DOI: https://doi.org/https://doi.org/10.57187/smw.2023.40112
Contact tracing (CT) has been an important part of strategies to reduce the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections and control the coronavirus disease 2019 (COVID-19) pandemic [1–2]. Forward CT involves identifying contact persons whom the index case might have infected, whereas backward CT involves finding the source of the infection of the index case. Backward tracing appears to be more effective for controlling COVID-19, based on the principle that the person who infected the index is likely to have had more contacts than the index themself [3–4]. In Japan, public health experts have suggested that forward CT is only feasible below a 7-day incidence threshold of 15 cases per 100,000 people (Hitoshi Oshitani, personal communication, 7 February 2022). The ‘test-trace-isolate-quarantine’ strategy, which includes CT activities, can break chains of transmission [5]. During the COVID-19 pandemic, authorities were under pressure to quickly establish CT systems and deal with surges in cases. Although public health authorities quickly trained individuals to join CT teams [6], there was a great need to incorporate a variety of digital tools into CT systems to reduce the high workload of tracers and make CT more efficient [7].
There is limited evidence on the performance of digital outbreak response tools [8]. According to the World Health Organization (WHO), outbreak response tools [9] are workflows that facilitate data entry and automated communication with cases and contacts, particularly when caseloads are high. They are also known as partly automated CT systems [8] as some parts of the system are automated, e.g. messages are sent automatically, whereas other tasks need to be done by a human, e.g. the index case providing contact details in an online form or an in-person interview. These digital systems are distinct from proximity tracing applications [10], which are used as complementary tools to notify users that they have been in close physical proximity to an infected person [9].
Several key performance indicators (KPIs) have been used to evaluate the performance of CT for COVID-19 [11–12] and can be categorised according to whether they measure the structure, process or outcome of a public health intervention like CT [13]. According to Swiss law, the cantonal physician’s office in Solothurn was responsible for implementing CT and they developed a partially automated CT workflow in their canton, which was introduced after the first wave of the COVID-19 pandemic. The objectives of the present study were 1) to describe the CT system developed by the Swiss canton of Solothurn and 2) to describe KPIs to analyse CT outcomes between the implementation of new CT software on 15 November 2020 (onset of the second wave) and 2 February 2022, at which point the Federal Office of Public Health (FOPH) of Switzerland ended the requirement to isolate [14].
We used CT data that were routinely collected between 15 November 2020 and 2 February 2022. More details about the canton of Solothurn, definitions and CT practices can be found in supplementary file 1 and in a previous analysis [15]. We did not prepare a statistical analysis plan in advance.
Solothurn is a mid-sized canton in Switzerland with a population of approximately 280,000 [16]. The official language is German and three quarters (213,800; 76%) of the population are Swiss, which is similar to Switzerland as a whole (supplementary table 1). Over the course of the COVID-19 pandemic, the Swiss government and the canton of Solothurn imposed several measures to reduce transmission of SARS-CoV-2 including restrictions on gatherings, mask mandates, school closures, travel quarantines, but also the provision of contact data based on the Epidemics Act law and FOPH directives [14, 17]. As of December 2022, 139,717 SARS-CoV-2 infections had been confirmed in Solothurn and 379 people reported to have died of COVID-19 [18]. The epidemiological landscape has changed with the emergence of new SARS-CoV-2 variants of concern (VOC) [19], and population immunity through infection and vaccines, licensed in Switzerland since December 2020.
Definitions
We defined an index case as an individual who tested positive for SARS-CoV-2 by a polymerase chain reaction (PCR) test or a positive rapid antigen test after their introduction at the end of 2020 within the testing criteria (adult within four days of symptom onset, non-healthcare worker, non-vulnerable person) with legal residency in the Swiss canton of Solothurn. Index cases were required to isolate for 10 days either from the date of the positive test result or the date of symptom onset. Close contacts were defined as Solothurn residents who had either spent at least 15 minutes with an index case at a distance of less than 1.5 metres without wearing a face mask or were living in the same household as the index case in the two days before testing positive [14].
Contact tracing practices and digital workflow
The CT team collected data from index cases and contacts through in-depth telephone interviews and the collation of self-reported information from online forms. After the first epidemic wave in spring 2020, the public health authorities of the canton of Solothurn implemented a process-oriented CT software “Straatos” at the beginning of the second epidemic wave in October 2020 [20–21] (see supplementary file 1 for details). Index cases received a link via text message to an online form after testing positive (supplementary files 1 and 2). Index cases reported their age, sex, place of residence, symptoms, date of symptom onset, potential source of infection, vaccination status, close contacts, and places visited before symptom onset or positive test if asymptomatic. The CT system then automatically emailed the index case an ‘administrative order’, digitally signed by the cantonal physician, which ordered the individual to isolate, based on the Epidemics Act [17]. The CT staff telephoned index cases within one day of their positive test result to confirm their responses to the online form, whenever workload allowed it (see contingency planning in supplementary file 3). Afterwards, the contact tracer would discuss the information provided with another member of staff. The index cases received an automated text message at the end of isolation on day ten (to 12 January 2022) or on day five (from 13 January 2022) to report any symptoms online. There was a three-step ‘traffic light’ system with levels one (indicating a low caseload and high staff availability) to four (indicating high caseload and low staff availability) to determine the extent to which staff would individually contact the index cases and contacts, check vaccination documents and test orders, and implement CT in institutional and business settings (supplementary file 3). The levels were not objectively defined; usually, the level increased at the beginning of a wave of infection. For example, at levels one and two, the CT staff contacted the index cases on days six and ten. However, at level three they only contacted them on these days when there were sufficient resources and at level four they did not contact them on these days.
