Performance and utility of contact tracing for COVID-19 in the WHO South-East Asia Region: implications for future pandemic preparedness

Summary

Contact tracing was widely implemented during the COVID-19 pandemic, but its real-world performance and utility for decision-making remain poorly understood. A qualitative study was conducted to appraise the performance and utility of contact tracing for COVID-19 in the WHO South-East Asia Region, based on interviews with government and non-governmental organisation technical staff and decision makers in Indonesia, Nepal and Thailand. Our findings highlight the good performance of contact tracing when sufficiently resourced and when case load is low, but reveal declining utility as case incidence increases. This study presents key definitions and a pragmatic approach for appraising contact tracing performance and utility throughout a major health emergency response. Countries should prospectively define objectives for contact tracing, establish monitoring and evaluation frameworks, adjust their contact tracing approaches informed by risk assessments, and consider other available public health interventions when its performance and utility decline.

Funding

Governments of Germany and Australia.

Background

Contact tracing was implemented worldwide to interrupt transmission and reduce deaths during the COVID-19 pandemic.¹ However, evidence regarding the performance of contact tracing for COVID-19 remains largely based on observational studies conducted early in the pandemic or estimated from modelling studies.^2,3 Few studies have described the implementation of contact tracing throughout the pandemic.²

In January 2025, the World Health Organization (WHO) released disease-agnostic contact tracing guidance that drew on lessons learnt from COVID-19 and other outbreaks.⁴ Owing to lack of evidence, key questions regarding the operational performance of contact tracing were not addressed in the WHO guidance. This included how the objectives of contact tracing should evolve over different epidemic or pandemic phases, whether alternatives to contact tracing may be more effective amid resource constraints, and the contribution of digital technologies to contact tracing performance. Furthermore, the WHO guidance found there were few quantitative indicators of contact tracing performance available,⁴ and performance indicators are not always directly comparable.5, 6, 7 A review of contact tracing indicators for COVID-19 reported that most contact tracing indicators were process or output indicators, and no indicators were defined or reported for financial resources, transmission chain interruption, or incidence reduction.⁸ Additionally, the perceived utility of data generated by contact tracing for public health decision-making during the pandemic was not addressed in the WHO guidance and has received limited attention in the literature. Finally, most studies of contact tracing have been conducted in high-income countries,⁹ with very limited evidence regarding the performance and utility of contact tracing in low-and-middle-income countries.

To address these gaps, this study aimed to appraise the performance and utility of contact tracing for COVID-19 in the WHO South-East Asia Region (WHO SEAR) from January 2020 to May 2023, with the goal of informing future epidemic and pandemic preparedness in the region. This study extends on the findings of a recent review of COVID-19 surveillance and contact tracing in this region¹⁰ by establishing key definitions and a pragmatic approach for describing and analysing the real-world performance of contact tracing and its utility for decision-making over different pandemic phases.

A qualitative case-based study was designed to explore contact tracing policy and implementation during the COVID-19 pandemic in the WHO SEAR. The study period was divided into the alert phase (from January 2020 to 11 March 2020, when WHO first described COVID-19 as a pandemic), the event phase (from 12 March 2020 to early/mid-2022, including the emergence and spread of several key variants of concern), and the transition phase (from early/mid-2022 to 5 May 2023, when WHO rescinded the Public Health Emergency of International Concern declaration). The timing of phases varied between countries, as some countries in the region shifted their response measures earlier than others.

Guided by a comprehensive literature review,¹⁰ this study adapted frameworks used to describe and evaluate surveillance system performance¹¹ and contact tracing performance indicators frequently used during the COVID-19 pandemic⁸ to define key domains of contact tracing performance. Performance was defined as comprising three components, as follows:

• •Accuracy: The extent to which contact tracing correctly identified, notified, and managed individuals who were at risk of infection due to exposure to confirmed cases during the case's infectious period (noting the precise definition of a contact varied over time and between countries).
• •Timeliness: The relative speed and ease with which the COVID-19 contact tracing system identified exposed contacts in time to reduce the risk of onwards transmission (for example, by directing contacts to quarantine before onset of their infectious period).
• •Resources: Considers the level of resources (human, technological, financial) required to implement contact tracing over different phases of the pandemic.

The utility of contact tracing data for decision-making was defined as the perceived effectiveness, demand for, and uses of contact tracing data over different phases of the pandemic by public health decision-makers.

