Increasing the calculated plasma concentration of propofol has been shown to increase choice reaction time and visual and auditory response times. We studied the relationship of reaction to a vibrating handset as the effect-site target-controlled propofol concentration was incrementally increased in 20 patients during sedation, before induction of general anaesthesia. The reaction time increased, initially slowly and then more rapidly, as the calculated effect-site concentration of propofol increased, until the reaction to the vibrating handset was lost at a mean (SD) propofol effect-site concentration of 2.0 (0.6) μg.ml−1. The loss of response to verbal contact occurred at a propofol effect-site concentration of 2.4 (0.5) μg.ml−1. Reaction time may be of use clinically to warn of impending loss of verbal contact.
Patient-maintained sedation with propofol allows the patient to control a target-controlled infusion of propofol, and hence his/her level of sedation, using a hand-held demand button. We previously found this technique to be safe and effective in patients having oral surgery and it allowed more rapid titration of, and recovery from, sedation when compared with operator-administered midazolam [1–3]. In volunteer studies, however, some subjects were able to reach potentially unsafe levels of sedation (minor arterial oxygen desaturation or loss of verbal contact) when deliberately attempting to over-sedate themselves. This was despite altering the lockout time on a plasma concentration-driven patient-maintained sedation system [4, 5], or delivering propofol using a target effect-site (brain) concentration-driven system . In that study, some volunteers were able to press the drug demand button successfully despite being over-sedated, suggesting that the inability to do so cannot be relied upon to prevent over-sedation.
Previous studies have shown that choice reaction time and visual and auditory response times all increase as the calculated plasma concentration of propofol increases [7–10]. The aim of this study was to examine the effect of incrementally increasing the calculated effect-site concentration of propofol on the patient’s reaction time, particularly as the patient approaches loss of verbal contact and hence over-sedation. The primary objective of this study was to characterise the relationship between the predicted effect-site concentration of propofol, reaction time and the patient’s level of sedation with particular reference to the point of loss of verbal contact. With this knowledge, reaction time might be useful in a feedback loop with the aim of preventing loss of verbal contact and improving the safety of a patient-maintained sedation system.
Patients of ASA physical status 1 or 2 and undergoing elective surgery under general anaesthesia were recruited following local Research Ethics Committee approval and informed written consent.
A target-controlled infusion system (Graseby 3500; Smiths Medical International Ltd, Ashford, UK) based on the Marsh pharmacokinetic model with a t½ ke0 (half-life of the equilibration constant) of 2.6 min  was used to increase the target effect-site concentration of propofol incrementally until the patient lost consciousness, after which a laryngeal mask airway (LMA) was inserted and anaesthesia and surgery continued as normal. Software and hardware modifications were made to enable reaction time monitoring. A push- button handset was connected to the infusion system, which in turn was connected to a laptop computer via a serial port. The handset was programmed to vibrate at intervals of 1 min and measurement of reaction time in milliseconds was recorded as the time taken to press the push-button on the handset once it had vibrated. The measured reaction times and the corresponding calculated effect-site concentrations were logged and displayed on a laptop computer in a spreadsheet.
Ethics Committee approval was initially obtained to recruit 20 patients. We planned to record two baseline reaction times before commencing the propofol infusion. The two times were to be averaged and used as that particular patient’s baseline reaction time. However, after 10 patients, it became clear that there was some learning involved in responding to the handset vibration and that two baseline reaction times would be insufficient to obtain an accurate average individual baseline reaction time. In addition, the initial increases in target effect-site concentration of propofol were made in increments of 0.5 μg.ml−1. It became evident that these increments were too large to allow us to monitor events accurately in the important time window when loss of verbal contact occurred. Further local Research Ethics Committee approval was gained for an additional 10 patients and, thus, to recruit a total of 30 patients (10 using the original protocol and 20 with the new protocol using three baseline reaction time measurements, and smaller (0.2 μg.ml−1) incremental increases of the target effect-site concentration of propofol).
An intravenous cannula was sited and the patients were monitored throughout the study period using ECG, peripheral blood oxygen concentration, and non-invasive blood pressure measurements. These values and the respiratory rate were recorded before administration of propofol and every 5 min throughout the study and thereafter. Oxygen was administered using a Hudson mask at 4 l.min−1 until the point where verbal contact was lost, after which a standard anaesthesia facemask was used to provide oxygenation via a Bain system until the patient was ready for LMA insertion.