Close contacts identified through forward CT received a text message with a link to an online form (supplementary file 2). The text of the message instructed them to quarantine at home and get tested five days after their last contact with the index case. Contacts reported their age, sex, vaccination status, workplace or school, and email address. Contacts were also emailed an ‘administrative order’ [17], which included information on what to do if they developed symptoms, current recommendations by the FOPH, links to the FOPH website, and a telephone number and email address if they needed more information. The contacts were called by telephone in the first two days of quarantine and at the end of the quarantine, depending on the caseload. Contacts received an automated text message on day five of quarantine to remind them to get tested. At the onset of the omicron wave (27 December 2021), the workload was judged to be too high to continue to call contacts at the start of the quarantine period by telephone. From January 2021, contacts could leave quarantine on day 7 if they tested negative for SARS-CoV-2. After this date, contacts were only telephoned at the end of quarantine in exceptional circumstances.
The CT team in Solothurn started backward tracing (in addition to forward tracing) on 15 November 2020. Index cases reported locations and events visited in the 10 days prior to testing positive using an online form (supplementary table 2 and file 1) and provided data were reviewed during the telephone interview. The CT software automatically identified index patients who could belong to a cluster because of attendance at the same event, bar, school, nursing home, or sharing a residential building (supplementary file 1). When staff capacity allowed, the CT team contacted the venues at which index cases had reported visiting in the ten days before infection. Based on this information, a mobile testing team (run by the two cantonal testing centres) was sent out.
We defined 16 KPIs: 2 for the structure, 11 for processes and 3 for outcomes of CT (supplementary table 3) [13]. The KPIs were based on previous research [11, 22] and an unpublished list of COVID CT indicators that had been discussed early in the pandemic [12]. ‘Structure’ indicators relate to the setting, including human resources and equipment in a CT context. They included quantification of the proportion of individuals using the Swiss Covid digital tracing application, used to notify individuals when they had been in close proximity to a positive case [23], and the capacity of the CT workforce over time, measured as the number of full-time equivalent (FTE) staff. “Process” indicators measured the speed and completeness of investigating, testing and CT. “Outcomes” of CT refer to measures that should indicate whether a chain of transmission could have been broken. Most (13) KPIs for COVID-19 CT had associated targets (supplementary table 3) [11–12, 22]. Index cases without a case date (date of first contact with the index case) were removed. Dates that were 25 days before or after the automatically generated case date were removed if we deemed them likely due to a typographical error.
We describe the level of engagement with CT in Solothurn at each stage in a cascade of CT processes. We also describe the characteristics of the index cases and compare them with Swiss national data (from the FOPH). We report their reported source of infection, activities or locations where they spent time in the ten days before testing positive, their number of close contacts and the characteristics of the close contacts. Several SARS-CoV-2 variants were circulating at the same time; we have indicated the dominant viral variant based on national testing data (supplementary file 1). We calculate KPIs by period of dominant viral variant and plot KPIs against the incidence rate of SARS-CoV-2 over the entire period. All analyses were conducted using R (version 3.5.1). The analysis code is available at https://github.com/leonieheron.
In accordance with the Swiss Epidemics Act, informed consent is not required for the collection or processing of personal data in the context of outbreak investigations and containment of infectious diseases. We obtained approval from the Ethics Committee of Northwestern and Central Switzerland (EKNZ, reference nº 2022-00261, www.swissethics.ch) to analyse CT data and publish only anonymised data.
Over sixty thousand (n = 60,378) positive tests were recorded in the canton of Solothurn from 15 November 2020 to 2 February 2022 (figure 1). The CT team contacted almost all positive cases (95%, n = 57,363) by text message through the CT software and, if workload allowed, additionally by phone. The remaining 5% of cases came from nursing homes. Nursing home residents were called individually (via the health service of the nursing home), and these calls were documented outside of the CT system in November and December 2020. A total of 57,360 individuals (all but 3) submitted the online form. The index cases reported 71,809 contacts, all of whom were contacted by the CT team. The majority of contacts (93%, n = 66,934) submitted an online form. Only 66,261 of the 71,809 (92%) contacts reported by the index cases completed an online form. 5548 contacts were not linked to any index cases, most of them came from outside the canton (n = 4481). The other contacts (n = 1167) were either referred by the FOPH because they were on a flight with a confirmed infection aboard or the reason was unknown. Nine percent (n = 5830) of the contacts who completed the online form later tested positive themselves and became index cases.
Figure 1Contact tracing cascade in Solothurn from 15 November 2020 to 2 February 2022 (SARS-CoV-2 index cases and contacts). Boxes are sized in proportion to the number of individuals at each stage, except for the final box which is as narrow as the text allows.
