Skip to main content
White Paper

The up and down impact of the pandemic on health care costs

Health and Benefits|Integrated Wellbeing
COVID 19 Coronavirus

September 29, 2020

2021 health care costs remain uncertain. This paper evaluates four potential scenarios on future care utilization in the COVID-19 environment.

Executive summary

Health care plan sponsors may see an unprecedented decrease in year-over-year medical costs in 2020, as system capacity shifts and fear of contracting COVID-19 in medical settings drives a significant volume of foregone and deferred care. Significant uncertainties remain however, including the course of the pandemic, the availability of effective vaccines and treatments, and changes in the health care delivery system that could impact future health care costs.

Willis Towers Watson has evaluated a set of potential future care utilization scenarios contemplating a variety of patterns of infection and care return. Across our scenarios, 2021 costs to employer plans are expected to be slightly higher (0.5% to 5.0%) than the non-pandemic baseline projection. Nevertheless, when 2020 and 2021 are combined, all scenarios show cost reductions (–2.8% to –3.8%) relative to the non-pandemic baseline. The baseline comparison from which these estimates were developed reflects projected costs for 2020 and 2021 assuming the pandemic never occurred. Employers should consider these scenarios as they budget for and measure the performance of their health care plans in the upcoming year.


Introduction

Health care costs have increased each year since 1960 (Centers for Medicare and Medicaid Services, National Health Expenditure Accounts [NHEA] [1960 – 2019]), when the United States began tracking the national health expenditure per capita. Due to COVID-19, it is possible that 2020 will be the first time the national health care expenditure will be lower than the preceding year. So far, the additional medical costs associated with the testing and treatment of COVID-19 have been more than offset by significant reductions in utilization across many service categories.

Because of uneven rates of infection, the exact impact on utilization and total costs varies geographically. Early on, specific areas, like New York City, saw so many cases that some hospitals breached capacity limits. In contrast, health care systems across the country saw significant utilization declines as care was forgone or deferred in anticipation of an infection surge that did not materialize. Now, as some procedures that were deferred in the early months of the pandemic are rescheduled, infections are rising in different geographies. As the infection travels, patients and providers must decide whether to postpone non-COVID-19 care again or move ahead despite substantial community transmission.

What we know now

Overall health care claims in the second quarter of 2020 were dramatically lower than in previous years. Figure 1 shows actual 2020 paid claims per employee per month (PEPM) for large employers compared with a seasonality-adjusted projection. Utilization shifts usually take one to two months to become evident in paid claim reporting due to the lag in claim processing.

Payment data shows health care claims in Q2 2020 were dramatically lower than in previous years.
Figure 1. Change in monthly paid claims per employee (2020 year to date)

Source: Willis Towers Watson 2020 Health Care Financial Benchmarks Survey (N = 208, 1.2 million employees, $1.1 billion in monthly claims)

While infection outbreaks were concentrated within certain geographies in the second quarter, the impact of care deferral is evident across all geographies, likely due to both provider closures and patient fears (Figure 2). Deferred utilization was likely highest in the Northeast, though direct COVID-19 claims partially offset the PEPM claim declines. The pandemic’s impact on medical costs will vary geographically, as local public responses and infection levels continue to differ.

Figure shows the impact of care deferral is evident across all geographies.
Figure 2. Monthly average of change in paid claim amount (by region)

Source: Willis Towers Watson 2020 Health Care Financial Benchmarks Survey (N = 208, 1.2 million employees, $1.1 billion in monthly claims)

The likelihood of deferred care returning to the system will vary based on the type of care. In fact, we expect that a significant portion of deferred care will be completely forgone and never return. Further, the return time will depend on status of the pandemic, system capacity, public policy, patient willingness to visit care settings and the urgency of the need for care. The system’s capacity limit can lengthen the time over which deferred care can be delivered, especially if capacity is constrained due to new infection surges.

The largest decreases in utilization have been in preventive care and other office-based services, as noted in Figure 3. But we have even seen decreases in emergency room visits and inpatient admissions for such emergency conditions as heart attacks and strokes.

