Expect the Unexpected: A Role for Behavioral Economics in Understanding the Impact of Cost- Sharing on Emergency Department Utilization

July/August 2010, Vol 3, No 4 - Business
Albert Tzeel, MD, MHSA, CPE, FAAPL
Regional Medical Director Senior Products
North Florida Humana, Tampa
Jack Brown, BA
Download PDF

We economists always think that waving money under people’s noses will make them behave according to some theoretical script. But we’ve tried that with healthcare for decades and it hasn’t worked. Healthcare isn’t just about financial incentives; it’s also about anxiety, fear, habit, guan-xi—a Chinese word that, loosely translated, means “family or business ties”—and professional pride. We’ve neglected to study the impact that human behavior has on the system, and I don’t think we can do much to improve it unless we address the noneconomic dimensions of healthcare issues.1 —Uwe Reinhardt, PhD

"Expect the unexpected, or you won’t find it.”2 This admonition, offered by the Greek Heraclitus more than 2500 years ago, still rings true today. Employer-sponsored healthcare coverage continues to subscribe to classical economics with respect to demand and price. Many employers providing insurance coverage believe that if they shift costs to those employees who consume medical care by increasing copayments or coinsurance, then those individuals will respond to a new price point for medical services by curbing their own demand for healthcare.3-7

Choices on whether to utilize healthcare can be made in a rational context when one does not have the pressure of a limited time horizon in which to make the decision. In an emergency situation, however, seconds count in the choice to seek care: whether a true emergency exists lies within the discretion and perception of the consumer making that decision.

O’Grady and colleagues showed that cost-shifting has an impact on emergency department (ED) use.8 Selby and colleagues showed that after introducing a copayment ranging from $25 to $35 for using the ED, utilization decreased significantly, with the largest decreases in lesser severity of illness.9 Yet, even with such cost-shifting, ED utilization continues to increase.10-12 If cost-shifting by itself cannot control increasing ED utilization, what else could be playing a role? Do people behave differently from what is predicted under classical economics? And, if so, why? Are people irrational when seeking ED services? Given the premise that people are irrational decision makers when it comes to ED utilization, and as part of a broader educational initiative to decrease ED utilization in southeastern Wisconsin, we at Humana wanted to confirm that increased cost-sharing (ie, ED copayment increases) for ED services indeed resulted in decreases in ED utilization.

Methods

Data Patterns
Early in 2002, we reviewed 2001 data for Humana Milwaukee HMO ED utilization. Through this review we identified a cohort of 797 unique members for a selfcare promotion. Each of these individuals had at least 2 nonemergent ED visits that were amenable to self-care, according to informational books we provided to all of them,13,14 as part of our study. Many in the cohort had ED copayment adjustments for a given benefit year through their employer-sponsored health insurance instituted to mitigate premium cost increases.
We followed this cohort for 3 consecutive 12-month blocks. At the end of each 12-month period, in addition to assorted demographic data, we collected information on ED utilization and on benefit changes (including EDcopayments) for members who remained with Humana throughout that entire 12-month period. Collecting this information for consecutive years allowed for direct comparison with each preceding period.

Data analysis at the conclusion of each time period showed that of the original 797 participants, 415 (52.1%) remained covered by Humana after the first year, 322 (40.4%) remained covered after the second year, and 194 (24.3%) remained Humana members when data were collected after the third year.

Statistical Analysis
We realize that observational analysis could not substitute for a well-designed study, but we were compelled to examine the findings statistically to see if the data showed something of potential interest and, if it did, to proffer a possible explanation. 

Statistical analyses were performed using Microsoft Excel 97 for Windows and XLSTAT, a Microsoft Excel add-in package.15 We assessed changes in the cohort’s ED encounters from the previous year to the current one relative to ED copayment changes, using matched pairs t-test.

