultry farmers on infected farms and by 75.5% of the 604 persons brought in to assist with outbreak control on infected farms. More than 2/3 of the persons examined who were taking antiviral prophylaxis had interrupted use at least once.

In 2004, when the H7N3 avian influenza outbreak struck British Columbia, public health authorities rapidly implemented a regimen of antiviral prophylaxis for all workers and farmers deemed to be at risk for avian influenza. Key findings regarding compliance with antiviral prophylaxis during this outbreak were that Tamiflu^® use was significantly greater among those who had direct contact with infected poultry (71% vs 13%), 64% of farmers citing direct contact with AI-infected birds used antivirals compared to 79% of workers involved in euthanizing poultry. While most people took antivirals, less then half of those who had direct contact with known AI-infected birds (only 44%) took Tamiflu^® every day while in contact with poultry.¹⁶

7.The risk of resistance

Currently, the risk of developing resistance to Tamiflu^® is low. This is because use is generally limited and appropriate. Tamiflu^® is used primarily in long term care facility outbreaks following laboratory confirmation; and compliance is high as the medications are often directly administered by health professionals. The strains that have developed resistance to Tamiflu^® are typically “unfit”, meaning their ability to spread from person to person and cause morbidity and mortality are low. Resistance to Tamiflu^® has been reported in 0.4 - 1% of adults¹⁷ and 4-18% of children.¹⁸ This difference may be due to some partial immunity in adults who have been exposed to similar influenza viruses in the past. Resistance to Tamiflu^® has been reported in H5N1 patients, associated with treatment failure and death.¹⁹ To date, no cases of resistance have been reported from prophylactic therapy.

Implications for a pandemic: Evidence and inference

It is important to be very clear: there is no direct evidence for the safety or efficacy of neuraminidase inhibitors during a pandemic. So what TGAP is doing is entering the realm of inference. Inference – or extrapolating from evidence – is in fact done all the time. But this must be done consciously, and with the awareness that some experts are more confident in extrapolating further than others. This explains why different experts can reach different conclusions based on the same evidence – because their inferences arising from those data are different. Here are the inferences TGAP has drawn.

1. Initially, it is highly probable that Tamiflu^® used either as pre-exposure prophylaxis or post-exposure prophylaxis will be efficacious against a pandemic virus.

Based on two good quality trials of pre-exposure prophylaxis, two good quality trials of post-exposure prophylaxis and the mechanism of action of neuraminidase inhibitors, TGAP can be fairly confident that, at least initially, Tamiflu^® will have efficacy as a prophylactic agent. Although this is widely accepted as a “given”, two trials are really minimally adequate evidence on which to base this inference. Nonetheless, it must be kept in mind that these trials were largely on healthy adults and children, therefore inference of efficacy will need to be stretched even more to include many of the vulnerable populations that are at increased risk of complications from influenza (as per Table 5).

2. It is likely that the efficiency of neuraminidase inhibitors will be better during a pandemic than during seasonal influenza.

The assessment of efficiency using number needed to prophylax to prevent one case is influenced by both the efficacy of the drug and the clinical attack rate. The clinical attack rate during a pandemic will, by definition, be higher than during seasonal influenza. No simple calculation can be done for this, as we cannot equate clinical attack rate with rates of symptomatic, lab-confirmed influenza in RCTs. However, we can be reasonably certain that fewer people would need to be prophylaxed to prevent one case and that post-exposure prophylaxis would remain more efficient than pre-exposure prophylaxis.

3. Overall, oseltamivir appears to be a safe drug, however, new adverse effects may arise or adverse effects that are known may appear more frequently during a pandemic.

Tamiflu^® is generally well-tolerated, serious side effects are rare, and in general, adverse events are less frequent during a daily prophylactic dose than in a twice daily treatment dose. However, due to the difficulties in standardizing and reporting safety data, and the relatively few trials that have been done that have collected safety data, it would not be surprising if new adverse events were noted or the known adverse events, such as nausea and abdominal upset, were more common than reported at this time. Serious adverse events, such as Stevens Johnson Syndrome, may occur with widespread use in Canada, and the risk of this in a well population taking prophylaxis, must be weighed against the likely benefit.

4. Compliance will likely be an issue.

The lack of compliance is well-known to occur with therapeutic regimes, to one degree or another. If the risk of anthrax did not result in high compliance with exposed postal workers; if the risk of H5N1 still resulted in only moderate compliance by poultry workers, and in the Netherlands, then it is reasonable to expect that variable compliance for a prophylactic regime may occur during an influenza pandemic. This is an issue for two reasons. First, it is likely that with variable compliance, “real world” effectiveness will not be as good as “best case” RCT efficacy. Second, incomplete compliance increases the risk of resistance. The greater the non-compliance, the greater the theoretical risk of developing resistance. Finally, the cost of prophylaxis—and the opportunity costs involved in increasing the NAS for the purposes of prophylaxis—must be weighed against the expected benefit that would arise from variable compliance with a preventive regime.

