Automated Symptom Monitoring for COVID-19 Contacts in Maine, May–June 2020

A report from the first months of the pandemic, posted as an MMWR Early Release on August 3, 2020.

Summary

• Known before: finding the contacts of COVID-19 patients and asking them to quarantine can reduce spread of SARS-CoV-2.
• New here: Maine added automated symptom monitoring to its contact tracing, and it was well received — 96.4% of monitored contacts agreed to it — and it promptly identified new cases. Of 1,622 people enrolled, 190 (11.7%) developed COVID-19.
• What it means: prompt case investigation can quickly identify contacts and recommend quarantine, reducing further exposure. Automated tools offered in several languages and formats may improve contact tracing and reduce the need for resources such as staff.

Why Maine automated

Quarantining exposed people (contacts) for 14 days after exposure reduces transmission, and contact tracing lets health departments tell contacts about quarantine and watch their symptoms to catch new cases early. Maine identified its first case on March 12, 2020, but staffing limits meant the Maine Center for Disease Control and Prevention (Maine CDC) could not monitor contacts consistently. On May 14, 2020, it began enrolling contacts in Sara Alert (MITRE Corporation), an automated, web-based symptom monitoring tool.

How it worked

1. Investigators interviewed each patient about their contacts — people within 6 feet of an infectious person for 15 minutes or more (30 minutes before May 29) — and the data reached the contact tracing team within 24 hours.
2. The team called each contact to explain quarantine and enroll them. Contacts who refused or could not be enrolled because of language barriers were to be monitored directly by staff.
3. Enrollees got a daily symptom questionnaire until their quarantine ended, by their choice of text message, texted weblink, e-mailed weblink or phone call. Monitored symptoms were cough, difficulty breathing, fever, chills, rigors, muscle pain, headache, sore throat and new loss of taste or smell.
4. Anyone who reported symptoms or tested positive got an epidemiologic investigation; staff called or texted people who didn't answer within 24 hours. Enrollees with no symptoms were released automatically at the end of quarantine.
5. Enrollment records were matched to the state's disease surveillance system to find cases, hospitalizations and deaths.

Who was enrolled

From May 14 to June 26, Maine enrolled 1,622 contacts of 614 patients — 2.9 per patient on average (range 0–31).

• 475 (29.3%) were enrolled within 2 days of their last exposure, including 153 (9.5%) on the same day — a likely sign of ongoing exposure.
• 1,564 (96.4%) accepted automated monitoring; 58 (3.6%) needed direct monitoring.
• Preferred methods: text message (60.2%), texted weblink (21.1%), phone call (7.8%), e-mailed weblink (7.3%). Most (59.0%) preferred an evening contact.

What it found

• 231 (14.2%) enrollees reported symptoms or tested positive. 41 of them turned out not to have COVID-19 — 24 tested negative and 17 had symptoms outside the case definition — and were re-enrolled for the rest of their quarantine.
• 190 (11.7%) met the case definition: 127 (66.8%) confirmed and 63 (33.2%) probable.
• Among them, the median age was 32; 52.1% were female; of those with race recorded, 52.7% were white and 43.5% Black or African American.
• Of 165 who reported how they were exposed, 112 (67.9%) were exposed at home.
• 136 (74.3%) reported symptoms. Four (2.1%) were hospitalized and one (0.5%) died.
• Maine reported 1,869 cases from May 14 to July 10, so about 10% of the state's patients were found among Sara Alert enrollees.

What it means

Tracing and monitoring contacts encourages exposed people to quarantine and lets health departments find symptomatic people quickly. Offering choices of method, time of day and language may help enrollment: for most of the study, Sara Alert sent messages in English only; Spanish was added on June 10, and French and Somali after the study ended.

Automation reduces the staff needed, but not to zero: enrolling contacts, directly monitoring those who don't take part and following up non-responders still take work — about 500 person-hours a week at Maine CDC. Automated tools can support traditional contact tracing, but they cannot replace the large, trained public health workforce a full COVID-19 response needs, and jurisdictions need continued support to build that capacity.

Limitations

1. The total number of contacts could not be determined: some records named places rather than people, some contacts had no working phone number or e-mail, and an unknown number refused and were never enrolled.
2. The data did not separate contacts lost to follow-up from those removed after reporting symptoms, so compliance is unclear.
3. Testing was not required, so untested people with asymptomatic infection were missed.
4. Whether people actually followed quarantine advice was not assessed.

Sources

Based on Krueger A, Gunn JKL, Watson J, et al., "Characteristics and Outcomes of Contacts of COVID-19 Patients Monitored Using an Automated Symptom Monitoring Tool — Maine, May–June 2020," MMWR Morbidity and Mortality Weekly Report, volume 69, Centers for Disease Control and Prevention; a work of the United States government in the public domain. Percentages are as published, among people for whom each item was recorded.