 
Figure 3. Post-COVID-19 decrease in weekly utilization, by service*

Breaking down utilization in the presence of COVID-19

To illustrate the future impact of the pandemic on health care use, relative to a pre-pandemic normal level, we considered the following general categories of health care utilization. This segmentation is informative given that utilization within each category will impact the health care system at different times:

Direct COVID-19 utilization: Immediate care for COVID-19 patients (e.g., testing, outpatient services and hospitalizations)

Non-deferred utilization: Immediate care for non-COVID-19 patients that can’t be deferred (e.g., appendectomy)

Deferred utilization: Generally, non-urgent care for non-COVID-19 patients that can be deferred and returns at a later point (e.g., knee replacement, vaccinations)

Forgone utilization: Non-urgent care for non-COVID-19 patients that will never be delivered (e.g., urgent care for self-limited disease or periodic visits for chronic disease)

Figure 4 lists several utilization factors that we expect will drive changes in health care use throughout the pandemic:

A variety of possible utilization scenarios may drive changes in health care use throughout the pandemic.
Increased use Decreased use
Health care delivery system
  • Hospitals may prioritize higher-cost procedures to drive up unit cost even as use falls
  • Telemedicine might supplement, rather than replace, in-person care
  • Physicians who are income-targeting might order more tests/interventions
  • Provider closures/mandated restrictions
  • Limited hospital capacity and decreased staffing available for non-COVID-19 conditions
  • Reduction in health care worker productivity
Patients
  • Cost of diagnosing and treating COVID-19
  • Return of deferred care
  • Excess demand/complexity due to preventive or diagnostic care missed
  • Concern over loss of employer coverage
  • Increased need for mental health services
  • Fear of contracting COVID-19 in a care setting
  • Inability to travel to see your provider (public transit/ride-sharing services closed, children at home)
  • Economic uncertainty often leads to decreased utilization
  • Some services are no longer needed (urgent care for a resolved illness) or no longer appropriate (curative surgery for someone with cancer that has spread)

Possible utilization scenarios

As a result of the recent broad, negative economic conditions, many employers are looking for their health care plans to deliver value at even lower cost. Designing plans that lower employer costs without shifting expense to members or sacrificing quality requires an assessment of future health care utilization patterns. Those patterns of use will be a function of care for COVID-19 as well as care deferred or foregone.

We have constructed the following utilization scenarios employers should consider as they design health care plans in a COVID-19 environment:

Scenario A: Early control. This scenario assumes the wave we saw in March/April was a one-time event. Thanks to mitigation efforts, COVID-19 remains at a manageable level going forward. A substantial portion of the care deferred from the initial wave takes place later in 2020 and in the beginning of 2021.

Scenario B: Variable hot spots. While the flu typically emerges in discrete waves, COVID-19 may not. This scenario assumes, at least until a vaccine appears, that some level of COVID-19 infection persists. Hot spots may appear and disappear within certain geographies, but the national infection level remains flat. Patients and providers adjust to this “new normal.”

Scenario C: Widespread new infections. COVID-19 remains at manageable levels through the summer, but we see a second wave in late 2020 (perhaps exacerbated by the seasonal flu or school reopenings) that is as bad or worse than the first wave. After this second wave, some level of infection persists as in Scenario B above.

Scenario D: Widespread new infections with multiple peaks. As in 1918, the first wave turns out to be small when compared with the second wave that hits in the fall and winter. A third and final wave occurs in the spring. After this third wave, some level of infection persists as in Scenario B above.

We compare projections of costs in the presence of the pandemic under four different scenarios to a common, non-pandemic baseline projection, which increases over time due to normal health care cost trend, as illustrated in Figure 5.

Cost projections in the presence of the pandemic will likely increase over time due to normal health care cost trends.
Figure 5. Baseline projection of cost. Magnitude and duration vary across scenarios

Our scenarios differ only with respect to their assumed future utilization patterns. Several factors that could impact cost of care are not considered in this model:

  • We have not assumed future changes in unit cost beyond normal health care cost trend. The pandemic has had an enormous impact on the health care delivery system, which could well lead to failures of some lower-cost providers and market consolidation that could lead to higher unit costs.
  • We did not contemplate the cost impact of shifts to virtual care or changes in the utilization of certain service categories, such as increases in use of services associated with behavioral health.
  • We did not incorporate the cost or impact of new treatments or vaccinations that might become available. A potential vaccination would not only impact cost but also impact the relative likelihood of each of the scenarios.

