This study was approved by both the Institutional Review Board of CDC-Atlanta and the Ethical Review Committee of the Kenya Medical Research Institute (KEMRI).

Informed written consent was obtained from all participants. For children, written consent was obtained from parents or guardians.

We conducted the study in a large informal urban settlement in Nairobi, in an existing population-based surveillance system that has been previously described [14]. Consenting participants who presented to a field clinic, known as Tabitha Clinic, within the surveillance site with signs and symptoms consistent with influenza-like illness (ILI) or severe acute respiratory illness (SARI) had oropharyngeal (OP) and nasopharyngeal (NP) specimens taken. The specimens were tested for influenza by rRT-PCR at the KEMRI/CDC-K laboratory in Nairobi.

An ILI case was defined as a patient with fever ≥38°C and cough or sore throat for all ages. For SARI, the definitions varied by age group. In children <5 years old, SARI was defined as cough or difficulty breathing along with a danger sign; unable to drink or breast feed, lethargic or unconscious, vomiting everything, convulsions, nasal flaring, grunts, chest indrawing, stridor in a calm child or oxygen saturation ≤90%. In people ≥5 years, SARI was defined as fever ≥38°C plus cough or shortness of breath or chest pain or oxygen saturation ≤90%.

Patients who came to Tabitha Clinic between October 14, 2009, and November 25, 2009, who had ILI or SARI and NP/OP specimens that tested positive for pH1N1 were recruited for the study. Patients who were positive for pH1N1 were contacted at their homes by field workers and requested to return to the clinic for follow up every 2 days. During the initial visit and subsequent visits, a trained clinician recorded signs and symptoms and collected both NP and OP specimens. Patients who had 2 consecutive rRT-PCR negative specimens were released from the study. A patient was considered to have pH1N1 RNA if a specimen collected on that day was positive for pH1N1 by rRT-PCR. Patients were not followed over the weekend.

Specimens were obtained according to the following procedure: for OP specimens, a sterile nylon-tipped plastic-shafted OP swab touched the back of the oropharyngeal mucosal membrane for 3–5 seconds and then was placed into a cryovial containing 1 mL of viral transport media (VTM). VTM was prepared at the KEMRI/ CDC-K laboratory using the standard WHO protocol that includes bovine serum albumin and veal infusion broth supplemented with amphotericin B (www.who.int/csr/resources/publications/surveillance/Annex8.pdf). Freshly prepared refrigerated VTM was used for up to 3 months. For NP specimens, a polyester-tipped flexible aluminum-shafted NP swab was inserted into the nose to the nasopharynx, where it was rotated 180 degrees and left in place for 3–5 seconds. The NP swab was inserted into the cryovial containing the OP swab from the same patient. The specimens were labeled and transported at 4°C to the KEMRI/CDC-K laboratory where they were tested for influenza A and pH1N1 using rRT-PCR within 24 hours.

The rRT-PCR testing was performed using the CDC pH1N1 testing protocol [15]. Briefly, total RNA was extracted from 100 µL aliquots of each specimen using QIAamp viral RNA minikit (Qiagen Inc., GmbH, Germany) according to the manufacturer's instructions. One step rRT-PCR was carried out using the AgPath kit (Applied Biosystems, California, USA). For influenza A detection, the primers used were 5′GAC CRA TCC TGT CAC CTC TGA C as the forward, and 5′ TG CAG TCC TCG CTC ACT GGG CAC G as the reverse. The influenza A detection probe was 5′ TGC AGT CCT CGC TCA CTG GGC ACG. For pH1N1, we used the 5′GTG CTA TAA ACA CCA GCC TYC CA as the forward primer, 5′ CGG GAT ATT CCT TAA TCC TGT RGC as the reverse primer, and 5′CA GAA TAT ACA TCC RGT CAC AAT TGG ARA A as the probe. Specimens were also tested for seasonal influenza A H1 and H3. Following the reverse transcription step, a typical 45 cycle PCR reaction was run and fluorescence was read at the annealing/extension step. Appropriate negative and positive control specimens were run alongside each reaction. The results were recorded as cross-over threshold (C_T) values. Any influenza A C_T value <40.0 was recorded as positive; C_T value 40.0–44 were considered indeterminate, and those without a C_T reading were recorded as negative. A specimen was considered to be pH1N1 positive if both the influenza A and the pH1N1 C_T values were <40.0 as described in the CDC protocol of real time RT-PCR for influenza A(H1N1) [16].