Before sedation, three baseline reaction times were recorded, of which the second and third were averaged and used as that patient’s baseline reaction time. Propofol sedation was commenced at a target effect-site concentration (Ce) of 1 μg.ml−1 and the handset set to vibrate at intervals of 1 min to measure reaction time. The target effect-site concentration was then manually increased in increments of 0.2 μg.ml−1 every 2 min to allow equilibration at each increment of the plasma and the effect-site concentrations, and stabilisation of the clinical effect. The patient’s sedation score (using the Observer’s Assessment of Alertness/Sedation scale (OAAS) ) was also recorded along with each reaction time measurement. The increases in the target effect-site concentration continued, even if the patient could no longer respond to the handset vibration (defined as no response or reaction time > 6 s), until the patient reached an OAAS score of 1 (unresponsive) and was ready for LMA insertion, after which the study ended and the patient’s surgery continued as planned.
This study was a pilot study and a power calculation was not performed. Descriptive data were calculated using Excel 2008 for Mac (v12.3.1 Microsoft Corporation Redmond, WA, USA). Statistical analyses were performed using SPSS (v 14.0; IBM Corporation, Armonk, NY, USA). The comparisons of reaction time baselines were performed using paired Student’s t-tests. Probit analysis was undertaken using proportions of patients at each effect-site concentration to construct dose–response curves.
In the 10 patients who were initially studied (in whom Ce propofol increments of 0.5 μg.ml−1 were used), the increase in reaction time was extremely variable. The mean (SD) reaction time at the increment 0.5 μg.ml−1 before the increment at which there was loss of response to the vibrating handset was 188 (96) % of baseline. However, three out of 10 patients had a shorter reaction time than their measured baseline time. This suggested that there might be a learning effect when reacting to handset vibration. On closer analysis of the two baseline measurements, the second measurement was an average of 81% (95% CI 65–98%, p = 0.03) of the first, confirming a learning effect.
Using the revised protocol, 20 patients scheduled for minor surgery were studied: 13 men and seven women. The mean (SD) age was 35 (9) years; weight 82 (17) kg; baseline peripheral blood oxygen saturation 99 (0.7) %; heart rate 70 (11) beats.min−1; and mean arterial pressure 108 (14) mmHg. The mean (SD) second reaction time was 536 (161) ms and the third was 539 (187) ms, with an average baseline reaction time of 538 (162) ms. The third baseline measurement was an average of 102% of the second baseline measurement (95% CI 92–112%, p = 0.91). During the study period, the mean (SD) fall in peripheral blood oxygen saturation was 0.4 (1.1) %, the mean (SD) fall in heart rate was 1 (10) beats.min−1 and the mean (SD) fall in mean arterial pressure was 21 (12) mmHg. All patients completed the study period successfully without any adverse events.
The mean (SD) effect-site concentration of propofol at which the patients stopped responding to the vibrating handset was 2.0 (0.6) μg.ml−1. The relationship between effect-site concentration and reaction time for each individual is shown in Fig. 1. As the effect-site concentration of propofol increased, sedation scores for each individual patient reduced until they lost verbal contact. The mean (SD) effect-site concentration at which the patients lost verbal contact was 2.4 (0.5) μg.ml−1. Two patients had already lost verbal contact at the point at which they lost response to the vibrating handset. Probit analysis was used to analyse the proportion of patients not responding to the vibrating handset and verbal contact at each increment of propofol. The constructed dose–response curves are shown in Fig. 2. The effect-site concentration at which half of patients (Ce50) lost response to the vibrating handset was calculated as 1.6 (95% CI 1.5–1.7) μg.ml−1, and Ce50 for loss of verbal contact was calculated as 2.3 (95% CI 2.2–2.4) μg.ml−1.
Because there was a large variation in the effect-site concentration of propofol at which each patient stopped responding to the vibrating handset, we reanalysed our results using the effect-site concentration at this point (loss of response to vibrating handset, CeLOR) as the index point and plotted the relationship between the reaction time for all patients and the effect-site concentration for the five increments of 0.2 μg.ml−1 before the loss of response to the vibrating handset (that is from 1.0 to 0.2 μg.ml−1 less than this point for each patient) as shown in Fig. 3. The OAAS score at each of these sedation increments is shown in Fig. 4.