Scenario A: Early control

This scenario assumes that both the COVID-19 cases and the health system response peaked in April and will be effectively gone by the fourth quarter of 2020. Of course, evidence is already emerging that makes this scenario seem unlikely, but it provides a baseline with which to measure other alternatives.

In this single-wave scenario, shown in Figure 6, the cost of COVID-19 care (yellow) has a meaningful effect on total medical spend, but it is offset in 2020 by the massive decline in claims due to deferred and forgone care (magenta). Beginning in April, some of the deferred care (blue) begins to return until the pent-up demand is satisfied around year-end. The black dashed line represents the impact of all these factors on total medical spending.

In a single-wave scenario, the cost of COVID care would have a meaningful effect on total medical spend.
Figure 6. Scenario A: Early control

Scenario B: Variable hot spots

This scenario assumes a continued presence of COVID-19 through at least 2021, with no effective herd immunity achieved through broad infection or through vaccination. It can be pictured as the continued reemergence of infection hot spots in different geographies, prompting public response and care deferral until the situation improves.

As this pattern settles in, we see the emergence of a new normal, where utilization has shifted downward. In the absence of herd immunity, which may only come to pass after several years, and while fear of care settings persists, this reduction could lead to higher-value health care delivery, as we prioritize the care that yields the most benefit. However, it’s also possible that the care that is not delivered includes substantial amounts of high-value care, so we could achieve cost savings while gaining less benefit from our medical care system (Figure 7).

A variable hot spots scenario in which utilization has shifted downward could lead to higher value health care delivery.
Figure 7. Scenario B: Variable hot spots

Scenario C: Widespread new infections

Several factors make a second wave possible in the fall/winter of 2020: increased hours spent indoors, fatigue from prevention measures, reopening of schools and students traveling to universities, and the potential that seasonal influenza will make diagnosis and treatment more difficult.

Even if new COVID-19 cases rise to the levels seen in April, it’s unlikely the response will be as uniform. Patients have had months to be desensitized to news about rising cases, and providers have learned how to take precautions and remain partially open for non-COVID-19 care. Responses will also likely be more limited to the local infection landscape rather than a national elective procedure shutdown as seen in March/April (Figure 8).

If we experience a second wave if widespread new infections, it’s unlikely the response will be as uniform.
Figure 8. Scenario C: Widespread new infections

Scenario D: Widespread new infections with multiple peaks

This final scenario illustrates the impact of a second, severe infection wave in the fall of 2020 followed by a slightly less severe third wave in the winter, like the 1918 pandemic. No specific evidence suggests the COVID-19 infection pattern in the United States will follow this model; however, it is worth illustrating as it is commonly referenced as a comparator for COVID-19 (Figure 9).

This figure shows 1918 influenza pandemic death toll
Figure 9. 1918 influenza pandemic death toll

Once again, utilization reductions outweigh increases as infection waves emerge. In this scenario, the backlog of deferred care grows with each wave, increasing the volume of deferred care that is unlikely to return. The combination of excess demand from 2020 and excess COVID-19 claims in 2021 worsens the case for employers in 2021. Still, even in this example, claims in 2021 only rise about 5% above pre-pandemic projections for 2021 (Figure 10).

Experiencing widespread new infections with multiple peaks would increase the volume of deferred care that is unlikely to return.
Figure 10. Scenario D: Widespread new infections with multiple peaks

What does this mean for employers?

Economic pressures challenge employers to demand performance and value from their health plans, even in a low or negative cost trend environment. The current cost reductions driven by COVID-19 are not due to improved system efficiency or improved health outcomes. Significant underlying challenges remain, and employers will need to continue to work to overcome them.

