Dobson | DaVanzo has a working hospital demand model (stress-tested by state health officials and academic researchers) as well as locally adjusted risk assessment models to help our clients with COVID-19 planning and response activities. We are finding that local context matters tremendously. While there are publicly available models which forecast surge hospital demand, these models can show very different projections. As such, it can be helpful to work with an analyst to model your specific understanding of the unfolding pandemic while applying information specific to your operational context. Dobson | DaVanzo works with clients to find the best possible data in their state, examine the differential spread of the disease and data by county, and include information about health system capacity to build actionable model outputs. We are also working to extend our modeling to post-acute care and other services as it relates to the risk of providers, their patients and the recovery of those infected by COVID-19.
Dobson | DaVanzo has constructed a discrete time-series epidemiological model to estimate potential demand for hospital services due to the COVID-19 pandemic in working with a State Hospital Association. The process involved consulting with state public health departments and university epidemiologists. The Dobson | DaVanzo model considers interventions, such as social distancing, while applying county- and state-level demographic information.
Accounting for local concerns in the model is a critical step in addressing:
- Where does the most at-risk populations live in the state and how does that relate to area hospital capacity?
- How does a state’s demographics potentially affect the spread and impact of the COVID-19?
- Where did the infections start in the state? This enables us to anticipate differential spreads by county or sub-region, where state and municipality public health interventions may have been implemented simultaneously.
Our knowledge of the disease thus far and lessons from modeling have led us to make two key observations:
- As we notice an exponential growth in the diagnosed COVID-19 infection rates in terms of doubling time in days (exponential growth),there are two important factors that should be considered in future case count estimates: a) number of current cases, and b) the infection rate. This observation implies that applying effective countermeasures early on is much more effective at stopping the spread than the same measures applied later.
- We also know that with an incubation phase of up to 14 days, communities need a sustained effective intervention to start seeing positive results in two weeks to a month’s time after the implementation of the intervention. If the interventions are not sustained for a long period of time on first implementation, they will not be effective. This observation implies that absolute minimum effective public health response must be sustained at high intensity for over a month, but 6 weeks is safer to stop most tertiary infections that will happen in that time. It also implies that once the infection rate slows, contact tracing and other traditional epidemiological approaches will be critical to prevent a quick resurgence.
Below is a chart showing our infection model under the following scenarios:
1) no or ineffective intervention,
2) effective, intense intervention (modeled from Wuhan data),
3) the same intervention, but implemented 5 days later, and
4) the same intervention implemented 10 days later.
Dobson | DaVanzo staff are rich in public health policy research capability and training to assist states, hospitals, and communities in public health preparedness by modeling the impact of COVID-19. Our public health projects are led by Dr. DaVanzo who has a Ph.D. in Public Health, and comprises a team of staff with Ph.D. in Economics and Statistics and Masters in Public Health degrees.
Send us an email at email@example.com if you would like to discuss how we can assist in the modeling for your medical demand or surge planning over the immediate or longer-term future.