Prolonged Perioperative Use of Pregabalin and Ketamine to Prevent Persistent Pain after Cardiac Surgery

Persistent postsurgical pain is common and has long-term effects on quality of life.¹  Defined as a new pain developing postoperatively in and around the incision site and persisting for at least 3 months after surgery, it is difficult to treat once established. Prevention of this phenomenon therefore seems attractive given the considerable impact on quality of life.

Up to half of all patients undergoing any type of surgery to the chest may be at risk, and over half of these cases will demonstrate features of neuropathic pain.²  Postoperative pain can persist for many years—for at least 5 yr after breast surgery, for example—with effects on quality of life.^3,4  Long-term data for pain after cardiac surgery are limited, but level 1 clinical trial data reveal a prevalence of 27 to 41% at 3 postoperative months.^5–7 

Surgical incision is believed to cause hyperalgesia and sensitize the central nervous system.¹  The gabapentinoids are effective in neuromodulating these processes during the treatment of established neuropathic pain.⁸  They have also been shown to suppress central sensitization in other centrally driven processes, such as chronic cough, leading to reduced symptoms as well as improved quality of life.⁹ 

Studies of the preventive effects of gabapentinoids have been limited in terms of duration of perioperative administration, rarely extending beyond a few days.¹⁰  Pregabalin has improved bioavailability, efficacy, and tolerability, as compared to gabapentin, which may be important when considering its prolonged and prophylactic use in pain-free surgical patients.¹¹  The concept of preventive or protective analgesia is better established with some neuromodulating analgesics, such as ketamine,¹²  but surprisingly few studies have taken the approach of combining agents, even in established neuropathic pain, although the exceptions have stood out for their efficacy.^13–15 

Not all surgical patients are destined to develop persistent pain. Predicting a patient’s susceptibility may inform the risk: benefit evaluation of any mitigating strategy or medication, especially if the latter has potential for side effects. Preoperative challenges to the nervous system with experimental pain may predict the subsequent development of pain persistence.¹⁶  Sensory testing can also be repeated after surgery to examine putative mechanisms for the transition to persistent pain states.

The aim of this study was to assess the effect of a prolonged regimen of preventive analgesia on chronic pain outcomes, as well as to determine risk factors and potential predictors for this phenomenon. We hypothesized that the use of pregabalin alone or in combination with ketamine would lower the prevalence of persistent pain after cardiac surgery, as compared to usual care.

The Heart Surgery and Persistent Postsurgical Pain (Heart PPPAIN) study was a prospective, double-blind, randomized, parallel arm, three-group, placebo-controlled trial of preventive analgesia alongside a mechanistic study for persistent pain. The objective of the trial was to test the superiority of pregabalin used alone or in combination with ketamine compared to usual care in preventing pain at 3 and 6 months after cardiac surgery. We screened patients, aged 18 to 80 yr, scheduled to undergo elective surgery via sternotomy, at the two London heart centers of St. Bartholomew’s Hospital and the London Chest Hospital.

We excluded patients who had undergone sternotomy previously, who gave a history of chronic pain, and who regularly used pain medication (other than paracetamol and nonsteroidal antiinflammatory drugs). As pregabalin is renally cleared, we used an estimated glomerular filtration rate of less than 60 ml/min to exclude patients or to withhold single postoperative doses, based on twice-daily blood testing. If this value was less than 30 or renal replacement therapy was required, we withdrew patients from the study.

All participating patients gave written, informed consent for this clinical trial, conducted in accordance with the original protocol, available on request from the authors and approved by the UK National Research Ethics Service and the Medicines and Healthcare Products Regulatory Authority. The study was registered on ClinicalTrials.gov (NCT01480765).

As set out in figure 1, 150 patients were recruited and block-randomized 1:1:1 (in groups of 30) to one of the following three treatment groups, using a computer-generated randomization sequence, created and managed in a blinded manner by the Barts Trials Pharmacy:

1. Usual care: received usual care of regular paracetamol and patient-controlled morphine analgesia and 1 preoperative day and 14 postoperative days of placebo capsules, as well as 48 postoperative hours of placebo (normal saline) intravenous infusion.

2. Pregabalin group: received usual care and 1 preoperative day and 14 postoperative days of 150-mg pregabalin capsules, alongside 48 postoperative hours of placebo (normal saline) infusion.

3. Pregabalin and ketamine combined group: received usual care and 1 preoperative day and 14 postoperative days of 150-mg pregabalin capsules, alongside 48 postoperative hours of ketamine infusion at 0.1 mg · kg⁻¹ · h⁻¹.

Allocation concealment was achieved by the use of study capsules and intravenous infusions with identical appearance for active and placebo drugs. Pregabalin study capsules were supplied by Pfizer (Surrey, UK) with no other contribution to the design, conduct, analysis, or publication of this trial. Sealed 50-ml syringes containing clear ketamine or placebo (0.9% saline) solution were prepared in a blinded manner by the clinical trials unit at St. Bartholomew’s Hospital, with no other involvement in patient care. Per-patient release of drug to the research assistants (after evaluation of eligibility, informed consent, and enrolment of participants) ensured that blindness was maintained throughout until all follow-up assessments were completed, and the data sets were locked and submitted to the trials pharmacy for release of the randomization code.

