Measurement of skin conductance responses to evaluate procedural pain in the perioperative setting

1 Introduction

Pain is an unpleasant sensation, ranging from slight discomfort to intense suffering. There is a large degree of variability between subjects in the relationship between a peripheral stimulus and the resulting pain experience [1]. The subjective nature of pain makes objective, quantitative measurements challenging. Assessment of pain using scales based on self-report is widely used. The most commonly used pain intensity scales are numeric rating scale (NRS, 0–10) and visual analogue scale (VAS, 0–100 mm), 0 representing “no pain” and 10 or 100 mm representing “worst pain imaginable” [2]. These pain assessment scales are based on the subjective evaluation of pain and require proper instruction and intact cognitive functions. The NRS scale has also been used to measure the subjective perception of anxiety [3], [4]. A number of methods to detect pain independent of patient participation have been introduced. These have been based on varying principles such as brain imaging, electroencephalography, biochemistry or autonomic responses such as blood pressure, heart rate, heart rate variability, or pupil size. Although some of these methods show promising results, none of them have gained universal acceptance.

The skin conductance algesimeter (SCA) is a device capable of continuously monitoring changes in skin conductance measured as skin conductance responses per second (SCR). SCR reflects activation of the emotional part of the sympathetic nervous system [5]. Skin sympathetic nerve activation causes the palmar and plantar sweat glands to be filled, and SCR then increases transiently before the sweat is reabsorbed and SCR decreases again. When short-lasting outgoing sympathetic nervous bursts occur, changes in SCR will follow [6], [7]. An increase in the number of SCR can therefore be interpreted as increased skin sympathetic nerve activity [6], [7]. Functional MRI of the brain during acute pain in awaked volunteers demonstrates that the skin conductance activity increases with pain-evoked brain responses, consistent with how SCR reflects pain-related autonomic processes [8]. The palmary and plantary SCR are transmitted by acetylcholine acting on muscarinic receptors, and are therefore expected not to be influenced by adrenergic receptor active agents or neuromuscular blocking drugs. SCR is not affected by normal variations in surrounding temperature [5]. Baroreceptor reflexes have little or no influence on SCR [9], [10]. Moreover, the SCR occurs very fast, within 1–2 s, to painful stimuli [5]. Studies suggest that other emotional stressors than pain, such as intellectual exertion, influence skin conductance activity [11]. Even though anxiety is not found to influence SCR [12], the SCA has not been validated just prior to major surgical procedures such as heart surgery. Previous studies in the postoperative setting have found that a cut-off value of SCR ≥0.20 distinguishes no-to-mild pain from moderate-to-severe pain [13], [14], [15], [16]. An NRS score of >3 is recognised by many health care institutions as signifying moderate to severe pain and defined as the threshold to initiate pain intervention [14]. SCR has been shown to detect pain in adults and children postoperatively [13], [14], [15], and in children and adults during procedures in intensive care [16], [17]. By monitoring SCR, pain can potentially be detected early and independent of patient cooperation. It could therefore facilitate more timely diagnosis and effective management of pain.

The current study tested for the first time if SCR can detect a nociceptive stimulus (arterial cannulation) in patients being prepared for major surgery.

2 Methods

The study was approved by the Regional Committee of Medical and Health Research Ethics, South East Norway, on May 15, 2013. ClinicalTrials.gov registration number is NCT01865344. The study complies with the Declaration of Helsinki, and all patients gave written informed consent prior to participation.

2.2 The skin conductance algesimeter – SCA

The SCA device monitors changes in skin conductance measured as SCR. The SCA consists of three components: (i) Three electrodes that are attached to the patient’s palm and are connected to (ii) a measurement unit which controls the application of a small electrical signal between the electrodes in order to evaluate skin conductance, and (iii) software for analysing the electrical signals, loaded on a computer with a monitor (Fig. 1).

Fig. 1:

Components of the skin conductance algesimeter. The skin conductance algesimeter consists of these components: measuring unit (MU), power supply for MU, electrode cable, communication cable, display PC with software, power supply for PC.

