The Role of Forecasting in Improving Hospital Operations

As we head into winter here in Boston, a favorite topic of school-age kids is soothsaying when and whether school will be closed for snow.  As soon as news breaks of an approaching winter storm, instagram switches from the Kardashians and Oscar snubs to weather maps and polar vortexes in an attempt to figure out if school is open or closed.

Why the uncertainty?  All the schools are working off the same weather forecast.  It’s not like one school has access to a “special” weather forecast other schools do not.  Yet unless the storm is really major, some schools stay open while others stay closed.  If everyone has the same data, what accounts for this discrepancy?

The decision to close or stay open is based on more than just the forecast.  School leadership  must also take into account the transportation risk of staying open, the impact on working parents by closing, and even the reputational outcome of the previous decision to remain open or close.  

This is because a forecast is not a decision.  It’s a key piece of data used to support a decision, but it is not the only consideration.  When hospitals first start using Hospital IQ’s forecasting tools, they sometimes conflate these forecasts with decisions.  They think the algorithms and data removes all uncertainty about how to proceed.  But a forecast is only as good as the hospital leadership behind it.  The true value of a reliable forecast is the ability to increase the lead time and confidence of a decision.  A canonical example in a hospital is whether or not to execute a surge plan in response to a forecast shortage of inpatient beds.  Sometimes this is triggered by an external disaster of such proportions the decision is clear (e.g., a literal train wreck), but other times are more ambiguous.  Surge plans usually require some combination of calling in per-diem and locums staff, required overtime, expensive short-term equipment rentals, and even cancelling elective surgeries.  If a surge plan is implemented but turns out to not be necessary, there are often cost overruns and staff dissatisfaction.  If the surge plan isn’t implemented and it turns out to have been necessary, patients are put at risk as well as heightened pressure on limited staff.

However, a census forecast does not always determine whether or not to initiate a surge plan just as a weather forecast cannot always decide if school should be open or closed.  Forecasts are very powerful tools that provide a warning or indicator which can help streamline the management process and provide additional critical lead time to address the situation and avoid or minimize any negative impact.   They also give management the opportunity to fine-tune the surge plan, such as only opening half a flex unit or transferring less acute patients to a satellite location that may have more bed capacity.  As implementing a surge plan is rarely a clean binary go / no-go decision, software can help tease these apart.

Futurists talk about computers replacing humans and making decisions without supervision.  As this example illustrates, we still have a long way to get to that future. In the meantime, hospitals need to learn to use a census forecast in the same way schools have learned to use the weather forecast — as the basis for sound decision-making that ultimately protects everyone’s safety and well-being by allowing the hospital to run as efficiently as possible.