Hospital volume or physician caseload and patient outcome associations have been extensively studied, all-causes mortality being the most frequently used outcome of medical care. More than 100 studies have demonstrated some association between low-volume healthcare providers and a relatively poor outcome, 1 and most of these studies have focused on hospital volumes of major surgical procedures and in-hospital or 30-day mortality. 2 However, similar associations have also been found for nonsurgical conditions requiring hospitalization such as trauma, subarachnoid hemorrhage, acute myocardial infarction (AMI), respiratory distress, cirrhosis, cancer, acquired immune deficiency syndrome, and adult and neonatal intensive care. 2–4 The association between low volume and poor outcome has also been found for individual physicians performing vascular surgery, 5 general surgery, 6 plastic surgery, 7 orthopedic surgery, 8 trauma care, 9 endoscopic procedures, 10,11 or coronary angioplasty. 12,13
Based on these findings, several researchers have recommended selective referrals of patients from low-volume to high-volume hospitals, also described as “evidence-based hospital referrals,” as the remedy for poor outcome. 1,2,4,14–19 Birk-meyer 17 has suggested that patients be educated to select a superior hospital or physician for a procedure on the basis of provider volume. Implementation of these suggestions and recommendations would pose a drastic healthcare policy decision that would lead to a major change in the healthcare system with significant political and economic implications. Before any policy decisions are made, the evidence supporting the patient referrals or public release of provider volumes should be examined for their reliability and scientific value.
Ideally, a public health risk factor should be an established cause of adverse outcomes before interventions are implemented to eliminate it. The provider volume has not been established as the cause or one of the main causes of adverse outcome. In healthcare research, only prospective, controlled trials can provide conclusive and reliable evidence of causality. Accordingly, if increasing the provider volume results in a significant decrease in adverse outcome (relative to the outcome for the control provider), low volume could be regarded as a cause of the adverse outcome. Even better evidence of causality could come from a “crossover” trial in which the provider volume is increased and then decreased below the baseline volume. No such experimental data on the effect of provider volume on the outcome have been documented in the medical literature.
In the absence of experimental data, objective judgment is used to draw inferences about causality 20–23 from the available evidence. This commentary evaluates the provider volume and mortality association, and the policy implications of the volume-based patient referrals are discussed.
Temporality of Variables
The risk factor or the potential cause must precede the outcome of interest. Almost all the studies published in the literature were based on cross-sectional data on the provider volume and outcome. It could be that high-volume hospitals attract greater volumes because of their better record in terms of the outcomes. 2 Similarly, physicians with a reputation for superior patient outcome could attract high caseloads. Consequently, higher volume might not be the cause of better outcome.
Plausibility
Although hospitals generally provide equipment, facilities, and technical staff for diagnostic and therapeutic procedures, the in-hospital mortality could reflect the skill and experience of the attending physicians and the supporting clinical teams. 1,5,24 A high hospital volume per se could not result in a better outcome.
Specificity
As seen in the data on the quality of inpatient care for AMI, 4,25 there are always some outliers within the aggregated groups of providers. The results of vascular surgery performed in some low-volume hospitals can be as good as those in many high-volume hospitals, 5 or a high-caseload surgeon in a community hospital could have the worst results. 26 Conversely, low-volume providers can have an excellent outcome. 27 The same surgeon could perform procedures in a large, high-volume hospital as well as in a small, low-volume hospital. 28 Some surgeons with low caseloads include those who are newly trained but have state-of-the-art skills, and very experienced surgeons with good outcome records could be low-volume providers because they have reduced their workload before retirement. 28 It has also been suggested that individual surgeons’ caseloads within high- or low-volume hospitals is not critical for the level of an outcome. 29 High volume is not well defined, and in some situations, it cannot be defined. 30 High-volume hospitals in one area could be low volume in other areas. Apparently, there is no consistency between studies with respect to what constitutes high volume or low volume, and there is no scientific basis for it. Apparently, the researchers have set arbitrary cutoff points to suit their study results.
