Clinical characteristics, predictors, and performance of case definition—Interim results from the WHO global respiratory syncytial virus surveillance pilot

Abstract Background The lack of a uniform surveillance case definition poses a challenge to characterize the epidemiology, clinical features, and disease burden of the respiratory syncytial virus (RSV). Global standards for RSV surveillance will inform immunization policy when RSV vaccines become available. Methods The WHO RSV surveillance pilot leverages the capacities of the Global Influenza Surveillance and Response System (GISRS). Hospitalized and non‐hospitalized medically attended patients of any age were tested for RSV using standardized molecular diagnostics throughout the year in fourteen countries. An extended severe acute respiratory infection (extended SARI) or an acute respiratory infection (ARI) case definition was used that did not require fever as a criterion. Results Amongst 21 221 patients tested for RSV between January 2017 and September 2018, 15 428 (73%) were hospital admissions. Amongst hospitalized RSV‐positive patients, 50% were aged <6 months and 88% <2 years. The percentage of patients testing positive for RSV was 37% in children <6 months and 25% in those aged 6 months to 2 years. Patients with fever were less likely to be RSV positive compared to those without fever (OR 0.74; 95% CI: 0.63‐0.86). For infants <6 months, 29% of RSV ARI cases did not have fever. Conclusion Requiring fever in a case definition for RSV lowers the sensitivity to detect cases in young children. Countries should consider ways to leverage the GISRS platform to implement RSV surveillance with an augmented case definition amongst the young pediatric population.


| INTRODUC TI ON
Globally, RSV is the leading infectious cause of respiratory morbidity and mortality in children aged <5 years. Annually, there are an estimated 33 (uncertainty range 21-50) million episodes of RSV-associated acute lower respiratory infection (ALRI), 3.2 (uncertainty range 2.7-3.8) million RSV hospitalizations, and 59 600 (uncertainty range 48 000-74 500) in-hospital RSV deaths. 1,2 RSV vaccine research and development efforts have progressed significantly in recent years, with some nineteen vaccines and two new generation monoclonal antibody candidates in various stages of clinical trials. It is possible that a maternal RSV vaccine to prevent RSV infection in young infants may be licensed in the next few years. 3 Though RSV disease occurs across all ages, 4-6 it disproportionately affects children <2 years. 7 The WHO Strategic Advisory Group of Experts on Immunization (SAGE) has recommended improved case definitions, surveillance, and disease estimates for RSV especially in low-and middle-income countries where the burden is likely to be high. 8 There is a need for improved RSV surveillance to better understand seasonality and disease burden in different countries. 9 RSV surveillance is often a by-product of influenza-like illness (ILI) or severe acute respiratory infection (SARI) surveillance. 10,11 However, these case definitions require a history of fever or measured fever and consequently may miss up to a significant proportion of RSV infections especially in younger children. [12][13][14] In 2016, WHO piloted a RSV surveillance strategy that leverages the capacities of the Global Influenza Surveillance and Response System (GISRS) using a broadened case definition. 15 It aimed to establish laboratory and epidemiological standards for RSV detection to improve the understanding of seasonality, disease burden, and age-groups at highest risk. Using the data collected in this pilot study, this paper aims to describe the clinical predictors for RSV presentation and evaluate the performance of the extended SARI and ARI case definitions for RSV surveillance.

| ME THODS
Surveillance sites from fourteen countries, from all six WHO regions, participated in the pilot study. Countries were selected based on having a WHO-designated National Influenza Centre and/or a national public health laboratory, a strong national influenza surveillance system and an interest in participation in the pilot. Countries were required to test 1000 patients annually for RSV (250 patients in each of the four age groups-<6 months, 6 months to <5 years, 5 to <65 years, and 65 years and more).  16 Specimens from patients with reported or measured fever were additionally tested for influenza and subtyped using standardized rRT-PCR assays. 17

| Laboratory testing
All laboratories participated in an external quality assurance program   England reported data only from their General Practitioner sentinel surveillance ( Abbreviations: ARI, acute respiratory infection; ILI, influenza-like illness; RSV, respiratory syncytial virus; SARI, severe acute respiratory infection. a Apnea defined as temporary cessation of breathing from any cause. b Shock defined as lethargy, fast breathing, cold skin, prolonged capillary refill or weak pulse. logistic regression was used to determine clinical predictors for RSV cases. The SARI and ILI case definitions were evaluated against the extended SARI and ARI definitions that did not require fever. The relative sensitivity was estimated as it was not known how many RSV cases were missed who did not fit the extended SARI or ARI case definitions. The positive predictive value (PPV) was calculated for the age-specific percent positivity cumulated over the study period. We also evaluated the individual perfor-

