Health Policy Brief September 2019 Where Do Patients Go? How Patients Choose Between Care Settings for Minor Illnesses and Injuries Dana B. Mukamel, Alpesh Amin, Heather Ladd, and Dara H. Sorkin ‘‘ As ... managed care models emphasizing cost efficiencies have evolved, so have SUMMARY: Having a bad cold with high fever, chills, and headache on the weekend. Waking up in the middle of the night with stomach cramps, nausea, and diarrhea. These are examples of medical conditions that in the past might have sent patients to the emergency room if they felt they could not wait until their their responses, we created a publicly available simulation model (https://www.medicine.uci.edu/ iteqc/wheredopatientsgo) that allows users to understand what care settings are most likely to be chosen by different populations when faced with different clinical situations. The model can inform policymakers and researchers in their alternative care doctor was able to see them during normal efforts to understand the complex relationships ’’ office hours. That was then. Today, patients between out-of-pocket costs, wait time, severity settings. have alternatives: Urgent care centers, retail of the medical condition, and population clinics, virtual physicians who diagnose over the characteristics that influence the choice of care internet, nurse advice lines, and­ in some major — setting. The general findings of the study are cities—physicians who make house calls. But that patients are more likely to choose less costly what care settings are patients likely to choose? care settings and settings with less wait time, To answer this question, we surveyed more than but that cost is usually more important than wait 5,000 employees at the University of California, time. Both those choices depend on the severity Irvine campus, a diverse population representing of the injury or the disease. Travel time does not all socioeconomic strata. Based on an analysis of affect choice. P atients seeking care for minor illnesses (such as colds, eye infections, or skin rashes) or for minor injuries (such as sprains The most established of these alternatives are urgent care centers and retail clinics. Urgent care centers—based in hospitals, stand-alone or cuts) have traditionally sought care at their clinics, and sometimes physicians’ offices physician’s office or a hospital emergency —offer evening and weekend hours and are room (ER). As the health care market has thus available to patients at times when become more competitive and managed care most physicians are not. These centers also models emphasizing cost efficiencies have accept walk-ins—that is, patients who do not evolved, so have alternative care settings. have a long-standing relationship with the These alternatives tend to be much more physician. There are an estimated 7,400 to accessible to patients in terms of both 8,100 urgent care centers nationally,1,2 and geography and hours of operation. They often the number continues to grow. The current are also less expensive and are therefore more growth rate is estimated at 300 to 600 new attractive to patients. centers per year, and growth is expected to continue at the same rate.1 In addition to 2 UCLA CENTER FOR HEALTH POLICY RESEARCH physician care, the centers may offer services $233 million, representing an 89% growth in such as lab testing, X-rays, and other “low- revenue and a 54% growth in visits compared end” diagnostics. with 2016.9 Retail clinics also usually provide evening Another care setting option that has recently and weekend service. These tend to be located been developed revives the old practice of in geographically accessible locations such as offering physician home visits.10,11 Heal, supermarkets, pharmacies, retail malls, and the first company to offer this service, has large box stores. Unlike urgent care centers, a consumer-facing application that for $99 retail clinics do not all have a physician lets patients schedule a family doctor or on site. They are often staffed by a nurse pediatrician home visit within 20 to 60 practitioner or a pharmacist, depending minutes from the time they place the order, on the services they offer. The number of between the hours of 8 a.m. and 8 p.m.12 The retail clinics is estimated at around 2,000,3 company was started in Los Angeles in 2015 with visits estimated at 6 million per year.4 and has since expanded to 16 other large The most well-known retail clinics are the metropolitan areas. ‘‘ estimated 1,100 CVS Health MinuteClinics.5 How do patients choose between care- In addition to In addition to these brick-and-mortar settings, patients can seek care from several