key: cord-0825370-1qq9osfx authors: Escuyer, Kay L.; Waters, Christine L.; Gowie, Donna L.; Maxted, Angie M.; Farrell, Gregory M.; Fuschino, Meghan E.; St. George, Kirsten title: The assessment of data sources for influenza virologic surveillance in New York State date: 2016-11-14 journal: Influenza Other Respir Viruses DOI: 10.1111/irv.12433 sha: c56567e91abd7957f4544c4f93d7b2d8b2131a2c doc_id: 825370 cord_uid: 1qq9osfx BACKGROUND: Following the 2013 USA release of the Influenza Virologic Surveillance Right Size Roadmap, the New York State Department of Health (NYSDOH) embarked on an evaluation of data sources for influenza virologic surveillance. OBJECTIVE: To assess NYS data sources, additional to data generated by the state public health laboratory (PHL), which could enhance influenza surveillance at the state and national level. METHODS: Potential sources of laboratory test data for influenza were analyzed for quantity and quality. Computer models, designed to assess sample sizes and the confidence of data for statistical representation of influenza activity, were used to compare PHL test data to results from clinical and commercial laboratories, reported between June 8, 2013 and May 31, 2014. RESULTS: Sample sizes tested for influenza at the state PHL were sufficient for situational awareness surveillance with optimal confidence levels, only during peak weeks of the influenza season. Influenza data pooled from NYS PHLs and clinical laboratories generated optimal confidence levels for situational awareness throughout the influenza season. For novel influenza virus detection in NYS, combined real‐time (rt) RT‐PCR data from state and regional PHLs achieved ≥85% confidence during peak influenza activity, and ≥95% confidence for most of low season and all of off‐season. CONCLUSIONS: In NYS, combined data from clinical, commercial, and public health laboratories generated optimal influenza surveillance for situational awareness throughout the season. Statistical confidence for novel virus detection, which is reliant on only PHL data, was achieved for most of the year. ESCUYER Et al. influenza test data. 2 The consequent network and data monitoring systems have facilitated the detection of numerous important events and viral changes, including the rapid identification in 2009 of the pandemic influenza strain (A/H1pdm09). In 2010, to address concurrent fiscal constraints and emerging diseases, the CDC and the Association of Public Health Laboratories (APHL) initiated the Influenza Virologic Surveillance Right Size Project to assess the vast and complex national surveillance system, determine the most efficient means to monitor influenza activity, and establish a standard reference for the CDC and state PHLs. 2 The Influenza Virologic Surveillance Right Size Roadmap (1st Edition released in 2013) attempted to guide surveillance toward a more systematic and statistically relevant process. The roadmap includes sample size calculators, developed to estimate the appropriate numbers of samples needed to achieve influenza surveillance with statistical confidence for situational awareness and rare/novel influenza event detection. The roadmap proposed identification of alternate, non-PHL, data sources as a means to augment state PHL data and, in turn, enhance national surveillance. Data generated from the NYS PHL, the Wadsworth Center, were measured against sample numbers calculated with the computer models for influenza situational awareness and rare/novel event detection. New York State alternate data from clinical and commercial laboratories were analyzed for integrity and impact on influenza situational awareness. Regional NYS PHL data were assessed for its impact on rare/novel event detection. New York State Department of Health scientific staff in the Virology Laboratory at the Wadsworth Center, in partnership with epidemiologists from the Bureau of Communicable Disease Control (BCDC), evaluated influenza testing practices, regulations, infrastructure, data collection, and reporting. Additionally, surveillance policy, potential future ideal practices and systems, and likely hurdles that may impede implementation were discussed. The NYS Wadsworth Center PHL performs influenza testing on specimens received through the Influenza-like Illness Network (ILINet) and Emerging Infections Program (EIP), in addition to samples received from non-EIP hospitals, student health clinics, veteran administration (VA) centers, long-term care facilities, correctional facilities, and occasionally commercial laboratories. The ILINet is an outpatient influenza surveillance program supported by CDC in all states. 3 The CDC supports the EIP, a program within FluServ-NET, 3 The Electronic Clinical Laboratory Reporting System (ECLRS) provides an electronic system for prompt and protected transmission of reportable disease information to the NYSDOH, local health departments, and the NYCDOHMH. 