Michigan

Thank you for clarifying that the antigen positives are classified as probable cases! To clarify why we asked, we were initially led to believe that confirmed cases included antigen results because on their dashboard, MI defines “confirmed cases” as “individuals with a positive diagnostic test for COVID-19,” and the definition of diagnostic tests on the dashboard includes antigen tests. (Attached are screenshots of the wording). We’ve seen a few states consider antigen positives as confirmed cases and deaths (for example, Indiana), so we wanted to confirm whether MI was after seeing this wording on the dashboard.

Per the CSTE case definition, individuals in Michigan who have a positive antigen test result are classified as Probable cases.

Thank you for pointing this out! We can fix that language

Based on definitions on your dashboard, we believe that MI is counting any individual with a positive antigen test or PCR test as a confirmed case. However, the most recent CSTE guidelines recommend counting individuals with a positive antigen test as a probable case. A) Is our understanding of MI’s probable case definition correct? B) If so, why did MI choose to count antigen positive individuals as confirmed cases instead of probable cases?

Michigan does not count antigen positive cases as confirmed cases. That is clear on our website I’m not at all clear about how this is being interpreted. We do not count antigen positive cases as confirmed here in Michigan per the CSTE guidance (you can note that I am one of the CSTE authors of that guidance)

This is how we report.

Daily, we give an update on NEW information that has come in since our last summary (based on referral date, not onset date… ie. what is transmitted, not exactly when the event took place) BUT we only report on the CONFIRMED cases here. Cases resultant from the transmission of laboratory referrals of pcr testing.

(unnamed 1.jpeg)

Let’s take a look at how we report cases based on the onset date (or test date in the absence of an onset date) on our website too.

image001.png image002.png

The confirmed and probable cases are literally distinct thoughout our reporting page.

We understand the difference between the 2 hospitalization tables provided here. Will you release this data in ARCGIS/files?

Not possible at the moment.

Why do the public use datasets have different cumulative numbers than the dashboard? What is their verification process between the two sources? Context: On 10/17 the MI dash updated late and we used numbers from data downloads. These numbers were higher than the updated dash numbers. Why is this?

So, usually, the public-use data sets, I think get posted a little faster. And then the dashboard typically updates really quickly. But for some reason that day, it just took a while for the the dashboard to fully update. Sometimes it'll update in pieces. And so I think what happened was the date got updated. And then the data. Because I had like an email the following morning about that. The dashboard not mentioning the public-use data sets, but by the time I went to look at it, it had fully updated. So sometimes it's just an issue, it's just giving you a little more time.

I would say almost always the public-use datasets get posted usually first. Almost always first. And they get posted, like all at the same time by the same person. So you won't have a problem with a public-use datasets where like, one sheet will be updated and the others are still waiting to update. So they get updated, like, bam, all at once. So if you want to go by that, it's probably pretty safe.

All the cases/deaths by race values add up to total cases except for probable deaths by race -- why is this, can they account for the missing deaths? Or should we just be counting them in the Unknown category?

The reason why you're seeing the probable deaths not add up to the total by race is because we suppress counts if they're below six, just to protect confidentiality, and our probable deaths are our smallest total. So when we split it out by race, then we lose some deaths that happen in categories with very few deaths.

Would it be possible to report the ethnicity data as whole numbers instead of (or as well as) percentages - we lose a lot of information because of the rounding.

We did that because very early on, those were reported out on numbers for the really, really specific, and we felt identifiable. And so, some of the asks you're making to change, kind of how we report things after we've been reporting it the same way, I think might be a little awkward with some of our media colleagues that have been taking that off our system every day. So ,what I was thinking about today was: Do you think it would be helpful if rather than report numbers, if we were to just bump it out and report say, go out with the percentages two decimal places to help? Because I know that percentages, those crude numbers get a little more awkward once we get above you know, a hundred thousand, 150,000, 160,000 cases. So, rounding errors can be big, by one percentage point. But if we were to add some decimals to that, and we could look into that. That might help, do you think?

Would it be possible to report Asian and Native Hawaiian and Pacific Islander as separate categories, in line with the Census categories?

1: I know, we talked about this a lot working with some folks and some other programs early on in determining the categories we were going to use and some of that was based on testing value and things like that. But that would be, I think, honestly a difficult change to make at this juncture after reporting on the same categories for so long. Sarah, I don't know, but do you feel any different?

