Abstract
This study investigates the incidence of Class B respiratory infectious diseases (RIDs) in China under the Coronavirus disease 2019 (COVID-19) epidemic and examines variations post-epidemic, following the relaxation of non-pharmaceutical interventions (NPIs). Two-stage evaluation was used in our study. In the first stage evaluation, we established counterfactual models for the pre-COVID-19 period to estimate expected incidences of Class B RIDs without the onset of the epidemic. In the second stage evaluation, we constructed seasonal autoregressive integrated moving average intervention (SARIMA-Intervention) models to evaluate the impact on the Class B RIDs after NPIs aimed at COVID-19 pandemic were relaxed. The counterfactual model in the first stage evaluation suggested average annual increases of 10.015%, 78.019%, 70.439%, and 67.799% for tuberculosis, scarlet fever, measles, and pertussis respectively, had the epidemic not occurred. In the second stage evaluation, the total relative reduction in 2023 of tuberculosis, scarlet fever, measles and pertussis were − 35.209%, − 59.184%, − 4.481%, and − 9.943% respectively. The actual incidence declined significantly in the first stage evaluation. However, the results of the second stage evaluation indicated that a rebound occurred in four Class B RIDs after the relaxation of NPIs; all of these showed a negative total relative reduction rate.
Similar content being viewed by others
Introduction
The Coronavirus disease 2019 (COVID-19) pandemic, caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), occurred at the end of 20191. This pandemic has resulted in a sharp rise in the global morbidity and mortality2. In January 2020, as the virus spread across China, the country immediately implemented emergency non-pharmaceutical interventions (NPIs) to mitigate its impact, including social distancing, mandatory face masks, school closures, bans on mass gatherings, and the suspension of public transport3,4. Studies have shown that these NPIs aimed at COVID-19 epidemic also have a certain impact on the inhibition and reduction in the incidence of respiratory diseases because the same transmission route5,6,7. Lai et al. reported social distancing substantially reduced the incidence rate of tuberculosis during the epidemic8.
Respiratory infectious diseases (RIDs) have continuously presented challenges for public health and social development since the twentieth century. In China, notifiable infectious diseases are categorized into A, B, and C classes, with Class B RIDs often posing a higher mortality risk and more significant health impact than Class C. Except for COVID-19, Class B RIDs include 9 types of notifiable RIDs (infectious atypical pneumonia, human highly pathogenic avian influenza, influenza A H7N9, tuberculosis, measles, pertussis, scarlet fever, diphtheria, epidemic cerebrospinal meningitis). Tuberculosis, scarlet fever, measles and pertussis are the four major Class B RIDs in China, which have resulted in a serious public health burden in China for a long time. The research by Liu et al. in Jingzhou of Hubei province concluded that compared with the actual values of 6–53 weeks in 2020 and the same period in 2015–20199, among infectious diseases of different transmission routes, the incidence rate of RIDs intervened by NPIs had the greatest decrease, with a relative reduction rate of 61.53%. He et al. showed that scarlet fever and pertussis decreased by 73.13% and 76.63% respectively in China in 202010. A survey conducted in Anhui province during the epidemic from 2020 to 2021 showed that the incidence of tuberculosis had decreased, but the elderly people between 66 and 75 years old were the largest proportion of patients11. And research on Children’s Hospitals in Jinan, Shandong Province, China, indicated that six common RIDs in children including measles, declined before and during the COVID-19 epidemic12. However, the vast majority of research was limited to cross-sectional statistical comparisons, without considering pre-existing potential short-term and long-term trends13.
In order to better understand the impact of NPIs on the incidence of Class B RIDs in China, it is of great significance to establish appropriate predictive analysis models. The interrupted time series (ITS) design, a robust quasi-experimental method, is optimal for inferring causality when randomized trials are impractical14. In addition, ITS can vertically track the results before and after intervention for analysis, rather than being limited to horizontal research. Consequently, ITS analysis was used in our study, which has been commonly applied in estimating the influence of large-scale health interventions15,16,17.
There are two key parts to establish ITS analysis. The first part involves establishing a counterfactual model. The definition of “counterfactual” is to extrapolate potential trends before intervention to the post-intervention and to predict Yt (“counterfactual”) without intervention using the autoregressive integrated moving average (ARIMA) models, so as to determine how the observed values deviate from this prediction. Secondly, we need to define the expected impact types of interventions on outcomes and establish the intervention impact models18. While segmented regression was widely used in ITS analysis, it assumes linear trends which do not fit the seasonal and cyclical nature of most infectious diseases. Therefore, a seasonal autoregressive integrated moving average (SARIMA) model was carried over into our study as it is able to eliminate seasonality and periodicity through differencing as well as identifying the long-term trend of the data.
A distinctive feature of our research is the two-stage evaluation, which is not found in previous studies. It not only includes the evaluation of the counterfactual part, but also the evaluation of changes after the intervention was relaxed, illuminating the intervention's influence on the event under study.
