Of the 173 persons from 47 participating households, 35 with pH1N1 PCR–confirmed persons (index case-patients) were referred by their treating physicians. Among the 138 other participants (household or community contacts), 73 had respiratory symptoms for <7 days at the time of enrolment and 32 (44%) of 73 were positive by pH1N1 PCR (Figure 1, Table 1). Cell culture was also positive for 97% (31/32) of these pH1N1 PCR–confirmed cases. Of the 32 pH1N1 PCR–confirmed cases, 78% (25/32) had fever at some point since illness onset; at the time of specimen collection, 94% (30/32) had cough and 34% (11/32) were still febrile. Specimens were retested with a matrix PCR. All results from the pH1N1 PCR–positive participants were also positive by matrix PCR; of those who were negative by pH1N1 PCR, 37% (15/41) were positive. Of the 15 participants positive by matrix PCR, 13% (2/15) were positive by cell culture. Of the 47 confirmed cases (32 by pH1N1 PCR and 12 by matrix PCR), cell culture positivity varied from 69% to 87% for specimens obtained within 4 days after symptom onset and dropped to 33%–40% for specimens obtained days 5 and 6 after symptom onset (Figure 2).

Of the 67 patients whose results were positive by pH1N1 PCR (35 case-patients initially referred by their physicians plus 32 detected among contacts), 62 were identified <7 days after symptom onset and 43 had a second swab collected on day 8 (Figure 1). Of these 43 cases (core participants), 47% were children <10 years of age, 26% were 10–17 years of age, and 28% were adults (>18 years of age) (Table 1). On day 8, 18 (42%) were still positive by pH1N1 PCR, and 14 others were positive only by matrix PCR, for a total of 74% (32/43) positive by any PCR method on day 8. (Table 2) Virus culture was positive for 19% (8/43) of the patients: culture was positive only for pH1N1 PCR–positive cases (8/18, 44%) but for none of the cases positive only by matrix PCR. PCR and virus culture positivity rates did not differ among age groups (Table 2). Only 5% (2/43) of case-patients were still febrile on day 8, but 91% (39/43) were still coughing. None of the 8 patients who had a positive virus culture on day 8 were febrile, but 7 (88%) were still coughing. Another swab was repeated on day 11 for 16 of the 18 case-patients who were positive by pH1N1 PCR on day 8 and 14 (88%) were still positive by at least 1 PCR (4 by pH1N1 PCR and 10 by matrix PCR). However, no specimen was positive by cell culture. On day 11, a total of 12 (86%) of the 14 PCR–positive case-patients were still coughing.

Of the 73 symptomatic participants tested within 7 days after symptom onset, 15 of the 47 PCR–positive case-patients (in gray in Figure 1) were detected only by matrix PCR (2/15 cell culture positive) (Table 1). Of specimens positive by pH1N1 PCR, 97% (31/32) were culture positive at diagnosis, 44% (8/18) on day 8, and none on day 11 of illness. Of those whose results were positive by matrix PCR, 13% (2/15) were culture positive at diagnosis, and none were positive on days 8 or 11 of illness. Because virus culture was much less frequently positive for specimens positive only by matrix PCR (2/39, 5%) than for specimens positive by pH1N1 PCR (39/54, 72%), the 19% virus culture positivity on day 8 among core participants (all pH1N1 PCR positive) overestimated the true proportion positive at day 8. Assuming that none of the 15 case-patients with matrix PCR–positive results and 19% of the 32 case-patients with pH1N1 PCR–positive results would shed live virus on day 8, we can estimate that 6 [(19% × 32 pH1N1 PCR positive) + (0 × 15 matrix PCR positive)] of the 47 (13%) case-patients would still be positive by virus culture on day 8. If, instead, we assume that all 73 symptomatic participants were infected by pandemic (H1N1) 2009 and that positive cell culture on day 8 would be found only in patients positive by pH1N1 PCR, then 8% (6/73) still would be shedding live virus 1 week after illness onset. The real cell culture positivity rate on day 8 for all pandemic (H1N1) 2009–infected patients thus probably ranges from 8% to 13%.

Human challenge studies with seasonal influenza have shown that virus shedding after day 7 is rare (2), but clinical studies have shown that shedding may persist beyond that period in some populations, such as elderly persons, immunocompromised patients, and children (3–5,10–12). In a study among hospitalized persons infected with seasonal influenza A viruses, 54% remained positive by PCR beyond 7 days after symptom onset, and 29% were positive by cell culture (13). In another study, elderly hospitalized patients infected by influenza A (H3N2) viruses had higher virus loads than did outpatients, and their PCR positivity rate 1 week after disease onset was still high (57%) (10).

