Your top resources for Covid-19 response and resilience

To be helpful, a leading indicator must be predictive, easily accessible, and up to date. There is no perfect indicator, and health systems should use a combination of metrics that are accessible to them to understand the likelihood of an upcoming surge.

Our team has compiled three categories of leading indicators below—clinical, business activity, and mobility—that strategists can use to project the likelihood of an outbreak.

How Covid-19 will impact disruptive market entrants

Clinical indicators

The most obvious way to assess the prevalence of Covid-19 in a region is by tracking testing data and the percentage of individuals reporting symptoms. An increase in these clinical indicators is likely to lead to an increase in patients who must be hospitalized days into the future.

There are several methods of tracking this data, each with its own strengths and weaknesses:

• Surveillance testing: This type of testing refers to determining the infection rate of a random sample of hospital staff or the general population over time. Note, surveillance testing could be complementary to internal testing data you may have access to through hospital labs.
  
  • Pros: local, community-level data; highly predictive of true number of infections; identifies asymptomatic patients
    
    
  • Cons: cost of testing supplies and staff to administer tests; limited sample size
• Call center reports: Use these reports to calculate the number of people calling contact centers to inquire about symptoms and testing options.
  
  • Pros: local data; ability to probe patients for further information
    
    
  • Cons: limited sample size; self-reported data; not perfectly predictive because Covid-19 symptoms overlap with common illnesses; requires manual data collection and data pulls from call center staff
• Chatbot reports: Use these reports to total the number of people reporting symptoms through bots on health system websites.
  
  • Pros: cost- and time-effective collection method; local data
    
    
  • Cons: limited sample size; self-reported data; inability to probe patients for personalized information; not perfectly predictive because Covid-19 symptoms overlap with common illnesses
• Routine wellness checks: Use wellness checks to identify the number of people self-reporting symptoms through apps such as ProtectWell or upon entry into health care facilities.
  • Pros: local data; capability to use Covid-19 specific prompts
    
    
  • Cons: not perfectly predictive because Covid-19 symptoms overlap with common illnesses; limited sample size; self-reported data
• Consumer pulse surveys: Use regular outbound surveys to consumers to determine how many people are reporting symptoms among the general population. 
  
  • Pros: capability to sample the general population instead of just your organization's patient population
    
    
  • Cons: resource intensive to administer and analyze survey results; self-reported data; sample may be prone to bias; not perfectly predictive because Covid-19 symptoms overlap with common illnesses
• Remote monitoring data: Use this data to get the number of individuals reporting symptoms through remote monitoring devices (ex: reports of fever from smart thermometers).
  
  • Pros: free public resource; data updated daily and available at the county level; less bias compared to self-reported data
    
    
  • Cons: sample size limited to patient population with the remote monitoring device; relies on users to track data regularly; not perfectly predictive because Covid-19 symptoms overlap with common illnesses
• Google trends: Use Google to identify the number of people searching for Covid-19 information relevant to symptomatic individuals (ex: "Covid-19 symptoms," "coronavirus symptoms," "hospitals that test for Covid-19 near me").
  
  • Pros: free; easily measurable daily; city-level data
    
    
  • Cons: details interest of the virus's symptoms, not its prevalence; likely very low predictive quality

Of course, there are shortcomings to using symptoms as a leading indicator. According to the World Health Organization, "80% of infections are mild or asymptomatic." This high rate of asymptomatic cases means that tracking only symptoms could underrepresent the magnitude of the outbreak. Therefore, additional indicators that measure the likelihood of transmission of the virus can be used to get a more complete picture of your community's risk.

Business activity and social gathering indicators

As states go through a phased reopening process, businesses and places of public gathering are slowly increasing their capacity. And although these locations have installed added safety precautions and social distancing measures, the mere fact that they are open increases the likelihood of person-to-person contact, and thus transmission of the virus due to imperfect safety protocols and/or lack of compliance with social distancing measures.

Beyond taking local policies into account regarding business and school opening guidelines, there are several data sources that you can leverage to get insight into the extent of person-to-person interactions occurring in the community.

• Consumer spending: Use this indicator to learn the percentage change in personal consumption at retailers, restaurants, and entertainment venues. Credit card companies, restaurant reservation platforms, and hotel booking platforms are some possible sources for this information. For example, one study by JPMorgan found that higher restaurant spending was correlated with a rise in new infections three weeks later.
  
  • Pros: good proxy for risk of exposure to the virus
    
    
  • Cons: lack of regular, updated, and localized data; correlation is not always causative; not perfectly predictive since it does not consider social distancing practices; difficult to determine relative risk of exposure to virus at different venues
• Visits to public venues: This indicator refers to the number of visits to venues such as libraries, workplaces, and transit stations. You can access this information via Google, based on Google Maps data, here.
  
  • Pros: free; updated regularly; available at county level; good proxy for risk of exposure to the virus; data trended over time
    
    
  • Cons: correlation is not always causative; not perfectly predictive since it does not consider social distancing practices; difficult to determine relative risk of exposure to virus at different types of venues

Mobility indicators

The correlation between mobility and likelihood of viral transmission was made clear as state and federal officials instituted stay-at-home orders, international travel restrictions, and domestic travel quarantine requirements. Tracking mobility, or lack thereof, could be a good proxy for forecasting likelihood of a local surge—but it still has significant limitations. Mobility is simply a measure of someone's movement, not a measure of someone's level of interaction with others or risk-taking activities. Therefore, mobility metrics must be contextualized with social distancing compliance to be truly predictive.

A number of sample indicators are listed below and available from the Maryland Transportation Institute.

• Intercounty/interstate travel: Track the number of individuals traveling between county and state lines.
  • Pros: free; county-level data; can track likelihood of imported cases; can distinguish between work and non-work-related trips; can track longitudinally over time
    
    
  • Cons: not perfectly predictive; does not indicate where people are traveling from or risk of viral transmission
• Residents staying home: This indicator refers to the percentage of individuals making zero trips more than one mile away from home.
  • Pros: free; county-level data; good proxy for level of social distancing in the community; can track longitudinally over time
    
    
  • Cons: not perfectly predictive; does not account for high-risk activities close to home (ex: social gatherings)
• Trips and mileage per person
  
  • Pros: free; county-level data; good proxy for level of social distancing in the community; can track longitudinally over time
    
    
  • Cons: not perfectly predictive; does not distinguish between modes of movement (i.e., biking vs. driving) and risk-level of activity

Next steps

While predicting the future with absolute certainty is, of course, impossible, leading indicators can give hospitals and health systems valuable insight and time for planning for local surges. Choose a variety of indicators that give you the most up-to-date and complete picture of the health status and risk-level of your local community and track these indicators on a regular basis to determine how you should adjust volume forecasts and operational plans over time.

Your top resources for Covid-19 response and resilience

Get best practices and expert insights for safely treating Covid-19 patients, protecting and empowering staff, and navigating the road ahead for your organization.

Access Now