Hospital Closure and Hospital Choice: How Hospital Quality and Availability will Affect Rural Residents

By Deepak Premkumar, Dave Jones, and Peter F. Orazem

The population shift from rural to urban regions has decreased the population density around hospitals in small towns and rural areas. At the same time, the availability of improved road systems that lower travel times, an improved ability to deliver health services via the Internet, and larger urban-rural gaps in access to the latest medical technologies may make urban hospitals more attractive for rural patients. Following a pattern of decline that started in the 1970s, these factors have led to a steady decrease in the number of rural hospitals over the last two decades—since 1990, the number of rural hospitals has decreased 20 percent while the number of urban hospitals has only decreased 3.5 percent (Figure 1).

To help stop the decline in the number of rural hospitals, in the 1990s Medicare enacted the Critical Access Hospital program. With rural hospitals being particularly dependent on publicly subsidized healthcare—almost 60 percent of their revenue comes from Medicare and Medicaid—the program was devised to prop up hospitals in isolated areas where residents had few other healthcare options.

Under its original rules, these hospitals had to be located at least 35 miles away from any other hospital, which means only about one-third of the nation’s 1,300 Critical Access Hospitals would have qualified under the original rules. However, the law was amended to allow states to designate “necessary provider” hospitals, which lessened or removed proximity restrictions.

More recently, federal budgetary constraints have led to renewed interest in re-imposing the more stringent rules, which would lead to further closure of rural hospitals in Iowa and elsewhere.

This study estimates how rural patients make tradeoffs between hospital quality and distance in deciding whether to choose the nearest hospital or to travel farther for an alternative. We base our analysis on an empirical model that estimates the sensitivity of rural choice of local, urban, or specialized research hospitals on distance to, and quality of, each of the three hospital options. We derive estimates of hospital choice for inpatient visits, for outpatient visits, separately for the most commonly diagnosed illnesses, and for emergency or nonemergency admissions. We use these estimates to simulate how potential hospital closings will alter hospital choices made by rural Iowa patients. We illustrate how two hospital closing scenarios: (a) closing 25 percent of the lowest quality rural hospitals; and, (b) closing 15 percent of the least-used rural hospitals in Iowa, affect the average distance to, and quality of, the chosen hospital.

Few studies have evaluated the role of hospital quality in patient choices; however, this is likely to be a key factor explaining the incentives to bypass rural hospitals. Liu et al. (2007) surveyed 647 hospital inpatients for their assessments as to why patients would bypass a local hospital. Following the lack of local specialists, the second-most common reason cited for bypassing a local hospital was poor reputation or quality of local care.

Health Grades, Inc. compiled the data on hospital quality. There are significant quality differences between hospitals, exemplified by the company’s simple one-to-five-star rating system. To avoid missing data, we used the two most common ailments, heart failure and pneumonia, to measure hospital quality. Table 1 illustrates a pronounced rise in quality when comparing rural hospitals to urban or research hospitals, with urban hospitals actually marginally outperforming research hospitals.

To estimate the tradeoffs between the two factors in hospital choice we were granted access to the Iowa Hospital Association recorded visits for all Iowa hospitals occurring between January 1, 2002 and December 31, 2002. The database includes 209,687 inpatient records for patients treated and discharged during this period and 138,685 outpatient records. Inclusion in the outpatient database does not require admission or release from the hospital, but only that the patient received treatment at an Iowa hospital. We then divided hospitals into rural, urban, and research groups. Rural hospitals are designated by the population density of the hospital county. Urban hospitals reside in counties containing a metropolitan statistical area. The research hospitals are located in Des Moines and Iowa City.

Our focus is on the determinants of hospital choice for rural patients—defined as those whose residence is in a zip code region listed as rural by the US Census in 2000. Distance was calculated as the straight line distance from the latitude and longitude of the patient’s home zip code to the latitude and longitude of the nearest rural, urban, and research hospital even if none of those hospitals were chosen.

