|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The observable tasks do not tell the whole story of what the anesthetist is doing. As Drui and coworkers[100] suggested, there is mental activity going on even when the anesthetist appears idle. What, then, is the mental workload of administering anesthesia? Mental workload is another concept that is readily understood but difficult to define precisely. There are various ways to measure workload, none of which is ideal.
The primary task performance measure assesses the subject's performance on standard work tasks as they are made progressively more difficult by increasing the number of tasks, the task density, or the task complexity. At first, the subject is able to keep up with the increasing task load, but at some point the workload exceeds the ability to manage it, and performance on the standard tasks decreases. The disadvantage of the primary task performance measure is that, in many complex task domains, there is no accepted way to measure objectively the subject's performance on the primary work tasks except by loading to the point of catastrophic performance failure, which is easily detected. In actual high-risk domains, one cannot allow primary task performance to degrade, and one certainly cannot allow it to fail catastrophically. Although experiments of this type are in principle feasible using anesthesia simulators, they have not yet been attempted.
A more useful technique, secondary task probing, tests the subject with a minimally intrusive secondary task that is added to the primary work tasks. The secondary task is a simple one for which performance can be objectively measured, and the subject is instructed that the primary tasks of patient care take absolute precedence over the secondary task.[235] Therefore, assuming that the secondary task requires some of the same mental resources as the primary task, the performance of the subject on the secondary task is an indirect reflection of the spare capacity available to deal with it; thus it is an inverse measure of primary task workload (the greater the spare capacity, the lower the primary workload). Secondary tasks such as reaction time (with or without choice), finger tapping, and mental arithmetic have been used for this technique in the psychology laboratory, in high-fidelity simulators, and in some field studies of actual work situations.[235]
Gaba and Lee[236] presented two-digit addition problems on a computer screen placed in the anesthesia workspace at random approximately every 45 seconds. The delay in responding to problems and the number of problems skipped were logged over time and were correlated with a concurrent six-category task analysis. The dynamic ebb and flow of mental workload during cases of varying complexity was documented. For example, cardiopulmonary bypass was confirmed as a time of very low workload for the anesthetist, whereas the induction of anesthesia was confirmed as a period of high workload. Manual tasks and conversing with the attending were correlated with a delayed or absent response to the secondary task.
Subsequent studies by the UCSD/VA-Stanford group[109] [110] and the UCD group[230] [234] [237] used the reaction time to a changing display in or around the clinical monitors as a secondary task to assess mental workload and/or vigilance. For the UCSD/VA-Stanford group, the display was a red light placed next to the main physiologic monitor. This secondary task was analogous to, but totally separate from, the standard tasks of clinical work. When the secondary task was embedded in the regular work tasks, it was called an embedded task. The UCD group used a secondary task involving recognition of changes in the alphanumeric display of an unused channel of an actual clinical monitor (a parameter labeled "Vig" on the monitor changed values from "5" to "10"). This task was only partially embedded, because although it did involve an actual clinical monitor, it displayed on an otherwise unused channel and was not of clinical significance. The VA-Stanford group has been experimenting with fully embedded secondary tasks during simulator anesthetic cases, in which the values of actual clinical variables can be manipulated at will to evaluate the subject's response time to the excursion of data values beyond predefined reporting thresholds.
The mean response time to the red light used by the UCSD/VA-Stanford group was markedly less than 60 seconds for experienced subjects in both the induction and postinduction (maintenance) phases, but it was much higher for novice residents during the induction phase ( Fig. 83-8 ). The probe was not given frequently enough to track the ebb and flow of workload. The response to the UCD task typically (56%) occurred within 60 seconds, but 16% of stimuli overall (27% during the induction period) were not responded to within 5 minutes. The conclusion was that spare capacity may be limited by the workload of the case during certain periods of anesthesia care.