In all the scenarios we considered, costs for 2020 are projected to be below baseline. COVID-19 may be driving the first reduction in per capita health care costs in the United States that we have ever witnessed. As we assess 2021, the return of deferred care and/or the continued presence of infections is likely to increase costs — but not by enough to offset the potential decrease from 2020. If we consider both years together, every scenario predicts total costs to decrease between 2.8% and 3.8% (Figure 11).

Figure 11. Projected impact of COVID-19 on per capita health care costs

*All impacts shown relative to typical utilization baselines in the absence of the pandemic.
2020 impact* 2021 impact* Two-year net impact*
Scenario A –3.3% +0.5% –2.8%
Scenario B –3.9% +1.2% –2.7%
Scenario C –5.8% +2.7% –3.1%
Scenario D –8.8% +5.0% –3.8%

In some respects, this could be viewed as a reserving exercise rather than a pricing/projection one. Some share of the claims that were expected to be incurred in 2020 were deferred. Depending on the scenario, we don’t know if they will return in late 2020 or in 2021, nor do we know what share of the deferred claims will return at all. Nevertheless, a reserve for a portion of expected costs not incurred in 2020 may be appropriate to offset costs for that care returning in 2021.

The impact of COVID-19 will differ based on geography and industry. One organization may experience the “early control” scenario, while others weather multiple waves of infections and closures. If ever there were a time to gain a deeper understanding of your demographics and your claims, as well as implement an analytic strategy to identify problems and monitor the effectiveness of solutions, it is now. In the absence of reliable macro-market trends, employers will need to rely increasingly on more detailed analyses of their own data and benchmarks customized to their profile. A comprehensive analytic approach from data input, through an analytic framework, to actionable output will reveal opportunities for employers to drive value to participants.

It will not be sufficient for employers simply to ignore 2020 and wait to “return to normal.” COVID-19 is driving broad change and volatility, which demands effective measurement. During this period of transition there will be many changes to the way business, health care and employee benefits function. Some of these (mask wearing, temperature checks, provider closures) will cease when the pandemic is over, while others (telehealth, remote working, provider restructuring) might endure. Employers will need to understand the rapidly changing health care market landscape as well as the shifting needs and risk profiles of their workforces.

Methods, assumptions and sources

The results of the model are meant to represent the national average of incurred medical and pharmacy employer health spend. Actual employer results will vary significantly by geography, population density, industry and demographics. This model does not attempt to consider the long-term effects from the pandemic (e.g., shift to telehealth, provider contracting, hospital consolidation, long-term health effects for those with severe COVID-19, changes to population mortality/morbidity).

All assumptions in the model are benchmarked to the national response to COVID-19 in April 2020. Results have been calibrated to align with actual emerging claims experience, as reported above.

Decreased utilization by scenario

Figure 12. Projected health-care utilization rates based on possible infection scenarios through 2021

Utilzation and costs by scenario are benchmarked to the national response to COVID-19 in April 2020.
2020 2021
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Scenario A 0.00 0.00 0.65 1.00 0.50 0.31 0.28 0.17 0.12 0.07 0.03 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Scenario B 0.00 0.00 0.65 1.00 0.50 0.31 0.28 0.17 0.15 0.14 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03
Scenario C 0.00 0.00 0.65 1.00 0.50 0.31 0.28 0.17 0.33 0.67 0.30 0.15 0.12 0.09 0.08 0.07 0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03
Scenario D 0.00 0.00 0.65 1.00 0.50 0.31 0.28 0.17 0.29 0.96 0.63 0.38 0.10 0.65 0.38 0.19 0.14 0.06 0.05 0.05 0.04 0.04 0.03 0.03
Contact Us


Each month, we assume both patients and the health system are 10% less sensitive to the infection rate. In other words, the number of new cases would need to be about twice as high in November 2020 to drive the same level of deferred claims we saw in April. By applying this scaling factor, we can impute the underlying COVID-19 claim rate (relative to April).