Methods
Data collection
We obtained monthly reported case counts of tuberculosis, scarlet fever, measles, and pertussis for China spanning from January 2004 to December 2023, sourced from the National Health Commission of the People’s Republic of China (http://www.nhc.gov.cn/). The population data used to calculate incidence rate were retrieved from the National Bureau of Statistics website (http://www.stats.gov.cn/). The initiation of the Wuhan lockdown in January 2020 served as the intervention time point and China’s comprehensive relaxing of epidemic restrictions in January 2023 served as the relaxation of COVID-19 pandemic NPIs. Accordingly, we designated January 2004 to December 2019 as the pre-intervention period, January 2020 to December 2022 as the during-intervention period, and January 2023 to December 2023 as the post-intervention period. In this time series analysis, we employed 228 monthly observations to construct our models and used the subsequent 12 monthly observations to ascertain the impact on Class B RIDs following the relaxation of COVID-19 pandemic NPIs. For this study, we used new cases of tuberculosis, scarlet fever, measles, and pertussis reported nationwide during the study. There was no missing data in our study.
Study design
The research process is described in Fig. 1 below. Initially, we calculated the incidence rate of four major Class B RIDs and compared the changes of incidence rate before and during the epidemic. Subsequently, we conducted a two-stage evaluation. The first stage involved constructing a “counterfactual” model by using pre-intervention observed data to estimate and evaluate the monthly incidence changes of the four major Class B RIDs without onset of epidemic. In the second stage, we used pre-intervention and during-intervention observed data to establish a seasonal autoregressive integrated moving average intervention (SARIMA-Intervention) model to estimate and evaluate the monthly incidence rate changes of the four major Class B RIDs after the NPIs were relaxed.
Statistical analysis
To control the confounding factors in population size, we calculated monthly and yearly incidence rate for analysis, which the average value of monthly and yearly new reported cases divided by the total population (per 100,000 population). The growth rate = [(the average incidence rate between 2020 and 2022 − the average incidence rate between 2004 and 2019)/the average incidence rate between 2004 and 2019] × 100%. We utilized the Box–Cox transformation to stabilize the variance in the monthly incidence rates of measles, scarlet fever, and pertussis, which demonstrated instability over time. However, the tuberculosis data did not require this transformation. We performed ordinary differencing on the time series to remove its trend, and used seasonal differencing to eliminate its seasonality, in order to make it stable. Augmented Dickey–Fuller (ADF) Unit Root tests were used to detect whether the time series is stationary19. Evidence of the analysis of stability in the data has been included in the supplementary material. The counterfactual model selection was guided by the Akaike Information Criterion (AIC), while the Ljung–Box test was used for model residual analysis, with P > 0.05 indicating white noise residuals20. In addition, root mean square error (RMSE) and mean absolute percentage error (MAPE) were applied to evaluate the goodness-of-fit of constructed models, with lower RMSE and AIC values indicating a better fit. MAPE values were categorized as accurate forecasting (< 10%), good forecasting (10–20%), reasonable forecasting (20–50%), and inaccurate forecasting (> 50%).
In the two-stage evaluation, relative reduction as an evaluation indicator was used to estimate the effect of NPIs before and after epidemic on the incidence rate of Class B RIDs. Relative reduction (%) = [(number of expected incidence − number of observed incidence)/number of expected incidence] × 100%.
In our study, all data analyses and model construction were performed in R software version 4.2.3. P < 0.05 indicates the difference was statistically significant.
Counterfactual model building
A widely used method for constructing ARIMA model is called the Box Jenkins method, which is one of the most common methods applied to predict time series21. It is a well-established approach for ARIMA model identification and selection, parameter estimations and model diagnosis22. Given the pronounced seasonality of the diseases under investigation, we utilized a multiplicative seasonal autoregressive integrated moving average (SARIMA) model to construct the “counterfactual” model. This model incorporated both non-seasonal and seasonal components, denoted as SARIMA (p, d, q) × (P, D, Q) s, where p is the non-seasonal autoregressive order, d is the non-seasonal differencing order, q is the non-seasonal moving average order, P is the seasonal autoregressive order, D is the seasonal differencing order, Q is the seasonal moving average order, and s represents the seasonal cycle.
The counterfactual model was constructed between January 2004 and December 2019 based on the SARIMA model to forecast the expected monthly incidence rates of the four major Class B RIDs from January 2020 to December 2022, assuming the COVID-19 pandemic had not occurred.
SARIMA-Intervention model building
SARIMA-Intervention models based on the ITS analysis as the intervention impact models were established between January 2004 and December 2022 to estimate the influence of the COVID-19 epidemic on the four major Class B RIDs. Furthermore, predicted number estimated by the models between January 2023 and December 2023 were compared with the actual number to evaluate the impact on the four major Class B RIDs after the NPIs were relaxed.