The performance and utility of contact tracing were further considered according to the intended coverage of contact tracing, which was broadly categorised as either comprehensive, with the intention to initiate contact tracing for most or all cases, or restricted, where risk-based approaches were used to determine whether to initiate contact tracing for particular cases or settings.

Indonesia, Nepal, Thailand were selected for case studies based on their representation of varied COVID-19 epidemic contexts, diverse social, cultural, religious, political, and economic contexts, as well as different public health system structures and varying health system capacities (e.g. regarding diagnostic testing), factors expected to have also influenced the performance and utility of contact tracing (Fig. 1).

Fig. 1.

Key event related to the COVID-19 pandemic are indicated for each case study country, annotated on a timeline of weekly COVID-19 case reports to WHO.¹² Note that case counts reported to WHO are not adjusted for population size or intensity of surveillance, and that dates of key events indicated are approximate given compressed time scale displayed on horizontal axis.

Key informant interviews (KIIs) were conducted online and in-country from September to November 2023. Topic guides for KIIs were developed based on a region-wide literature review¹⁰ and refined iteratively to illuminate perspectives regarding the key domains of contact tracing performance being explored.

Key informants were identified based on a pre-specified purposive sampling strategy. For each country, we aimed to conduct KIIs with the following stakeholder mix:

• •3 KIIs with senior public health decision makers (1 at national level, 2 at sub-national level).
• •3 KIIs with senior technical staff with responsibility for COVID-19 surveillance, including use of digital technologies (1 at national level, 2 at sub-national level; prioritising staff who had been in position during most or all of the COVID-19 pandemic).
• •3 KIIs with senior technical staff with responsibility for contact tracing, including use of digital technologies (1 at national level, 2 at sub-national level; prioritising staff who had been in position during most or all of the COVID-19 pandemic).
• •1 KII with the national focal point for digital health, where this role existed.
• •1 KII with WHO national focal point for the COVID-19 response.
• •2 KIIs with development partners, bilateral agencies, and/or NGOs involved in COVID-19 surveillance and contact tracing.

Government stakeholders were identified and invited to participate in coordination with WHO Country Offices. NGO representatives were invited to participate in the study directly by the researchers. Interviews were conducted in English or in national languages with live simultaneous interpretation, according to the preferences of the interviewees. The research team comprised epidemiologists, public health specialists and qualitative health researchers with considerable experience working in the South-East Asia region. For each interview, one researcher with training in epidemiology or public health and one researcher with training in qualitative research methods participated. With participants’ consent, interviews were audio-recorded. Interviewers also took notes during interviews and completed debrief forms after each interview to assist refinement of the methodology and aid later contextualisation of themed findings.

Interview recordings were automatically transcribed using Otter™ transcription software (Otter.ai (2023), Mountain View, CA, www.otter.ai) and then manually reviewed and corrected. Transcripts, notes and debrief forms were uploaded to Dedoose™ qualitative data management and analysis software (Dedoose Version 9.0.107 (2023), Los Angeles, CA: SocioCultural Research Consultants, LLC, www.dedoose.com).

A coding frame to enable thematic analysis was developed a priori based on reviewed literature and refined based on independent coding of two transcripts by two research team members. The draft coding frame was reviewed and revised by three research team members who had conducted KIIs online and in-country. Memos were used to record and develop emerging themes, including links to relevant excerpts from the transcripts/notes.

Findings were anonymised with respect to the three case study countries, presented as Country A, B, and C, and jurisdictional level of key informants, to reduce the risk of re-identification.

The study was approved by the Research Ethics Review Committee of the World Health Organization Regional Office for South-East Asia (ID 2023.29.MC). Plain language summaries were made available to participants in local languages. Verbal informed consent was sought from all participants in online and in-person interviews. Additional verbal consent was sought to audio-record interviews. The consent process was documented by the interviewers. Participants were free to withdraw their consent at any time.

40 KIIs were conducted across the three case study countries (Table 1), of which 37 were conducted in person in-country, three were conducted online, and 16 conducted as group interviews. Themes are presented based on analysis of data from the KIIs from the three case study countries and reviewed literature from all WHO SEAR Member States. Illustrative quotes are shown in Tables 2 and 3, respectively.

Table 1.

Key informant interviews held across case study countries.

Countrya	National government	Subnational government	Non-government organizationb	Total
A	4	5	4	13
B	7	6	3	16
C	8	2	1	11
Total	19	13	8	40

^aCountries are anonymised in the findings to avert the risk of reidentification of key informants.

^bIncludes World Health Organization Country Offices.

Table 2.

Illustrative quotes from key informants regarding performance of contact tracing.

