RSS Feeds
-
The Authors Respond
Submit responseThe Authors reply:
Dr. Lobdell points out challenges inherent to interpreting results from analyses using highly detailed claims data. While our approach did not exclude patients on amiodarone or pressors specifically (1), it did exclude patients with a wide range of contraindications (hypotension, heart failure, and bleeding). Nevertheless, we may have misclassified a small number of patients with acceptable contraindications, perhaps partially explaining the inconsistent relationship between individual measures and mortality seen. However, it is less obvious how misclassification would produce a strong association between overall quality and improved outcomes.
While our risk adjustment models did not contain the clinical data used in the Society of Thoracic Surgeons approach, we saw no statistical association between surgeon or hospital volume and severity of illness as measured by APR-DRG risk of mortality score. As a result, we feel it would require a very large number of patients with important and unmeasured comorbidities to influence our results unduly. Interestingly, the difference between clinical data-driven models (such as STS’s) and administrative data-driven models power is relatively small when comparisons are made at the aggregate level (2) – such as in our study.
While Dr. Lobdell raises points potentially useful in interpreting our study findings, we would strongly disagree that our results are driven entirely by poor quality data or inadequate risk adjustment. It is highly unlikely that the strong associations we saw between highest quality care and improved outcomes in our study are due to poor quality data. Rather, we hope our study prompts further examination of the benefits of improving the reliability of perioperative care, and compels the development of systems that maximize care quality from start to finish.
References:
1. Auerbach AD, Hilton JF, Maselli J, Pekow PS, Rothberg MB, Lindenauer PK. Shop for quality or volume? Volume, quality, and outcomes of coronary artery bypass surgery. Ann Intern Med. 2009;150(10):696-704.
2. Pine M, Jordan HS, Elixhauser A, et al. Enhancement of claims data to improve risk adjustment of hospital mortality. JAMA. 2007;297(1):71-6.
Conflict of Interest:
None declared
-
Shopping for Data: Quality or Volume?
Submit responseThe manuscript focuses our attention on important process and outcome measures in surgical coronary artery revascularization, but fails to address some very practical limitations of the voluminous administrative data. Specifically, quality measures related to statins, aspirin, and beta -blockers within two days of surgery is of limited value if not taken in context. Not uncommonly, a clinical situation may exclude a patient from one of the treatments, yet the exclusion criteria do not address these issues. Examples include: 1) typically, patients that remain mechanically ventilated would not be given a lipid-lowering agent, 2) patients with significant post-operative thrombocytopenia or coagulopathy may not be given aspirin until the platelet count increases or coagulopathy improves, 3) beta-blockers would not be administered when inotropes (mediated through beta-receptor agonists such as dopamine, dobutamine, epinephrine, and norepinephrine) are continued. Similarly, patients prophylaxed or treated with amiodarone for atrial fibrillation would not commonly be given beta-blockers.
A generally accepted risk model, such as the Society of Thoracic Surgeons-National Cardiac Database, would also have added significantly to value of the comparisons. Bias cannot be excluded, since some institutions or surgeons may operate on a higher percentage of "risky" patients than others.
Unfortunately, without the aforementioned contextual information, it is difficult to determine the value of this investigation. Let us suggest that as we shop for data, quality is more valuable than volume.
Conflict of Interest:
None declared