Predictability
Some of the studies of volume–outcome association reported in the literature did not adjust or fully adjust for the confounding effects of age, the severity of illness, comorbidity, and clinical process factors. In a recent study of trauma centers, 31 the in-hospital mortality adjusted for the severity of injury and patient characteristics was associated with volume only for patients at high risk of death. In a study of inpatient care for AMI, 32 those elderly patients in low-volume hospitals who were transferred to high-volume hospitals had less severe illness and comorbidity than those not transferred. The former group of patients had been seen early and treated with thrombolytic therapy before transfer. Consequently, it was not surprising to find relatively low mortality among the transferred patients, and most of this mortality deficit was clearly not attributable to the care provided at the high-volume hospitals. Furthermore, the frequency of prescribing appropriate medications at discharge was similar in both the low-volume and high-volume hospitals. Adjustments to volume–mortality associations for the severity of illness and comorbidity were clearly not sufficient to control confounding. Risk estimates based on administrative data, such as Dudley et al.’s estimates, 2 are misleading because appropriate risk adjustment requires clinical data. 33
In another study, inpatient care for AMI according to the clinical guidelines accounted for only one third of the survival benefit attributable to high-volume hospitals after adjustments for other clinical, demographic, and health system-related variables. 4 This finding clearly suggests that hospital volume was not the major cause of in-hospital mortality. The direct measures of the quality of medical care such as the clinical processes are known to be more reliable predictors of the outcome. 33–35
The Medicare Part A claims data on the adverse outcome of carotid endarterectomy procedure show that the number of acute strokes during the hospital stay for endarterectomy and the length of stay have been decreasing over the years. 36 It appears that the rate of postoperative stroke was a function of the length of stay. Similarly, the length of stay would affect in-hospital mortality. Because hospital volume is inversely associated with the length of stay, 31 it could confound the hospital volume and mortality associations. 37
Consistency
Some of the published studies do not show associations between low mortality and high volume of certain procedures or therapies performed in hospitals. 1,38–41 At least 25 of the 128 studies reviewed by Dudley et al. 2 did not show statistically significant differences in surgical mortality between low-volume and high-volume hospitals. These 25 studies included all of the 4 studies of emergent abdominal aortic aneurysm repair surgery and the only study of emergent coronary angioplasty. Some studies of physician caseload also did not find any association with the outcome, for example, for coronary artery bypass graft and pediatric surgery. 42,43 Mayo et al. 44 found no significant difference between high-volume and low-volume hospitals or between high-caseload and low-caseload physicians with respect to stroke rate following carotid endarterectomy. Khuri et al. 33,45 analyzed data on 68,631 cases of 9 major surgical procedures performed in 44 Veterans Affairs medical centers and found no association between any procedure or the specialty volume and 30-day mortality. Another review of 88 studies showed no association of outcome with hospital volume in 22% of the studies and with physician caseload in 26% of the studies. 46 Furthermore, 63% of these studies were limited to New York State so that their results could not be generalized. In 2 studies, the risk of adverse outcome was associated with the severity of the disease or injury, not the hospital or physician volume. 47,48 One study demonstrated association of 30-day mortality with physician caseload, but not with hospital volume 30 or vice versa. 29 In a prospective observational study of nearly 500 community hospitals, a volume–outcome association existed for some conditions/procedures but not for others. 49 In addition to these studies, there could be many other unpublished studies that failed to demonstrate such a relationship. 46 All these data clearly demonstrate that there is no consistency between the studies of provider volume and outcome, and that the observed associations in some studies most probably were secondary associations, not causal associations.
Strength of Association
In several studies such as the studies of stroke following carotid endarterectomy, the association between provider volume and outcome was weak. Such statistically significant associations might not be of sufficient magnitude to be clinically significant. 28 A review of 88 studies 46 showed that many studies were small and of inferior quality design, and the effect of volume was most marked for only the complex surgical procedures for cancer.