Quote ID	Quote
C1	“For example, they [contact tracers] take photos of the migrant workers that go [into] a database. They've got to link that with results [of testing] and then link their forms to the migrant worker who is positive. So it is really hard at that time.” (interview 15)
B1	“Most of the people have like a temporary address and they get tested somewhere and the information is like the permanent address they put, and while we do the contact tracing they say that no there is no person here, like so that type of problem we face.” (interview 28)
A1	“[Contact tracers] not from health backgrounds didn't fully understood what is the purpose of contact tracing. People who are contacted or exposed need to have proper [support to] stay at home ….daily life support. But … people who were contacted feel that it's, it's like a crime. And then so, at that time, it developed a situation of stigma, and people tend to hide. When actually, the idea is how to develop empathy, and it's not happening” (interview 7)
B2	“I do admit that there are a lot of lapses in that case investigation and contract tracing, so much so that it was well organised in small towns and villages, but larger cities like [capital city], [regional city] and all where there are a large chunk of people who are temporary residents, or who stay in rentals and all, many of them were missing [from] these activities. Because we didn't have definitive data on how many people reside in each house” (interview 33)
A2	“So for example, if you have to find 10 [contacts] … there's no guidance for example, to prioritise the household contacts or prioritise the people they were at work with. It's just find the 10” (interview 11)
A3	“Using the manual [system] we had to write lots of things and try to verify serial numbers with the ID, we had issues. We didn't have the actual database, especially concerning the citizens, the residents. Using the apps, because it's been connected to the government, we have information on every individual: the name, the age, the address, everything, instead of us doing it manually. So in terms of speed, using the app, it was way better.” (interview 2)
C2	“We couldn't get data back from [CT apps]. We asked [central government] but they couldn't pull it out. It wasn't specific enough. It wasn't timely.” (interview 21)
B3	“There are 200 health workers, but in the frontline, there were only 10 people were working. The rest were not agreeing to work. The 10 health workers, they were supposed to be provided with N95 masks and sanitised or thermal gun. Even to provide this equipment to them was a big challenge …” (interview 35)
B4	“But this has truly required a sizable engagement of the whole of the society to really tackle this contact tracing … To be very honest, it's impossible job to do what we were asking our frontline care providers to do.” (interview 39)
A4	“Because there were just too many cases, we couldn't input all of them. We would get, like, 500 in one day. So imagine doing contact tracing to so many people who were in contact. I lost a lot of sleep.” (interview 13)
C3	“Second thing, we had many databases that we had to join … sometimes at the beginning it wasn't that easy. They didn't match to each other and we had a lot of problems with the IT techniques to connect the data to each other–we don't have IT staff. We're only epidemiologists. We have some programmers, but they're not data scientists. There's a lot of problems about the data management and the database management that we had to solve and it takes time to fix those problems that we had during the crisis.” (interview 15)
C4	“So if you develop a system and it's good for catching all the variables, it works when you have the small number of cases. But then when the cases increase, it's not just the technology that struggles to keep up. It's the people and their workloads. How do you have time to do the contact tracing and the entering [of] the data?” (interview 18)
A5	“To fulfil the criteria, sometimes by the end, we kind of made it up” (interview 4)
B5	“Many police authorities were also involved along with the health workers for this contact tracing team … When there was an outbreak amongst travellers from [neighbouring country] and belonging to a certain community … involving police authorities … created a kind of a panic situation among those communities. So that was one of the biggest lessons on that, that, you know, outbreak investigation needs to happen more in a more of a humanitarian [way], [with a] health workers aspect rather than more of a security aspect. So, that was also one of the lessons learnt.” (interview 27)
A6	“We need to do more approaches to local government, local figures, in order to determine what is best for their community.” (interview 10)

Table 3.

Illustrative quotes from key informants regarding the utility of contact tracing data for decision-making.

Quote ID	Quote
C5	“Each district followed the same protocol but then response depended on results–they could decide based on results” (interview 21)
B6	“So the monitoring and evaluation component was completely lacking, like measuring the performance indicator of these provinces … all these components were not happening for contact tracing … due to the lock downs and everything happening at the same time. So it's quite tricky to monitor contact tracing during different phases of pandemic.” (interview 27)
C6	“We found from contact tracing that most of the high risk contacts were household contacts so we learnt from our own data who that were the most important people to target” (interview 17)
A7	“So when the spread was massive, contact tracing became meaningless.” (interview 11)
C7	“Contact tracing was doable in the early stages because most cases were travellers … We did contract tracing as though for an emerging infectious disease … In early March there was a big cluster, and this was our first major contact-tracing exercise … The mindset changed during Delta wave in mid-2021. Delta came in April, but took around two months before rising sharply. At that time, we realised that contact tracing and containment strategies were no longer appropriate in a pandemic situation, so we changed strategy.” (interview 24)
B7	“We did not scrap the policy. So officially policy was there. But in reality, contact tracing, wasn't there.” (interview 14)
B8	“That was political also, if you say that you are not doing contact tracing. It would not have sounded good at that time. That's why it continued but not with stronger efforts.” (interview 33)

Implementation of contact tracing

Key informants in all three countries reported that contact tracing was first implemented in the alert phase, with timing varying in accordance with the date of first detection of COVID-19 within each country. All three countries conducted case investigations and contact tracing for laboratory-confirmed COVID-19 cases, however there was variability within countries and over time regarding whether contact tracing was performed for suspected cases, or cases that tested positive through self-administered rapid antigen tests. All three countries initially directed contacts to quarantine, often in government-run facilities, and later introduced home-based quarantine. In the first few months of the pandemic, contacts were frequently directed to quarantine without undergoing testing, or prior to their test result being known, due to testing capacity shortfalls. By the late event and transition phases, quarantine requirements were relaxed or removed for contacts classified as low risk (e.g. vaccinated contacts). Contact notification systems also changed over time in accordance with the overall risk management approach in each country. For example, in some contexts, cases were directed to self-notify their contacts, and/or contacts were advised to self-test once rapid antigen tests were widely available. Traditional and social media outlets were also used to announce recent public exposure locations to alert potential contacts. Contact follow-up included monitoring compliance with quarantine and/or testing requirements, depending on the pandemic phase. In all three countries, contact tracing efforts de facto reduced or largely ceased once the Delta Variant of Concern was widespread, due to lack of sufficient capacity or strategic orientation away from contact tracing.

All three countries launched contact tracing at national level initially, before decentralising implementation and decision-making to provincial and local health authorities. Contact tracing protocols were revised over time, in accordance with emerging information about COVID-19 transmission characteristics, in response to rising incidence, and in conjunction with broader adjustments to the pandemic response in each country. Initially, all three countries utilised manual contact tracing systems (paper-based and/or Microsoft Excel or Google spreadsheets), but ultimately introduced digital technologies to support contact tracing. These included electronic information systems to assist with data entry and linkage for contact tracers, as well as smartphone apps designed for use by the general public to support detection of possible exposures. Communications software were also used by contact tracers and public health personnel to coordinate contact tracing within and between subnational areas.

Performance of contact tracing

Contact tracing performance was reported to vary over the pandemic phases and across case study countries. During the alert phase, there were several instances of relative success applying contact tracing to reduce or eliminate local transmission clusters when implemented in conjunction with other public health and social measures, including international and domestic border control restrictions. However, there was a clear consensus across key informants from all countries that contact tracing was ineffective once case incidence exceeded contact tracing capacity. Key informants reported that capacity was exceeded in the event phase in each country, including in some settings prior to the emergence and spread of the Delta Variant of Concern.

Accuracy

Perceived accuracy of contact tracing varied substantially through the pandemic phases in all three countries. Generally, once contact tracing protocols were established, there was higher perceived accuracy early in the pandemic when COVID-19 case incidence was low, with declining accuracy due to incomplete contact tracing and limited coverage as case incidence progressed. In all three countries, there were key population groups that were under-represented or missed in contact tracing. For example, in Country C, a particular weakness was noted for contact tracing amongst migrant workers, due to language barriers, lack of national identification cards, use of multiple names, overcrowded residences, and other challenges (Table 2, C1). In Country B, the lack of a national identification system also constrained the accuracy of contact tracing, due to errors in identification of the primary residence and hence exposure sites associated with a confirmed case (Table 2, B1). In Country A, migrant workers were issued with a national identification card, but initially this was a paper-based system, which key informants reported led to challenges with identity verification. This improved with a shift towards digital identity records from 2021 onwards.