Results

Demographics
As noted, study participants had to remain active Humana members for an entire 12-month block of time. Because the study period overlapped calendar years, we lost some members at the time of their employer group’s renewal (especially on January 1). Table 1 outlines the demographic data for 322 members who were active through the first 2 years, categorized by age and by sex.
Table 1
Table 1
Changes in ED Utilization Associated with Copayment Adjustments
In reviewing the data, we found a specific pattern in each year: the groups who had an ED copayment increase had an increased number of ED visits (Table 2). As shown in Table 2, in the first year, if the ED copayment increased in a given period (relative to the 12-month period before it), ED utilization actually increased by a factor of nearly 12. Given the low rate of ED utilization in the group with the increasing ED copayment in the preceding period, this finding might have simply been a case of regression to the mean. But what if we find similar patterns during the following 2 years?
Table 2
Table 2

In the second year, a similar pattern emerged. For the 2 groups whose ED copayment did not increase during this period, decreases in ED utilization occurred. However, in the third group, whose ED copayment increased during the period, we observed a significant (P = .022) and rather large increase in ED utilization—that is, an average increase of nearly 1 full ED visit for each person in that group, for a total of 46 ED visits.

A comparable pattern continued during the third year. In the groups whose ED copayment did not increase, we found either a small, nonsignificant increase or a significant decrease in ED utilization. The remaining group continued to show that when the ED copayment increased in a given period, a noteworthy, albeit not significant (a = .05, P = .09) increase in ED utilization occurred.

All 3 time frames showed the same counterintuitive pattern: ED copayment increases were associated with increased ED utilization. Although there were no individuals who had ED copayment increases in 2 consecutive periods during any of the 3 intervals, we observed consistency among unique individuals’ responses to ED copayment increases.

Discussion

Significant Implications for Employers
Our original presumption was that individuals respond to having their ED copayment increased by decreasing their ED utilization, according to classical economics. But that is not what our data revealed to us year after year. Why?

This inconsistency has tremendous implications for employer groups that provide health benefits to their employees. A given employer may choose to rely on the tenets of classical economics (ie, increasing ED copayments to promote decreased ED utilization). Yet, without a general understanding of key ideas from behavioral economics, that employer may not necessarily achieve the desired results.

Several behavioral concepts may influence this finding and promote the association that our results show for ED utilization.

Prospect Theory: Behavioral Economics Explains Emergency Care Utilization

A review of our results showed an association in which, on average, if the ED copayment increased, then ED utilization increased as well. We speculate that there is a legitimate reason for this finding, and we defer on this to behavioral economics, specifically to “prospect theory.” This theory arose from experimental proof that people repeatedly violate the principles of expected utility theory when making particular decisions.

Prospect theory was first proposed by Kahneman and Tversky in 1979.16 Although a formal discussion of prospect theory is beyond the scope of this article, 2 of its key elements are pertinent to this discussion.

First, according to prospect theory, a person multiplies his or her distinct expected utility by a subjective probability. However, individuals tend to distort those probabilities by overweighting low-probability events and underweighting high-probability events (eg, if a child only has the symptoms of fever, neck pain, and difficulty swallowing, the more likely probability of streptococcal pharyngitis may be underweighted, whereas the less likely probability of bacterial meningitis may be overweighted).

This finding would imply that the probability of a particular healthcare event (eg, vomiting in gastroenteritis, generalized fever, ear pain) being extremely serious or even life-threatening would most likely be exaggerated by the individual or the caregiver. Burns and colleagues corroborate that the overweighting of the high-risk, low-probability event tends to promote a utilization profile of increased consumption.17

Second, individuals view these expected utilities as changes from a reference point.18 With respect to “changes from a reference point,” Kahneman and Tversky propose that individuals make decisions as if they had a “value” function for gains and losses (Figure).16 The horizontal axis in the Figure reflects either monetary gain (to the right) or monetary loss (to the left) relative to one’s reference point (ie, the origin). The value function shows that individuals perceive losses as more significant than equivalent gains: the value function increases the slope as one moves to the right, until the origin, and decreases the slope as one moves to the right, away from the origin.