Overall, the distribution of index cases according to age and sex was similar to that of the general population in Switzerland (table 1, supplementary table 1). The proportion of children (aged 0–17 years) in CT increased from 7% (n = 503) before the periods of VOCs to 30% and 27%, respectively, when delta and omicron were the dominant VOCs. The increasing proportion of children in CT partly reflects relaxations in COVID-19 measures in Swiss schools, which were replaced with the repeated testing of students in June 2021, and most children were not vaccinated, meaning that they were not exempt from quarantine. The symptoms most commonly reported by the index cases were cough (46%), runny nose (39%) and sore throat (34%) although the distribution of symptoms changed as new viral variants emerged (table 1). Loss of taste or smell was more common in the period before the alpha VOC emerged (23%) compared with 12% overall. In the period during which alpha was the dominant VOC, relatively more people reported no symptoms (27% compared with 18% overall). During the period when delta was the dominant VOC, more people reported a cough or a runny nose (49% and 43% compared with 46% and 39% overall). Finally, while omicron was the dominant VOC, a sore throat was relatively more common (39% compared with 34% overall).
The proportion of vaccinated individuals increased over time. By the omicron period, 48% of all index cases had been vaccinated with at least two doses. While only 0.1% (n = 65) and 1.3% (n = 737) of the age and gender data were missing, respectively, 7.1% (n = 4062) of the data on symptoms were missing. If participants did not answer the question on vaccination status, we assumed that they had not been vaccinated.
Table 1Characteristics of index cases referred to contact tracing in Solothurn from 15 November 2020 to 2 February 2022 by periods of different viral dominance.
Variable | Level | Pre-VOC | Alpha | Delta | Omicron | Entire period |
15 Nov 20 – 7 Feb 21* | 8 Feb 21 – 27 Jun 21 | 28 Jun 21 – 26 Dec 21 | 27 Dec 21 – 2 Feb 22 | 15 Nov 20 – 2 Feb 22 | ||
Total | 7494 | 4964 | 15,609 | 29,296 | 57,363 | |
Age, years | 0–17 | 503 (7%) | 838 (17%) | 4626 (30%) | 7856 (27%) | 13,823 (24%) |
18–29 | 1547 (21%) | 1000 (20%) | 2745 (18%) | 6070 (21%) | 11,362 (20%) | |
30–39 | 1292 (17%) | 879 (18%) | 2582 (17%) | 5355 (18%) | 10,108 (18%) | |
40–49 | 1116 (15%) | 771 (16%) | 2233 (14%) | 4342 (15%) | 8462 (14%) | |
50–59 | 1385 (18%) | 749 (15%) | 1674 (11%) | 3325 (11%) | 7133 (12%) | |
60–69 | 914 (12%) | 440 (9%) | 961 (6%) | 1524 (5%) | 3839 (7%) | |
70–79 | 463 (6%) | 189 (4%) | 478 (3%) | 492 (2%) | 1622 (3%) | |
≥80 | 271 (4%) | 91 (2%) | 300 (2%) | 287 (1%) | 949 (2%) | |
Gender | Female | 3495 (51%) | 2373 (48%) | 7695 (49%) | 13,748 (47%) | 27,311 (48%) |
Male | 3295 (49%) | 2572 (52%) | 7846 (50%) | 14,085 (48%) | 27,798 (49%) | |
Other/unknown | 0 (0%) | 5 (0%) | 57 (0%) | 1455 (5%) | 1517 (3%) | |
Symptoms** | Cough | 3263 (44%) | 2082 (42%) | 7632 (49%) | 13,175 (45%) | 26,152 (46%) |
Runny nose | 2701 (37%) | 1571 (32%) | 6662 (43%) | 11,499 (39%) | 22,433 (39%) | |
Sore throat | 2088 (28%) | 1365 (28%) | 4714 (30%) | 11,360 (39%) | 19,527 (34%) | |
Fever | 2253 (31%) | 1369 (28%) | 5271 (34%) | 9714 (33%) | 18,607 (33%) | |
Sweating | 1914 (26%) | 1271 (26%) | 4308 (28%) | 8751 (30%) | 16,244 (28%) | |
General malaise | 1422 (19%) | 804 (16%) | 2721 (17%) | 5276 (18%) | 10,223 (18%) | |
Other symptoms | 1799 (24%) | 1050 (21%) | 2737 (18%) | 4424 (15%) | 10,010 (18%) | |
Loss of taste or smell | 1698 (23%) | 632 (13%) | 2673 (17%) | 1988 (7%) | 6991 (12%) | |
Nausea | 765 (10%) | 493 (10%) | 1736 (11%) | 3335 (11%) | 6329 (11%) | |
Diarrhoea | 692 (9%) | 416 (8%) | 1283 (8%) | 2146 (7%) | 4537 (8%) | |
Difficulty breathing or shortness of breath | 570 (8%) | 316 (6%) | 902 (6%) | 1586 (5%) | 3374 (6%) | |
Elevated heart rate | 200 (3%) | 83 (2%) | 285 (2%) | 491 (2%) | 1059 (2%) | |
Pneumonia | 36 (0%) | 17 (0%) | 89 (1%) | 34 (0%) | 176 (0%) | |
Acute respiratory distress | 40 (1%) | 8 (0%) | 48 (0%) | 40 (0%) | 136 (0%) | |
Oxygen required | 4 (0%) | 7 (0%) | 6 (0%) | 0 (0%) | 17 (0%) | |
Respiratory failure | 6 (0%) | 3 (0%) | 1 (0%) | 0 (0%) | 10 (0%) | |
No symptoms reported | 1239 (17%) | 1313 (27%) | 3257 (21%) | 3800 (15%) | 9609 (18%) | |
Vaccination status*** | Vaccinated (2 doses) | 9 (0%) | 233 (5%) | 4246 (27%) | 14,143 (48%) | 18,631 (33%) |
* See supplementary material for definition of periods of different viral variants.
