Investigation on the possibility of dynamic COVID-Zero strategy in China: a population-based transmission model analysis and economic evaluation

Abstract

Objective

To explore the feasible and cost-effective intervention strategies to achieve the goal of dynamic COVID-Zero in China.

Design

A Susceptible–Exposed–Infectious–Recovered model combined economic evaluation was used to generate the number of infections, the time for dynamic COVID-Zero and calculate cost-effectiveness under different intervention strategies. The model simulated the 1 year spread of COVID-19 in mainland China after 100 initial infections were imported.

Interventions

According to close contact tracing degree from 80% to 100%, close contact tracing time from 2 days to 1 day, isolation time from 14 days to 7 days, scope of nucleic acid testing (NAT) from 10% to 100% and NAT frequency from weekly to every day, 720 scenarios were simulated.

Outcome measure

Cumulative number of infections (CI), social COVID-Zero duration (SCD), total cost (TC) and incremental cost-effectiveness ratio.

Results

205 of 720 scenarios could achieve the total COVID-Zero since the first case was reported. The fastest and most cost-effective strategy was Scenario 680, in which all close contacts were traced within 1 day, the isolation time was 14 days and 10% of the national population was randomly checked for NAT every day. In Scenario 680, the CI was 280 (100 initial infections) and the SCD was 13 days. The TC was ¥4126 hundred million and the cost of reducing one infection was ¥47 470. However, when the close contact tracing time was 2 days and the degree of close contact tracing was 80%–90%, the SCD would double to 24–101 days and the TCs increased by ¥16 505 to 37 134 hundred million compared with Scenario 680.

Conclusions

If all close contact was controlled within 1 day, the rapid social COVID-Zero can be achieved effectively and cost-effectively. Therefore, the future prevention and control of emerging respiratory infectious diseases can focus on enhancing the ability of close contact tracing.

Strengths and limitations of this study.

• The method of the study combined population-based transmission model analysis and economic evaluation.

• Three measures undertaken during the dynamic COVID-Zero strategy—close tracing, isolation, and nucleic acid testing (NAT)—were considered in our model and quantified the reduction in transmission resulting from these measures.

• The parameters of cost used in the model, including NAT cost, isolation cost, medical cost and gross domestic product, were collected based on the Chinese government, reflecting the general situation in China.

• The transmission model was national-level and did not discuss the differences among regions and the effect of age structure.

Introduction

With the rapid mutation of SARS-CoV-2, global containment of COVID-19 has become increasingly difficult. Since the rampant epidemic of the Omicron variant, two different countermeasures have been highlighted in the international response to COVID-19 prevention and control. The USA, Canada, Britain and other countries^1–3 have cancelled COVID-19 prevention and control measures such as wearing masks, mass nucleic acid testing (NAT) and self-quarantine to prepare for long-term coexistence with the SARS-CoV-2 variants. This is because these measures have caused huge losses to people’s spirit and life,⁴ society and economy. However, there are other countries which insist on combining COVID-19 epidemic prevention restrictions.^5–7 Among them, China proposed the dynamic COVID-Zero strategy in August 2021 to deal with the regular epidemic prevention and control of COVID-19. The dynamic COVID-Zero strategy⁸ mainly includes vaccines, therapeutics and non-pharmacological interventions (NPIs). NPIs include large-scale, high-frequency SARS-CoV-2 NAT, long periods of infection isolation, strict management of close contacts, social distancing control, travel-related measures and personal protection measures. This strategy summarises various prevention and control measures China has taken to contain the COVID-19 epidemic at a lower cost and in a shorter time.⁹

