Comparison of outpatient medically attended and community‐level influenza‐like illness—New York City, 2013‐2015

Background Surveillance of influenza‐like illness (ILI) in the United States is primarily conducted through medical settings despite a significant burden of non‐medically attended ILI. Objectives To assess consistency between surveillance for respiratory viruses in outpatient and community settings using ILI surveillance from the Centers for Disease Control and Prevention Influenza Incidence Surveillance Project (IISP) and the Mobile Surveillance for Acute Respiratory Infections (ARI) and Influenza‐Like Illness in the Community (MoSAIC) Study. Methods The Influenza Incidence Surveillance Project conducts ILI surveillance in 3 primary care clinics in New York City, and MoSAIC conducts community‐based ILI/ARI surveillance through text messaging among a cohort of New York City residents. Both systems obtain respiratory specimens from participants with ILI/ARI and test for multiple pathogens. We conducted a retrospective review of ILI cases in IISP and MoSAIC from January 2013 to May 2015 with descriptive analyses of clinical and laboratory data. Results Five‐hundred twelve MoSAIC and 669 IISP participants met an ILI criteria (fever with cough or sore throat) and were included. Forty percent of MoSAIC participants sought care; the majority primary care. Pathogens were detected in 63% of MoSAIC and 70% of IISP cases. The relative distribution of influenza and other respiratory viruses detected was similar; however, there were statistically significant differences in the frequency that were not explained by care seeking. Conclusions Outpatient and community‐based surveillance in the one found similar timing and relative distribution of respiratory viruses, but community surveillance in a single neighborhood may not fully capture the variations in ILI etiology that occur more broadly.


| INTRODUCTION
Surveillance for influenza-like illness (ILI) activity in the United States traditionally relies on reports of medically attended visits, including outpatient, and emergency room visits and hospitalizations. 1 The majority of ILI is mild and self-limited, and many patients never seek care in a medical setting. 2 Surveillance for medically attended ILI misses cases of non-medically attended ILI in the community and underestimates the true prevalence of ILI in the population. Understanding community ILI can shed light on the full burden of influenza. Furthermore, primary care clinics, a common setting for outpatient ILI surveillance, may not adequately capture cases who could seek care in other settings, such as emergency departments or urgent care clinics. In addition, systems that focus on medically attended ILI may have other biases (eg, access to care) and delays associated with care seeking that may affect sensitivity of testing and may not reflect influenza and other respiratory virus activity in the broader community. [3][4][5] A comparison between community surveillance and outpatient medically attended ILI surveillance systems could help to determine how well medically attended surveillance systems reflect respiratory viral activity in the community.
Community surveillance for ILI in the United States has been performed periodically in several cohort studies in various small geographic areas [6][7][8][9] ; broader scale community surveillance is generally cost prohibitive. The Mobile Surveillance for Acute Respiratory Infections and Influenza-Like Illness in the Community (MoSAIC) Study is one recent prospective household-based cohort study established in 2012, in a neighborhood in New York City to obtain community-level incidence of influenza and other respiratory viruses. 8 In 2009, New York City, along with other participating sites, began the Influenza Incidence Surveillance Project (IISP). IISP is designed to determine the incidence of medically attended ILI and the proportion of ILI attributable to influenza and other respiratory diseases using provider estimates of patient populations. 10 Using these two systems in the single geographic region of New York City, we sought to compare symptoms as well as the frequency, proportion, and seasonal distribution of respiratory pathogens associated with ILI between individuals with medically attended ILI and broader community-level ILI across three influenza seasons to better understand how current ILI surveillance in medical clinics reflects activity seen at the community level.