setting options? Not much is known about how patients ... brick-and- virtual settings. Nurse advice lines­ through — choose care settings. Very few studies have mortar settings, which a nurse can be reached by phone, addressed this question, and these studies patients can seek mostly for triage purposes—have been available for many years. Originally started have tended to examine partial choices— namely, one of the new alternative settings care from several for pediatric care,6 these lines now target for care, such as a retail clinic, compared ’’ virtual settings. adult patients and are included in many insurance packages. to the traditional physician’s office.13 These studies were also limited to small convenience samples14 or were conducted in Two more recent options include “virtual other countries15,16 and may therefore not be physician visits,” in which patients contact applicable to patients in the U.S. physicians over the internet using their computer, tablet, or smartphone,7 and home This scant literature does not offer much visits by physicians that are arranged via information for policymakers. Even though the internet. Virtual visits can be conducted several studies suggest that costs at these from any location where there is an internet settings might be lower than at physicians’ connection, thus offering extreme geographic offices and ERs,17,18 it is not clear whether flexibility. Extended hours are also offered, patients are inclined to choose them when adding to the attractiveness of this care they need care.19 Studies suggest that the setting. Further, information technology market penetration of these care settings today includes the use of high-resolution is slow and that it is concentrated among cameras that make it possible for physicians younger20 and, in some cases, high-income to diagnose and treat a large number of patients.21 ailments, such as skin conditions requiring visual inspection of the patient’s skin. Should policies be implemented to encourage faster adoption of these The leading company in this space is alternative settings for care? Teladoc.8 In 2017, the company had more The answer to this question depends on the than 23 million members, each of whom cost and quality of the newer care settings made an average of 1.5 virtual visits per relative to the traditional physician’s office year. Its revenues in that year exceeded and the ER. However, the extant literature UCLA CENTER FOR HEALTH POLICY RESEARCH 3 does not provide an unequivocal answer. patients from choosing a particular care Some studies find that costs are lower and setting. Thus, patients choosing between quality is better in urgent care centers, retail alternative care settings would be expected clinics, and virtual physician visits, but only to choose the one closest to their residence some of the time, for some segments of the or the one easiest to get to, all else being populations, and some diagnoses.22 Yet, equal. many insurers are acting on the assumption that beneficiaries should be encouraged to To quantify the strength of these levers, seek care in these settings and are adopting one typically would calculate “elasticities,” benefits designed to incentivize such self- namely, the response that a change in each referrals.23,24 one of these levers creates in self-referral behavior. For example, we might ask how Can policy influence the choice of care many patients in a given population with setting? a set of prespecified characteristics (such To determine how policy can effectively as age, gender, and other socioeconomic influence the choice of care setting, we characteristics) will decide to self-refer to need to first identify the policy levers that urgent care centers and retail clinics instead are likely to change self-referral patterns of a physician’s office if the out-of pocket when patients seek care for minor illnesses costs at the physician’s office were to increase ‘‘ and injuries, and then quantify the change by 20%. required in these levers to bring about the desired outcomes. Predicting where patients will go. We conducted We conducted a large survey, with more Economic theory suggests three such policy than 5,000 respondents, to understand a large survey levers: what drives patients’ care-setting choice. to understand 1. The patient’s out-of-pocket costs. For The study specifics are described in the next what drives section. Based on the findings, we developed patients who are uninsured or who have patients’ care- ’’ a publicly available simulation model that not yet met their insurance deductible for the year, the market price equals the out- predicts how any given