Although China successfully contained the domestic spread of the Delta variant based on the dynamic COVID-Zero strategy, the emergence of the Omicron variant led to renewed domestic COVID-19 risk. For example, the Omicron variant was first imported into Hong Kong on 27 November 2021, with a total of 330 880 confirmed cases as of 3 May, and then spread sporadically in mainland China.^{10 11} Since the first confirmed case was reported in Jilin province on 1 March 2022,¹² 40 242 people were confirmed as of 3 May.¹³ Shanghai reported its first COVID-19 infection on 10 March 2022. By 3 May, 59 605 people¹⁴ were confirmed and the cumulative number of infections (CI) had exceeded 600 000.^{15 16} Several studies had shown that eliminating a respiratory virus such as SARS-CoV-2 was unrealistic,¹⁷ and it would take even 3 months to reach the COVID-Zero target in areas with high infection rates.¹⁸ Under such circumstances, many people questioned that China’s dynamic COVID-Zero strategy was unsustainable, stating that China should be prepared to live with COVID-19.¹⁹

Existing researches mainly focus on the review of dynamic COVID-Zero strategy in China, and the economic evaluation combined with dynamic model is still rare. The model researches mainly simulated the epidemic trend after the implementation of mitigation measures in China,²⁰ or predicted the daily number of new infections and the turning point of the epidemic in a certain region under the dynamic COVID-Zero strategy.²¹ For cost-effective analysis of NPIs, studies have focused on implementing a single NPI at the national or provincial level or considering multiple NPIs in a community.²²

The dynamic COVID-Zero strategy was not specific to certain towns and townships, but a general policy across the country. Therefore, this study adopted a population-based model to explore the feasibility and cost-effectiveness of the dynamic COVID-Zero strategy under different combinations of COVID-19 prevention and control strategy strengths at the national level.

Methods

Research scenario

In this study, we included measures for close tracing, isolation and NAT. The detailed information of these single measures is shown in table 1. Combining all these measures in the actual situation, 720 joint scenarios were developed, which are listed in online supplemental table 1. To assess the effectiveness of single or joint NPIs, we set different comparison scenarios. The comparison scenario of the joint scenarios was Scenario 481. In Scenario 481, the degree of close contact tracing was medium, the close contacts were traced within 2 days, the isolation time was 14 days and 10% of the national population was randomly checked for NAT every 7 days. When discussing the effectiveness of a single NPI, under the premise of controlling other NPIs unchanged, the comparison scenarios of a single NPI were middle degree of close contact tracing, tracking close contacts with 2 days, the 7 day isolation time, doing NATs among 10% of the national population and doing NATs every 7 days, respectively.

Table 1.

COVID-19 surveillance scenarios

Non-pharmaceutical interventions	Classification	Explanation
Close contact tracing degree	Low, middle and high	A low, middle and high level of close contact management indicates that 80%, 90% and 100% of close contacts were managed.
Close contact tracing time (day)	2, 1.75, 1.5, 1.25 and 1	It refers to the time that close contacts are tracked and managed.
Isolation time (day)	14, 10 and 7	Confirmed cases and close contacts should be isolated or quarantined for 7, 10 or 14 days in specific locations.
NAT scope	0.1, 0.2, 0.3, 0.5, 0.75 and 1*	It refers to the nationwide for NAT sampling proportion.
NAT frequency (day)	7 days, 5 days, Once in 3 days, Once a day	It refers to the time interval between NATs.

*When the close contact tracing degree is low, the NAT scope is 0.5, 0.75 and 1; when the close contact tracing degree is middle, the NAT scope is 0.1, 0.2, 0.3, 0.5, 0.75 and 1; when the close contact tracing degree is high, the NAT scope is 0.1, 0.2 and 0.3.

NAT, nucleic acid testing.

Model

A population-based transmission dynamics model was used to generate the number of infections in China under different scenarios. The population was divided into three groups, namely general population, close contact population and quarantined population. In each group, let S, E, I, H and D denote the state to be susceptible, exposed, infected, in hospital and dead, respectively. Infections (I) could be infectious (figure 1 and online supplemental figure 1). The model simulated the 1-year spread of COVID-19 in mainland China after 100 initial infections were imported. The detailed description and parameters of the model are shown in online supplemental tables 2 and 3. The asymptomatic infection ratio ($\rho$) and fatality rate ($r_f$) were derived from the current COVID-19 vaccination situation in China, which meant that the current vaccination was considered.