Later, in order to evaluate the concordance between rRT-PCR testing and viral culture for pH1N1, influenza virus isolation in Mardin-Darby Canine Kidney (MDCK) cells was attempted for the first rRT-PCR-positive specimen, the last rRT-PCR-positive specimen, and the first rRT-PCR-negative specimen from each enrolled patient. Confluent monolayers of MDCK cell line growing in T25 cell culture flasks were used. Media were removed from the flasks. Each flask was inoculated with 100 µL of specimen and inoculum was allowed to adsorb on the cells for 30 min at 37°C. Following adsorption, 6 mL of Viral Growth Medium [DMEM supplemented with HEPES buffer, bovine serum albumin, L-Glutamine, trypsin, and antibiotic-antimycotic (Sigma-Aldrich, Inc., St. Louis, MO, USA)] was added and cultures were observed daily for cytopathic effect (CPE) for up to 6 days, at which point all the cultures were harvested. Cultures that showed observable CPE by microscopy were subjected to hemagglutination assay using guinea pig red blood cells. Supernatants from cultures that showed no CPE were subjected to a second passage, following which cultures with observable CPE were subjected to hemagglutination assay. Cultures that had no observable CPE after the second passage were considered negative by viral isolation.

Data were collected in three ways. First, information related to the patient's initial visit was recorded on computers by clinicians at Tabitha Clinic. Second, during the first follow-up clinic visit and subsequent visits, information about signs and symptoms was recorded on paper questionnaires and entered into a Microsoft Access database. The signs and symptoms recorded were fever, temperature ≥38°C, cough, sneezing, runny nose, vomiting, diarrhea, chest pain, and difficulty breathing for all patients. In addition, the following symptoms were recorded for patients ≥5 years; earache, sore throat, headache, chills, muscle pain, and abdominal pain. Third, laboratory data were recorded into Freezerworks software (Dataworks Development, Inc.). Data were combined into a central database for analysis.

We considered the date of onset of illness to be the date on which the patient reported the first symptom associated with the illness, according to the history taken during the patient's initial clinic visit. For possible initial symptoms, we considered fever, cough, and difficulty breathing for patients <5 years old. In patients ≥5 years old, these symptoms and sore throat were considered as possible initial symptoms. Patients were considered to have no detectable virus RNA once they had two consecutive specimens that tested negative for pH1N1 by rRT-PCR. We defined the last day of virus detection as the midpoint between the dates of collection of the last rRT-PCR pH1N1-positive specimen and the first rRT-PCR pH1N1-negative specimen. We were unable to follow all patients until they had two negative rRT-PCR tests because some did not return to the clinic for the necessary follow-up visits. The time to cessation of pH1N1 virus detection for these patients was right censored at their last pH1N1 rRT-PCR-positive test date. Thus, the duration of pH1N1 virus detection for these patients was only known to occur after their drop out time and was treated accordingly in the survival analysis below.

Comparisons of duration of pH1N1 virus detection between groups, described by the median and mean, were conducted using the log-rank test. A multivariable Cox regression model was used to determine the association between gender and age with duration of pH1N1 virus detection. A Kaplan-Meier plot was used to determine duration of pH1N1 virus detection. Odds ratios were calculated to determine the odds of isolating pH1N1 virus from rRT-PCR-positive specimens collected over time. All analyses were conducted with SAS version 9.1. Findings were considered statistically significant if the resulting p-value was <0.05.