population with setting choice. specific sociodemographic and economic of-pocket cost. For all others, the out-of- characteristics will choose to seek care pocket cost is a significantly lower fraction when faced with several different injuries of the market price, typically ranging or diseases and varying out-of-pocket costs from 20% to 50%. or wait times. This simulation offers users 2.The wait time for an appointment the opportunity to gain insight into how until the patient can be seen and a population defined by them will behave receive care. For example, for an illness under different price and wait time scenarios. or injury that occurs on the weekend, when physicians’ offices are closed, the The simulation allows for the prediction traditional options are either to wait of care-seeking patterns for the following until Monday or go to the ER. All new assumptions/scenarios: alternatives offer care on evenings and/or •Choice of populations weekends, which may be more attractive to patients. •Ten different clinical scenarios 3. Travel time to the care setting. Just as •Eight care settings wait time might be a barrier to choosing •Three different price levels for out-of- a care setting, having to travel a long pocket costs distance—or to have a long travel time because of heavy traffic—could deter •Three different wait-time levels 4 UCLA CENTER FOR HEALTH POLICY RESEARCH Exhibit 1 Descriptive Statistics Respondents UCI Population (N=19,449) N % N % Gender (N=5,201) Female 3,439 66.1 11,378 58.5 Male 1,743 33.5 8,071 41.5 Other 19 0.4 NA NA Age (N=5,202) 18-24 828 15.9 4,880 25.1 25-39 1,906 36.6 6,699 34.4 40-64 2,236 43.0 7,197 37.0 65 + 232 4.5 673 3.5 Race/Ethnicity (N=4,884) White 2,030 41.6 7,189 37.0 Asian/ Pacific Islander 1,472 30.1 6,064 31.2 Hispanic 976 20.0 4,343 22.3 African American or Black 108 2.2 622 3.2 Other 298 6.1 1,231 6.3 Education (N=5,200) Graduate/professional degree (MBA, MS, MD, PhD, etc.) 1,934 37.2 College degree (BA, BS) 1,660 31.9 Associate’s degree / Post-high school training / Some college 1,249 24.0 but no degree High school degree / High school equivalency / Did not com- 357 6.9 plete high school Percent of Federal Poverty Level (N=4,363) 400% + 2,554 58.5 300-399% 718 16.5 150-299% 632 14.5 0-149% 459 10.5 Speak English At Home (N=4,958) Yes 4,484 90.4 No 474 9.6 Marital Status (N=5,197) Married or live with partner 3,096 59.6 Divorced, separated, or widowed 509 9.8 Never married 1,592 30.6 Rating of General Health (N=5,199) Excellent or Very Good 3,465 66.6 Good 1,457 28.0 Fair or Poor 277 5.3 Has a Personal Doctor (N=5,121) 4,435 86.6 Aware of Provider Type (N=5,200) Urgent Care 5,032 96.8 Retail Clinics or Minute Clinics 2,306 44.4 Virtual physician visit or Teladoc or visit with physician by 3,295 63.4 smartphone or video chat Nurse Advice Line 3,695 71.1 Physician visit at your home 2,843 54.7 Received Medical Care from Provider Type in Past 12 Months Emergency Room (N=5,084) 580 11.4 Physician’s Office (N=5,107) 4,277 83.8 Urgent Care (N=5,090) 1,336 26.3 Retail Clinic (N=5,083) 508 10.0 Virtual Physician Visit (N=5,075) 155 3.1 Physician Home Visit (N=5,075) 11 0.2 Called a Nurse Advice Line in the Past 12 Months (N=5,082) 686 13.5 All comparisons between the analysis sample and the UCI population are statistically significant at the 0.001 level except for the Asian and other categories of race, which are 0.097 and 0.565, respectively. NOTE: Because of rounding, some total may not equal 100%. UCLA CENTER FOR HEALTH POLICY RESEARCH 5 The simulations will provide users with graphs depicting how many people would choose each care setting under the conditions the user selects. This will allow users to download and print their data. It will also allow users to change the assumptions of their scenario, thus giving a sense of which vary more, patients would include travel time when making decisions on care settings. Summary and Conclusions The self-referral patterns of patients experiencing minor injuries or illnesses have changed in response to the availability of new ‘‘We found that the most important policy lever is out-of-pocket levers are more important than others. care settings that offer increased convenience costs. Wait time in terms of geography and time. Insurers, also affected ’’ The site has limitations: interested in keeping costs down, have •The simulations are approximate and are encouraged these trends by adjusting benefits patients’ choice. not intended as an actual planning tool. and copayment structures to make these alternatives financially attractive compared to •Simulations reflect the findings from our the traditional