Title: Medicaid and Moral Hazard: Covering Emergency Department Visits Increases Emergency Department Visits…or Not?
Abstract: 1.Klein et al1Klein E.Y. Levin S. Toerper M.F. et al.The effect of Medicaid expansion on utilization in Maryland emergency departments.Ann Emerg Med. 2017; 70: 607-614Abstract Full Text Full Text PDF PubMed Scopus (33) Google Scholar conducted a retrospective cross-sectional study to understand the effect of Medicaid expansion on emergency department (ED) visits in Maryland, using an administrative claims database that included all visits to any of the 48 EDs in the state.A.What were the 2 main questions the authors sought to answer? What was the subject (unit of observation) for each of the main study objectives? Which of the 2 questions does Figure 3B in the article by Klein et al answer? Interpret Figure 3B.B.Many research studies attempt to make statements on a population that are based on making “statistical inferences” using information on a randomly collected “sample” from the “population.” In this study, did the authors use statistical inference to determine the effect of Medicaid expansion on ED utilization? Were the data a sample of the population being studied?C.Did the authors account for differences in the state population of Maryland between the 2 study periods? How would you interpret the results if you knew there had been a large decrease in the population between the 2 periods? What if there had been a large increase?2.Generalizability is important when research is analyzed. As the authors describe, Maryland is somewhat unique in its approach to health care compared with other states, and this may limit the generalizability of what was observed in Maryland.A.What is the Health Services Cost Review Commission of Maryland? Does such an entity exist in other states? How did the existence of such an entity affect the ability to perform this study?3.The Oregon Health Insurance Experiment involved a randomized controlled trial that assigned participants to Medicaid according to a lottery in 2008. The study found a nearly 40% increase in ED visits per covered person compared with controls who did not receive Medicaid.2Taubman S.L. Allen H.L. Wright B.J. et al.Medicaid increases emergencydepartment use: evidence from Oregon’s Health Insurance Experiment.Science. 2014; 343: 263-268Crossref PubMed Scopus (371) Google Scholar Therefore, many policy experts expect that Medicaid expansion will translate to an increase in ED visits overall. In fact, another recent study of the effect of the Patient Protection and Affordable Care Act demonstrated a significant increase in ED utilization in Illinois.3Dresden S.M. Powell E.S. Kang R. et al.Increased emergency department use in Illinois after implementation of the Patient Protection and Affordable Care Act.Ann Emerg Med. 2017; 69: 172-180Abstract Full Text Full Text PDF PubMed Scopus (48) Google ScholarA.Using complete data on all EDs in Maryland, the authors show that there was no increase in overall arrivals to the ED, despite the increase in Medicaid enrollees from 23.3% to 28.9%. What are some theories to explain why this did not create an increase in ED visits at hospitals?B.During the study period, Medicaid-covered ED visits increased by 5.6%; however, visits covered by commercial insurance did not increase in Maryland with passing of the Patient Protection and Affordable Care Act. How does this compare with the Illinois study? How may this have affected the results?4.Moral hazard is defined as “lack of incentive to guard against risk where one is protected from its consequence.” A major policy question of studies such as this one and the Oregon Health Insurance Experiment is, does expanding Medicaid increase moral hazard? Put more bluntly, policymakers want to know whether enrolling in Medicaid makes patients more likely to utilize the ED just because they have protection from financial consequences as opposed to perceiving an emergency.A.One major assumption of measuring moral hazard simply as ED utilization for nonurgent conditions is that Medicaid patients have “reasonable” access to other options of health care. Is this assumption valid? What factors influence reasonable access? How does avoiding this assumption change the question that policymakers should be asking in regard to moral hazard?B.How might a patient’s health status or expectations affect his or her decision to sign up for Medicaid? What is self-selection bias? What can you learn from the admission rates of the newly enrolled Medicaid group compared with the admission rates of those who remained uninsured?C.Do hospitals have an incentive to help patients enroll in Medicaid? Is this incentive different in Maryland compared with other states? Does this incentive depend on patients’ health status? Q1. Klein et al1Klein E.Y. Levin S. Toerper M.F. et al.The effect of Medicaid expansion on utilization in Maryland emergency departments.Ann Emerg Med. 2017; 70: 607-614Abstract Full Text Full Text PDF PubMed Scopus (33) Google Scholar conducted a retrospective cross-sectional study to understand the effect of Medicaid expansion on emergency department (ED) visits in Maryland, using an administrative claims database that included all visits to any of the 48 EDs in the state. Q1.a What were the 2 main questions the authors sought to answer? What was the subject (unit of observation) for each of the main study objectives? Which of the 2 questions does Figure 3B in the article by Klein et al answer? Interpret Figure 3B. To understand whether Medicaid expansion in Maryland increased the total number of statewide ED visits, Klein et al proposed 2 specific questions. The first is as follows: Comparing the two 18-month periods before and after expansion, was the increase in proportion of Medicaid-covered visits in an individual ED