Contact tracing, a key element of nonpharmaceutical interventions (NPIs), has been demonstrated as an effective approach in reducing the spread of emerging infectious diseases (EIDs), such as COVID-19 in the early phase of the global pandemic in 2020, when effective vaccines and antiviral therapies were not yet available [1,2]. Individuals who are more likely to contract SARS-CoV-2 can be identified through contact tracing. Supported by this knowledge, measures such as quarantine, isolation, symptom monitoring, and sanitary practices can be applied to forestall a community-acquired epidemic [3-13]. However, acquiring the information required for precise NPIs through traditional contact tracing is a time-consuming and labor-intensive technique that may impede the timely implementation of containment measures.

As an efficient alternative, digital contact tracing uses big data analytics to locate all potential contacts by integrating accessible information. With this detailed knowledge, the two factors, contact rate and transmission probability, embedded in the effective reproductive number can be minimized to forestall the outbreak of an EID such as COVID-19 in a community. Since the beginning of the COVID-19 pandemic, digital contact tracing has been proposed and implemented in a variety of nations and regions [5,8,10,11,13-23]. The implementation of contact tracing in the context of containment measures may vary from time to time, not only due to the transmissibility and severity of dominant COVID-19 variants of concern (VOCs) but also the availability of vaccines and antiviral therapies [24,25]. Nevertheless, there have been few attempts to examine the details of implementations of digital contact tracing based on empirical cases of efforts to contain COVID-19 outbreaks [14,16,18,19,22].

Taiwan implemented digital contact tracing based on multiple information sources during the early phases of the COVID-19 pandemic. Before the occurrence of community-acquired clusters, digital methods were used to trace potential contacts from the Diamond Princess cruise ship [5]. These digital techniques included sensor data, such as GPS information from shuttle buses, highway electronic toll collection systems, credit card transaction logs, closed-circuit television records, and mobile position data, for tracing and providing warning messages to potential contacts [5]. Supported by the Taiwan Communicable Disease Control Act [26], consent for the retrieval of individual information related to the containment of EID outbreaks under government auspices was waived. Taiwan’s experience in containing the severe acute respiratory syndrome outbreak in 2003 and the early adoption of digital contact tracing during the COVID-19 pandemic, as justified by the Taiwan Communicable Disease Control Act, have established a strong foundation for the mandatory provision of individual data to the government with the aim of containing EIDs and preserving collective benefits. This acceptance within the Taiwanese population also provides a solid basis for the evolution of digital contact tracing.

Since the initial identification of community-acquired transmission of COVID-19 in January 2020 in Changhua, Taiwan, a series of contact tracing measures have been implemented to mitigate the spread of the virus in the community. To ensure the efficient deployment of NPIs, the Changhua Health Administration (Changhua County Public Health Bureau) has adopted digitally transformed approaches since early February 2020. This study aimed to demonstrate the design and implementation of digital contact tracing. The evolution of this digital cluster investigation was further illustrated by 3 empirical SARS-CoV-2 infection clusters in Changhua. The effectiveness in containing a community-acquired outbreak by using digital contact tracing was evaluated by using effective reproductive number as an indicator.

On January 20, 2020, the first domestic case, which was also the first community-acquired transmission of COVID-19 in Taiwan, occurred in Changhua. This COVID-19 cluster involved an index case identified through symptom-based surveillance for COVID-19 and a household member of this index case [27]. Following this household transmission, the Changhua County Public Health Bureau switched from traditional contact tracing to a digital strategy to effectively contain the spread from household to community. The framework and the implementation of this digital transformation are described in the following sections.

To assess the effectiveness of digital contact tracing in containing the transmission of COVID-19, a comparison between the effective reproduction number (Rt) and the basic reproduction number (R0) was performed. R0 is determined by 3 factors: contact rate, transmission probability for each contact, and duration of infection, as follows:

R0 = contact per day × transmission probability of each contact × infectious period in days (1)

In a population that is vulnerable to SARS-CoV-2 infection, the value of R0 represents the average number of cases infected for each symptomatic case. Effective interventions can reduce each R0 factor, leading to a lower Rt value. An Rt value <1 suggests that the disease will eventually vanish without causing a large-scale community-acquired outbreak.

The study was approved by the Taipei Medical University joint institutional review board (N202007018). The provision of individual information, including health data, TOCC, and contact history was mandatory during the outbreak period as justified by the Taiwan Communicable Disease Control Act [26]. According to the act, consent for the retrieval of individual information related to the containment of EID outbreaks can be waived under government auspices.

