Cardiac Revascularization in Specialty and General Hospitals
http://www.100md.com
《新英格兰医药杂志》
To the Editor: We investigated whether the findings of Cram et al. (April 7 issue)1 regarding severity of illness, mortality, and length of stay among Medicare patients undergoing revascularization in specialty hospitals could be generalized to an all-payer population. We used data for January 2002 through September 2004 from Solucient's all-payer Projected Inpatient Database, which contains information on more than 17 million discharges in the United States annually and has been used for the analysis of patients with cardiac disease.2 We identified specialty hospitals using the 2003 General Accounting Office definition,3 modified to include only hospitals with a majority of patients in Major Diagnostic Category 5 (denoting those with diseases of the circulatory system). Nonspecialty hospitals were in the same hospital-referral regions.4
We found that patients who underwent percutaneous coronary intervention (PCI) or coronary-artery bypass grafting (CABG) at specialty hospitals were less severely ill than those who underwent one of these procedures at general hospitals, on the basis of the probability of death according to All Patient Refined Diagnosis Related Groups.5 Although the unadjusted mortality rate for PCI was lower for specialty hospitals, no significant difference remained after adjustment for patient characteristics, admission source, hospital-referral region, payer, and volume of procedures. Unadjusted and adjusted lengths of stay for PCI and unadjusted length of stay for CABG were significantly lower for specialty hospitals. The adjusted length of stay for CABG did not differ significantly between the two types of hospital (Table 1).
Table 1. Risk of Death, Mortality Rate, and Length of Stay in Patients Undergoing PCI and CABG in Specialty Hospitals, as Compared with Nonspecialty Hospitals.
Janet K. Young, M.D., M.H.S.A.
David A. Foster, Ph.D., M.P.H.
Sivana T. Heller, M.D., M.P.H.
Solucient
Ann Arbor, MI 48108
jyoung@solucient.com
References
Cram P, Rosenthal GE, Vaughan-Sarrazin MS. Cardiac revascularization in specialty and general hospitals. N Engl J Med 2005;352:1454-1462.
Young JK, Foster DA. Cardiovascular procedures in patients with mental disorders. JAMA 2000;283:3198-3198.
Specialty hospitals: geographic location, services provided, and financial performance. Washington, D.C.: General Accounting Office, October 2003:1-41. (GAO-04-167.) (Accessed June 9, 2005, at http://www.gao.gov//new.items/d04167.pdf.)
Dartmouth atlas of health care. Hanover, N.H.: Dartmouth Medical School Center for the Evaluative Clinical Sciences, 2003. (Accessed June 9, 2005, at http://www.dartmouthatlas.org/.)
All Patient Refined Diagnosis Related Groups definitions manual, version 15.0. Wallingford, Ct.: 3M Health Information Systems, 1998.
To the Editor: In the article by Cram et al., the confidence intervals of the adjusted odds ratios are consistent with a large survival benefit when revascularization is performed in specialty hospitals: a 31 percent reduction in death for PCI and 28 percent for CABG. When adjustment is also made for the hospital's volume of procedures (inappropriately, given its role in the causal pathway related to hospital quality and efficiency), the reduction is 17 percent for PCI and 26 percent for CABG. The analysis omits longer-term survival and clinically specific process-of-care data in the public domain. In this regard, Baylor Heart and Vascular Hospital has consistently achieved 100 percent performance for these measures while disseminating its best practices to improve care across Baylor hospitals (www.hospitalcompare.hhs.gov/). The low power to detect differences in length of stay is reflected in the confidence intervals of the adjusted odds ratios, which are consistent with a 15 percent shorter length of stay for specialty hospitals. Finally, cost analyses might reveal more efficient resource use within a given length of stay, as has been observed for the use of pharmaceuticals and other supplies in the Baylor cardiovascular specialty hospital.
David J. Ballard, M.D., Ph.D.
Baylor Health Care System
Dallas, TX 75206
To the Editor: Cram et al. used Medicare Provider Analysis and Review (MedPAR) data to examine relationships between specialty hospitals and general hospitals. Although the study was well conceived, we are concerned about its total reliance on Medicare-specific administrative data. As acknowledged by the authors, claims data may be inadequate to reflect severity of illness meaningfully. This inherent deficiency of administrative data may be understated in the study by Cram et al.
We believe that the conclusions are mitigated by the shortcomings of administrative data. We strongly advocate the use of clinical data in investigations linked to severity of illness. The advantages of clinical data have been clearly illustrated in recent years,1,2 and significant differences in patient volume and outcome have been demonstrated in comparisons of clinical and administrative data.2
We recognize that data from administrative databases provide reasonable information for the assessment of general quality, but the inherent disadvantages of claims data necessarily limit the conclusions. Accordingly, we maintain that data from clinical databases provide the optimal analytic tool for investigations in cardiac surgery.
Fred H. Edwards, M.D.
Society of Thoracic Surgeons
Chicago, IL 60611
Karl F. Welke, M.D.
Oregon Health and Science University
Portland, OR 97239
Sidney Levitsky, M.D.
