Much is still unknown about the ripple effects of the coronavirus outbreak. The trajectory of the virus, the timing of the economic impact, even election policy mandates could all dramatically impact healthcare costs in the United States for years to come.
In the first of a two-part series, Leader’s Edge sat down with the Claros Analytics team—Todd Owen, President & CEO; Gary Miller, Head Actuary; and Dan Wolsk, Chief Revenue Officer—to hear how they’re approaching what previously seemed unimaginable: measuring and managing the cost implications associated with a global pandemic.
Gary: Actually, I’ve been thinking about this quite a bit. With our software, I do a good deal of user support, and it’s given me an opportunity to think through how I would build a model to calculate the impact on costs.
You have to consider the variables to create a range of assumptions. Let’s take the geographic region of New York. You have the plan, plan sponsor, where the participants are located. With that, you need to ask: what has been the impact of the virus on that particular geographic region? Clearly, if you’re in New York, the answer is going to be different compared to a client or a plan in Nebraska. Number one is to consider the location and the overall exposure of that geographic area to the virus.
Another important variable is your population. Just how vulnerable is a particular population?
The next thing I would do is think through the impact on different parts of care utilization. Take medical inpatient stays. We know elective surgeries are being put off, but there still could be some emergencies. Then you have to consider the COVID-19 specific hospitalizations in that calculation. The model will then give an estimate of what a normal, baseline utilization looks like.
The impact is going to be very different in the short term versus the long term. Right now, we’re anecdotally seeing huge reductions to normal monthly plan costs because of all of the [tabled] procedures.
That reduction in cost is going to translate into a reduction in utilization. The lower the claim level, the bigger impact the plan design has on cost. In a low-claim environment—which COVID-19 has created—the participant is actually going to be picking up all costs except for the coronavirus claims where the insurance companies have said, “we’re going to pick up 100% of those.”
When we model the long-term impact, we have to ask questions like, are we going to get back to normal, and how fast? A big assumption there is, how much capacity do the providers in that geographic region have to handle this pent-up demand? Overall, the results of those calculations are what’s known as a mature impact, or what we see on an annualized basis.
Dan: I want to highlight one point you mentioned, which I think is pretty significant: this type of analysis typically occurs on a plan-year basis. And, what you’re suggesting is that the only way to look at this is on a monthly basis, where you adjust a number of these assumptions as you go. That’s a very different way of looking at this dynamic.
What sort of assumptions do you make about changes in capacity? In pricing? [Some] providers are going into significant financial distress in this current environment. That puts demand way above supply at some point in the future.
Gary: Because of overall medical trend, prices continue to increase over time. Will that financially relieve hospitals? Will they want to renegotiate network contracts?
Gary: Everything that I have seen so far is anecdotal. Let’s say, hypothetically, we’re looking at emergency room visits. Initially, my estimate of the reduction to emergency room visits is 50%. I would create a range of 40%-60% reduction. At the end of the day, there’s always judgment involved because no estimate is perfect. The important thing is you need the ability to modify those assumptions, and the quicker the better because this information is going to be changing very, very rapidly.
Dan: What do you think this dynamic does to the decision to self-fund vs. fully insure?
Gary: It’s a good question. We’re still trying to understand the layers of risk involved. If you have a stop-loss deductible of $50,000, what will the leveraging effect be on excess claims? Longer term, we’ll most likely see an increase in stop loss policies, but claims are claims. If risks are changing for self-funded groups, they’ll change for fully-insured groups too. For fully-insured groups, you could see their community ratings be affected.
Todd: This is also an opportunity for plan sponsors who regarded choosing to be fully insured as the riskless option to consider the alternative. Fully-insured plans still pay their full premiums even though costs and claims have dropped dramatically. Self-funded groups could capitalize on this moment.
Gary: Right now, self-funded plans could be building up a surplus.
Todd: Self-funding is not a panacea, but it could be worthwhile for benefits advisors to explore that idea with plan sponsors.
Gary: It also depends on exposure of the self-funded plan sponsor. From what I’ve heard from a few clients, self-insured employers haven’t incurred a lot of costs specific to the virus. If a self-funded employer has a contract with a stop-loss carrier and decides to make changes to the plan mid-course, those changes could be denied by the carrier unless they’ve decided to cover that change.
Gary: My gut feeling is absolutely. We will see long-lasting changes when it comes to mental health. I would model that as increase to normal utilization by creating a benchmark. How would a population normally utilize mental health services? What have you seen in your data? A lot of times though, actual claim data is not credible depending on the size of your group and how finely you cut the data—if you look at just one piece of it, like inpatient stays. That’s why you would need a benchmark and to look at historical data. But based on those two things, I would expect to see an increase in mental health and substance abuse claims.
Dan: I could see it impacting and expanding the population of people accessing those services, which could deepen the mental health crisis further by potentially hampering access for those that were already using those services. I think understanding the cost implications through a model would be very complicated.
Dan: An increase in telehealth could very well interact with the behavioral health aspect we just discussed.
Gary: Telehealth is becoming more popular, but it’s historically not been a standalone variable that we take into consideration. It’s going to take a while before it becomes a significant part of overall plan costs. That may be a cultural change going forward.
The way I would model telehealth costs is by doing research around pricing and usage. How much is telehealth being used right now? What is the average cost per visit? Does it rival an in-person primary care visit? If it grows as a practice, will it take away costs from other categories, like ER or urgent care visits? It wouldn’t surprise me if there were savings to uncover going forward.
Dan: Even though we’re thinking about commercially insured populations, changes in the delivery system and the openness of providers to offer these types of services also may be impacted by Medicare policies, which are temporarily making it easier to provide telemedicine.
Gary: We focus on the commercial population, but commercial costs are highly impacted by Medicare. That public program essentially controls its own trend. Depending on how providers react to the moves Medicare makes during this pandemic, they might later pass on costs to the commercial population through government cost shifting. Medicare and Medicaid populations have been disproportionately impacted by the virus. We know fully-insured carriers are covering different types of populations.
Gary: Think of variables like mental health and provider capacity and how they might impact costs. What is the direction of the change going to be? Then, what is the magnitude? As Dan talked about before, we are looking at the cost impact on a monthly basis now to try to understand how direction and magnitude might shift.
I’ve been advising customers to create a budget for the upcoming year as if COVID-19 never happened. What would your budget rates be if you calculated them based on non-coronavirus claims? You’ll be able to calculate based on these assumptions how much of a surplus you have. You can then monitor when the utilization might be coming. There will be a transition where you will see a crossover as capacity constraints come into play. If inpatient hospital stays drop by 60% this year, they will not be able to increase by the same amount next year.
Todd: I want to add some context around one type of impact. I think this has been under-represented in the media and public discourse so far: Providers are hurting here. Effectively, if it’s not an emergency or it’s not COVID-19, it’s not happening. Even the emergency traffic is way down. Elective procedures, routine procedures, going to the labs—all of those services have halted, and yet they are where the hospitals make money and cover their costs. How can or should providers be supported? How is this going to impact how they set their pricing?
Dan: It’s not simply that we’re afraid to get care. When we’re all cooped up in our houses, we’re drastically reducing our overall need for healthcare services regardless of our anxieties about being treated for those conditions.
Todd: It’s also tough to look back at the last global pandemic—the influenza outbreak in 1918—and draw comparisons. At that time, records were kept completely differently, and people behaved and lived differently. But if we go back to other recent disruptive events, like after 9/11 or even the 2008 financial crisis, we saw from the data a notable drop in utilization of medical services as a function of those events. Now that March 2020 claims data is being reported, some clients are starting to see dramatic drop in reported claims.