Assessing the impact of COVID-19 interventions on influenza-like illness in Beijing and Hong Kong: an observational and modeling study

Table 1 The potential impact of each individual and combined NPI on weekly ILI counts in Beijing and the Hong Kong SAR, 2020–2021

Interventions	Beijing	Hong Kong SAR
C1: Closings of schools	− 0.086 (− 0.158, − 0.014)*	0.053 (0.008, 0.099)*
C2: Closings of workplaces	− 0.128 (− 0.188, − 0.069)*	0.059 (0.021, 0.097)*
C3: Cancelling public events	− 0.178 (− 0.234, − 0.121)*	0.004 (− 0.043, 0.051)
C4: Limits on gatherings	− 0.046 (− 0.074, − 0.018)*	− 0.001 (− 0.036, 0.035)
C5: Closing of public transport	− 0.096 (− 0.177, 0.015)− *	0.426 (0.272, 0.579)*
C6: Staying in place or at home	− 0.081 (− 0.139, − 0.022)*	0.186 (0.090, 0.282)*
C7: Restrictions on internal travel	− 0.166 (− 0.224, − 0.108)*	0.139 (0.000, 0.278)
C8: Restrictions on international travel	0.059 (0.002, 0.115)*	− 0.113 (− 0.161, − 0.065)*
H6: Mask wearing outside the home	− 0.145 (− 0.270, − 0.021)*	0.071 (0.003, 0.139)*
C12: Closings of schools and workplaces	− 0.168 (− 0.245, − 0.091)*	0.062 (0.018, 0.107)*
C34: Cancelling public events or gatherings	− 0.099 (− 0.140, − 0.057)*	0.005 (− 0.050, 0.059)
C567: Restrictions on internal movement	− 0.088 (− 0.118, − 0.057)*	0.419 (0.246, 0.592)*
All NPIs as a whole	− 0.282 (− 0.378, − 0.186)*	0.079 (− 0.018, 0.177)

The numbers presented are the coefficient with its 95% CI (data in the round bracket) of each intervention in GAM models (Eq. 8), to reflect their potential impact on weekly ILI counts. A positive coefficient represents that the intervention, or its intensity might not be statistically associated to the decrease of weekly ILI counts across the whole study period, while a negative coefficient represents the implementation of COVID-19 NPIs might be associated with the reduction of ILI in the city. Besides, population density change and non-parametric terms of s(time), s(mean temperature change), and s(relative humidity change) were also incorporated into the GAM models to adjust for potential confounding factors between cities. NPI non-pharmaceutical intervention, ILI influenza-like illness, Hong Kong SAR Hong Kong Special Administrative Region, GAM generalized additive model. CIs, Confidence intervals; COVID-19, Coronavirus disease 2019. *P < 0.05

ISSN: 2049-9957