This study describes the duration of pH1N1 virus RNA detection, the correlation between rRT-PCR-positive results and virus isolation, and clinical symptoms associated with pH1N1 virus detection in patients living in a densely populated community with low socioeconomic status in sub-Saharan Africa. Pandemic H1N1 RNA was detected from respiratory specimens by rRT-PCR for a median duration of 8 days but up to 17 days after symptoms onset. This duration of virus RNA detection is similar to the median of 6 and 8 days reported in studies from China and Hong Kong, respectively 
. In our study, we did not find any differences in the duration of pH1N1 RNA detection among various age groups or between males and females. These results were similar to those reported by a study in Hong Kong, which reported no correlation between influenza viral load and age 
, and one from Canada which showed no differences in shedding between children and adults 
. In contrast, studies in Hong Kong and China found that younger age and male gender were risk factors for prolonged pH1N1 virus detection 
. Our study population, was mostly children (79% of the patients were <14 years old), thus associations between age and duration of pH1N1 RNA virus detection were difficult to assess. In addition, our study population included outpatients only, the China and Hong Kong studies mentioned above included hospitalized patients 
Cough, fever, and runny nose were the most common clinical symptoms associated with pH1N1 virus RNA detection. This is similar to studies carried out in Korea 
and Hong Kong 
. While sore throat was commonly associated with pH1N1 patients in other studies 
, in our study, only 29% of the pH1N1-positive patient visits from patients ≥5 years old were associated with this symptom. In our study, a substantial proportion of patients continued to shed the virus after respiratory symptoms had resolved. Viable pH1N1 virus, which was potentially infectious, was isolated from one-third of specimens obtained from recovered patients. Therefore, some pH1N1 patients who were no longer symptomatic may still be shedding viable pH1N1 virus. In addition, half of rRT-PCR-positive samples from patients who were on day 8 to day 10 after symptom onset were culture-positive. In 2009, in the early stages of the H1N1 pandemic, WHO recommended that pH1N1 patients remain isolated for 7 days or until symptoms resolved 
. While these guidelines may be appropriate for the community, in a healthcare setting, where the goal is to prevent as much spread as possible, other measures need to be considered 
Of the 2 HIV-positive patients in the study, the patient who was not on antiretroviral therapy (ART) had detectable pH1N1 virus RNA for 16 days, much longer than the median duration for the study population. This is slightly longer than was shown in a study of HIV-positive school-aged children carried out in Germany in which the median time from symptoms onset to first negative rRT-PCR result was 9 days (Range, 5–14 days) and cultures become negative after 6 days (Range 3–11 days) 
. Other studies have shown that immunosuppressed individuals may shed influenza virus longer than the general population 
. In a case report from the United States of America, 2 cancer patients who were severely immunosuppressed were shown to continue having detectable pH1N1 virus RNA for 5 and 6 weeks after initial diagnosis 
. In previous studies, severely immunocompromised patients have been shown to shed seasonal influenza for weeks to months 
. The issue of prolonged shedding in immunocompromised patients is especially relevant in sub-Saharan Africa, where over 22 million people are infected with HIV and only 30% are on ART 
. In Kenya, the HIV prevalence in persons aged 15–64 is 7.1% 
, and in the study site, HIV prevalence is 14% (KEMRI/CDC-K, unpublished data). In our study, we identified 2 HIV-positive patients among 16 tested, one of whom had prolonged shedding. More research should be conducted to better understand the extent of pH1N1 viral shedding in untreated HIV-infected patients and those on ART therapy.
PCR, which detects viral nucleic acid instead of infectious viral particles, is more sensitive than virus culture in detecting influenza 
. However, because it does not detect viable whole virus, people who have respiratory specimens that are rRT-PCR-positive may not harbor a sufficient amount of viable virus to infect other people. The relationship between virus titers and influenza transmissibility is not known. However, serial interval studies suggest that most seasonal influenza household transmission occurs in the first 3 days after the index case's illness onset 
. In our study, we successfully isolated pH1N1 virus from 95% (81/85) of rRT-PCR-positive specimens taken from day 0–3 after symptom onset, but only 18% (3/17) rRT-PCR-positive specimens taken ≥11 days after symptom onset were culture-positive. These findings suggest that if a patient has a respiratory specimen taken early in the course of illness that is positive for pH1N1 by rRT-PCR, that patient is likely shedding live virus. In contrast, a rRT-PCR-positive result from a sample taken later in a patient's course of illness may not mean that the patient is still shedding live virus.
Our study had some limitations. First, our study population was mainly comprised of persons <14 years, and the oldest patient was 41 years old. Thus, we were not able to evaluate viral shedding patterns in elderly individuals. Second, since few people in our study had known underlying medical conditions, we were not able to determine the impact of co-morbidities, including HIV infection, on duration of pH1N1 viral shedding. Some patients whose HIV status was unknown may have been HIV-positive; this could explain the slightly longer duration of rRT-PCR positive pH1N1 results in our study compared to other studies. Third, there was an overlap between the patients who presented with ILI and SARI, thus it was difficult to analyze data for each of the syndromes separately. Finally, 20% of patients enrolled in the study did not fully complete the study. Therefore, we have little information on these patients regarding shedding. We used right-censoring in our analysis for these patients. These results were comparable to those obtained from the 85 patients who completed the study.
In this study, we show that pH1N1 shedding patterns in an impoverished, densely populated urban community in Nairobi, Kenya, are similar to those described in studies in more affluent countries in temperate and subtropical areas of the world. Because of unique co-morbidities in sub-Saharan Africa compared to other areas of the world 
, more research is needed to characterize shedding dynamics and impact on disease transmission for pH1N1 infection in African communities.