[Survey data as a way to estimate the prevalence of COVID-19. A pilot study in the city of Madrid]

Objective: This research (EP-Covid19-Madrid) was inspired on the lack in the middle of March 2020 of data on COVID-19 produced from representative samples. Its goal was to evaluate the potential of interviewing such type of samples in order to assess the incidence and prevalence of epidemics as COVID-19.
Methods: The sample was of 211 households in the city of Madrid, with one informant for all household members (571). Households were selected through random generation of phone numbers, informants through sex and age quotes. A questionnaire was applied on April/3/20, with a list of symptoms and basic socio-demographic questions for the informant and a general question on COVID-19 for co-residents. Data was analyzed through cross-tabulations and logistic regression.
Results: Prevalence for individuals. On April/3/20, 10,9% of people 18 years and older living in Madrid reported symptoms compatible with COVID19 (SCC19). Occurrence of SCC19 was similar for both sexes, being respectively above and below the mean for the 40-49 (18,9%) and for the >69 (4%) age groups, showing no relation with household size, but being associated with economic activity (19% among working population) and, even more strongly, with the fact of living with symptomatic co-residents (52%). As for households, there was one member with SCC19 in 17% of households. In 8% of the households two or more members presented SSC19, in fact representing 42,9% of the people in these multiply affected households.
Conclusions: Prevalence of SSC19 was much higher than officially reported for COVID-19, although surprisingly low for people of >65. This prevalence associates with work and intra-home transmission. The inter and intra-home prevalence rates (17% and 42,9% respectively) might be useful to assess the proportion of asymptomatic carriers. These results would benefit from confirmation in larger surveys, preferably also including COVID-19 serological testing.