**Time Lines of Infection and Disease in Human Influenza: A Review of Volunteer Challenge Studies**

**<nobr>Fabrice Carrat<sup>1</sup><sup>,2</sup><sup>,3</sup></nobr>, <nobr>Elisabeta Vergu<sup>1</sup><sup>,2</sup><sup>,4</sup></nobr>, <nobr>Neil M. Ferguson<sup>5</sup></nobr>, <nobr>Magali Lemaitre<sup>1</sup><sup>,2</sup></nobr>, <nobr>Simon Cauchemez<sup>5</sup></nobr>, <nobr>Steve Leach<sup>6</sup></nobr> and <nobr>Alain-Jacques Valleron<sup>1</sup><sup>,2</sup><sup>,3</sup></nobr>**<sup>1</sup> Universit? Pierre et Marie Curie-Paris6, UMR-S 707, Paris, France

<sup>2</sup> INSERM, UMR-S 707, Paris, France

<sup>3</sup> Assistance Publique H?pitaux de Paris, H?pital Saint-Antoine, Paris, France

<sup>4</sup> Math?matiques et Informatique Appliqu?s Unit, Institut National de la Recherche Agronomique, Jouy-en-Josas, France

<sup>5</sup> Department of Infectious Disease Epidemiology, Faculty of Medicine, Imperial College London, London, United Kingdom

<sup>6</sup> Health Protection Agency, Centre for Emergency Preparedness and Response, Porton Down, Salisbury, Wiltshire, United Kingdom

Correspondence to Dr. F. Carrat, Epid?miologie, Syst?mes d'Information, Mod?lisation, UMR-S 707, Facult? de m?decine Saint-Antoine, 27, rue Chaligny, 75571, Paris Cedex 12, France (e-mail: carrat@u707.jussieu.fr<script type="text/javascript"><!-- var u = "carrat", d = "u707.jussieu.fr"; document.getElementById("em0").innerHTML = '<a href="mailto:' + u + '@' + d + '">' + u + '@' + d + '<\/a>'//--></script>).

*Received for publication September 20, 2007.*

*Accepted for publication November 30, 2007.*

<!-- null -->

<table bgcolor="#e1e1e1" cellpadding="0" cellspacing="0" width="100%"> <tbody><tr><td align="left" bgcolor="#ffffff" valign="middle" width="5%"></td> <th align="left" valign="middle" width="95%"> ABSTRACT </th></tr></tbody></table> <table align="right" border="1" cellpadding="5"><tbody><tr><th align="left"> TOP

ABSTRACT

INTRODUCTION

MATERIALS AND METHODS

RESULTS

DISCUSSION

References

</th></tr></tbody></table>

The dynamics of viral shedding and symptoms following influenza<sup> </sup>virus infection are key factors when considering epidemic control<sup> </sup>measures. The authors reviewed published studies describing<sup> </sup>the course of influenza virus infection in placebo-treated and<sup> </sup>untreated volunteers challenged with wild-type influenza virus.<sup> </sup>A total of 56 different studies with 1,280 healthy participants<sup> </sup>were considered. Viral shedding increased sharply between 0.5<sup> </sup>and 1 day after challenge and consistently peaked on day 2.<sup> </sup>The duration of viral shedding averaged over 375 participants<sup> </sup>was 4.80 days (95% confidence interval: 4.31, 5.29). The frequency<sup> </sup>of symptomatic infection was 66.9% (95% confidence interval:<sup> </sup>58.3, 74.5). Fever was observed in 37.0% of A/H1N1, 40.6% of<sup> </sup>A/H3N2 (