Figure 1.

Dynamic model transfer chart.

Cost

The economic data were derived from fieldwork and previous literature (online supplemental table 4). The NAT cost (NC) included sample collecting, processing, and testing; information registration; treatment of medical waste; receiving clinical consultation; and personnel salary.²³ The isolation cost were obtained by multiplying the daily isolation cost (DIC) per capita^{24 25} by cumulative under control people (CUP). And the medical cost was the average medical cost (AMC)²⁶ multiplied by CI. The direct costs included quarantine or isolation costs, the cost of NAT and the cost of medical treatment. The indirect costs contained socioeconomic losses due to labour shortages and were obtained by multiplying the daily gross domestic product (GDP) per capita by CUP by isolation time.^{27 28} It contained isolated infections and quarantined close contacts. As the study period was 1 year, the discount rate was not considered.

Outcome indicators

1. CI: the CI during the monitoring.

2. Cumulative number of deaths (CD): the cumulative number of agents of state D during the health monitoring.

3. Social COVID-Zero duration (SCD): the duration was from ‘the first infection report’ to ‘no infection in society report’, which meant that all new confirmed cases or asymptomatic infections were found in isolation and control places, and there would be no positive infections in society.

4. Total COVID-Zero duration (TCD): the duration was from ‘the first infection report’ to ‘no infection report’, which meant that there were no new confirmed cases or asymptomatic infections in society.

5. Total cost (TC): it included direct costs and indirect costs.

   $${\displaystyle \mathrm{T}\mathrm{C}=\mathrm{T}{\mathrm{C}}_{\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{s}}+\mathrm{T}{\mathrm{C}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{s}}}$$
   $${\displaystyle \mathrm{T}{\mathrm{C}}_{\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{s}}=\mathrm{T}{\mathrm{C}}_{\mathrm{N}\mathrm{A}\mathrm{T}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}}+\mathrm{T}{\mathrm{C}}_{\mathrm{i}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}}+\mathrm{T}{\mathrm{C}}_{\mathrm{m}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{t}}}$$

6. Incremental cost-effectiveness ratio (ICER): compared with comparators, it was the ratio of the variation of TC to the reduction of CI.²⁹ Therefore, when the TC and the CI of the n^th scenario decreased, the larger the ICER, the more cost-effective the n^th scenario was.

   $${\displaystyle \mathrm{I}\mathrm{C}\mathrm{E}\mathrm{R}=\frac{\mathrm{\Delta }\mathrm{T}\mathrm{C}}{\mathrm{\Delta }\mathrm{C}\mathrm{I}}=\frac{\left(\mathrm{T}{\mathrm{C}}_{\mathrm{n}}-\mathrm{T}{\mathrm{C}}_{\mathrm{c}}\right)}{\left(\mathrm{C}{\mathrm{I}}_{\mathrm{n}}-\mathrm{C}{\mathrm{I}}_{\mathrm{c}}\right)}}$$

   (where n represented the n^th scenario and c represented the comparators.)

Statistical analysis

This study implemented the simulation of the dynamic model, the sensitivity analysis and subsequent data processing in MATLAB R2021a (Mathworks, Natick, Massachusetts, USA) and Excel 2010 (Microsoft Corporation, Redmond, Washington, USA). At the same time, for the optimisation strategy, the ${\displaystyle {\mathrm{R}}_0}$ package in R V.4.2.2 (GNU project, Auckland, New Zealand) was used to calculate time-dependent regeneration number (R_t).³⁰ Considering the uncertainty of different SARS-CoV-2 mutants and economic costs, univariate sensitivity analyses were performed to explore the impact of the perturbations on basic reproductive number (${\displaystyle {\mathrm{R}}_0}$), infectious period (IP), daily contacts per person (c), median incubation period (MIP), NAT detection time, asymptomatic infection ratio ($\mathrm{\rho }$), NC, DIC, AMC and per capita GDP to test the robustness of the optimal scenario. The value range of parameters is shown in online supplemental table 2.

