Time series data available for the monthly reported and confirmed cases of 23 class B notifiable infectious diseases in China’s mainland, from April 2017 to September 2021, were obtained from the National Health Commission of the People’s Republic of China. The data set is open to the public around the world and is reported by the Chinese Centre for Disease Control and Prevention (CDC) each month. These 23 diseases are HIV/AIDS, hepatitis (including hepatitis A virus, HAV; HBV; hepatitis C virus, HCV; and hepatitis E virus, HEV), measles, hemorrhagic fever, dengue and severe dengue, rabies, Japanese encephalitis, anthrax, Shigella spp. or Entamoeba histolytica, tuberculosis, typhoid and paratyphoid fever, pertussis, neonatal tetanus, scarlet fever, brucellosis, gonorrhea, Treponema pallidum, leptospirosis, schistosomiasis, and malaria. The data sampling rate was 1 time point per month (12 time points per year) from the monthly report of the National Health Commission of the People’s Republic of China. We used 2 criteria to select these diseases. First, the maximum number of infectious cases each month in recent years should be larger than 10. Second, the time points should be continuously publicly reported within the years of interest. We were mainly interested in how other class B infectious diseases might be influenced by policies related to COVID-19, considering that COVID-19 is also classified as class B. We did not include class A diseases due to their low incidences. Class C diseases, such as the flu, were not included, because they are less fatal and controllable and would not have the same impact as class B diseases.

Indicators of government response in China were taken from the Oxford COVID-19 Government Response Tracker [44]. In this work, we use the stringency index (all closure indicators, such as lockdown policies and travel bans, and health system policies that record public information campaigns and contact tracing), which records the strictness of lockdown-style policies. The index scores the level of government responses between 0 and 100. The higher the score is, the stricter the government interventions were (Multimedia Appendix 1).

For this study, we used public data from the National Health Commission of the People’s Republic of China. Our study did not involve any intervention on human participants. This study was approved by the Ethics Committee of Beijing Anding Hospital, Capital Medical University, China.

We defined a new concept named the ERT, which originates from the event-related potential (ERP) in neuroscience [45]. The ERT describes the direct impact of specific events on reducing the number of infectious diseases. This event could be a pharmaceutical or nonpharmaceutical intervention to prevent the spread of infectious disease. In this study, the specific event is the strong prevention and control policies implemented at the outbreak of the COVID-19 epidemic, which are an NPI. The ERT is the lowest increase of an outbreak in the period of 6 months after the outbreak of COVID-19. The trough amplitude before COVID-19 is the lowest value of the infection in the 3-year period before COVID-19 (Equation 1). The peak amplitude (Equation 2) before and after the COVID-19 outbreak is the highest value of the infection before and after COVID-19. We also calculated the trough ratio index as the ratio of troughs before and after the outbreak of COVID-19 (Equation 3).

The tuning curve of the monthly infected cases illustrates the essential profile of the outbreak of each disease in China, which gives a direct picture of the situation each month based on the historical data. We assumed that all infectious diseases included in this study have a similar trend each year for the years of observation (Multimedia Appendix 2), similar to previous studies [18]. Thus, we took the monthly average number of infected cases and computed them into a tuning curve (Equation 4). Each infectious disease in this study has a tuning curve before and after the COVID-19 outbreak, and the oscillatory pattern within a year is clear.

where N is the number of years.

Two indices of the disease were defined: preferred month and infection selectivity (Equation 5), which are important indicators of the infectious property of the epidemics caused by a disease in a year. The preferred month index is defined as the month in a year that has the most cases of infections. The infection selectivity index is defined as (1 – ratio of the minimum and the maximum number of infected cases in a year). If the selectivity index is closer to 1, it means outbreaks only occur in specific months. If the selectivity index is closer to 0, it means that outbreaks occur throughout the year.

The oscillatory property of an infectious disease is an important indicator of the regular fluctuations and recurrence of epidemics. To better quantify these fluctuations, we used spectrum analysis. Similar methods have been used in classic and modern studies in the field of infectious diseases [38-42] and some other biological research [46,47]. Spectrum analysis is a technique for decomposing complex signals into simpler signals based on Fourier transform (Equation 6). Many biological signals can be expressed as the sum of various simple signals of different frequencies and produce information about a signal at different frequencies (eg, amplitude, power, intensity, phase).

The power spectral density (PSD; Equation 7) for each infectious disease before and after the outbreak of COVID-19 was computed using the multitaper method with the Chronux toolbox [48], an open source, data analysis toolbox [49]. Power spectra of the time series data (infected cases of each disease) were calculated in 2 time periods (2017-2020 and 2020-2021).

where W_T(t) is 1 within the arbitrary period and 0 elsewhere, and T is centered about some arbitrary time t=t₀.