The SARIMA-Intervention model is as follows:
In this Eq. (1), Yt is the tth observation of the time series Y, which consists of a “noise” component after excluding the effect of intervention, Nt, and an intervention output, yt. The yt term can be written as follows:
In this Eq. (2), where §t is intervention input (i.e. step, pulse function), B is a backshift operator (BbYt = Yt−b), δ is the decay rate (0 < δ < 1), ω (B) and δ (B) represent respectively polynomials in B of degree s and r, ω (B) = ω0 − ω1B − … − ωsBs, δ (B) = 1 − δ1B − … − δrBr. In practice, δ (B) and ω (B) normally take a simple form: δ (B) = 1 − δ1B, ω (B) = ω0. The Nt term can be written as follows:
In this Eq. (3), φ (B) and θ (B) represent respectively MA and AR polynomials of the qth and pth orders: φ (B) = 1 − φ1B − φ2B2 − … − φpBp, θ (B) = 1 − θ1B − θ2B2 − … − θqBq. The at term is a series of white noise.
The purpose of conducting ITS analysis to assess interventions is to quantify the impact of implementing the intervention on a given outcome, commonly referred to as the “intervention effect”. The SARIMA model offers a considerable advantage in constructing more possible intervention impact forms between the intervention variables and the outcome variables through a “transfer function”23. Transfer functions describe the relationship between the outcome series Yt and the intervention variables. \(\frac{\upomega (\text{B}){\text{B}}^{\text{b}}}{\updelta (\text{B})}\) mentioned above represents the general form of a transfer function. The SARIMA-Intervention models in our study utilize two primary intervention variables: step function and pulse function.
-
Step function: A sudden, sustained change which takes the value of 0 prior to the intervention and shifts to 1 subsequently.
-
Pulse function: A sudden, temporary change which takes the value of 1 at the time of intervention, and 0 otherwise.
Incorporating transfer functions with the elementary step function and pulse function to represent a range of possible impacts to test these potential effects, in this study we discussed five possible forms of intervention suggested by Box and Tiao24. Comprehensive details pertaining to the identification and estimation of both noise and intervention components within the ARIMA interrupted time series model are expounded upon in elsewhere25. Below, we consider the five possible forms used in our study:
-
Model 1: an immediate change of ω0 units after intervention and persists permanently: Yt = ω0I1t + at.
-
Model 2: an immediate change of ω1 units that presents gradual change at the rate of δ units after the intervention until it reaches a new level: Yt = [ω1/(1 − δΒ)] I1t + at.
-
Model 3: the time series increase or decrease immediately by ω0 after the intervention and return to baseline immediately: Yt = ω0I2t + at.
-
Model 4: the time series increase or decrease by ω1 after the intervention and decay by (1 − δ) each time point: Yt = [ω1/(1 − δΒ)] I2t + at.
-
Model 5: a complex combination of ω0I1t and [ω1/(1 − δΒ)] I2t: Yt = ω0I1t + [ω1/(1 − δΒ)] I2t + at.
Where Yt signifies the incidence rate of four major Class B RIDs at time t, I1t symbolizes the step function, I2t symbolizes pulse function, ω0 indicates zero-order input parameter of a transfer function, ω1 indicates first-order input parameter of a transfer function.
We constructed five above models for each of the four major Class B RIDs. Model selection, guided by parameter estimation and the AIC value, was applied to evaluate the goodness-of-fit of the constructed models and to select the optimal models. We aimed to choose the model with statistically significant parameter estimates and the smallest AIC value as the optimal model.
Results
Incidence rate of four major Class B respiratory infectious diseases in China from 2004 to 2022
The changes in incidence rate of four major Class B RIDs comparing with pre-intervention and during-intervention are shown in Table 1. Between 2020 and 2022, the average yearly incidence rates of four major Class B RIDs are respectively tuberculosis (57.06/100,000), scarlet fever (1.60/100,000), measles (0.07/100,000) and pertussis (1.27/100,000). In contrast, from 2004 to 2019, the average yearly incidence rate for tuberculosis (95.65/100,000), scarlet fever (3.44/100,000) and measles (4.01/100,000), there are a marked reduction in the amount in 2020–2022, decreased respectively by 40.35%, 53.49%, and 98.25%. However, pertussis showing a large growth rate (159.18%). By reviewing the literatures, we found that it may be related to the recurrence of pertussis since 2014, as highlighted in a study by Wang et al.26, the incidence of pertussis was 538% higher in 2019 compared to 2004 in China.
Counterfactual model evaluation results
As shown in Table 2, the value of MAPE indicated a good fit of the constructed models. What’s more, the Ljung-Box test showed that the residual sequences of the models are white noise. Parameter estimations for the counterfactual models of four RIDs shown in Table S1 indicate that all the parameter estimates are statistically significant.