| MoSAIC surveillance
The MoSAIC study methods have been described previously. 8 Briefly, MoSAIC is a CDC-funded community-based study which performs surveillance for acute respiratory illness (ARI), including influenza-like illness (ILI), in a neighborhood in New York City yearround with the goal to assess ARI and ILI rates and etiology in the community. The cohort consists of approximately 250 households and is a primarily Latino population. Demographic information is collected upon enrollment. Households receive twice-weekly text messages, asking if anyone has runny nose, congestion, sore throat, cough, body aches, or fever and report either "yes" or "no." Home visits are performed by research staff to obtain a nasal swab from any participants meeting the ARI criteria (2 symptoms including fever, runny nose/congestion, sore throat, cough, and/or myalgia) with symptoms lasting < 5 days and who are still symptomatic. The ARI criteria for infants (less than 12 months of age) also include runny nose/congestion alone. Respiratory swabs are analyzed in a research laboratory using a BioFire FilmArray multiplex polymerase chain reaction (PCR) assay (BioFire Diagnostics LLC, Salt Lake City, UT) detecting influenza A virus (H1, H1N1pdm09, and H3), influenza B virus, adenovirus, coronaviruses (229E, HKU1, NL63, and OC43), enteroviruses/rhinoviruses, human metapneumovirus (HMPV), human parainfluenza viruses (HPIV) types 1-4, respiratory syncytial virus (RSV), Chlamydia pneumoniae, Mycoplasma pneumoniae, and Bordetella pertussis. Further information on symptoms and care-seeking behaviors is collected by interview at illness end.
We limited our analysis to episodes meeting ILI criteria (fever with cough or sore throat) for appropriate comparison to ILI within IISP.

| IISP surveillance
The CDC's IISP is a population-based outpatient surveillance network operating year-round, for syndromic ILI with systematic laboratory testing for influenza viruses, including three primary care clinics in New York City. 10 Since 2009, participating clinics report the weekly number of ILI and all-cause visits. Patients with symptom onset within 7 days of presentation are included, and ILI is defined as fever with cough or sore throat among patients ≥ 2 years of age, and fever with ≥1 of cough, sore throat, nasal congestion, or rhinorrhea among children < 2 years

| Study populations and symptoms
During the surveillance period, 334 households were followed by While all participants included in this analysis met the case definition of ILI (fever and cough or sore throat), there were significant differences in symptoms reported by participants in MoSAIC as compared with participants in IISP (Table 2). Differences persisted even after adjusting for age; ILI cases in MoSAIC were more likely to report rhinorrhea (81% vs 62%, P < .01) and less likely to report sore throat (47% vs 61%, P < .01) or myalgia (23% vs 46%, P < .01) compared with those in IISP. To further explore if these difference were due to care-seeking behavior, we examined reported symptoms among only MoSAIC cases that reported seeking care for their illness and found that differences compared with the IISP group persisted ( Table 2).   Coronavirus 25 (7) 47 (  10 (7) 1 (1) 16 (10) 11 (4) 14 (10) 6 (4)

| Limitations
Our analysis had several limitations. Firstly, we restricted to an ILI definition to focus on influenza, however, may not be optimal for the other viral pathogens tested. Our objective was to compare the pathogens detected within the context of ILI between the two systems rather than the full burden of all pathogens. Data on vaccination were incomplete in both IISP and MoSAIC, limiting our ability to control for vaccination in the analysis. In addition, we did not take possible household clustering into account among MoSAIC participants who were living in the same residence.

| CONCLUSIONS
Surveillance through community cohorts can more fully capture local incidence and etiology of ILI and has the potential to provide further information beyond what is capable in medically attended surveillance, such as transmissibility 8,[19][20][21] and may add to our understanding of the full burden of influenza. While the timing of influenza and other virus detections were captured equally well in community and outpatient surveillance, the proportions of viruses detected varied between the community and outpatient clinics. There may be local geographic and/or social mixing differences affecting incidence of circulating viruses at a very local level and outpatient surveillance across more clinics may better reflect viral circulation in the larger community. Community and medically attended surveillance are valuable and complementary for understanding full ILI incidence, etiology, and burden.