Hospital Volume Not Equal to Quality?

View on the News

The Devil in the Details

Procedural volume and its relationship to outcome has enormous implications, not just for patients undergoing surgical treatments, but in planning future surgical services. This study is important in respect of the large data set analysed and the finding that volume was only weakly associated with mortality compared to other factors such as patient co-morbidity. This raises questions about the current trend across the developed world to move towards high volume surgical units.

As impressive as the study initially appears there are some weaknesses that need to be considered. We of course only know about patients who were operated on and know nothing about patients declined surgery. There are studies that indicate patients are more likely to be declined appropriate surgery in units with less experience and lower volumes. Also of course many low volume units will refer high risk or complex cases to larger more experienced centres. It may be expected that some of this effect would be picked up in the Agency for Health Care Research and Quality comorbidities category (AHRQ) but the most usual reason for transfer to a regional unit is because of surgical complexity rather than patient co-morbidity.

This study also does not allow for any clinical governance measures that may be in force. Most if not all units with outcomes worse than average will take measures to address the problem ranging from ceasing undertaking the procedure to better case selection or improved team structure. At the time this study was conducted it is inconceivable to assume that no such measures were in action. It is incorrect then to conclude that this is an unselected study of the effects of volume. In the United Kingdom following concerns raised about outcomes from aortic surgery prospectively collected national data showed a very strong association with volume and outcome. The highest volume units had mortality 2.5 times lower than the lowest volume units (Outcomes after elective repair of infra-renal abdominal aortic aneurysm, March 2012; www.vascularsociety.org).

The authors conclude that volume is a poor indicator of quality of health care. This may be correct, but it is hard to suggest that based on this study as the authors did not measure factors such as length of hospital stay, re-admission rates, costs or patient experience of their care. In a review of cardiovascular services in London, not only did unit volume for AAA repair influence mortality it had a strong effect on efficiency. While patient outcome must be the main factor in planning service configurations, other issues such as cost and manpower need also to be considered.

Finally there are some practical issues. Small volume units can get good results. A unit that does 5 aortic aneurysm repairs in a year with no mortality will look very good. However, if they have a single mortality they will become one of the worst performing units overnight. Performance of units can only be assessed on large numbers. If a small volume unit is underperforming it may take a long time to identify that. There cannot be a surgical team in the world that is not aware that the more they work together to undertake complex surgical interventions the better they function across the board. So while the results of the study need to be considered carefully they do fly in the face of what most surgeons know in their hearts.

Dr. C. P. Shearman is Professor of Vascular Surgery, University of Southampton, United Kingdom, and an associate medical editor of Vascular Specialist.


 

If the causative factors could be identified, "then we could take the experience of high-volume centers and translate that to everybody else, so everybody could have good outcomes," he said.

According to Dr. Livingston, of the University of Texas, Dallas, previous studies relied on statistical modeling of the mortality relationship. "Those models are only as good as the model can represent the data," he said, and very few have been rigorously assessed to see how well they describe the phenomenon that they're trying to describe. This paper "should serve as the template for what everyone should do when they're performing volume outcome studies or any kind of regression analysis," he added.

The authors reported that they have no financial disclosures.

Pages