VA may open surgical outcomes database

Risk-adjusted system covers all major procedures

A vast database of surgical outcomes developed by the Veterans Affairs may soon open up to subscribers from private hospitals.

In a pilot project, three hospitals have been selected to submit data via the Internet to the VA’s National Surgical Quality Improvement Program (NSQIP), a risk-adjusted outcomes monitoring system that began in 1991 and now encompasses 123 VA medical centers and more than 700,000 surgical cases. (The three hospitals are the departments of surgery at the University of Michigan in Ann Arbor, Emory University in Atlanta, and the University of Kentucky in Lexington.)

The VA database is unique not only in its size, but in its scope as well. The centers report data on 50 preoperative variables, 43 operative, and 40 postoperative variables for each patient. Virtually all major surgical procedures are assessed, including the subspecialty areas of general surgery, orthopedics, cardiac surgery, urology, neurosurgery, otolaryngology, noncardiac thoracic, and plastic surgery.

"We had been literally bombarded by questions from various providers in the private sector regarding the feasibility of using the VA models in their own private institutions," says Shukri F. Khuri, MD, FACS. Khuri is a cardiac surgeon who is chairman of the NSQIP and chief of surgery at the Boston VA Health Care System. "We feel this is quite applicable to the private sector."

The NSQIP has already set up a secure Web site to capture the data from private institutions. While the data will be compared to national VA benchmarks, the private information will remain separate, with confidentiality assured, says Khuri, who is also vice chairman of the department of surgery at Brigham and Women’s Hospital in Boston and professor of surgery at Harvard Medical School.

"The NSQIP will be purely a repository," Khuri says. "The data are owned by the institutions that enter data into it. If an institution wants to release it’s own data, that’s a different story."

Criticism prompted outcomes system

The VA’s quality improvement program grew out of a storm of criticism and a congressional mandate in the mid-1980s, when the news media published high mortality rates from cardiac surgery that were not adjusted for severity of illness.

The VA had no risk-adjustment model, and there were no national benchmarks for mortality rates or risk adjustment. They had no data to refute the negative assertions.

Congress demanded proof of improved quality, and, after some false starts, by 1990, the VA recognized the need for a comprehensive response. The National VA Surgical Risk Study focused on cardiac surgery performed at 44 medical centers and sought to validate a model of risk adjustment.

That risk-adjustment model, which has been expanded to include non-cardiac surgeries, forms the foundation of the NSQIP system, says Khuri.

"The risk-adjusted observed-to-expected ratios in mortality and morbidity were indeed reflective of the quality of care," he says. "If an institution was a high outlier — with statistically significant higher mortality than that expected — then it is very likely one would find suboptimal surgical care.

Without a valid system of risk-adjustment, comparisons among hospitals would be meaningless, notes Khuri. "If you do not risk adjust, you can err almost 60% in judging the outlier status of an institution. You have a 60% chance of mislabeling them [as high or low outliers]."

Of course, collecting quality data is also vital to such a project. The VA ensured the integrity of its data collection by requiring a full-time nurse reviewer at every institution. The nurse collects the forms that are filled out on every patient preoperatively or postoperatively and inputs the information. The nurse transmits data periodically to a central data center.

Collecting data from administrative records or even retrospectively from medical records is inadequate, says Khuri. "If you really want to do proper risk-adjustment, you need to collect data on a prospective basis and have a dedicated nurse."

In the early days of the cardiac project, different centers devoted different resources to the data collection, recalls Jeannette Spencer, RN, MS, CS, the national coordinator of NSQIP who is based at the West Roxbury (MA) VA Medical Center. When the VA began requiring a full-time nurse reviewer, "the compliance skyrocketed from 70% of data collected to 99%."

The quality of the data also improved with special training of the reviewers. "All of them had to utilize the same definitions," she says. "They were able to collect the data from each institution more accurately and more reliably."

The three private hospitals that are now joining the project are required to devote a nurse to the data collection effort.

If NSQIP eventually is expanded to include subscriber hospitals around the country, that requirement will remain, says Khuri. The expense of the dedicated employee is more than worth it, he says. "The cost for the VA for 123 centers, mostly salaries for these nurses, is no more than $5 million. If you calculate that per case captured, you’re talking about $32 per operation. This is a cost which is less than two prolene sutures."

Database leads to better outcomes

Since 1994, the VA medical centers have reduced 30-day mortality from major surgery by 9% and morbidity by 30%. While Khuri and his colleagues acknowledge that surgical advances and other factors may contribute to the improved outcomes, they see clear evidence that NSQIP has paid off.

Chiefs of surgery receive annual reports and some less detailed information quarterly. High outliers receive help in identifying quality improvement goals and processes.

The low outliers become benchmark institutions. "We have gone to the low outliers and asked them to come up with lists of processes or structures that the surgeons felt were aspects of good practices of care," says Khuri. "We felt it was our job to disseminate these as good practices."

The result, says Khuri, is a system that can monitor outcomes and provide models of success. "We think we have a system that can compare quality of care using risk-adjusted outcomes, which also provides the tools with which one can enhance and improve the quality of surgical care," he says.