Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website

IIR 05-229 – HSR&D Study

New | Current | Completed | DRA | DRE | Portfolios/Projects | Centers | Career Development Projects

IIR 05-229
Intra-Operative Predictors of Adverse Outcomes
Terri G. Monk MD BS
Durham VA Medical Center, Durham, NC
Durham, NC
Funding Period: October 2006 - September 2011

BACKGROUND/RATIONALE:
Intra-operative variables amenable to intervention by the anesthesiologist/anesthetist, such as heart rate, blood pressure, temperature, and arterial oxygen saturation, are important determinants of both short- and long-term outcomes from surgery after adjusting for important patient and procedural characteristics.


OBJECTIVE(S):
The research objectives of this retrospective cohort study are to:
1. Analyze data from disparate anesthesia information management systems (AIMS) to understand the variability in data recording;
2. Develop and apply preliminary data standards that will allow the merging of data from disparate AIMS;
3. Merge archived AIMS data on intra-operative physiology from approximately 29,000 patients from six Veterans Health Administration (VA) medical centers with the VA National Surgical Quality Improvement Program (NSQIP) data on patient-related risks, operative procedures, and outcomes; and
4. Perform multivariable analyses to assess the associations between altered intra-operative physiology and adverse perioperative outcomes.



METHODS:
Research Design: Retrospective cohort study.

The data used for this study came from two sources-the Anesthesia Information Management System (AIMS) data from the Anesthesiology Services at the six participating VA medical centers for the years 2001 to 2008, and data from the same time period and the same six VA medical centers on surgical cases entered into the VA Surgical Quality Improvement Program (VASQIP) database.

The AIMS and VASQIP databases were matched using patient Social Security number and date of surgery. Only those operations that had both the AIMS and VASQIP data available were retained. For patients with more than one operation in the database, only the first operation for each patient was retained for analysis. Non-physiologic blood pressure values were excluded. These included systolic arterial pressures (SAP) below 20 and above 300, diastolic arterial pressures (DAP) below 20 and above 200, and mean arterial pressures (MAP) below 20 and above 233.

Means and standard deviations for SAP, DAP, and MAP were calculated using all pressure values for all patients. High and low blood pressure thresholds were established by calculating the pressures above (hypertension) and below (hypotension) 1.0, 1.5, and 2.0 standard deviations from the means. The threshold groups were defined to make them mutually exclusive; i.e., patients with pressures beyond 2.0 standard deviations were excluded from the analyses of patients with pressures beyond 1.0 and 1.5 standard deviations, and patients with pressures beyond 1.5 standard deviations were excluded from the analyses of patients with pressures beyond 1.0 standard deviation. Abnormal intra-operative blood pressures for each patient were then summarized by calculating the areas under and over the thresholds (AUT and AOT in mm Hg x minutes), time duration that the blood pressure was under or over the thresholds (TUT and TOT in minutes), and average blood pressure beyond threshold when the pressures were under and over the thresholds (PUT and POT in mm Hg). Thus, there were 54 summary measures of a patient's experience with intra-operative hypertension or hypotension (AUT, AOT, TUT, TOT, PUT, POT for 1.0, 1.5, and 2.0 standard deviations for SAP, DAP, and MAP, 6 x 3 x 3 = 54) that were used in logistic regression analyses (along with the VASQIP pre-operative patient risk factors) to predict 30-day postoperative mortality.

The whole data set was split in half randomly for development and test data sets. The logistic regression models to predict 30-day postoperative mortality using the VASQIP risk factors and each of the intra-operative blood pressure variables were first run on the development data set, and then repeated on the test data set.

FINDINGS/RESULTS:
The AIMS data included 46,496 operations from 30,650 patients. The VASQIP data included 24,548 operations from 20,523 patients. Once matching occurred, there were 23,167 operations from 19,387 patients for whom we had both AIMS and VASQIP data. We then selected the first operation for each patient, leaving a total of 19,387 operations. There were no intra-operative BP data for 198 operations, leaving a final sample size for analysis of 19,189.

Mean age of the sample was 59 years, and 93% were male. Race/ethnicity was unknown for 15% of the sample, but for those with known race/ethnicity, 73% were Caucasian and 23% were African-American. About 7% of the sample were emergency operations. About 13% of the sample were ASA class 4-5, 58% class 3, and 29% class 1-2. The 30-day postoperative mortality rate was 1.7% and overall postoperative morbidity rate was 12.6%. As expected, all operative variables and postoperative outcomes were well balanced between the development and test data sets.

As expected, all ten VASQIP preoperative risk factors were highly predictive of 30-day postoperative mortality. In addition, even after including the ten VASQIP preoperative risk factors, systolic hypotension is highly significantly predictive of 30-day mortality. Further, it is not until the patient is in the 4th quartile group of AUT that systolic hypotension becomes predictive of mortality (Odds ratio = 4.0, 95% CI = 2.5-6.4). For patients in the 4th quartile, the median AUT was 51.6 mm Hg x min. (Range, 25 to 3363), the median TUT of 8.2 minutes (Range, 5 to 197), and the median PUT of 62.5 mm Hg (Range, 44 to 65). There is virtually no effect of hypertension on 30-day postoperative mortality for any of the hypertension measures (area, time, or average pressure over threshold for SAP, MAP, or DAP). Practically all 95% confidence intervals for the odds ratios include 1.0.

IMPACT:
Our initial analyses indicates that intra-operative systolic hypotension is predictive of 30-day mortality after major surgery. However, hypotension is only a significant factor when it is severe and in the range of a median systolic BP of 62.5 mm Hg for a median time of 8.2 min. The occurrence of hypertension was not related to mortality. The information from our analysis indicates that intra-operative hypotension, especially if severe, should be promptly treated. However, this retrospective analysis is hypothesis generating and prospective trials are needed to determine if the prompt treatment of intraoperative hypotension will improve outcomes. Future analyses of this database will determine if hemodynamic management impacts postoperative complications or late mortality.

PUBLICATIONS:

Conference Presentations

  1. Bronsert M, Nguyen J, Sum-Ping J, Mangione M, Kazdan D, Bentt D, Henderson W, Hammermister K, Monk TG. Challenges of summarizing physiologic data from anesthesia information systems. Abstract 11-A-4962. Paper presented at: American Society of Anesthesiologists Annual Meeting; 2011 Oct 15; Chicago, IL.
  2. Bronsert M, Hammermeister K, Henderson W, Mangione M, Nguyen J, Sum-Ping J, Bentt D, Kazdan D, Monk TG. Challenges of summarizing physiologic data from anesthesia information systems. Paper presented at: Joint Statistical Annual Meeting; 2011 Aug 2; Miami Beach, FL.


DRA: Health Systems
DRE: Prevention
Keywords: Adverse Event Monitoring, Adverse events, Informatics, Outcomes - Patient, Quality Improvement, Risk Factors, Surgery
MeSH Terms: none

Questions about the HSR&D website? Email the Web Team.

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.