Early in January 2020, the Changhua County Public Health Bureau identified a man with no previous travel history as a contact of a COVID-19 case to assess the risk of SARS-CoV-2 infection (case 2; Figure S2A in Multimedia Appendix 1). Contact tracing was initiated following the detection of an imported COVID-19 case with a travel history to Wuhan, China, who arrived in Taiwan on January 21, 2020. Figure S2A in Multimedia Appendix 1 depicts the timeline of this clustered event. When the individual (referred to as case 1 in Figure S2A in Multimedia Appendix 1) arrived in Taiwan, she was residing with a family member while under home quarantine in Changhua. Owing to fever and respiratory symptoms, on January 25, 2020, case 1 sought medical advice at Changhua Christian Hospital. Case 1 was subsequently isolated and received treatment in a negative pressure ward at Changhua Christian Hospital owing to a positive TOCC (as an international traveler from Wuhan, China; a hotspot in early 2020) history and being on the fourth day of quarantine. A positive RT-PCR result for SARS-CoV-2 was confirmed for case 1 on January 27, 2020 [27]. A household contact (case 2) was deemed to be at extremely high risk due to prolonged close contact with case 1. Despite only exhibiting very mild nonspecific respiratory symptoms, case 2 tested positive for COVID-19 on January 28, 2020, becoming the first case of household-acquired transmission in Taiwan [27].

The conventional contact tracing approach revealed that between January 21, 2020, and January 25, 2020, case 1 visited various places, including a traditional market, an electrical appliance shop, a supermarket, and a convenience store within the residential community. It is important to note that case 1 consistently wore a mask during these visits. In addition, case 2, a household contact of case 1, had visited multiple hardware stores and a department store before being confirmed as a COVID-19 case. A total of 36 contacts were identified through comparison of the contact tracing information for these 2 cases. All of these contacts were placed under quarantine and subsequently tested for COVID-19 using RT-PCR testing. Over a 14-day quarantine period, they were also monitored for COVID-19 symptoms. All contacts tested negative for COVID-19. Furthermore, thorough sanitation measures were implemented in the exposed environments and shops. These specific locations were then communicated to the community residents through mass media channels to ensure that individuals who were potentially exposed practiced self–health management.

After this household transmission of COVID-19, the Changhua County Public Health Bureau collaborated with the public health, civil affairs, and police departments to create a contact tracing system to contain the outbreak. However, owing to the extensive and multisource information required for effective contact tracing, the process is time consuming and challenging, despite the integrated efforts across sectors. Using the first COVID-19 cluster as an example, to assess the risk of disease transmission and contain an outbreak, the contact tracing process must identify potential contacts across various settings, including households and public gatherings such as supermarkets and social events. Drawing insights from the contact tracing efforts during the first community-acquired COVID-19 transmission in Changhua, a process aided by IT and the integration of big data from 3 sectors has been designed to enhance the efficiency of information gathering.

The third COVID-19 case in Changhua (CHC), termed CHC3, was a man aged 62 years who was admitted to the intensive care unit on February 3, 2020, owing to rapid progression from an influenza-like illness to severe pneumonia with respiratory distress. Upon admission, a preliminary diagnosis of viral pneumonia, initially suspected to be caused by influenza, was made. Unfortunately, the disease progressed rapidly, and on February 15, 2020, when the COVID-19 RT-PCR test result returned positive, the patient passed away. CHC3 was the first COVID-19 case detected through community-acquired pneumonia surveillance in Changhua.

Upon confirming CHC3 as a COVID-19 case, the Changhua County Public Health Bureau promptly initiated contact tracing with the assistance of digital technology. Within 36 hours, the officials had identified the index case of the cluster. The tracing revealed that CHC3 had come into contact with a passenger who had visited China, Hong Kong, and Macau and was suspected of contracting the virus there. Although the passenger tested negative for COVID-19 in an RT-PCR test on February 16, 2020, a subsequent serological test for immunoglobulin G against SARS-CoV-2 returned positive, indicating that the passenger was indeed the index case of the cluster. Figure S2B in Multimedia Appendix 1 illustrates the timeline of COVID-19 transmission, reconstructed through digital contact tracing.

Further contact tracing for CHC3 identified 4 asymptomatic household members positive for COVID-19 (cases 4-7 in Changhua; Figure 5). As CHC3 had worked as a pak-pai taxi chauffeur during the transmission period, the Changhua County Public Health Bureau had to rely on both traditional and digital methods to conduct occupational contact tracing, involving the collection of anonymous data. With the assistance of integrated big data and digital contact tracing, the bureau was able to trace and identify >600 individuals who had occupational or social contacts with the cluster. The contact tracing process included several steps: clarifying CHC3’s family contact structure (Figure 5), identifying risk information using digitalized systems (Figure 6), and tracking the history of contacts using a digital contact tracing system (Figure 7). Following these steps, the pathways of contacts within households, workplaces, health care systems, and social occasions were elucidated, resulting in a total of 665 contacts traced within 30 days for this cluster (from January 13, 2020, to February 12, 2020). Table 1 summarizes the details about all the contacts traced in this clustered event and the results of RT-PCR tests, including the types and positive rates of cases in Changhua.