physician’s office. Yet patients study and are therefore subject to the same are slow to respond, and many continue to limitations as the study. In particular, make the traditional choice of going to a they reflect health care–seeking behaviors doctor’s office or an ER. The trends away and perceptions of individuals living in from the traditional choices may increase in Southern California in late 2016. To the the future, if and when patients become more degree that individuals in other parts of comfortable with alternative settings for care the country have different behaviors in and the services these provide. Insurers may seeking health care or that perceptions have be able to accelerate these trends by further changed since the end of 2016 (when the changing benefit structures to make these survey that underlies the simulations was care settings even more financially attractive conducted), the estimates provided by these to patients. simulations may be less accurate. Methodology Study Findings During the fall of 2016, we conducted a survey of We found that the most important policy all employees at the University of California, Irvine, lever is out-of-pocket costs. The less costly a a large campus with more than 20,000 employees. care setting is, the more likely patients were The university encompasses many schools, including those dedicated to arts and sciences, medicine, law, to seek care in that setting. This finding, education, engineering, and nursing, as well as large however, was modified by the medical programs in public health and pharmacy. It has condition for which the patient sought care. its own police force, housing services, hospitality Wait time also affected patients’ choice: The services, retail shops, transportation, janitorial and longer the wait time, the less likely patients groundskeeping divisions, building maintenance were to choose that care setting. This finding and engineering, child care services, retail shops, and other entities. It employs individuals from all walks was also tempered by the severity of the of life, from highly paid physicians and lawyers to medical condition, with patients willing to lower-income workers, some of whom may not speak pay more and less willing to face a long wait English. It can be viewed as a “mini city” that has a time as the perceived severity of the condition population as diverse as the population of Southern increased. California, where the university is located. On the other hand, we did not find that While the university’s population is not representative of the nation as a whole, our findings travel time mattered to patients at all. This have been weighted to account for distributional finding might be due to the fact that all our differences between our survey respondents and the survey repondents were urban, and travel U.S. population. Weighting was done for differences times did not vary substantially (up to 30 in age by gender, race/ethnicity, education, income, minutes). It is possible that in rural areas household size, and regions of the country, using the where travel time tends to be longer and to Random Iterative Method. 6 UCLA CENTER FOR HEALTH POLICY RESEARCH Exhibit 2 Where Do Patients Go? Inputs Decision Model Outputs Severity Illness/Injury Timing Patient Care Setting Characteristics Physician’s Office Age Emergency Room Gender Patient’s Urgent Care Race/Ethnicity Decision Retail Clinic Education Virtual Physician Visit Income Nurse Advice Line Marital Status Self-care/ Wait & See Language Spoken at Home Policy Levers Out-of-Pocket Costs Wait Time Travel Time ‘‘ The trends away from the traditional choices (going to a doctor’s office Of the 21,037 employees (including student employees and retirees) who were invited to participate in the survey, 5,451 (26%) responded. Of those, 4,855 had complete data included in the analyses presented. All respondents were given a series of 10 scenarios. Each scenario depicted a clinical situation describing of day and the day of the week (i.e., working hours versus evening or weekend), out-of-pocket costs, wait time, and travel time. Respondents were asked to choose the care setting they preferred for each scenario. They were also asked about their socioeconomic status and the language they spoke at home. A description of the or the ER) may the onset of an illness (such as a cold or eye redness), a chronic condition that required monitoring, sample studied appears in Exhibit 1. increase in the stomachache with diarrhea, the need for a flu The data were then analyzed statistically. For each future, if and immunization, allergies, the need for a physical examination, or