There are several problems with these workload studies. One is interference with the "response channel." If responding to the probe requires manual activity with a mouse or keyboard (as in the studies of Gaba and Lee and the UCD group), it cannot be performed whenever the subject is occupied with a manual task. This is especially true during a sterile procedure. Therefore it may be
Figure 83-8
A test of the anesthetist's vigilance. Shown is the
mean reaction time of novice residents and experienced anesthetists in response to
the illumination of a red light placed next to the electrocardiogram monitor display
during actual ambulatory surgery procedures. In both groups the reaction was faster
in the maintenance phase than during induction of anesthesia. Novices reacted significantly
more slowly than did experienced anesthetists. Because the distribution of reaction
times was skewed and non-gaussian, error bars are not shown, and nonparametric statistics
were used for hypothesis testing.
An additional problem with these studies is that even these simple secondary tasks were intrusive when repeated frequently. Thus, there was a tradeoff between the temporal resolution of the measurement and its intrusiveness. There is also controversy about whether these probes measure "vigilance" or "workload," although the same techniques probably measure both aspects of performance. When probes occur infrequently, are subtle, have multiple response channels, and are performed with a low level of existing workload, they are more likely to measure vigilance; when they are frequent, readily detectable, require manual response, and are performed during a high workload period, they probably are more indicative of spare capacity and workload.
Slagle and colleagues[206] carried out an assessment of an established clinical task analysis methodology regarding intrarater and interrater reliability. These investigators had one trained observer rating 20 routine anesthetics, first rating cases in the OR and then rating the same cases from a videotape, with another observer rating the same videotapes twice. A computerized task analysis program with 38 task categories was used. The results showed good intrarater reliability and also a high concordance between real-time and video analyses. That finding is important because real-time observation is unfeasible in many circumstances. This study had problems in analyzing parallel tasks because the technique of "toggling between the task categories at a rate proportional to the time spent on each task" resulted in big differences in "task duration" and "task occurrences" between the two raters. This problem of recording parallel tasks (two or more) was solved by Manser and Rall in Tuebingen.[111] [112]
A third modality of workload assessment consists of subjective measures in which individuals are asked, either in retrospect or in real time, how much load they were or are under during actual work situations. Subjective measures complement objective measurements because the subjective perceptions of the anesthetist may be an important source of stress and anxiety; conversely, an anesthetist may subjectively underestimate the workload in settings in which objective measurements demonstrate a marked reduction in spare capacity.
Various scales have been proposed to measure dimensions of mental workload that are, in theory, different.[235] However, Gaba and Lee[236] adapted a set of workload scales from those used at the National Aeronautics and Space Administration (NASA)[238] and showed that workload ratings on each of the scales were highly correlated. In subsequent studies by the UCSD/VA-Stanford group, a single dimension of overall load was assessed using an asymmetric numeric scale that minimized biases caused by a tendency to group responses at the middle and extremes of symmetric scales. This group demonstrated that a neutral observer can estimate the subjective workload of the anesthetist in real time with a high correlation to the self-rated workload of the subject.[109] Again, as could be expected, subjective workload was highest during induction and emergence from anesthesia, especially for novices.
The final set of techniques for assessing workload consists of physiologic measures. Visual or auditory evoked potentials have been used successfully to assess mental workload, but this technique can only be used in a static laboratory environment. Heart rate is a relatively easily measured variable that may be altered by mental workload. Toung and associates[239] [240] showed that the anesthetist's heart rate increases at the time of intubation, and the amount of increase is inversely related to the amount of overall medical training. Azar and associates[241] found that anesthesia faculty members' heart rate and blood pressure increased during induction of anesthesia, and one individual developed significant ST-segment depression. Bitetti and coworkers [242] confirmed that heart rate changes occurred during anesthesia, but these did not always correlate with contemporaneous self-recordings of "stress."
Because of the many factors that affect heart rate, the beat-to-beat variability of the heart rate is thought to be a better indicator of mental workload. [235] The frequency components of heart variability can be separated by spectral analysis; a component at 0.1 Hz has been linked to mental workload. Although several groups have acquired heart rate data on anesthesiologists, none has reported an analysis of the workload-related frequency components.
|
|
|
|
|
|
|
|
|
|
|
|
|