*p*= 0.86), and 7.5% of B infections (

*p*= 0.001). The<sup> </sup>total symptoms scores increased on day 1 and peaked on day 3.<sup> </sup>Systemic symptoms peaked on day 2. No such data exist for children<sup> </sup>or elderly subjects, but epidemiologic studies suggest that<sup> </sup>the natural history might differ. The present analysis confirms<sup> </sup>prior expert opinion on the duration of viral shedding or the<sup> </sup>frequency of asymptomatic influenza infection, extends prior<sup> </sup>knowledge on the dynamics of viral shedding and symptoms, and<sup> </sup>provides original results on the frequency of respiratory symptoms<sup> </sup>or fever.<sup> </sup>

influenza, human; signs and symptoms; virus shedding

Abbreviations: CI, confidence interval; GEE, generalized estimating equations; HAI, hemagglutination inhibition; SD, standard deviation

<!-- null -->

<table bgcolor="#e1e1e1" cellpadding="0" cellspacing="0" width="100%"> <tbody><tr><td align="left" bgcolor="#ffffff" valign="middle" width="5%"></td> <th align="left" valign="middle" width="95%"> INTRODUCTION </th></tr></tbody></table> <table align="right" border="1" cellpadding="5"><tbody><tr><th align="left"> TOP

ABSTRACT

INTRODUCTION

MATERIALS AND METHODS

RESULTS

DISCUSSION

References

</th></tr></tbody></table>

The threat of a human influenza pandemic has dramatically increased<sup> </sup>in recent years, and many countries have now developed pandemic<sup> </sup>preparedness plans following World Health Organization guidelines<sup> </sup>(1). Measures to reduce the spread of influenza within a given<sup> </sup>population, based on treatment or prophylaxis with antiviral<sup> </sup>medications, isolation, quarantine, or other social-distancing<sup> </sup>measures, are considered at various phases of the plans, as<sup> </sup>they might play a major role by reducing transmissibility. The<sup> </sup>effectiveness of these measures would depend greatly on the<sup> </sup>possibility of identifying infectious individuals and on how<sup> </sup>or when influenza virus is transmitted between individuals (2?5).<sup> </sup>

One critical question is whether the latent period overlaps<sup> </sup>the incubation period or, in other words, how onset of infectiousness<sup> </sup>overlaps onset of symptoms, if any (6). Another critical issue<sup> </sup>is the duration of infectiousness, which determines, among other<sup> </sup>things, the duration of treatment, prophylaxis, or isolation.<sup> </sup>

Influenza infectiousness is usually equated to the presence<sup> </sup>of virus shedding. A recent report from the World Health Organization<sup> </sup>(7) concluded that influenza virus shedding can be detected<sup> </sup>24?48 hours before clinical onset, and that it peaks during<sup> </sup>the first 24 hours of illness. Shedding usually lasts less than<sup> </sup>5 days, but it may be higher and longer in children. The incubation<sup> </sup>period is reported to average 2 days (range: 1?4 days).<sup> </sup>These data are derived from expert opinions or may have involved<sup> </sup>observational and experimental studies without any attempt to<sup> </sup>use systematic review; they are not supported by high-quality<sup> </sup>evidence.<sup> </sup>

The frequency of asymptomatic infection is also a critical parameter<sup> </sup>for interventions involving contact tracing. Modeling studies<sup> </sup>used frequencies of between 30 percent and 50 percent (2?5,<sup> </sup>8). However, these percentages come from pre- and post-influenza-season<sup> </sup>serologic studies, and they may therefore be subject to recall<sup> </sup>bias when individuals were asked whether they had had influenza-like<sup> </sup>illness during winter (9) or to classification bias due to lack<sup> </sup>of sensitivity of laboratory tests (10).<sup> </sup>

Experimental influenza virus infection of healthy volunteers<sup> </sup>provides a unique opportunity to describe the natural history,<sup> </sup>as 1) the date of infection is known with certainty, 2) shedding<sup> </sup>and symptoms are recorded prospectively, and 3) participants<sup> </sup>are usually selected with low pre-hemagglutination inhibition<sup> </sup>(HAI) antibody titers.<sup> </sup>

