A pluralistic approach to interpreting surveillance data: The likely decline in the occurrence of invasive pneumococcal disease in Japan during the COVID-19 pandemic.

Purpose: To systematically investigate whether the decline in invasive pneumococcal disease (IPD; caused by Streptococcus pneumoniae) notifications during the COVID-19 pandemic, based on Japan's national infectious disease surveillance system, was due to a reduction in incidence.

Methods: Considering potential surveillance biases from changes in healthcare-seeking, testing/diagnosis, or reporting, we compared the following for the 12-month period before vs. after the state of emergency declaration (week 15, 2020): notification rates of IPD, invasive Haemophilus influenzae disease (IHD), and legionellosis (clinically similar bacterial diseases, but unlike IPD/IHD, legionellosis is not transmitted person-to-person and serves as a negative control); surveillance timeliness indicators; notifications restricted to severe cases; and S. pneumoniae detections accounting for test frequency.

Results: Following the declaration, IPD notifications decreased by 60 % (notification rate ratio = 0.39, 95 %CI = 0.37-0.42). While IHD also decreased markedly (0.37, 0.31-0.44), legionellosis declined little (0.83, 0.78-0.88). None of the diseases showed delays in timeliness. Restricted to severe cases, respectively similar trends were observed for IPD, IHD, and legionellosis, with respectively similar death trends in census data. Additionally, decrease in S. pneumoniae detections greatly exceeded that in test counts, substantially decreasing test positivity.

Conclusions: Based on our pluralistic approach, reduced IPD notifications could not be explained by surveillance biases and was likely due to decreased incidence.