Surgical Caseload and the Volume-Outcome Effect

The observation that patients who receive care from high-volume providers typically have better outcomes compared with those treated by low-volume providers is called the volume-outcome effect.5,6 The volume-out-come effect was first observed in neurosurgery in relation to aneurysm care7 and has been subsequently confirmed for many other types of neurosurgical procedures, including craniotomy for many individual types of tumor.8–15

Because demonstration of the volume-outcome effect requires knowledge of outcomes across many providers (hospitals or surgeons), it is usually studied using administrative databases. This has some specific consequences that affect the design and interpretation of volume-out-come studies. First, outcome measures in such studies are not usually disease-specific. This means that mortality, adverse hospital discharge disposition, or cost measures, such as length of stay or hospital charges, are more likely to be the endpoint of an administrative database study, rather than extent of resection, specific neurological outcomes, or disease-free survival. Second, volume-outcome studies are nonrandomized. This means that if patients who obtain care from low-volume providers differ systematically from those who are cared for by high-volume providers, the study will be biased. Accurate risk adjustment for such “casemix differences” are critical in volume-outcome studies. Third, use of administrative databases also typically forces the use of coding systems that have not been designed to study specific diseases, and may have other shortcomings. In meningioma studies, for example, one consequence is that many studies cannot distinguish hemiparesis on admission from weakness that results from surgery. Vision loss is likely to be crudely coded as affecting one or both eyes, rather than specific information about acuity or fields. Many complications and adjunct procedures, such as embolization, may be systematically undercoded in administrative databases.

Despite these drawbacks, studies on short-term outcomes after meningioma surgery8,12 have shown better patient outcomes after surgery at higher-volume hospitals or from higher-volume surgeons. Curry et al8 studied 15,028 admissions for meningioma craniotomy in the United States from 1988 to 2000. In multivariate analysis adjusted for multiple casemix factors, they found significantly lower in-hospital mortality after surgery at higher-volume hospitals (p = 0.01) ( Fig. 43.1 ). There was a 26% reduction in mortality for each 10-fold increase in hospital case volume. The highest-volume quintile of hospitals in this study had 24 or more meningioma craniotomies per year. Bateman et al reported similar results in a cohort of adult meningioma patients treated from 1998 to 2002.12 Surgeon volume had no effect on mortality in the study by Curry et al, but both hospital and surgeon caseload were significantly (p < 0.001) related to less frequent adverse hospital discharge disposition (i.e., other than directly home). For most individual complications studied (such as hematomas, thrombotic complications, and transfusions), there were no differences between high- and low-volume providers. Paradoxically, neurological complications were more frequent with high-volume hospitals and surgeons. This could reflect a more complex case mix treated by high-volume providers. Only 6% of neurological complications were followed by in-hospital death at high-volume hospitals, compared with 19% at low-volume hospitals; this effect (better “rescue” at high-volume hospitals) has been identified in studies of general surgical procedures as well.16

Other studies of short-term outcomes after brain tumor craniotomies have identified similar volume-out-come effects after surgery for primary brain tumors,9 metastases,11 acoustic neuromas,17,18 adult brain tumors as a whole,13–15 and pediatric brain tumors.10 Most brain tumor volume-outcome studies to date have been from the United States, although in other medical areas the effect has been demonstrated in many different national health care systems.

The volume-outcome effect has usually been explained as a combination of the well-established existence of a learning curve for most surgical procedures (“practice makes perfect,” see later discussion) and from the systematization of care in the form of formal checklists or less formal habits that accompany any frequent activity.19 Other possible explanations for better outcomes at higher-volume hospitals include “structural” factors describing hospital resources, such as the presence of advanced operating room technology, dedicated intensive care units and intensivists, 24-hour resident coverage, and the number and level of training of nurses.20–22 Finally, the “perfect makes practice” or “selective referral” effect describes the ability of established high-volume centers to attract good-prognosis patients from a broad area, whereas lower-volume hospitals are more likely to receive the sickest, most emergent patients with a given diagnosis from a limited geographical catchment area.23 For example, lower-volume meningioma centers and surgeons treated a significantly higher proportion of patients with emergent or urgent presentation in the study by Curry et al (although conversely, complex patients with neurofibromatosis 2 tended to receive care at high-volume referral centers).8

Some surgeon-specific factors may also help to explain the volume-outcome effect. Integrated Medical Learning (IML) results from the 2007 CNS meeting demonstrated that surgeons with high-volume tumor practices displayed significantly greater knowledge about current neurooncology publications.24 Of four questions in this survey study on current literature knowledge, 54% of surgeons who saw fewer than 10 brain tumor patients per year scored zero correct, whereas 75% of surgeons who saw 50 brain tumor patients or more per year answered all questions correctly.