Patient and public involvement

There was no patient or public involvement in this research.

Results

Overall trajectory

Overall, 720 scenarios were simulated based on frequency and scope of NAT, degree and time of close contact tracing and duration of isolation. The CI was 274–1 412 600 000 and the CD was 2–11 300 800. The TC was ¥590–807 417 hundred million. The ICER was between ¥−2760.73 and 1438.63 million with the comparison Scenario 481, as shown in online supplemental table 5 and figure 2.

Among them, 205 scenarios could reach the state of total COVID-Zero (no new infections per day), with the TCD ranging from 17 to 364 days and the SCD ranging from 13 to 217 days. In these 205 scenarios, the CI ranged from 274 to 5471 and the CD was 2–44. The TC ranged from ¥590 to 41 260 hundred million and the ICER was between ¥39.56 and 47.72 thousand compared with the Scenario 481, as shown in online supplemental tables 6 and 7. There were 12 scenarios for social COVID-Zero within 2 weeks, 59 scenarios for social COVID-Zero within 3 weeks and 103 scenarios for social COVID-Zero within a month, as shown in online supplemental table 8.

Health effectiveness

Based on the actual data in mainland China, the 59 scenarios could reach the state of social COVID-Zero within 3 weeks. Under the 59 scenarios, the CI and CD declined with fluctuation as NPIs became more stringent (figure 2A,B). The CI ranged from 274 to 992, and the CD was between 2 and 8. Meanwhile, the CI in Scenarios 300, 360, 680, 700 and 720 was at low levels, with CI values of 292, 287, 280, 277 and 274, respectively. Across the 59 scenarios, SCD ranged from 13 to 21 days, with the TCD ranging from 17 to 221 days. It was worth noting that Scenarios 620, 640, 660, 675, 679, 680, 695, 699, 700, 715, 719 and 720 could reach the state of total COVID-Zero within 14 days, as shown in table 2.

Figure 2.

Distribution of different outcome indicators in 59 scenarios.

Table 2.

COVID-19 transmission and cost-effectiveness of 59 scenarios

Scenario	COVID-19 transmission				Cost-effectiveness
	Cumulative number of infections	Cumulative number of deaths	Social COVID-Zero duration (day)	Total COVID-Zero duration (day)	Total cost (hundred million)	ICER (10 000 per person)
481*	1 412 600 000	11 300 800	365	–	674 796	–
120	337	3	20	22	41 260	3.9563
179	372	3	21	23	41 260	3.9563
180	329	3	19	21	41 260	3.9563
239	345	3	21	23	41 260	3.9563
240	308	2	18	20	41 260	3.9563
280	487	4	19	69	30 945	4.0293
298	362	3	20	22	41 260	3.9563
299	325	3	18	20	41 260	3.9563
300	292	2	16	18	41 260	3.9563
340	425	3	17	54	30 945	4.0293
355	472	4	20	25	20 630	4.1023
357	392	3	21	23	41 260	3.9563
358	353	3	19	21	41 260	3.9563
359	318	3	17	18	41 260	3.9563
360	287	2	15	17	41 260	3.9563
580	992	8	16	221	8253	4.7175
600	697	6	16	137	12 378	4.6883
610	884	7	20	75	826	4.7700
614	635	5	20	69	2064	4.7613
615	522	4	15	57	2063	4.7613
619	411	3	16	53	4126	4.7467
620	346	3	13	42	4126	4.7467
630	851	7	20	70	1651	4.7642
634	602	5	20	62	4126	4.7467
635	501	4	15	52	4126	4.7467
638	496	4	21	63	8252	4.7175
639	390	3	16	47	8252	4.7175
640	333	3	13	38	8252	4.7175
650	821	7	20	65	2476	4.7584
654	573	5	19	56	6189	4.7321
655	482	4	15	47	6189	4.7321
658	458	4	20	55	12 378	4.6883
659	371	3	15	41	12 378	4.6883
660	321	3	13	33	12 378	4.6883
665	786	6	19	44	590	4.7717
670	666	5	17	39	826	4.7700
673	585	5	21	42	2064	4.7613
674	472	4	16	32	2063	4.7613
675	407	3	14	27	2063	4.7613
678	382	3	17	31	4126	4.7467
679	317	3	14	23	4126	4.7467
680	280	2	13	18	4126	4.7467
685	779	6	19	42	1180	4.7675
690	659	5	17	37	1651	4.7642
693	571	5	21	39	4127	4.7467
694	465	4	16	30	4126	4.7467
695	403	3	14	25	4126	4.7467
698	371	3	17	28	8252	4.7175
699	313	3	14	21	8252	4.7175
700	277	2	13	18	8252	4.7175
705	772	6	19	41	1769	4.7634
710	653	5	17	36	2476	4.7584
713	558	4	20	36	6190	4.7321
714	459	4	16	28	6189	4.7321
715	398	3	14	23	6189	4.7321
717	453	4	21	37	12 378	4.6883
718	361	3	17	26	12 378	4.6883
719	308	2	14	20	12 378	4.6883
720	274	2	13	17	12 378	4.6883

