Traveler-derived importation risk underestimates regional disease activity: Evidence from Okinawa, Japan

Traveler-based surveillance may infer infection risk at places of origin, yet its validity may depend on local testing practices and travel behavior. We evaluated whether the importation risk estimated from travelers can reliably reflect population-based reported case rates across regions of departure within Japan.

Using Okinawa Prefecture as a sentinel destination, we compiled laboratory-confirmed COVID-19 cases diagnosed during stays in Okinawa among residents from outside the prefecture and paired them with monthly inbound passenger volumes by origin region. We estimated monthly importation risk (imported cases per passenger) and compared it with reported case rates at origin. We assessed regional and wave-specific heterogeneity and conducted lag sensitivity analyses by shifting origin case time series forward by unshifted (0-day), 7-day, and 14-day shifts before monthly aggregation.

Importation risk and origin case rates were positively associated but heterogeneous by region. Overall correlation was robust to timing offsets (Lag 0: 0.60 [0.46-0.71]; Lag 7: 0.56 [0.42-0.68]; Lag 14: 0.48 [0.32-0.61]). High-volume regions contributed most imported diagnoses, yet alignment with origin case rates varied across regions and waves, from consistently strong to weak or inconsistent. Patterns were consistent with heterogeneity in surveillance intensity and travel behavior.

Traveler-based importation risk was generally lower than and did not uniformly track origin reported case rates across regions and epidemic waves. It should therefore be interpreted as a complementary sentinel signal rather than a stand-alone proxy for regional disease activity, particularly when routine surveillance is delayed or uneven.