Please share the website with your family and friends! We can tell if your area has an outbreak or not if more people take the survey. You contribute directly to our fighting the COVID spread, quickly, free, all while staying safe!

Yes! Please come by every day to help us continue to have quality data. But at the least, please report when your symptoms change. Daily reporting lets us know if you are still negative. Bookmark the site or set a daily reminder to visit.

Presumed COVID = people with fever and cough + people answering "yes" to question 2 who also have fever, cough, or difficulty breathing + COVID Positives.

Confirmed cases come from people who have been tested by PCR. Due to the lack of tests and costs, most people with COVID-19 will never be tested. As a result, we may actually have many more people with COVID-19 than is shown in the health department statistics. This ambiguity overwhelms all others and will not be fixed by simply increasing the availability of tests. Instead, we just ask people directly without depending on testing. Epidemiology is the science of estimating actual cases from some other data. Historically, cases were counted using clinical symptoms and signs. With the advent of genetic testing, we have focused on confirmed cases and abandoned the traditional measures of doctors' judgements. But with the lack of widespread genetic testing, we should not use confirmed cases as the only method of tracking the spread of disease. Alternatively, we can return to the historical ways of epidemiology and count clinical cases.

We know the "confirmed cases" statistic is extremely low compared to the number of people who have this disease. This number is limited by the number of tests available, false negatives, inconsistencies in reporting, and delays in performing the tests. The CountCOVID survey gives an independent indication of the new cases in an area that does not depend on testing. We can tell if more or less people are showing symptoms, if the rate is going up or down, or if be the earliest indicator of an outbreak. For places like New York, we can tell if herd immunity may be reached.

Since the disease is spreading rapidly in days, we need to count the change in cases quickly. The internet can gather millions of pieces of data quickly. Self-reporting by crowd sourcing can give us an approximation of the spread in the fastest time possible, rather than clipboard surveys by epidemiologists. The collective reporting by many people may give us a better measure of actual cases in the community in real-time. We think a website is the most accessible way of gathering this information, especially if the questions are short and few.

These initiatives were announced recently and will not be ready for months. We need information now as the disease spreads, not just capabilities to provide possibilities in hindsight. Our survey gives info on the spread immediately if thousands of people participate now. Our survey asks for voluntary self-reporting that is anonymous. The projects described by the big companies focus on tracking individuals and your movements using confirmed cases, which still give a limited view into the scope of the problem. And while passive, these big initiatives have privacy implications.

The answers are chosen to give accuracy. Since COVID-19 has fever more than 90-96% of the time, this question is sensitive. We will miss the rare instance of an asymptomatic carrier (false negative of 4-10%). However, not all people with fever will have COVID-19 (false positives). Persistent cough reduces the false positives as seasonal flu and allergies typically have sneezing. Similarly, loss of smell is specific to COVID-19 compared to seasonal flu and allergies (reduces false positives). A combination of fever and cough and loss of smell is consistent with a clinical diagnosis of COVID-19. Combining the information reduces both false negatives and false positives to improve accuracy. Using just a few questions, we can determine the incidence of COVID-19 in a population.

Deaths may be the endpoint of severe COVID-19 that began three weeks ago. COVID deaths cannot tell us how many actual cases are in an area now, even if we knew the Case Fatality Rate (CFR). Extrapolations of prevalence from local deaths is many weeks late and not accurate.

Surveys depend on completion. The more questions, the fewer people complete. Our test users indicated that 6 questions were reasonable, but even just 10 questions were too many. We sought to ask a small number of questions to be statistically significant. There are other groups asking pages of questions with few responses. Similarly, we find that many people are resistant to downloading an app.

COVID-19 has been in the US population and appears to be more prevalent than flu since March. This survey attempts to capture current people with possible COVID-19.

We use a mixture of symptoms (Subjective information) and signs (Objective information) to identify likely cases of COVID-19. Symptoms are sense of smell and difficulty breathing that the patient feels. Signs are more things that others could observe such as measuring temperature, breath holding, and COVID-19 testing. We will report out these responses separately so physicians and epidemiologists can weigh their value in their assessment and plans.

Prevalence counts who has or had COVID-19. Prevalence is the proportion of current cases (new and pre-existing) at a given time in a population. We counted the prevalence in April. Incidence is the attack rate of new cases. Incidence is the number of new cases over a time period for a population. Since May, we have changed the question to symptoms in the past two weeks. This incidence reflects the spread of the disease and how fast it is spreading. For more information, visit the CDC's whitepaper on the subject.

Note that exposure likely takes place 3-7 days before onset of symptoms.
The current CountCOVID questions are most analogous to widespread PCR testing results, often reported as New Confirmed Cases per 100,000. A PCR test is likely positive only for the first 3 weeks after onset of symptoms. Antibody detection (Serological testing) may be positive only for weeks 3-6 after onset of symptoms. IgG antibodies may last longer, but this has not been shown. Our prevalence results in April are more analogous to antibody detection.