<!-- null -->

<table bgcolor="#e1e1e1" cellpadding="0" cellspacing="0" width="100%"> <tbody><tr><td align="left" bgcolor="#ffffff" valign="middle" width="5%"></td> <th align="left" valign="middle" width="95%"> MATERIALS AND METHODS </th></tr></tbody></table> <table align="right" border="1" cellpadding="5"><tbody><tr><th align="left"> TOP

ABSTRACT

INTRODUCTION

MATERIALS AND METHODS

RESULTS

DISCUSSION

References

</th></tr></tbody></table>

**Search strategy and identification of studies**

A literature search was carried out using the PubMed database,<sup> </sup>with the keywords ("human influenza" (text word)

*or*"influenza,<sup> </sup>human" (MeSH terms))

*and*("volunteer" (all fields)

*or*"experimental"<sup> </sup>(all fields)

*or*"deliberate infection" (all fields)

*or*"shedding"<sup> </sup>(all fields)

*or*"symptoms" (all fields)). We limited our search<sup> </sup>to English-language papers published between 1965 and 2005.<sup> </sup>A total of 827 papers were selected (figure 1). We included<sup> </sup>any study with any design in which a subgroup of participants<sup> </sup>was challenged with a wild-type influenza virus and for which<sup> </sup>there was at least one type of outcome measure, that is, viral<sup> </sup>shedding or symptoms. We identified additional articles by searching<sup> </sup>the reference lists of articles. We also made a hand search<sup> </sup>in textbooks on influenza. We did not specifically consult world-leading<sup> </sup>specialists on influenza, but a bibliographic search on their<sup> </sup>names was performed. No attempt was done to retrieve primary<sup> </sup>data from the original studies. Two of us (F. C., M. L.) read<sup> </sup>all the studies retrieved in the search and applied the inclusion<sup> </sup>criteria. Differences were resolved by discussion and consensus.<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View larger version**(11K):

<nobr>[in this window]

[in a new window]

[Download PowerPoint slide]

</nobr> </td><td align="left" valign="top"> FIGURE 1. Identification of eligible articles.

</td></tr></tbody></table> </td></tr></tbody></table></center>

**Data abstraction**

For each paper, we collected information on the year of publication,<sup> </sup>the number of study subgroups, the participants' characteristics,<sup> </sup>the number of participants, the type and subtypes of wild-type<sup> </sup>influenza virus used for challenge, the route of inoculation,<sup> </sup>the inoculated dose expressed in median tissue-culture infective<sup> </sup>doses, the duration of follow-up, and a summary of how clinical<sup> </sup>and virologic data were collected (refer to "Supplementary Material").<sup> </sup>(This information is posted on the

*Journal*'s website (http://aje.oxfordjournals.org/).)<sup> </sup>Note that one study can involve several subgroups of volunteers<sup> </sup>challenged with different influenza viruses and, conversely,<sup> </sup>several articles can describe the identical subgroup of volunteers<sup> </sup>but different outcomes (e.g., viral shedding or illness). Influenza<sup> </sup>virus infection was defined as a greater than fourfold rise<sup> </sup>in pre-HAI antibody titers or viral shedding (positive nasal<sup> </sup>wash cultures) at least 1 day after inoculation.<sup> </sup> We extracted the following effect measures for each subgroup:<sup> </sup>proportion of infected individuals among those challenged, proportion<sup> </sup>of infected participants who shed virus (positive nasal wash<sup> </sup>on at least one occasion at least 1 day after inoculation),<sup> </sup>duration of viral shedding (time from inoculation to the first<sup> </sup>negative nasal wash with no subsequent positive washes), and<sup> </sup>proportion of infected participants who developed symptoms (any,<sup> </sup>systemic, or fever, respiratory, or nasal symptoms). We also<sup> </sup>described the dynamics of viral shedding, expressed in terms<sup> </sup>of the log-scale viral titer, and the dynamics of symptoms.<sup> </sup>Because various methods were used for scoring of symptoms, we<sup> </sup>normalized each study curve to its maximum clinical score of<sup> </sup>signs and symptoms. Summary curves were calculated as the weighted<sup> </sup>average of curves with weights equal to the number of individuals<sup> </sup>who were considered in each study curve. A further measure of<sup> </sup>shedding of interest is the first moment of the viral shedding<sup> </sup>curve. Recent work that estimated the generation time of influenza<sup> </sup>from household study data also showed that the resulting generation<sup> </sup>time distribution was quantitatively similar to viral shedding<sup> </sup>curves from experimental infection studies in human volunteers<sup> </sup>(4). This result is supportive of the hypothesis that infectiousness<sup> </sup>is proportional to viral shedding. Under this hypothesis, the<sup> </sup>average delay from a person's being infected to that individual's<sup> </sup>infecting other people, that is, the generation time (