Figure
Figure

Baron describes this Figure as meaning “that, for sim-ple gambles, subjects tend to avoid risks in the domain of gains and tend to seek risks in the domain of losses where gains and losses are defined in terms of expected monetary change from their reference point.”18

An increased ED copayment, relative to the previous copayment, would in theory tend to make an individual “risk-seeking,” because a potential loss, coupled with an exaggerated probability for a potentially adverse event, drives ED utilization.

Perceived value versus cost.
In other words, there is more perceived value (because of the higher ED copayment/“ loss”) for that ED visit than existed before the copayment change. In addition, from a cognitive standpoint, the individual can rationalize perceived value in the decision to seek services despite a greater personal financial cost. As early as 1969, Doob and colleagues noted that according to cognitive dissonance theory, the more something costs, the more people find value in it, and the more they can feel internal pressure to buy it.19

Prospect theory’s role may perhaps be better viewed through the following example. Two families each have a young child who complains of a nonspecific sore throat on a Sunday afternoon. Each family is concerned that the child has strep throat, which may develop into rheumatic fever or worse. For family A, the most recent insurance changes resulted in a $50 decrease in ED copayment (from $100 to $50). For family B, the most recent insurance changes have resulted in a $50 increase in ED copayment (from $50 to $100).

Family A may view this event as a choice between 2 alternatives—going to the ED removes a certain monetary gain (compared with last year) for diagnosing a potentially low-probability event—estimated in the past as a probability of 11% for strep throat for all ED visits in which the chief complaint is sore throat, and as 15% to 36% in the pediatric population.20,21 Family B may also view this event as a choice between 2 alternatives—going to the ED may yield a greater monetary loss (compared with last year), but the cost of not going may be even higher—the financial cost and/or disability incurred by the child whose strep throat develops into something worse, such as acute rheumatic fever in 0.01% of all childhood throat infections.20,21

For this example, we use a cost of approximately $25,000 for each case of acute rheumatic fever (adapted from Webb and others).22 From an expected value standpoint, family A now views the decision as a choice between a certain “gain” of $50 versus only a possible “gain” of $2.50 (0.000 X $25,000). The $2.50 is considered the expected financial value for diagnosing 1 case of acute rheumatic fever; it represents the “cost” of removing uncertainty in diagnosis.

Certain versus possible gain.
This value in information gain does not include additional potential costs of the disability or of the financial impact as a result of missed time off from work. Family A is risk-averse and would wait to see if the symptoms worsen before seeking ED care, because a sure gain of $50 is preferable to a possible gain of only $2.50. Family B, however, views the choice as one of a sure loss of $50 compared with only a risk of a loss of $2.50. For them, this choice is easy—taking a chance on losing only $2.50 is significantly better than losing $50 for sure.

Studies also show that anything constraining one’s sense of an unlimited future shifts motivations and priorities away from a future orientation toward a present orientation.23 Family B, then, is “risk-seeking” and therefore seeks ED care for their child. If urgent care were substituted for ED care, neither family in our example would change its decision unless the urgent care copay/personal cost was less than $2.50. Even rapid-care clinics found in most national drugstore chains charge more than that for a typical visit. In fact, our data showed that the greater the dollar differential between ED visit copays and copays at alternate sites of care (eg, office visits or urgent care), the more likely the member was to go to the ED.

The Psychology of Time: Behavioral Economics Explains Non–Emergency Care Utilization
Another way of looking at ED-seeking behavior that would be consistent with our data is to apply the “psychology of time.” For the purposes of explaining the patterns we saw earlier, the economic curves of demand versus price perform adequately for most conditions in which the majority of the “cost” to the individual is time, because the reference value for the ED copayment did not change.

Tucker and Davison suggest that, “When monetary costs to consumers are minimal or nonexistent, time costs function to reduce and regulate demand….Once time is ‘spent,’ it cannot be recovered as one might recover a monetary loss or replace a tangible good.”24 Time, then, becomes the “currency of choice” when money is not a factor.