The fitted numbers of the constructed models in 2004–2019 were highly close to the actual numbers. However, the predicted numbers in the absence of COVID-19 for the four diseases exceed the actual numbers in 2020–2022 (Fig. 2). Our findings are consistent with previous related studies and exhibit similar predictive trends27. The incidence of tuberculosis and measles continued to decline, following expected trends, while scarlet fever and pertussis initially drop below expected values, and then showed an upward trend, which showing a rebound in 2021 and 2022.
Each month’s actual incidence, predicted incidence, and relative reduction (%) of four RIDs in China from 2020 to 2022 are shown in Table S2. For tuberculosis, the annual relative reduction between 2020 and 2022 are 9.639%, 7.709% and 13.118% respectively. Correspondingly, the reductions for scarlet fever are 81.211%, 70.992% and 79.575% respectively; for measles, 60.612%, 66.032% and 57.909% respectively; and for pertussis, 79.830%, 81.862% and 45.001% respectively. That is to say, the overall relative reductions between January 2020 and December 2022 for tuberculosis, scarlet fever, measles and pertussis indicated that 10.015%, 78.019%, 70.439% and 67.799% respectively should be increased without the COVID-19 outbreak. Except for a negative relative reduction for tuberculosis in March 2021 and 2022, positive relative reductions were observed in all other months for the four diseases.
Overall, the relative reduction for scarlet fever has been at a higher level, while it was lowest for tuberculosis. The relative reduction for measles declined towards the end of 2020 and 2021, which may be related to the relaxation of NPIs, and as for pertussis, the incidence rate from 2020 to 2021 significantly reduced with a high relative reduction, even with the seasonal trend eliminated.
SARIMA-Intervention model evaluation results
The impact of the intervention on tuberculosis and scarlet fever began immediately in January 2020, with a zero month lag and thus b = 0. In contrast, there was a 1-month delay in measles and pertussis of intervention output for the influence of COVID-19 epidemic on the incidence rate and thus b = 1. Final Noise and Intervention Models for the four RIDs from 2004 to 2022 were shown in Table 3. In the meantime, the final diagnostics using the Ljung–Box test indicate the estimated models were statistically adequate. Here, a total of 228 observations were utilized, ensuring sufficient data.
As for tuberculosis, Model 1 best fitted the data and showed statistically significant parameter estimates. It showed that the incidence decreased immediately (ω0 = − 1.135, P < 0.01) in January 2020 with a sustained low level after the decline. The four other models were considered; the models all showed that the parameters were not statistically significant (detailed in Table S3).
In terms of scarlet fever, Model 5 best fitted the data and showed statistically significant parameter estimates. It showed that a permanent impact of ω0 = − 1.712 units was realized in January 2020 and remained at a low level after the decline. A temporary impact of ω1 = 1.317 units was also realized in January 2020 but began to decay at δ = 0.261 immediately. The four other models were considered; unexpectedly, however, the results of Model 3 showed a significant increase in incidence (ω0 = 0.234, P = 0.039) and Model 4 with incorrect value (δ = − 0.385, P = 0.127) were rejected. Model 1 and 2 showed statistically significant parameter estimates, however, we choose the best Model 5 with the least AIC value (AIC = − 155.59) and optimal goodness of fit (detailed in Table S4).
Regarding measles, Model 2 best fitted the data and showed statistically significant parameter estimates. It showed that the incidence decreased immediately (ω1 = -2.608, P < 0.001) in February 2020 and began to decay gradually at δ = 0.378. The four other models were considered; unexpectedly, however, the results of Model 4 with incorrect value (δ = 1.002, P < 0.001) and Model 5 with a statistically insignificant parameter (δ = 0.235, P = 0.241) were rejected. Model 1 and 2 showed statistically significant parameter estimates, however, we choose the best Model 2 with the least AIC value (AIC = 328.69) and optimal goodness of fit (detailed in Table S5).
As for pertussis, Model 2 best fitted the data and showed significant parameter estimates. It showed that the incidence decreased immediately (ω1 = − 0.753, P < 0.001) in February 2020 and began to decay gradually at δ = 0.755. The four other models were considered; however, the results showed that the parameters of Model 1 (ω1 = − 0.417, P = 0.128), Model 3 (ω1 = − 0.089, P = 0.618) and Model 4 (ω1 = − 0.037, P = 0.721, δ = − 0.813, P = 0.268) and Model 5 (ω0 = 0.770, P = 0.988, ω1 = − 1.188, P = 0.981, δ = 1.003, P < 0.001) were insignificant, so we reject them (detailed in Table S6).
The best fitted models for the four Class B RIDs in China were used to predict, which were demonstrated in Fig. 3 and Table 4. As indicated in the Fig. 3, on the whole, there is a downward trend for tuberculosis, measles and scarlet fever in 2023, while, pertussis is on the rise, still lower than pre-COVID-19 years. That is, if the NPIs continue to exist, the incidence of four Class B RIDs in China is likely to continue this downward trend.
