Circulation 60,475 • Volume 15, No. 4 • Winter 2000

Peri-anesthetic Data Management: Report on the Retreat of the Board of Directors of the Anesthesia Patient Safety Foundation

J.S.Gravenstein, MD

An Overview

As the opening of a significant policy initiative, the APSF Board of Directors prior to its Annual Meeting in October, 2000, held a retreat to consider thoroughly the subject of “Data Management” and its potential patient safety implications. Over the years, the Anesthesia Patient Safety Foundation has repeatedly drawn attention to the fact that anesthesia is not as safe as possible and that we lack data to document the magnitude of the problem and identify its roots. When in late 1999 the Institute of Medicine issued its dramatic report on tens of thousands of preventable deaths in American hospitals, it also called for the collection of data to pinpoint the mechanisms that cause so many patients to suffer harm or die prematurely. Clinicians, investigators, and representatives from industry were invited to present vignettes that identified the existing problems, described the hoped for ideal, delineated the current state of technology, its solutions and challenges, offered a look at the problem of outcome studies and concluded with a description of how to collect and analyze data. Dr. Robert Stoelting, President of APSF, moderated the meeting.

Dr. Jeff Cooper (Partners Healthcare, Massachusetts General Hospital, Boston), who many years ago introduced us to the concept of the critical incident, set the tone of the meeting by pointing out that all is not well in the world of anesthesia. He asked: “Is anesthesia safe enough yet?” The most optimistic statistics bandied about lack solid data and may indeed not be true. We still lack credible data and the statistic of one anesthetic death in 200,000 is probably false. Growing production pressure and anesthesia in out-of-the-way settings threaten safety. Anesthesiology and APSF have spearheaded the safety movement. Dr. Cooper believes automated perioperative data collection is needed to learn about the true details of critical incidents and disasters.

Dr. Matt Weinger (Department of Anesthesiology, University of California, San Diego), talking about utopia, imagined an accurate, available when needed, prioritized and comprehensive data system. Today we get data (the BP = 120/80 mmHg) instead of information (the BP is normal) and instead of knowledge (the clinical implications of this information isÉ). We also need population statistics obtained from a large database enriched by evidence-based guidelines. Small portable sensors and tracking devices would cater to patients and physicians. DNA readings taken from the fingertip assure the patient’s identity and guarantee confidentiality. Retinal displays would project information on the physician’s retina, simultaneously presenting the clinical scene. Mixed reality would superimpose video images of the airway during laryngoscopy and haptic displays would guide the anesthesiologist. A “flight” recorder would store information as long as necessary. All of this would be wireless and non-invasive. Continuous plasma drug and electrolyte levels will be enhanced by real-time measures of intracellular function (e.g., myocardial cellular stress-activated protein kinase levels). The future system will access many sources of information and seamlessly collect data throughout the perioperative phases.

Speakers from industry brought the audience back to earth, describing the current systems.

First, Dr. Chester Phillips from Agilent regretted that many consider the pre-anesthetic assessment as gruntwork. Outcome statistics are based in part on a comprehensive pre-anesthetic assessment. We lack standards for a pre-anesthetic evaluation. In modern anesthesia, the person providing the anesthesia may not have collected the pre-anesthetic information adding to the importance of structure and standards. Production pressure adds to the problem and delays are not tolerated even though data are missing. With standards lacking, customers request and industry offers configurability, which introduces diversity and hinders studies of large groups of patients. Dr. Phillips concluded, “We need to engineer a preanesthesia process and set some basic standards. In addition, organizations like APSF could promote the teaching of medical informatics in the anesthesia curricula.”

Ms. Lori Cross from Datex-Ohmeda spoke next on physiologic monitoring during anesthesia. We have a difficult information paradox, i.e. when we have all the time in the world, we have no critical data to record, but during moments of crisis, we have no time to collect data. Ms. Cross presented four key points: First, define the needed data precisely, recognizing that, for example, blood pressure data will be influenced by the method, site, and the context of measurement. Second, reduce artifacts at the source. All data sources are not created equal and artifact rejection can vary widely from system to system. Third, collect the right amount of data, which depends on the user and the circumstances. The granularity of the traditional anesthesia record is insufficient for an assessment of physiologic processes. An optimal monitoring system must eventually bring all these different requirements under one hat. Fourth, inspect and protect what you collect, with quality control and confidentiality being key issues. Progress will be enhanced through the establishment of a standard anesthesia data dictionary eventually resulting in a voluntary global anesthesia database for multi-center research and outcome studies.

Next, Mr. Reuben Derderian from North American DrŠger Company addressed the topic of monitoring the anesthesia machine and ventilator. Microprocessors in modern anesthesia equipment have laid the foundation for data and voice and possible video recorders helpful in the analysis of disasters. A host of data are already captured, including error codes, alarm settings, messages, and waveforms. Most manufacturers still have their own communication protocols. Uniformity will come only when the profession adopts standards and defines data-capture requirements. Expenses are not limited to the front-end data standardization and collection, but extend to the archiving, maintenance, and evaluation of the data. The most modern equipment is slow to migrate into the non-hospital-based anesthesia care sites, the very area in greatest need of excellent data capture.