2: I think we'd run into an issue with sample size, which is one of the reasons they're collapsed together. So that's, that's kind of where we're at.

1: And again, a lot of it was sample size back in the day, and some of those are not as impactful now that we have so many cases. But now some of the issue may be changing our reporting this far into the game.

On the chart, there is an * next to the LTC cases and deaths totals. What does “includes data that is undergoing additional validation” refer to? What is the additional validation?

So, each day the skilled nursing facilities report data in. And medical services administration staff(so, not people in this phone call, but others), they call back and verify if there's questions or oddities about it. You know, if you get data that somehow, like, your cumulative has dropped over time or something like that, they'll be calling and checking back on that....

So, it's kind of their version of "data may change."

Can you release your LTC data in a spreadsheet format? Right now we run a script to gather the data.

I can send that over as an asked to medical services administration, they've been putting it up as HTML, because that's sort of the simplest version that they have of doing things. So, I can send that over as a question to, to them.

We have on record that your total diagnostic tests on the dashboard counts tests rather than individuals. So, if people are tested multiple times for coronavirus, they will get counted multiple times in that figure. We are wondering how individuals would be counted if an individual goes in for testing multiple times on the same day, will they be counted once or twice?

Test results are dependent on their message date, which is the date the test result is received by the state health department. If an individual is tested multiple times on the same day at the same lab, the message date will almost always be the same and the test results will be deduplicated. It will count as one test. If an individual is tested multiple times in one day at different labs, the CLIA ID number from the lab, which is used to deduplicate test results, will be different. Here, the test results are counted as two tests even if they come back on the same message date because they are coming from separate labs. It would be rare for specimens collected on the same date at the same lab to come back on different message dates. Diagnostic testing for the same person(two PCR tests) occurring at different labs on the same date would also be rare.

If an individual is swabbed multiple times in one test (e.g. one nasopharyngeal swab and on oropharyngeal swab), and the samples are run separately, will that get counted as one or two tests? If that individual is counted twice, do all laboratories consistently provide data at sufficient granularity to count samples rather than tests?

Early on in the pandemic, multiple swabs in a test was occurring, though that doesn't happen so much now. When it was occurring, it was treated in the same way that two tests in one day at the same lab would be treated(since it would follow the same message date rules) - the specimens would be deduplicated and counted as one test. We're just not seeing people have multiple specimens taken on the same visits now.

We do have sample level source information in our HL7 feeds though we haven't profiled it as MI doesn't currently see it as important.

We have learned that some states are excluding repeat positive results for an individual after their first positive test even while counting repeat negative tests. Does your total diagnostic tests metric count results after the first positive?

It's a test-based statistic, so we're not removing anyone from the numerator or denominator depending on their prior result. It counts subsequent positives, and it counts subsequent negatives. It just counts all tests. Here we're referring to diagnostic(PCR) tests. Our surveillance system doesn't count them as cases. We have a 3-month rule for repeat positive tests. We'll introduce a case as a new case if it's a positive 3 months after their first positive specimen. There can be a lot of testing in between that time, and we don't count those as new cases.

In MI’s reasons for why probable cases fluctuate (tab “Learn More” of your dashboard) The dashboard says: “As additional information is gathered, the case could be reclassified to confirmed, remain as probable, or be determined as not a case.” What circumstances would lead MI to determine a probable case is not a case?

If they have an exposure history to meet the epi link and symptoms to meet the CSTE definition of a probable case, then they are a probable case. If they then get a negative PCR test, then they are determined to not be a case.

Michigan is lumping PCR and antigen. Can you separate this reporting?

1: So you know, we do not include antigen tests in our calculations for positivity, we base that strictly on the PCR test.

2: Right now. We're tracking that volume. And has it been excessive? So we have the ability to do it, we have it come to an agreement on how right how we want to report that, I would say, and really I would go back to Sarah and Jim, on whether we plan on publishing that right.

1: Yeah, I think we're discussing that, especially with the introduction of a, you know, hundreds of thousands of binax tests to the state, plus for reporting from those. So I think reporting is going to be a challenge with the binax tests. So those instant read tests. That's certainly going to be a challenge. But we're distributing those and the feds are distributed those to long term care. And we're looking at a bunch of different locations that are going to report back to us, so I think we're going to be looking to breaking that out kind of like we break out in the ____ counts where we have the serology and then the PCR. We'll have an antigen color to it. (He's referring to the graph on the dash).