The results listed in Table 4 suggested that, overall, it showed a rebound in the incidence rates for four Class B RIDs after the NPIs were relaxed from January 2023 to December 2023 and the total relative reduction for tuberculosis, scarlet fever, measles and pertussis were − 35.209%, − 59.184%, − 4.481% and − 9.943% respectively, indicating higher actual than predicted incidences. To some extent, it more fully validated the positive impact of NPIs taken against the COVID-19 on Class B RIDs.
Discussion
In our time series investigation, we conducted a two-stage evaluation, constructing both counterfactual and SARIMA-Intervention models. These models were used to estimate the changes in incidence of four Class B RIDs that are attributable to the interventions and pandemic. In the first stage evaluation, we found that the observed incidence of all four diseases was consistently below predicted values, with the exception of tuberculosis in March 2021 and 2022, where the actual incidence exceeded expected levels (detailed in Table S2). In the second stage evaluation, we observed a marked resurgence in the incidence of the four RIDs from January to December 2023, following the easing of NPIs (detailed in Table 4). This pattern indicates that the NPIs implemented in response to the COVID-19 epidemic played a positive role in reducing Class B RIDs in China.
The reductions in tuberculosis, scarlet fever, measles and pertussis after the epidemic have also been revealed in other countries. Research from the United States observed a notable decrease in RIDs cases, with the incidence of measles, scarlet fever, pertussis, and tuberculosis dropping by 93%, 79%, 50%, and 8%, respectively, in Thailand under public health interventions28. Comparable findings have emerged from South Korea29 and Japan30.
Our study corroborates the significant disruption of COVID-19 on the reporting of tuberculosis, scarlet fever, measles, and pertussis cases in China. In terms of the reasons why RIDs are affected during the epidemic, we have discussed as follows. Firstly, measures such as mask-wearing, hand hygiene, respiratory etiquette, and the widespread use of antiviral medications have reduced the transmission of respiratory pathogens in general. Additionally, by reducing gatherings of people, the spread of all respiratory infections has been curtailed. Regarding the main transmission site of RIDs, the closure of schools has significantly reduced the transmission of respiratory infections for children and adolescents31,32. Secondly, the COVID-19 pandemic has placed significant strain on healthcare resources, potentially weakening the capacity for detection and diagnosis of other respiratory infectious diseases. As medical attention and supplies were redirected towards the novel coronavirus, the surveillance and reporting systems for other respiratory infectious diseases may have been compromised, leading to a notable decline in reported cases33. Furthermore, heightened public awareness and proactive health measures post-epidemic have contributed to disease prevention34. Finally, stringent containment and prevention strategies likely delayed the diagnosis and, consequently, the reporting of infectious diseases, leading to diminished reported incidence rates. For instance, research from Shanghai, China intimates that during the epidemic the detection and treatment of tuberculosis would be delayed, and if exists the reduction in the number of COVID-19, it would lead to increase the detection rate of tuberculosis35. A finding in China concluded that the stronger the prevention and control measures, the lower the incidence of seven respiratory diseases27.
The resurgence of certain RIDs following the relaxation of NPIs could be attributed to three primary factors. For the first reason, the concept of “immunity debt” was first proposed by French scholars in May 2021 to illustrate that children’s lack of immunity to conventional infectious diseases is caused by the lack of immune stimulation to pathogens caused by NPIs during the epidemic36. “Immunity debt” may be responsible for the huge surge in RSV infections among children in many European countries after the relaxation of NPIs37. However, British researchers found that certain vaccine-preventable diseases, including mumps, measles, pertussis and tuberculosis were consistently suppressed after NPIs were relaxed38, lower compared to the pre-COVID-19 period, consistent with our research results, nevertheless, this does not mean that a rebound could not be observed beyond a longer observation period39. For the second reason, with the relaxation of non-pharmaceutical intervention measures, such as reducing social distancing and canceling mask mandates, increased contact between people provides opportunities for the spread of respiratory diseases. The reopening of schools and gatherings in collective places create conditions conducive to the transmission of respiratory diseases, especially among children, who are more susceptible to infections due to their immature immune systems. For the third reason, the COVID-19 pandemic prompted an improvement in surveillance efforts, leading to the observation of more RIDs cases even after the NPIs were relaxed. Additionally, following the COVID-19 pandemic, surveillance for other diseases returned to normal levels, which may have facilitated the detection of diseases that were not prioritized during the pandemic. This shift in surveillance focus could indeed contribute to the perceived resurgence in RIDs cases.