We performed Pearson correlation to measure the relationship of several indices (ERT, selectivity, oscillatory strength, and mean infected number) before and after the COVID-19 outbreak. Pearson correlation was also performed in the correlation analysis between trough ratio and selectivity, between the change in power and change in infected numbers, and between the change in power and change in trough amplitude. Spearman correlation was performed to measure the relationship between the trough amplitude and the peak amplitude before and after the COVID-19 outbreak. The significance (P value) of the correlation was corrected with Bonferroni correction.

We performed an independent-sample t test to compare the difference between several indices (trough amplitude, peak amplitude) before and after the COVID-19 outbreak and test the difference in the trough ratio between diseases with a different preferred time of outbreak. The pairwise t test was performed to compare the oscillatory power and the average infected number before and after the COVID-19 outbreak.

This study analyzed monthly data from April 2017 to September 2021 of confirmed cases of 23 class B notifiable infectious diseases in China’s mainland. After the COVID-19 outbreak, most class B infectious diseases had an obvious sudden trough, which we defined as the ERT (see Figure 1A for 3 typical examples). The stringency index of China showed that during the COVID-19 pandemic, the strict NPIs in China have always been existing (Multimedia Appendix 1), which allows us to analyze the long-term effect after the COVID-19 outbreak.

We showed several examples of diseases that had an obvious ERT after the COVID-19 outbreak (HCV, tuberculosis, and gonorrhea); see Figure 1A, and see Multimedia Appendix 3 for all diseases. To compare the time series of all diseases on a notionally common scale, we normalized the time series of the number of infected people by subtracting the mean number of infections before COVID-19 and dividing it by its SD. Hence, the mean number of infections before COVID-19 was 0 for all the diseases (Figure 1B). The pattern shows an obvious and sudden decrease in confirmed cases after the COVID-19 outbreak (see Figure 1B). To investigate whether the peak and trough amplitudes may change due to the outbreak, we compared the differences between peak and trough amplitudes before and after the event (policy in response to COVID-19). Results showed that the amplitude of both peak (P<.01) and trough (P<.01) significantly decreased, which indicated that the outbreak strongly moderated the oscillation amplitude (see Figure 1C). We then calculated the correlation between the normalized trough before and after the outbreak event, and we found that the trough after the outbreak was significantly correlated (r=0.88, P<.001) to the trough before but the peak was not correlated (r=–0.15, P=.48; see Figure 1D).

The ERT might be affected by the basic properties (infection selectivity and preferred outbreak time) of infectious disease outbreaks. To further clarify potential factors that would cause an ERT, we determined the property of oscillations for infectious diseases in a year by defining 2 indicators: infection selectivity and preferred outbreak time of the infectious disease. We selected 3 infectious diseases that have different selectivity as examples (Figure 2A,B; see Multimedia Appendix 4 for all diseases). The infection selectivity index is defined as (1 – ratio of the minimum and the maximum number of infected cases in a year). If the selectivity index is closer to 1, then the shape of the tuning curve is sharper (eg, Japanese encephalitis), and vice versa (eg, HEV). The preferred month index is defined as the month in a year that has the most cases of infections. Results showed that there was a significant increase in infection selectivity after the outbreak of COVID-19, and infection selectivity before and after the outbreak was positively correlated (with Bonferroni correction; Figure 2C). When we compared the selectivity before the outbreak and the trough ratio, we found that the stronger the infection selectivity, the smaller the trough ratio (Figure 2D). We also conducted a partial correlation analysis between infection selectivity and trough ratio, controlling for the preferred time of outbreak, which was significant (r=–0.58, P=.004). The association between infection selectivity and trough ratio confounded by seasons was, however, weak.

We quantified the oscillatory strength of outbreaks using power spectrum analysis (Figures 3A and 3B; see Multimedia Appendix 5 for all diseases). We then explored the relationship between infected cases and their corresponding strength of oscillatory power before and after the COVID-19 outbreak. Results indicated that the oscillatory strength (r=0.83, P<.001) and mean infected cases (r=0.95, P<.001) before the COVID-19 outbreak were significantly positively correlated to the indices after the COVID-19 outbreak (Figure 3C), showing that the stronger the oscillatory power was before the outbreak, the stronger it was after the outbreak. The same was true for mean infected cases.