** Note that some people experienced more than one symptom and so the percentages for symptoms reported do not add to 100%.
*** Cases answered the question “Are you fully vaccinated?”. Those who replied “Yes” were considered to be vaccinated with at least two doses.
VOC: variants of concern
Eleven thousand (n = 11,072, 19%) index cases reported at least one group activity in the ten days before symptom onset or positive test (table 2). Most cases (n = 46,288; 81%) did not report any activities. Of those who did, most reported only one activity (67%) although the maximum was 23. The most common activities reported in the last 10 days were being at a fitness centre or doing sports, private parties, working, or dining with others at home or in a restaurant. During the period when omicron was the dominant VOC, a much higher proportion of people did not report any information on activities. Out of the index cases who reported attending a certain location, 30% reported that they knew of a positive case who had also attended. The proportion increased as the pandemic continued: in 13% (n = 430) of venues visited and reported by index cases, the index case knew of a positive case in attendance during the pre-VOC period, 8% (n = 284) in the period during which alpha was the dominant VOC, 34% (1139) while the delta VOC was dominant, and 45% (n = 1516) while the omicron VOC was dominant. Most index cases reported their likely source of infection (91%, n = 52,350). The most common responses were family or friends (39%), school (13%), work (9%), or shopping/public transport (5%). Fourteen percent of respondents could not classify their source of infection and in one-fifth the information was missing.
Table 2Activities reported in the ten days prior to a positive SARS-CoV-2 test by individuals in Solothurn from 15 November 2020 to 2 February 2022 by time period of different viral dominance.
Activity* | Pre-VOC | Alpha | Delta | Omicron | Total |
Fitness centre / sports | 792 (4%) | 1005 (8%) | 2612 (7%) | 2203 (5%) | 6612 (6%) |
Private party | 1491 (7%) | 442 (3%) | 1677 (5%) | 2213 (5%) | 5823 (5%) |
Work | 1743 (8%) | 956 (7%) | 1705 (5%) | 1339 (3%) | 5743 (5%) |
Dining with friends (home) | 1494 (7%) | 679 (5%) | 1257 (3%) | 2159 (5%) | 5589 (5%) |
Dining at a restaurant | 817 (4%) | 87 (1%) | 1076 (3%) | 1291 (3%) | 3271 (3%) |
Bar / club | 113 (1%) | 1 (0%) | 591 (2%) | 1027 (2%) | 1732 (1%) |
School | 193 (1%) | 199 (2%) | 394 (1%) | 253 (1%) | 1039 (1%) |
Sporting event (spectator) | 3 (0%) | 21 (0%) | 366 (1%) | 386 (1%) | 776 (1%) |
Choir/singing | 10 (0%) | 15 (0%) | 348 (1%) | 71 (0%) | 444 (0%) |
Foreign travel | 30 (0%) | 8 (0%) | 125 (0%) | 0 (0%) | 163 (0%) |
Public event | 3 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 3 (0%) |
Other | 3220 (15%) | 2731 (21%) | 8343 (23%) | 2611 (6%) | 16,905 (14%) |
Missing | 11,711 (54%) | 7065 (54%) | 18,212 (50%) | 33,806 (71%) | 70,794 (60%) |
Total | 21,620 (100%) | 13,209 (100%) | 36,706 (100%) | 47,359 (100%) | 118,894 (100%) |
* Some index cases reported more than one activity. Most index cases did not report any activities (n = 46,288; 81%).
VOC: variants of concern
Self-reported source of infection correlated with the types of activities reported in the ten days prior to testing positive (supplementary table 2).
Figure 2 shows results for four main KPIs, two for process (P6, percentage of test results returned on same or next day; P8, percentage of contacts notified on same or next day after interview with index) and two for outcomes (O1, percentage of new cases who were already in quarantine by time of positive test, i.e. new cases who were previously identified as contacts; O2, percentage of index cases in isolation by day of testing). The missing data for each KPI are outlined in supplementary table 5. Less than 1% of the data were missing for almost all KPIs, except for structure KPI S1 and process KPI P5 (7% and 5% missing data, respectively).
Figure 2Process and outcome key performance indicators of contact tracing during the COVID-19 pandemic in the canton of Solothurn.The coloured solid lines indicate the percentage of the key performance indicators over time. The thicker black line shows the change in SARS-CoV2 caseload. The Greek symbols α, δ and ο indicate the periods during which the alpha, delta and omicron SARS-CoV-2 variants of concern (VOCs), respectively, were dominant. The dashed line indicates the point at which the system contacted individuals automatically (1 January 2022).