Figure 13. Projected COVID-19 claim rate based on possible infection scenarios through 2021

Utilzation and costs by scenario are benchmarked to the national response to COVID-19 in April 2020.
2020 2021
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Scenario A 0.00 0.00 0.65 1.00 0.56 0.38 0.39 0.25 0.20 0.13 0.07 0.04 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Scenario B 0.00 0.00 0.65 1.00 0.56 0.38 0.39 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Scenario C 0.00 0.00 0.65 1.00 0.56 0.38 0.39 0.25 0.56 1.25 0.63 0.35 0.30 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Scenario D 0.00 0.00 0.65 1.00 0.56 0.38 0.39 0.25 0.49 1.80 1.31 0.87 0.27 1.85 1.19 0.66 0.54 0.27 0.25 0.25 0.25 0.25 0.25 0.25

COVID-19 costs of COVID-19 care by scenario

This model recognizes the time required for patients and providers to recover from shocks and the time course of COVID-19 infections and treatments. The level of utilization reduction in each month is indexed to the impact seen in April (1.00). We assume a 40% reduction in that month, a 15% reduction the following month and a 5% reduction two months later. This reflects both the time it takes to close/reopen/restaff and the time required for patients to feel comfortable returning. These reductions will compound. For example, we assume April claims to be 1.00 * (1-40%) = 60% of baseline, while May (indexed at 0.50) is (1-40% * 0.50) * (1-15% * 1.00) = 68% of baseline. COVID-19 claims follow a similar rule, though costs are additive: (4% * 0.50) + (2% * 1.00) = +4% in May.

Figure 14. Compounding claims reductions

Claims reduction COVID-19 claims
Month 0 –40% +4%
Month 1 –15% +2%
Month 2 –5% +0%
Month 3 0% 0%

Pent-up demand follows a Markov model, based on the new claims deferred each month. Thirty percent of these missed claims are assumed to be forgone outright (e.g., condition resolves, care no longer needed, patient decides to wait until next appointment). The other 70% is added to the pool of “pent-up demand.” Each month, 50% of the pent-up demand is assumed to be delivered, to a maximum of 20% of health system capacity. This is shown as “R.” The other 50%, along with any excess that is not accommodated that month, is rolled over to the next month (representing the delay in rescheduling care). The rest [1-R] is “unmet demand.” Ninety percent of this will be rescheduled to the next month, but 10% will be added to the forgone care. As a result, the longer pent-up demand goes unmet, the more likely services will not be delivered.

The figure shows a pent-up demand model, based on the new claims deferred each month.
Figure 15. Markov model for deferred claims

Disclaimer

In preparing the results presented in this report, we have relied upon information from external sources and publications. We have reviewed this information for overall reasonableness and consistency but have neither audited nor independently verified this information. The accuracy of the results presented in this report is dependent upon the accuracy and completeness of the underlying information. While we consider the results shown in this report to be reasonable financial results, a different set of results could also be considered reasonable based on a range of reasonable assumptions for each measurement.

Footnotes

*Sources:
(1) Retail/Mail order Rx, 30-day-equivalent scripts (2020 data shown relative to same week in 2019)
IQVIA National Prescription Audit

(2) ED visits (2020 data shown relative to baseline week March 1 – 7, 2020)
CDC Morbidity and Mortality Weekly Report
Impact of the COVID-19 Pandemic on Emergency Department Visits — United States January 1, 2019 – May 30, 2020
https://www.cdc.gov/mmwr/volumes/69/wr/mm6923e1.htm (June 12, 2020)

(3) Breast cancer screenings (2020 data shown relative to mean weekly screening volume 2017 – Jan 19, 2020)
Epic Health Research Network
Delayed Cancer Screenings — A Second Look
https://ehrn.org/delayed-cancer-screenings-a-second-look/ (July 17, 2020)

(4) In-person ambulatory visits (2020 data shown relative to baseline week March 1 – 7, 2020)
Commonwealth Fund
The Impact of the COVID-19 Pandemic on Outpatient Visits: Practices Are Adapting to the New Normal
https://www.commonwealthfund.org/publications/2020/jun/impact-covid-19-pandemic-outpatient-visits-practices-adapting-new-normal (June 25, 2020)