*These comparisons were Scenario 481.

There were three low degree close contact tracing scenarios, 12 middle degree close contact tracing scenarios and 44 high degree close contact tracing scenarios. The CI for the middle degree scenarios of close contact management ranged from 287 to 487, and the CI for the low degree of close contact tracing scenarios was between 329 and 372, as shown in table 2. The CI decreased significantly in the high degree of close contact tracing scenarios compared with the middle degree of close contact tracing scenarios (online supplemental figure 3A). The trend was the same for close contact tracing time reduced to 1 day compared with 1.25 days (online supplemental figure 3B).

The expanded scope of NAT could effectively reduce the CI and TCD and make up for the lack of close contact tracing. For example, when the degree of close contact tracing was 100% and NAT was conducted every 2 days, as the NAT scope was expanded from 10% to 30%, the CI ranged from 398 to 573, 403 to 602 and 407 to 635, respectively, and the TCD ranged from 23 to 56, 25 to 62 and 27 to 69, respectively. For the middle degree of close contact tracing, the Scenario 360 with 100% NAT scope, resulted in 287 people infected, and achieved total dynamic COVID-Zero in 17 days, indicating that 32.47% of infections were avoided and 68.53% of TCD was shortened compared with the Scenario 340 with 75% NAT scope, as shown in table 2.

Cost-effectiveness

Cost-effectiveness analysis was used in this study to evaluate the feasibility of different-strength NPIs scenarios. The TC decreased significantly as the degree of close contact tracing increased and the TC increased with the increasing NAT scope (figure 2E). The TC ranged from ¥590 to 41 260 hundred million among the 59 scenarios.

When considering the effects of joint NPIs among the 59 scenarios, compared with Scenario 481, the ICERs were all greater than zero and ranged from ¥39.56 to 47.72 thousand per person, indicating that all 59 scenarios could reduce infections and costs and were cost-effective (table 2). ICER was the greatest in Scenario 355 among the low or middle degree of close contact tracing scenarios. And ICERs were the same in the low degree of close contact tracing scenarios.

When only considering the close contact management degree, taking the middle close contact tracing degree scenarios as the control, the ICERs of scenarios with high degree close contact tracing were all greater than zero and the range was ¥0.0426–0.1125 million, as shown in online supplemental table 9. Simultaneously taking the scenarios of managing all close contacts within 2 days as a control, the ICERs were all greater than zero, and the range was ¥0.0674–1263.6888 million. And taking the scenarios of the 7-day isolation time as a control, the ICERs ranged from ¥0.0543 to 713.6364 ten million.

