An ARIMA Model to Quantify the Impact of COVID-19 Nonpharmaceutical Interventions on the Epidemiology of Varicella in China, 2005–2020

Abstract Background: Varicella is a common respiratory infectious disease. Since 2020, non-pharmaceutical interventions targeting SARS-CoV-2 greatly affected the activity of other respiratory-transmitted diseases. The aim of our study was to quantify changes of varicella activity in China associated with implementation of COVID-19 non-pharmaceutical interventions (NPIs) and raise awareness of the possibility of an “immunity debt” that may increase risk of varicella outbreaks and activity. Methods : We developed and fit an autoregressive integrated moving average (ARIMA) model with historical data on monthly varicella outbreak cases in China that were reported between January 2005 and December 2018. The selected, optimal model was used to predict outbreak cases of varicella from January 2019 to June 2020. We evaluated the reliability of the model by comparing the actual and predicted values for 2019. We then used the model to compare the actual and predicted values for first half of 2020 to quantify the impact of the COVID-19-NPIs on varicella. Results : Four models passed parameter tests (all p<0.05) and the residual tests (all p>0.05): ARIMA（2，1，2)(1，1，0) 12 , ARIMA（2，1，2)(1，1，1) 12 , ARIMA（2，1，2)(2，1，0) 12 , and ARIMA（2，1，2)(0，1，2) 12 . The model, ARIMA (2,1,2) (1,1,0) 12 was determined to be the optimal model, with a high goodness of fit R 2 (0.488) and a small, standardized BIC (14.174). We showed that the optimal model fit well because actual outbreak cases in 2019 were within 95% confidence intervals of the model-predicted outbreak cases for 2019. During 2020 however, actual outbreak cases were fewer than the number of cases predicted at the lower limits of the model’s 95% confidence intervals, and actual 2019 cases were 93.37% lower than the number of cases predicted for the first half of 2020 had there been no COVID-19 NPIs. Conclusions: It is feasible to quantify the impact of COVID-19-targeted NPIs on varicella outbreak cases using a time series model. Our optimal ARIMA model showed that varicella activity was significantly decreased by COVID-19 NPIs. The marked absence of varicella during the first half of 2020 likely led to an immunity debt, thus increasing the likelihood of future varicella outbreaks, off-season varicella activity, and pressure on medical services when COVID-19 NPIs are relaxed.