NP washes were performed with the patient seated and their head in a semi-Fowler (sniffing) position, holding a sterile cup under their nose. With the patient leaning forward, occluding one nostril, and holding his/her breath, saline was pushed into the open nostril. The wash was collected in the sterile cup. The wash was drawn from the cup into a syringe (recover 1–2 cc) and squirted into cryotube and placed on ice. NP specimens were batched and stored at −70°C.

We collected NP washes every 3 days for 21 days following the date of influenza confirmation. We allowed a 1-day window period on either side of the scheduled study visit date to accommodate weekends, holidays, emergencies, and missed visits. If a participant continued to experience symptoms at the day-21 visit, we collected an additional final set of NP washes at day 28. All NP wash specimens, including the initial screening specimens, were split into at least two aliquots and stored at −70°C until a later time when they were run in batches sequentially for RT-PCR and viral culture at a single central laboratory. All specimens underwent RT-PCR for influenza A and B, with subtyping for influenza A-positive specimens using a laboratory-developed test. RT-PCR cycle threshold (Ct) values were recorded and assumed to be a correlate for influenza viral RNA load values. Influenza virus type and subtype were recorded. If an NP wash was positive for influenza by RT-PCR, an aliquot was cultured for virus isolation. Virus was isolated using the CDC protocol for virus propagation in MDCK cells and RMK cells using traditional cell culture tubes as well as by R-Mix culture, which is a commercial system that consists of centrifuging the specimen onto a cell monolayer containing a mixture of cells optimized for respiratory virus recovery and then staining the monolayer with monoclonal antibodies specific for viral antigens. Three wells were inoculated and then read at days 1, 2, and 3. Specimens that were positive by >2 methods were considered positive. Resistance testing was performed using sequencing using CDC PCR/sequencing primers for amplifying and sequencing fragments A, B, and C of the N gene. PCR and resistance testing were performed at the Special Project's Laboratory at Washington University. Viral culture was performed at the Clinical Virology Laboratory at St Louis Children's Hospital.

To assess the effect of influenza on immune status and on control of HIV infection, serum specimens were drawn for CBC and CD4 cell count at enrollment and at the patient's next HIV clinic visit at least 60 days after and closest to 90 days after the onset date of ILI (two samples total). Plasma specimens were drawn for HIV viral load quantification at the enrollment visit and subsequent study visits at day 9 and day 21 and at the patient's next HIV clinic at least 60 days after and closest to 90 days after the onset date of ILI (four samples total) (Table 1). CBC, CD4 cell count, and HIV viral load quantification were performed at local laboratories that served each participating HIV specialty clinic.

Continuous variables were summarized as medians with interquartile ranges (IQRs), and percentages for each category of categorical variables were calculated. Characteristics were compared between patients with and without influenza using Fisher's exact test. We considered p values less than .05 statistically significant. We defined the duration of influenza virus shedding as the interval from date of ILI onset to the date of last culture-positive NP wash. This definition of shedding duration was used for all analyses. However, to correlate symptoms with shedding, we defined the duration of influenza virus shedding as the interval from date of ILI onset to the date of last PCR-positive NP wash, as PCR is the screening test for influenza, and thus, most clinically relevant in this context. We used the Wilcoxon rank-sum test to examine factors that may have affected influenza virus shedding.²⁵ We calculated correlation coefficients to assess the linear association between variables. All analyses were conducted in SAS 9.2 (SAS Institute, Cary, NC).

From October 2010 to April 2011, we enrolled 961 participants in syndromic surveillance and diagnosed 20 patients with PCR-confirmed influenza infection; all 20 agreed to enroll in the shedding study. Compared with persons without influenza, persons with influenza were more likely to be older, use injection drugs, have a lower median CD4 cell count, and were less likely to have had an influenza vaccination in the past 12 months (Table 2). For the 20 patients with influenza, demographic and clinical characteristics were as follows: median age 48 years (IQR 43–53), 60% male, 50% non-Hispanic black, 44% current smokers, 95% prescribed HAART, 85% virologically suppressed (HIV RNA <400 copies/ml), median CD4 cell count 317 cells/mm³ (IQR: 190–544), and median follow-up time after confirmation of influenza infection 21 days (IQR: 19–22). The most common symptoms reported were cough (100%), fever (90%), myalgias (80%), sore throat (75%), and headache (70%) (Fig. 1). Five patients with influenza had a CD4 cell count <200 cells/mm³, of whom four contributed more than one specimen for the shedding study.

Of the 20 persons with influenza, 19 had influenza A (10 with H1N1pdm09, seven with H3N2, and two with subtype unknown) and one had influenza B. Nine (45%) patients had received pandemic monovalent H1N1 vaccine, of whom eight had also received 2010–2011 seasonal trivalent influenza vaccination. Nine patients received oseltamivir within 48 h of ILI onset. Median days between symptom onset and clinical evaluation differed for patients with and without oseltamivir use (2 days vs. 4.5 days, p = .025). With regard to comorbidities, nine patients had liver disease (six with hepatitis C and three with hepatitis B), seven had cardiovascular disease, three had a history of asthma, and three had an AIDS-defining cancer.

Median duration of influenza detection by PCR was 10 days (IQR: 6–15 days). Median duration of influenza shedding by viral culture was 3 days (IQR: 0–5 days). No patient shed influenza virus as detected by viral culture for >7 days. Median duration of any ILI symptom from ILI onset was 14 days (IQR: 12–26 days). At day 10 after ILI onset, the median RT-PCR Ct value was 37.8; 1 (20%) specimen had a Ct value <30. The median Ct values for RT-PCR-positive but culture-negative specimens (based on repeat testing at sequential visits) were 36.2 (IQR: 31.1–38.9) for influenza A, 36.6 (IQR: 31.1–40.6) for influenza B, and 34.0 (IQR: 31.4–36.5) for influenza A/H1N1.

Median days of shedding were similar among patients with and without prior influenza vaccination in the past 12 months (0 day vs. 4 days, p = .448), among patients with and without HIV viral suppression (2 days vs. 6 days, p = .053), and among patients who were eligible for oseltamivir treatment (seen within 72 h of symptom onset) who did and did not receive oseltamivir (5 days vs. 0 days, p = .083). Those on oseltamivir had similar CD4+ cell counts, 312 cells/mm³ (176–444) vs. 443 (300–568), p = .428. Duration of symptoms were also similar among persons who did and did not receive oseltamivir (13 days vs. 20.5 days, p = .302). CD4 cell count is correlated with shedding time, r = −0.481, p = .043. Only one patient whose CD4 cell count was 176 cells/mm³ and HIV viral load was 1,025 copies/ml shed viable influenza virus 2 days after his symptoms resolved (total duration of influenza shedding = 6 days). One patient whose CD4 cell count was 128 cells/mm³ and HIV RNA was <20 copies/ml died 1 day after diagnosis of influenza and pneumonia. There were no hospitalizations. Among ILI symptoms, only a longer duration of fever correlated with a longer duration of influenza virus shedding (Fig. 2).