We developed models for each of the four major Class B RIDs, with five different models constructed for each disease. In 2023, although the annual relative reduction for these four major Class B RIDs was negative, the model of tuberculosis consistently under-predicted the incidence. We can see from Table 4 that in every month of 2023, the relative reduction for tuberculosis is negative, while the other diseases have both negative and positive values. There may be several reasons for this phenomenon. Firstly, the time series of tuberculosis incidence is different from that of several other diseases. Since 2005, its incidence rate has shown a stable downward trend. During the COVID-19 pandemic from 2020 to 2022, its response to the effects of NPIs was also significantly less than that of several other diseases. The counterfactual model in the first stage evaluation suggested that, had the epidemic not occurred, the average annual increases for tuberculosis, scarlet fever, measles, and pertussis would have been 10.015%, 78.019%, 70.439%, and 67.799%, respectively. Based on the goodness-of-fit of the optimal models for the four RIDs, we ultimately selected Model 1 (Yt = ω0I1t + at) for tuberculosis, Model 5 (Yt = ω0I1t + [ω1/(1 − δΒ)] I2t + at) for scarlet fever, and Model 2 (Yt* = [ω1B/(1 − δΒ)] I1t + at) for both and measles and pertussis in the second stage evaluation. The model of tuberculosis is different from the other models. It means an immediate change of ω0 units after intervention and persists permanently. Secondly, from Fig. 2, we can observe these four RIDs and find that in the first stage evaluation, although NPIs reduced the incidence of tuberculosis, its seasonality remained intact, while the seasonality of the other diseases was noticeably disrupted. In the second stage evaluation, the SARIMA-Intervention model's forecast results show that the seasonality of tuberculosis is also complete, while the seasonality of the other models will exhibit some drift. Consequently, the other diseases have both negative and positive relative reduction values. Finally, based on the principles outlined in the methods section for the SARIMA-Intervention model, this model is designed for processing a single time series, requiring only the time series data of the incidence rates for the relevant diseases as input. Consequently, during the process of establishing the SARIMA-Intervention model in this study, it was not possible to include other external factors that affect these four RIDs, which may also lead to certain deficiencies in the precision of the models established.
Strengths and limitations
To our knowledge, this is the first study to quantify effects of NPIs after they were relaxed on major Class B RIDs using a dataset spanning over a decade of national infectious disease surveillance. Most importantly, a distinctive feature of our study is the use of ITS design for conducting a two-stage evaluation.
Notwithstanding its contributions, our study is not without limitations. Firstly, our studies only roughly summarized the effects of NPIs, which evaluated at some large social points without delving into the nuances of specific interventions. Secondly, our research was based on ecological research, which involves difficulty to control confounding factors and ecological fallacies. Thirdly, policy shifts and improved surveillance efforts during the study period also influenced the intensity of NPIs, warranting further consideration in future analyses. Lastly, the monthly data of our study were from across China, which did not take into account the differences in incidence in different regions.
Conclusion
In conclusion, our findings illustrate the influence of the COVID-19 epidemic on major Class B RIDs in China, with a pronounced decline in incidence rates from 2020 to 2022 during our first-stage evaluation. At the same time, our study further validated the epidemic’s impact on Class B RIDs in the second stage evaluation, which indicated four Class B RIDs showing a rebound after the relaxing of NPIs from January 2023 to December 2023.
Data availability
Data are available in a public, open access repository. The data that support the findings of this study are openly available from the website of the National Health Commission of the People's Republic of China (http://www.nhc.gov.cn/). The population data used to calculate incidence rate were retrieved from the National Bureau of Statistics website (http://www.stats.gov.cn/). The data and R scripts for the counterfactual model and the SARIMA-Intervention model have been provided as supplemental files.
References
Wang, C., Horby, P. W., Hayden, F. G. & Gao, G. F. A novel coronavirus outbreak of global health concern. The Lancet 395, 470–473. https://doi.org/10.1016/S0140-6736(20)30185-9 (2020).
Wang, H. Estimating excess mortality due to the COVID-19 pandemic: A systematic analysis of COVID-19-related mortality, 2020–21. Lancet 399, 1513–1536. https://doi.org/10.1016/s0140-6736(21)02796-3 (2022).
Shen, J. et al. Prevention and control of COVID-19 in public transportation: Experience from China. Environ. Pollut. 266, 115291. https://doi.org/10.1016/j.envpol.2020.115291 (2020).
Liu, N. N. et al. COVID-19 pandemic: Experiences in China and implications for its prevention and treatment worldwide. Curr. Cancer Drug Targets 20, 410–416. https://doi.org/10.2174/1568009620666200414151419 (2020).
Chow, E. J., Uyeki, T. M. & Chu, H. Y. The effects of the COVID-19 pandemic on community respiratory virus activity. Nat. Rev. Microbiol. 21, 195–210. https://doi.org/10.1038/s41579-022-00807-9 (2023).
Dadras, O. et al. Effects of COVID-19 prevention procedures on other common infections: A systematic review. Eur. J. Med. Res. 26, 67. https://doi.org/10.1186/s40001-021-00539-1 (2021).
