An analysis of test costs vs. hospital size; hospital mergers may increase rather than reduce overall spending and testing, this investigator asserts.

An analysis of test costs vs. hospital size

Many hospital administrators and laboratory directors are considering consolidations to counter hyperinflation of medical costs. They believe that fusing smaller organizations together will save money. They may be mistaken.

Evidence suggests that big hospitals and, to a certain extent, big laboratories are less economical to run than small ones. A diseconomy of scale became apparent when I analyzed data from the American Hospital Association's 1982 Guide to Hospitals. I studied bed size, admissions, personnel, and expenditures for 600 teaching and community hospitals across the country.

Some of the results were predictable. For example, bed size, admissions, and expenditures are significantly greater in teaching than in nonteaching hospitals, and these figures are also higher for urban than rural settings.

What is startling, however, is the finding that overall it costs much more to run one large hospital than two small ones, as you can see in Table I. Operating one 600-bed facility, for example, costs almost $67 million a year and requires a staff of 2,143. But two 300-bed hospitals can be run for about $10 million less with 209 fewer staffers. A 700-bed hospital employs 14 times more staff than a 100-bed facility and costs 29 times more to run.

A number of known factors contribute to this diseconomy of scale, among them the greater organizational complexity in large hospitals, their more marked fluctuations in occupancy, and a heavier percentage of high-cost patients. Patients in the high-cost category--neonates, those with cancer, and people who have vascular surgery or are suffering an acute phase of a chronic illness-- generally require a longer hospital stay and greater intensity of care than other patients. This difference in case mix is reflected to some extent in the annual number of admissions per bed, which decreases from 50 in a 100-bed hospital to 38 in a 300-bed hospital, at which point a plateau is reached for hospitals of larger size.

Even these considerations cannot wholly explain a fourteenfold increase in expenditures from a 100-bed to a 400-bed hospital--or the fact that it costs $10 million more to run an 800-bed hospital than to run two 400-bed hospitals.

Disproportionate increases in staff needs are not the explanation, either. While there is a significant increase in the number of full-time equivalents (FTEs) per bed between a 100-bed hospital (1.8 FTEs/bed) and a 400-bed hospital (3.4 FTEs/bed), the number begins to level off at this point, and no further significant diseconomies occur as bed size goes up.

On the theory that the type and location of a hospital has an effect on its costs, 1 studied a smaller subset of the original group of hospitals to find our how these factors influenced expenditures. For this study, I analyzed data from 234 of the 600 hositals mentioned earlier--104 teaching and 130 nonteaching community hospitals (see Table II). The results indicate that teaching hospitals average twice the bed capacity and admissions per year of nonteaching hospitals, but have three times more personnel and total expenditures. Teaching hospitals also spend 70 per cent more per admission than do community hospitals.

Difference in location also seems to affect costs. For example, community hospitals in metropolitan areas average 50 per cent more beds than those in rural areas, but their staffing and total expenditures are twice as high. Also, expenditures per admission are 38 per cent higher for metropolitan than for rural community hospitals.

Another match-up produces a significant finding: When 300-bed teaching hospitals are compared with 600-bed teaching hospitals and 300-bed community hospitals with 600-bed community hospitals, we discover much less discrepancy in relative costs (Table III).

It is clear, then, that the diseconomy of scale we are investigating is due in part to the fact that most large hospitals are one kind of institution, metropolitan teaching, while most small hospitals are another kind, community. The diseconomy fades when we consider the teaching and community categories separately, but note that there is still no economy of scale--no savings as hospitals grow larger.

A number of factors relating to the clinical laboratory (and presumably other ancillary services) serve to drive up costs in larger institutions. Put simply, there are more tests available in large hospitals, and availability leads to increased utilization. In order to assess laboratory utilization, I evaluated data from the CAP and the Intersociety Committee on Pathology Information.

In general, teaching hospitals perform an average of 50 per cent more tests than do community hospitals of comparable size. However, the most interesting finding in this study is that, while teaching hospitals use the laboratory increasingly more efficiently as they get larger, community hospitals show the opposite pattern.

That is, the number of tests per admission decreases with bed size in teaching hospitals, but increases with bed size in community hospitals (Tables IV and V). Thus, a 200-bed teaching hospital has approximately 75 tests per admission as opposed to 30 for the same size community hospital. At the 800-bed level, however, tests per admission drop to 60 in teaching hospitals but increase to 42 in community hospitals.

Although the data give no clue to why community hospitals become less economical in their use of laboratory tests as they increase in size, it's interesting to speculate about the pathologist's effect on utilization. The diseconomy of scale may reflect the significant role the pathologist plays in guiding test ordering in small institutions, a role that diminishes in a larger community setting.

Pathologists have a less aggressive role in large noneteaching hospitals for several reasons. First, there is usually a large surgical pathology load, with a resulting decrease in commitment to the clinical laboratory.

Second, the medical staff in a large community hospital is likely to be more aggressive and competitive and less inclined to consult the pathologist. Third, because of past reimbursement formulas, there has been no economic incentive for pathologists on community hospitals to be effective managers of clinical laboratory services--in fact, just the opposite.

The way specific tests are used also provides a clue to increased costs in larger institutions. I have logarithmically plotted the costs per bed size of multiphasic profiles and CK isoenzymes. Figure I compares the cost of an "ideal' profile of 18 tests (line A) with the actual cost, represented by line B. The actual cost is a good deal higher because more tests are added to the profile as bed size increases.

Figure II illustrates much the same pattern with CK isoenzymes. Line A reflects a near economy of scale at a uniform standard of six tests per patient, regardless of bed size. In actuality, the number of tests per patient grows with bed size and allocation of resources, as indicated in line B.

Obviously, a variety of factors contribute to hyperinflation of costs in hospitals and laboratories. But the unavoidable implication is that we cannot significantly reduce costs by consolidating our medical institutions.

What then is the solution? I will offer some suggestions for reducing laboratory costs in a second, concluding article, which will appear in next month's issue of MLO.

Table: What it takes to run a hospital for a year

Table: How location affects hospital costs

Table: Teaching vs. community hospitals

Table: Lab utilization in teaching hospitals

Table: Lab utilization in community hospitals

Table: Figure I; Multiphasic profile costs

Table: Figure II; CK isoenzyme costs