Original Articles
MANUALRECORDKEEPINGIS NOT NECESSARY FOR ANESTHESIAVIGILANCE Robert G. Loeb, MID
Loeb RG. Manual record keeping is not necessary for anesthesia vigilance. J Clin Monit 1995;11:9-13 ABSTRACT.Objective. The goal of this study was to determine whether the intraoperative vigilance of anesthesia residents is different when they keep a manual record than when an assistant performs the charting. Methods. A total of 9 anesthesia residents were studied during 36 general anesthesia cases ori ASA class 1 or 2 patients. In half of the cases, the resident performed all record keeping. In the other half, the anesthesia record was kept by a human assistant. Vigilance was measured as detection rate and response time for the resident to detect a simulated abnormal value displayed on the physiologic monitor. For analysis, anesthesia cases were divided into stages of induction, maintenance, and emergence. Results. Response times and detection rates were not different when record keeping was performed by an assistant, rather than by the clinician. Shorter cases were associated with longer median response times (i.e., lower vigilance) during the maintenance phase, but only when record keeping was done manually. Conclusions. The results demonstrate that anesthesia residents are equally attentive to an experimental signal displayed on an electronic monitor while manually charting as they are when an assistant keeps the record. This brings into question the contention that eliminating the record-keeping task will result in a reduced level of vigilance. KEY WORDS.Records: anesthesia. Equipment: monitors. Anesthesiologists: performance. Vigilance: measures.
INTRODUCTION
Between 10% and 15% o f the intraoperative time o f anesthesiologists is spent on creating the anesthetic record [1-3]. A u t o m a t e d record keeping has been proposed as a w a y to reduce record-keeping time, which could be better used to observe patients and procedures [1,4]. H o w e v e r , critics have challenged that vigilance is impaired b y r e m o v i n g the anesthesiologist f r o m the data-logging process [5,6]. In this study, a previously described m e t h o d for evaluating intraoperative vigilance [7] is used to assess vigilance w h e n the anesthesiologist writes the anesthesia record and w h e n it is written b y an assistant. METHODS
From the Department of Anesthesiology, University of California, Davis Medical Center, Sacramento, CA. ReceivedJul 16, 1993, and in revised form Mar 1, 1994. Accepted for publication Mar 14, 1994. Address correspondence to Dr Loeb, Department of Anesthesiology, University of California, Davis Medical Center, Sacramento, CA 95817.
With H u m a n Subjects R e v i e w C o m m i t t e e approval,* we studied 9 residents, all o f w h o m agreed to p e r f o r m a vigilance task during administration o f anesthesia. Five o f the residents were clinical anesthesia year 1 (7 to 8 m o n t h s o f anesthesia training at the start o f the study), one was clinical anesthesia year t w o (19 months), and three were clinical anesthesia year three *UC Davis Human Subjects Review Committee, Protocols 91-676R and 92-676R, last approved August 28, 1992. Copyright © 1995 by Little, Brown and Company
9
10 Journal of Clinical Monitoring Vol 11 No 1 January 1995
(31 to 32 months). The residents were specifically instructed, both verbally and in writing, that patient care was to take precedence over the vigilance task. Data were collected over a t w o - m o n t h period during 36 general anesthetics performed in operating rooms o f our outpatient and minor surgery facilities. The experiment was a 2 × 2 × 3 factor, withinsubject design (Table 1) [8]. Vigilance was measured under two experimental conditions. In condition 1 (no scribe), the resident kept a manual anesthesia record. In condition 2 (scribe), a human assistant relieved the resident o f record-keeping tasks. T o control for the effect o f patient illness, the residents were studied under each condition during two procedures on patients of different physical status (American Society o f Anesthesiologists' [ASA] physical status 1 or 2). T o Control for the effect o f the phase of the anesthetic, each case was divided into induction, maintenance, and emergence. The induction phase began when the electrocardiograph was attached to the patient and ended 5 min after intubation. The maintenance phase then began and lasted until emergence, which started 10 min before the conclusion o f surgery and finished when the electrocardiogram was disconnected. One of two people, the investigator or a trained college student, acted as the "scribe" when