*T<sub>g</sub>*), can<sup> </sup>be calculated as<sup> </sup>

<table border="0" width="100%"><tbody><tr><td align="center"> </td></tr></tbody></table>where

*V*(

*t*) is the absolute level of viral shedding at time<sup> </sup>

*t*postinfection.<sup> </sup>

**Statistical analysis**

Summary effect measures were calculated. For summary means,<sup> </sup>a random-effect model was used, and the effect of covariates<sup> </sup>was assessed using the chi-square test for heterogeneity (11).<sup> </sup>For summary proportions, we used a binomial generalized estimating<sup> </sup>equations (GEE) model with an exchangeable correlation structure<sup> </sup>(12). The "clustered" effect was subgroup defined as the set<sup> </sup>of individuals challenged with the same influenza strain in<sup> </sup>each study. The GEE model gives population-averaged parameters,<sup> </sup>and it has been used for meta-analysis of rates (13, 14). The<sup> </sup>influence of a covariate on the effect measure was tested with<sup> </sup>the Wald chi-square test by introducing the covariate as a predictor<sup> </sup>of the effect measure in the GEE model. For all comparisons<sup> </sup>involving the type or subtype of influenza virus, A/H1N1 was<sup> </sup>chosen as the reference group. Statistical tests were two tailed,<sup> </sup>with a type I error risk of 5 percent.<sup> </sup>

<!-- null -->

<table bgcolor="#e1e1e1" cellpadding="0" cellspacing="0" width="100%"> <tbody><tr><td align="left" bgcolor="#ffffff" valign="middle" width="5%"></td> <th align="left" valign="middle" width="95%"> RESULTS </th></tr></tbody></table> <table align="right" border="1" cellpadding="5"><tbody><tr><th align="left"> TOP

ABSTRACT

INTRODUCTION

MATERIALS AND METHODS

RESULTS

DISCUSSION

References

</th></tr></tbody></table>

**Search results**

Seventy-one papers describing 56 different studies, 79 different<sup> </sup>subgroups, and 1,280 different participants were considered.<sup> </sup>In all the studies, the participants were young adults aged<sup> </sup>between 18 and 40 or 50 years, except in one study (15), where<sup> </sup>the age ranged up to 65 years. A total of 199 (12 subgroups)<sup> </sup>participants had pre-HAI antibody titers to the challenge strain<sup> </sup>of at least 1/16 or were not selected according to their pre-HAI<sup> </sup>antibody titers, and these subjects were excluded from summary<sup> </sup>analyses of viral shedding or illness; 1,081 participants (67<sup> </sup>subgroups) had a pre-HAI antibody titer that was considered<sup> </sup>unprotective (<1/16). A total of 532 volunteers were challenged<sup> </sup>with an A/H1N1 virus, 473 with an A/H3N2 virus, 86 with an A/H2N2<sup> </sup>virus, and 189 with a type B virus. The routes of challenge<sup> </sup>were intranasal instillation in most studies. Throat sprays<sup> </sup>were also used in three studies (16?18), and aerosol inhalation<sup> </sup>was used in one study (19). The inoculum ranged between three<sup> </sup>and 7.2 log<sub>10</sub> median tissue-culture infective doses. Most papers<sup> </sup>reported ethics committee approval or collection of written,<sup> </sup>informed consent from each participant. In almost all studies,<sup> </sup>participants were individually confined for 1 week. Most studies<sup> </sup>included daily follow-up with daily nasal washing and collection<sup> </sup>of clinical signs and symptoms. The follow-up period ranged<sup> </sup>from 3 days prior to inoculation to 14 days after inoculation.<sup> </sup>

