During the 2014–2015 US influenza season, 256 cases of influenza-associated acute parotitis occurred among residents of 27 states and were reported to CDC. Reported cases ranged widely in age but occurred primarily among school-aged children. Patients described painful facial swelling, consistent with acute parotitis, which developed shortly after the onset of systemic or respiratory symptoms. Swelling lasted a median of 4 days before resolving. Seven cases were hospitalized during their illness, including 2 patients who were admitted to the intensive care unit.

Prior to the 2014–2015 influenza season, only 18 cumulative cases of influenza-associated parotitis were reported during the 1975–1976, 1984–1985, 2004–2005, and 2007–2008 influenza seasons in the United States, Canada, Chile, and Spain [4–10]. Additional laboratory testing revealed influenza A(H3N2) virus infections in all cases tested (15/15). Furthermore, the predominant influenza viruses during these seasons were influenza A(H3N2) viruses [17–20]. Additionally, circulating viruses characterized during these seasons had evidence of antigenic differences (drift) from prior circulating influenza A(H3N2) viruses [18–20].

Influenza A(H3N2) viruses were also the predominant viruses circulating during the 2014–2015 influenza season in North America and Europe. The majority of circulating A(H3N2) viruses were antigenically and genetically drifted from the influenza A(H3N2) component in the 2014–2015 Northern Hemisphere influenza vaccine and from prior circulating A(H3N2) viruses [21]. Further genetic characterization of A(H3N2) viruses in the United States demonstrated that the majority of the drifted viruses were in a newly emerged HA genetic group, labelled 3C.2a. Other HA genetic groups of the A(H3N2) viruses also circulated during the season, including 3C.2b, 3C.3, 3C.3a, and 3C.3b, with local variability [22]. The 3C.3 and 3C.3b viruses are closely related, yet distinct, from 3C.2a viruses in phylogenetic analysis, with several amino acid changes at antigenic sites [16, 22, 23].

In addition to the cases reported to the CDC, cases of influenza-associated parotitis were also reported in Canada, England, Scotland, and separately in Pennsylvania during the 2014–2015 influenza season [24–27]. Combined with our report, 99% (258/261) of all patients with influenza-associated parotitis reported during the 2014–2015 influenza season had influenza A virus infections and 95% (246/258) were infected with A(H3N2) viruses. Though infrequent during the 2014–2015 season, parotitis was also described in 1 patient after influenza A(H1N1pdm09), reported from Scotland [25], and in 1 patient with B virus infection, reported here. Of the 246 A(H3N2) viruses, 114 were further sequenced and 96% (109/114) were classified in the 3C.2a HA genetic group. Our case ascertainment efforts identified 2 patients infected with A(H3N2) viruses in the 3C.3b HA genetic group, both reported from Pennsylvania, which had a greater proportion of 3C.3b viruses circulating than other parts of the country [22], and 1 patient infected with an A(H3N2) virus from the 3C.3 HA genetic group.

Our case-control study revealed a significant association of parotitis with male sex. We are not aware of anatomical differences in male and female parotid glands that could account for this association [28], and male predominance among patients with mumps parotitis, to our knowledge, has not been described [29–31]. The recent report from British Columbia, Canada, noted 88% (14/16) of cases were males [24]. Additionally, we found an association between a history of prior mumps infection and influenza-associated parotitis, mainly among adults; the etiology and clinical relevance of this finding is unclear. Patients with influenza-associated parotitis, more so than controls, were less likely to receive influenza antiviral therapy but were more commonly given an antibiotic during their illness, possibly because healthcare providers were concerned about bacterial etiologies of parotitis. We do not know for certain whether antibiotics were appropriate for a patient; however, nationally we see that antibiotics tend to be overprescribed for patients with influenza [32].