condition 2 was studied. The scribe was instructed to maintain the anesthesia record as autonomously as possible. Vital signs were copied from the electronic monitors. Therapeutic maneuvers, such as administration o f drugs and alteration of machine settings, were recorded without resident intervention. The scribe also wrote down spoken input from the anesthesia resident. This approach simulates a sophisticated automated record-keeping system that does not require user training and flawlessly accepts voice input. In each case, the resident performed a previously described vigilance task [7]. The task involved monitoring an experimental numeric value displayed on our standard physiologic monitor (Puritan Bennett model 240, Wilmington, MA). This value was normally "5"; but, at random intervals (between 5 and 15 min), it would Table 1. Distribution of Vigilance Measurements within Blocks of the Experimental Design ASA 1 (n = 18 cases) No scribe Induction 14 (n = 18 cases) Maintenance 137 Emergence 8 Scribe Induction 11 (n = 18 cases) Maintenance 37 Emergence 10
ASA 2 (n = 18 cases) 10 110 11 12 56 10
change to "10." The resident was instructed to respond to the higher value by pressing a button on the table of the anesthesia machine. Response time, between the initial display o f the value "10" and the press of the button, was recorded as a measure o f vigilance performance. If the higher value was not detected within 5 min, the display reverted to "5," the response was scored as a "miss," and a response time o f 300 sec was assigned. To be included in the analyses, both the initiation of the abnormal display and the ensuing 5-min period had to be within a single phase-of-procedure period. Also, data were not included for analysis if collected during periods while the resident was on a break. The chi-square test was used to compare hit versus miss rates between groups. Differences in response times between groups were determined using the Kruskal-Wallis nonparametric test with post hoc comparisons. The post hoc comparisons were evaluated using a Bonferroni correction for multiple comparisons. Significance was assumed for all tests to be a p value less than 0.05. Response times are summarized by the median value.
RESULTS Table 1 shows the number o f stimuli analyzed in each block of the experimental design. A total o f 539 stimuli were presented; 113 were not included in the analysis because they spanned phases-of-procedure or occurred while the subject was on break. O f the 426 stimuli that were analyzed, 290 were from the "no scribe" condition and 136 were from the "scribe" condition. Table 2 compares the patients and cases in the "scribe" and "no scribe" conditions. The maintenance phase was longer for cases performed without a scribe (162 -+ 124 min) than for cases performed with a scribe (69 - 40 min). This accounts for the larger number o f stimuli analyzed in the "no scribe" condition during the maintenance phase (see Table 1). Independent of record-keeping condition, detection rates were lower during induction and emergence than during maintenance (Table 3). Similarly, the response times were longer during induction and emergence than during maintenance (see Table 3). There were no significant differences in detection rates between the scribe and no-scribe conditions, except during the emergence phase (Table 4). There were also no significant differences in response times between the two record-keeping conditions (see Table 4). H o w ever, there was an insignificant tendency toward improved vigilance (higher detection rate and shorter re-
Loeb: Record Keeping and Vigilance
Table 2. Patient and Surgical Case Characteristics
No Scribe Patient characteristics Age (mean _+ SD) Number under 18 years Number of male patients Number of female patients Phase-of-procedure duration (min) Induction (mean -+ SD) Maintenance (mean +- SD) Emergence (mean -+ SD) Surgical procedure performed Retinal Dental Nasal or sinus Middle ear Mouth or upper airway Pelvic laparoscopy Other gynecologic Arthroscopy Other orthopedic Soft tissue
Table 4. Detection Rates and Response Times by Phase-of-Procedure and Record-Keeping Condition
Scribe
32 _+ 18 4 9 9
33 "+ 20 3 11 7
16 -+ 4 162 _+ 124 17 _+ 8
15 + 4 69 + 40 18 "+5
2 1 2 1 2 3 1 2 2 2
11
0 3 3 2 3 3 0 1 0 3
sponse time) during each phase-of-anesthesia when record keeping was performed by the scribe. There was a correlation between median response time during maintenance and the duration o f the maintenance phase (Fig). However, this effect was only observed in cases where record keeping was done by the resident (no scribe condition).