Structure indicators
The CT workforce changed over time in response to the changing incidence rate of SARS-CoV-2 infection. The full-time equivalents of contact tracers increased from 35.4 to 36.3, decreased to 34.2 while delta was the dominant VOC, and then increased to 42.1 during the omicron wave (structure KPI S2, figure 3, supplementary table 4). Less than a third (28%) of index cases reported that they had installed the Swiss Covid app on their smartphone (structure KPI S1, supplementary table 4).
Figure 3SARS-CoV-2 infections and the contact tracing workforce in the canton of Solothurn from 15 November 2020 to 2 February 2022, inclusive. The graph is shaded red according to the SARS-CoV-2 caseload. The dashed line indicates the contact-tracing workforce in full-time equivalent (FTE) and the solid line indicates the cases per workforce. The Greek symbols α, δ and ο indicate the periods during which the alpha, delta and omicron SARS-CoV-2 variants of concern (VOCs), respectively, were dominant.
Process indicators
The process KPIs indicated that performance changed over the course of the pandemic. During the first waves of the pandemic and the period during which the alpha VOC was dominant, almost all test results were returned on the same or next day (process KPI P6, figure 2, supplementary table 4). The time taken to return PCR test results increased while delta was the dominant VOC and only partially recovered afterwards (process KPI P2, supplementary table 4 and supplementary figure 1B). Similarly, almost three-quarters of contacts were notified within 24 hours of the CT interview by phone with the index before the emergence of viral variants (process KPI P8, figure 2, supplementary table 4). While the alpha, delta and omicron VOCs were dominant, the proportions decreased to 64%, 36% and 54%. The days taken between the lab processing the result and the subsequent isolation of the index case also appeared to increase while delta was the dominant VOC (process KPI P4, supplementary figure 1D).
Nevertheless, the CT team successfully contacted almost all positive cases within 24 hours (99%) throughout the pandemic (process KPI P7, supplementary table 4). On average, at least 70% of new cases were tested and interviewed within 3 days of symptom onset across the entire period (process KPI P10). The contacts began quarantine a median of 3 days after the index case began experiencing symptoms (process KPI P3, supplementary table 4, supplementary figure 1C). This time period includes the time it takes to test the index case after symptom onset, the time to receive the results, and the time for CT and quarantine orders to be sent to contacts.
Index cases reported between 0 and 69 close contacts, although most did not report any (52%, n=29,754, process KPI P11). Two-thirds (65%) of index cases did not report any contacts in the pre-VOC period and the proportion steadily decreased to 40% in the omicron period (process KPI P9, figure 4). Of those who did report contacts, the median number (interquartile range [IQR]) of contacts for each period were 2 (1–4) for the pre-VOC periods and while alpha was the dominant VOC and 2 (1–3) for the periods during which delta and omicron were the dominant VOCs.
Figure 4Number of contacts reported by SARS-CoV-2 index cases in Solothurn from 15 November 2020 to 2 February 2022. The horizontal bars indicate the median, the box indicates the interquartile range and the dashed lines indicate the minimum and maximum values, with outliers removed.
Outcome indicators
Overall, the percentage of index cases in isolation by day of testing remained at almost 100% throughout the period of analysis (outcome KPI O2, figure 2). However, the percentage of new cases already in quarantine by time of positive test, i.e. new cases who had previously been identified as contacts (outcome KPI O1) indicated that the CT system may have become less efficient as the pandemic continued. In the pre-VOC period and while alpha was the dominant VOC, about a fifth of index cases were already in quarantine by the time of the positive test (outcome KPI O1, figure 2, supplementary table 4). This percentage was highest (over 30% weekly) in February–March 2021 and lowest (<5% weekly) during May–August 2021. In the following waves, only 13% and 5% of index cases were already in quarantine (outcome KPI O1, supplementary table 4).
Nine percent of contacts tested positive for SARS-CoV-2 overall (outcome KPI O3, supplementary table 4). The proportion of contacts who tested positive for SARS-CoV-2 was highest during the period when alpha was the dominant VOC (13%) and lowest in the omicron wave (6%). From 31 May 2022 contacts who had been vaccinated or had tested positive in the 10–180 days before contact were exempt from quarantine and not under any obligation to get tested.
The CT system in Solothurn used a partially automated workflow, which likely broke some chains of SARS-CoV-2 infection between 15 November 2020 and 2 February 2022, based on the high percentage of test results returned on the same or next day (KPI P6), the high percentage of contacts notified within 24 hours of the CT interview with the index (KPI P8), the high percentage of index cases in isolation by day of testing (KPI O2) and other indicators.
The two main process KPIs indicated that the system was faster during the earlier stages of the pandemic: more than 95% and 97% of test results were returned on the same or the next day during the period before the VOCs and while alpha was the dominant VOC (process KPI P6). However, during the periods when delta and omicron were the dominant VOCs, the percentages dropped to 41% and 52%, highlighting the burden placed on the overall system (testing and CT) during periods of very high caseloads. Furthermore, a higher proportion of contacts was notified on the same day or next day after the interview with the index (process KPI P8) during the pre-VOC and alpha phases compared with the delta and omicron phases.
