Editor’s Note: Recipients of APSF Research Grants are asked to summarize their projects in the Newsletter.
Development of a population-based epidemiologic study of morbidity and mortality related to surgery and anesthesia was carried out with the funding from an APSF Research Grant. Although there have been numerous studies of adverse perioperative outcomes, none has been performed on well-defined, well-described populations.
There is a need to study the anesthesia outcomes of patients in unique populations who undergo anesthesia care because it is difficult to conduct continuous surveillance of persons in referral practices who die or who have morbidities during the extended perioperative period.
Individuals may have been discharged to home or other facilities prior to death or the occurrence of major morbidity, records may be unavailable, or personal contact may be lost. For these reasons, the accurate incidence of these events and the perioperative period over which they occur are infrequently reported. Further, because these are rare events, it is difficult to accumulate sufficient statistics to allow analyses for common ecologies and predictive risk factors.
Factors Unique to Olmsted County
Two factors unique to Olmsted County allow these types of analyses. First, the medical care of all residents of Olmsted County, Minnesota (population 128,W) is continuously monitored and indexed. Over the last 33 years, less than 0.1% of Olmsted County residents have been lost to continuous surveillance of their ongoing medical care. Second, the large anesthetic caseload of the Mayo Clinic and Olmsted Community Hospital of more than 67,000 cases annually provides sufficient numbers of low-frequency events for evaluation of common etiologies and predictive factors.
Specifically, the grant was used to integrate the capabilities of the Mayo Clinic and Olmsted Community Hospital anesthesia databases with the Rochester Epidemiology (Surveillance) Project. Once integrated, the resulting data sets will be used to determine predictive risk factors, common etiologies, incidence, and time course of perioperative mortalities and major morbidities of persons undergoing anesthesia care.
Briefly, multiple local and national data resources have been integrated into a network system with two gigabytes of disk space and served by the Mayo main computer facilities. This integrated system can be used to cross-reference data sources, a very desirable epidemiologic capability. How it works can be shown with a hypothetical retrospective study to evaluate the incidence of pulmonary embolism after abdominal hysterectomy in females greater than 70 years of age, and any association with general or regional anesthetic techniques. For the retrospective study, all patients fulfilling the study population criteria and who received either regional or general anesthetics are identified by the demographic and surgical/anesthetic databases. This group can then be matched against a Master Billing record database to determine those patients who had either a V/Q scan and/or a pulmonary angiogram in the postoperative period. As the number of qualifying patients gets smaller with each epidemiologic sweep, more detailed computerized data can be accessed. For example, complete radiologic interpretations can be retrieved for those patients who had V/Q scans or pulmonary angiograms.
The uniformity of data and the facility and speed of its retrieval have eliminated the labor intensity barriers that have inhibited many such studies in the past. Similar integrated systems can be used prospectively for data retrieval. Patients may be randomized, then followed with minimal delay by using these same databases. Many of these databases have on-line features.
Pilot Projects
Several pilot studies were undertaken to confirm the accuracy of population selection and data retrieval by the integrated system. Clerk verifiers were used to glean selected medical records and surgical schedules for data related to demographics, intraoperative surgical and anesthetic management, and postoperative outcomes (in-hospital and out-of-hospital). These pilot studies identified the strengths and weaknesses of the computerized data. Adjustments and refinements of the systems resulted in marked improvement in data reliability. Specificities and sensitivities of most data components essential for the study of preoperative morbidity and mortality are now greater than 97%.
After the pilot projects, we performed a retrospective study using this system to determine the incidence and outcomes of intraoperative pulmonary aspirations in all patients who underwent elective or emergency general anesthetics from July 1985 to June 1991. This project was not restricted to Olmsted County patients because we wanted to test the data retrieval of the entire system.
Pulmonary aspiration, with bilious secretions or particulate matter present in the tracheobronchial tree, occurred in 62 of 215,488 of these patients (I in 3476); one died intraoperatively from exsanguination during emergency repair of a ruptured abdominal aneurysm. Of the remaining 61 patients, 40 (66%) did not develop either a symptomatic cough or wheeze, hypoxia while breathing room air, or radiographic abnormalities within two hours of pulmonary aspiration. Those 40 patients had no respiratory sequelae. The remaining 21 patients needed intensive care observation for hypoxia and/or pneumonitis. Three of these died from pulmonary insufficiency.
Based on these data, patients with clinically apparent pulmonary aspirations but who do not develop either hypoxia, radiographic abnormalities, or symptomatic, cough or wheeze within two hours after aspiration will probably not have respiratory sequelae. If -any of these findings are present, however, approximately two-thirds of these patients will need ventilatory support and their risk of developing pneumonitis or respiratory distress syndrome is approximately 50%.
The APSF grant provided a portion of the resources used to develop the software for our integrated data system. The grant funded several pilot and preliminary studies to confirm the sensitivity and specificity of the system’s electronic data retrieval. Because of the generous support of the APSF, studies which in the past would have been inhibited by the extraordinary time requirements for data retrieval may now be attempted.
Dr. Warner, Mayo Clinic, Rochester, MN was a recipient of a 1990 APSF Research Grant,