Those getting tested for COVID-19 and want to reduce the chances of a false negative might want to get swabbed in the early afternoon—statistically speaking, 2:06 p.m. looks optimal.

That’s according to the first study to examine the effect of time of day on COVID test accuracy, reported in the Journal of Biological Rhythms by a team at Vanderbilt University Medical Center.

Tracking COVID-19 test results by the time of day that samples were collected from patients, the researchers report a 1.7-fold variation in the adjusted probability that a PCR (polymerase chain reaction) test would return a positive result: odds peaked at 2:06 p.m., when a positive result was 1.7 times more likely than at the daily low, shown in a graph to occur around 2 a.m. (the emergency room never sleeps).

The VUMC team studied results of COVID tests performed on adults at VUMC from March through August of 2020—86,342 tests in all. Some 8.7% of the tests were positive for SARS-CoV-2. The analysis was adjusted for effects of patient age, sex, race and testing location, among other factors.

“Our question was whether there was some daily pattern of positive versus negative test result rates, perhaps due to circadian regulation of the immune system or cycles of viral shedding in people with COVID infections,” said medical data scientist Thomas Lasko, MD, PhD, associate professor of Biomedical Informatics. “It turned out that the exceptional data resources we have here at VUMC allowed us to glimpse the answer to that question in the first weeks of the pandemic, and further analysis found an intriguing effect that became clearer as more people were tested.”

False negative results are common in clinical testing for SARS-CoV-2, the virus that causes COVID. In August of 2020, in the Annals of Internal Medicine, a team at Johns Hopkins School of Medicine reported a literature review and pooled analysis of test accuracy for SARS-CoV-2, finding a median false-negative rate on PCR tests of 67% on day four of the infection, dropping to 38% on the day of symptom onset (day five of the infection) and to 20% on day eight.

Lasko led the study with epidemiologist and emergency medicine specialist Candace McNaughton, MD, PhD, MPH, now associate professor of Medicine at the University of Toronto (and formerly at VUMC).

“The daily pattern we found for these test results could be down to a host of patient factors that vary over the day, factors which are already known to pose various clinical implications. These factors involve not only the immune system but the autonomic nervous system, body temperature, fasting/feeding cycles, hormones and so on,” McNaughton said, adding that further studies are needed.

“As the pandemic plays out, confirmation of our result would carry important implications for public health and diagnostic strategies, not least in terms of optimizing test collection times,” she said.