When it comes to the effect of the COVID pandemic, one can say two things with certainty: occupancy has been negatively impacted in most facilities and states, and that recovery has been slow and will be ongoing for the foreseeable future. An interesting question is what has the causal impact of COVID been on occupancy?
To answer this question, I fit a model where, for each state, I projected daily census numbers from the date of the first confirmed COVID case in the state to March 31, 2021, conditional on the daily census numbers for the state observed from January 1, 2017, to the date of first confirmed COVID case in the state. The projection was not a point estimate. Rather, the model created thousands of alternative future realities, or counterfactuals: that is what the model thinks would have happened if COVID did not occur. From those counterfactuals, I can then derive an upper and lower bound of credibility of what would have happened sans COVID. I can then compare the observed vs counterfactual – and the difference is our estimated causal impact.
(The counterfactual is the gray area, the observed daily census numbers are in maroon, and the date of the first confirmed COVID case is the red horizontal line)
The relative cumulative effect:
The absolute cumulative effect (in terms of stay-days):
These numbers reflect a challenging SNF market in terms of occupancy. What is striking is the asymmetry of decline versus recovery; the decline was multiplicative and recovery has been additive. The implication of this asymmetry is profound not only for SNF operators but for SNFists as well.