Chen, B. et al. Changes in incidence of notifiable infectious diseases in China under the prevention and control measures of COVID-19. Front. Public Health 9, 728768. https://doi.org/10.3389/fpubh.2021.728768 (2021).
Lai, C. C. & Yu, W. L. The COVID-19 pandemic and tuberculosis in Taiwan. J. Infect. 81, e159–e161. https://doi.org/10.1016/j.jinf.2020.06.014 (2020).
Liu, T., Wu, Y., Chen, Q., Huang, J. & Luo, M. Impact of non-pharmaceutical interventions on incidence of notifiable infectious disease in Jingzhou, Hubei. Dis. Surveill. 37, 1198–1204. https://doi.org/10.3784/jbjc.202202170045 (2022).
He, Y. et al. Collateral impact of COVID-19 prevention measures on re-emergence of scarlet fever and Pertussis in Mainland China and Hong Kong China. Int. J. Environ. Res. Public Health 19, 9909. https://doi.org/10.3390/ijerph19169909 (2022).
Chen, S., Wang, X., Zhao, J., Zhang, Y. & Kan, X. Application of the ARIMA model in forecasting the incidence of tuberculosis in Anhui during COVID-19 pandemic from 2021 to 2022. Infect. Drug Resist. 15, 3503–3512. https://doi.org/10.2147/idr.S367528 (2022).
Wang, B. et al. Epidemiological characteristics of common respiratory infectious diseases in children before and during the COVID-19 epidemic. Front. Pediatr. 11, 1212658. https://doi.org/10.3389/fped.2023.1212658 (2023).
Soumerai, S. B., Starr, D. & Majumdar, S. R. How do you know which health care effectiveness research you can trust? A guide to study design for the perplexed. Prev. Chronic Dis. 12, E101. https://doi.org/10.5888/pcd12.150187 (2015).
Lopez Bernal, J., Cummins, S. & Gasparrini, A. The use of controls in interrupted time series studies of public health interventions. Int. J. Epidemiol. 47, 2082–2093. https://doi.org/10.1093/ije/dyy135 (2018).
Wagner, A. K., Soumerai, S. B., Zhang, F. & Ross-Degnan, D. Segmented regression analysis of interrupted time series studies in medication use research. J. Clin. Pharm. Therap. 27, 299–309. https://doi.org/10.1046/j.1365-2710.2002.00430.x (2002).
Bernal, J. L., Cummins, S. & Gasparrini, A. Interrupted time series regression for the evaluation of public health interventions: A tutorial. Int. J. Epidemiol. 46, 348–355. https://doi.org/10.1093/ije/dyw098 (2017).
Schaffer, A. L., Dobbins, T. A. & Pearson, S. A. Interrupted time series analysis using autoregressive integrated moving average (ARIMA) models: A guide for evaluating large-scale health interventions. BMC Med. Res. Methodol. 21, 58. https://doi.org/10.1186/s12874-021-01235-8 (2021).
Lopez Bernal, J., Soumerai, S. & Gasparrini, A. A methodological framework for model selection in interrupted time series studies. J. Clin. Epidemiol. 103, 82–91. https://doi.org/10.1016/j.jclinepi.2018.05.026 (2018).
Dickey, D. & Fuller, W. Distribution of the estimators for autoregressive time series with a unit root. J. Am. Stat. Assoc. 74, 427–431. https://doi.org/10.2307/2286348 (1979).
Zeng, Q. et al. Time series analysis of temporal trends in the pertussis incidence in Mainland China from 2005 to 2016. Sci. Rep. 6, 32367. https://doi.org/10.1038/srep32367 (2016).
Hartmann, D. P. et al. Interrupted time-series analysis and its application to behavioral data. J. Appl. Behav. Anal. 13, 543–559. https://doi.org/10.1901/jaba.1980.13-543 (1980).
Box, G. E. P., Jenkins, G. M., Reinsel, G. C. & Ljung, G. M. Time Series Analysis: Forecasting and Control 5th edn. https://doi.org/10.1111/jtsa.12194. (Wiley-Blackwell, 2016).
Helfenstein, U. The use of transfer function models, intervention analysis and related time series methods in epidemiology. Int. J. Epidemiol. 20, 808–815. https://doi.org/10.1093/ije/20.3.808 (1991).
Box, G. E. P. & Tiao, G. C. Intervention analysis with applications to economic and environmental problems. J. Am. Stat. Assoc. 70, 70–79. https://doi.org/10.1080/01621459.1975.10480264 (1975).
McDowall, D., McCleary, R. & Bartos, B. J. Interrupted Time Series Analysis (Oxford University Press, 2019).
Wang, M. et al. ARIMA and ARIMA-ERNN models for prediction of pertussis incidence in mainland China from 2004 to 2021. BMC Public Health 22, 1447. https://doi.org/10.1186/s12889-022-13872-9 (2022).
