This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Vigoda, M. M. Articles by Lubarsky, D. A. Search for Related Content PubMed PubMed Citation Articles by Vigoda, M. M. Articles by Lubarsky, D. A. Related Collections Economics and Health Care Research

---

ECONOMICS, EDUCATION, AND POLICY

Failure to Recognize Loss of Incoming Data in an Anesthesia Record-Keeping System May Have Increased Medical Liability

Department of Anesthesiology, Perioperative Medicine and Pain Management, Center for Informatics and Perioperative Management, University of Miami/Jackson Memorial Medical Center, Miami, Florida

Address correspondence and reprint requests to Michael M. Vigoda, Department of Anesthesiology, University of Miami/Jackson Memorial Hospital, 1611 NW 12th Ave. (C-301), Miami, Florida 33136. Address e-mail to Mvigoda{at}med.miami.edu.

Abstract

Automated anesthesia record-keeping systems (AARKs) are increasingly being used. There is a perception that AARKs may limit medical liability. We report a case in which our AARK may have increased our medical liability exposure. Nine months after a patient suffered a serious intraoperative complication, the anesthesiologist was named (as one of several defendants) in a claim alleging failure to properly monitor anesthetic care. One reason why the anesthesiologist was named related to a gap of 93 min in which no vital signs were documented in the anesthesia record. Relying on the physiological monitors to assess the patient's condition, the anesthesiologist did not recognize the interruption of data transmission, because the "active" medication window obscured the graphical display of the vital sign window.

Automated anesthesia record-keeping (AARK) applications free anesthesiologists from manually recording vital signs. Feldman (1) suggests that these systems are useful for managing malpractice risk. We describe a case that highlights the need for verifying the completeness of the anesthesia record when an AARK is used.

When initially launched, our AARK application (Chart+ version 6.3; Picis, Wakefield, MA) had a display that resembled the standard paper-based anesthesia record (Fig. 1). The home screen displayed O₂, N₂O, and anesthetic gas concentrations in the top half of the monitor screen, with the vital signs in the lower half.

View larger version (132K): [in this window] [in a new window]   Figure 1. Home screen of Automated Anesthesia Record-Keeping System displays graphical trend of vital signs in the lower half of the screen.

Other views can be displayed to aid the user in documentation (events or medications) or viewing data (input/output, physiological values in tabular format, medications administered etc.). Nine months after a case was completed, a claim was filed against the surgeon because of a negative outcome. No finding of malpractice was evident. The focus shifted to the anesthesiologist, primarily because no vital signs were displayed on the printed anesthesia record for more than 90 min.

Case Report

A 58-yr-old man with a brain tumor (status, post-aborted previous craniotomy and subsequent radiation therapy) was referred to our institution for treatment. In addition to standard monitoring, an arterial line and multi-orifice central line were placed. After an uneventful anesthetic induction, the patient was placed in the sitting position. A Doppler was used for detection of venous air embolism.

Three hours after the induction of anesthesia, a nurse anesthetist logged in to relieve an anesthesia resident for lunch. This returned the monitor display to the home screen.

The nurse anesthetist noticed the absence of incoming real-time data. Technical support was notified, and the problem was corrected. Unfortunately, the event was not communicated to the attending, and the missing data were not manually entered.

The case concluded with the patient breathing spontaneously but with the absence of motor function. It was felt that the patient was still emerging from the medications administered during the 7-h neurosurgical procedure. When the patient awoke, it was determined that he was quadriplegic.

The anesthetic record in the patient's chart showed many 5-min intervals for which there were no recorded real-time data. After the claim was filed, our investigation showed that 93 min of data were missing, displayed in graphical (15-min intervals) and tabular (30-min intervals) format in Figures 2 and 3.

View larger version (105K): [in this window] [in a new window]   Figure 2. Graphical display of real-time data. Note columns indicate 15-min intervals. In the printed anesthesia record, the columns represent the traditional 5-min interval. This view displays the relevant missing data in a condensed format.

View larger version (87K): [in this window] [in a new window]   Figure 3. Tabular display of real-time data. Physiological data can also be displayed in a tabular format, which can be customized by the user. In this instance, we used 30-min intervals.