5. The risk of resistance will rise during a pandemic.

Although the current risk is low, the risk of resistance will rise with a pandemic, and will continue to rise over the course of a pandemic. There are several reasons for this:

a. Widespread use will put evolutionary pressure on the pandemic virus to develop mutations that confer drug resistance and provide a competitive advantage to a resistant strain that retains its transmissibility amongst humans.

b. Lack of pre-existing immunity will tend to lead to higher viral replication which may increase the chances of emergence of resistant mutants, thus one might expect a resistance rate more similar to that of children during the use of antivirals for treatment in annual epidemics (4-18% vs. 0.4-1%for adults) If a “fit” resistant strain develops to Tamiflu^® (i.e., easily transmissible), it will be equally applicable to both treatment and prophylaxis.

c. Variable compliance from missed doses or incomplete treatments will foster persistent viral replication and increase opportunity for resistant strains to emerge through mutations.

d. There is a potential that if people receive prophylaxis after they have been infected with the virus but are not yet symptomatic, they may be receiving a sub-therapeutic dose of antiviral medication. This could foster the emergence of a resistant strain. This is a theoretical risk that has not been documented in trials to date. Specifically, during post-exposure prophylaxis in household studies, the prophylaxis dosage is given under conditions where there may be viral replication in the close contacts, no evidence of resistant viruses have been found. However, this may in part be due to the better compliance seen under trial conditions. In the real world, if people on prophylaxis have been infected and then miss doses, or take it intermittently, they will be creating conditions that will foster resistant strains emerging.

David Heymann from the World Health Organization has noted that:

“Resistance to anti-infective drugs is an urgent public health problem threatening the treatment and control of infectious diseases...[including] those that pose pandemic threats such as the H5N1 strain of avian influenza.”²⁰

A recent modelling study has identified a concerning scenario if resistance develops in health care workers who are on pre-exposure prophylaxis. This could result in the rapid spread of drug resistance to patients and other health care workers due to frequent contacts of health care workers with both patients and staff in the health care setting.²¹ Multiple drug regimes are starting to be considered as a strategy against the development of resistance and developments in this area should be monitored.

Conclusions

There is minimally adequate, good quality RCT evidence that the neuraminidase inhibitors are efficacious in preventing seasonal influenza when taken for either pre-exposure or post-exposure prophylaxis. During a pandemic, adverse events and lack of compliance are to be expected, and the risk of antiviral resistance is likely to increase. As such, there is uncertainty as to how these studies, conducted under ideal conditions during seasonal influenza, will translate into "real world" effectiveness during a pandemic. A prudent approach to population-based use of these drugs for prophylaxis is merited. Strategies to avoid or moderate the risk of resistance, such as promoting education to optimize compliance, are advisable.

Further research is needed to assess the efficacy of neuraminidase inhibitors in vulnerable populations, especially the elderly and the immunocompromised. A review and critical appraisal of modelling studies on the use of antivirals during a pandemic may further inform population-based policy development. Monitoring trends in the use of multi-drug regimes to prevent development of resistance is indicated. Finally, as real time surveillance for adverse events will be of critical importance, this capacity needs to be established well in advance of a pandemic.

Table 3: Summary of Randomized Control Trial (RCT) evidence for the use of neuraminidase inhibitors to prevent influenza (high quality trials in bold)

Tamiflu^®= oseltamivir, Relenza^® = zanamivir, SLCI = symptomatic lab confirmed influenza

Study (Noting Jadad score, antiviral studied, author and year of publication)	Results= inc. of SLCI in prophylaxis & placebo groups	Relative efficacy (p value)	Absolute risk reduction	No. needed to treat to prevent
PRE-EXPOSURE Healthy people
1: Tamiflu®: Kashiwaga >00	1.3% T: 8.5% placebo	85% (p=0.003)	7.2%	1 / 7.2 = 13.9
4: Tamiflu®: Hayden >99 (6 weeks)	1.2% T; 4.8% placebo (n=1040)	74% (p<0.001)	3.6%	1 / 3.6 = 27.8
4: Relenza®: Monto >99 (4 weeks)	2% R; 6% placebo (n=553)	67% (p<0.001)	4%	1 / 4 = 25
People with health conditions
4: Relenza®:Campbell >02 (4 weeks)	<1% R; 1.3% placebo (n=1678)	84%(p<0.001)	1.1%	1 / 1.1 = 90.9
Retirement homes
4: Tamiflu®: Peters >01 (6 weeks)	0.4%T; 4.4% placebo (n=276)	91% (p=0.002)	4%	1 / 4 = 25
OUTBREAK: Long Term Care
4: Relenza®: Ambrosaitis >05 (14 days)	6% R, 9% placebo(n=242)	Not significant	not appl	not applicable
3: Relenza®: Gravenstein >05 (14 days)	3% R, 8% rimantadine(n=238)	61%(p=0.038)	5%	1 / 5 = 20
POST-EXPOSURE Household contacts
2: Tamiflu®: Hayden >04(10 days)	4%T; 13% exp tx(n=410)	68% (p=.0017)	15.1%	1 / 15.1 = 6.6
3: Tamiflu®: Welliver >01(7 days)	1.4%T;12.6% placebo(n=497)	89% (p<0.001)	11.2%	1 / 11.2 = 8.9
4: Relenza®: Hayden >00(10 days)	6% R; 16% placebo (N=414)	79% (p<0.001)	10%	1 / 10 = 10
4: Relenza®: Monto >02 (10 days)	4% R; 19% placebo(n=661)	81% (p<0.001)	15%	1 / 15 = 6.6
2: Relenza®: Kaiser >00:(5 days)	2% R, 6% placebo(N=144)	Not significant	not appl	not applicable

