large share of individuals with pancreatic cancer are surgically treated in lower-volume hospitals. In the six studies described in Table 5.10, estimates varied from one-quarter to slightly more than one-half. Policies to shift care to higher-volume facilities would likely affect many individuals.

Specialization

Specialization of Facilities

Although specialization is difficult to separate analytically from volume, some feel that specialized cancer care facilities can provide better care, in part because of the concentration of expertise and resources. Some areas have regionalized services, which attempt to triage care so that the most complex cases are referred to the facility with the most sophisticated level of resources. Investigators have assessed the relationship between specialization and outcomes using different measures: tertiary care status, academic affiliation, or even hospital size.

The relationship between specialization and the processes and outcomes of cancer care was recently examined in a review of 46 empirical studies reported in the literature since 1980 (Grilli et al., 1998). Almost all of the studies reviewed found lower mortality associated with care provided by specialized centers and clinicians. For breast cancer, where there were a number of methodologically sound studies, a pooled estimate of the effect of specialization showed that specialized cancer care was associated with an 18 percent reduction in five-year mortality. In general, Grilli judged the evidence far from conclusive because of major methodological flaws and speculated that publication bias favoring specialized centers may have accounted for observed trends. Grilli concludes that the widespread belief that cancer patients are better off when treated in specialized centers is not supported by available evidence. What follows is a review of recent studies of the effects of specialization on cancer care processes and outcomes.

Munoz and colleagues (1997) from the National Cancer Institute (NCI) assessed in detail the process of ovarian cancer treatment for 785 women selected from SEER sites in 1991. Only about 10 percent of women with presumptive Stage I and II, compared to 71 percent with Stage III and 53 percent with Stage IV disease received recommended staging and treatment. The absence of lymphadenectomy and assignment of histologic grade were the primary reasons women with presumptive Stage I and II disease did not receive recommended staging and treatment. The principal deficiency in Stage III and IV disease was withholding of platinum-based chemotherapy in older women. The only provider variable assessed was whether care was provided at a hospital with a residency program. The odds ratio for receiving appropriate care was 1.9, or almost double, if care was given at a hospital with a residency program.

Gordon and colleagues (1995) looked at the effects of regionalization of care by comparing hospital mortality, length of stay, and costs of care for pancreatic cancer at one specialty center, Johns Hopkins (271 patients), in 1988-1993 with all other hospitals throughout the State of Maryland (230 patients in 38 hospitals) using hospital discharge data. The results were striking: in-hospital mortality 2.2 versus 13.5 percent (a relative risk of 6.1 [confidence interval {C.I.}] 2.9-12.7), average length of stay (23 versus 27 days), and average total charges were 20 percent greater outside Johns Hopkins. A follow-up study by Gordon et al. (1998) attributes 61 percent of the decline in the Maryland in-hospital mortality for pancreas surgery (17.2 to 4.9 per-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement