Updated 4/14/20
Should the Best Available Data or Worst Case Scenarios Drive Public Health Policy?
Imagine you’re an elected official and you need to increase hospital capacity to respond to a pandemic.
Imagine you don’t have a national health system like the UK, so you need to issue executive orders to compel the hospitals you regulate to increase capacity at their own expense.
How do you decide what to order the facilities to do? What percentage increase in their bed and ICU capacity should you require them to implement? How many ventilators should you acquire?
State-of-the-art modeling using the best available data may be your best bet for informing important policy decisions.
At the beginning of an epidemic there may be unreliable data with which to use in the model. As the epidemic progresses more data becomes available, it allows you to refine your model and your policy directives.
That’s where we are right now. It’s time to use better modeling to improve the evidence base for policy directives.
On March 26, the governor issued an executive order directing AZ hospitals to increase their bed and ICU capacity by 50% by April 24. Half of that increase needs to be in place by yesterday (April 10). The media release announcing the Order said that it’s based on a worst-case scenario.
But is it best to base public policy on a worst-case scenario?
That depends on who is paying the invoice, doesn’t it? If you’re a hospital that’s responsible for complying with the order, perhaps you’d rather see it based on more likely scenarios using models with contemporary data. If you’re an elected or appointed official who’s not paying the bill, perhaps you’d rather see the Order based on a worst case scenario to cover your downside risk.
Our elected and appointed health officials and public health staff are busy putting together a host of response plans designed to minimize the health impact of the virus. Modeling should be a key element in those plans. It’s always best to use evidence-based criteria in planning, especially when you’re asking private hospitals to expand their capacity at their own expense.
New evidence suggests our interventions are working
Indeed, there’s evidence that our social distancing interventions are working, providing new information that our policy-makers should consider as they consider their interventions.
For example, an Arizona public health associate professor has released a COVID-19 disease outbreak outlook that suggests our Arizona interventions are working. Joe K. Gerald, MD, PhD, acting in a personal capacity, studied data from Arizona COVID-19 cases and states… “Mounting evidence indicates that social distancing, including the current stay-at-home order, is slowing the spread of new infections.” He also says the “lag between new infections and hospitalizations and ICU admissions means that the pace of these outcomes will increase for the next 1-3 weeks before slowing.” Here’s the analysis and discussion (results from 4/14/20).
The results also suggest that our aggregate hospital system is already adequate to handle the peak number of cases. With that in mind, does it still make sense to require hospitals to cancel all of their elective procedures and increase their bed and ICU capacity by 50%? Those directives are putting intense financial stress on our hospital system and at this point are likely doing more harm than good (as long as folks still continue to practice good social distancing).
We urge our policy makers in Arizona to tap our University expertise (like Dr. Gerald) and use their analyses to adjust their interventions and directives. Evidence based policy-making demands it.
P.S. Here’s a good article from the Arizona Republic that talks about the financial impact that these Executive Orders are having on our hospital systems.
P.S.S. Here’s a good high-level 10-minute You Tube video about how models are developed and can (and should) be used.
P.S.S.S. The ADHS began displaying COVID-19 cases by Primary Care Area or ZIP Code ion 4/12, including hospital capacity data. That data is posted here.