Dose–Response
A linear or dose–response relationship is convincing evidence of a robust association. Because provider volume and mortality are continuous variables, volume could be regressed on death rate. Causality would be indicated if a unit change in volume results in a unit change in outcome. Most of the studies that showed a provider volume–outcome association dichotomized the measure of volume. In studies that used multiple categories of provider volume or volume as a continuous variable, its association with adverse outcome was not linear, so that the medium-caseload surgeons achieved the best outcome 50,51 or the worst outcome. 29,34 The quality of care provided by those low-volume providers who are affiliated with high-volume providers such as teaching hospitals could be similar to that provided by the high-volume providers. This effect was clearly evident in a study of liver transplantation procedures in which the low-volume centers that were affiliated with high-volume centers had 1-year mortality rates similar to that of the high-volume centers. 52
Policy Implications
This discussion suggests that healthcare provider volume is a nonspecific, indirect, and unreliable measure of provider performance. The temporal relationship and the biologic gradient of provider volume and outcomes are not established. Their associations are implausible, inconsistent, and often weak. Consequently, causality is not established. Patient referrals from one provider to another are a critical issue. If such a policy, based on insufficient or weak evidence, is implemented, it can have disastrous effects.
The low-volume providers are already referring patients to experts for the treatment of rare, serious diseases and complex procedures for which they are not equipped. For example, low-volume hospitals and surgeons must be referring Medicare patients requiring liver, heart, lung, or heart–lung transplantation to high-volume providers, because the Medicare program restricts these procedures to be performed by only those medical centers that can demonstrate a volume of 10 or more procedures per year. 53 The data on the performance of healthcare providers are also being collected and used for quality improvement and released for consumer use. 54 Consequently, public release of data on provider-specific volume and additional patient referrals from low-volume to high-volume providers are not justified on the basis of the published studies. Besides these concerns, there are 3 good reasons for not implementing such a policy.
First, the release of provider volume data for patients and their physicians to choose better providers is an impractical and unrealistic proposition. Acutely ill patients cannot timely access, interpret, and use provider-specific data. 32,55 The policy of routine referrals based solely on provider volume would also be misguided, inappropriate, and not feasible. Emphasis on volume will push some hospitals to merge and increase their overall volume without paying any attention to improving the quality of care or patient outcome. It has also been suggested that healthcare providers might perform medically unnecessary procedures merely to increase their volume. 28
Second, several authors have speculated the adverse effects of referral of patients from high-volume to low-volume providers. For example, large numbers of referred patients could disrupt clinical processes and create waiting lists in high-volume hospitals, limit competition, increase healthcare costs, and decrease the quality of care in and patient satisfaction with these hospitals; consequently, low-volume hospitals would lose specialists and economic viability. 2,46,56 Others have suggested that increases in volume could strain provider resources and actually worsen patient outcome. 50,57 If referrals were made solely on the basis of volume, patients at some high-volume hospitals with above-average risk-adjusted mortality rates would be cared for by providers with inferior performance, 28 low-volume hospitals with good outcomes would be treated unfairly, 58 and some low-volume hospitals would be turned into very low-volume hospitals. 59 Furthermore, patient referrals would deny preference for and access to local healthcare providers, increase transport costs, and more importantly, increase the critical transport time, particularly in rural and low-volume areas. 55 For example, intravenous tissue plasminogen activator and thrombolytic therapy soon after the onset of acute stroke and AMI, respectively, can reduce morbidity and mortality in eligible patients. 60,61 Patients from remote areas referred for such therapy would be exposed to morbidity and mortality risks en route to high-volume hospitals because of the transportation delays. 4,62 Consequently, referrals could do more harm than good to patients.
Third, if the quality of care provided by the low-volume providers is improved, their patient outcomes would improve to a level in which patient referrals would not be necessary. This common sense approach was clearly demonstrated by Sallano et al., 63 who reanalyzed the coronary artery bypass graft procedure mortality data for New York State; they found that a statewide quality improvement program had eliminated the previously observed procedure volume–outcome association. A recent study of survival following lung resection for cancer in relation with hospital volume also suggests that “rectifiable variations in care may account for differences in survival.”57 Clinical guidelines are available for most procedures and clinical processes. Many quality improvement models have been tested in a variety of clinical settings, and some have been found to be effective. These intervention strategies include telemedicine, 64 low-caseload physicians working with high-caseload physicians, 35 collaborative quality improvement projects, 54 attracting experienced physicians to low-volume areas, and provider education. Recently, Congress passed a bill entitled “Medicare, Medicaid and SCHIP Benefits Improvement and Protection Act of 2000.” This legislation modifies and improves Medicare policies that will increase disproportionate share hospital add-on payments for rural hospitals, reimburse physicians at 115% of the fee schedule, and pay emergency room physicians and ambulances in the rural areas at a reasonable cost. However, more needs to be done to reduce the variance between high- and low-volume providers with respect to the quality of care they provide and its outcome.
