Mr. Chairman, Members of the Subcommittee, thank you for giving me the opportunity to make this presentation before you today:

I am Jeffrey H. Silber, M.D., Ph.D., Director of the Center for Outcomes Research at The Children’s Hospital of Philadelphia, and Associate Professor of Pediatrics and Anesthesia at The University of Pennsylvania School of Medicine and Associate Professor of Health Care Systems at The Wharton School. I am also an attending physician in pediatric oncology at The Children’s Hospital of Philadelphia.

I have been conducting medical outcomes studies using data from the Health Care Financing Administration since 1987, and have developed numerous tools for the proper adjustment of outcomes data so that meaningful comparisons across providers can be made. I have published widely in this field.

In 1992 we published a paper in the journal Medical Care, using Medicare data, which showed that hospitals with higher percentages of board certified anesthesiologists had lower rates of death in those patients with complications (otherwise known as lower “failure-to-rescue” rates). In 1995, we published a second study in the Journal of the American Statistical Association. That study found similar results using different data. These studies interested the American Board of Anesthesiology (ABA), and as a result, in July of 1995 our group, through The Children’s Hospital of Philadelphia and The University of Pennsylvania, was awarded an $88,000 grant from the ABA to explore the influence of board certification in more detail. From that ongoing study, though not directly requested by the ABA, and using methodology developed as part of other Agency for Healthcare Research and Quality supported studies, our group has recently completed a study of medical direction provided by the anesthesiologist on patient outcomes.

This afternoon I would like to share some very interesting findings from that research, briefly discuss the methodology, then talk about the significance of this work as it relates to current policy questions. Before I do, let me say a few words about the history of outcomes research. Outcomes research techniques have been used since 1968 when Lincoln Moses and Frederick Mosteller, two renowned statisticians, published a now famous report from the National Halothane Study, an observational study assessing the safety of the then new anesthetic agent Halothane. In that report, it was noted that some hospitals had very different deaths rates than other hospitals. Moses and Mosteller performed numerous statistical adjustments, many of which we still use today, and concluded that differences in adjusted mortality rates may reflect differences in quality of care.

Over the past 32 years, literally hundreds of studies have been performed using large data sets across hospitals looking at many different medical questions concerning quality of care. The study I will discuss today is one of many such studies that use large data bases with various forms of medical data to measure differences across providers or hospitals.

Outcomes research uses large numbers of observations in order to detect small effects not readily apparent at any single hospital or within any single provider group. While the data in these large outcomes studies is usually not as refined as in smaller chart review studies, the large sample size often allows us to gain insight into differences in quality of care and outcomes that would not be apparent using other methodology.

Our study showed that the lack of an anesthesiologist was associated with an increase of 2.5 excess deaths per thousand patients, and an even higher number, 6.9 deaths per thousand patients, when there were complications.

We also found that three provider level factors remained significantly associated with lower mortality rates after full model adjustment: (1) higher registered nurse-to-bed ratio; (2) larger hospital size and (3) the personal performance or medical direction by an anesthesiologist. All three factors were significant and independently related to lower mortality.

These study results are cause for concern, and raise important questions regarding the quality of care delivered to Medicare patients undergoing general surgical and orthopedic procedures who did not have an anesthesiologist personally perform or medically direct their anesthesia care. Here is how we developed the study and the methodology used.

Today, anesthesia services for surgical procedures may or may not be personally performed or medically directed by anesthesiologists. Our study compared the outcomes of surgical patients whose anesthesia care was personally performed or medically directed by an anesthesiologist (“directed cases”) with the outcomes of patients whose anesthesia care was not personally performed or medically directed by an anesthesiologist (“undirected cases”). Under HCFA billing rules, personal performance and medical direction require direct and extensive involvement of the physician in the anesthesia procedure.

Medicare claims records were analyzed for all elderly patients in Pennsylvania who underwent general surgical or orthopedic procedures between 1991-1994. The study involved 194,430 directed and 23,010 undirected cases across 245 hospitals. Outcomes studied included death rate within 30 days of hospital admission, in-hospital complication rate and the failure-to-rescue rate (defined as the rate of death after complications).

Cases were defined as being either “directed” or “undirected”, depending on the type of involvement of the anesthesiologist as determined solely by HCFA billing records. Outcome rates were adjusted to account for the severity of each patient’s medical condition and for other provider characteristics using logistic regression models. The final model included 64 patient and 42 procedure covariates plus an additional 11 hospital characteristics often associated with quality of care. Numerous alternative models were developed, using different data elements and subsets of the full data set. These are reported at great length in our soon-to-be-published paper in the journal Anesthesiology. The results from these other adjustments confirmed our main findings.

After extensive adjustments for patient and hospital characteristics, we found that lack of direction by an anesthesiologist was associated with an increase of 2.5 excess deaths per 1000 patients (1 excess death in 400 cases). This corresponded to an adjusted death rate of 3.49 % in the directed group and 3.74 % in the undirected group. We further found that lack of an anesthesiologist was associated with 6.9 excess deaths per 1000 patients with complications (1 excess death in 145 cases with complications).

After appropriate adjustments, we saw no difference in the rates of complications between the directed and undirected groups. However, as we have published in numerous articles prior to this study, complication rates found in Medicare data should not be used for assessing quality, due to imprecision in the coding of these complications. Our previous work has shown that adjusted complication rates are almost never correlated with adjusted mortality rates, and that adjusted complication rates are best thought of as a severity of illness indicator.

The methodology used for this study was standard for claims based outcomes research analyses. The techniques of adjustment used in this study are well known, commonly used methods that appear in the medical and statistics literature. Nevertheless, confirmatory studies should be conducted. Such studies ideally should involve case-control methodology to most efficiently abstract patient charts in the directed and undirected groups. If such studies were to be done as the next logical step in my research agenda, my belief is that we would observe similar results. However, such studies would provide us with greater confidence concerning this important topic.

From my perspective, there are three important policy issues raised by these results. (1) Crucial quality of care results need to be addressed regarding anesthesiologist direction of anesthesia care. (2) Our results need to be confirmed by other studies, some involving direct chart review. (3) We need to ask why there are these differences in adjusted mortality and failure to rescue across hospital and provider type, and we need to develop better systems that reduce such differences. Reducing the differences will clearly improve the quality of medical care for all Americans.

Thank you Mr. Chairman and members of the committee. I am ready to answer your questions.