To determine the differences between oscillatory power before and after the COVID-19 outbreak and between mean infected cases before and after the COVID-19 outbreak, we also split the data and compared the indices before and after the event. Consistent with our hypothesis, results showed both decreases in oscillatory power and mean infected cases after the COVID-19 outbreak (Figure 3D). We further examined the relationship between the change in power between the change in infected numbers and trough amplitude. Results showed that the change in power and the change in infected numbers was significantly correlated (r=0.92, P<.001). The more the change in oscillatory power, the more changes in the number of confirmed cases (Figure 3E). However, the change in power was not related to the change in trough amplitude (r=–0.37, P=.08 with Bonferroni correction; Figure 3F). In sum, the COVID-19 outbreak reduced the outbreaks of class B notifiable infectious diseases, as indicated by oscillatory power and mean infected cases.

In this study, we defined several novel concepts and robust metrics (ERT, selectivity of infection, preferred time to outbreak, oscillatory strength of the infectious disease) to quantify and capture the temporal characteristics of infectious disease outbreaks and event-related fluctuations in China. Our results showed that a clear ERT occurred for most class B infectious diseases after the COVID-19 outbreak under the strict public health policy. We further found that the ERT was related to the nature of diseases, such as their infection selectivity and preferred outbreak time. However, their oscillatory strength was somehow unrelated. We also compared these indices of the infectious diseases before and after the outbreak of COVID-19. The impact of the COVID-19 outbreak influenced the infectious diseases by reducing the trough amplitude, mean infected cases, and oscillatory strength but increasing infectious selectivity (see Figure 4).

To the best of our knowledge, this is the first study to systematically investigate the influence of the COVID-19 outbreak on the temporal characteristics of other class B infectious diseases in China, including both respiratory infectious diseases and other types, such as those transmitted through sex, body fluids, the digestive tract, contact, and mosquitos. The key contribution of this study is that several new concepts were purported, such as the ERT, selectivity of infectious diseases, the preferred outbreak time, and the power strength of infectious oscillation. Some previous studies have investigated basic properties of a few infectious diseases in China [18,22-25,50-52] and countries worldwide [39,40,53-55], and NPIs to mitigate COVID-19 could have affected the transmission dynamics of influenza and other respiratory diseases [17,27-29,56,57]. However, previous research did not quantify the reduction using a specific temporal index. We first defined the ERT in the field of infectious diseases to capture the immediate influence of the strong interventions related to the public health events, such as the prevention policy on disease outbreaks. The ERT could measure the temporal feature when studying the effect of some specific interventions in the future, which increases 1 dimension (temporal dynamics) compared with a simple reduction number.

Another novel finding of this study is that we also built up a connection between the ERT and some other important indicators (selective property and oscillatory property), which were neglected in prior works. In this study, we found that the ERT is related to the selectivity (Figure 2D) of an infectious disease, which gives a new understanding of how an epidemic could be more easily controlled (when a disease has high selectivity). In the future, infectious selectivity would play a more important role than before, especially when combined with the tuning curve of a disease. It would depict new pictures of the basic property of each disease and give more practical guidance on the prevention and control of epidemics. The oscillatory properties of infectious diseases were also analyzed in some previous studies [18,22-25,38-40,53-55,58-60], which could be driven by both natural [11,61,62] and human [63-68] factors. However, prior studies did not investigate the influence of COVID-19 measures or other NPIs to control epidemics on the oscillatory strength of infectious diseases systematically. Our results indicate that the oscillatory strength significantly decreased after the COVID-19 outbreak, which was accompanied by a decrease in the mean infections. This finding supports the conceptual hybrid model [18]. We also found that the oscillatory strength before the COVID-19 outbreak did not correlate to the change in trough, which further suggests that the ERT is not related to some seasonal factors but more to the measurement of the COVID-19 outbreak. The oscillatory phenomenon of population-based epidemics would be the new impetus for the study of public health. In the future, this index could be connected to more natural and human factors, which would contribute to constructing a more generic stimulated model to explain history and predict the future situation.

One limitation of our study is that the data we used are from the entire mainland China but are not specific to different provinces or cities, which may lack spatial resolution. Another limitation of our study is that the results were based on a macroscopic rather than a microscopic view of most class B infectious diseases. Further studies are needed to clarify the deeper underlying mechanisms of the COVID-19 pandemic. With these findings, we could better provide the government with recommendations on the optimal timing to intervene before achieving herd immunity, thereby helping to design fit-for-purpose policies.

In sum, the study developed a new and potentially universal approach to revealing the dynamics of infectious diseases. The transmissibility and severity of infectious diseases fluctuate regularly. The introduction of the concept of the ERT in infectious diseases can better capture the immediate influence of interventions related to previous public health events. Our results confirmed that early commencement of strong public health interventions has additional benefits on other infectious diseases.