**Infection and viral shedding**

The overall proportion of influenza virus infection in individuals<sup> </sup>with pre-HAI antibody titers of <1/16 was 88.2 percent (95<sup> </sup>percent confidence interval (CI): 83.9, 91.4) (refer to Supplementary<sup> </sup>Material). Viral shedding was found in 93.1 percent of participants<sup> </sup>infected with A/H1N1, 92.5 percent with A/H3N2 (

*p*= 0.71 vs.<sup> </sup>A/H1N1), and 83.9 percent with A/H2N2 virus (

*p*= 0.14) and was<sup> </sup>lower in participants infected with a type B virus (81.5 percent,<sup> </sup>

*p*= 0.014) (table 1).<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View this table:**

<nobr>[in this window]

[in a new window]

</nobr> </td><td align="left" valign="top"> TABLE 1. Frequency of viral shedding (positive nasal wash on at least one occasion at least 1 day after inoculation) in healthy volunteers with low pre-hemagglutination inhibition antibody titers after experimental influenza virus infection<sup>*</sup>

</td></tr></tbody></table> </td></tr></tbody></table></center>

**Dynamics of viral shedding in infected volunteers**

The dynamics of viral shedding are summarized in figure 2. The<sup> </sup>A/H1N1 and A/H3N2 curves showed a sharp increase during the<sup> </sup>first day following inoculation, and they reached their maximum<sup> </sup>values during the second day. Return to baseline values was<sup> </sup>obtained by day 8. The summary curves did not differ markedly<sup> </sup>according to influenza virus type or subtype, although A/H3N2<sup> </sup>infections gave sustained high viral titers by comparison with<sup> </sup>A/H1N1.<sup> </sup> <!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View larger version**(11K):

<nobr>[in this window]

[in a new window]

[Download PowerPoint slide]

</nobr> </td><td align="left" valign="top"> FIGURE 2. Summary curves of viral shedding in experimental influenza virus infection, according to the virus type or subtype. Eight curves (116 participants who shed influenza virus) for A/H1N1 subtype (21, 22, 24, 26, 32, 74?76) and four curves (41 participants) for A/H3N2 subtype (25, 28, 40, 45) were averaged, and one curve (eight participants) was plotted for the B type (24). Bold curves correspond to weighted averages of study curves (standard error).

</td></tr></tbody></table> </td></tr></tbody></table></center>

On average, viral shedding was detected 1 day after inoculation.<sup> </sup>The distribution of the first observation of shedding was day<sup> </sup>1 in 64 (83 percent) participants, day 2 in 11 (14 percent)<sup> </sup>participants, and day 3 in two (3 percent) participants (1.1<sup> </sup>days, on average) (20). After challenge with A/H2N2 virus, four<sup> </sup>(40 percent) of the volunteers who shed the virus had culture-positive<sup> </sup>nasal washes by day 1 and 100 percent by day 3 (16). In two<sup> </sup>other studies, 14 (74 percent) (21) and 12 volunteers (86 percent)<sup> </sup>(22) had positive viral culture on the first day after inoculation.<sup> </sup>In the latter study, however, viral cultures became positive<sup> </sup>on days 4 and 5 in two volunteers.<sup> </sup> When calculating the duration of viral shedding, we excluded<sup> </sup>three studies because of missing or inconsistent standard deviation<sup> </sup>values (23?25), and 23 study subgroups (375 participants)<sup> </sup>were considered. The mean duration of viral shedding was 4.80<sup> </sup>days (95 percent CI: 4.31, 5.29) and did not differ according<sup> </sup>to the influenza virus types or subtypes: 4.50 days (95 percent<sup> </sup>CI: 3.71, 5.28) for type A/H1N1, 5.14 days (95 percent CI: 4.48,<sup> </sup>5.80) for type A/H3N2 (

