To our knowledge, this study is the first to present a systematic framework of key performance indicators to assess contact tracing performance with actual country data, as soon as the contact tracing program was designed and gradually implemented. As major agencies providing advice on COVID-19 have updated their contact tracing recommendations [4, 26], our study provides an opportunity to critically evaluate our rating system. As the Centers for Disease Control and Prevention emphasizes, the use and evaluation of key indicators in contact tracing programs for COVID-19 is important because they help detect areas for improvement and put in place implement changes to strengthen the program.
Despite the increase in new cases of SARS-CoV-2 during our study period (particularly from May to October 2020, before the start of the third wave), the four KPIs included in our contact tracing program increased significantly over time, closely approaching the set targets. Our goals have been set so that contact tracing plays a key role in controlling the pandemic, explaining its ambitious goals. For example, linked to KPI1 (percentage of cases with identified contacts), its target was set at 80% instead of 75% as proposed by the Harvard TTSI strategy. . Localized clusters of cases and outbreaks can put a strain on limited human resources and therefore reduce the system’s ability to complete case investigation and contact identification in a timely manner, as we have observed, through example, in Lleida in July and August 2020, a region that suffered severe COVID-19 outbreaks explaining its low KPI1 values (Table S2). In addition, our targets have been rationalized according to the socio-demographic characteristics of Catalonia. For example, Laxminarayan et al.  describes an average of 7.3 CC per case in two Indian states (KPI2), whereas in our context 4.6 was the average over the study period, close to other European parameters, such as 2.4 CC per case in Italy .
The usefulness of stratifying the results according to the exposure framework of household, social and non-family/non-social contacts is similar to that of other national strategies such as the UK [29, 30]. The dominance of the home environment in our study is likely related to the nature of the restrictions imposed by COVID-19 on mobility and social distancing. The domestic setting, which is more predominant among CCs aged 40 to 59, could be explained by the habits of the Spanish population and increasing telework since the pandemic. Social background was more common among CCs aged 20-29, an age group prone to outdoor activities and gatherings . This finding illustrates the need to tailor communication strategies to target audiences, as evidence suggests that the effectiveness of the response depends in part on how communities receive, perceive and act on information provided by governments and organizations. other agencies. . In addition, KPI2 has proven to be very handy in identifying potential super-diffusion events and has helped in the early investigation of clusters and the detection of high-risk contacts, towards which the epidemiologist can take countermeasures. additional controls, such as massive and urgent screening of the population. .
In KPI3, we assessed the percentage of new contacts traced and quarantined by contact tracers, identifying the need for government support (regarding social or housing support for quarantine compliance). After the summer easing of restrictions, the increase in the number of cases from July to October (from 607 cases/day in July to 2943 cases/day in October) put the system under pressure, but the system responded by increasing the number of investigators. KPI3 improved to reach the planned target of 70% (72.7%) in December (Table S1).
Regarding KPI4, the proportion of CCs developing symptoms of COVID-19 in May-July was 10.7% on average compared to August-December 8.2%, because in May-July only cases symptomatic were tested and their CC questioned. To assess the secondary attack rate of SARS-CoV-2, we monitored the variable collected during case investigation to find out whether new cases arise among contacts previously identified by the program. We observed a significant increase over time in this indicator (from 33.9% in May to 57.9% in November), reflecting the effectiveness of the program in interrupting the chains of transmission of SARS-CoV-2 . The universal CC test implemented in early August has probably helped improve this indicator, helping to detect asymptomatic cases.
Prompt testing, case investigation, and contact elicitation and tracing are critical to successfully interrupting community transmission. Since completeness data is largely covered by the KPI framework used in our program, in order to further improve CT monitoring, we propose to extend it with timeline-related KPIs as shown in the adapted Table 3. guidelines from international authors. [26, 34]. The new framework offers a set of new process indicators prioritized by case incidence. After this study, we reviewed the indicators used in 2020 and in 2021 it was proposed to add a new indicator framework (Table 3). During a period of high incidence (> 3,000 cases/day): “time between onset of case symptoms and diagnosis”; “time between case diagnosis and contact interview and quarantine”; “CC per case disaggregated by risk category and settings” and “% of CC confirmed as new SARS-CoV-2 infection”. In a time of “low – moderate risk transmission”, in addition to the main KPIs and the previous new indicators, four additional ones are proposed: percentage of cases without CC to report”; “percentage of CCs tested for SARS-CoV-2”; “percentage of monitored CCs respecting quarantine measures” and “percentage of vaccinated CCs”, vaccination in Spain started on December 27, 2020.
According to Kretzschmar et al. and other international public health agencies  CT will only be effective if the time from onset of case symptoms to receipt of test results is less than 3 days, and time from receipt of test results to contact quarantine is less than 1 day. Additionally, we suggest two scenarios in the evaluation of KPIs based on the weekly incidence of SARS-CoV-2.
Measuring process indicators pertaining to “time from onset of case symptoms to diagnosis, and interview with case and quarantine of contacts” will help identify bottlenecks in the system and to carry out appropriate measurements in each geographical zone according to the zone in which the delay is identified. Due to our two separate COVID-19 databases, storing case or CC information hampered its calculation, but at the time of writing this manuscript it was a merger of the two databases and these measures have been incorporated into our framework.
The centralization of contact management via a single platform allowing diligent evaluation of KPIs is fundamental . A special TB monitoring and evaluation team has been set up within PHAC to assess KPIs daily and issue weekly reports with overtime trends. A fire code was established for the four indicators presented and, under predefined thresholds, the territories concerned were closely monitored in order to adjust the control strategy, either at regional or national level.
Additionally, scaling the CT workforce with ongoing training and support, as well as integrating it with primary care providers and PHAC, has been a cornerstone of the program. Towards the recommended minimum of 30 tracers per 100,000 inhabitants, in an emergency situation our workforce has gradually increased at all levels (primary care, hospital structures and epidemiological surveillance units).
Several limitations related to information systems hinder the systematic evaluation of TCs and KPIs in our context. First, the existence of decentralized data systems for certain professional settings and nurseries means that these subsets of exposures are partially connected to the MCC platform and therefore underrepresented in our study. Second, the lack of digital scanner integration to supplement manual efforts, as Catalonia has consistently only implemented a mobile app to track the symptoms of confirmed cases (App STOP COVID19 CAT). Finally, to ensure compliance with the “General Data Protection Regulation”, no information that could identify the index case was provided to the contact during the initial notification call. Such a legal constraint meant that the date of last exposure, from which the 14-day quarantine was calculated, could not be disclosed. which interrupted the estimate of quarantine compliance.