On September 27, 2009, after the hospital infection prevention office at PHSKC was notified about a retreat participant with acute influenza A, all HCP who had attended the work retreat were contacted. HCP who were scheduled to be working were assessed to determine the presence of respiratory symptoms. All HCP with any respiratory symptoms were tested for influenza infection by direct fluorescent antibody (DFA) testing or real-time RT-PCR at Hospital A; negative DFA results were confirmed by real-time RT-PCR (5). All viruses were subtyped.

The retreat occurred from 8:00 am to 5:00 pm daily for 5 days, with optional evening social activities. Forty-six persons participated in retreat activities. Fourteen persons were facilitators for the retreat, did not attend all of the retreat events, and typically participated for <4 hours/day when attending. The remaining 32 core participants attended all 5 days of the retreat and participated in evening activities. Most core participants also stayed overnight in 1 cabin on Thursday, September 24.

On Friday, September 25, the last day of the retreat, the first retreat participant with documented illness experienced a cough. The following day, this person had fever and was DFA positive for influenza A. By Monday, September 28, active surveillance of the 46 retreat attendees by Hospital A staff identified 20 persons with respiratory symptoms; 19 were core participants and 1 was a facilitator. Influenza A virus was detected by DFA or real-time RT-PCR in 17 symptomatic retreat attendees; all were core participants. All viruses from these 17 persons were confirmed by real-time RT-PCR to be pandemic (H1N1) 2009 virus. Thus, of the 32 core participants at the retreat, 17 (53%) were virus positive; 7 of the 17 had measured fever and cough or sore throat. The other 3 participants with respiratory symptoms were virus negative; all were afebrile.

All HCP who were virus positive were treated with oseltamivir, 75 mg orally 2×/d for 5 days, within 48 hours of symptom onset. In accordance with CDC recommendations at that time, ill HCP were excluded from work for 7 days after symptom onset. Core participants and facilitators who did not become symptomatic or positive for influenza A were administered antiviral chemoprophylaxis (oseltamivir, 75 mg orally, 1×/d for 10 days) and were not excluded from work.

Questionnaires were completed by 45 (98%) of 46 retreat attendees, including 16 of 17 persons who were positive for pandemic (H1N1) 2009 virus infection. Sixteen infected HCP provided serial nasal wash samples and completed a symptom log for the duration of their illness. One virus-infected participant did not complete a questionnaire but did provide serial nasal wash specimens. Another participant who was virus positive completed a questionnaire but did not provide serial nasal wash specimens.

HCP infected with pandemic (H1N1) 2009 virus ranged in age from 26 to 33 years (median 28.5 years). Sixteen (94%) infected HCP were women compared with 67% of core attendees who did not meet the case definition (p = 0.06). Three (18%) infected HCP had an underlying medical condition compared with 1 (7%) of 15 core attendees who did not meet the case definition (p = 0.60). Of those infected HCP who completed the questionnaire, cough, myalgias, and headache were the most commonly reported symptoms (Table), with >2 of these symptoms reported by all 16 HCP, and all 3 symptoms reported by 13 (81%) HCP. According to questionnaire data or employee health records of all 17 infected HCP, fever >100.5°F was measured by 7 (41%) infected HCP, and an additional 5 (29%) infected HCP reported subjective fever. HCP who reported fever typically documented the fever only during the first 1–2 days of illness, and no one reported measuring an oral temperature >100.5°F for >4 days. All HCP reported completing antiviral treatment as prescribed, and no deaths or hospitalizations occurred.