Recommendations
The medical profession should take the lead to ensure good quality of medical care for all without curtailing access and to develop reliable measures of the quality of care. The emphasis should be on improving the quality of surgical and nonsurgical care, which in turn will improve the outcome regardless of the level of provider volume. All healthcare providers should be educated regarding the quality of care they provide and how it can be measured and improved. Not all hospitals will buy into improving and providing superior services in all areas. Pooling resources and merging services should be the goal for those low-volume providers. They should form partnerships or consortia so that neighboring hospitals share patients, skilled physicians, and services. Although accreditation and licensing agencies should require minimum standards of care, they should not interfere in hospitals’ and physicians’ practices and business decisions. Improvements in the quality of care provided by low-volume providers should be voluntary, but the profession should find a way for compliance with standard clinical guidelines. Policies that will lead to closing hospitals because of their low volume are unnecessary and undesirable. Further research should be conducted to identify the modifiable risk factors for poor quality of care and poor outcome. Meanwhile, the policy of further patient referrals based on healthcare provider volume should be abandoned.
References
1. Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation between surgical volume and mortality. N Engl J Med 1979; 301: 1364–1369.
2. Dudley RA, Lohansen KL, Brand R, et al. Selective referral to high-volume hospitals: estimating potentially avoidable deaths. JAMA 2000; 283: 1159–1166.
3. Phibbs CS, Bronstein JM, Buxton E, et al. The effects of patient volume and level of care at the hospital of birth on neonatal mortality. JAMA 1996; 276: 1054–1059.
4. Thieman DR, Coresh J, Oetgen WJ, et al. The association between hospital volume and survival after acute myocardial infarction in elderly patients. N Engl J Med 1999; 340: 1640–1648.
5. Hertzer NR. Presidential address: outcome assessment in vascular surgery—results mean everything. J Vasc Surg 1995; 21: 6–15.
6. Gordon TA, Burleyson GP, Tielsch JM, et al. The effects of regionalization on cost and outcome for one general high-risk surgical procedure. Ann Surg 1995; 221: 43–49.
7. Witt PD, Wahlen JC, Marsh JL, et al. The effect of surgeon experience on velopharyngeal functional outcome following palatoplasty: is there a learning curve? Plast Reconstr Surg 1998; 102: 1375–1384.
8. Lavernia CJ, Guzman JF. Relationship of surgical volume to short-term mortality, morbidity, and hospital charges in arthroplasty. J Arthroplasty 1995; 10: 133–140.
9. O’Keefe GE, Jurkovich GJ, Copass M, et al. Ten-year trend in survival and resource utilization at a level I trauma center. Ann Surg 1999; 229: 409–415.
10. Freeman ML. Complications of endoscopic biliary sphincterotomy: a review. Endoscopy 1997; 29: 288–297.
11. Rabenstein T, Schneider HT, Nicklas M, et al. Impact of skill and experience of the endoscopist on the outcome of endoscopic sphincterotomy techniques. Gastrointest Endosc 1999; 50: 628–636.
12. Shook TL, Sun GW, Burstein S, et al. Comparison of percutaneous transluminal coronary angioplasty outcome and hospital costs for low-volume and high-volume operators. Am J Cardiol 1996; 15: 331–336.
13. McGrath PD, Wennberg DE, Dickens JD, et al. Relation between operator and hospital volume and outcomes following percutaneous coronary interventions in the era of the coronary stent. JAMA 2000; 284: 3139–3144.
14. Grumbach K, Anderson GM, Luft HS, et al. Regionalization of cardiac surgery in the United States and Canada: geographic access, choice, and outcome. JAMA 1995; 274: 1282–1288.
15. Hannan EL. The relation between volume and outcome in health care. N Engl J Med 1999; 340: 1677–1679.
16. Bodenheimer T. The American health care system: the movement for improved quality in health care. N Engl J Med 1999; 340: 488–492.
