JMIR Publications

ABSTRACT

Background:

Crowdsourcing is a useful way to rapidly collect information on COVID-19 symptoms. There are potential biases and data quality issues that can be introduced, however, given the population that chooses to participate in crowdsourcing activities and common strategies used to screen participants based on their previous experience.

Objective:

Study objectives were to: 1) build a pipeline to enable data quality and population representation checks in a pilot setting prior to deploying a final survey to a crowdsourcing platform, 2) assess COVID-19 symptomology among survey respondents that report a previous positive COVID-19 result, and 3) assess associations between symptomology groups and underlying chronic conditions and adverse outcomes due to COVID-19.

Methods:

To build a pipeline, we developed a Web-based survey and hosted it on the Amazon Mechanical Turk (MTurk) crowdsourcing platform. We conducted a pilot study August 5, 2020 to August 14, 2020 to refine filtering criteria to our needs before finalizing the pipeline. The final survey was posted late-August to December 31, 2020. Following data collection, we performed hierarchical cluster analyses to identify COVID-19 symptomology groups and logistic regressions for hospitalization and mechanical ventilation outcomes. Last, we performed a validation of study outcomes by comparing our findings to those of others reported in systematic reviews.

Results:

The crowdsourcing pipeline facilitated piloting our survey study and revising the filtering criteria to target specific MTurk experience levels and to include a second attention check. We collected data from 1,254 COVID-19 positive survey participants and identified six symptomology groups: abdominal and bladder pain; population with flu-like symptoms - loss of smell/taste/appetite; hoarseness, sputum production; joint aches, stomach cramps; skin or eye dryness, vomiting; and no symptoms group. The risk factors for adverse COVID-19 outcomes differed for different symptomology groups. The only risk factor that remained significant across four symptomology groups was influenza vaccine during the previous year (OR1=6.22 95%CI 2.32-17.92; OR2=2.35 95%CI 1.74-3.18; OR3=3.7 95%CI 1.32-10.98; OR4=4.44 95%CI 1.53-14.49). Comparison with systematic reviews indicated that our findings regarding symptoms of abdominal pain, cough, fever, fatigue, shortness of breath and vomiting as risk factors for COVID-19 adverse outcomes were concordant with others. Some high-risk symptoms including bladder pain, dry eyes or skin, and loss of appetite were found in our study, were reported less frequently by others and were not considered previously in relation to COVID-19 adverse outcomes.

Conclusions:

This work demonstrated that a crowdsourced approach was effective to collect data to assess symptomology for COVID-19. Such a strategy may be considered by others to facilitate quick and cost-effective assessments in a rapidly changing spectrum of infectious disease, and changing societal and environmental settings.