Among 7 HCP who measured a fever >100.5°F during their illness, 5 (71%) were virus positive by real-time RT-PCR and 3 (43%) were positive by rapid culture >24 hours after defervescence, when they met 2009 CDC criteria for returning to work. Of the 9 HCP who did not measure a fever >100.5°F during their illness and therefore did not meet CDC exclusion criteria, 7 (78%) were positive by real-time RT-PCR; 6 (67%) were also positive by rapid culture at the time of collection of the initial serial nasal wash sample >3 days after symptom onset. Samples from 12 (75%) of 16 HCP (95% confidence interval 48%–93%) were positive by real-time RT-PCR, and samples from 9 (56%) of 16 HCP (95% confidence interval 30%–80%) were positive by rapid culture at the time they met 2009 CDC criteria for returning to work (24 hours after defervescence). From the onset of symptoms, the duration of viral shedding determined by real-time RT-PCR results ranged from 3 to 13 days compared with 3–10 days by rapid culture results. Among 7 infected HCP, virus was detected by real-time RT-PCR and rapid culture for the same number of days. Among 8 HCP, virus was detected ≈2 days longer by real-time RT-PCR than by rapid culture. In 1 HCP, virus was detected by rapid culture 3 days longer than by real-time RT-PCR.

Figure 1

Figure 1. Survival analysis model of pandemic (H1N1) 2009 virus shedding over time among infected health care personnel, Seattle, Washington, USA. Survival curves were modeled on data for 16 persons who became infected...

In our initial model using data from all 16 participants, which included test type (real-time RT-PCR vs. rapid culture), presence of fever (measured and subjective), and viral load as predictors of shedding duration, test type was the only significant variable (p = 0.02). Our results indicated no association between fever (measured and subjective) and shedding duration (p = 0.65) or between viral load and shedding duration (p = 0.74). Our final model, which included only test type as a predictor of shedding duration, indicated that shedding duration was significantly different between rapid culture and real-time RT-PCR (Figure 1). Shedding duration measured by rapid culture was 0.73× higher than shedding duration measured by real-time RT-PCR (p = 0.02). Eight of 16 HCP remained positive by rapid culture 4.8 days from symptom onset compared with 6.6 days by real-time RT-PCR.

Figure 2

Figure 2. Virus RNA concentrations over time among health care personnel infected with pandemic (H1N1) 2009 virus, Seattle, Washington, USA. Each colored line represents a virus RNA concentration for an infected person tested...

An overall decrease in viral RNA concentration with time was measured and quantified by real-time RT-PCR (Figure 2). The person with the highest concentration of viral RNA on the first day of serial sampling also had one of the longest shedding durations by real-time RT-PCR. However, viral load was not a significant predictor of shedding duration in our analysis. Measured fever and subjective fever were not correlated with viral load (p = 0.41 and 0.12, respectively).

We describe a pandemic (H1N1) 2009 outbreak with a high attack rate that involved relatively young, healthy HCP after a hospital-associated retreat. Because of the single source of infection in an enclosed environment, we were able to prospectively monitor study participants to document duration of clinical symptoms and laboratory parameters. Infected HCP often did not report a fever during their illness. Virus infection was documented by real-time RT-PCR and rapid culture for prolonged periods, even though all infected HCP were treated with oseltamivir within 24–48 hours of illness onset. The shedding duration measured by real-time RT-PCR was significantly longer than that measured by rapid culture.

A total of 17 (53%) of 32 participants who attended all 5 days of the retreat became infected with virus. This rate of secondary transmission among core attendees is higher than that of other reports of secondary attack rates, which have been <35% (2,8–10). The high attack rate we observed might be explained by a susceptible population without immunity to pandemic (H1N1) 2009 virus, active surveillance for infection, or by prolonged close contact among participants, including sharing a cabin. Although viral load of the index case on day 1 of illness was not quantified, the initial viral load (5.03 log₁₀ RNA copies/mL) in this person on day 5 of illness was higher than the day 5 mean and median viral loads reported in other studies (4,11,12), which suggested that a high viral load might have contributed to the high attack rate in this outbreak. These findings highlight the potential for rapid spread of influenza among populations with ongoing close contact. In the absence of an available vaccine, handwashing and respiratory hygiene practices (e.g., cough etiquette and exclusion or isolation) should be emphasized in settings where persons might be in prolonged close contact with one another (e.g., hospitals, schools, or shelters).

This study had several limitations. Although all persons received standardized instructions to minimize variation in collection methods, nasal washes were self-collected and had the potential for variability in collection technique. Self-collection of nasal washes might have led to underestimation of presence or quantity of influenza virus. Other studies have demonstrated increased sensitivity with nasal washes, compared with nasal swabs, in respiratory virus detection (4,13), and we believe that self-collection of nasal washes likely increased compliance with study procedures.