an injury (such as a bad a cut that scenario, we estimated models that predicted simultaneously the likelihood that a patient would when patients might require stiches, a bad fall with a suspected broken bone, or a burned hand). We also included choose any of the eight care settings, given out- of-pocket costs, wait time, and travel time. For become more chest pain, which we expected all respondents to example, for the scenario that described having comfortable. ’’ recognize as a symptom that would necessitate treatment in the ER. In addition to a description of the clinical condition, a bad stomachache with severe diarrhea in the middle of the night, we estimated a model that predicted how likely it was for a patient to go to the physician’s office, the ER, an urgent care center, all scenarios included information about the time UCLA CENTER FOR HEALTH POLICY RESEARCH 7 or a retail clinic; to call a virtual physician on a Funder Information smartphone or a nurse on the advice phone line; to The survey underlying the simulation and the request a house call by a physician; or to just wait development of the simulation were funded by the and see whether the condition would resolve on its Robert Wood Johnson Foundation (grant #73053). own. By estimating these models simultaneously, we The views expressed here do not necessarily reflect mimicked the actual choice that patients make when the views of the Foundation. they consider all the choices they have at the same time and trade them off, one against the other, based Author Information on the characteristics of each care setting and of cost Dana B. Mukamel, PhD, is a professor of Medicine, and time. Public Health, and Nursing in the Department of Medicine, Division of General Internal Medicine The models we estimated also took into account the at University of California, Irvine, and director characteristics of the patient: age, gender, education, of the iTEQC Research Program at University of income, race/ethnicity, marital status, and whether California, Irvine. Alpesh Amin, MD, is the Thomas or not English was spoken at home. We did not & Mary Cesario Chair of Medicine and a professor include insurance status. This was not necessary, as and the executive director of the Hospitalist we explicitly told the respondents what assumptions Program, Department of Medicine at the University to make about their out-of-pocket costs. of California, Irvine. Heather Ladd, MS, is a senior statistician in the Department of Medicine, Division Exhibit 2 depicts the underlying model and the of General Internal Medicine, iTEQC Research assumptions we made. The model assumes that Program, University of California, Irvine. Dara H. the self-referral decisions made by patients are Sorkin, PhD, is an associate professor of medicine dependent on three types of inputs: the specific and associate director of the iTEQC Research illness or injury and its timing, the patient Program at the University of California, Irvine. characteristics, and the policy levers. These are combined in the decision model, which we Suggested Citation estimated statistically from the survey data, to give the outputs that are the probabilities of self- Mukamel DB, Amin A, Ladd H, and Sorkin DH. referrals to each care setting. (The specific models 2019. Where Do Patients Go? How Patients Choose and further description of the methods can be found Between Care Settings for Minor Illnesses and Injuries. in the manuscript “Patients’ Preferences Over Care Los Angeles, Calif.: UCLA Center for Health Policy Settings for Minor Illnesses and Injuries.”22) Research. Endnotes 1 Urgent Care Association of America. 2017. Industry FAQs. http://www.ucaoa.org/?page=IndustryFAQs. Accessed We hope to get your feedback June 29, 2017. on the website. 2 Weinick RM, Bristol SJ, DesRoches CM. 2009.Urgent Care Centers in the U.S.: Findings From a National Survey. BMC Health Services Research.9:79. We invite you to visit our website 3 Ashwood JS, Gaynor M, Setodji CM, Reid RO, Weber (https://www.medicine.uci.edu/iteqc/ E, Mehrotra A. 2016. Retail Clinic Visits for Low-Acuity Conditions Increase Utilization and Spending. Health wheredopatientsgo), where we have Affairs (Project Hope) 35(3):449-455. included a short survey asking for your 4 Mehrotra A, Lave JR. 2012. Visits to Retail Cinics Grew opinions about the simulation and its Fourfold from 2007 to 2009, Although Their Share of Overall Outpatient Visits Remains Low. Health Affairs usefulness. We would greatly appreciate (Project Hope) 31(9):2123-2129. your feedback, which will help us 5 Sarasohn-Kahn J. July 25, 2016. Retail Clinics Continue to Shape Local Healthcare Markets. http://www. in developing future projects and healthcareitnews.com/blog/retail-clinics-continue-shape-local- ensuring that they are more relevant healthcare-markets. Accessed June 29, 2017. 