*p*= 0.22, random-effect model), and 3.70<sup> </sup>days (95 percent CI: 1.73, 5.66) for type B virus (

*p*= 0.46).<sup> </sup>

Regarding the maximum duration of viral shedding, most volunteers<sup> </sup>had stopped shedding virus by day 6 or 7 (26, 27). However,<sup> </sup>longer durations are not rare: In one study subgroup, five (20<sup> </sup>percent) participants still shed influenza B virus on day 8<sup> </sup>after inoculation (15), and durations of A/H3N2 viral shedding<sup> </sup>ranging up to 9 days have been reported (28). In another study,<sup> </sup>three (30 percent) participants shed A/H2N2 virus until day<sup> </sup>10 (16). In this latter case, however, the results were controversial:<sup> </sup>Volunteers were placed in isolation in groups of three, so that<sup> </sup>reinfection cannot be excluded.<sup> </sup>

The mean generation time calculated from viral shedding curves<sup> </sup>was 2.3 days (range: 1.5?2.7 days) for type A/H1N1, 3.1<sup> </sup>days (range: 2.2?4.0 days) for type A/H3N2, and 3.4 days<sup> </sup>for type B virus. Across all studies, the mean generation time<sup> </sup>was 2.5 days.<sup> </sup>

**Factors influencing viral shedding**

There are few studies of factors associated with viral shedding.<sup> </sup>A dose-ranging study showed that the duration of shedding was<sup> </sup>proportional to the intranasal dose (29). Several studies suggested<sup> </sup>that volunteers were partially immune to the contemporary strain<sup> </sup>even though they had a low level of HAI antibodies in their<sup> </sup>serum before challenge (23, 30), thus explaining why some infected<sup> </sup>volunteers shed a small quantity of wild-type virus over a short<sup> </sup>period. This was supported by a significant difference in preexposure<sup> </sup>HAI antibody titers between participants with influenza virus<sup> </sup>infection who shed (average log titer = 0.7 (standard deviation<sup> </sup>(SD): 0.2)) and those who did not (average log titer = 1.5 (SD:<sup> </sup>0.4)) (31).<sup> </sup>

**Clinical illness**

**Any symptoms.**

Thirty-eight subgroups (522 infected individuals) were considered<sup> </sup>(table 2). The proportion of symptomatic infection (any symptoms)<sup> </sup>was 66.9 percent (95 percent CI: 58.3, 74.5). No significant<sup> </sup>difference was noted according to the virus type (refer to table 2<sup> </sup>for

*p*values) or the initial infectious dose (

*p*= 0.12).<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View this table:**

<nobr>[in this window]

[in a new window]

</nobr> </td><td align="left" valign="top"> TABLE 2. Proportion of volunteers who developed clinical illness after experimental influenza virus infection<sup>*</sup>

</td></tr></tbody></table> </td></tr></tbody></table></center>

**Respiratory symptoms.**

Upper respiratory symptoms, defined as nasal stuffiness, runny<sup> </sup>nose, sore throat, sneezing, hoarseness, ear pressure, or earache,<sup> </sup>were most frequent. The proportion of upper respiratory symptoms<sup> </sup>was 58.8 percent (95 percent CI: 45.5, 70.8) (table 3). A lower<sup> </sup>proportion was noted among participants infected with A/H3N2<sup> </sup>virus compared with those infected with A/H1N1 virus, but the<sup> </sup>difference must be interpreted with care as only three small<sup> </sup>subgroups were considered for A/H3N2 infections.<sup> </sup> <!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View this table:**

<nobr>[in this window]

[in a new window]

</nobr> </td><td align="left" valign="top"> TABLE 3. Proportion of volunteers who developed upper respiratory tract illness after experimental influenza virus infection<sup>*</sup>

</td></tr></tbody></table> </td></tr></tbody></table></center>

Lower respiratory symptoms were defined as cough, breathing<sup> </sup>difficulty, and chest discomfort. Frequencies of lower respiratory<sup> </sup>symptoms were reported in six subgroups (four A/H1N1, one A/H3N2,<sup> </sup>one B; 119 infected participants). The proportion of lower respiratory<sup> </sup>symptoms was 21.0 percent (95 percent CI: 14.0, 30.3) and did<sup> </sup>not differ between virus types and subtypes or according to<sup> </sup>the inoculated dose.<sup> </sup>

**Fever.**

Defined as a temperature of 100?F or 37.8?C or above,<sup> </sup>fever was reported in 34.9 percent (95 percent CI: 26.7, 44.2)<sup> </sup>of infected individuals (table 4). A lower proportion of fever<sup> </sup>in influenza B infection and a higher proportion in A/H2N2 infection<sup> </sup>were found as compared with A/H1N1 infections. A negative link<sup> </sup>was found between the dose and the proportion with fever (per<sup> </sup>log<sub>10</sub> median tissue-culture infective dose increase: odds ratio<sup> </sup>= 0.56, 95 percent CI: 0.42, 0.73;

*p*< 0.001).<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View this table:**

<nobr>[in this window]

[in a new window]

</nobr> </td><td align="left" valign="top"> TABLE 4. Proportion of volunteers who had fever (>100?F or >37.8?C) after experimental influenza virus infection<sup>*</sup>

</td></tr></tbody></table> </td></tr></tbody></table></center>

**Ear symptoms.**

Otologic manifestations were frequently reported. Ear pressure<sup> </sup>abnormalities were observed in 33?73 percent of 26 infected<sup> </sup>placebo recipients (32?35), and earache has been reported<sup> </sup>in 33?47 percent of adults (33). In one study, four participants<sup> </sup>(15 percent) developed signs of otitis media between days 5<sup> </sup>and 7 after infection (36).<sup> </sup>

**Dynamics of symptoms**

An increase in the average total symptoms score was noted by<sup> </sup>day 1 after inoculation in A/H1N1 and A/H3N2 infections (figure 3).<sup> </sup>Total scores peaked by day 2 or day 3 and returned to baseline<sup> </sup>values by day 8. Individual incubation times were reported in<sup> </sup>16 men who developed febrile illness after being inoculated<sup> </sup>with A/Bethesda/10/63 (H2N2) virus (37): Three men had an incubation<sup> </sup>period of 1 day, nine of 2 days, and four of 3 days (average:<sup> </sup>2 days). In another subgroup, illness began an average of 1.7<sup> </sup>days after challenge (38).<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View larger version**(11K):

<nobr>[in this window]

[in a new window]

[Download PowerPoint slide]

</nobr> </td><td align="left" valign="top"> FIGURE 3. Summary curves of total symptoms scores in experimental influenza virus infection, according to the virus type or subtype. Seven curves (134 infected participants) for A/H1N1 subtype (21, 22, 26, 76, 77) and eight curves (68 participants) for A/H3N2 subtype (18, 25, 40, 43, 45, 48, 78) were averaged, and one curve (11 participants) for A/H2N2 subtype (16) and one curve (15 participants) for B type (17) were plotted. A total score of 1 corresponds to the maximum reported score value (refer to Materials and Methods).