The primary output from the device is the average number of SCR. The output is a floating average over the last 15 s and is updated every 2 s. Previous studies in the postoperative setting (13–16) have found that a number of SCR of 0.20 or higher corresponds to an NRS above 3. The equipment is CE certified.

2.3 Study procedure

After enrolment in the study, when the patients arrived in the operation room they were placed on the operating table and monitored with ECG and pulse oximetry. Continuous recording with the SCA, using three electrodes placed on the palmar side of the left hand in accordance with the manufacturer’s recommendations, was started. Then, after confirming a stable signal on the SCA monitor, a 2-min period lapsed where the patients were asked to relax and not move (Fig. 2). To minimise all stimuli, the patients were not addressed in this period. After this first stabilisation and recording period, the patients were asked about anxiety and pain, using the NRS (0–10). Then the arterial cannulation was performed in the radial artery. No local anaesthetics were used. The time for start and finish of radial cannulation was marked on the SCA. Immediately after arterial cannulation the patient was asked about perceived pain and anxiety, using the NRS. On the first postoperative day during a period of no other stimuli, monitoring with the SCA was repeated together with recordings of perceived pain and anxiety, using the NRS. During all recordings the patients were blinded for the SCA readout.

Fig. 2:

Timeline of the study. The timeline illustrates the a priori defined periods for measuring of skin conductance responses with the skin conductance algesimeter, and registration of reported pain and anxiety with the numeric rating scale (NRS).

3 Results

Twenty patients were included, 80% male, age 61 [51–73.5] years [median (IQR)]. All but one patient had the arterial line placed in the right radial artery. Due to technical problems on the right arm, the arterial line of this patient was placed in the left radial artery. Arterial line cannulation lasted 43.5 [28.5–60] s for the individual patients. There were no adverse effects related to the measurements of SCR. SCR increased significantly from 0.27 [0.20–0.27] before to 0.33 [0.30–0.37] during arterial cannulation (p<0.01) (Fig. 3). Self-reported pain also increased significantly, from NRS 0 [0–0] to 5 [3], [4], [5], [6], p<0.01, during arterial cannulation. The patients reported a moderate level of anxiety that was unchanged from before to during arterial cannulation: 3.5 [2], [3], [4], [5] vs. 3.5 [1], [2], [3], [4], [5], p=0.96. After induction of anaesthesia, median SCR declined significantly to 0.00 [0.00–0.07] when compared to before and during arterial cannulation values (p<0.01 for both comparisons). On the first postoperative day, the reported NRS was 2.0 [0.5–2.0]. At the same time SCR was 0.13 [0.00–0.20]. This value, however, was significantly lower than before arterial cannulation, 0.27 [0.20–0.27], p<0.01, even though no patients reported pain before cannulation. Reported anxiety was also lower on the first postoperative day than before arterial cannulation, 0.5 [0.0–2.0] vs. 3.5 [2.0–5.0], p<0.01. When using the suggested threshold of 0.20 SCR, which is established for postoperative pain, and examining all the individual data points, the following results occurred: before arterial cannulation no patients reported NRS for pain above 3. At this time only four patients (20%) had SCR lower than 0.20. During arterial cannulation, all 20 patients had SCR equal to or above 0.20, and 11 (55%) had NRS for pain above 3. Ninety seconds after induction of anaesthesia, 16 patients (80%) had SCR lower than 0.20. On the first postoperative day, 17 patients reported NRS for pain equal to or lower than 3. Of these, 11 (65%) had SCR lower than 0.20. Of the three patients reporting NRS for pain above 3, 2 had SCR of 0.20 or higher. In total we have 60 pairs of NRS pain reports and measurements of SCR. Of the 14 reports of NRS above 3, 13 had a correlating SCR of 0.20 or higher, giving a sensitivity of 0.93 (0.68–1.0). Of the 46 reports of NRS of 3 or lower, 15 had a correlating SCR per second of or below 0.13, giving a specificity of 0.33 (0.25–0.35). During arterial cannulation, SCR was 0.20 or above in all cases. Post hoc analysis demonstrated that changing the cut off value to SCR ≥0.33 for all the data would have resulted in a sensitivity of 0.79 and a specificity of 0.78.

