Abstract / Summary
Introduction: Japan's COVID-19 Contact-Confirming Application (COCOA) stopped functioning in November 2022, and the records held on users' devices were lost with it. We describe the exposure records preserved by two users during the seventh wave of COVID-19 in Japan and the changes around the revision of full mandatory case reporting on September 26, 2022. Methods: We analyzed exposure windows (date, duration, and minimum attenuation) contained in files exported from the app. Observer 1 worked at a hospital in Tokyo and commuted almost exclusively by private car; Observer 2 lived in Kyoto Prefecture and spent most of the time at home. Seven-day rolling sums of exposure windows were compared descriptively with daily new case counts in each prefecture. Results: During the common pre-revision period (April 3 to September 25, 2022; 176 days), Observer 1 recorded 358 exposure windows and Observer 2 recorded 261. The rolling sums rose in July and fell in September, as did new case counts (r-squared = 0.807 and 0.397), but week-on-week changes corresponded weakly (r-squared = 0.183 and 0.002). Of the windows in this period, 97.2% and 93.1% had a minimum attenuation of 65 dB or more, the bucket given the lowest weight in the notification decision. After the revision, Observer 1 recorded three windows in 27 days (45 in the preceding 27 days), and Observer 2 recorded none in 23 days (25 in the preceding 27 days). Conclusions: The exposure-window records of two users rose and fell with the seventh wave, although the correspondence of week-to-week changes was limited. Recorded windows decreased after the revision, but the effects of the policy change, the supply of diagnosis keys, app usage, and operating-system updates cannot be separated. This descriptive report does not evaluate the preventive effectiveness of COCOA. It shows that indicator definitions, retrieval dates, and the history of software and policy changes are essential when records left on users' devices are analyzed after the fact.