Michigan’s coronavirus data reporting has gone through substantial changes since the first case was reported during a press conference on March 10th, 2020. The State’s online dashboard contains seven tabs of data, which can be overwhelming without the proper context, so this post will walk users through the different available data.
When you first access Michigan’s coronavirus data at https://www.michigan.gov/coronavirus/, you get a cumulative understanding of the state and counties’ experiences with coronavirus since March.
If you scroll down and click on “See Cumulative Data”, you can access the state’s dashboard. The dashboard’s data is currently updated every day except Sunday.
Tab 1: Current Status
This tab defaults to the same map as on the first page, total cases by county. However, unlike the landing page, it also gives users the option to view cumulative deaths. Users can also select Preparedness Regions or counties of interest to get a more detailed understanding of their local community.
This data can help residents in different counties understand their county’s proportion of overall cases and deaths, both confirmed and probable. The state’s definitions of confirmed versus probable deaths and cases are included in the dashboard to help clarify the exact measurement. It’s important to note that the cumulative case and death data are the raw numbers and don’t take into account population sizes, which we’ve explained before is an important context for understanding this data. D3 used data from Johns Hopkins to create a new interactive map. The data reported here is also provisional and subject to changes in the future as reporting systems improve and mistakes are corrected.
Tab 2: Daily Cases
When someone has a confirmed covid case, they are asked when their symptoms first appeared. Understanding the date of onset helps smooth out the delays in test reporting and show a more accurate curve of case rates over time because individuals are generally understood to be contagious prior to symptom onset.
For example, if someone starts having symptoms today, they would get tested tomorrow, and wait for another 2-5 days for their test results. When the test result is recorded as positive, they would be interviewed about their date of onset. If a lab has a significant backlog, it could take a week or more for the test to be processed and appear in the case data. So while the date of onset is retrospective, it does highlight the trend of cases over time more clearly because it does not rely on systematic reporting from labs that have inconsistent timelines for results. However, since this data is self reported, it is potentially biased based on how each individual identifies their onset of symptoms. Interestingly, fewer people start experiencing symptoms on Saturday and Sunday compared to the rest of the week.
Another important thing to remember about date of onset is that the most current day will always have a low number of cases listed because the testing data can lag. As mentioned before, the date of onset is only collected after a positive test is confirmed. So if someone is tested today and waits 3 days for their positive result, their date of onset might not be recorded for a week. This is critical when looking at the most recent data as the last few days will always be artificially lower rates as people wait for positive test results for their date of onset to be recorded.
This tab also allows a closer investigation of deaths by day. It breaks down deaths by confirmed and probable. The probable deaths have only been tracked since April 5, 2020 and include “those who have COVID indicated on their death certificate, but have not had a positive diagnostic test”. Deaths have declined for a number of possible reasons including: fewer cases in the state compared to March, better treatment plans for severely ill individuals, more testing of asymptomatic and less sick individuals, and a lag between testing positive and death.
Similarly to Current Status, this data is filterable by preparedness region and county. There is also a filter by date slider above the table that can be used to narrow the selection of dates and make it easier to see daily counts. By hovering over any of the bars, you can see the actual count of symptoms on day of onset or number of confirmed/probable deaths.
The MI Start Map offers breakdowns of daily cases per million at the county and state level. In this data, the State indicates that geographically a community is high risk for community transmission with over 40 new daily cases per million, medium risk below 20 cases per million, and low risk below 7 cases per million.
Tab 3: Cumulative Trends
Similarly to the previous tabs, you can sort this data by preparedness region or county and zoom in on specific dates.
Cumulative cases and deaths among confirmed cases helps show the overall impact in our community. Similarly to the daily cases data in the Daily Cases tab, cases are represented by date of onset, not when the positive test was confirmed. It helps put the growth of cases into perspective compared to different points in time. In some data sources, the data is converted to a logarithmic scale, which changes the perspective of the growth in cases. You can compare the different methods of reporting at WorldOMeter’s Michigan dashboard.