6 Marklund B, Strom M, Mansson J, Borgquist L, Baigi A, and useful for policymakers. We would Fridlund B. 2007. Computer-Supported Telephone Nurse also appreciate direct feedback to Triage: An Evaluation of Medical Quality and Costs. Journal of Nursing Management 15(2):180-187. dmukamel@uci.edu. 7 Uscher-Pines L, Mehrotra A. 2014. Analysis of Teladoc Use Seems to Indicate Expanded Access to Care for Patients Without Prior Connection to a Provider. Health Affairs (Project Hope) 33(2):258-264. 8 Teladoc Health. About Our Company - Teladoc 2017. http://s21.q4cdn.com/672268105/files/doc_ financials/2017/2017-TDOC-Annual-Report.pdf. Accessed August 15, 2019. UCLA CENTER FOR HEALTH POLICY RESEARCH 10960 Wilshire Blvd., Suite 1550 Los Angeles, California 90024 9 Teladoc Health 2018. Teladoc Announces Full-Year 17 Gordon AS, Adamson WC, DeVries AR. 2017. and Fourth-Quarter 2017 Results. http://ir.teladoc.com/ Virtual Visits for Acute, Nonurgent Care: A Claims news-and-events/investor-news/press-release-details/2018/ Analysis of Episode-Level Utilization. Journal of Teladoc-Announces-Full-Year-and-Fourth-Quarter-2017- Medical Internet Research 19(2):e35. Results/default.aspx. Accessed March 20, 2018. 18 Weinick RM, Burns RM, Mehrotra A. 2010. Many The UCLA Center 10 Kavilanz P. 2017. These Doctors Do House Calls Emergency Department Visits Could Be Managed for Health Policy Research —For Just $99. http://money.cnn.com/2015/03/03/ at Urgent Care Centers and Retail Clinics. Health is part of the smallbusiness/startup-doctors-heal/index.html. Accessed Affairs (Project Hope) 29(9):1630-1636. UCLA Fielding School of Public Health. June 30, 2017. 19 Mehrotra A, Wang MC, Lave JR, Adams JL, 11 Heal Inc. 2017. See a Doctor Now—In the Comfort McGlynn EA. 2008. Retail Clinics, Primary of Your Home. On-Demand House Calls, Every Day Care Physicians, and Emergency Departments: A From 8 a.m.to 8 p.m. https://heal.com/. Accessed June Comparison of Patients’ Visits. Health Affairs (Project 30, 2017. Hope) 27(5):1272-1282. 12 Jolly J. An Uber for Doctor Housecalls. New 20 Mehrotra A, Paone S, Martich GD, Albert SM, York Times. May 5, 2015. https://well.blogs.nytimes. Shevchik GJ. 2013. Characteristics of Patients com/2015/05/05/an-uber-for-doctor-housecalls/. Accessed Who Seek Care via eVisits Instead of Office Visits. June 29, 2017. Telemedicine Journal and e-Health 19(7):515-519. The analyses, interpretations, conclusions, 13 Ahmed A, Fincham JE. 2010. Physician Office vs. 21 Ashwood JS, Reid RO, Setodji CM, Weber E, and views expressed in this policy brief are Retail Clinic: Patient Preferences in Care Seeking for Gaynor M, Mehrotra A. 2011. Trends in Retail those of the authors and do not necessarily Minor Illnesses. Annals of Family Medicine 8(2):117- Clinic Use Among the Commercially Insured. represent the UCLA Center for Health Policy 123. American Journal of Managed Care 17(11):e443-448. Research, the Regents of the University 14 Wang MC, Ryan G, McGlynn EA, Mehrotra A. 22 Mukamel DB, Ladd H, Amin A, Sorkin DH. 2019. of California, or collaborating 2010. Why Do Patients Seek Care at Retail Clinics, Patient Preferences Over Care Settings for Minor organizations or funders. and What Alternatives Did They Consider? American Illnesses and Injuries. Health Services Research. Apr 29 Journal of Medical Quality: The Official Journal of the [Epub ahead of print]. PB2019-6 American College of Medical Quality 25(2):128-134. 23 Chou S-CA, Schuur JD. Beware This Insurer’s Sneak Copyright © 2019 by the Regents of the 15 Edwards L, Thomas C, Gregory A, et al. 2014. Are Attack on Emergency Care Coverage. March 9, 2018. University of California. All Rights Reserved. People With Chronic Diseases Interested in Using https://www.statnews.com/2018/03/09/anthem-insurance- Telehealth? A Cross-Sectional Postal Survey. Journal emergency-care/. Accessed July 18, 2018. Editor-in-Chief: Ninez Ponce, PhD of Medical Internet Research 16(5):e123. 24 Japsen B. 2015. UnitedHealth Widens Telehealth 16 Russo L, Campagna I, Ferretti B, et al. 2017. What Coverage to Millions of Americans. https:// Drives Attitude Towards Telemedicine Among www.forbes.com/sites/brucejapsen/2015/05/05/ Phone: 310-794-0909 Families of Pediatric Patients? A Survey. BMC unitedhealth-widens-telehealth-coverage-to-millions-of- Fax: 310-794-2686 Pediatrics 17(1):21. americans/#6e32303b1cb7. Accessed July 18, 2018. Email: chpr@ucla.edu healthpolicy.ucla.edu