Interpretation of CT performance takes into consideration that the targets were proposed early in the COVID-19 pandemic and may not have been validated. Of the 16 KPIs that we assessed, 13 had at least one suggested target in the literature (supplementary table 3) [11–12, 22]. Five KPIs achieved the suggested targets overall and during each of the periods of different viral variants (process KPIs P1–3, P5 and P7; supplementary table 4, supplementary figure 1) and two had mixed results during the epidemic waves (process KPIs P4 and P6). The process KPIs P1–3 and P5 quantified the time taken between symptom onset, testing, lab results, isolation of the index and quarantine of the contact: all remained within the suggested targets of less than or equal to between one and five days. Process KPIs P4 and P6 were the only KPIs that varied widely from the suggested targets of ≤1 day and 90–100%, respectively. The median number of days between index test results and isolation was over one overall but remained at the target of one day for three out of four periods (process KPI P4).
Some KPIs that did not meet suggested targets indicate challenges at the testing laboratory and the CT workforce when delta and omicron were the dominant VOCs. Six KPIs were below the targets, according to the limited literature (structure KPI S2, process KPIs P8–10 and outcome KPIs O1 and O3). The caseload changed considerably during this period, so it was not clear what caused the test results to be delivered more slowly (process KPI P6). The number of people working in the CT team was below the suggested target of 30 tracers per 100,000 population, which corresponds to 64 FTE in Solothurn [11] (structure KPI S2). However, the number of tracers needed is debated and in addition to caseload depends on the degree of CT automation and team organisation, the level of personal contact required for interviews and telephone calls, and the availability of office space and funding. Less than 80% of contacts were notified on the same day or day after the interview with the index (process KPI P8), but the proportions were higher in the pre-VOC and alpha phases. More than 25% of cases did not report any contacts (process KPI P9), most likely due to the rising reluctance of the population to report contacts and the high volume of cases that restricted the ability of the CT team to contact index cases to verify information. Ultimately, it can be argued that above a certain number of cases per day and population, and particularly in the context of highly transmissible SARS-CoV-2 VOCs such as omicron, forward CT is no longer feasible. This is supported by a recent genomic analysis of sequences obtained in 2020, which found that CT likely slowed transmission during the summer of 2020, when there were few cases, but not during the second wave in autumn/winter 2020-2021, when there was a very high number of cases [24]. Therefore, public health experts support focusing on cluster identification and outbreak investigations when cases are rising [3–4].
The outcome indicators are closest to measuring the containment of COVID-19 by CT activities. The main outcome KPI O2 which indicated the percentage of index cases in isolation by day of testing remained high throughout the pandemic (almost 100% in all time periods), whereas the proportion of new cases who were already in quarantine by the time of the positive test (outcome KPI O1) was relatively low (range 5–22%). The target of 80% has been suggested [12], but not validated. We observed that less than 80% of new cases were already in quarantine by the time of the positive test. It is feasible that many residents would leave the canton on an almost daily basis for work or education. Contacts who were resident in other regions of Switzerland were under the jurisdiction of a separate canton and thus it was not possible to link many index cases and contacts. Therefore, we expect that the percentage may be misleadingly low because the data were not linked with other regions in Switzerland. Furthermore, according to the suggested target, we observed that 9% of contacts overall tested positive for SARS-CoV-2, higher than a target proportion of <1% (outcome KPI O3). However, this target is not validated and may result in a larger number of contacts being quarantined.
Contact tracers in other countries reported similar findings to ours. Contact tracers in the United States and Spain found that the performance of CT was worse while caseloads were higher [25–26]. The US team cited overwhelmed staff, unable to conduct thorough interviews due to time pressure, as a possible reason [25]. They also found that different viral dynamics in new variants complicated CT because of increased transmissibility and possibly shorter incubation period [25]. The Catalonian team responded by increasing their workforce [26]. They promote the use of constant monitoring using KPIs to allow for regular evaluation of the CT system and the epidemiological situation [26]. In another US study, contact tracers in New York were able to use their data to confirm that there was more SARS-CoV-2 transmission at known places of interest in the city [27]. Some reports on CT highlighted areas where we might have improved data collection. For example, the US team collected objective data on test results from contacts to monitor the contacts’ outcomes: the prevalence ratio of SARS-CoV-2 infection was much higher in contacts compared with non-contacts when viral transmission was higher in the community [25]. In Taiwan, they benefitted from centralised digital tools, with a unique nationwide CT platform, linking various data sources, including information from telephone companies, and using a smartphone-based real-time locating system to track contacts [7]. The CT team reported a subsequent increase in self-reported updates of health status from 22.5% to 61.5% via automatic text message or web applications in Taiwan during the COVID-19 pandemic (7), reducing the pressure on the CT workforce.
Our study provides an overview of a new digitised CT system established in an urgent and ever-changing real-life situation and proposes a set of important KPIs for evaluation purposes. To our knowledge, we present the first analysis to transparently report the performance of a CT workflow in Switzerland. The strengths of this analysis were the availability of CT data over a long period of 15 months, nearly the entire period of CT in the canton of Solothurn. The dataset includes the CT workforce data and covers periods of different viral VOCs. However, the study is limited by data that are self-reported, missing or not requested. For example, we noticed that people became less willing to disclose information on infection sources and did not report activities as readily as the pandemic progressed. In contrast, index cases reported more close contacts in the later stages, but that may be because most of the close contacts had recovered or had been vaccinated and were therefore no longer ‘at risk’ of being quarantined (vaccinated and recovered persons could be exempted from quarantine from 31 May 2021). Reporting morale likely decreased and many index cases decided to circumvent the system by notifying friends and family themselves. A further limitation is that the CT system only covers the canton of Solothurn, which may not reflect other cantons and other CT workflows. However, the canton of Solothurn is a mid-sized canton, with demographics representative of Switzerland. We did not have access to data from other Swiss cantons for comparison.