| Clinical characteristics
Amongst hospitalized patients, RSV percent positivity was highest (37%) in infants <6 months, 25% in children 6 months to 2 years, and 12% in children 2-5 years age. RSV percent positivity was about 4% in adults and older adults aged 65 years and more. RSV percent positivity was lower but showed similar age trends amongst non-hospitalized patients. In contrast, percent positivity for influenza amongst hospitalized patients was 3%-7% in children <2 years, 11% in adults aged 18 to <65 years, and 14% in older adults aged over 65 years ( Figure 2). The age-group stratified RSV percent positivity varied across countries. RSV predominated over influenza in children <2 years age (especially infants <6 months) whereas the reverse was seen in older adults aged 65 years and more. This trend was seen for both hospitalized and non-hospitalized patients across all the participating countries ( Figure 3).
Amongst hospitalized patients <6 months, apnea was significantly more prevalent amongst RSV-positive infants (8.3%) than those who tested negative (6.0%) (odds ratio (OR) 1.42; 95% CI: The presence of either reported or measured fever was significantly less likely amongst RSV cases than RSV-negative children <2 years than in those aged 6 months to 2 years. Similar trends in clinical predictors were seen in older children, adults, and older adults but were generally not statistically significant. Presence of a pre-existing illness was not significantly associated with RSV infection except in children aged 5-18 years. Clear statistically significant trends for RSV infection for most clinical predictors could not be ascertained for patients for any of the age-groups attending outpatient clinics in a primary care setting, but these analyses were based on smaller numbers of patients and the confidence intervals on odds ratio estimates were generally wide.
F I G U R E 1 Age distribution of specimens tested, WHO RSV surveillance, 2017-18

| Case definition
Amongst hospitalized infants <6 months age, 29% of those RSV cases that otherwise had been captured by an extended SARI case definition were missed after inclusion of measured or reported fever. The positive predictive value (PPV) of the SARI case definition that included fever was 34% for infants <6 months and 24% for children between 6 months to <2 years. The ability of fever to correctly identify the presence or absence of RSV infection was poor (area under the curve (AUC) -0.31). The percent of missed cases was lower (18-20%) in children 6 months to <5 years age. Apnea and sepsis individually had low sensitivity (8.3% and 4.7%) but were highly specific (93% and 89%) for RSV infection in infants <6 months. The addition of wheeze to the extended SARI case definition reduced the sensitivity to 36% in hospitalized infants <6 months age and to about 32-33% in children aged 6 months to 5 years. Similar trends were seen amongst non-hospitalized children, albeit with lower certainty. Amongst hospitalized older adults aged 65 years and more, the sensitivity of the SARI case definition after inclusion of fever and wheeze was 78% and 40%, respectively. Amongst non-hospitalized older adults aged 65 years and more, the addition of fever reduced the sensitivity to 27%. However, the addition of wheeze to the extended SARI case definition increased the sensitivity to 98%, respectively (Table 5).

| D ISCUSS I ON
The lack of a global uniform surveillance case definition for RSV complicates the interpretation of surveillance data. In the WHO RSV surveillance pilot, the use of an extended SARI or an ARI case definition substantially increased the number of RSV infections detected also seen in other studies. 12  including wheeze as a criteria reduces its sensitivity in children. 25 Apnea, though it lacked sensitivity, was a significant clinical predictor for severe RSV infection requiring hospitalization in infants <6 months. 22,[26][27][28] The reason for sepsis to be significantly less common amongst RSV-positive hospitalized young infants in this study in contrast to other studies 29,30 is unclear.
The RSV surveillance case definition is not intended to modify or replace the SARI or ILI case definition for influenza surveillance.
Countries reported challenges in the adoption of the extended SARI case definition in the early stages of the RSV surveillance which were resolved through training. In practice, the physician or nurse in a sentinel hospital engaged in both RSV and influenza surveillance, screened patients with acute onset cough or shortness of breath, and collected an appropriate respiratory specimen.
Information on the presence or absence of fever was recorded in the specimen requisition form and sent to the laboratory along with the respiratory specimen. At the laboratory, all the specimens were tested for RSV, whereas those specimens from patients with fever were additionally tested for influenza. Moreover, the results for influenza were reported to FluNet only for those patients with fever. Notwithstanding the additional burden of reporting, this ensured that the RSV surveillance did not disturb the influenza surveillance system but complemented it by targeting a very young age-group that is important, yet often underrepresented in influenza surveillance.
The WHO RSV surveillance pilot had several limitations, and its early findings need to be interpreted with some caveats.