For logistical reasons, persons were not tested daily, which affected our ability to more precisely define duration of shedding. We chose to use an accelerated failure time model with a Weibull survival function in our analysis because this method accounts for interval censoring. However, our model did not account for the paired nature of the data, in that each sample was tested by rapid culture and real-time RT-PCR. Our unpaired analysis likely led to an overly conservative p value, indicating that a statistically significant p value in our analysis also might have been detected by an analytic technique that accounts for pairing. The small sample size of this study might have decreased the power to detect associations observed in previous studies even if they existed, such as an association between fever and shedding duration or fever and viral load (12).

In addition, viral load was not measured on the day of symptom onset in our study but began on the first day that the participant provided serial nasal wash specimens. This delayed measurement might have affected the ability to detect associations between viral load and shedding duration and between fever and viral load. Furthermore, our findings should be considered in light of the fact that all participants were treated with antiviral medication, which might have affected relationships between symptoms, shedding duration, and viral load. Previous research has shown that oseltamivir reduces the duration of symptoms and influenza virus shedding (14,15). We did not prospectively ask about the use of fever-reducing medications among infected HCP, and therefore we were unable to definitively assess the potential effect of their use on our findings. Considering that in 6 of 7 HCP who had measured fever during their illness, the fever had resolved at least 36 hours before testing was performed, we do not believe that prospective assessment of the use of fever-reducing medications would have changed our conclusions.

Most characteristics of infection were similar to those of previous reports of pandemic (H1N1) 2009 virus infections (16–18), including an outbreak among HCP in Kenya (8). Approximately one third of infected HCP in our study did not have measured or subjective fever during their infection. This observation is consistent with those of previous reports (19,20). Those HCP who had an oral temperature >100.5°F typically documented fever only during the first 1–2 days of infection, although this finding might be explained by oseltamivir treatment that resulted in a shorter duration of symptoms. Fever has been demonstrated in some studies (12,21) to correlate with influenza virus shedding. However, we observed that fever was not statistically associated with viral shedding duration (i.e., those HCP who had a fever were not more likely to be positive by real-time RT-PCR or rapid culture longer than those HCP who never had a fever). The association between fever and virus shedding duration might have been modified by antiviral treatment; the small sample size also might have affected the power to detect an association between fever and shedding duration if one existed. In our study, 50% of participants remained positive at 6.6 days by real-time RT-PCR, which is consistent with the average shedding duration reported by Suess et al. (6.6 days) (4), but considerably longer than 4.5 days reported by Carrat et al. (11).

Most HCP in our study were shedding virus that was detected by real-time RT-PCR and rapid culture after they met CDC criteria for returning to work. Our findings support those of a study that found that persons with influenza infections who shed virus cannot be reliably identified by using fever alone (11). These results raise essential considerations regarding exclusion policies for infected HCP. Because febrile and afebrile HCP had similar virologic shedding durations and viral loads, the absence of influenza by real-time RT-PCR or culture might be preferable to the absence of fever as a criterion for HCP who are returning to work in settings where they place others at high risk. A critical caveat is that positive influenza real-time RT-PCR or rapid culture results do not necessarily indicate that the patient is capable of transmitting virus to others. Although reported transmission dynamics of pandemic (H1N1) 2009 virus are similar to those of seasonal influenza viruses (4), the generalizability of our findings to seasonal influenza viruses is uncertain.

Our findings, although limited to pandemic (H1N1) 2009 virus infections, suggest that HCP who return to work after other influenza infections might continue to excrete viable virus and reinforce the need for adherence to infection control measures to prevent transmission in the workplace. All HCP who return to work after influenza infection should practice frequent hand and respiratory hygiene and cough etiquette. Because many HCP were afebrile and others might still be infectious >24 hours after defervescence, our results support the overriding need for influenza vaccination of HCP as the preferred prevention method in health care settings. Further research is needed to determine whether detection of influenza virus and clinical symptoms correlates with infectivity with or without antiviral use.