This may explain why, when an ED copayment change was not the key issue (ie, it either had not changed or had decreased), individuals tended to consider time their “currency” and might have chosen to use self-care strategies over seeking ED care.

In fact, probably because of time’s lack of fungibility, studies show that in hypothetical choice situations, individuals tend to be either more risk-averse or risk-neutral about potential time losses, but they tend to be more risk-seeking with respect to monetary losses.25 Although Leclerc and colleagues based their conclusions on hypothetical choices,25 these findings have tremendous implications for employer groups who provide health benefits.

A given employer may choose to rely on the tenets of classical economics (ie, increasing ED copayments to promote decreased ED utilization, or promotion of a wider gap between ED copayments and office visit copayments to drive utilization toward the physician’s office as an ED substitute). Yet, without a general understanding of, as well as an incorporation of, some basic ideas from behavioral economics, those employer groups may not necessarily achieve the results they desire.

Phelps once asked if anybody behaved as a “rational economic actor” in the healthcare market.26 The answer comes from Heraclitus’ quote at the beginning of this present article, “expect the unexpected, or you won’t find it.”2

Limitations
As with any study, there are potential limitations to interpreting our findings. First, we did not collect data on certain specific characteristics of the cohort. Oftentimes, these traits may influence patient behavior and may help to explain why ED encounter rates varied at the onset of our review period and in response to changing ED copayments. It is known, for example, that patient characteristics, such as socioeconomic status and chronicity of illness, can impact ED utilization.27,28 However, all the individuals for whom we reviewed data were insured and therefore relatively insulated from the actual cost of the ED visit (outside of an expected copayment). We would therefore expect that the results would tend to be biased toward an increased ED copayment yielding decreased ED utilization and not the other way around (assuming the effect of classical economics).

Second, a selection bias may also be noted, because the cohort from which all members were drawn came from an initial pool of “nonemergent ED utilizers.” Such a cohort may prejudice the results.29 However, just as we see a potential association in our data between ED utilization and increasing ED copayments, the door nowopens for assessing other areas of potential individualdriven overutilization with increased cost-sharing (eg, specialist office visits).

Third, we cannot state with complete certainty that we did not observe a simple regression to the mean. Regression to the mean effects, however, can take several years to manifest, and it is therefore possible that this cohort could see a return to baseline ED utilization over the next several years.30

Conclusion
Analyzing how individuals reconcile their cost for emergency care relative to obtaining emergent services can lead to the discovery of pertinent facts that can have some bearing on the control of healthcare costs. First and foremost, there is a distinct role for behavioral economics within the sphere of controlling healthcare costs. To achieve behavior changes in their covered populations, employers and payers must understand how people view medical risk, how they make decisions regarding financial tradeoffs, and how information alters these perceptions. Attempting to influence one area, such as personal financial responsibility, without fully understanding the systemic implications, can be shortsighted.

Second, although classical economic theory still exerts the most influence in health economics, to affect behavior change one must realize that people do not tend to be rational actors when it comes to their health.

After taking account of the potential impact of behavioral economics in ED or other medical care utilization, further evaluation and analysis are certainly warranted to better understand how individuals choose to respond to increasing financial responsibility in the context of overall medical care. If employers and payers want to contain increasing ED (or potentially other medical care) utilization, they will need to consider the implications of behavioral economics in their benefit design and programmatic decisions.  

Acknowledgment
Financial support for this article was provided by Humana Insurance Corporation, a subsidiary of Humana Inc, Louisville, Kentucky (doing business in Wisconsin). The funding agreement insured the authors’ independence in designing our pilot study, interpreting the data, and writing the article.

Disclosure Statement
Dr Tzeel and Mr Brown did not report any potential conflict of interest.