CT is a useful tool for reducing the transmission of infectious diseases when the incidence rate is below a manageable threshold. This description of CT in a Swiss canton illustrates how partial automation of CT contributed to real-time monitoring and surveillance of SARS-CoV-2 throughout the pandemic. The findings support the use and monitoring of CT data to inform modelling studies and public health measures in real time. Sudden exponential increases in caseloads challenged the system although automation ensured that CT could continue. Our study also shows the need for standardised benchmarks across CT to facilitate cross-region and cross-border cooperation, taking into account the resource level of the country, which will require an electronic data capture system to allow real-time data extraction. Adding genomics data to the CT workflow might further support and advance identification of transmission clusters [15, 24, 28]. Digitised CT workflows can be used for other existing or re-emerging infectious diseases such as measles or mpox, but also for emerging infections in the years to come [28]. In the context of pandemic preparedness, we believe that periods of relatively low transmission provide a good opportunity to analyse and compare CT performance, to improve the systems and to reach a consensus on targets. Modelling studies could help public health authorities to understand where to focus attention to maximise interruption of transmission. Efficient CT systems should automate workflows as far as possible to adapt to high incidence rates, save on human resources, and generate KPIs to monitor the system’s performance.
We would like to thank all persons who contributed data to this study. We are also indebted to the cantonal physician’s office of the canton of Solothurn and the contact tracing team for their diligent daily work and the immense efforts to continuously improve the CT workflow.Authors' contributions: Conception and design: LH, CM, NL, LF. Contact tracing data collection: CM, EB, BD, LF. Statistical analysis: LH, CM. Wrote the first draft of the paper and revised it based on comments from all authors: LH, CM, NL, LF. All authors reviewed and approved the final version of the manuscript.
LF is supported by the National Institute of Allergy and Infectious Diseases (NIAID) through grant no. 5U01-AI069924-05. LH was supported by the Swiss National Science Foundation (grant no. 176233) and the European Union Horizon 2020 research and innovation programme (grant no. 101003688).
All authors have completed and submitted the International Committee of Medical Journal Editors form for disclosure of potential conflicts of interest. No potential conflict of interest related to the content of this manuscript was disclosed.
1. Aleta A, Martín-Corral D, Pastore Y Piontti A, Ajelli M, Litvinova M, Chinazzi M, et al. Modelling the impact of testing, contact tracing and household quarantine on second waves of COVID-19. Nat Hum Behav. 2020 Sep;4:964–71. 10.1038/s41562-020-0931-9
2. Kucharski AJ, Klepac P, Conlan AJ, Kissler SM, Tang ML, Fry H, et al.; CMMID COVID-19 working group. Effectiveness of isolation, testing, contact tracing, and physical distancing on reducing transmission of SARS-CoV-2 in different settings: a mathematical modelling study. Lancet Infect Dis. 2020 Oct;20(10):1151–60. 10.1016/S1473-3099(20)30457-6
3. Kojaku S, Hébert-Dufresne L, Mones E, Lehmann S, Ahn YY. The effectiveness of backward contact tracing in networks. Nat Phys. 2021 May;17:652–8. 10.1038/s41567-021-01187-2
4. Raymenants J, Geenen C, Thibaut J, Nelissen K, Gorissen S, Andre E. Empirical evidence on the efficiency of backward contact tracing in COVID-19. Nat Commun. 2022 Aug;13(1):4750. 10.1038/s41467-022-32531-6
5. Ashcroft P, Lehtinen S, Bonhoeffer S. Test-trace-isolate-quarantine (TTIQ) intervention strategies after symptomatic COVID-19 case identification. PLoS One. 2022 Feb;17(2):e0263597. 10.1371/journal.pone.0263597
6. Koetter P, Pelton M, Gonzalo J, Du P, Exten C, Bogale K, et al. Implementation and Process of a COVID-19 Contact Tracing Initiative: Leveraging Health Professional Students to Extend the Workforce During a Pandemic. Am J Infect Control. 2020 Dec;48(12):1451–6. 10.1016/j.ajic.2020.08.012
7. Jian SW, Cheng HY, Huang XT, Liu DP. Contact tracing with digital assistance in Taiwan’s COVID-19 outbreak response. Int J Infect Dis. 2020 Dec;101:348–52. 10.1016/j.ijid.2020.09.1483
8. Braithwaite I, Callender T, Bullock M, Aldridge RW. Automated and partly automated contact tracing: a systematic review to inform the control of COVID-19. Lancet Digit Health. 2020 Nov;2(11):e607–21. 10.1016/S2589-7500(20)30184-9
9. World Health Organization. Digital tools for COVID-19 contact tracing: annex: contact tracing in the context of COVID-19, 2 June 2020 [Internet]. World Health Organization; 2020 [cited 14 December 2022]. Available from: https://apps.who.int/iris/handle/10665/332265
10. Daniore P, Ballouz T, Menges D, von Wyl V. The SwissCovid Digital Proximity Tracing App after one year: were expectations fulfilled? Swiss Med Wkly. 2021 Sep;151(35-36):w30031. 10.4414/SMW.2021.w30031
11. Vogt F, Kurup KK, Mussleman P, Habrun C, Crowe M, Woodward A, et al. Contact tracing indicators for COVID-19: rapid scoping review and conceptual framework. PLoS One. 2022 Feb;17(2):e0264433. 10.1371/journal.pone.0264433