Tab 4: Demographics
This tab allows users to view cases and deaths by age, sex, and race. In an important addition, the cases per million is also included, which allows more accurate comparisons by providing context to the number. Cases per million accounts for Michigan’s actual population in each demographic, which helps understand disproportionate impacts. For example, we’ve explained before how raw case numbers can hide outbreaks in rural areas. If 100 people are sick in a community of 1,000, the community is going to be impacted significantly differently than in a community of 100,000.
Similarly, other demographics have different sized populations in the state. For example, about 78.7% of Michigan residents are White and 13.8% are Black. If we compare just the overall number of cases, it appears that White residents have a much higher rate of cases than Black residents (51,285 vs. 24,937). However, when looking at cases per million, the story flips with Black residents having 16,595 cases per million and White residents with 6,385. Early on in the pandemic, it became clear that coronavirus was impacting certain communities at a much higher rate and this data can help understand where to direct resources for medical research, testing, treatment, and contact tracing.
It’s also important to note that there is data about Hispanic/Latino and Arab ethnicity underneath the dashboard in static tables. The race/ethnicity data has reliability issues, with a substantial percentage of cases and deaths assigned to the unknown category. However, this has significantly improved since April when the data was first reported as we talk about in our last blog post about the evolution of data.
Tab 5: Total Testing
Testing is an important metric in the state’s response to covid. In this tab, the State reports diagnostic and serology tests that have been reported each day. Diagnostic tests are conducted to determine if an individual has an active infection and, thus, needs to isolate to avoid spreading COVID-19 to others. Serology tests identify antibodies in the bloodstream that indicate an individual likely had an infection previously.
Like the other tabs, you can obtain data for a specific preparedness region or county and zoom into smaller time frames. This can help a resident understand the proportion of tests that their county has compared to the number of cases or total population. Understanding where tests are being distributed can also lend more information about the pattern of cases being identified because if there aren’t enough diagnostic tests in a community, their case rate will be artificially lower.
Another important context that diagnostic testing can provide is around daily case rates. If there is a large spike or dip in cases on a particular day, checking the diagnostic testing count for around that time period can help identify if the change in case rate might be driven by a change in diagnosis versus increased/decreased community spread.
Tab 6: Diagnostic Testing
This is the final tab with data in the dashboard. Diagnostic tests, as mentioned before, show how many individuals have active coronavirus infections. This tab shows the percent of diagnostic tests that are positive, one of the governor’s key indicators of our successful fight against community spread.
The MI Start Map considers anything below 3% positive to indicate a low risk of community spread, 3-7% medium risk, 7-10% medium-high risk, 10-20% high risk, and over 20 percent to be very high risk. So while the state’s percent positive rate has hovered around 2-4% since July, users can also zoom into their particular county or preparedness region to understand how testing in their community is doing. If a community is experiencing high or low rates of positives, it could impact decisions about what types of activities are safe to participate in.
According to the MI Start Map: “When testing levels are stable, this metric identifies increases or decreases in the epidemic relative to other causes of respiratory infection. This metric is commonly used in community monitoring studies of influenza and other respiratory viruses. Evaluations of percent positivity (number of positive tests / total number of tests) also enable a look at adequacy of testing in a population.”
For example, from the above data, Detroit residents are the only community that is providing close to adequate testing for the community. The three counties all fall into the Medium or Medium-High risk categories, indicating that there is a risk of community spread in these counties and that additional testing might be necessary to identify all the cases.
Tab 7: Learn More
The final tab includes some important information and definitions about the data contained in the previous six tabs, including links where you can learn more information about Covid-19 as well as the different departments and systems that are working together to make this data access possible.
You might also notice other links above the dashboard to more data such as “Data About Places” or the public use data files located below the dashboard. These are additional rich sources of local information that we will tackle in a future blog post as the information contained in them requires in-depth explanations like this!