Optional strategy

Scenario 680 was the fastest and most cost-effective scenario to achieve dynamic social COVID-Zero, in which the degree of close contact tracing was 100%, all the close contacts were traced within 1 day, the isolation time was 14 days, and 10% of the national population was randomly checked for NAT every day. In Scenario 680, the CI was 280 and fell below the lower quartile of CI for 205 scenarios (online supplemental table 6). The R_t of Scenario 680 was significantly lower than that of the comparison scenario (online supplemental figure 4). Scenario 680 could achieve the goal of dynamic social COVID-Zero within 14 days, with a TCD of 18 days. The cost of reducing an infection was ¥47 466.76, which cost 4126 hundred million in total. However, in practice, the ability to trace close contacts was affected by various factors.

When the ability of close contact tracing was 80%–90%, Scenario 360 was the fastest to achieve dynamic social COVID-Zero, in which all the close contacts were traced within 1 day, the isolation time was 14 days and the whole population was checked for NAT every day. Scenario 355, which carried out NATs every 2 days, was the most cost-effective. Scenario 355’s ICER was ¥41 022.86 and was the largest of scenarios with low or middle close contact tracing degrees (table 2). When the degree of close contact tracing was 80%, NAT must be conducted on all staff and the frequency must be higher than once every 2 days to achieve total COVID-Zero (table 3).

Table 3.

The relationship between the degree of close contact tracing and NAT

Close contact tracing degree	Close contact tracing time (day)	Isolation time (day)	NAT scope	NAT frequency (day)
0.8	1.5	7	1	2
0.8	1.75	10	1	2
0.8	1.75	14	1	2
0.8	2	14	1	1
0.9	1.25	7	0.75	1
0.9	1.25	7	1	5
0.9	2	7	1	2
0.9	2	7	1	1
0.9	1.25	10	0.75	2
0.9	1.75	10	0.75	1
0.9	1.25	10	1	7
0.9	1.5	10	1	5
0.9	2	10	1	2
0.9	2	10	1	1
0.9	1	14	0.5	1
0.9	1	14	0.75	7
0.9	1.5	14	0.75	2
0.9	1.75	14	0.75	1
0.9	1.25	14	1	7
0.9	1.5	14	1	5
0.9	2	14	1	2
1	1	7	0.2	2
1	1.25	7	0.3	2
1	1.75	10	0.1	7
1	1.75	10	0.2	7
1	2	10	0.2	1
1	1.75	10	0.3	7
1	2	10	0.3	2
1	2	14	0.1	7

NAT, nucleic acid testing.

When the degree of close contact tracing was 80%–90% and close contact tracing time was 2 days, Scenario 356 was the first to achieve social COVID-Zero with 14-day isolation time and NAT every day for whole population. Scenarios 231, 291 and 351 were the most cost-effective, in which the whole population was checked for NAT every 2 days. The values of CI and SCD in Scenario 356 were 436 and 24, respectively. ICERs in Scenarios 231, 291 and 351 were ¥41 023 and were the largest of scenarios with low or middle close contact tracing degrees (online supplemental table 7).

Sensitivity analysis

Sensitivity analysis showed that the optimal scheme had no fluctuation with parameter change. The value range of each parameter in sensitivity analysis was ranging from 0.75 to 1.25 times of original parameters. ${\displaystyle {\mathrm{R}}_0}$ and IP had a greater impact on the CI, ranging from 230 to 334 and from 353 to 239, respectively. MIP had a greater impact on the SCD which ranged from 10 to 15 days and TCD from 13 to 25 days. And ${\displaystyle {\mathrm{R}}_0}$, IP and $\mathrm{\rho }$ had the greatest impact on ICER, ranging from 6.0522 to 2.9027, 4.1201–3.4625 and 5.0360–3.0136, respectively. More specific results are found in online supplemental table 10.

