High-Deductible Health Plans and Emergency Department Use

In this issue of JAMA, Wharam and colleagues¹ report findings from an observational study that assesses the relationship between transition to a high-deductible health plan and use of emergency departments (EDs). The authors use a sophisticated administrative data set, a large sample of patients enrolled in a modern employer-sponsored health plan, and impeccable health services research methods. Their major conclusion is that individuals whose health insurance coverage was switched to a high-deductible health plan, compared with those in a traditional health maintenance organization, reduced overall ED visits by about 10%, mainly for conditions of low and indeterminate severity.

Some may infer from these findings that high-deductible health plans are safe, at least because they do not appear to deter necessary emergency visits. However, even though the study methods are sound, the data are as good as possible, and the study limitations are clearly identified, there are important issues that must be addressed before these findings can inform health policy. Under US Food and Drug Administration regulation, safety and efficacy must be evaluated before new drugs are approved for use in health care markets. However, health policy changes do not require such rigorous evaluation before implementation, although these policies can either substantially help or harm people,² just as drugs can.

If an evaluation of a health policy change is conducted, it often resembles the study by Wharam et al.¹ The design is quasi-experimental; the evaluation occurs after the policy has been implemented; because of severe data limitations, the outcomes measured, if any, are not comprehensive; and as in this study, the outcomes often do not measure patients' health, morbidity, or even mortality. There is no way of knowing whether all evaluations are even submitted for publication, including those that are not consistent with the researcher's point of view. Perhaps the time has come to register, at the time of human participants approval, all evaluations of changes in health policy, just as clinical trials are registered at the time of inception.³

Despite these limitations, Wharam et al¹ have added new data from a contemporary health system experience that confirm previous findings from the RAND Health Insurance Experiment (HIE)⁴ about how health plan deductibles and co-payments affect ED use. The HIE, conducted almost 30 years ago, remains the largest social experiment in health policy ever carried out. It randomized families to different health insurance plans based on amount of out-of-pocket expenditures and used comprehensive and clinically detailed measures to determine how insurance generosity affected cost, use, quality, and health.

The HIE found that increased out-of-pocket expenditures substantially reduced use, except for hospitalizations among children. However, reduction in use did not affect the health status of the average individual because those assigned to more generous health plans did not receive substantially better quality of care than those in less generous plans, and reductions in use appeared virtually equal for both appropriate and inappropriate care. The major exception was ED use, which was reduced much more for conditions that did not require ED care.⁵ For example, the number of lacerations that required suturing did not depend on the generosity of the ED user's insurance plan, but the number of abrasions treated in the ED was decreased substantially.

The world of medicine has changed dramatically since the HIE was conducted, and presenting to an ED quickly after the onset of many acute conditions is now more important than it was then. New information is needed in the current clinical environment about patients' decisions to use an ED as a function of their insurance plan as well as the outcomes resulting from those decisions. In addition to reporting changes in ED visit rates, Wharam et al¹ also measured hospital admission rates associated with the ED visit as a possible indicator of patient outcomes. As the authors acknowledge, they were unable to assess whether patients experienced adverse events or increased mortality by forgoing some ED visits.

The authors report that patients in individual high-deductible health plans had increased odds of having ED visits during the period that they had not met their deductible compared with the period after their deductible was exceeded. The authors also provide some data on payment for ED visits by member and by plan, and demonstrate that virtually all reductions in ED visits were for repeat ED visits. This presumably reflects the increased knowledge that patients had regarding the cost of an ED visit. However, the authors do not report how much a person had to pay for the ED visit or how much money a person believed would be charged for the visit prior to going to the ED. This is extremely important because Wharam et al found a 10% reduction in ED use vs a 23% reduction found among Kaiser patients when a $50 to $100 co-payment for ED use was implemented.⁶ Thus, the full effect of a high-deductible health plan on ED use might first be seen at the beginning of the insurance year among those people who actually know the amount they would have to pay out of pocket.

In addition, in the current study, there are no data on whether a patient delayed seeking care for time-sensitive symptoms, such as chest pain, or for outcomes for individuals in the high-deductible health plan who had fewer visits classified as low severity. Even individuals in the high-deductible health plan who had fewer ED visits classified as low severity might have experienced some adverse events, for example, had longer duration of untreated pain.

The findings of Wharam et al¹ that ED utilization decreased after the switch to high-deductible health plan coverage should not be used to support the notion that implementing a high-deductible health plan will help reduce ED overcrowding. Since the high-deductible health plan reduced ED visits by about 10%, and applied only to individuals who have both health insurance and a generous health insurance plan, the overall population effect will be much less. Thus, high-deductible health plans will do little to diminish the overcrowding and long waiting times that currently characterize many EDs.

In addition, the algorithm used by Wharam et al¹ to classify ED visits was intended primarily to categorize ED use as a way to gauge access to primary care services for specific geographic locations.⁷ However, Wharam et al used a modification of the algorithm to make inferences about the severity of ED visits. Visits classified as low severity (based on retrospective coding after the ED visit was completed) include chief complaints that have broad differential diagnoses (based on symptoms at the time of initial ED presentation). Headache, for example, can be caused by subarachnoid hemorrhage or tension. In a patient with a severe headache, it may be difficult for emergency physicians to differentiate a sentinel bleed from a more benign cause of headache, and it is nearly impossible for a patient to do so. If health plans use financial disincentives to discourage people from using the ED, then same-day urgent care services must be available within those health plans. According to Wharam et al this was not necessarily the case for some of the patients who were part of the health plan.

Goldman et al⁸ demonstrated that increasing co-payments from $5 to $10 per prescription is associated with decreased use of effective medication among individuals with chronic diseases. Similarly, it is inconceivable that high-deductible health plans will not affect the health of some patients. While it is doubtful that paramedics will ask critically ill patients whether they can afford the deductible for using the ED before transporting them, it is conceivable that patients with early symptoms of stroke or myocardial infarction will delay going to the ED because they know it will cost them a lot of money. However, such delays will lead to organ damage, worse future health status, and increased risk of death.

On the other hand, some patients receive a good deal of inappropriate care⁹ and some of that medical care can actually be harmful. In the HIE, medical care use was reduced without affecting health because less generous insurance prevented some adverse outcomes resulting from unnecessary or too much medical care. However, some individuals also did not receive needed care, and thus were less healthy at the end of the experiment. If costs are to be contained by less generous health insurance, some patients will be harmed. The health care system needs to be reengineered so that unnecessary and harmful care is eliminated and effective care is provided more efficiently.

Ideally, future studies examining health plan policies that influence ED use would begin by stating that, “Even though patients did not have to pay any more out of pocket for their care when they came to the ED, vs a physician's office, all unnecessary care has been eliminated and virtually every possible outcome has been achieved at a price that the local community considers good value for money.” Such a goal is attainable if the ED were used not only as the place to receive high-quality, life-saving care for emergency conditions, but also as a place for high-quality, timely, and inexpensive care for less serious acute conditions. Such a transformation could be facilitated, for example, by the use of nurse practitioners, guidelines, protocols, and standardized triage procedures for selected conditions.¹⁰ If such a transformation were realized, perhaps the US health care system would be that much closer to meeting the needs of all of the people all of the time.