References

1.McCue MT. Whither sociology? Princeton economist Uwe Reinhardt says healthcare needs to account for human behavior, not just dollar signs. Managed Healthcare Executive. September 1, 2009. http://managedhealthcareexecutive.modernmedicine.com/mhe/Executive+Profile/Whither-sociology/ArticleStandard/Article/detail/121923?searchString=whithersociology. Accessed May 1, 2010.
2. Von Oech R. Expect the Unexpected (Or You Won’t Find It). San Francisco, CA: Berrett-Koehler; 2002.
3. Huskamp HA, Deverka PA, Epstein AM, et al. The effect of incentive-based formularies on prescription-drug utilization and spending. N Engl J Med. 2003;349:2224-2232.
4. Piette JD, Heisler M, Wagner TH. Problems paying out-of-pocket medication costs among older adults with diabetes. Diabetes Care. 2004;27:384-391.
5. Lohr KN, Brook RH, Kamberg CJ, et al. Use of medical care in the Rand Health Insurance Experiment: diagnosis- and service-specific analyses in a randomized controlled trial. Med Care. 1986;24(suppl 9):S1-S87.
6. Cherkin DC, Grothaus L, Wagner EH. The effect of office visit copayments on preventive care services in an HMO. Inquiry. 1990;27:24-38.
7. Simon GE, VonKorff M, Durham ML. Predictors of outpatient mental health utilization by primary care patients in a health maintenance organization. Am J Psychiatry. 1994;151:908-913.
8. O’Grady KF, Manning WG, Newhouse JP, Brook RH. The impact of cost sharing on emergency department use. N Engl J Med. 1985;313:484-490.
9. Selby JV, Fireman BH, Swain BE. Effect of a copayment on use of the emergency department in a health maintenance organization. N Engl J Med. 1996;334:635-641.
10. Young GP, Wagner MB, Kellerman AL, et al. Ambulatory visits to hospital emergency departments. Patterns and reasons for use. 24 hours in the ED study group. JAMA. 1996;276:460-465.
11. Tufts Managed Care Institute. Emergency department utilization: trends and management. November-December 2001. www.thci.org/downloads/topic1112_01.pdf. Accessed July 7, 2010.
12. US Government Accountability Office. Report to the Chairman, Committee on Finance, US Senate. Hospital emergency departments: crowding continues to occur, and some patients wait longer than recommended time frames. GAO 09-347. April 2009. www.gao.gov/new.items/d09347.pdf. Accessed June 28, 2010.
13. Pantell RH, Fries JF, Vickery DM. Taking Care of Your Child: A Parent’s Guide to Complete Medical Care. Reading, MA: Perseus Books; 1999.
14. Vickery DM, Fries JF. Take Care of Yourself. Cambridge, MA: Perseus Books; 2001.
15. XLSTAT Statistical Software. Paris, France: Addinsoft, 1995-2004.
16. Kahneman D, Tversky A. Prospect theory: an analysis of choice under risk. Econometrica. 1979;47:263-292.
17. Burns Z, Chiu A, Wu G. Overweighting of small probabilities. Prepared for Wiley Encyclopedia of Operations Research and Management Science. March 15, 2010. http://faculty.chicagobooth.edu/george.wu/research/papers/burns%20chiu%20wu%202010%20Overweighting%20of%20Small%20Probabilities.pdf. Accessed July 7, 2010.
18. Baron J. Thinking and Deciding. New York, NY: Cambridge University Press; 2000:268.
19. Doob AN, Carlsmith JM, Freedman JL, et al. Effect of initial selling price on subsequent sales. J Pers Soc Psychol. 1969;11:345-350.
20. Kaufman A, Murray D, Starita L, Brickner PW. Streptococcal sore throat followup program in a hospital clinic, New York City. Public Health Rep. 1975;90:369-372.
21. Linder JA, Bates DW, Lee GM, Finkelstein JA. Antibiotic treatment of children with sore throat. JAMA. 2005;294:2315-2322.
22.Webb KH. Does culture confirmation of high-sensitivity rapid streptococcal tests make sense? A medical decision analysis. Pediatrics. 1998;101:E2.
23. Zimbardo P, Boyd J. The Time Paradox. New York, NY: Free Press; 2008.
24. Tucker JA, Davison JW. Waiting to see the doctor: the role of time constraints in the utilization of health and behavioral health services. In: Bickel WK, Vuchinich RE, eds. Reframing Health Behavior Change with Behavioral Economics. Mahwah, NJ: Lawrence Erlbaum; 2000:219-264.
25. Leclerc F, Schmitt BH, Dube L. Waiting time and decision making: is time like money? J Consumer Res. 1995;22:110-119.
26. Phelps CE. Health Economics. NewYork, NY: HarperCollins; 1992.
27. Sun BC, Burstin HR, Brennan TA. Predictors and outcomes of frequent emergency department users. Acad Emerg Med. 2003;10:320-328.
28. Wong MD, Andersen R, Sherbourne CD, et al. Effects of cost sharing on care seeking and health status: results from the Medical Outcomes Study. Am J Public Health. 2001;91:1889-1894.
29. Zenner PA, Parke R, Mirkin DP, for Milliman. Insight into two analytical challenges for disease management. Milliman Research Report. April 2004. http://publications.milliman.com/research/health-rr/archive/pdfs/Two-Analytical-Challenges-Disease-Management-RR04-01-04.pdf. Accessed August 11, 2004.
30. Johnson A. Measuring DM’s net effect is harder than you might think. Manag Care. 2003;12:28-32.