associated with an increased total number of visits to that ED? Although the study was observational, the authors took advantage of a natural experiment of Medicaid expansion. To understand whether Medicaid expansion increased ED use in Maryland, they assessed whether the proportional change in Medicaid-covered visits in an ED was positively associated with the change in total visits in that same ED. That is why the subjects in this study were each of the 48 EDs in the state. In the postexpansion period, all 48 EDs had an increase in Medicaid-covered visits (x axis of Figure 3B from the article by Klein et al) and a decrease in uninsured visits (y axis of Figure 2B). Figure 3B plots each ED on the graph by the percentage increase in Medicaid visits and the percentage of change in total visits; the best-fit line shows no association between increased proportion of Medicaid visits and an increase in total visits. The second question was, in patients who visited the ED in both periods who were initially uninsured, what was the effect of being newly insured by Medicaid on the total number of ED visits compared with patients who remained uninsured or obtained commercial insurance? For this question, the study subjects were individual patients. Out of the 185,631 uninsured patients who visited the ED in the pre-expansion period, 67,788 (37%) visited again and were included in this analysis. Table 2 categorizes these patients into 4 categories, and allows comparison of patients newly insured in Medicaid with other initially uninsured patients. Patients newly enrolled in Medicaid have higher use than those who remain uninsured or commercially insured. Q1.b Many research studies attempt to make statements on a population that are based on making “statistical inferences” using information on a randomly collected “sample” from the “population.” In this study, did the authors use statistical inference to determine the effect of Medicaid expansion on ED utilization? Were the data a sample of the population being studied? Statistical inference is used in research to understand characteristics of a population by observing a sample of that population. To answer question 1, the authors selected individual EDs as subjects, and so the population they intended to study was all of the EDs in the state. They did not sample from the population, but actually obtained data from all 48 EDs. Therefore, they could draw conclusions about their population without using inference. Q1.c Did the authors account for differences in the state population of Maryland between the 2 study periods? How would you interpret the results if you knew there had been a large decrease in the population between the 2 periods? What if there had been a large increase? There is no indication that the authors accounted for changes in the population of Maryland between the periods. This could have affected the study results if people entering or leaving the state were not representative of the people currently living in the state, especially in regard to their insurance status. For example, if a large company moved its headquarters into downtown Baltimore, recruited 10,000 people from other states, and provided them with private health insurance, then the EDs in the area may experience a decrease in proportion of Medicaid visits but an increase in total visits. This may artificially suppress the effect of Medicaid expansion. This is unlikely to be relevant in this study, however, given that total statewide ED visits did not change (Figure 3A). Last, US Census Bureau data do not indicate that there were large changes in the overall population. Q2. Generalizability is important when research is analyzed. As the authors describe, Maryland is somewhat unique in its approach to health care compared with other states, and this may limit the generalizability of what was observed in Maryland. Q2.a What is the Health Services Cost Review Commission of Maryland? Does such an entity exist in other states? How did the existence of such an entity affect the ability to perform this study? When assessing whether the effects reported in this study of Medicaid expansion would be generalizable to other states, we must consider how hospital-based health care is financed in Maryland. The Health Services Cost Review Commission (HSCRC) is an organization that was established by an act of the Maryland legislature in 1971.4Murray R. Setting hospital rates to control costs and boost quality: the Maryland experience.Health Aff (Millwood). 2009; 28: 1395-1405Crossref PubMed Scopus (67) Google Scholar It is an independent state agency of 7 commissioners appointed by the governor. The agency is tasked with setting a hospital-adjusted rate for each inpatient, hospital-based outpatient, and ED service. The implication for any given ED is that all payers (self-payers, private insurers, and Medicare/Medicaid) pay the same facility fee for each level of ED visit. To decide on payment rates, the HSCRC created a centralized, comprehensive, and uniform data infrastructure of hospital operations. The resulting data set includes every ED visit in the state. The availability of this data set allowed a precise measurement of ED volume and payer mix at each ED during periods before and after Medicaid expansion. In 2014, Maryland began a cost-control experiment involving caps on hospital revenues,5Rajkumar R. Patel A. Murphy K. et al.Maryland’s all-payer approach to delivery-system reform.N Engl J Med. 2014; 370: 493-495Crossref PubMed Scopus (70) Google Scholar but this study was conducted before this major change. The all-payer model described above may still limit the generalizability of this study to other states. Theoretically, an all-payer model should only prevent cost shifting6Reinhardt U.E. Analysis and commentary: the many different prices paid to providers and the flawed theory of cost shifting: is it time for a more rational all-payer system?.Health Aff (Millwood). 