Phase-of-Procedure Induction Maintenance Emergence
Condition
Median Response % Time Number Missed (sec)
No scribe Scribe No scribe Scribe No scribe Scribe
24 23 247 93 19 20
33% 18% 1% 0% 26%a 0%"
98.5 80 20 17 90 40
~Difference between "Scribe" and "No scribe" conditions, p < 0.05.
eliminating the record-keeping task will result in a reduced level o f vigilance. The results also suggest that manual record keeping impairs vigilance during short procedures, and that automated record keeping reduces this effect. The Figure demonstrates that response times were longer during the maintenance phase o f shorter procedures, but only when the resident charted manually. Anesthesia residents spend a greater percentage o f time on record keeping during maintenance than during induction [3]. Therefore, record keeping should cause greater w o r k load during the maintenance phase o f shorter cases, because there is less time to complete the task. This additional workload was only experienced by the residents who charted manually. We assessed vigilance using a simple task: intraopera-
DISCUSSION It has been speculated that automated anesthesia record keeping impairs the vigilance o f anesthesiologists by removing the clinician from the data-logging process [5,6]. To study this, we objectively measured the response time o f residents to a simulated event when they did or did not write their o w n anesthesia records. The results demonstrate that the anesthesiology residents were not less vigilant when the record-keeping task was removed. This brings into question the contention that
180 +
No Scribe
O Scribe v
120 f--
g cc 60 g
+
+
1; Table 3. Detection Rates and Response Times by Phase-of-Procedure 10
Median Response Time Phase-of-Procedure
Number
% Missed
(sec)
Induction Maintenance Emergence
47 340 39
26% a 1% 13% a
90a 19 45~
aDifferent from maintenance, p < 0.05.
100 Duration of Maintenance Phase (min)
1000
Anesthesia residents' median time to respond to a simulated signal during the maintenance phase of anesthesia versus the duration of the maintenance phase (logarithmic scale). There was a correlation between median response time and duration of the maintenance phase in cases where residents did record keeping (R = 0.59; p = 0.02); however, there was no correlation between the two when a scribe wrote the record (R = 0.10; p = 0.70).
12 Journal of Clinical Monitoring Vol 11 No I January 1995
tive monitoring o f a binary signal displayed on a physiologic monitor. Admittedly, anesthesia vigilance is more complex; it is the attention to and mental integration of numerous complex signals to maintain an awareness of the clinical situation [9]. The best way to assess anesthesia vigilance would be to measure the clinician's response time and accuracy to detect naturally occurring intraoperative events, such as hypotension, patient movement, or a disconnection o f the intravenous line. But it would be difficult to design prospective studies using such measures, because the incidence and timing of such events vary from case to case. Also, the measurements would never be accurate, because onset o f and response to an event are imprecise and ill defined. For instance, if one wished to study vigilance toward hypotension, h o w would the onset o f hypotension be defined and h o w would the clinician's recognition o f the event be identified (given that the clinician may detect the event, but choose not to treat it)? For these reasons, we measured reaction time and accuracy to detect an artificial event in the real environment. The advantages of using an artificial event are that it is defined, the timing is precise, and the occurrence is controlled. The limitations are that the artificial signal has no bearing on patient well-being, and the task is overly simple. Nonetheless, we feel that our task, and others like it [10,11], provide a way to assess the anesthesiologist's attention to routine stimuli accurately in the clinical setting, which is an important component o f anesthesia vigilance. This was not a study o f the effect o f current electronic record keepers on vigilance. Rather, we investigated whether the clinician must perform the record-keeping task to maintain vigilance. Unlike current computerbased record keepers, which require substantial data entry and effort to use, our human record keeper required no user training, automatically charted administration of drugs, and flawlessly accepted voice input from the clinician. While such a sophisticated electronic system may be available in the future, it is not achievable with present technology. The effort o f using current electronic record keepers may potentially impair vigilance more than does manual record keeping. The study had a n u m b e r o f methodological limitations. First, the subjects were not asked to record the vigilance n u m b e r on the anesthesia record. This weakens our conclusions regarding the relationship between logging data and vigilance. Second, the cases performed with a scribe tended to be shorter than those where the resident kept the record. However, this would tend to bias the manual record keeping group toward improved vigilance, since median vigilance tended to be better during longer cases (see Fig). Finally, the presence of a