From the 106 patients enrolled to the study, 449 specimens were collected including the initial specimen from which pH1N1 was detected; 200 (44%) were positive for pH1N1 by rRT-PCR. Virus isolation was attempted in 262 of the 449 (58%) specimens collected. The specimens cultured included the 106 initial rRT-PCR-positive specimens, 56 final rRT-PCR-positive specimens (fifty patients had only one rRT-PCR-positive specimen), and100 first rRT-PCR-negative specimens from the enrolled patients. No rRT-PCR-negative specimens were obtained from 6 of the patients. Of the 162 rRT-PCR-positive specimens, pH1N1 virus was isolated from 132 (81%) specimens. Of rRT-PCR-positive specimens taken on day 0–3 after illness onset, 81/85 (95%) were culture-positive, as were 37/40 (93%) taken on day 4–7, 11/20 (55%) taken on day 8–10, and 3/17 (18%) taken ≥11 days after illness onset (Figure 2). Viral isolation was successful in 100/106 (94%) of the rRT-PCR positive specimens collected at the initial clinic visit. Viral isolation was equally successful in samples with C_T values<25 and those with C_T values 25–30 (92% vs. 84% p = 0.25), but more successful for samples with C_T values<25 compared to those with C_T values 30–39 (92% vs. 63%, p = 0.009). Real time RT-PCR-positive specimens collected from patients on day 0–3 after illness onset were 7.5 times more likely to be culture-positive than rRT-PCR-positive specimens collected on day 4–7 (CI: 1.4–38.9; P<0.012), 17.6 times more likely to be culture-positive than rRT-PCR-positive specimens collected on day 8–10 (CI: 3.2–96.1; p<0.001), and 142.9 times more likely to be culture-positive than specimens collected from patients ≥11 days after illness onset (CI: 2.4–833.3; P<0.001). Of the 100 rRT-PCR negative specimens, 6(6.4%) were culture-positive.

The median number of days pH1N1 virus RNA was detectable in patients specimens 8 days (95% CI: 7–10 days) after symptom onset. There was no statistically significant difference in the duration of pH1N1 virus detection between children <5 years old and those 5 to 14 years or persons ≥15 years. There was no difference in the median duration of pH1N1 virus detection between males and females. In the majority (58%) of patients, pH1N1 virus RNA was detected for ≥7 days, and in 16% of patients for ≥14 days (Figure 3). In general, the mean C_T value increased with time. One of the HIV-positive patients who was on antiretroviral therapy and had a CD4 count of 17cells/mm³ had detectable pH1N1 virus RNA for 4 days after symptom onset. The other HIV-positive patient, who was not on antiretroviral treatment and whose CD4 count was unknown, had detectable pH1N1 RNA for 16 days after symptom onset. These two patients completed the study with two consecutive pH1N1-negative specimens.

There were 200 patient visits that were associated with pH1N1-positive rRT-PCR results. We were able to link laboratory results with clinic visit data on symptoms for 186/200 patients. From the linkable patient visits, there were 69/186 pH1N1-positive rRT-PCR results from patients <5 years old and 117/186 from patients ≥5 years old. The most common signs and symptoms in the pH1N1-positive patients <5 years old were cough [52/69 (75%)], runny nose [47/69 (68%)], reported fever [44/69 (64%)], and temperature ≥38°C [40/69 (58%)]. The most common signs and symptoms in the patients ≥5 years old were cough [83/117 (71%)], temperature ≥38°C [63/117 (54%)], reported fever [61/117 (52%)], and runny nose [60/117 (51%)] (Table 2).

Eighteen (17%) pH1N1 enrolled patients who were initially symptomatic and pH1N1-positive continued to be rRT-PCR-positive after all of their signs and symptoms had resolved for a median of 9 days (interquartile range 7–10 days) after all signs and symptoms had resolved. Viable pH1N1 virus was isolated from specimens obtained from 7/18(39%) of these patients.