Challenges with confidentiality, trust, and stigma further constrained the accuracy of contact tracing. For example, key informants reported that operators of unlicensed and illicit businesses in Country C were reluctant to cooperate with contact tracers, and undocumented migrant workers avoided presenting for testing and case investigation. Across countries, ethnic and religious minority groups were viewed as less likely to self-present for case investigation or share information with contact tracers, due to fear of COVID-19 as well as the impact of containment strategies such as quarantine. This was exacerbated by the lack of financial and social support for quarantined contacts and their families, leading to under-reporting of cases and contacts, as well as non-compliance with quarantine (Table 2, A1). Across all three countries, contact tracing accuracy was perceived to be higher outside of major population centres, due in part to lower case incidence, but also due to a more stable population with known addresses, and the existence of traditional governance structures and roles such as village heads with whom contact tracers could coordinate (Table 2, B2).

During the event phase, as case incidence overwhelmed contact tracing capacity, key informants in all countries described a formal or de facto shift towards restricted rather than comprehensive contact tracing. In Country C, this was formalised as a risk-based targeting approach early in the alert phase onwards, with the contacts of the first locally detected cases categorised as high risk or low risk contacts based on their exposure history (for example, seating position on an incoming passenger plane with a confirmed case onboard). Country C later expanded its risk-based targeting as part of its overall strategy to mitigate COVID-19 spread whilst maintaining economic activity. In Country A, key informants reported that contact tracing efforts shifted towards identifying the targeted number of contacts per case (20 contacts per case, or 10 contacts per case, depending on policy in place at the time) without necessarily adhering to a risk-based strategy for contact identification (Table 2, A2).

Timeliness

Several key informants reported that delays in the initiation of contact tracing activities, the availability of protocols, and the notification of the first confirmed case reduced the timeliness of contact tracing in the alert phase in Country A and Country B. During the event phase, key informants reported that contact tracing was often conducted days or weeks after the relevant exposure period due to long turnaround times for test results—particularly before the introduction of rapid antigen tests—as well as human resources constraints and other challenges.

Timeliness was partly improved through increased human and financial resources as well as scaling up testing infrastructure. For example, in Country A, substantial delays in testing persisted until the initiation of PCR testing at provincial reference laboratories from the second half of 2020 onwards; previously, all samples were sent to the national capital for testing. During this period, contacts were directed to quarantine but could not be tested, which key informants reported reduced situational awareness about COVID-19 and compliance with quarantine in the population. Adoption of digital technologies designed to support contact tracers’ workflow reportedly improved timeliness, particularly with the integration of surveillance and contact tracing systems (Table 2, A3). However, smartphone applications designed for use by the general population were reported to have no impact on contact tracing timeliness in all three countries (Table 2, C2). At population level, timely contact tracing could not be sustained in the face of rapidly rising case incidence, which overwhelmed contact tracing resources.

Resources

Resource constraints limiting contact tracing arose early in the alert phase in Countries A and B, before COVID-19 incidence escalated substantially. Early challenges included the lack of provision of personal protective equipment and other measures for contact tracers (Table 2, B3). In all three countries, the contact tracing workforce was rapidly expanded amidst very challenging conditions to scale up contact tracing in the event phase. Despite immense efforts to continue contact tracing, key informants in all countries reported that the contact tracing workforce was quickly over-burdened and under-resourced relative to the immensity of the contact tracing task in the event phase and early transition phase (Table 2, B4).

Other challenges in all three countries included the burden of contact tracing data entry, management, and linkage (Table 2, A4). These challenges were only partially alleviated through introduction of digital technologies designed to support data entry and management for contact tracers, which typically took several months to more than one year to deploy (Table 2, C3). For example, in Country A, a national electronic information system was launched in late 2020 for contact tracing, which was initially integrated with surveillance data and ultimately with vaccination and other key databases. Key informants reported that this system improved timeliness and scalability of contact tracing, and that its utility was enhanced through its compatibility with multiple device types, offline access, and automated reporting of aggregate performance indicators. However, this and similar systems in the other two countries could not address the shortage of contact tracers to conduct case investigations during periods of moderate to high COVID-19 incidence (Table 2, C4). Resource challenges were not ameliorated by the introduction of smartphone apps designed for use by the general population. Key informants reported that these smartphone apps were often not integrated into subnational or national contact tracing systems and in many cases, key informants were not aware of how to access app data to support contact tracing. Furthermore, population-based uptake of these apps and use of contact tracing functions was very low. For example, key informants in Country C estimated there were only 10,000 users of a national Bluetooth proximity-tracing app. In Country A, a smartphone app was introduced and relatively widely used for its case monitoring and vaccination status functions, but there was low or no awareness amongst most key informants of its contact tracing functions, despite proximity tracing functions purportedly having been included.