12. Frieden T. COVID Contact Tracing Indicator list for consideration [Unpublished work]. 2020.
13. Donabedian A. The quality of care. How can it be assessed? JAMA. 1988 Sep;260(12):1743–8. 10.1001/jama.1988.03410120089033
14. Federal Office of Public Health FOPH. Coronavirus: Measures and ordinances [Internet]. Federal Office of Public Health FOPH; 2022 [cited September 20 2022]. Available from: https://www.bag.admin.ch/bag/en/home/krankheiten/ausbrueche-epidemien-pandemien/aktuelle-ausbrueche-epidemien/novel-cov/massnahmen-des-bundes.html
15. Anderegg N, Schwab T, Borcard L, Mugglin C, Keune-Dübi B, Ramette A, et al. Population-based SARS-CoV-2 whole genome sequencing and contact tracing during the COVID-19 pandemic in Switzerland. J Infect Dis. 2023;228(3):251-260.
16. Kanton Solothurn. Bevölkerungszahlen [Internet]. Kanton Solothurn; 2022 [cited 20 September 2022]. Available from: https://so.ch/verwaltung/finanzdepartement/amt-fuer-finanzen/statistikportal/bevoelkerung/bevoelkerungszahlen/
17. Federal Office of Public Health FOPH. Communicable Diseases Legislation – Epidemics Act, (EpidA) [Internet]. Bern: Federal Office of Public Health; 2020 [cited 13 December 2021]. Available from: https://www.bag.admin.ch/bag/en/home/gesetze-und-bewilligungen/gesetzgebung/gesetzgebung-mensch-gesundheit/epidemiengesetz.html
18. Federal Office of Public Health FOPH. COVID-19 Switzerland [Internet]. Swiss Confederation; 2022 [cited 21 October 2022]. Available from: https://www.covid19.admin.ch/en/epidemiologic/
19. World Health Organization. Tracking SARS-CoV-2 variants [Internet]. 2022 [cited 12th August 2022]. Available from: https://www.who.int/activities/tracking-SARS-CoV-2-variants/
20. CumulusPro. Straatos. CumulusPro; 2022.
21. GmbH SP. Menschen. Prozesse. Digitalisierung. . Lenzburg: Strub & Partner GmbH; 2022.
22. Resolve to Save Lives. Covid-19 Contact Tracing Playbook [Internet]. 2022 [cited 11 August 2022]. Available from: https://contacttracingplaybook.resolvetosavelives.org/checklists/metrics
23. Federal Office of Public Health FOPH. Coronavirus: SwissCovid app [Internet]. 2023. Available from: https://www.bag.admin.ch/bag/en/home/krankheiten/ausbrueche-epidemien-pandemien/aktuelle-ausbrueche-epidemien/novel-cov/swisscovid-app-und-contact-tracing.html
24. Nadeau SA, Vaughan TG, Beckmann C, Topolsky I, Chen C, Hodcroft E, et al. Swiss public health measures associated with reduced SARS-CoV-2 transmission using genome data. Sci Transl Med. 2023;15(680):eabn7979. 10.1126/scitranslmed.abn7979
25. Borah BF, Pringle J, Flaherty M, Oeltmann JE, Moonan PK, Kelso P. High Community Transmission of SARS-CoV-2 Associated with Decreased Contact Tracing Effectiveness for Identifying Persons at Elevated Risk of Infection – Vermont. Clin Infect Dis. 2022;75(S2):S334-7.
26. Herrero M, Ciruela P, Mallafré-Larrosa M, Mendoza S, Patsi-Bosch G, Martínez-Solanas È, et al.; Epidemiological Surveillance Network of Catalonia. SARS-CoV-2 Catalonia contact tracing program: evaluation of key performance indicators. BMC Public Health. 2022 Jul;22(1):1397. 10.1186/s12889-022-13695-8
27. Pei S, Kandula S, Cascante Vega J, Yang W, Foerster S, Thompson C, et al. Contact tracing reveals community transmission of COVID-19 in New York City. Nat Commun. 2022 Oct;13(1):6307. 10.1038/s41467-022-34130-x
28. Walker A, Houwaart T, Finzer P, Ehlkes L, Tyshaieva A, Damagnez M, et al.; German COVID-19 OMICS Initiative (DeCOI). Characterization of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection Clusters Based on Integrated Genomic Surveillance, Outbreak Analysis and Contact Tracing in an Urban Setting. Clin Infect Dis. 2022 Mar;74(6):1039–46. 10.1093/cid/ciab588
The appendix is available in the pdf version of the article at https://doi.org/10.57187/smw.2023.40112.