We have built this dashboard to enable local governments, communities, and workplaces to enact locally appropriate health- and economic-protective policies for workers during this period of economic recovery. This dashboard provides specific information about occupational risk factors and a county level COVID and unemployment assessment.
Mapping tools and dashboards have been widely utilised to share information about COVID-19, with some dubbing this the ‘dashboard pandemic’ (Everts, 2020). Many people hope that these tools will help communities plan for ‘“return to normal” conditions’ (Smith and Mennis, 2020: 2). To allow communities in Washington state to thrive following the COVID-19 pandemic, and associated socio-economic uncertainty, both public health and economic concerns need to be addressed. We have built this dashboard to enable local governments, communities, and workplaces to enact locally appropriate health- and economic-protective policies for workers during this period of economic recovery. This dashboard will provide specific information about workplaces and occupational risk factors and will work alongside the Washington state Covid-19 Risk Assessment Dashboard, which provides information about COVID risks and particular vulnerabilities.
We have gathered data at the county level to assess local COVID risk and unemployment levels. Occupation-specific risk levels have been determined by the frequency of proximity to others and exposure to infection. These three factors (COVID risk, unemployment and occupation risk) are individually classified on a 5-point risk scale (lowest risk, low, medium, high, highest), and then combined to determine a risk matrix for each occupation in each county. This will allow locally appropriate policy responses for employers and county level officials. The smart dashboard will rely on the most recent data to inform these risk matrices, to respond to the crisis and recovery in real time. To learn more about our methodology please click here.
To allow local officials and employers to make informed decisions, we have included data obtained via Twitter to show prominent feelings about COVID in different parts of WA state. This represents real-time, local experiences of COVID-19 and can help inform those making public health and economic policy decisions.
This dashboard focusses on occupational and geographical risk levels. However, we recognize that there are increased vulnerabilities for some groups. People with underlying medical conditions such as hypertension, asthma, chronic lung disease, obesity and diabetes; people over the age of 65, and those who are immunocompromised are at greater risk for increased complications from COVID-19 (Smith & Mennis, 2020: 4; Garg et al, 2020: 458). Additionally, there is evidence that people of color are at increased risk for contracting the virus, being hospitalized and dying from it, which could be linked to living in densely populated areas, systemic racism and discrimination, or a higher presence of comorbid health conditions (Hawkins, 2020: 817). While our risk matrices do not factor in individual or demographic risks, we have included a page of additional sources which provide more information on this to allow employers and individuals to make a more informed decision of their COVID vulnerability in respect to their occupational and geographic risk.
We hope this dashboard will aid public health planning in Washington state and that the methodology might inspire similar dashboards in other states. Below explains in detail the methodologies on how different risk indicators and data are calculated and visualized.
Methodology
This smart dashboard uses county level data to assess local COVID risk, unemployment and occupation-specific risk levels. Together, they allow users to determine geographic and workplace COVID risk. These three factors are individually classified on a 5-point risk scale (lowest risk, low, medium, high, highest), and then combined to determine a risk matrix for each occupation in each county. Below there are more details about how we gathered data and determined each risk factor.
Many public health dashboards have been developed in response to COVID-19, which has led to a concern from the academic community and members of the public that the creators will be able to act as gatekeepers of information which they would not normally be able to access (Cinnamon, 2020). This in turn would allow these organizations to control the narrative of the pandemic. We did not collect any of our own data, instead relying on that collected by the Washington state government and its partner agencies, and other scholars. We also have a publicly available csv file which contains all of the data we have used to create this dashboard. By making our data easily accessible and our methodology transparent, we hope to avoid being gatekeepers of any important information about the pandemic. Additionally, we recognize that we have presented a particular view of the pandemic, by the nature of creating a dashboard which contains particular information. To overcome this as much as possible, we have been clear about what this dashboard is useful for and areas where other sources of information might be more helpful (see About this Dashboard). In many cases, we have tried to link other sources which we think might be valuable to those who use this dashboard, that are beyond the scope of our own work.
COVID-19 Risk
We used COVID data from Washington State Department of Health, which we combined with population data from Washignton Demographics Website at the county level. This allows users to understand the COVID case numbers in the context of the community they are looking at. Counties with more people may have higher COVID cases even if the rate of COVID is lower than places with a similar COVID case number, but a smaller population. County level COVID data, placed in the context of population data, should allow local policy makers to make appropriate public health decisions for their jurisdiction. As Schneider (2020: 301) notes, there is a data gap due to lack of widely available testing so we cannot fully assess the reliability and validity of the COVID testing data at this stage, however, this is the most reliable COVID data we could find for Washington state and believe it speaks to general COVID trends by county.
The indicator we are using is the incidence rate, which is defined as a measure of the frequency with which a disease or other incident occurs over a specified time period. Through this indicator, we can know about the most updated covid case trend. For some counties, the confirmed cases may increase sharply at the beginning but get controlled now. For example, the King county has a high rate of aggregate confirmed case / population which could be categorized into high risk. However, when looking at the most recent case / population, it indicates the risk level of king county is moderate.
| 0 - 0.1 | 0.1 - 0.2 | 0.2 - 0.3 | 0.3 - 0.4 | 0.4+ | |
|---|---|---|---|---|---|
| Incidence Rate | Lowest | Low | Moderate | High | Highest |
Unemployment
Job security is another facet of a worker health that needs to be considered (Baker, 2020: e1). Several researchers have noted a link between job insecurity and adverse physical and mental health outcomes (ibid), making it a crucial part of the public health and economic landscape to consider at this moment in time. To calculate job security, we used unemployment claims from the Washington State Employment Security Department. This data allows users to see which counties have experienced the highest economic disruptions, and highlights areas which may have additional physical and mental health burdens due to job losses. We calculate the overall unemployment risks (lowest, low, moderate, high, highest) based the following chart.