¹Relative Efficacy = Placebo - Px/ Placebo Absolute risk reduction = placebo-prophylaxis

“Number needed to treat” (prophylax) = 1/Absolute risk reduction

Table 4: Reported Adverse Events Following the Use of Neuraminidase Inhibitors

(“NR = not reported, SAEs = Serious Adverse Events)

Study	Common Adverse Events				SAEs
	Nausea	Vomiting	Abd pain	Diarrhea
Product Monograph
T= Tamiflu® (prophylaxis)	7%	2%	2%	3%	<1%
R= Relenza® (inhalations)	3%	1%	<1.5%	3%	<1%
Pre-exposure Healthy people
1. T: Kashiwagi, 2000	3.9%	1.3%	5.2%	4.5%	NR
2. T: Hayden 1999	12.1%	2.5%	NR	NR	NR
3. R; Monto, 1999	NR	NR	NR	<1%	<1%
People with health conditions
4. R: Campbell 2002	N&V 2%	N&V 2%	NR	2%	1%
Retirement homes
5. T: Peters 2001	3.3%	<2%	NR	4.3%	NR
6. T: Gravenstein 2000	NR	NR	NR	NR	NR
Outbreak LTC Residents
7. R: Ambrozaitis 2005	NR	NR	NR	NR	2.5%
8. R: Gravenstein 2005	NR	NR	NR	NR	1.7%
Post-exposure
9. T: Hayden, 2004	NR	NR	NR	NR	NR
10. T: Welliver, 2001	5.5%	NR	NR	NR	NR
11. R: Monto, 2002	NR	NR	NR	NR	<1%
12. R: Hayden, 2000	12.1%	2.5%	NR	NR	NR
13. R: Kaiser, 2000	NR	NR	NR	NR	NR

Table 5: Vulnerable populations and Antiviral Prophylaxis:

People at increased risk for influenza complications from CPIP (2004) Annex G: Clinical Care Guidelines and Tools, p. 17) compared with information in product monograph for Tamiflu^® (oseltamivir) and Relenza^® (zanamivir)

People at increased risk (Based on 2004 CPIP)	Tamiflu® Product Monograph	Relenza® Product Monograph
Age < 2 years	Use of Tamiflu® in children under 1 year of age is contraindicated.@	Relenza® is indicated... in adults and children 7 years and older...@
Age > 65 years	Efficacy of Tamiflu® in the treatment of elderly patients has not been evaluated.@	In 2 studies of Relenza® for prophylaxis in the nursing home setting, efficacy was not demonstrated.@
Pregnancy (2nd and 3rd trimester)	Insufficient data are available in pregnant women taking Tamiflu Tamiflu® to enable an evaluation of the potential for Tamiflu® to cause fetal malformations or fetal toxicity.@	The safe use of Relenza® during pregnancy has not been established. There are no adequate and well controlled studies of Relenza® in pregnant women@
Bronchopulmonary diseases: asthma, bronchitis, bronchiectasis, emphysema, cystic fibrosis	Efficacy of Tamiflu® in the treatment of subjects with chronic cardiac disease and/or respiratory disease has not been established... Safety has been demonstrated in elderly residents of nursing homes who took Tamiflu® for the prevention of influenza. Many of these individuals had cardiac and/or respiratory disease.@	There have been reports of patients being treated for influenza who have experienced bronchospasm and decline in respiratory function... there have been cases of respiratory arrest, including deaths, in which a contribution from Relenza® cannot be excluded.@
Cardiovascular disease	See above.	Not addressed.
Metabolic diseases: e.g. diabetes	Tamiflu®...contains sorbital... unsuitable for subjects with hereditary fructose intolerance.@	Not addressed.
Renal diseases	No dosing recommendation is available for patients with end stage renal disease and for patients with creatinine clearance < 10 mL/min@	Not addressed.
Malignancies	Not addressed.	Not addressed.
Immunodeficiency, AIDS, immunosuppression, transplant recipients	Efficacy of Tamiflu® for treatment or prevention of influenza in immunocompromised patients has not been established.@	Due to the limited number of... immunocompromised patients who have been treated, it has not been possible to demonstrate the efficacy and safety of Relenza® in these groups.@
Diseases of the blood, anemia, hemoglobinopathy, and malignancies	Not addressed.	Not addressed.

[Previous] [Table of Contents] [Next]