Regardless of the causes of the volume-outcome effect, can it be used as a tool for improving the quality of meningioma surgery in a population? Most answers to this type of question for other specific operations have taken one of two forms. First, demonstration of a volume-out-come effect for a procedure is often followed by a call to “regionalize” the procedure by restricting its performance to high-volume centers.25,26 Such initiatives have been performed by government mandate for some expensive, high-risk procedures, with mixed results.27–30 Regionalization can cause disadvantageous, unintended consequences, such as increased difficulty coordinating care with local primary providers, longer travel time for patients, and the general deterioration of skills in the general community when emergency care is necessary without transfer to regional centers; in addition, some patients simply prefer to receive care close to home.31–34 Careful study of both intended and unintended consequences is necessary before mandating regionalization of a procedure.35 That said, spontaneous regionalization of some neurosurgical procedures, including meningioma surgery, seems to be already taking place in the United States, following trends in other major surgical procedures such as cancer surgery.36,37 Between 1988 and 2000, the median annual hospital caseload of meningioma craniotomies in the United States increased from six to 10 cases per year; in 2007, the increase had continued to 15 per year (Barker FG, unpublished data, Nationwide Inpatient Sample [NIS], Health-care Cost and Utilization Project, Agency for Healthcare Research and Quality, Rockville, MD). In 1988, 10% of meningioma patients underwent surgery at hospitals where there was no other meningioma craniotomy that year, compared with 3.5% in 2000 (2.8% in 2007; Barker FG, unpublished data, NIS).8 Similar spontaneous centralization (i.e., not in response to government mandate) took place in the United States between 1988 and 2000 for pediatric tumor craniotomies10 and craniotomies for adult primary brain tumor.9 The forces causing these shifts in practice have not been studied.

The other general way to use the volume-outcome effect to improve care is to study processes of care that characterize high-volume centers and care delivered by specialists38 and use them to improve care at low-volume centers.39,40 This is sometimes called “floating all boats.” It is possible that the periodic published reports of “good outcomes at low-volume aneurysm centers”41–44 may result from conscious or unconscious applications of this strategy. At the individual practitioner level, a parallel is the high-quality care delivered by low-volume surgeons who have participated in a formal credentialing process, as in the context of a clinical trial.45–48 For meningioma care, processes that characterize high-volume providers have not been studied, nor have there been published reports of high-quality care achieved specifically at low-volume centers. In general terms, one example of possible meningioma process research would be to study the use of a fixed protocol for meningioma surgery resembling the “surgical checklists” that have been studied in general surgery.19 Postcraniotomy infection has been shown to be more common in brain tumor craniotomy than after other craniotomies,49 and a standardized protocol ensuring timely administration of antibiotics50 could easily be developed and tested. Protocols to retain patients through long-term follow-up periods after surgery could also be studied.

The demonstration that higher-volume providers are associated with better outcomes offers the possibility to use volume as a surrogate for quality of care, allowing investigators to ask questions about social variables (such as race and income) that could predict better access to high-quality care for meningiomas. In 2003, Curry et al demonstrated that high-volume meningioma surgery was associated with white race, private insurance, residence in higher-income areas, and lesser general medical comorbidity in the United States, 1988 to 2000.8 These disparities have demonstrable impact on patient outcomes: In the United States, 1988 to 2004, the same group demonstrated higher inpatient mortality for black patients compared with whites after meningioma craniotomy (2.7% vs 1.9%).51 The difference in mortality was significant when meningioma results were pooled with results for primary brain tumors, metastases, and acoustic neuromas, with an odds ratio of 1.6-fold higher mortality for blacks compared with whites (p < 0.001).51