In this prospective study, the proportion of pandemic (H1N1) 2009–infected persons still shedding replicating virus on day 8 varied from 8% to 13%, with no difference between children and adults. None were still shedding infectious virus on day 11. With seasonal influenza, virus shedding may be longer in children because they have less preexisting immunity that would limit replication than in adults. However, children and adults <50 years of age appear equally susceptible to infection with pandemic (H1N1) 2009 virus, which would support our findings of comparable virus replication and shedding across age groups studied (1).

Our study had some limitations. First, our small sample size and study design may have limited our ability to directly measure culture positivity on day 8. Only a small number of patients had a specimen collected on day 8 and even fewer on day 11. In retrospect, a better design would have been to collect specimens from all 73 symptomatic household members on day 8, irrespective of the initial pH1N1 PCR result. That design would have enabled a more direct estimate of the proportion of patients who were culture positive on day 8, rather than the indirect approach we used. However, our extreme scenario (which assumes that all 73 symptomatic contacts were infected) provides the minimal positivity rate on day 8, and testing of all 73 on day 8 could only have found a proportion equal to or greater than our 8% estimate.

Second, our sampling methods could have influenced positivity rates. Although collection of NP specimens with a flocked swab is one of the best methods for obtaining specimens to detect influenza, those specimens might have been improperly collected by the nurses. Suboptimal collection of swabs would have yielded false-negative PCR or cell culture results, which in turn would have underestimated the proportion of patients shedding virus on day 8.

Third, PCR testing with the matrix PCR was conducted retrospectively on frozen specimens, and only 5% of those were positive by virus culture. A greater proportion of virus culture specimens might have been positive if those specimens had been processed immediately instead of going through a freeze-thaw cycle (14). Moreover, our study included only ambulatory patients ,whereas studies of seasonal influenza that include hospitalized or immunocompromised persons show prolonged shedding, contributing to the impression that our findings most likely underestimate the true proportion of case-patients still shedding virus on day 8. The strengths of our study include its prospective design in a family setting and its use of various methods, including 2 PCR assays and virus culture, to detect pandemic (H1N1) 2009.

Our results are consistent with other reports of virus shedding in pandemic (H1N1) 2009–infected patients. In Singapore, among 70 pandemic (H1N1) 2009–infected patients treated with oseltamivir and swabbed daily until virus clearance, 37% were PCR positive on day 7 of their illness and 9% on day 10 (15). No virus culture was performed in that study, so we cannot estimate the proportion of patients shedding infectious virus at these time points. However, even with oseltamivir treatment, the positivity rate by pH1N1 PCR on day 7 was similar to our own (42%) on day 8, and we can thus infer that the cell culture positivity rates also would be similar. In China, among 421 patients with serial swabs tested by real-time PCR but not cell culture, the median time from onset of disease to negative test result by real-time PCR was 6 days (range 1–17 days), indicating that 50% of patients were shedding virus >6 days (16).

A study conducted by Witkop et al. during a pandemic (H1N1) 2009 outbreak at the US Air Force Academy showed that 29% (31/106) of afebrile patients and 19% (11/58) of patients who had been symptom-free for 24 hours still shed viable pandemic (H1N1) 2009 virus. In their study, 24% of 29 swabs collected on day 7 and 13% of the 16 swabs collected on day 8 of illness were culture positive, despite the large proportion of patients prescribed antiviral drugs (17).

No definitive test is available for assessing the real contagiousness of a patient. The presence of replicating, and therefore infectious, influenza virus is an absolute prerequisite for contagiousness, but it does not necessarily imply it. Contagiousness depends on many factors, including viral load and presence of clinical characteristics contributing to spread of droplets (such as coughing, rhinorrea, or sneezing) and is affected by the number and proximity of contacts between a case-patient and a susceptible person. Nevertheless, our study raises concerns about current recommendations for self-isolation until only 24 hours after fever has subsided (6). With pandemic (H1N1) 2009, fever generally persists 1–4 days and may be absent in 6%–11% of patients (1,15). In our study, of the 32 pH1N1 PCR–positive household members who had been symptomatic for <7 days, 78% had fever at any time since onset of their illness, but only 34% were still febrile on the day they tested positive. Nonetheless, 97% of specimens obtained from these patients were positive by cell culture. Our sample size was insufficient to directly compare PCR or culture positivity by fever status or other symptom or severity indicator at specimen collection or as a component of the overall illness.

Before policy implications can directly follow from these findings, the association of self-isolation with substantial social impact needs to be carefully weighed against the possible benefits of reducing community transmission. In the general population, a 1-week self-isolation period seems more likely to prevent transmission than does isolation until fever has resolved. However, given that 8%–13% of patients may still shed infectious virus on day 8, longer periods of self-isolation for persons expected to come into contact with vulnerable persons (e.g., pregnant women, newborns, or immunocompromised persons) also may be prudent.