Related Items
The Challenge of Variable Costs in Decisions Based on Cost-Effectiveness Evidence: A Case Study for Brodalumab
Diana Brixner, RPh, PhD, Gary Oderda, PharmD, MPH, Joseph Biskupiak, PhD, MBA, Douglas S. Burgoyne, PharmD, FAMCP, Steven G. Avey, RPh, MS, FAMCP, Steven R. Feldman, MD, PhD
February 2019 Vol 12, No 1 published on February 6, 2019 in Business, Case Study, Online First
Potential Cost-Savings with Once-Daily Aminomethylcycline Antibiotic versus Vancomycin in Hospitalized Patients with Acute Bacterial Skin and Skin Structure Infections
Kenneth LaPensee, PhD, MPH, Thomas Lodise, PharmD, PhD
December 2018 Vol 11, No 9 published on December 27, 2018 in Business, Original Research
Healthcare Costs and Absenteeism Among Caregivers of Adults with Partial-Onset Seizures: Analysis of Claims from an Employer Database
Richard A. Brook, MS, MBA, Krithika Rajagopalan, PhD, James E. Smeeding, RPh, MBA
November 2018 Vol 11, No 8 published on November 30, 2018 in Business, Original Research
Economic Burden of Adverse Events Associated with Immunotherapy and Targeted Therapy for Metastatic Melanoma in the Elderly
Sameer R. Ghate, PhD, Zhiyi Li, MBA, MA, Jackson Tang, MSc, Antonio Reis Nakasato, MD
October 2018 Vol 11, No 7 published on November 8, 2018 in Business, Original Research
Computed Tomography Colonography Less Costly Than Colonoscopy for Colorectal Cancer Screening of Commercially Insured Patients
Tia Goss Sawhney, DrPH, FSA, MAAA, Bruce S. Pyenson, FSA, MAAA, David Rotter, PhD, Michele Berrios, Judy Yee, MD
October 2018 Vol 11, No 7 published on November 8, 2018 in Business, Original Research
Last modified: September 6, 2013
  •  Association for Value-Based Cancer Care
  • Value-Based Cancer Care
  • Value-Based Care in Rheumatology
  • Oncology Practice Management
  • Rheumatology Practice Management
  • Urology Practice Management
  • Inside Patient Care: Pharmacy & Clinic
  • Lynx CME