This case highlights a number of issues that arose as a result of using an electronic (as opposed to a paper) anesthetic record. Over reliance on the AARK software resulted in a lack of recognition of missing data in the anesthetic record and the anesthesiologist's failure to check the accuracy of the record. Whereas this oversight was the initial focus in the plaintiff's strategy, a number of other documentation concerns (described below) surfaced upon further examination. Cumulatively, these discrepancies presented our defense team with significant challenges, which resulted in their decision to settle the case during the trial phase.

Two factors could have contributed to the patient's quadriplegia. Patient positioning is always a concern in the sitting position because excess traction on the cord may be injurious. The plaintiff asserted that there were episodes that should be interpreted as "hypotensive episodes," which could lead to cord ischemia. Had the trial continued, we were prepared to present a series of 14 uncomplicated sitting craniotomies performed by the same surgeon, which contained several episodes with arterial blood pressures lower than the lowest recorded arterial blood pressure in this case. We had planned to use our anesthesia information management system (AIMS) as a way to refute the plaintiff's claim that a particular arterial blood pressure was evidence of cord ischemia.

The plaintiff suggested that missing data were evidence that the anesthesiologist did not provide standard of care to the patient, which dictated that "every patient receiving anesthesia shall have arterial blood pressure and heart rate determined and evaluated at least every five minutes" (2). Subsequent investigation indicated that the data were not stored in the database or in the hard drive in the operating room.

Several entries (recording the adjustment of narcotic infusion rate and the results of an arterial blood gas) during this 93-minute time gap indicate that both the AARK software and the monitor display screen were working. The most likely cause of the problem was the accidental disconnection of a cable connecting the physiological monitor to the AARK. Our follow-up investigations suggested that the home screen window, which normally contains a graphical display of the vital signs, was covered by another window.

Whereas any of the windows (e.g., documentation, medications, input/output, tabular format, etc.) can be moved, closed, or resized, most users keep the medication/fluids window as the active window because most of the documentation entries in the middle of the case are typically medications or fluids. The software does not revert to the home screen (graphical display of vital signs) after a period of inactivity. In the absence of a need to access the home screen, the anesthesiologist did not recognize the interruption in data transmission.

Poor ergonomic placement of the AARK monitor may also have contributed to the lack of recognition of a problem. Our anesthesiologists monitor the patient's vital signs in the same manner that those without AARKs assess their patients—by looking at the patient and the physiological monitors. In this case, three different types of monitors were used: Datex for CO₂; Spacelab for heart rate, respiratory rate, and arterial blood pressure; and Ohmeda for anesthetic vapor concentration and O₂/N₂O.

The configuration in use during this case required placement of the AARK monitor to the right of the anesthesia machine. With the anesthesia machine placed on the anesthesiologist's right side, this required the anesthesiologist to turn his back to the patient when he was charting or reviewing the anesthetic record. Thus, it was routine for anesthesiologists to look at the physiological monitors to assess the real-time data.

When we purchased new anesthesia machines, the placement of the AARK was corrected so that there was a line of sight configuration. Whereas our anesthesiologists continue to use the physiological monitors as their main data source, this arrangement permits verification of the completeness of the record in a more user-friendly manner.

Because of this instance of data transmission failure, the vendor designed an alert (Fig. 4). This notification appears on top of all windows and alerts the anesthesiologist whenever there is an interruption of incoming data for 10 minutes or more. Manual entry of the data into the AARK system is possible because it is still stored in the physiological monitor's memory.

View larger version (104K): [in this window] [in a new window]   Figure 4. Notification of the absence of incoming data stream from physiological monitors. This alert was developed by the vendor after we highlighted the need for data-checking by the automated anesthesia record-keeping software.

The original alert appeared when no data streams were transmitted from the Spacelab monitor to the AARK. This was a compromise between the vendor and ourselves because we required a quick solution to the problem. Our preferred solution, in which an alert would appear if any data stream were interrupted, was implemented a short time afterward (Fig. 5). This improvement not only provided closer detection of data stream interruptions, but also allowed our technical support staff to remotely diagnose (and correct) the problem in the event that the user was unaware of which particular data stream was affected.

View larger version (120K): [in this window] [in a new window]   Figure 5. Modification of the alert. The vendor developed a modification of the alert several months later, which identifies the specific data stream that has been interrupted. This allows the technical support staff to remotely troubleshoot the problem.