Zuo, Z. et al. Trends in respiratory diseases before and after the COVID-19 pandemic in China from 2010 to 2021. BMC Public Health 23, 217. https://doi.org/10.1186/s12889-023-15081-4 (2023).
Ndeh, N. T., Tesfaldet, Y. T., Budnard, J. & Chuaicharoen, P. The secondary outcome of public health measures amidst the COVID-19 pandemic in the spread of other respiratory infectious diseases in Thailand. Travel Med. Infect. Dis. 48, 102348. https://doi.org/10.1016/j.tmaid.2022.102348 (2022).
Kim, D. H., Nguyen, T. M. & Kim, J. H. Infectious respiratory diseases decreased during the COVID-19 pandemic in South Korea. Int. J. Environ. Res. Public Health 18, 6008. https://doi.org/10.3390/ijerph18116008 (2021).
Shimizu, K., Teshima, A. & Mase, H. Measles and rubella during COVID-19 pandemic: Future challenges in Japan. Int. J. Environ. Res. Public Health 18, 9. https://doi.org/10.3390/ijerph18010009 (2020).
Huang, S., Zhang, P., Hong, X. & Zhong, C. Characteristics of public health emergencies at schools in Hubei Province, 2004–2013. Chin. J. School Health 36, 113–115. https://doi.org/10.3784/j.issn.1003-9961.2014.11.017 (2015).
You, Y. et al. Scarlet fever epidemic in china caused by Streptococcus pyogenes serotype M12: Epidemiologic and molecular analysis. EBioMedicine 28, 128–135. https://doi.org/10.1016/j.ebiom.2018.01.010 (2018).
He, H. et al. The decline in immunity and circulation of pertussis among Chinese population during the COVID-19 pandemic: A cross-sectional sero-epidemiological study. Vaccine 40, 6956–6962. https://doi.org/10.1016/j.vaccine.2022.10.020 (2022).
Han, B. et al. Public awareness, individual prevention practice, and psychological effect at the beginning of the COVID-19 outbreak in China. J. Epidemiol. 30, 474–482. https://doi.org/10.2188/jea.JE20200148 (2020).
Wu, Z. et al. Impact of the COVID-19 pandemic on the detection of TB in Shanghai, China. Int. J. Tuberc. Lung Dis. 24, 1122–1124. https://doi.org/10.5588/ijtld.20.0539 (2020).
Cohen, R. et al. Pediatric Infectious Disease Group (GPIP) position paper on the immune debt of the COVID-19 pandemic in childhood, how can we fill the immunity gap? Infect. Dis. Now 51, 418–423. https://doi.org/10.1016/j.idnow.2021.05.004 (2021).
Cohen, R., Pettoello-Mantovani, M., Somekh, E. & Levy, C. European pediatric societies call for an implementation of regular vaccination programs to contrast the immunity debt associated to coronavirus disease-2019 pandemic in children. J. Pediatr. 242, 260–261. https://doi.org/10.1016/j.jpeds.2021.11.061 (2022).
Hayes, L. J. et al. Impact of the COVID-19 pandemic on the circulation of other pathogens in England. J. Med. Virol. 95, e28401. https://doi.org/10.1002/jmv.28401 (2023).
Cohen, P. R. et al. Trends in pediatric ambulatory community acquired infections before and during COVID-19 pandemic: A prospective multicentric surveillance study in France. Lancet Reg. Health Eur. 22, 100497. https://doi.org/10.1016/j.lanepe.2022.100497 (2022).
Funding
This research was supported by the National Natural Science Foundation of China (71974199) and the Liaoning Provincial Department of Science and Technology Support Fund Project for High-Quality Development of China Medical University (2023JH2/20200054).
Author information
Authors and Affiliations
Contributions
NW, PG, YR, DH, and WW contributed to the conceptualization of the study. NW, ZW, and SA played a key role in data collection and database creation. For statistical analysis, NW, ZW, SA, and WW took the lead. The initial draft of the manuscript was jointly written by NW, PG, YR, DH, and WW, with all authors engaging in subsequent editing and revisions. After thorough review, all authors gave their approval for the final manuscript. YR, DH, and WW, serving as guarantors, assume full responsibility for the study’s content, managing the work, and having the authority over data access and publication decisions.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Ethical approval
Research institutional review boards of China Medical University approved the protocol of this study and determined that the analysis of the publicly available data on Class B RIDs cases in China was part of the ongoing public health surveillance for notifiable infectious diseases, and thus was exempt from institutional review board assessment. All data were supplied anonymously and analyzed in an anonymous format, without access to personal identifying or confidential information.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Wu, N., Guan, P., An, S. et al. Assessing the impact of COVID-19 non-pharmaceutical interventions and relaxation policies on Class B respiratory infectious diseases transmission in China. Sci Rep 14, 21197 (2024). https://doi.org/10.1038/s41598-024-72165-w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-72165-w
- Springer Nature Limited