2011; 30: 2125-2133Crossref PubMed Scopus (54) Google Scholar; for example, lower payments from Medicaid being balanced by higher charges to private insurers. However, effects on ED use may exist, and, if present, the mechanisms are likely to be complex. ED provider incentives are unlikely to explain variation in ED use because physician fees are not regulated by the HSCRC and the Emergency Medical Treatment and Labor Act ensures ED services regardless of insurance status. ED use differences in Maryland are more likely to be explained by the all-payer model’s effects on downstream factors such as inpatient bed availability, boarding of admitted patients, and adequacy of nurse staffing. These factors may be affected if the calculations used to set rates are inaccurate or flawed. Q3. The Oregon Health Insurance Experiment involved a randomized controlled trial that assigned participants to Medicaid according to a lottery in 2008. The study found a nearly 40% increase in ED visits per covered person compared with controls who did not receive Medicaid.2Taubman S.L. Allen H.L. Wright B.J. et al.Medicaid increases emergencydepartment use: evidence from Oregon’s Health Insurance Experiment.Science. 2014; 343: 263-268Crossref PubMed Scopus (371) Google Scholar Therefore, many policy experts expect that Medicaid expansion will translate to an increase in ED visits overall. In fact, another recent study of the effect of the Patient Protection and Affordable Care Act demonstrated a significant increase in ED utilization in Illinois.3Dresden S.M. Powell E.S. Kang R. et al.Increased emergency department use in Illinois after implementation of the Patient Protection and Affordable Care Act.Ann Emerg Med. 2017; 69: 172-180Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar Q3.a Using complete data on all EDs in Maryland, the authors show that there was no increase in overall arrivals to the ED, despite the increase in Medicaid enrollees from 23.3% to 28.9%. What are some theories to explain why this did not create an increase in ED visits at hospitals? Despite expectations set by the landmark randomized controlled trial in health policy in Oregon, there was no significant increase in ED visits in the postexpansion period in Maryland. Given the robust nature of HSCRC data, it seems unlikely that the reason was due to issues with data validity. Oregon had previously expanded Medicaid to poor adults within 110% of the poverty level, but had frozen this expansion from 2004 to 2008 because of budget issues.7James J. The Oregon health insurance experiment.Health Aff (Millwood). 2015; 3: 1057-1106Google Scholar In 2008, the state announced a lottery to give coverage to certain eligible patients and not others because of budget restrictions. Subsequently, more than 85,000 people signed up for a waiting list for Medicaid. This was the population from which the Oregon study was randomized through a lottery system. Arguably, the population from which the sample was selected was different from the Maryland pre-expansion population; those who entered the Oregon lottery may have been more likely to seek care, given that they were motivated to sign up. Another major difference is that the present study excluded the first 6 months of enrollment from the postexpansion period, when uptake into Medicaid was expected to be highest. These “early-signup” patients and visits may have been like the lottery winners who received Medicaid in Oregon, but they were purposefully excluded in the Maryland study to avoid this initial spike and measure longer-term effects. Yet another explanation could be that uninsured patients in Maryland are less sensitive to costs and are charged less per ED visit compared with those in Oregon; evidence indicates that the HSCRC rate setting in Maryland prevents exposure to arbitrary “charge-master” prices.8Anderson G.F. MarketWatch: from “soak the rich” to “soak the poor”: recent trends in hospital pricing.Health Aff (Millwood). 2007; 26: 780-789Crossref PubMed Scopus (102) Google Scholar Last, differences in factors such as ED wait times may provide some patients incentive to avoid the ED and attempt to seek alternative care; according to Medicare Hospital Compare data, Maryland’s ED wait times are the longest in the nation.9Groeger L, Tigas M, Wei S. ER wait watcher. Available at: https://projects.propublica.org/emergency/. Accessed November 28, 2017.Google Scholar Q3.b During the study period, Medicaid-covered ED visits increased by 5.6%; however, visits covered by commercial insurance did not increase in Maryland with passing of the Patient Protection and Affordable Care Act (ACA). How does this compare with the Illinois study? How may this have affected the results? Dresden et al3Dresden S.M. Powell E.S. Kang R. et al.Increased emergency department use in Illinois after implementation of the Patient Protection and Affordable Care Act.Ann Emerg Med. 2017; 69: 172-180Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar reported an increase in ED visits after implementation of the ACA in 2014. Using an interrupted time-series analysis of nearly all hospitals in Illinois, they calculated the average change in slope of mean monthly visits covered by Medicaid before and after ACA implementation. Using this method, they demonstrated a small but significant increase in overall ED visits (1.47 visits/1,000 residents) in the post-ACA period, which were attributed to Medicaid patients who were ultimately discharged. Overall, after implementation, there was a 25% increase in ED visits paid for by Medicaid in Illinois, along with a 3% increase in private-insurance-paid visits. Maryland’s Medicaid expansion was far more modest, making meaningful comparison between