HEALTH INSURANCE, RACE, AND EMERGENCY ROOM UTILIZATION.(Brief Article)

INTRODUCTION

It is well recognized that the billing costs at the time of a patient's hospital discharge vary not only by medical diagnosis and procedure, but also by the type of insurance. [1] In addition to variations in medical product utilization that can be attributed to differences in medical diagnoses, medical procedures, and type of insurance, medical service utilization also appears to vary with socioeconomic characteristics including race, gender, income, and employment status. In particular, several studies have shown that blacks, on average, receive less medical care than whites, and the poor receive less than the non-poor. [2]

The current study attempts to determine whether race and other demographic variables and insurance type, after controlling for medical diagnoses and procedures, have significant effects on the utilization of emergency health care services. Although there is extensive evidence that demographic characteristics and insurance type have an impact on overall health care utilization, much less work has been done on the relationship between emergency room utilization and insurance type and demographic characteristics.

In this study, I begin with the assumption that emergency room service may differ in at least three important ways from hospital outpatient, clinic, or private physician care. First, patients who use emergency room service have very limited control over the physician who treats them. Second, the emergency department medical equipment is more likely to be available to all emergency room patients, regardless of insurance type. Third, the time constraint on urgent and emergent care restricts the hospital's ability to restrict medical service access. In addition, the county hospital that provided the data for this study has, like many public hospitals, a commitment to provide quality service to the residents of its community.

If time constraints are more important factors in emergency room medical decisions than in non-emergency room medical decisions, then the impact of insurance type on hospital discharge costs should be close to zero. That is, the cost of treatment for any patient should depend upon the diagnoses and medical procedures employed, rather than on the type of insurance. Moreover, the demographic characteristics of the patient should influence emergency care utilization only to the extent that such variables serve as proxies for the patient's medical attributes and the level or type of medical services; and only in this way should they influence the cost of medical treatment.

Although these relationships between medical service utilization and illness apply to virtually all medical services, the amount of time available for decision-making activity in the emergency room, depending on the triage process, is, on average, of shorter duration. It seems likely, therefore, that the level of emergency room service is determined by the hospital or physician rather than by the patient. That is, emergency room patients, including the uninsured, are less likely than non-emergency room patients to review price before purchasing medical services. As a consequence, emergency room outcomes, with respect to charges and the corresponding services, reflect the medical choices--type of services, intensity of service, and quality of service--of the hospital-physician and not the patient. Thus, the medical services received by any emergency room patient with urgent or emergent needs are more likely to reflect the doctor's decision than the medical services administered under non-emergent and non-urge nt conditions. If insurance type is not a determinant of emergency room protocol or plays only a minor role in such protocols, then insurance type should not impact on the type of services, intensity of service, and quality level of the services received by the uninsured relative to the insured. Thus, under the assumption of homogeneous emergency room inputs--same medical staff and same equipment--then differences in billing and length of stay should be due to differences in diagnoses, procedures, or the intensity of care, but not due to differences in insurance type. [3]

Using a database of emergency room visits to a county hospital in southern Michigan in 1995, I use a regression model to estimate the discharge costs to patients as a function of the patient's primary diagnosis, the procedures performed, the length of stay, and the age, race, gender, and labor market status of the patient. In addition, I use a logit model to estimate the probability that a patient will remain overnight as a function of the patient's primary diagnosis, the procedures performed, and the age, race, gender, and labor market status of the patient.