As shown in Table 1, hospital choices do not differ much between inpatient and outpatient treatments. Almost 70 percent of rural residents choose a rural hospital for inpatient and outpatient service. Urban hospitals serve 12 percent of rural residents and 18 percent are served by research hospitals. The average rural patient lives about five miles from a rural hospital, but lives 51 miles from the nearest urban hospital and 71 miles from the nearest research hospital.

The results of our model of inpatient and outpatient hospital choice are presented in Table 2. The key variables of interest are distance to and quality of the nearest hospital of each type. Distance is the single largest driving factor in the choice of hospital. At sample means, a 10 percent increase in distance lowers the probability of choosing that hospital type for inpatient services by 12.9 percent. Hospital choice is less sensitive to quality, although a tradeoff between distance and quality is apparent. A 10 percent improvement in quality increases likelihood of choosing that hospital by 2.3 percent for an inpatient procedure. Hospital demand for outpatient services is also sensitive to distance, but not quality. Our findings suggest that women, older patients, and patients who do not pay through insurance are more distance sensitive.

We expect that patients with severe or time-sensitive needs might be more sensitive to distance and less sensitive to quality. For inpatient hospitalizations, the three admission codes—ordered from most to least critical—are emergency, urgent, and elective. Consistent with our expectations, emergency and urgent admissions are much more sensitive to distance than elective (Table 3). A 10 percent increase in distance leads to a 17.5 percent and 16.1 percent reduction in the probability of choosing a hospital for emergency and urgent patients respectively, while it only leads to a 8.3 percent drop for elective procedures. Choice of where to receive emergency and urgent care is also sensitive to quality, while choice of hospital for elective procedures is virtually unaffected by quality. For patients with insurance, quality is more important for both emergency and elective procedures.

A notable finding of this paper is that the quality of a health institution is an important factor in hospital choice, and that patients assess tradeoffs in distance and quality when deciding where to get hospital services. The tradeoff is most salient for inpatient treatments and for emergency or urgent care. Proximity largely drives hospital choice for elective procedures and outpatient services. Our results are consistent with previous research that concludes patients with severe or complicated issues will seek out higher quality care, while people with time-sensitive conditions and the elderly are more distance sensitive. Our findings also illustrate that patients with insurance coverage are more sensitive to quality.

Our simulations show that closing 15 percent of the least-used rural Iowa hospitals results in a marginal increase in distance (around 1.8 miles) and a small decrease in quality, while closing 25 percent of the lowest quality hospitals results in a marginal increase in distance (around 2.9 miles) and a significant increase in quality.

To analyze differential impacts, we separate the analysis by inpatient-outpatient, admission type, and diagnosis. Closing the 15 percent least-used hospitals have more pronounced effects on expected quality and distance for outpatient admissions over inpatient, while closing the 25 percent lowest-quality hospitals have similar magnitudes for both.

When segregating by type of admission (emergency, urgent, or elective), we found that closing the 15 percent least-used hospitals increased expected distance the most for elective procedures. The reductions in quality were largest for the urgent and emergency patients who originally chose a closed hospital.

On the other hand, closing the 25 percent lowest-quality hospitals resulted in a substantial rise in expected quality coupled with only a slightly greater increase in expected distance. For the elective admission type, there is no significant change in expected distance with patients still benefiting from the higher quality. For emergency and urgent admission types, the increased distance is partially offset by large gains in expected quality. As a result, closing the lowest quality hospitals is a better policy prescription, providing a substantial increase in quality with only a marginally higher increase in distance.

Footnotes

1. We were able to get information on hospital quality based on heart and pneumonia deaths for 117 of the 119 hospitals in Iowa. Quality measures based on other criteria were missing for at least 31 percent of the hospitals. ↩

2. This is consistent with reported hospital infection rates for the University of Iowa Hospitals, which were higher than for urban hospitals, possibly because the research hospitals treat more complicated cases. ↩

Suggested citation:

Premkumar, D., D. Jones, and P. Orazem. 2017. "Hospital Closure and Hospital Choice: How Hospital Quality and Availability will Affect Rural Residents." Agricultural Policy Review, Winter 2017. Center for Agricultural and Rural Development, Iowa State University. Available at www.card.iastate.edu/ag_policy_review/article/?a=61.