The estimated contact rate and overall positive rate were 22 per day (ie, 665/30, representing that 665 contacts took place during the 30-day period) and 0.6% (4/665), respectively. Assuming a 7-day infectious period [33,34], Rt was thus estimated as 0.92 (contact rate of 22 per day × transmission probability of 0.006 per contact × infective period of 7 days), indicating the successful containment of community-acquired outbreak for this clustered event.

On July 29, 2020, a Belgian engineer working in Changhua, Taiwan, tested positive for COVID-19 during his routine RT-PCR test before an international flight (reported as case number 469 by the Central Epidemic Command Center in Taiwan). Case 469 had arrived in Taiwan on May 1, 2020, and had been in the country for 3 months (Figure S2C in Multimedia Appendix 1). Therefore, the patient was classified as a domestic case with a potential exposure period to infection from January 25, 2020, to July 8, 2020. The infectious period was estimated to be from July 9 to 29, 2020 (Figure S2C in Multimedia Appendix 1).

Initial investigations revealed that case 469 had not consistently worn a mask in public areas and had a sociable lifestyle. Figure S2C in Multimedia Appendix 1 displays the timeline of this case, including specific exposure instances. To prevent a community-acquired outbreak, the Changhua County Public Health Bureau immediately initiated contact tracing procedures, using digital contact tracing to collect information about contacts in hotels, workplaces, gyms, bars, and social activities related to case 469. A total of 462 potential contacts were identified, and the numbers and containment measures are summarized in Table 2. Of the 462 contacts, 47 (10.2%) were placed in self-isolation, and 415 (89.8%) were advised to practice self–health management. None of these contacts (0/462, 0%) tested positive for SARS-CoV-2 infection in RT-PCR tests.

In this study, we have demonstrated the implementation of digital cluster investigation by using 3 empirical SARS-CoV-2 infection clusters in the early phase of the pandemic, when D614G was the dominant strain. However, the optimization of digital contact tracing requires updated information, varying with different VOCs from D614G and Alpha until Omicron and its subvariants to provide precise containment, including NPIs, antiviral therapy, and vaccination.

The implementation of digital contact tracing was made possible by integrating the public health, civil affairs, and police departments with the help of IT to collect detailed TOCC information at the individual level that could be retrieved using the NHI platform to achieve the goal of efficiently and precisely containing the COVID-19 outbreak in a community. The analysis focused on 3 empirical clusters of COVID-19 cases in Changhua, Taiwan, which included the first family cluster infection, the first community cluster infection in Taiwan, and a foreign worker cluster with an extensive history of contacts. These clusters served as examples of how digital contact tracing was effectively used for EID containment. Although there was no evidence of transmission within the household cluster and foreign worker cluster, the community cluster in Changhua (stemming from CHC3) had an estimated Rt of 0.92. This suggests confined transmission within the community following a prompt intervention aided by digital contact tracing.

Following the successful experience in using digital contact tracing for containing COVID-19 among the large number of contacts (n=627,386) [5], a series of digital approaches was adopted. These included the incorporation of COVID-19 into the TRACE platform that has been established since 2017 [35]; development of a Bluetooth-based contact tracing platform, social distancing 2.0 [36]; a process of digital transformation for disease control [37]; and digital health governance during COVID-19 [38].

Since the onset of the COVID-19 pandemic in 2020, various frameworks and methods for digital contact tracing have been developed [8,10,11,13-23]. In addition to techniques aimed at enhancing the efficiency of identifying individuals at risk of infection owing to contact with pathogens, we have demonstrated a comprehensive strategy that integrates this information with traditional approaches. This strategy includes specific implementation steps for using big data analytics to curb the spread of COVID-19 within a community.

Despite the benefits of digital contact tracing in monitoring and preventing COVID-19 transmission, its implementation is still debatable because of ethical concerns [10,39,40]. The main issue lies in the violation of individual autonomy and privacy as a result of collecting the detailed information required for tracing the history of contact and assessing the risk of disease spread. These humanitarian aspects were emphasized during the implementation of digital contact tracing in the early phase of the COVID-19 pandemic. Ethical guidelines for contact tracing apps were proposed in 2020 [40]. These guidelines outline 4 major principles for collecting individual data: necessity; proportionality; sufficient effectiveness, timeliness, popularity, and accuracy; and temporariness. The digital contact tracing implemented in Taiwan fulfills all of the 4 principles. By using the well-established public service of the NHI, authorized professionals may assess various information, including travel, quarantine location, and health status, using a smart IC card for identity and media for data retrieval. The retention of tracing data for no more than 28 days ensures compliance with the principle of temporariness. These approaches guarantee both flexibility and security in digital contact tracing. Nevertheless, close monitoring for following these ethical guidelines has been implemented for all forms of digital contact tracing.