8 Clinical and commercial laboratories submit influenza-positive test results electronically to the NYSDOH via ECLRS. Each ECLRS report contains specimen-level data, including name, DOB, sex, address, home phone, county of residence, reporting laboratory, ordering physician, specimen source, testing method, results, specimen collection date, and report date. New York State Department of Health staff review the submitted data to determine whether it meets the case definition for laboratory-confirmed influenza, defined as a positive influenza laboratory test result with at least one of the following methods: culture, enzyme immunoassay (EIA), direct immunofluorescence assay (DFA), immunofluorescence assay (IFA), RT-PCR, immunohistochemistry (IHC), or influenza virus antigen detection systems (IVADs, also known as rapid influenza diagnostic tests, RIDTs). If the case definition is met, an influenza case report is created in the Communicable Disease Electronic Surveillance System (CDESS). The system automatically deletes duplicate CDESS case reports on the same patient. If the Wadsworth Center tests and does not confirm an initial positive influenza result from another laboratory, the initial F I G U R E 1 NYS map showing the 39 counties of 57 total outside of NYC that contribute to the ILINet and EIP influenza virologic surveillance networks. Counties participating in the EIP program are clustered around the cities of Albany and Rochester. The distribution of the ILINet primary care practitioners is indicated by number in each county and include the following practice types: pediatrics, family practice, internal medicine, student health, urgent care, obstetrics/gynecology, allergy and asthma, ear nose and throat, employee health, infectious disease, and pulmonology. Gray counties do not have providers enrolled in either the ILINet or EIP. The NYS map also depicts the 11 NREVSS laboratories and the 11 WHO collaborating laboratories, which include the NYS PHL in Albany and the three regional PHLs in Erie and Westchester Counties and NYC test is considered a false-positive result and the original influenza case report is revoked. Communicable Disease Electronic Surveillance System allocates ECLRS positive influenza laboratory results to disease classification codes for influenza type A or B, influenza type not specified, A/H1pdm09 subtype (since 2009), and H7N9 (since 2013). October 2014 pursuant to discussions from this project. While PHLs including Wadsworth use the CDC influenza rtRT-PCR panel to detect and subtype influenza viruses, clinical laboratories use a variety of testing methods, which may or may not include subtyping. The majority of influenza molecular assays generate results in approximately 1-8 hours and are capable of detecting influenza viruses with very high sensitivity and specificity; some tests also identify subtypes. 9 Growth and isolation of influenza viruses in culture may take 7-14 days or longer, while IVAD kits provide results in 15-30 minutes. 10 Table 1 summarizes 2014 information on testing platforms and assays, the number of licensed clinical laboratories using them in NYS, test complexity, and the influenza types/subtypes and other respiratory pathogens detected. Revised recommendations released in 2014 11 advised using only PHL data generated with molecular methods for the assessment of sample sizes for detection of a rare/novel influenza event. Three regional PHLs exist in NYS: the Erie County PHL in western NY, the NYC PHL, and Westchester PHL downstate. Sample sizes and confidence levels were evaluated with the calculators for all available data generated with the CDC influenza rtRT-PCR assay from Wadsworth and the regional PHLs. Pooling state surveillance data into national aggregates produce large enough sample sizes to meet recommended confidence levels for the detection of a rare/novel influenza event. During peak season when influenza positivity is 20% or greater, the Right Size Roadmap recommends that states calculate sample sizes sufficient to have 95% confidence in detecting one novel virus among 700 influenza-positive specimens. Prior and post-peak influenza activity, when positivity is less than 20%, the Roadmap recommends a detection threshold of one novel virus of 200 influenza-positive specimens, while for offseason and summer periods, a threshold of one novel virus among four influenza positive samples is recommended. The goal for NYS is to obtain recommended sample sizes for the state population of approximately 20 million, which would ensure optimal detection thresholds with ≥95% confidence and ≤5% margin of error (MOE), for both situational