Discussion

We simulated 720 scenarios, and 205 of 720 scenarios could reach the state of total COVID-Zero within a year, and 59 of 720 scenarios could reach social COVID-Zero in 3 weeks since the first case was reported. Scenario 680, in which the close contact tracing degree was high, the close contacts were tracked within 1 day, the isolation time was 14 days and 10% of the national population was randomly checked for NAT every day, was the first to achieve dynamic social COVID-Zero and most cost-effective COVID-19 prevention and control strategy. It enabled timely detection of infections, controlled the development of the epidemic and implemented the dynamic COVID-Zero strategy. Like the Omicron variant pandemic in Beijing that started on 22 April 2022, the state of social COVID-Zero was reached on 27 May after rapid management of close contacts and high-frequency NAT.³¹

Controlling all close contacts within 24 hours was the priority of COVID-19 prevention and control. Beijing adhered to the ‘24-hour disposal rule’, immediate epidemiological investigation, tight control, fast detection within the specified time to complete the epidemiological investigation, close contact determination, location, transport, isolation and other measures.³² In the scenarios that achieved total COVID-Zero within one year, 58.08% (119/205) scenarios were 100% degree of close contact tracing, and 26.83% (55/205) scenarios were close contact tracing within 24 hours. Affected by the spillover from Hong Kong, Shenzhen was hit by a round of Omicron in early March. The city achieved the state of dynamic social COVID-Zero within 1 week by taking measures such as lockdown, rapid epidemiological investigation and high-frequency NAT. The role of epidemiological fast-tracking should not be ignored, and each confirmed case should be followed up in groups to identify contacts.³³

The results of this study indicated that it was necessary to carry out the high-frequency NAT. NAT sampling circles with 15 min walks have been established in provincial capitals and 10 million population cities, which provided conditions for large-scale high-frequency NAT in a short time.³⁴ The 2022 Beijing Winter Olympics applying China’s governance models was successfully held.³⁵ At the airport entry with the 1.76% positive rate, the NAT closed-loop positive rate was only 0.007% due to NAT once a day.³⁶ Under regular prevention and control measures, routine NAT screening-detected many of China’s first COVID-19 cases. For example, it was necessary to implement regular high-frequency NAT in the first case in Shenzhen on 18 January and the first case in Qingdao on 11 March.^{37 38}

Dynamic COVID-Zero strategy needs the support of the government, the publicity of media and the cooperation of the public, which requires outstanding organisational power and appeal.³⁹ When the degree of close contact management could not reach 100%, the goal of social COVID-Zero could be achieved by expanding the scope of NAT, such as Scenarios 180, 340, 357–360. For instance, the Chinese Government stressed that local governments should determine and dynamically adjust testing strategies and population coverage based on the needs of epidemic prevention and control and testing capacity, and clarified the capacity requirements for expanding the detection range.⁴⁰ Besides, in Liaoning province, Xi'an city, the local government conducted NATs for all people in the areas where the cases occurred, effectively fulfilled the goal of dynamic COVID-Zero.^{41 42} Although in the above scenario the CI was relatively small and social COVID-Zero could be achieved quickly, the TC was higher and ICER was smaller, indicating that the cost effect was poor compared with Scenario 680.

This study has the following limitations. First, due to the continuous emergence of SARS-CoV-2 variants such as Omicron, initial infection rate, IP, basic reproduction number and asymptomatic infection ratio might change, and even beyond the sensitivity analysis results. Second, most of non-sanitary side effects of the ongoing COVID-19 pandemic, especially the socioeconomic costs remain hidden for now. The direct costs including NAT, isolation and medical costs and indirect cost impact vary from different regions and age groups. Third, many of Omicron’s deaths in China were among the elderly, and the model did not consider the effect of age structure, which could affect the estimation of CD. In the future, more exploration could focus on different variants, COVID-19-related costs and building an aged-based model.

Conclusions

This was a pioneering study to evaluate the dynamic COVID-Zero strategy. If all close contact was controlled within 1 day, the goal of rapid social COVID-Zero can be achieved effectively and cost-effectively. The future prevention and control of emerging respiratory infectious diseases can focus on enhancing the ability of close contact tracing, increasing the speed of close contact tracing, expanding the scope of NAT and increasing the frequency of NAT. Reallocation of resources to cost-effective strategies that increased the frequency of NAT and reduced the time required for close tracing would increase health effects and save costs.

Data availability statement

All data relevant to the study are included in the article or uploaded as an online supplemental information.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.