The transition from traditional contact tracing to a digital strategy supported by big data analytics was developed using detailed information to assess the risk of disease spread, on the basis of which an effective containment measure could be applied; this serves as the first illustration of household clustered transmission. The system integrated big data and various digital contact tracing resources, offering a solid foundation for prompt action, which is crucial for stopping the spread of COVID-19 within a community.

In the second illustration of community cluster infection, collaboration between the public health sector and the police department enhanced the digital contact tracing, allowing 665 contacts to be traced and quarantined and further cases to be efficiently isolated. The likelihood of a community-acquired outbreak being contained is thanks to this improved digital contact tracing for the family and occupational contacts of the pak-pai taxi chauffeur, who covered Changhua county and Taichung city (a metropolitan area close to Changhua county, as illustrated in Figure S3 in Multimedia Appendix 1). For instance, case 7 in Changhua, a family member and also a contact of CHC3, who was later confirmed as a COVID-19 case, resides in Taichung and works there. Without the digital contact tracing system, it would have been impossible to precisely implement isolation and quarantine measures and quickly identify the contacts of case 7 in Changhua.

The final investigation focused on a series of incidents linked to a foreign engineer. The index case had an extensive network of contacts and had traveled to Changhua county, Taipei city, and Yilan county during his infectious period (Figure S2C in Multimedia Appendix 1). The digital contact tracing system enabled the reconstruction of potential contact pathways and timelines, facilitating local health authorities in promptly implementing containment measures. Owing to these precise containment measures supported by digital contact tracing, in addition to the widespread use of masks, monitoring for COVID-19 symptoms including body temperature, and social distancing measures, the risk of a large-scale community-acquired outbreak in Taiwan remains low.

These initial outbreaks in Changhua were contained to a small number of clusters that were manageable for Changhua’s health care infrastructure owing to the combination of digital contact tracing for targeted monitoring and surveillance and universal NPIs governed by a national level-3 alert [30,31]. The large-scale community outbreak caused by the Alpha VOC, which struck Taiwan between May 2021 and July 2021, was also mitigated by using this digital contact tracing method.

Incorporating the proposed digital contact tracing method into our public health strategy offers an efficient tool for managing community-acquired outbreaks. Moreover, the synergistic use of this digital approach alongside comprehensive databases holds significant promise in the timely identification of emerging variants or subvariants (such as XBB.1.5, which descends from Omicron BA.2) of EIDs. These variants may originate from imported cases or emerge within our domestic setting, akin to the scenario of Alpha and Omicron outbreak that occurred in Taiwan [31]. Drawing lessons from the COVID-19 pandemic, we recognize the importance of monitoring asymptomatic and presymptomatic imported cases upon their arrival in Taiwan. This can be efficiently achieved by harnessing immigration registries and implementing community surveillance mechanisms for tracking COVID-19 transmission. Integrating these data sources with our NHI database platform, which also encompasses communicable disease reporting systems, empowers us to facilitate early interventions. This proactive approach is instrumental in halting further transmission stemming from widespread community-acquired COVID-19 outbreaks and, ultimately, mitigating the risk of another EID pandemic. Without loss of generalizability, we used COVID-19 as an illustration for triggering the digital transformation of contact tracing. With the advent of digital contact tracing, timely decision-making can facilitate the containment and mitigation of EID.

In this study, we used Rt as an indicator to evaluate the effectiveness of contact tracing. However, in addition to digital contact tracing, there are many factors that affect Rt, such as the strategies for testing and isolating high-risk contacts, transmissibility and pathogenesis of VOCs, and accessibility of medical and public health resources. Evaluating the performance of digital contact tracing, using indexes such as the number of contacts required to identify a confirmed case, and performing sensitivity and specificity evaluations can help monitor the efficiency and adequacy of the system. These performance metrics also aid health policymakers in allocating limited resources, including those needed for isolation and testing, and the solicited indirect costs as a result of all warning messages.

In communities with a low risk of EID transmission, stringent criteria for determining the contacts eligible for NPIs are warranted, whereas the opposite holds true for areas experiencing outbreaks. These criteria are also influenced by the transmissibility and severity of an EID. Balancing the digital contact tracing system to address the trade-off between identifying all contacts to reduce the risk of missing individuals who are potentially infected and targeting a subgroup of contacts with high risk of infection is the objective of an ongoing study.

In conclusion, we have demonstrated with the COVID-19 cluster investigation how the contact tracing process can be digitized by integrating multisectoral databases to provide the necessary information for risk assessment and decision-making in a timely manner. The better use of the contact tracing technique facilitates the efficient containment of community-acquired outbreaks and the prevention of subsequent large-scale outbreaks, especially when facing EIDs in the future.