Our investigation is subject to limitations. First, interview responses may be subject to recall errors because most participants were interviewed 1–3 months after their illnesses, and misclassification might occur with self-reported exposures. Second, no cases had laboratory evidence of mumps infections; however, detection of mumps virus in buccal and serum specimens is challenging, particularly among vaccinated persons [33–35]. So, while unlikely given that none of the cases had epidemiologic links to mumps outbreaks, we could not definitively rule out mumps. Third, we used different strategies to find cases and controls, and this differential ascertainment might have contributed to observed differences in self-reported symptoms. Fourth, case specimens were primarily buccal swabs, and control specimens were nasopharyngeal or oropharyngeal swabs. Buccal swabs are currently not an approved specimen type for clinical influenza diagnostic testing; however, low cycle threshold values from case buccal swabs tested by RT-PCR suggest a large amount of influenza virus was present (median cycle threshold, 29.1, range, 19.1–38.5 among 98 specimens). Fifth, while influenza A(H3N2) virus was detected in all subtyped case specimens, fewer case than control specimens had complete genetic sequences. Next-generation sequencing has a lower limit of detection than RT-PCR; therefore, it is likely that case specimens that could not be sequenced contained lower viral loads. This might result from physiologic differences in viral shedding from the parotid gland or with the timing of specimen collection during the course of illness. Finally, additional sequencing investigations of case buccal specimens are ongoing to determine what bacterial pathogens may be codetected and whether these pathogens may contribute to the propensity to develop parotitis with influenza virus infection [36].

Additional investigations of influenza-associated parotitis are warranted to better understand the etiology and epidemiologic features of this uncommon clinical finding. After finding few epidemiologic risk factors in our study, we focused on describing the sequence differences in the influenza virus. Specifically, we looked for differences in the virus’s HA gene because of this gene’s role in viral attachment to host epithelial cells and its propensity for genetic change. We did not identify any changes in the HA gene that were specific to case viruses; however, detailed sequence analysis of the 7 other influenza gene segments might help further explain differences in the case and control influenza viruses. Furthermore, there may be host cell polymorphisms, environmental factors, or unmeasured exposures that may be risk factors for influenza-associated parotitis. Further studies would also be helpful to understand how influenza viruses interact with epithelial cells on the parotid gland and whether influenza viruses circulating in 2014–2015 had different tropism to the parotid gland.

We believe that this outbreak is reflective of increased occurrence during the 2014–2015 season rather than an artifact of enhanced surveillance and case finding efforts. Multiple independent reports from public health departments and clinicians were received by CDC’s Influenza Division and Division of Viral Disease’s mumps investigation team prior to setting up nationwide active case finding. However, it is possible that influenza-associated parotitis was detected more often because targeted mumps investigations and reductions in mumps incidence during the past few decades has made non-mumps parotitis easier to detect. Efforts to establish population-based surveillance of influenza-associated parotitis could help describe its incidence, associated complications, and the public health impact during influenza seasons.

In conclusion, this outbreak of influenza-associated parotitis is the largest ever reported. Compared with recruited controls and broader virologic surveillance in the United States, influenza-associated parotitis seems to occur primarily with influenza A(H3N2) virus infection, although it does not appear to be exclusively associated with the newly emerged 3C.2a HA genetic group of A(H3N2) viruses. Our findings suggest that including influenza in the differential diagnoses among patients who present with acute parotitis may be prudent during influenza season, particularly during seasons dominated by influenza A(H3N2) virus infection and when respiratory symptoms precede parotitis. Testing for influenza would support appropriate treatment with influenza antiviral treatment, according to treatment guidelines [37], and may also reduce inappropriate use of antibiotics. Additionally, differentiating mumps-associated parotitis from influenza-associated parotitis is important from a public health perspective. Currently, mumps is relatively uncommon and prone to outbreaks. When mumps is suspected in a person, an intensive public health response may be warranted, including contact tracing, vaccination clinics, and patient isolation. In the United States, we experience annual widespread epidemics of influenza, and such a public health response is not generally needed. Due to these substantial differences, public health officials should consider influenza virus infection when investigating occurrences of parotitis during the influenza season.