2: We have dozens of labs that are reporting testing in the state of Michigan. But as of last Friday, we only had nine labs, for example, that were reporting antigen tests across the state of Michigan. So just to put it in perspective, this volume isn't there, but we know that it's coming.

Would it be possible to have hospitalizations and tests by race and ethnicity included in what they report?

The hospitalization data does not have race. We have no way of providing that to you. For testing: It's about it's only reported about 50% of the time. And so we we have not made the decision to stratify out, you know, the testing by race because of that.

Why is the testing data only 50% complete race/ethnicity?

Speaker 1: I mean, we can hypothesize: it varies by lab, it varies by time, we have a number of, I'm speaking on behalf of Sara and Jim, a number of data quality improvement projects going on with our major, you know, most ___ labs. And some of it is really just in the lift from the reporters in transferring over that data, should they have it on the requisition form. And in other instances, we hear from the labs that they didn't have it on the requisition form. So there's a variety of reasons, but we haven't been tracking it from the very beginning. But it has not hit over 50% of the tests coming through with a valid race or ethnicity.

2: We've, we have struggled to get, and I've seen marked improvements and getting the core demographics or contacting the cases, with our electronic lab reports....When local health ___ gets a referral, they leave that information as part of their public health follow up, and so it doesn't stay in the void forever. So we've had significant improvement and provision of telephone numbers and addresses for cases, so we can route them to the correct local health jurisdictions for case follow up and investigation.

1: We did prioritize as a part of our data improvement projects, with the labs, this provision of a valid patient address and a valid patient phone number, obviously, because of the criticality for contact tracing. We'd love to have improvements in in race and ethnicity. But if we had to bug them and push them for something, first, the priority was for the patient address and phone number.

We are currently interpreting total diagnostic tests as referring to the total number of tests conducted. Is this correct? If so, do you have plans to release the total number of unique individuals tested?

I believe the total diagnostic test data is total tests performed, not individuals tested. would know more. I only receive the aggregate data after MAG has cleaned and prepared. I don’t have any information on the plans of the department to release data on unique individuals tested. SLC might be able to provide more insight into that question.

Do you perform any deduplication to reach total diagnostic tests? If so, how? (i.e. instances swabbed per day/week; recording only one positive test per person, but multiple negative tests for the same person; etc.)? When reporting testing results, if the same person gets different test results on different days, would you report raw results as part of the daily positives or do you apply any kind of logic such as reporting only the first negative or positive result per individual?

Again, I only receive the aggregate data after it’s been prepared by MAG. I’m not sure how they handle the circumstances described above. I would forward this question to .

We’re also interested in testing data in units of “testing encounters”, which can be defined as “the number of unique people who have been tested per day.” Under this metric, if a person is being tested once today and once again the following week, that counts as two tests, however, if the same person has two samples tested at a testing center visit, then that only counts as one testing encounter. Are you willing to or are you planning to release testing encounters? If so, will you be able to provide historical data consistently in an automation friendly form, like an ArcGIS or CKAN API or a direct download of CSV or JSON?

Sorry, again this is not something I have any insight into. I can’t speak to what the department would or wouldn’t be willing to release. Based on the question above, if MAG could analyze the raw data to create unique testing encounter and provided me that data, I could provide it in various forms, given that MDHHS was willing to release the data. These are all hypotheticals though, as I do not currently receive unique testing encounter data.

For the ethnicity percentages on the main Michigan data page, are those calculated from the total cases, the total confirmed cases, or some other number?

For the ethnicity percentages, this is for confirmed cases only.

What challenges are you facing in adjusting to the new HHS guidelines? Which metrics that you report were impacted, or could be impacted by HHS procedure changes? How does this affect the reporting relationship between hospitals and DOH? Are there changes in regulations on which hospitals are now required or not required to report to state DOH?

Earlier this month, MDHHS began reporting a combined total for confirmed and suspected adult COVID-19 cases as a result of changes in how the federal government requires us to post the data. MDHHS modified all variables to align with federal reporting requirements to minimize the reporting burden on hospitals. Following feedback from stakeholders, including media, we again began providing confirmed cases in a separate category along with other data elements we had previously been reporting in addition to the newly required data.

