Who should use this?
These dashboards are for anyone who wants to understand how different test strategies will affect their budget and the overall rate of infection in their institution.
We expect they will be especially useful for coordinating organization-wide testing plans for schools, nursing homes, and other institutions where a common group of people interact on a daily basis.
What is it good for?
There are many strategies that will reduce infection in your population. Any testing will reduce your disease burden. Using this tool to compare pooling, test accuracy, testing frequency, and costs will allow you to choose the strategy best suited to your needs and your community
How should I use this tool?
You should first input all known quantities for your community. For any testing strategy you implement, we suggest you consider high, medium, and low community prevalence and compare that to "no testing."
I have a big population to consider! Why can I only input a population of up to 15,000?
This model assumes that everyone in the population is mixing together regularly. That is, everyone in the population you are considering is equally likely to mix with everyone else. This is very unlikely for populations that are so large. Instead, one should consider smaller populations and multiply. Say you are a school district with 770,000 students. Instead of using a population size of 770,000 you should think about 154 schools of 5,000 students each. Use a population of 5,000 in the dashboard and multiply results by 154 as appropriate.
We assume daily symptom tracking is taking place for all testing scenarios (other than "no testing"). For the simple dashboard we assume 40% of cases are asymptomatic in concordance with the "best guess" provided by the CDC. This number likely depends on the population in which you are testing.
We also assume that 66% of symptomatic infections are caught via symptom tracking based on this paper.
What is test pooling?
From the FDA's website:
Pooling samples involves mixing several samples together in a "batch" or pooled sample, then testing the pooled sample with a diagnostic test. This approach increases the number of individuals that can be tested using the same amount of resources. For example, four samples may be tested together, using only the resources needed for a single test. However, because samples are diluted, which could result in less viral genetic material available to detect, there is a greater likelihood of false negative results, particularly if not properly validated. This method of pooling samples works well when there is a low prevalence of cases, meaning more negative results are expected than positive results.
Because individual samples are mixed together, if a pooled test is positive, confirmatory tests are needed to determine which individual(s) in the group are positive. For information on the discount rate for pooling which is responsible for the false negatives mentioned above see the section below on testing strategy and disease parameters.
How big should my test pool be?
The optimal size for a test pool is not the same for every community and depends on factors like population size, infection rates within the community, and type of test being administered.
What is confirmatory testing?
Confirmatory testing requires individuals to take a second test if they or their pool tests positive for the virus. This method catches the false positives within a community and prevents people from going into quarantine unnecessarily.
We assume everyone in a pool is retested individually. More sophisticated confirmatory testing strategies exist such as sub-pooling of positive pools. See here and here. Such strategies may lower costs but are more complicated to implement.
What assumptions are being made for the simple calculator?
For the simple calculator we make assumptions about turn around time, sensitivity, specificity, and cost. For specifics see the Resources page.
We also assume that intially, 0 people are infectious and 0 people are immune and 100% of your population will comply with isolation and testing. We assume default values for disease parameters as found in the detailed calculator and as explained on the Resources page.
To pick your own values visit the detailed calculator.
Are there limitations I should be aware of?
We implemented an SIR model that assumes homogeneous mixing in the population. In basically every situation this will be an overestimate. For a population with low prevalence, dynamics will be driven by external community prevalence. This quantity is, of course, unknown for any future period.
We assume that once a person is no longer infectious, they are immune for the remainder of the simulation.
May I have access to your code?
The code to build these dashboards was written in Dash using Python and is available upon request.
How is this calculator different from others?
The primary purpose of this calculator is to inform decision-makers about how different testing strategies will perform in their populations of interest. An important feature is the ability to see how the number of infectiuos people in your population will change over time and how effective various testing strategies can be in terms of cost, number of infections avoided compared to no testing, and number of false positive tests expected.
I have more questions!
Contact us at email@example.com