17. Birkmeyer JD. High-risk surgery—follow the crowd. JAMA 2000; 283: 1191–1193.
18. Hillner BE, Smith TJ, Desch CE. Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care. J Clin Oncol 2000; 18: 2327–2340.
19. Hewitt M. Interpreting the Volume–Outcome Relationship in the Context of Health Care Quality: Workshop Summary. Washington, DC: National Academy of Sciences; 2000.
20. MacMahon B, Pugh TF. Epidemiology: Principles and Methods. Boston: Little, Brown & Company; 1970.
21. Susser M. Causal Thinking in the Health Sciences: Concepts and Strategies of Epidemiology. New York: Oxford University Press; 1973.
22. Rothman KJ. Modern Epidemiology. Boston: Little, Brown & Company; 1986.
23. Wan TTH. Assessing causality: foundations for population-based health care managerial decision making. In: Oleske DM, ed. Epidemiology and the Delivery of Health Care Services: Methods and Applications. New York: Kluwer Academic/Plenum Publishers; 2001: 75–98.
24. Begg CB, Cramer LD, Hoskins WJ, et al. Influence of hospital volume on mortality following major cancer surgery. JAMA 1999; 281: 1375.
25. Sheikh K, Bullock C. Urban–rural differences in the quality of care for Medicare patients with acute myocardial infarction. Arch Intern Med 2001; 161: 737–743.
26. Gibbs BF, Guzzetta VJ. Carotid endarterectomy in community practice: surgeon specific versus institutional results. Ann Vasc Surg 1989; 3: 307–312.
27. Schwarz RE, Ellenhorn JD. Influence of hospital volume on mortality following major cancer surgery. JAMA 1999; 281: 1374.
28. Hewitt M. Interpreting the volume–Outcome Relationship in the Context of Health Care Quality: Workshop Summary. Washington, DC: Institute of Medicine; 2000.
29. Cebul RD, Snow RJ, Pine R, et al. Indications, outcomes, and provider volumes for carotid endarterectomy. JAMA 1998; 279: 1282–1287.
30. Tu JV, Austin PC, Chan BTB. Relationship between annual volume of patients treated by admitting physician and mortality after acute myocardial infarction. JAMA 2001; 285: 3116–3122.
31. Nathens AB, Jurkovich GJ, Maier RV, et al. Relationship between trauma center volume and outcomes. JAMA 2001; 285: 1164–1171.
32. Mehta RH, Stalhandske EJ, McCargar PA, et al. Elderly patients at high risk with acute myocardial infarction are more frequently transferred from community hospitals to tertiary centers: reality or myth? Am Heart J 1999; 138: 688–695.
33. Khuri SF. Invited commentary: surgeons, not General Motors, should set standards for surgical care. Surgery 2001; 130: 429–431.
34. U.S. Congress, Office of Technology Assessment. Identifying Health Technology That Work: Searching for Evidence. Washington, DC: U.S. Government Printing Office; 1994. Publication OTA-H-608.
35. Naylor CD. Public profiling of clinical performance. JAMA 2002; 287: 1323–1325.
36. Sheikh K. Sex differences in utilization of carotid endarterectomy, 30-day post-operative mortality and stroke. Neurology 2001; 56 (suppl 3): A160.
37. Jollis JG, Romano PS. Volume–outcome relationship in acute myocardial infarction. JAMA 2000; 284: 3169–3171.
38. Luft HS. The relation between surgical volume and mortality: an explosion of causal factors and alternative models. Med Care 1980; 18: 940–959.
39. Moscovice I. Strategies for promoting a viable rural health care system. J Rural Health 1989; 5: 216–230.
40. Shroyer AL, Marshall G, Warner BA, et al. No continuous relationship between veterans affairs hospital coronary artery bypass grafting surgical volume and operative mortality. Ann Thorac Surg 1996; 61: 17–20.
41. Reickert CA, Hirschl RB, Atkinson JB, et al. Congenital diaphragmatic hernia survival and use of extracorporal life support at selected level III nurseries with multimodality support. Surgery 1998; 123: 305–310.