Below is info on the previous data and new data we are now reporting: Previous Clinical Variables: # of ED Discharges - The number of patients discharged home (or equivalent) from ED with suspected or confirmed COVID-19 in past 24 hours. INCLUDE hospital urgent care and ED drive through. # in Critical Care - The CURRENT number of patients in critical care units with suspected or confirmed COVID-19. INCLUDE non-critical care areas being used for surge critical care. DO NOT INCLUDE ED boarding/overflow (patients currently admitted and awaiting bed). # on Ventilators - The CURRENT number of patients receiving mechanical ventilation with suspected or confirmed COVID-19. DO NOT INCLUDE ED boarding/overflow (patients currently admitted and awaiting bed) and surge critical care areas. # of Inpatients - The total number of positive tests for COVID-19 among admitted patients.

New Clinical Variables: Previous Day’s COVID-19-related ED Visits ( Enter the total number of ED visits who were seen on the previous calendar day who had a visit related to COVID-19 (meets suspected or confirmed definition or presents for COVID diagnostic testing – do not count patients who present for pre-procedure screening).) Total ICU adult suspected or confirmed positive COVID patients (Patients currently hospitalized in an adult ICU bed who have suspected or laboratory-confirmed COVID-19.) Hospitalized ICU adult confirmed-positive COVID patients (same as above, laboratory confirmed only) Hospitalized and ventilated COVID patients (Patients currently hospitalized in an adult, pediatric or neonatal inpatient bed who have suspected or laboratory-confirmed COVID-19 and are on a mechanical ventilator (as defined in 7 above).) Total hospitalized adult suspected or confirmed positive COVID patients (Patients currently hospitalized in an adult inpatient bed who have laboratory-confirmed or suspected COVID-19. Include those in observation beds.) Hospitalized adult confirmed-positive COVID patients (same as above, laboratory confirmed only) Total hospitalized pediatric suspected or confirmed positive COVID patients (Patients currently hospitalized in a pediatric inpatient bed, including NICU, newborn, and nursery, who are suspected or laboratory-confirmed-positive for COVID-19. Include those in observation beds.) Hospitalized pediatric confirmed-positive COVID patients (Patients currently hospitalized in a pediatric inpatient bed, including NICU, newborn, and nursery, who have laboratory-confirmed COVID-19. Include those in observation beds.)

Is Michigan performing pool testing?

We have one hospital lab that has an a EUA for this. We are in process of doing validations for pooling with multiple commercial and clinical labs in the state. It is very important to ensure the correct numbers in a pool based on prevalence of disease in a given area and other population factors to ensure that there is not a major loss in sensitivity that would result in accidentally missing patients that are positive. Based on volume of testing needs this will be very beneficial, but we need to cross check the populations, the sensitivities, the impact of different types of media singly and mixed on ability to detect, differences in asymptomatic vs symptomatic viral content for detection, etc. There are about a dozen sites that we are all working together on this aspect though.

Is Michigan conducting antigen testing? Are these testing results reported on the state site? How many tests have they conducted?

Serology testing is occurring in Michigan. Results are included on our website and marked serology on this page https://www.michigan.gov/coronavirus/0,9753,7-406-98163_98173---,00.html under Current Status and Total Testing tabs. Serology tests are counted separately from diagnostic tests.

What is the difference between the “COVID-19 metrics table” and the “Patient Census Table” on the “Statewide Available PPE and Bed Tracking” page? Which table should be used for the most up to date current and cumulative hospitalization figures (inpatient, ICU, and ventilation)?

The COVID Metrics table is reported by the Emergency Preparedness Regions and is updated M-F. The Patient Census Table is completed by Michigan Hospital Association and broken out by hospital system. It is updated twice a week. I would use the COVID-19 metrics table for most up-to-date info as it is updated more often.

Could the Asian & Pacific Islander categories to be separated to match the Census categories?

We are not able to separate Asian & Pacific Islander categories.

How often is MI updating demographic data?

The demographic info for cases and deaths is updated daily along with the other data on our website. It is located on this page https://www.michigan.gov/coronavirus/0,9753,7-406-98163_98173---,00.html under the Demographics bar on the chart. There are also additional data below the main chart of Hispanic/Latino ethnicity and deaths by Arab ethnicity.

Is the percentage of cases or deaths of all the reported cases or deaths (what is n=)?

The percentage of cases or deaths is for all the reported cases or deaths.