Fig. 3:

Median measured skin conductance responses per second (SCR) and median reported pain with the numeric rating scale (NRS) at the different time points.

4 Discussion

This study confirms the results from previous experimental and clinical studies that have consistently reported that acute pain causes an increase in skin sympathetic nerve activity mirrored by increased SCR [4], [8], [13], [14], [15], [16], [17]. During painful procedures both the median of the patients’ NRS and SCR indicated moderate or severe pain, and postoperatively both these pain assessment tools indicated no more than mild pain. The discrepancy between these two methods to assess pain occurred when assessing the patients just before they were to undergo major surgery, but without any nociceptive stimuli. At that time 80% of the patients had SCR ≥0.20, while the NRS demonstrated no pain.

When using the cut off established in the postoperative setting on all our data the SCR showed a good sensitivity (93%) to predict self-reported pain of more than 3 on the NRS. The specificity of the SCR to predict NRS >3, however, was low (33%). Because of low specificity, the SCR power to discriminate mild and no pain from moderate and severe was poor, especially before major surgery. Changing the cut off value to SCR ≥0.33 would have resulted in a marked improvement in discriminating power due to increased specificity and only moderately decreased sensitivity when all data were analysed.

It has previously been shown that the skin conductance activity may be influenced by other factors inducing emotional stress, such as intellectual exertion [11]. The increased SCR detected preoperatively, probably reflects the patients’ level of emotional stress just before going through major surgery.

However, it is an interesting finding that even though SCR was increased due to other factors before the nociceptive stimulus, there was still a significant rise in SCR during the painful procedure corresponding to the level of self-reported pain.

Due to the nature of the procedure, the nociceptive stimulus (arterial cannulation) did not have a standard, predefined length. The pain elicited by arterial cannulation may also vary in intensity during the procedure. The period of highest pain intensity might in some cases be very short. To compensate for this, we report the highest number of SCR in any 15-s period during cannulation. To eliminate bias as a result of this, the approach of reporting the maximum SCR in any 15-s period were used in all other analysing windows as well. The difference being that for the other analyses, the time window available for study was of a predetermined length of 60 s, which was often longer than the painful stimulus lasted. This method of analysing is expected to result in a somewhat higher number of SCR, than if we just used the average SCR throughout the duration of the analysing window or procedure. On average it is also expected to be higher than if we analysed a random 15-s period within the defined time windows. The advantage of this way of analysing is higher sensitivity, especially for short-acting pain. However, it also makes it more sensitive to the random fluctuations always present in a biological system. In former studies regarding SCR, the exact methods of choosing the time period to analyse have not been described. The absolute values of SCR in our study might therefore not be directly comparable to earlier studies. This might also partly explain the improvement in discriminating power observed when using a higher cut off value.

Some potential study limitations should be noted. The investigators were not blinded to the readout during registration of SCR. However, our approach, with a stringent a priori definition of which time windows to analyse, should eliminate any possible bias. We believe that this stringency is a significant strength of our study compared to earlier studies where the method of selecting the exact time window to analyse is not described. On the other hand, this approach has precluded us from manually excluding some measurements that probably are artificially high due to movement artefacts. Correcting for this might have slightly improved the specificity of the method, but at the same time this would have introduced a subjectivity that could reduce the reliability of the results. Diazepam is standard premedication at our institution, and was administered to all patients. Diazepam might have some anticholinergic activity, but the effect is likely minimal with therapeutic dosing [18]. If this had any influence on the skin conductance, it would be expected to slightly decrease SCR.

In conclusion, both NRS for pain and SCR increased from before to during nociceptive stimulation, and both indices indicated moderate pain. SCR had an overall good sensitivity to detect pain, but specificity was poor when all data where analysed. This low specificity was probably related to emotional alertness in patients waiting for major surgery. Measurement of SCR with a cut off value of SCR ≥0.20 should not be used to diagnose pain before major surgery. Whether a higher cut off value for SCR pre-operatively might result in a clinically useful sensitivity and specificity to diagnose pain remains hypothetical.