Society of Thoracic Surgeons
Chicago, IL 60611
References
Welke KF, Ferguson TB Jr, Coombs LP, et al. Validity of the Society of Thoracic Surgeons national adult cardiac surgery database. Ann Thorac Surg 2004;77:1137-1139.
Mack MJ, Herbert M, Prince S, Dewey TM, Magee MJ, Edgerton JR. Does reporting of coronary artery bypass grafting outcomes from administrative databases accurately reflect actual clinical outcomes? Presented at the 84th Annual Meeting of the American Association for Thoracic Surgery, Toronto, April 25–28, 2004.
The authors reply: We are pleased that Dr. Young and colleagues found the results of our analyses of data on Medicare patients to be largely generalizable to an all-payer population. Their analyses are timely, since roughly half of revascularization procedures are performed in non-Medicare patients.
We generally agree with the issues raised in Dr. Ballard's letter, but we would note that it is not customary to focus on the lower bounds of a 95 percent confidence interval when interpreting study results. A more balanced interpretation of our results is that the odds of death for patients at specialty hospitals was 11 percent lower for PCI and 16 percent lower for CABG after adjustment for patient characteristics. We agree that it is not unreasonable to argue that, because specialization is closely tied to procedural volume, comparisons of specialty and general hospitals should be based on risk-adjusted (but not volume-adjusted) mortality rates. However, given that some have argued that there is an inherent advantage in specialization above and beyond the "volume effect," we felt it was important to control for volume in our analyses in an attempt to tease apart these differences.
We believe that our analyses indicate that the lower unadjusted mortality at specialty hospitals is largely attributable to the lower average risk of the patients such hospitals treat and to the hospitals' greater procedural volumes. Perhaps most informative were the results of stratified analyses that directly compared lower-volume and higher-volume specialty and general hospitals and did not show significant differences. Finally, we agree with Dr. Ballard that other indicators of quality, such as process-of-care measures, which are difficult to derive from claims data, should ideally be examined.
We also agree with Dr. Edwards and colleagues that analyses based on administrative data are subject to a number of potential methodologic limitations when the quality of care is assessed. However, the lack of registries with detailed clinical information that can be used to track patient outcomes on a national level makes it impossible to conduct comparative analyses of specialty and general hospitals, since the validity of such analyses is predicated on the availability of data from all hospitals in the markets studied. We believe that, in the absence of national clinical registries, use of available administrative databases to provide empirical insight into important questions relevant to policy is preferable to not addressing these questions; however, the limitations of such analyses should be clearly recognized.
Peter Cram, M.D., M.B.A.
Gary E. Rosenthal, M.D.
University of Iowa College of Medicine
Iowa City, IA 52242
peter-cram@uiowa.edu
Mary S. Vaughan-Sarrazin, Ph.D.
Iowa City Veterans Affairs Center of Excellence
Iowa City, IA 52240
We found that patients who underwent percutaneous coronary intervention (PCI) or coronary-artery bypass grafting (CABG) at specialty hospitals were less severely ill than those who underwent one of these procedures at general hospitals, on the basis of the probability of death according to All Patient Refined Diagnosis Related Groups.5 Although the unadjusted mortality rate for PCI was lower for specialty hospitals, no significant difference remained after adjustment for patient characteristics, admission source, hospital-referral region, payer, and volume of procedures. Unadjusted and adjusted lengths of stay for PCI and unadjusted length of stay for CABG were significantly lower for specialty hospitals. The adjusted length of stay for CABG did not differ significantly between the two types of hospital (Table 1).
Table 1. Risk of Death, Mortality Rate, and Length of Stay in Patients Undergoing PCI and CABG in Specialty Hospitals, as Compared with Nonspecialty Hospitals.
Janet K. Young, M.D., M.H.S.A.
David A. Foster, Ph.D., M.P.H.
Sivana T. Heller, M.D., M.P.H.
Solucient
Ann Arbor, MI 48108
jyoung@solucient.com
References
Cram P, Rosenthal GE, Vaughan-Sarrazin MS. Cardiac revascularization in specialty and general hospitals. N Engl J Med 2005;352:1454-1462.
Young JK, Foster DA. Cardiovascular procedures in patients with mental disorders. JAMA 2000;283:3198-3198.
Specialty hospitals: geographic location, services provided, and financial performance. Washington, D.C.: General Accounting Office, October 2003:1-41. (GAO-04-167.) (Accessed June 9, 2005, at http://www.gao.gov//new.items/d04167.pdf.)
Dartmouth atlas of health care. Hanover, N.H.: Dartmouth Medical School Center for the Evaluative Clinical Sciences, 2003. (Accessed June 9, 2005, at http://www.dartmouthatlas.org/.)
All Patient Refined Diagnosis Related Groups definitions manual, version 15.0. Wallingford, Ct.: 3M Health Information Systems, 1998.