the 2 states very difficult. Q4. Moral hazard is defined as “lack of incentive to guard against risk where one is protected from its consequence.” A major policy question of studies such as this one and the Oregon Health Insurance Experiment is, does expanding Medicaid increase moral hazard? Put more bluntly, policymakers want to know whether enrolling in Medicaid makes patients more likely to utilize the ED just because they have protection from financial consequences as opposed to perceiving an emergency. Q4.a One major assumption of measuring moral hazard simply as ED utilization for nonurgent conditions is that Medicaid patients have “reasonable” access to other options of health care. Is this assumption valid? What factors influence reasonable access? How does avoiding this assumption change the question that policymakers should be asking in regard to moral hazard? This and many other recent studies have focused on quantifying the effects of insurance expansion on ED use. Another recent study by Nikpay et al10Nikpay S. Freedman S. Levy H. et al.Effect of the Affordable Care Act Medicaid expansion on emergency department visits: evidence from state-level emergency department databases.Ann Emerg Med. 2016; 70: 215-225.e6Abstract Full Text Full Text PDF Scopus (94) Google Scholar found an annual increase of 10 visits per 1,000 residents across 25 states that expanded Medicaid. Given the average rate of 200 visits per 1,000 residents, this was derived from comparing the 3.2% relative increase in expansion states with a 1.5% relative decrease in nonexpansion states. The real underlying question is whether “moral hazard” is caused by providing low-cost insurance to patients. Moral hazard is a term used to describe a tendency for patients to seek more health services because of less financial liability. Specifically, in ED patients, it refers to seeking expensive ED care without the perception of a medical emergency. Policymakers and payers, who are interested in decreasing costs, are understandably interested in this question. Given small to no increases in ED use after insurance provision, though, it may be more productive to assess what other factors affect ED use. For example, there is sufficient evidence that patients, both privately insured and with Medicaid, have difficulty obtaining timely outpatient care.11Miller P. Survey of physician appointment wait times and Medicare and Medicaid acceptance rates. Available at: https://www.merritthawkins.com/uploadedFiles/MerrittHawkins/Pdf/mha2017waittimesurveyPDF.pdf. Accessed January 24, 2018.Google Scholar, 12Chou S.-C. Deng Y. Smart J. et al.Insurance status and access to urgent primary care follow-up after an emergency department visit in 2016.Ann Emerg Med. 2017; (https://doi.org/10.1016/j.annemergmed.2017.08.045)Abstract Full Text Full Text PDF Scopus (22) Google Scholar They may perceive impending deterioration and wish to receive care from a physician before being sick enough to truly need emergency therapy, or wish to seek care in a facility that has sufficient imaging or laboratory capabilities. For certain patients, the only times that they are able to escape the demands of work and family are when most scheduled care is unavailable. Although cost sharing may play some role in the decision to seek ED care, it is not likely the most important driver of it. This is well illustrated by the small to no increases in ED visits across expansion states and the lack of uniformity in finding an increase in visits. Q4.b How might a patient’s health status or expectations affect his or her decision to sign up for Medicaid? What is self-selection bias? What can you learn from the admission rates of the newly enrolled Medicaid group compared with the admission rates of those who remained uninsured? In Maryland, despite no overall increase in ED visits, patients who presented in the pre-expansion period without insurance and then returned with Medicaid insurance highlight how a patient’s health status may affect his or her behavior to become insured. These patients may have discovered that they are ill, that they need medications, or simply need more medical attention. Of patients who returned with Medicaid, 11.7% were admitted compared with only 2.7% of patients who remained uninsured. This is clear evidence of self-selection bias; of all initially uninsured revisits, patients returning with Medicaid insurance were sicker, suggesting that they may have recognized their risk for health deterioration and enrolled to obtain services or protect themselves from financial risk. As described in question 4c, hospital incentives are also likely to play a role. Q4.c Do hospitals have an incentive to help patients enroll in Medicaid? Is this incentive different in Maryland compared with other states? Does this incentive depend on patients’ health status? Hospitals have an obvious financial incentive to enroll eligible patients in Medicaid, as opposed to providing uncompensated care. Visits can be retroactively paid for once an eligible patient is enrolled. Additionally, patients who have medical illnesses are likely to revisit the same hospital because of proximity. A patient who is uninsured and Medicaid eligible will likely receive significant encouragement and logistic assistance from the hospital to be covered, and will likely be insured on the next visit. This incentive is present in any state that has large numbers of Medicaid-eligible patients who are currently uninsured. As of now, 33 states have opted to accept federal funds and expand Medicaid eligibility to previously ineligible patients.13Henry J. Kaiser Family Foundation. Current status of state Medicaid expansion decisions. Available at: https://www.kff.org/health-reform/slide/current-status-of-the-medicaid-expansion-decision/. Accessed November 28, 2017.Google Scholar