Ann Thorac Surg 2001;72:340-341
© 2001 The Society of Thoracic Surgeons
Invited commentary
John D. Urschel, MDa
a Department of Surgery, McMaster University, St. Joseph Hospital, 50 Charlton Ave E, Hamilton, ON L8N 4A6, Canada
e-mail: urschelj{at}mcmaster.ca
There is an inverse relationship between hospital volume and in-hospital mortality for complex operations, such as esophagectomy for esophageal cancer. The report of Dimick and coauthors validates this relationship. Although this hospital volume-outcome relationship seems intuitively obvious the authors correctly point out several important uncertainties. To what extent are low-risk patients referred, or self-referred, to high-volume centers (selective referral hypothesis)? Can the relationship simply be explained by improvements in care brought about by institutional and health care team experience (practice makes perfect hypothesis)? Finally, is patient outcome dependent on hospital volume, surgeon volume, or a combination of the two? Health care policies are being developed based on the volume-outcome studies of Dimick and others, so a better understanding of the hospital volume-outcome relationship is needed.
The validity of volume-outcome studies can be assessed by posing four key questions: is the database accurate; are the patients similar; how were volume categories determined; and are the measured outcomes appropriate. Administrative databases usually capture major surgical procedures and mortality accurately, but the coding of secondary diagnoses is often suspect. This hampers the assessment of patient comorbidity and operative risk. Multivariate analysis is used to determine if patients in low and high volume hospitals are similar. Dimick’s data, and those from other investigators, show that high volume hospitals tend to attract patients with higher socioeconomic status, white race, and private insurance (selective referral). Determination of volume categories (defining numbers of patients for low and high categories) is a major problem in many volume-outcome studies. The cut off points should be specified a priori; otherwise the reader may wonder if the cut-off points were selected after the data analysis to maximize the volume-outcome relationship. Alternatively, volume can be studied as a continuous variable. Finally, operative mortality is a problematic outcome in all surgical outcome studies. Both in-hospital mortality and 30-day mortality should be captured but few administrative databases can do this. Future volume-outcome studies will hopefully address the common pitfalls of this type of research, and help us understand the reasons behind hospital volume-outcome relationships.
Related Article
-
Hospital volume is related to clinical and economic outcomes of esophageal resection in Maryland
- Justin B. Dimick, Stephen M. Cattaneo, Pamela A. Lipsett, Peter J. Pronovost, and Richard F. Heitmiller
Ann. Thorac. Surg. 2001 72: 334-340.
[Abstract]
[Full Text]
[PDF]