Dr. Charles McLeskey representing Abbott Laboratories spoke about drug identification. Medication errors loom large in the stories of medical disasters. Of the preventable 44,000 to 98,000 annual hospital deaths, the IOM reports that 7,000 are due to medication errors. Verbal orders in the OR should be repeated back. When writing orders, write “1 mg” without a decimal point and for 1/10 mg, write 0.1 mg to forestall it being misread as 1 mg. Pre-filled syringes might reduce but not eliminate the problem. When storing drugs, separate look-alikes. Intimidation of a younger colleague by an older one can lead to impaired performance and medication errors. Handheld automated physician order entry devices might contain menus of acceptable formulary medications and default dose ranges and offer smarts that raise case specific questions. Bar-coding technology might minimize medication errors and automatically call up patient information when a drug is being ordered.

Dr. Dave Schlotterbeck from Alaris Medical Systems discussed infusion pumps and patient safety. The typical hospital utilizes from three to seven different infusion devices, each with unique user interfaces. While today’s infusion pumps accurately deliver from a few drops to a liter or more/hour, this accuracy relates to volume of fluid delivered, not accuracy and appropriateness of the drug and dose. Hospitals desire a “one size fits all” pump configuration so the device can be used anywhere at any time. The complexity of the entire infusion system contributes to the high incidence of medication errors associated with IV infusions. Standardization of the user interface, matching performance characteristics to the area of use, and connecting infusion devices to physician order entry and pharmacy information systems are among the suggestions for addressing infusion-related medication errors. In addition, new approaches to training are required to insure users are competent.

The final two speakers addressed outcome, how to measure it and how to study mishaps and disasters.

Dr. Terri Monk (Department of Anesthesiology, University of Florida) spoke about outcome measurements, which must be clearly defined: Is the question important? Does it pass the “so what” test? Is the question answerable? Is it feasible? Passive and active data collection requires consideration. In the passive case-control study the investigator waits for the problem to develop. The prospective cohort study or a randomized clinical trial is active; e.g. data are collected before the problem develops. The IOM report, “To Err is Human,” calls for mandatory and voluntary error reporting systems to be developed and instituted probably in the near future. Data collection techniques include scanable coding forms. Computer-aided input at the site eliminates errors in data entry but requires complex safeguards. A web-based data collection anticipates electronic entry of data and automatically forwards it to a coordinating center. Site training can be done online, enabling private practice physicians to participate. Confidentiality must be guaranteed, aided by encryption and restricted access.

Dr. Robert Caplan (APSF Executive Committee At-Large Member, is staff anesthesiologist at the Mason Clinic and Professor of Anesthesiology at the University of Washington, Seattle) concluded with a discussion of data analysis, drawing on his enormous experience with the closed claim studies to stress the importance of data collection and analysis. First we need to be able to analyze data for events that have a high frequency, a high clinical impact, or both. A trend analysis follows. Then we check for the “undata”, i.e. unexpected events that are unexplained and possibly of great interest. Finally we frame a hypothesis that will be meaningful. A project of this dimension requires energy and considerable financial resource from industry and government. We need a consortium involving all affected parties and a stable corps of directors, analysts and visiting scholars to do justice to a large database. In addition to cost and manpower, confidentiality and liability need to be recognized as obstacles. Over the 15 years of the Closed Claims project, liability has not been an issue. A pilot study could assess the liability risks.

These presentations were followed by a lively discussion, which supported the concept of automated record keeping. No participant knew of supposedly flawed automated records that had biased the outcome of a malpractice suit. Because we lack systems independent of human input, we have no system free of error. The pro- and con of color-coding of drugs came up for discussion. Uniformly white labels would force clinicians to read the labels. Bar-coding of drugs would help with the tracking of administration of drugs and would open the possibility of tapping into a database on the clinical pharmacology as well as the patient’s drug history. It would not eliminate human errors.

APSF was called on repeatedly to lead the way in generating a data dictionary viewed as key in all attempts to establish a large database that would be essential for the collection and interpretation of comprehensive outcome data. The great cost of this effort would have to be shared by government, industry, foundations, such as APSF, and academic volunteers.

The meeting concluded with the decision to form two task forces under the aegis of APSF. One will deal with legal impediments to the collection of clinical data and critical incidents from our system that fosters fear of law suits instead of encouraging the central collection and analysis of data, and Dr. David Gaba will chair it. The other will work on the development of a patient safety data dictionary and will be chaired by Dr. Terri Monk.

Dr. Gravenstein, Graduate Research Professor, Department of Anesthesiology, University of Florida, was a founding member of the APSF Executive Committee and remains a respected international senior sage and crusader for anesthesia patient safety.