| Unemployment Insurance Benefit per Claimant (UIB) (across top) Unemployment Rate (UR) (going down) |
UIB in the 4th (highest) quartile of all counties in WA | UIB in the 2nd and 3rd quartile of all counties in WA | UIB in the 1st (lowest) quartile of all counties in WA |
|---|---|---|---|
| UR < 4.3% | Lowest | Lowest | Lowest |
| 4.3% <= UR < 5.8% | Low | Low | Moderate |
| 5.8% <= UR < 7.3% | Moderate | Moderate | High |
| 7.3% <= UR < 8.8% | High | High | Highest |
| 8.8% <= UR | Highest | Highest | Highest |
Occupational Risk
We used data from the Bureau of Labor Statistics which includes SOC codes to determine each occupation and survey data to assess many occupational factors. Following Baker et al’s (2020) precedent, we exclusively considered two of these factors to determine workplace COVID risk: risk of infection and proximity to the public. We plotted these factors against each other for each occupation on a scatter diagram. Occupations that scored above a ? for risk of infection were categorized as high risk, between ? and ? was considered medium risk and between ? and ? was considered a low risk. Similarly, for proximity to the public, ? was defined as high risk, ? was defined as medium risk and ? was defined as low risk.
| Proximity Rate(across top) Exposure Rate(Going down) |
0-24 Not around |
25-49 Don't work closely |
50-74 Shared office |
75-100 At arm's length |
|---|---|---|---|---|
| 0-24 Never | Lowest | Low | Moderate | Highest |
| 25-49 Once a month | Low | Moderate | High | Highest |
| 50-74 Once a week | Moderate | Moderate | High | Highest |
| 74-100 Daily | Moderate | High | Highest | Highest |
Overall Risk
Since each risk factor has it's own scale to classify in 5 different risk level. So we first evaluating each factors' risk level based on different standard (for example covid-19 is use incidence rate). Then we got the risk level for each factor in 5 levels and those levels get converted into scores : Highest - 5, high - 4, moderate - 3, low - 2, lowest - 1. Each risk factor took different proportions when calculating the overall risk. Both unemployment and occupatino tisk took 30% while occupational risk took 30%. The reason we divide the proportion in that way is because the infection risk has the strongest direct connection with all citizens and workers, then is the unemployemnet and occupation risk. Bbased on the number we got by adds up three different risks time their own proportion, we convert it back it different risk levels.
Overall risk = 40% infection risk score + 30% unemployment score + 30% occupation score
| 0 - 1 | 1 - 2 | 2 - 3 | 3 - 4 | 4 - 5 | |
|---|---|---|---|---|---|
| Overall risk level | Lowest | Low | Moderate | High | Highest |
Word Cloud
Twitter data are also used to generate word cloud images in order to visualize frequently mentioned terms in Twitter online discussion. Certain words that we considered irrelevant (such as 'the', 'is', 'https', and etc) from COVID19 are excluded from the word clouds. Word clouds are generated using an open-source Python library wordcloud. These images are available for both county level and state level and are updated monthly.
Social Media Analysis
Twitter data are crawled using a Python library Tweepy, which is an easy-to-use Python library for accessing the Twitter API. Using the API, the most recent Geotagged Tweets are collected and updated monthly for each county in Washington State using custom queries. The latest update was made on 2020/11/04. Aside from regular tweets, some tweets may include external links, references and also could be a retweet. This means that the same contents could be repeated in different tweets that are collected. Both past and lates tweets are accessible from our GitHub page under 'assets' folder. We use these collected tweets to perform sentiment intensity analasis. This score shows how much of the tweets are associated with positive/neutral/negative sentiment. We use VADER in Python to do this. This score is not available at county scale, and is only available at the state level due to insufficient amount of Tweets collected in less populated counties.
Word Cloud
Twitter data are also used to generate word cloud images in order to visualize frequently mentioned terms in Twitter online discussion. Certain words that we considered irrelevant (such as 'the', 'is', 'https', and etc) from COVID19 are excluded from the word clouds. Word clouds are generated using an open-source Python library wordcloud. These images are available for both county level and state level and are updated monthly.
Reference
[1] Everts, J. (2020) The dashboard pandemic, Dialogues in Human Geography, 10(2): 260-264.
[2] Garg, S., Kim, L., Whitaker, M., O’Halloran, A. Cummings, C., Holstein, R., Prill, M., Chai, S.J., Kirley, P.D., Alden, N.B., Kawasaki, B., Yousey-Hindes, K., Niccolai, L., Anderson, E.J., Openo, K.P., Weigel, A., Monroe, M.L., Ryan, P., Henderson, J., Kim, S., Como-Sabetti, K., Lynfield, R., Sosin, D., Torres, S., Muse, A., Bennett, N.M., Billing, L., Sutton, M., West, N., Schaffner, W., Talbot, H.K., Aquino, C., George, A., Budd, A., Brammer, L., Langley, G., Hall,
[3] A.J. & Fry, A. (2020) Hospitalization Rates and Characteristics of Patients Hospitalized with Laboratory-Confirmed Coronavirus Disease 2019 — COVID-NET, 14 States, March 1–30, 2020, Morbidity and Mortality Weekly Report, 69(15): 458-464.
[4] Hawkins, D. (2020) Differential occupational risk for COVID-19 and other infection exposure according to race and ethnicity, American Journal of Industrial Medicine, 63(9): 817-820.
[5] Smith, C. D. & Mennis, J. (2020) Incorporating Geographic Information Science and Technology in Response to the COVID-19 Pandemic, Preventing Chronic Disease, 17(E58): 1-7.