Increasing the calculated plasma concentration of propofol increases choice reaction time, and visual and auditory response times [7–10]. The aim of this study was to investigate the relationship between an increasing calculated effect-site propofol concentration, patient reaction time and sedation score more closely, in particular to ascertain if an individual patient’s increasing reaction time during sedation could be used to monitor their sedation level.
In the initial phase of this study, we demonstrated a significant change in measured baseline reaction time between the first and second measurement. This is a well-recognised effect when studying reaction times where learning or practice improves performance , but makes averaging these two values unsuitable for use as a baseline. In the next phase of the study, we measured three baseline reaction time measurements; there was no significant change between the second and third measurements, suggesting a more stable baseline. We therefore averaged the second and third reaction times to give the baseline against which the subsequent effects of propofol were compared.
Notwithstanding the unstable reaction time baseline in the initial phase, increasing the effect-site concentration of propofol in increments of 0.5 μg.ml−1 also appeared too large to provide detailed information on the change in reaction time before the loss of response to verbal contact. Therefore, we also adjusted the incremental increases in propofol concentration to 0.2 μg.ml−1, allowing us to study this relationship in greater detail. We demonstrated that patient reaction time increases slowly with initial increases in the effect-site concentration, until a point, which is very variable between patients, where reaction time lengthens rapidly (Fig. 1). These results are strikingly similar to those demonstrated by another group looking at the effect of propofol on reaction time .
Due to this wide variability, we reanalysed the data to look at the mean change in patients’ reaction time at the five incremental increases of effect-site concentration propofol leading up to concentration at which there was loss of response to the vibrating handset. This revealed a more constant relationship with reaction time initially lengthening slowly until one increment before the loss of response to the vibrating handset, at which the mean reaction time was 172% of baseline measurement (Fig. 3). Following this point, which appeared to indicate imminent loss of response, reaction time lengthened much more rapidly until there was no response to the handset. We hypothesise that within a patient-maintained propofol sedation system, a slowing from the baseline of less than this value of 172%, to allow for a margin of safety, could be fed back into the system to prevent any further increases in the effect-site concentration of propofol. Further studies would be required to evaluate the safety and utility of this observation in practice.
After constructing dose–response curves by probit analysis, we can estimate the likely therapeutic index of propofol for sedation. The Ce50 for loss of response to the vibrating handset and the Ce50 for loss of verbal contact are separated by 0.7 μg.ml−1. This is equivalent to more than three propofol increments of 0.2 μg.ml−1. This suggests that it should be possible to separate the clinical decrement of reaction time associated with sedation by propofol with loss of verbal contact by allowing small increments of propofol and checking the reaction time between these.
We also noted that in seven out of the 20 patients, the reaction time fell in the initial stages of sedation, especially up to an effect-site concentration of 1 μg.ml−1. This has not been previously reported to our knowledge [7, 8], and one explanation is that it could represent a learning effect on measuring reaction time after the initial baseline measurement (coincidental with propofol administration). More speculatively, it could represent improved performance process associated with anxiolysis. The relationship between psychomotor testing and anxiety/arousal appears complex [14, 15], may be U-shaped, and the interpretation of this phenomenon in the context of current knowledge is far from straight forward. A patient-maintained propofol sedation system may have to account for this, perhaps by resetting the baseline reaction time if subsequent reaction time measurements were less than 100% of baseline.
The main strength of this study is the association of the known slowing of reaction time with propofol administration to the transition from sedation to general anaesthesia, namely loss of verbal contact. The main limitation is the limited number of patients studied. A power calculation was not performed, because there were no similar studies available to use. However, a similar number of patients have been included in previous studies of this type . Our results should, therefore, be considered a pilot study and the data interpreted with caution, or used to power a larger study to confirm our findings.
In summary, increasing the effect-site concentration of propofol causes a prolongation of the measured reaction time in a characteristic manner. This happens at a lower dose than that at which the patient loses verbal contact, suggesting that measuring the reaction time could be used as an individualised patient feedback method to avoid loss of verbal contact.
We thank Martyn Gray of Anaesthesia Technology Ltd for the modification to the target-controlled infusion system to enable reaction time monitoring.
No external funding and no competing interests declared.