Communication and system failures exacerbated the technical failure of the software. The passage of time between the actual event and our investigation hindered the participants' recollections of the situation. In retrospect, there were several opportunities to fix the problem. The nurse anesthetist could have retrieved the values from the physiological monitor and manually entered the data. If the nurse anesthetist told the returning resident of the problem, the resident could have entered the values. Even if not told of the problem, the resident should have noticed the missing data when he logged in again. The failure of both these individuals to notify the attending demonstrates a serious communication problem. Finally, when it was noted that the patient had suffered an untoward outcome, the anesthesia team could have reviewed the record. Because the missing vital signs would have been retrievable from the physiological monitor, these values could have been entered into the record.

There were further unexpected issues with the AARK. The plaintiff also highlighted a mismatch of the timing of the entered note for assuming the sitting position and the actual movement of the transducer. There was no note saying that the transducer was moved, and the arterial blood pressure did not change at the moment the sitting position note was entered. Although this omission became the major focus of the plaintiff's case, we were able to show that the transducer was moved by having minute-by-minute vital signs, which showed a 40-mm Hg arterial blood pressure change over 1 minute. Finally, the entry of an untimed documentation note stating that the attending was present at emergence from anesthesia was entered shortly after surgery started. This was discovered after the plaintiff hired an expert in AIMS who demanded the computerized audit trail. We believe this is the first report of a plaintiff doing so; we do not expect it will be the last.

This sequence of note entry (all at once, editing throughout the case if required) is not uncommon at several institutions with several different types of systems but must be highly discouraged. The plaintiff claimed that the anesthesiologist's pattern of documentation reflected poorly on the anesthesiologist's veracity and called into question his assertions that he actually attended the case or zeroed the transducer at the auditory meatus.

The plaintiff's attack of the timing of the attending notes was difficult to rebut. Whereas the timing of entries on a paper-based record is not identifiable by a time stamp, prospective documentation nonetheless represents a poor practice. Using AARK systems will require modification of some established work habits, particularly because these systems time stamp all entries made to the record. Upon discovering this practice, we instituted a departmental policy, which prohibited documentation of the events that had not already occurred, even if solely done to ensure that there were no missing entries. We counseled our faculty that documentation does not need to take place at the precise moment an event/procedure has occurred but it cannot occur in advance.

Most AARKs allow the user to adjust the time of a documentation entry. It is possible to make an entry in the record, go back at some later point in time, and change the time. Some users might do this so that it would seem that the documentation occurred at the later time. This is not a recommended practice because the audit trail stores two times—the time the event was entered and the time that the user wants the record to reflect when the event occurred.

Because many institutions are considering purchasing/implementing electronic medical records, we offer some advice based on our experience.

Our investigations uncovered some serious deficiencies in documentation (e.g., timing of documentation entries, missing documentation entries, omission of demographic data, and failure to complete quality assurance forms). It is instructive to note that we would never have found these errors if we did not have an AIMS. Once we discovered the scope of the problem, we developed applications to mine the data to improve our department's performance. As an example, we run automated reports in the early morning hours to review cases performed in the past 24 hours. These reports detect omissions in billing, demographic, and quality assurance data. Most reports send an email to the resident/nurse anesthetist and attending anesthesiologist. The individual is expected to complete the record once they are made aware of the deficiency. We have seen dramatic improvements in the quality of the documentation using this approach.

We are developing a process to notify users of deficiencies in their documentation during the case. Our initial approach has been to customize pop-up reminders, which appear on the monitor when documentation entries have not been completed. These reminders appear every 30 minutes after a case starts until the required documentation has been performed. The reminders disappear after remaining on the screen for a minute, so as not to be intrusive to the user. Obviously, in designing real-time notifications, we must consider how to make the application help the user and not merely harass them.

Institutions considering implementing an AARK should have a close working relationship with their hospital's Chief Information Officer because Information Services will be responsible for maintenance and improvements. There are a number of concerns regarding the data archival process that merit consideration. Is there an offsite location where the data are backed up? Will the data be stored locally in the workstation in the operating room, and for how long? What will be the resolution of time intervals (10 seconds or 1 minute) for storing the data? Is there capability to archive recent cases with finer time cuts and then compact the data if there are no complications?

For those unfamiliar with AARKs, this case demonstrates that implementing electronic medical records is not simply a matter of replacing a paper form with an electronic copy. At first glance, the issue of access to all of the electronic elements within the database is likely to become a highly charged issue from two perspectives: (a) accessibility of the data to the plaintiff and (b) discoverability of the data in court. The century-old standard for documentation of vital signs in the paper-based anesthesia record is the five-minute interval. With the adoption of electronic records (where data are stored no longer than every one minute), it seems likely that the standard will change. This may cause some anesthesiologists to resist adopting electronic records. We suggest that this is not a rational response.

