Respiratory syncytial virus-associated deaths in the United States according to death certificate data, 2005 to 2016

Background and aims: In the United States, respiratory infections due to respiratory syncytial virus (RSV) cause an estimated 57 000 hospitalizations annually among children aged <5 years and 177 000 hospitalizations among adults aged ≥65 years. RSV-associated deaths are less well described. It will be important to establish a baseline of RSV-coded deaths prior to the introduction of vaccines, immunoprophylaxis products, and anti-viral therapies currently in development.

Methods: US death certificate data for all ages from 2005 through 2016 were compiled through the National Center for Health Statistics. Deaths with International Classification of Diseases codes of J12.1 (RSV-pneumonia), J20.5 (RSV-bronchitis), or J21.0 (RSV-bronchiolitis) assigned as either the underlying cause of death or a contributing cause of death were considered "RSV-associated" for this analysis.

Results: Among 30.5 million deaths, 1001 (.003%) were assigned an RSV-associated cause of death as follows: 697 (69.6%) RSV-pneumonia, 277 (27.7%) RSV-bronchiolitis, 17 (1.7%) RSV-bronchitis, and 10 (1.0%) with multiple RSV-associated causes. Most deaths were among children <5 (47.8%) and adults ≥50 (40.4%) years of age. Almost half (46.8%) had an RSV-associated cause as the primary underlying cause of death. The average annual number of RSV-associated deaths did not significantly change among those aged <5 and 5 to 49 years. However, RSV-pneumonia deaths among adults aged ≥50 years increased from 17.6 in 2005 to 2012 to 57.3 in 2013 to 2016 (P value <.0001).

Conclusions: From 2005 to 2016, the number of recorded RSV-associated deaths increased, primarily due to greater RSV-associated pneumonia deaths among older adults since 2013. The reasons for this increase are not clear but likely reflect increased testing for RSV among adults. The number of RSV-associated deaths according to death certificates compared with estimates derived from active, laboratory-confirmed surveillance and models using hospital administrative data suggests that counts from death certificates are a large underestimation, particularly among adults.