42. Clark RE. Outcome as a function of annual coronary artery bypass graft volume. The Ad Hoc Committee on Cardiac Surgery Credentialing of the Society of Thoracic Surgeons. Ann Thorac Surg 1996; 61: 21–26.
43. Hartz AJ, Kuhn EM, Krakauer H. The relationship of the value of outcome comparisons to the number of patients per provider. Int J Qual Health Care 1997; 9: 247–254.
44. Mayo SW, Eldrup-Jorgensen J, Lucas FL, et al. Carotid endarterectomy after NASCET and ACAS: a statewide study. J Vasc Surg 1998; 27: 1017–1022.
45. Khuri SF, Daley J, Henderson W, et al. Relation of surgical volume to outcome in eight common operations. Ann Surg 1999; 230: 414–432.
46. Auerbach AD. Localizing care to high-volume centers. In: Shojania KG, Duncan BW, McDonald KM, et al, eds. Making Healthcare Safer: An Evidence-Based Compendium of Patient Safety Practices. Rockville, MD: U.S. Department of Health and Human Services, Agency for Health Care Research and Quality; 2001: 209–213.
47. Wen SW, Simunovic M, Williams JI, et al. Hospital volume, calendar age, and short term outcomes in patients undergoing repair of abdominal aortic aneurysms: the Ontario experience, 1988–92. J Epidemiol Community Health 1996; 50: 207–213.
48. Richardson JD, Schmieg R, Boaz P, et al. Impact of trauma attending surgeon case volume on outcome: is more better? J Trauma 1998; 44: 266–271.
49. Farley DE, Ozminkowski RJ. Volume–outcome relationships and in-hospital mortality. Med Care 1992; 30: 77–94.
50. Tepas JJ, Patel JC, DiScala C, et al. Relationship of trauma patient volume to outcome experience: can a relationship be defined? J Trauma 1998; 44: 827–830.
51. Harmon JW, Tang DG, Gordon TA, et al. Hospital volume can serve as a surrogate for surgeon volume for achieving excellent outcomes in colorectal resection. Ann Surg 1999; 230: 404–411.
52. Edwards EB, Roberts JP, McBride MA, et al. The effects of the volume of procedures at transplantation centers on mortality after liver transplantation. N Engl J Med 1999; 341: 2049–2053.
53. U.S. Department of Health and Human Services, Health Care Financing Administration. Medicare program; criteria for medicare coverage of lung transplants. Federal Register 1995; 60: 6537–6547.
54. Sheikh K. Defining and achieving quality of medical care. Am J Med Qual 1998; 13: 59–62.
55. Sheikh K. Utility of provider volume as an indicator of medical care quality and for policy decisions. Am J Med 2001; 111: 712–715.
56. Finlayson SR, Birkmeyer JD, Tosteson AN, et al. Patient preference for location of care: implications for regionalization. Med Care 1999; 37: 204–209.
57. Bach PB, Cramer LD, Schrag D, et al. The influence of hospital volume on survival after resection for lung cancer. N Engl J Med 2001; 345: 181–188.
58. Dudley RA, Johansen KL. Invited commentary: physician responses to purchaser quality initiatives for surgical procedures. Surgery 2001; 130: 425–428.
59. Birkmeyer JD, Finlayson VA, Birkmeyer CM. Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative. Surgery 2001; 130: 415–422.
60. NINDS (National Institute of Neurological Diseases and Stroke) rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333: 1581–1587.
61. FTT (Fibrinolytic Therapy Trialists’) Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative review of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Lancet 1994; 343: 311–322.
62. Boersma E, Maas ACP, Deckers JW, et al. Early thrombolytic treatment in acute myocardial infarction: reappraisal of the golden hour. Lancet 1996; 348: 771–775.
63. Sollano JA, Gelijns AC, Moskowitz AJ, et al. Volume–outcome relationships in cardiovascular operations: New York State, 1990–1995. J Thorac Cardiovasc Surg 1999; 117: 419–428.
64. Wang DZ, Rose JA, Honings DS, et al. Treating acute stroke patients with intravenous tPA. The OSF Stroke Network experience. Stroke 2000; 31: 77–81.