Most of the time, anesthesiologists provide excellent care to their patients, and it is likely that the data will be more helpful than harmful in unequivocally describing the patient's condition on a minute-to-minute basis. Regarding those instances when a finer resolution (from five-minute to one-minute intervals) does indeed show that there was a deficiency in care, we must ask ourselves if ethically we should attempt to prevent the discovery of these data.

While we are not lawyers, our understanding is that all of the electronic entries that are stored as part of the permanent medical record are considered discoverable in court.

As more institutions adopt (or consider adopting) AARKs, we would hope that in the near future vendors incorporate data checking in their software applications. Until such time, we recommend that all users periodically review the graphical display of real-time data during the case. It is imperative that the record be reviewed before it is finalized.

The implementation of electronic medical records will not be without some unforeseen consequences, some of which may be incorrectly attributed to this transition. Thus, it is important to note that while the plaintiff's allegations concerning the missing data were problematic, the key event leading to the settlement was whether an arterial blood pressure of 90/60 mm Hg at the auditory meatus was reasonable. Our defense would have been easier if not for the cumulative impact of poor documentation practices, inadequate communication, and the absence of record review.

AARK systems should verify data capture. This could be a two-part process. Anesthesiologists should be notified about any incoming data stream interruption soon after it occurs to allow for correction during the case. At the end of the case, AARKs should authenticate that all data that should be continuously stored in the database are present. If not, there would still be the possibility of manually entering the data stored in the local hard drive or in the physiological monitor's backup storage. Besides the gap in real-time data, the plaintiff identified several problems with the documentation of care. Entries that omit relevant information or made in advance of the occurrence of specific events are difficult to defend. There is no substitute for careful reading of each record before it becomes final. The consequences of not doing so can be costly.

Footnotes

Presented, in part, at the American Society of Anesthesiologists annual meeting, October 22–26, 2005, Atlanta, Georgia.

Accepted for publication February 6, 2006.

References

1. Feldman J. Do anesthesia information systems increase malpractice exposure: results of a survey. Anesth Analg 2004;99:840–3.[Abstract/Free Full Text]
2. Standards for basic anesthetic monitoring. Approved by ASA House of Delegates. Amended on October 27, 2004.

This article has been cited by other articles:

				A. Eden, R. Pizov, L. Toderis, G. Kantor, and A. Perel The Impact of an Electronic Reminder on the Use of Alarms After Separation from Cardiopulmonary Bypass Anesth. Analg., April 1, 2009; 108(4): 1203 - 1208. [Abstract] [Full Text] [PDF]

				W. S. Sandberg Anesthesia Information Management Systems: Almost There Anesth. Analg., October 1, 2008; 107(4): 1100 - 1102. [Full Text] [PDF]

				T. R McLean Principle of robotic surgery litigation in the United States Clin Risk, September 1, 2008; 14(5): 179 - 181. [Abstract] [Full Text] [PDF]

				W. S. Sandberg, E. H. Sandberg, A. R. Seim, S. Anupama, J. M. Ehrenfeld, S. F. Spring, and J. L. Walsh Real-Time Checking of Electronic Anesthesia Records for Documentation Errors and Automatically Text Messaging Clinicians Improves Quality of Documentation Anesth. Analg., January 1, 2008; 106(1): 192 - 201. [Abstract] [Full Text] [PDF]

				W. D. Driscoll, M. A. Columbia, and R. A. Peterfreund An Observational Study of Anesthesia Record Completeness Using an Anesthesia Information Management System Anesth. Analg., June 1, 2007; 104(6): 1454 - 1461. [Abstract] [Full Text] [PDF]

				S. Kheterpal, R. Gupta, J. M. Blum, K. K. Tremper, M. O'Reilly, and P. E. Kazanjian Electronic Reminders Improve Procedure Documentation Compliance and Professional Fee Reimbursement Anesth. Analg., March 1, 2007; 104(3): 592 - 597. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Vigoda, M. M. Articles by Lubarsky, D. A. Search for Related Content PubMed PubMed Citation Articles by Vigoda, M. M. Articles by Lubarsky, D. A. Related Collections Economics and Health Care Research