Much is expected from the UK’s symptomatic Test and Trace system (sT&T), and all of us must play our part if it is to succeed in unlocking the many, while the infected, their households and close contacts are quarantined for up to 14 days.
In February and March 2020, those of us with symptoms were told to stay home and seek medical help only if symptoms did not improve within 7 days. In the new sT&T era, from June 2020, this has changed: if I experience at least one of three key Covid-19 symptoms – high temperature, new persistent cough, loss or change to sense of smell or taste – I must self-isolate for 7 days and immediately book a swab test, and all my household members must also remain at home.
If my swab tests positive, then my household members should remain in quarantine for 14 days from my date of symptom onset. But, I have an additional task, namely: to identify other close contacts in the two days prior to symptom onset and afterwards, so that they can be traced by the sT&T team on behalf of the National Health Service (NHS) and advised to quarantine themselves (but not their entire household) for 14 days from my date of symptom onset.
Speed is of the essence, here. The time from symptom onset to test booking (same day, day 1) to test result (next day, day 2) to close contacts traced and quarantined (day 3) is tight and may become tighter still as the sT&T team learns what works best for successful interaction with close contacts. Weekly public disclosure of robust information on the delivery of sT&T is critical for maintaining the confidence of the public and medical professionals.
Because sT&T has a centrally co-ordinated and (we trust) well-designed booking system, we should learn efficiently and quickly. Here are 10 likely topics for analysis.
1. Daily number of symptomatic persons who book a swab test (by booking date). For each such person, brief information for analysis might include: booking date, birth year, sex, absence/presence (and onset date) of each of the key symptoms; part-postcode of household, other persons normally resident in the household – number of children, adults aged 18-69 years, adults aged 70+ years; symptomatic person’s main occupation, part-postcode of main workplace.
2. Daily number of symptomatic persons with swab test result, positive or negative or inadequate for testing (by swab date and result date). For each such person, swab date, result date and test result are added to the booking date information already held.
The above quartet of dates (onset date, booking date, swab date, result date), together with part-postcode, enable analysis of waiting times, which may differ geographically and improve over time. Symptom sequence and covariate information, including occupation and geography, may be explanatory factors for being swab-test positive or for delays.
3. Daily number of identifiable close contacts outside of the household of newly swab-positive persons (by trace date). For each such newly swab-positive person, the number of their identifiable close contacts outside of their own household, together with the number of locations (and location type) where unidentifiable close contacts were encountered (approximate number), defines the newly swab-positive person’s close contact network outside of their own household and the estimated proportion of the network that is identifiable.
4. For each newly swab-positive person (by trace date), documentation of each identified outside close contact in terms of traced/not traced and if traced: date contact was made, contact’s birth year, sex, current absence/presence (and onset date) of each of the key symptoms; part-postcode of contact’s household, other persons normally resident in the contact’s household – number of children, adults aged 18-69 years, adults aged 70+ years; contact’s main occupation, part-postcode of main workplace; close contact intends/declines to remain quarantined at home until date (= 14 days after symptom onset date of the newly swab-positive person whose identified close contact this person is).
The number of identifiable close contacts per newly swab-positive person outside of their own household defines the extent of the tracing task and needs regular analysis, as the mean number may increase with each further easing of lockdown.
The higher the proportion is of close contacts who are identifiable, the more effective sT&T can be in quarantining contacts who have had a close encounter with a newly swab-positive-person.
But effectiveness also depends on contact being made by tracers with a high percentage of identified closed contacts, and their agreeing to be quarantined.
Agreeing and abiding by agreement are different: the extent of the difference could be assessed by a randomly timed follow-up to offer swab testing to quarantined individuals who remain asymptomatic/have not booked a test.
Covariate information about the newly symptomatic person, or about his/her close contacts outside of the home, may be explanatory for various of the above effectiveness proportions.
Geographic distance (or convergence) within the network of identifiable close contacts, their home locations or workplace locations, and those of their newly swab-positive contact, are potentially of interest for identifying at-risk locations, occupations or transmission risk across too many households.
Corresponding analyses across a series of networks may be additionally informative about potential occupational risks or about the number of identified households per close-contact network.
5. Importantly, sT&T seeks to document, and glean new intelligence – via household-membership, covariates, occupation and geographical location – about the percentage of identified and traced close contacts of a newly swab-positive case who develop symptoms and belong to a) the case’s own household versus b) other households. This requires analysis both within networks and across networks.
