Definition of healthcare‐associated influenza: A review and results from an international survey

Aim To describe definitions of healthcare‐associated influenza (HAI) in recent literature and in hospitals participating in a survey of Society for Healthcare Epidemiology of America (SHEA) Research Network (SRN) members. Method A review with PubMed search was undertaken to retrieve articles published between 2008 and 2016, focusing on the subject headings “influenza, human” and “cross infection.” Definitions of clinical influenza‐like illness (ILI) and HAI were identified. An invitation to participate in the survey was sent to 218 SRN members via email. Results Of 75 articles on HAI included in the review, 30 presented a standardized definition of clinical ILI based on fever (100%), cough (80%), and sore throat (70%). Forty studies (53%) contained a standardized HAI definition, grounded on threshold delay from admission in 29 of them, this delay ranging from 48 to 196 hour (median: 72 hour). Fifty‐five SRN members responded to the survey, with a standardized definition of HAI adopted by 76% of them. This definition was based on clinical features for 24%, virological features for 31%, and both for 45%. Fever (mean threshold: 38.0°C) was part of the definition for 82%. The features required most frequently in the clinical definition were cough (46%) and sore throat (26%). Median threshold delay between admission and symptoms onset adopted for HAI definition was 48 hour (range: 24‐96 hour). Conclusion This work underlined the heterogeneity of HAI definitions in different countries. A standardized definition would be helpful to evaluate HAI spread, outcomes in patients and healthcare systems, and the impact of prevention measures, including vaccination.


| INTRODUCTION
Healthcare-associated influenza (HAI) is associated with significant morbidity, mortality, and costs attributed to increased length of stay, but is likely to be under-recognized. 1 Early detection leads to rapid implementation of respiratory isolation precautions and prevents nosocomial transmission between patients and healthcare personnel (HCP). Timely recognition could obviate extra testing and treatment.
The need for a standardized definition of HAI has been underlined previously. 2

A synthesis using the Outbreak Reports and Intervention
Studies of Nosocomial Infection (ORION) statement highlighted the dearth of standardized information collected during HAI outbreak investigations, 3 which limits comparability between studies. Although standardized definitions of clinical influenza-like illness (ILI) were implemented in some observational and interventional studies, they differed from one another in terms of body temperature threshold, symptoms, and time interval between hospitalization and symptoms onset. 3 The World Health Organization, the Centers for Disease Control and Prevention (CDC), and the European Center for Disease Prevention and Control have proposed different definitions of community ILI. [4][5][6] To control the spread of HAI, definition of clinical cases should afford high sensitivity to avoid cases being missed. 3 To the best of our knowledge, no review has systematically reported clinical and virological criteria defining HAI. Reported HAI and contemporary definitions used by HCP might help to improve the reliability and homogeneity of HAI definition.
The purpose of this study was to report HAI definition: (i) in recent literature and (ii) in hospitals participating in the SHEA (Society for Healthcare Epidemiology of America) Research Network (SRN). The results might help to facilitate guideline updating on HAI and collaborative surveys and studies.

| METHODS
A literature review with PubMed search was undertaken with the medical subject headings (MeSH) "influenza, human" and "cross infection" to find English-and French-language articles published between January 1, 2008, and June 1, 2016. A previous synthesis of the literature done with similar MeSH terms by our team for articles published before 2008 had other objectives; these studies were not included in this review. 3 Titles and/or abstracts were analyzed to trace publications on influenza or ILI infections in hospital settings. Articles with healthcare-associated influenza or ILI cases as outcome were selected. Definitions of clinical ILI and HAI were described in selected articles. Country, study period, study design, population involved (patient and/or HCP) were also gathered.
An invitation to participate in the online survey was sent to all SRN members via e-mail on July 9, 2015, followed by two e-mail reminders. The survey was posted publicly on the SRN Web site (http://www. shea-online.org). The SRN is a consortium of more than 200 hospitals worldwide, with collaborative multicenter research projects in healthcare epidemiology. 7 The questionnaire asked about the demographic characteristics of respondent hospitals, definitions of clinical ILI and HAI, recommended tests, detection of clusters and notification (see Data S1).
Additional questions were asked concerning a particular definition to clinically suspect influenza in patient with clinical ILI. Statistics were reported as means (range), medians, and percentages to describe the study results.
A standardized definition of clinical ILI was given in 30 studies (40%). This definition required at least the presence of fever in all them ( Figure 1A). A fever threshold was defined in 19 studies (mean 37.9°C; range 37.5-38.0°C). Cough and sore throat were present in the definition in 24 and 21 studies, respectively (80% and 70% of those with definition). The association of different signs/symptoms (fever/cough/sore throat) was noted in 70% of definitions. The definition was less precise in 16 studies (21%): It comprised the following terms: "influenza-like illness," "respiratory infection," "acute respiratory illness," "signs and symptoms of influenza." Symptoms presented by  Overall, 76% of respondents worked with a standardized definition of HAI (N=42). This definition was based on clinical features only for 24%, on virological elements only for 31%, and on both categories for 45% (Table 1). Fever was part of the definition for 82% of those who had a standardized definition. The mean minimal threshold defining fever in patients and HCP free of antipyretics was 38.0°C (range: 37.0-39.0°C).

| SRN survey
The clinical features mostly required in the definition of ILI patients were cough (52%), sore throat (31%), and myalgia (19%, Figure 1B). Thirty-six percent of respondents used a particular definition to clinically suspect influenza in patient with clinical ILI. Among them, 94% took fever into account in their "clinical influenza" definition. Cough, sore throat, and myalgia were required symptoms in 75%, 45%, and 35% of respondents, respectively. Virological testing consisted of throat swabs for 2% of respondents, nasal or nasopharyngeal swabs for 69%, and swabs from both sites for 29%. The most frequently performed tests were molecular identification (RT-PCR, 77%) and antigen detection by rapid diagnostic test (19%). Virological samples were sent to a national reference center in 38% of healthcare settings. The mean minimum threshold delay between admission in the unit and symptoms onset defining