awareness and rare/novel event detection (Table 2 ). Computer modeling software for situational awareness with Calculator A establishes ideal sample sizes using unscreened MA-ILI specimens. For detection of a rare/novel event, the current Calculator B revised late 2015 uses only Flu+ specimens tested at state PHLs. New York State influenza test data were compared with the recommended sample sizes for situational awareness as determined with Calculator A (Figure 2 ). To avoid bias, specimens should preferably be unscreened, or a random sampling. The Wadsworth Center Virology Laboratory receives specimens for influenza testing from many sources including some that are prescreened by IVADs or other methods. During most weeks of peak influenza activity, sample sizes needed to achieve ≥95% confidence levels for situational awareness were obtained only with a combination of randomly submitted Flu+ and MA-ILI specimens. During peak season, the recommended sample sizes were not achieved with only MA-ILI specimens, or outside of peak season with Wadsworth test data alone. The WHO/NREVSS laboratories in NYS provide additional data with sufficient power to meet the optimal confidence levels and MOE For the detection of a rare/novel influenza virus, Wadsworth test data were compared to recommended sample sizes for NYS, aggregated on a national scale, determined from Calculator B ( Figure 5 ). The total number of Flu+ specimens tested at the Wadsworth Center was insufficient to detect a rare/novel event for influenza surveillance at the recommended confidence levels. To augment detection of a rare/novel influenza virus, NYS regional PHL influenza rtRT-PCR data supplemented the Wadsworth Center rtRT-PCR data ( Figure 5 ). From the last week of December 2013 through January 2014 with peak influenza activity, Flu+ data provided 86% to 94% confidence in the likelihood of detecting a novel virus present at 1/700 of cases. During low season, the recommended threshold for detection of a novel virus outside of peak season is 1/200 with a minimum Flu+ sample size of 37; sample sizes with 95% confidence were obtained for 4 of those 6 weeks ( detecting a rare/novel influenza virus, combined sample sizes from Wadsworth and the regional PHLs were sufficient to reach minimum confidence levels (≥85%) for recommended detection thresholds during peak weeks of influenza activity, and optimal (≥95%) confidence for the majority of low season and all of off-season, yet not throughout the year. their data shared with CDC by the NYSDOH. These ECLRS/CDESS data consist of only influenza-positive cases, and sufficient numbers are not attained for situational awareness in the off-season months. The CDESS Flu+ cases for NYS surveillance comprise a large alternate data source that is not currently transmitted to the CDC and provides an indicator of prevailing influenza strains. In the months preceding peak influenza activity, more samples were tested by IVAD than PCR, yet the reliability of the IVAD data is questionable. Some providers submit IVAD-positive specimens to Wadsworth molecular data were insufficient to achieve recom- Week ending Wadsworth total tested NYS Regional PHLs Total # Tested NYS PHLs Total # PosiƟve Flu + Sample size (95% Confidence) World Health Organization. Global Epidemiological Surveillance Standards for Influenza Influenza Virologic Surveillance Right Size Roadmap Centers for Disease Control and Prevention. Overview of Influenza Surveillance in the United States Influenza Virologic Surveillance Right Size Sample Size Calculators: User Guide Centers for Disease Control and Prevention. Guidance for Clinicians on the Use of RT-PCR and Other Molecular Assays for Diagnosis of Influenza Virus Infection Centers for Disease Control and Prevention. Rapid Diagnostic Testing for Influenza: Information for Clinical Laboratory Directors Influenza Virologic Surveillance Right Size Roadmap: Using Alternative Data for Influenza Virologic Surveillance Design and performance of the CDC real-time reverse transcriptase PCR swine flu panel for detection of 2009 A (H1N1) pandemic influenza virus Comparison of a singleplex real-time RT-PCR assay and multiplex respiratory viral panel assay for detection of influenza "A" in respiratory specimens. Influenza Other Respir Viruses Comparison of the Biofire FilmArray RP, Genmark eSensor RVP, Luminex xTAG RVPv1, and Luminex xTAG RVP fast multiplex assays for detection of respiratory viruses Performance of rapid influenza H1N1 diagnostic tests: a metaanalysis. Influenza Other Respir Viruses Accuracy of Rapid Influenza Diagnostic TestsA Meta-analysis Communicable Disease Control Statewide Influenza Surveillance Report for Week Ending The assessment of data sources for influenza virologic surveillance in New York State