</td></tr></tbody></table> </td></tr></tbody></table></center>

Systemic symptoms (fever, muscle aches, fatigue, headache) peaked<sup> </sup>earlier, by day 2 after inoculation, and resolved faster than<sup> </sup>respiratory or nasal symptoms (figure 4).<sup> </sup> <!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View larger version**(11K):

<nobr>[in this window]

[in a new window]

[Download PowerPoint slide]

</nobr> </td><td align="left" valign="top"> FIGURE 4. Summary curves of systemic symptoms (fever, muscle aches, fatigue, headache), respiratory symptoms, or nasal symptoms scores. Seven curves (159 infected participants) were considered for the systemic scores (20, 34, 74, 79?82), five curves (132 participants) for the nasal scores (20, 34, 79?81), and two curves (28 participants) for the respiratory scores (28, 75). A score of 1 corresponds to the maximum reported score value (refer to Materials and Methods).

</td></tr></tbody></table> </td></tr></tbody></table></center>

The mean duration of illness was rarely reported. The mean duration<sup> </sup>of illness was 4.4?5 days in 25 participants infected<sup> </sup>with A/H1N1 virus (34, 39), 3.7 days in seven participants challenged<sup> </sup>with A/H3N2 virus (40), 4.6 days in 13 participants infected<sup> </sup>with A/H2N2 virus (37), and 4.1 days in seven participants challenged<sup> </sup>with B virus (41).<sup> </sup>

**Factors influencing clinical illness**

The proportion and duration of illness were lower in the case<sup> </sup>of elevated pre-HAI titers. In two different studies with A/H3N2<sup> </sup>virus challenge, the pooled proportions of illness (any symptoms)<sup> </sup>were 57 percent (20 of 34) in participants with pre-HAI titers<sup> </sup><1/12, 52 percent (15 of 29) in participants with pre-HAI<sup> </sup>titers between 1/12 and 1/24, and 15 percent (two of 13) in<sup> </sup>participants with pre-HAI titers >1/24 (42, 43) (

*p*= 0.015;<sup> </sup>Cochran-Armitage chi-square for a trend). The mean duration<sup> </sup>of illness was 4.4 days (SD: 1.8) in participants with pre-HAI<sup> </sup>titers of 1/8 versus 1.0 day (SD: 1.4) in those with pre-HAI<sup> </sup>titers of >1/8 challenged with a wild-type A/H1N1 virus (34).<sup> </sup>

**Relation between viral shedding and illness**

Figure 5 describes the summary curves of viral shedding and<sup> </sup>total symptoms scores averaged over all influenza types and<sup> </sup>subtypes. The two curves showed similar shapes although viral<sup> </sup>shedding preceded illness by 1 day.<sup> </sup>

<!-- null -->

<center><table cellpadding="0" cellspacing="0" width="95%"><tbody><tr bgcolor="#e1e1e1"><td><table cellpadding="2" cellspacing="2"> <tbody><tr bgcolor="#e1e1e1"><td align="center" bgcolor="#ffffff" valign="top">

**View larger version**(12K):

<nobr>[in this window]

[in a new window]

[Download PowerPoint slide]

</nobr> </td><td align="left" valign="top"> FIGURE 5. Summary curves of viral shedding and total symptoms scores in experimental influenza virus infection. Thirteen curves were used for viral shedding (refer to figure 2 legend), and 17 curves were used for total symptoms scores (refer to figure 3 legend).

</td></tr></tbody></table> </td></tr></tbody></table></center>

There is limited information on viral shedding in volunteers<sup> </sup>who did not develop clinical illness. In one study, three participants<sup> </sup>infected with a A/H3N2 virus who did not develop clinical illness<sup> </sup>excreted the virus, but the quantity of shedding was not available<sup> </sup>(44). We found only two studies using A/H3N2 viruses. The mean<sup> </sup>quantity of virus in nasal wash fluids from volunteers who shed<sup> </sup>virus and developed illness (

*n*= 11) was from two log<sub>10</sub> to three<sup> </sup>log<sub>10</sub> times higher than in individuals who did not develop illness<sup> </sup>(

*n*= 14), and a positive correlation was found between the mean<sup> </sup>quantity of virus per positive specimen and severity of illness<sup> </sup>(45, 46).

<sup> </sup><!-- null -->

## Comment