To the Editor: In the article by Cram et al., the confidence intervals of the adjusted odds ratios are consistent with a large survival benefit when revascularization is performed in specialty hospitals: a 31 percent reduction in death for PCI and 28 percent for CABG. When adjustment is also made for the hospital's volume of procedures (inappropriately, given its role in the causal pathway related to hospital quality and efficiency), the reduction is 17 percent for PCI and 26 percent for CABG. The analysis omits longer-term survival and clinically specific process-of-care data in the public domain. In this regard, Baylor Heart and Vascular Hospital has consistently achieved 100 percent performance for these measures while disseminating its best practices to improve care across Baylor hospitals (www.hospitalcompare.hhs.gov/). The low power to detect differences in length of stay is reflected in the confidence intervals of the adjusted odds ratios, which are consistent with a 15 percent shorter length of stay for specialty hospitals. Finally, cost analyses might reveal more efficient resource use within a given length of stay, as has been observed for the use of pharmaceuticals and other supplies in the Baylor cardiovascular specialty hospital.
David J. Ballard, M.D., Ph.D.
Baylor Health Care System
Dallas, TX 75206
To the Editor: Cram et al. used Medicare Provider Analysis and Review (MedPAR) data to examine relationships between specialty hospitals and general hospitals. Although the study was well conceived, we are concerned about its total reliance on Medicare-specific administrative data. As acknowledged by the authors, claims data may be inadequate to reflect severity of illness meaningfully. This inherent deficiency of administrative data may be understated in the study by Cram et al.
We believe that the conclusions are mitigated by the shortcomings of administrative data. We strongly advocate the use of clinical data in investigations linked to severity of illness. The advantages of clinical data have been clearly illustrated in recent years,1,2 and significant differences in patient volume and outcome have been demonstrated in comparisons of clinical and administrative data.2
We recognize that data from administrative databases provide reasonable information for the assessment of general quality, but the inherent disadvantages of claims data necessarily limit the conclusions. Accordingly, we maintain that data from clinical databases provide the optimal analytic tool for investigations in cardiac surgery.
Fred H. Edwards, M.D.
Society of Thoracic Surgeons
Chicago, IL 60611
Karl F. Welke, M.D.
Oregon Health and Science University
Portland, OR 97239
Sidney Levitsky, M.D.
Society of Thoracic Surgeons
Chicago, IL 60611
References
Welke KF, Ferguson TB Jr, Coombs LP, et al. Validity of the Society of Thoracic Surgeons national adult cardiac surgery database. Ann Thorac Surg 2004;77:1137-1139.
Mack MJ, Herbert M, Prince S, Dewey TM, Magee MJ, Edgerton JR. Does reporting of coronary artery bypass grafting outcomes from administrative databases accurately reflect actual clinical outcomes? Presented at the 84th Annual Meeting of the American Association for Thoracic Surgery, Toronto, April 25–28, 2004.
The authors reply: We are pleased that Dr. Young and colleagues found the results of our analyses of data on Medicare patients to be largely generalizable to an all-payer population. Their analyses are timely, since roughly half of revascularization procedures are performed in non-Medicare patients.
We generally agree with the issues raised in Dr. Ballard's letter, but we would note that it is not customary to focus on the lower bounds of a 95 percent confidence interval when interpreting study results. A more balanced interpretation of our results is that the odds of death for patients at specialty hospitals was 11 percent lower for PCI and 16 percent lower for CABG after adjustment for patient characteristics. We agree that it is not unreasonable to argue that, because specialization is closely tied to procedural volume, comparisons of specialty and general hospitals should be based on risk-adjusted (but not volume-adjusted) mortality rates. However, given that some have argued that there is an inherent advantage in specialization above and beyond the "volume effect," we felt it was important to control for volume in our analyses in an attempt to tease apart these differences.
We believe that our analyses indicate that the lower unadjusted mortality at specialty hospitals is largely attributable to the lower average risk of the patients such hospitals treat and to the hospitals' greater procedural volumes. Perhaps most informative were the results of stratified analyses that directly compared lower-volume and higher-volume specialty and general hospitals and did not show significant differences. Finally, we agree with Dr. Ballard that other indicators of quality, such as process-of-care measures, which are difficult to derive from claims data, should ideally be examined.
We also agree with Dr. Edwards and colleagues that analyses based on administrative data are subject to a number of potential methodologic limitations when the quality of care is assessed. However, the lack of registries with detailed clinical information that can be used to track patient outcomes on a national level makes it impossible to conduct comparative analyses of specialty and general hospitals, since the validity of such analyses is predicated on the availability of data from all hospitals in the markets studied. We believe that, in the absence of national clinical registries, use of available administrative databases to provide empirical insight into important questions relevant to policy is preferable to not addressing these questions; however, the limitations of such analyses should be clearly recognized.
Peter Cram, M.D., M.B.A.
Gary E. Rosenthal, M.D.
University of Iowa College of Medicine
Iowa City, IA 52242
peter-cram@uiowa.edu
Mary S. Vaughan-Sarrazin, Ph.D.
Iowa City Veterans Affairs Center of Excellence
Iowa City, IA 52240