m e m b e r of the experimental team (the scribe) may have motivated residents to perform better when the scribe was present. However, in a previous study, we found no difference in resident's performance on the vigilance task when they were being observed (for task analysis) versus when they were not observed [12]. T w o previous studies, which claimed to investigate the effect o f automated record keeping on vigilance, used a flawed method o f assessment [13,14]. In both studies, vigilance was assessed as the resident's ability to recall the patient's current vital signs. However, vigilance and recall are not analogous tasks because they utilize different resources for information processing [15]. Vigilance requires attention, while recall needs both attention and m e m o r y . Also, these studies did not report at what phase of the cases vigilance was assessed or the difficulty of the cases. The experimental design must control both of these factors, since they affect the anesthesiologist's attention to physiologic monitors [7]. More recently, a well-controlled study was performed using a more appropriate method to assess anesthesia vigilance [16]. Response time to detect a vigilance signal was measured in anesthesia residents during cardiac surgery cases. The results o f that study are similar to ours. There was no significant difference in response times when residents used an electronic automated anesthesia record keeper compared to when they kept a manual record, although there was a tendency toward slightly faster response times with automated record keeping. In summary, this study presents evidence that manual record keeping is not essential to maintaining vigilance during outpatient surgery cases performed under general anesthesia. It remains to be seen whether our results may be generalized to more experienced anesthesiologists, more complex cases, or regional or local anesthetics. The author gratefully acknowledges the assistance of John Langan and Peter Set with data collection, and Yasser Sowb, MA, with development of the measurement tools. REFERENCES
1. McDonald JS, Dzwonczyk R, Gupta B, Dahl M. A second time-study of the anaesthetist's intraoperative period. Br J Anaesth 1990;64:582-585 2. Kennedy pJ, Feingold A, Wiener EL, Hosek RS. Analysis of tasks and human factors in anesthesia for coronaryartery bypass. Anesth Analg 1976;55:374-377 3. Dallen L, Nguyen L, Zornow M, et al. Task analysis/ workload of anesthetists performing general anesthesia. Anesthesiology 1990;73:A498 (Abstract) 4. Hamilton WK. The automated anesthetic record is inevitable and valuable. J Clin Monit 1990;6:333-334
Loeb: Record Keeping and Vigilance 13
5. Noel TA. Computerized anesthesia records may be dangerous (Letter). Anesthesiology 1986;64:300 6. Noel KR. Controversy in automated record keeping (Letter). J Clin Monit 1991;7:280 7. Loeb RG. A measure ofintraoperative attention to monitot displays. Anesth Analg 1993;76:337-341 8. Pedhazur EJ. Multiple regression in behavioral research: Explanation and prediction, ed 2. New York: Holt, Rinehart, and Winston, 1982:334-337 9. GravensteinJS, Weinger MB. Why investigate vigilance? J Clin Monit 1986;2:145-147 10. Cooper JO, Cullen BF. Observer reliability in detecting surreptitious random occlusions of the monaural esophageal stethoscope. J Clin Monit 1990;6:271-275 11. Weinger MB, Herndon OW, Zornow MH, et al. An objective methodology for task analysis and workload assessment in anesthesia providers. Anesthesiology 1994;80: 77-92 12. Loeb RG. Monitor surveillance and vigilance of anesthesia residents. Anesthesiology 1994;80:(in press) 13. Yablok DO. Comparison of vigilance using automated versus hand written records (Abstract). Anesthesiology 1990;73:A416 14. Dzwonczyk R, AllardJ, McDonaldJS, et al. The effect of automatic record keeping on vigilance and record keeping time (Abstract). Anesth Analg 1992;74:$79 15. Sanders MS, McCormick EJ. Human factors in engineering and design, ed 6. New York: McGraw-Hill, 1987: 57-68 16. Herndon OW, Weinger MB, Zornow MH, Gaba DM. The use of automated record keeping saves time in complicated anesthetic procedures (Abstract). Anesth Analg 1993;76:$140