Characterization of PTB incidence before and after the COVID-19 outbreak
Before the COVID-19 outbreak, a stable and periodical cycle of annual PTB incidence was observed from January 2005 to December 2019 (Figure 2A). The monthly incidence fluctuated from 5 to 13 per 100,000 every year. The highest incidence usually occurred in March or April and then it gradually decreased to the lowest value in February of the following year (Figure 2A). Therefore, a dramatic rise in PTB incidence from February to March was observed in almost every year (Figure 2A). The average increase in PTB incidence from February to March was 2.25, wherein the highest value was 5.87 in 2005 and the lowest value was 0.97 in 2009.Another slight increase in PTB incidence was usually observed from October to December each year. However, during the COVID-19 outbreak, PTB incidence sharply decreased when compared with the data from before the COVID-19 outbreak. As indicated in Figure 2, the average monthly PTB incidence from the previous 6 years (from 2005 to 2019) were 7.10, 6.89, 9.58, 9.41, 8.96, 8.71, 8.49, 8.25, 7.83, 7.22, 7.45, and 7.52, respectively, whereas these numbers in 2020 decreased to 4.83, 3.21, 5.25, 6.12, 5.96, 6.07, 5.94, 5.46, 5.39, 4.85, 4.97, and 4.58, respectively. Similarly, the monthly PTB incidence in February was the lowest in 2020, although there was an overall decline.
Predicted PTB incidence in 2020 with the non-interventional model
The remarkable overall decline in PTB incidence since the COVID-19 outbreak in China was influenced by the drastic intervention measures enacted to contain the novel coronavirus. The exact inference of intervention strategies conducted during the COVID-19 outbreak was quantified by constructing the ARIMA non-interventional model to predict the reference PTB incidence in 2020. As shown in Figure 2B, the incidence estimated with this prediction model was significantly higher than the actual observed incidence in each month. The predicted monthly PTB incidence from January 2020 to December 2020 was 5.65, 4.86, 6.82, 6.53, 6.39, 6.17, 6.05, 5.79, 5.45, 4.98, 5.13, and 5.02 respectively, whereas the actual observed value during these months was 4.83, 3.21, 5.25, 6.12, 5.96, 6.07, 5.94, 5.46, 5.39, 4.85, 4.97, and 4.58, respectively (Figure 2A and Table 1). The difference between the predicted PTB incidence and the observed value ranged from 0.06 (September 2020) to 1.65 (February 2020) per 100,000.
Table 1 Predicted monthly PTB incidence in 2020 and 2021 under the presence or absence of a persistent intervention
|
No intervention
(1/100,000)
|
Persistent strict intervention
(1/100,000)
|
Persistent normalized intervention
(1/100,000)
|
Month
|
2020
|
2021
|
2020
|
2021
|
2021
|
1
|
5.650
|
5.261
|
|
4.127
|
4.707
|
2
|
4.857
|
4.525
|
|
3.093
|
3.649
|
3
|
6.825
|
6.477
|
|
5.073
|
5.635
|
4
|
6.534
|
6.195
|
|
5.212
|
5.770
|
5
|
6.394
|
6.052
|
5.475
|
4.885
|
5.617
|
6
|
6.169
|
5.833
|
5.370
|
4.707
|
5.515
|
7
|
6.054
|
5.715
|
5.224
|
4.578
|
5.393
|
8
|
5.794
|
5.450
|
4.957
|
4.313
|
5.053
|
9
|
5.449
|
5.106
|
4.615
|
3.970
|
4.807
|
10
|
4.983
|
4.641
|
4.149
|
3.504
|
4.313
|
11
|
5.127
|
4.781
|
4.285
|
3.643
|
4.450
|
12
|
5.023
|
4.677
|
4.183
|
3.540
|
4.243
|
Notes: No intervention indicates the prediction model under no intervention. Persistent intervention is defined as the prediction model under a persistent intervention includes strict and normalized interventions.
PTB incidence from 2020 to 2021 predicted with the interventional model under strict intervention conditions
The significant decline in PTB incidence observed since the implementation of the strict intervention during the COVID-19 outbreak from January 2020 to April 2020 in China highlighted the importance of strict social interventions in preventing PTB. Therefore, the ARIMA interventional model was constructed with the incidence data from January 2005 to April 2020 to speculate the future tendency of PTB incidence under strict interventions. As illustrated in Figure 3A, the incidence predicted for the next 2 years would stay at a relatively low level when the strict interventional measures are maintained. The monthly incidence would be reduced by 1.03 per 100,000 in every month compared with those without intervention before the COVID-19 outbreak (Figure 3A and Table 1). The mean incidence from 2020 to 2021 predicted with this interventional model would decline to 4.51 per 100,000.
PTB incidence from 2020 to 2021 predicted with the interventional model under normalized intervention conditions
Since the COVID-19 epidemic had gone into remission in April 2020 in China, prevention and control strategies had also been adjusted as normal state, which continues to this day. These normalized intervention strategies were tested by comparing the predicted PTB incidence under different intervention strategies. As illustrated in Figure 3A, the incidence predicted for the next year would stay at a relatively low level when the normalized interventional measures are maintained. The annual incidence estimated 59.15 under normalized intervention per 100,000 in 2021, and the value would decline to 50.65 with strict interventions. The monthly incidence predicted with the intervention model under normalized intervention conditions would be generally larger than that under strict intervention conditions. Moreover, the difference among the annual incidence in 2021 of the two groups was statistically significant. The preventive effect on PTB incidence would clearly be better under a strict intervention strategy. In theory, a strict intervention is difficult to implement continuously. Thus, normalized intervention strategies are more conducive to promotion and implementation, which would represent a huge achievement for PTB prevention.
Model prediction effect evaluation
Until December 31, 2021, China has continued to implement normalized interventions for COVID-19. So we compared the predicted values and actual incidence which gained in 2021 to validate the accuracy of intervention model under normalized intervention conditions. Table 2 showed that the predicted values in the monthly incidence of Chinese PTB is generally within the 95% CI, indicating that the model has good prediction performance.
Table 2 Predicted and actual monthly PTB incidence in China from January to December 2021
Month
|
Predicted value(95% CI)
|
Actual value
|
1
|
4.707 (3.677, 5.738)
|
4.590
|
2
|
3.649 (2.613, 4.684)
|
3.925
|
3
|
5.635 (4.568, 6.702)
|
5.722
|
4
|
5.770 (4.679, 6.861)
|
5.704
|
5
|
5.617 (4.501, 6.733)
|
5.328
|
6
|
5.515 (4.375, 6.656)
|
5.232
|
7
|
5.393 (4.229, 6.557)
|
5.428
|
8
|
5.053 (3.865, 6.240)
|
4.813
|
9
|
4.807 (3.596, 6.017)
|
4.802
|
10
|
4.313 (3.080, 5.546)
|
4.347
|
11
|
4.450 (3.195, 5.705)
|
4.373
|
12
|
4.243 (2.966, 5.520)
|
4.376
|