Human resources to continue contact tracing in line with stated program objectives rapidly dwindled in Country A and Country B. In both countries, key informants reported that contact tracing objectives were not substantially revised as case incidence increased. In Country A, key informants reported that during the late event phase and into the transition phase, contact tracing was largely conducted to respond to nationally defined targets, which informed local ‘zoning’ measures. Informants reported that contact identification was often arbitrary rather than risk-based and that benchmarks were mostly not achievable (Table 2, A5). In Country B, contact tracers were also responsible for other key public health tasks, such as administering COVID-19 vaccines. In practice, the public health workforce de facto deprioritised or largely ceased contact tracing in favour of the vaccination program, even though contact tracing remained an official priority activity. These challenges were exacerbated by concurrent delays and gaps in surveillance, such as testing lags.

All three countries engaged the police and/or military to support contact tracing in the event phase. Community organisations and community volunteers were also engaged to varying extents in each country, depending on the strength and role of these sectors prior to the pandemic. Though the police and military provided a rapidly deployable workforce, key informants in all three countries reported that their involvement in contact tracing was not well supported or accepted amongst communities given the implied linkage of public health protection with security measures (Table 2, B5). In contrast, the involvement of community organisations and volunteers was better supported. In Country A, key informants reported that recruitment of community members as contact tracers was considered an asset to containment strategies overall, as community members were able to mobilise social support (such as providing food and other necessities) to enable individuals and households to comply with isolation and quarantine measures (Table 2, A6). In two countries, community members were often more trusted than security services or public health staff due to their position and status in traditional governance systems, which formed part of local cultural practices. In Country C, community health volunteers who formed an important part of village healthcare in rural areas prior to the pandemic were engaged to support contact tracing; their longstanding and positively perceived community role was considered an asset for improving support for and compliance with contact tracing. Despite numerous efforts to increase the contact tracing workforce, contact tracing capacity was exceeded in all three countries due to increasing case incidence, particularly during pandemic periods dominated by the spread of the Delta and Omicron Variants of Concern.

Utility of contact tracing data for decision-making

Contact tracing data was generated, retained, and used mainly at subnational level. For example, in Country C, national level key informants reported that the regulation, retention and use of contact tracing data at provincial level reduced national level oversight of contact tracing implementation and performance, but was more useful as it enabled contextual-relevant application (Table 3, C5). To the extent that contact tracing data were visible to national level decision makers, it was primarily used to inform decision-making about implementation or lifting of public health and social measures. For example, in Country A, case-to-contact ratios were reported to the national level from the event phase onwards, forming part of the risk matrix used to assign areas to low, medium, or high-risk transmission zones, with corresponding measures put in place. However, in general, in the absence of clearly stated objectives and benchmarks of contact tracing, there was limited or no performance monitoring, which constrained the utility of contact tracing data (Table 3, B6).

Contact tracing data sometimes provided other benefits for decision-makers. For example, in Country B, data on the workload of contact tracers supported the government to successfully advocate for increased funding and resources from international partners. In Country C, contact tracing data for the first few cases during the alert phase was used to generate early insights into the infectiousness of COVID-19 and set initial policies accordingly (Table 3, C6).

Overall, the utility of contact tracing data was perceived to be highest in the alert phase and early event phase in all three countries, and declined substantially due to insufficient capacity for timely and comprehensive contact tracing once community transmission was widespread (Table 3, A7). Utility was further constrained by challenges related to data management within and between jurisdictions and areas. For example, in Country A, key informants reported that sharing data between provinces (for example, for commuting workers) initially relied on individual connections and use of general messaging software to share case details. This was partially addressed by introduction of electronic information systems to manage contact tracing data, as described in the previous section. However, in Country B and C, ongoing challenges with linking surveillance to contact tracing data were reported despite introduction of digital technologies.

Countries took different approaches in response to the declining utility of contact tracing data. Notably, Country C largely switched to alternative strategies such as active surveillance and other strategies that aimed to reduce infections and deaths in the population (Table 3, C7). Key informants in Country C reported that contact tracing continued in lower risk rural areas and/or defined high-risk populations where considered effective, until the transition phase. In contrast, key informants in Country A and B reported that contact tracing was officially retained as national policy until late in the transition phase, but de facto phased out in favour of other interventions, such as COVID-19 vaccination programs (Table 3, B7). Key informants, particularly at subnational level, recognised the poor utility of contact tracing data as the event phase progressed, but reported limited impact of contact tracing data on decision-making (Table 3, B8). Changes were made to national guidelines in response to growing case numbers and the unmanageable burden of contact tracing in both countries, but these changes and revisions were not perceived to have readily filtered down to operational level.

