How many infectious people are likely to show up at an event?

Stephen Kissler and Yonatan Grad launched a Shiny app,

Effective SARS-CoV-2 test sensitivity,

to help you answer the question,

How many infectious people are likely to show up to an event, given a screening test administered n days prior to the event?

Here’s a screenshot.



The app is based on some modeling they did with Stan followed by simulation-based predictions. Here’s the medRxiv paper.

Stephen M. Kissler et al. 2020. SARS-CoV-2 viral dynamics in acute infections. medRxiv.

Users input characteristics of the test taken, the event size, the time the tests are taken before the event, the duration of the event itself, etc. The tool then produces plots of expected number of infectious people at your event and even the projected viral load at your event with uncertainties.

This obviously isn’t a tool for amateurs. I don’t even understand the units Ct for the very first input slider; the authors describe it as “the RT-qPCR cycle threshold”. They said they’d welcome feedback in making this more usable.

11 thoughts on “How many infectious people are likely to show up at an event?

  1. Roughly PCR works by doubling the number of copies of DNA in the vial each cycle. Typical max number of cycles allowed is 35-45. The machine detects the fluorescence of the DNA above some baseline brightness so the higher the counts the more amplification required to get above baseline, so the smaller the original quantity of DNA at the start, and it’s roughly exponential to the base 2

    • The part which isn’t clear to me is the choice of cycle thresholds CTs.

      Right now for covid testing each nation seems to be using a different CT.

      Does anyone know what CT is used for other PCR tests of viral diseases and a scientific procedure to set the right CT?

      How does one correlate the viral load of clinical significance?

      • It’s a good question, my collaborators are collecting serial samples, we know that contagiousness goes away around 10 days post symptoms (according to CDC) so with some data and a model I may be able to answer this in a couple weeks 🤪

      • Cts by themselves are not super informative for the main applications of COVID tests.

        In terms of clinical significance to the patient being tested, the Ct does not add much value – treatment options will depend on how symptoms develop. For contact tracing and preventing the patient from infecting others, a low Ct could indicate high viral load and presumably infectiousness, but a high Ct could mean any of the following: (1) patient is in early stages of disease and will soon become more infections; (2) patient is in late stages of disease and was more infectious in the past; or (3) poor quality sampling leading too a non-representative amount of viral RNA in the test vial. I suspect this last case to be a major factor.

        Tests are also valuable in aggregate for monitoring the course of an epidemic in order to predict hospital burden and inform public health guidelines. For this application, the most valuable metric is the total number of infections over time, so a binary positive/negative readout for each test is sufficient. I’ve seen an attempt to infer epidemic dynamics from the distributions of Cts across patients at a given time (growing epidemics should have more low Cts from recently-infected patients; waning epidemics should be shifted to higher Cts), but again the noise is enormous.

        Note also that assay parameters (such as maximum Ct) will not only vary across countries and diseases, but also from lab to lab depending on their protocols.

  2. The test seems to calculate mainly how many infectious people would show up at an event IF they had all been tested negative at N days previously. This calculation might be useful in some situations.

    But the number I would generally want to know is how many new infections would be expected to happen during an event when I don’t know their test status. Or equivalently, if I go to a party with 30 people indoors, how likely is it that I will get Covid?

    • ” if I go to a party with 30 people indoors,”

      If you go to such a party, it’s going to be populated by idiots who don’t care if they get covid, and you aren’t going to have much fun, because they’re going to be idiots. Anyone nerdy enough to read this blog isn’t going to have much fun at a party populated by idiots.

      Some questions answer themselves.

      As for the first case, folks going to events after many having been tested, we (well, the Republicans) have done the experiment, and lots of them got covid.

  3. Ct is the number of PCR cycles required to amplify a particular fragment above background; a lower Ct means there is more DNA present in the sample.. Since PCR is exponential, a decrease of 1 cycle (1 Ct unit) means twice the amount of starting DNA.

    • So as a policy maker how does one choose the right CT that will lead to positives of clinical significance?

      Is there consistency of CT choice across viral disease diagnostics?

      • Each diagnostic test does a validation where a range of calibration curve are set. Other measures such as level of detection, specificity and sensitivity are assessed. Typically a calibration is done with some of the wells on each test plate but this might be different now. So cycle threshold is not of interest in itself. It is a way to back calculate the amount of RNA. For a qualitative test this is set at a given level, usually the level of detection. For research projects Ct is typically treated as a truncated normal random variable.

Leave a Reply to Daniel Lakeland Cancel reply

Your email address will not be published. Required fields are marked *