6. Adherence to quarantine restrictions is also pertinent, as above, and can be assessed by offering a randomly timed follow-up to a) one or more randomly selected traced close contacts; and b) randomly selected quarantined individuals in cases’ own household.
7. Super-spreaders among newly swab-positive cases may be identified by the number of their identifiable, traced and quarantined close contacts and/or own household members who develop a key Covid-19 symptom, book a swab test, and are PCR-positive at test result. Explanatory factors which distinguish potential super-spreaders would be valuable intelligence, especially if those explanatory factors identified transmission pathways, occupationally or location wise, that could be counteracted.
8. Occupational information is important not only in the preceding context of super-spreaders but because sT&T destabilizes the testing pillars relied upon hitherto – that is, prior to June 2020. Occupational information should allow people tested by sT&T to be assigned, in effect, to the previous test tiers so that the number of swab tests and PCR-positive rates can continue to be compared weekly (by swab date) within test tier, both during lockdown (April and May) and as restrictions are relaxed.
9. Occupational information together with other covariates may also be useful for explaining why some individuals find themselves repeatedly identified as a close-contact of newly swab-positive cases. Possible explanations range from chance through transmission-prone behaviours, or occupational risk, on the part of the close contact to the possibility that the close contact is an infectious but asymptomatic Covid-19 carrier.
10. In principle, sT&T is kick-started when a swab-test is booked as soon as a person develops at least one of three key Covid-19 symptoms. However, around 40% of onward transmissions may occur pre-symptomatically. Offering a randomly timed swab-test to a random sample of quarantined close contacts of the newly swab-positive index case a) in households other than his/her own and b) in his/her own household would quantify the extra yield of asymptomatic PCR positives, and more timely T&T as a consequence, particularly for those who never developed symptoms within their assigned quarantine period.
Alongside sT&T sits an intended suite of surveillance studies, for example periodic swab-testing of: a) on-duty hospital-staff, b) outpatients, c) inpatients, d) on-duty care home staff, e) care home residents, f) other on-duty social care workers, and g) other at-risk occupations, including those whose contracts mean that they are not paid when quarantined or are obliged to take their quarantine-time as annual leave.
Those individuals who tested PCR-positive would then be referred to sT&T for tracing of their non-occupational identifiable close contacts. Hospitals, care homes or other settings might also institute an outbreak investigation or other actions to minimize potential transmission routes that surveillance had highlighted. Outbreak investigations may invoke the tracing services of sT&T to trace and follow-up identifiable close contacts of symptomatic persons who outside of the immediate outbreak cluster.
Auditing the performance of sT&T
There has been no mention, as yet, of whether symptomatic persons who are hospitalized with Covid-19 could have their sT&T record checked by clinicians to access the contemporaneous sT&T record of the patient’s sequence of key symptoms; and, thereby, also discover if the patient had been already referred to sT&T.
Knowing what percentage of working-age (versus older) patients who are hospitalized with Covid-19, on or after 8 June 2020, had immediately booked a swab-test and pre-registered their onset symptoms with sT&T is an important cross-check.
Office for National Statistics (ONS) Infection Survey
The ONS Infection Survey is another point of reference. The survey does not make explicit what proportion of its respondents in weeks 1–6 (mainly during May) had one of the three key symptoms that the sT&T scheme wishes immediately to be notified about. However, my surmise from the following facts –
- that 12% tested positive of those with at least one key symptoms on their swab date;
- 0.35% tested positive of those who did not have any of the three key symptoms on their swab date; and
- between 0.33% and 0.50% of those swab-tested in weeks 1–6 was positive
– is that, on a typical test date, only 100 per 10,000 persons in the community had at least one key symptom, of whom 12 tested swab-positive, whereas the majority of swab-positives (35) arose from the 9,900 who did not exhibit one of the three key symptoms on their swab date.
If my presumptions are roughly right, then sT&T will at best identify only about one-quarter of those who would indeed be swab-positive in the community. And identifying even this quarter might mean that the scheme had to allow capacity for 14,000 to 20,000 swab tests being booked per day by those experiencing at least one of three key symptoms: but only 12% of swab tests, when analysed, would be positive (1,700 to 2,400).
As the ONS Infection Survey progresses through summer and autumn, these estimates will change but they appear to provide a very useful benchmark for the performance of the Test and Trace scheme based on symptomatology.
About the author
Sheila M. Bird is formerly programme leader at the MRC Biostatistics Unit, Cambridge University. She is a member of the Significance editorial board and the Royal Statistical Society’s Covid-19 Task Force. She writes here in a personal capacity.