Discussion

This study aimed to assess performance and utility of contact tracing in the WHO SEAR encompassing insights from key stakeholders from national and subnational levels as well as non-government stakeholders. Previous studies comprised mainly modelling studies from high-income countries,¹³ as well as a small number of single-site or single country contact tracing evaluations from the WHO SEAR, often focussing on specific populations,¹⁴ settings,¹⁵ time periods,¹⁶ or contact tracing components.¹⁷

We found that contact tracing performance and its perceived utility for decision-making was highest in the alert phase and early event phase of the COVID-19 pandemic, then rapidly declined, despite substantial efforts to create surge workforces and introduce digital technologies to improve scalability of contact tracing. These findings were consistent across three countries with different epidemiological contexts, resources, and governance models. All three countries initially developed protocols and coordinated contact tracing at national level, but ultimately delegated substantial responsibility for implementation of contact tracing to provincial-level health authorities, at which point contact tracing data was mainly retained and used at provincial level. This was perceived to improve timeliness and utility of contact tracing data at operational level compared to central management, however gaps in data collection and performance monitoring hindered national oversight of contact tracing effectiveness. This was a consistent finding despite differences in the extent of decentralisation between the three countries. Nonetheless, central coordination of aspects of contact tracing, including updating protocols, revising the goals of contact tracing in accordance with changing epidemic characteristics, and enabling national-level data collection and retention to allow for real-time or retrospective evaluation of contact tracing remains an important part of a national health emergency response, especially given severe resource burdens and constraints experienced at operational levels. Despite the well-recognised poor performance and utility of contact tracing once community transmission was widespread, two of the three case study countries retained contact tracing officially as part of the national response, though, in practice, contact tracing activities reduced or ceased when capacities were overwhelmed. The third country initiated a strategic pivot away from contact tracing in favour of other interventions that were considered more effective from the mid-event phase onwards.

Strengths of this study include productive engagement with key informants from different jurisdictional levels and agencies in three countries who provided perspectives on contact tracing performance and utility for the entire pandemic period, rather than only a few months, as is common in existing literature. Data collection in-country at national and subnational level strengthened data quality and reliability, as it allowed for thorough triangulation and iterative refinement of topic guides across different key informants and settings. Key themes emerged quickly and consistently across countries and types of key informants, which further supports the validity of the findings through data saturation. The qualitative data captured in this study offers richly detailed insights on factors that affected the performance and utility of contact tracing for COVID-19, particularly in comparison to prior quantitative studies that focused on a limited set of purposefully selected quantitative indicators with limited generalisability beyond the specific setting, time period, or epidemiological context.

The inclusion of only three out of 11 WHO SEAR Member States as case study countries due to cost and feasibility constraints was a limitation of this study. However, this limitation was partially addressed by purposive selection of countries representing a range of epidemiological contexts, populations, and governance models, as well as by conducting a region-wide literature review to select priority topics and domains for the qualitative case studies.¹⁰ Of note, Indonesia transitioned from the WHO SEAR into the WHO Western Pacific Region after conclusion of data collection. The key informants were limited to government and non-governmental technical staff, therefore the perspectives of local level contact tracers, patients, contacts, and community members were not captured. This constrained the domains of contact tracing performance and utility that could be explored in this study; for example, there were no insights into how communities participated in and made use of local level contact tracing data. Further, there was no opportunity to quantitatively appraise contact tracing performance in the case study countries, due to data access restrictions as well as loss of data availability. For example, many electronic information systems and other digital tools for contact tracing had been withdrawn from use by the time of data collection, and/or data privacy measures had been in place that prevented retention and use of contact tracing data for research or evaluation purposes. Selection of key performance domains was guided by considerations of feasibility of qualitative data collection retrospectively; in a prospective study, it would be optimal to define a broader set of performance domains and collect quantitative as well as qualitative data. Finally, a comprehensive analysis of the contribution of digital technologies to contact tracing was beyond the scope of this study. However, digital contact tracing has been addressed more extensively in the literature¹⁸ than the performance and utility domains assessed in this study.

