With anecdotal reports of viral shedding from COVID-19 patients for several weeks, there is a need to quantify the prevalence of long-term SARS-CoV-2 shedding. Here, we characterize the temporal distribution of diagnostic SARS-CoV-2 PCR outcomes from nasopharyngeal swabs and associated EHR-derived features over two months for 874 COVID-19 patients with longitudinal data. Among a cohort of 379 COVID-19 patients with at least one positive follow-up SARS-CoV-2 PCR test, 53 patients remain SARS-CoV-2-positive after four weeks of initial diagnosis. Surprisingly, a majority of COVID-19 patients with long-term viral shedding are not hospitalized (40 of 53 patients), and have no enrichments among symptoms, demographics, or medical history. In a cohort of 370 COVID-19 patients that transition to a confirmed negative status, the upper bound of viral shedding duration has a mean of 21.2 days with standard deviation of 9.3 days. Of the 81 PCR-confirmed COVIDpos patients who have undergone serologic testing, 68 patients have developed anti-SARS-CoV-2 IgG to date, with a mean upper bound of time to seroconversion of 38.1 days (95% C.I. = 35.2-41.1 days). Given that SARS-CoV-2 PCR testing may detect replication incompetent virus and that serologic tests do not imply neutralizing immunity, we suggest that the development of novel assays for measuring infectious viral load in non-hospitalized long-term shedders may help mitigate community transmission. This study motivates a platform that can link longitudinal diagnostic and serologic testing with real-time epidemiological data, towards proactively identifying and managing emerging hotspots of COVID-19.