Conclusions

This study offers key definitions and a pragmatic approach for appraising the performance and utility of contact tracing for a pandemic respiratory pathogen like SARS-CoV-2 in real-world settings. Our findings show that that public health specialists and decision-makers affirmed good performance of contact tracing when done early and with resources commensurate to case incidence, but also reveal its declining performance and utility in the face of rapidly increasing case load. Follow-up studies expanding on issues identified in this study, including the role of national identification systems, residence status, community trust, and the role of digital technologies are recommended, especially studies that explore how these issues have evolved in the late- and post-pandemic period, for which little data are available in the literature. As part of preparedness efforts against future respiratory epidemics and pandemics, countries and international health agencies should prospectively define objectives for contact tracing guided through risk assessments. Monitoring and evaluation frameworks should be established to proactively appraise performance and utility of contact tracing against predefined objectives over different epidemic and pandemic phases. This includes the development of pathways and decision-making processes for adjusting contact tracing approaches as its performance and utility decline, through risk assessment, taking into consideration the epidemic context, disease severity, clinical and epidemic properties of the pathogen, and availability of resources. It may also involve shifting to risk-based contact tracing in defined populations or to other available public health interventions. Guidance for adjusting contact tracing approaches is especially pertinent when efforts to strengthen contact tracing through workforce expansion and introduction of digital technologies have not, or are unlikely to, meaningfully improve its performance and utility overall.

Contributors

Melanie Bannister-Tyrrell: Conceptualisation: YES; Data curation: YES; Formal analysis: YES; Funding acquisition: YES; Investigation: YES; Methodology: YES; Project administration: YES; Resources: YES; Software: NO; Supervision: YES; Validation: YES; Visualisation: YES; Writing—original draft: YES; Writing—review & editing: YES.

Kirsty Teague: Conceptualisation: NO; Data curation: YES; Formal analysis: YES; Funding acquisition: NO; Investigation: YES; Methodology: NO; Project administration: NO; Resources: NO; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Daniel Strachan: Conceptualisation: YES; Data curation: NO; Formal analysis: YES; Funding acquisition: NO; Investigation: NO; Methodology: YES; Project administration: NO; Resources: NO; Software: NO; Supervision: NO; Validation: YES. Visualisation: YES; Writing—original draft: NO; Writing—review & editing: YES.

Anna Barrett: Conceptualisation: NO; Data curation: YES; Formal analysis: YES; Funding acquisition: NO; Investigation: YES; Methodology: YES; Project administration: YES; Resources: NO; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Tiara Marthias: Conceptualisation: NO; Data curation: YES; Formal analysis: NO; Funding acquisition: NO; Investigation: YES; Methodology: NO; Project administration: NO; Resources: NO; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Clare E Strachan: Conceptualisation: NO; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: YES; Methodology: YES; Project administration: NO; Resources: YES; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Hannah Brindle: Conceptualisation: YES; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: NO; Methodology: NO; Project administration: YES; Resources: NO; Software: NO; Supervision: NO; Validation: NO; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Pawinee Doungngern: Conceptualisation: NO; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: NO; Methodology: NO; Project administration: NO; Resources: NO; Software: NO; Supervision: NO; Validation: NO; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Yuka Jinnai: Conceptualisation: NO; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: NO; Methodology: YES; Project administration: YES—specifically in facilitating the ethical approval; Resources: NO; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Mushtofa Kamal: Conceptualisation: YES; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: YES; Methodology: NO; Project administration: YES; Resources: NO; Software: NO; Supervision: NO; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Nishant Thakur: Conceptualisation: NO; Data curation: NO; Formal analysis: NO; Funding acquisition: NO; Investigation: YES; Methodology: NO; Project administration: YES; Resources: NO; Software: NO; Supervision: NO; Validation: NO; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Masaya Kato: Conceptualisation: YES; Data curation: NO; Formal analysis: YES; Funding acquisition: YES; Investigation: NO; Methodology: YES; Project administration: YES; Resources: YES; Software: NO; Supervision: YES; Validation: YES; Visualisation: NO; Writing—original draft: NO; Writing—review & editing: YES.

Florian Vogt: Conceptualisation: YES; Data curation: YES; Formal analysis: YES; Funding acquisition: NO; Investigation: YES; Methodology: YES; Project administration: NO; Resources: NO; Software: NO; Supervision: YES; Validation: YES; Visualisation: YES; Writing—original draft: YES; Writing—review & editing: YES.

Data sharing statement

To protect the privacy of interview participants and mitigate the risk of re-identification, raw qualitative data for this study including interview transcripts cannot be made publicly available or shared outside of the research team.

Declaration of interests

This study was funded by the Governments of Germany and Australia (Department of Foreign Affairs and Trade) via the World Health Organization Regional Office for South-East Asia. The funders had no involvement in the study design; the collection, analysis, or interpretation of the data; the writing of the report; or the decision to submit this paper for publication. The authors have no conflicts of interest to declare.