All patients completed baseline questionnaires of Quality of Life: EQ-5D (EuroQOL, The Netherlands), the Spielberger State Anxiety, and the Pain Catastrophizing Scale, as potential risk factors for the development of persistent pain.^17,18  On the day before surgery, we tested sensory responses to painful stimuli as set out below. This was done in a quiet environment with patients in a comfortable semireclined position and with both eyes closed. Four reference points were used for sensory testing of the planned incision site (5 cm from the midline of the sternum, at the level of the second and third rib bilaterally; marked as Xs in figure 1). Remote sensory testing was performed on the right forearm at the midpoint between the wrist and the elbow as a surrogate measure of central processing of pain.

This test measures sensitivity of pain pathways to increasing mechanical pressure. We used a handheld pressure algometer (fig. 2; Somedic AB, Sweden) to measure the pain pressure threshold at the same four standardized testing points on the chest, as well as the remote site. The diameter of the contact tip was 1 cm.²  A standard pressure of 30 kPa/s. We took the mean of four, random-ordered, measurement points as the pain pressure threshold.

Conditioned pain modulation is the physiologic engagement of the endogenous analgesic system to reduce pain intensity or increase threshold to pain detection.¹⁹  We calculated the conditioned pain modulation effect as the difference in algometer-derived pain pressure threshold readings, with and without the application of a conditioning remote noxious stimulus.

Ischemic arm pain was used as the conditioning stimulus. A blood pressure cuff was manually inflated to 250 mmHg to achieve an arm pain score on the numeric rating scale of 5/10. In refractory cases, after 15 min of inflation, the cuff was further inflated in 10-mmHg increments to attain this pain score. Pain pressure threshold measurements were repeated at this point to record the conditioned pain modulation effect.

We used 20 progressively stiffer monofilament von Frey fibers (Ugo Basile, Italy) to determine tactile pain detection thresholds at the sites described above. Ascending fibers were applied perpendicular to the skin for 1 s, to the point of deformation of the fiber, until the patient described pain, with the tactile pain detection threshold defined as the least force that elicits a sensation of pain on buckling of the standardized filament. This was determined by repetitive testing of ascending fiber sizes, until the same von Frey fiber elicits two similar responses in succession. All measurements were repeated at the remote site on the right arm.

Temporal summation to tactile stimulation is an indicator of a sensitized pain system.²⁰  We used the von Frey fiber one reading below the tactile pain detection threshold and stimulated at a frequency of 2 Hertz for 60 s on all test points. Pain score increases of more than 1 point were reported as positive temporal summation, as described in the literature.^21,22 

We retested pain responses at the sternotomy site, as well as remotely, on postoperative day 4 to assess for new postsurgical summation and changes in pain pressure threshold. In addition, a dynamic assessment of spreading sensitization was carried out to give an indication of zone of secondary hyperalgesia. We used the tactile pain detection threshold fiber but starting from an area free of pain—the lateral chest—and moved toward and perpendicular to the midline sternotomy. The fiber was advanced in 1-cm increments until the first sensation of pain was achieved. This distance from the midline sternotomy was measured using a disposable tape measure, and we used the sum of the four recordings as a measure of zone of secondary hyperalgesia.²⁰ 

Nursing staff administered the first study capsule (containing either 150 mg of pregabalin or placebo lactose) to all patients 2 h before surgery. Anesthesia was induced with propofol and fentanyl (restricted to a total intraoperative dose of 7.5-20 µg/kg) and maintained with isoflurane, before cardiopulmonary bypass, before converting to intravenous infusion of propofol for the remainder of the perioperative period. Remifentanil was prohibited in this study.

Cardiopulmonary bypass was established on all patients using moderate hypothermia (30° to 34°C), a membrane oxygenator, and a centrifugal pump. The intravenous infusion (of ketamine or placebo) was started at the end of cardiac surgery, once sternal closure with wires had commenced.

All patients remained sedated and ventilated for transfer to the cardiac intensive care unit after surgery, and extubation took place as per unit protocol. In addition to the trial regimen, all patients received the usual care of patient-controlled analgesia (morphine at 1 mg/ml per bolus with a lockout period of 5 min) and regular paracetamol at 1 g every 6 h for the duration of the hospital stay. During the recovery from surgery, study capsules were continued twice daily for 10 continuous days, followed by a dose reduction to 75 mg for days 11 and 12 and then 50 mg for days 13 and 14.

Chest drains were removed, as per usual care, with provision of patient-controlled morphine analgesia up to this point. Supplementary regular oral codeine was provided after drain removal, and in addition, oral tramadol was available on demand for breakthrough pain.

If patients were discharged from hospital before completion of the 14-day capsule regimen, any remaining doses were dispensed by the trials pharmacy for the patients to complete the course at home. Any unused capsules were returned to the trial pharmacy and recorded. Details of any missed doses during in-patient stay or early withdrawal from either drug were also recorded.

The primary outcome was the proportion of patients with clinically meaningful pain at 3 and 6 months after cardiac surgery. This was defined as a pain score on the numeric rating scale of greater than 3 out of a maximum score of 10, indicating moderate to severe pain intensity, after a functional assessment of three maximal coughs.²³  This was changed before the start of the study from pain scores alone, at rest and with cough, to better capture functional recovery and quality of life at 3 and 6 months.

Secondary outcomes included clinically meaningful acute pain scores at the sternotomy and saphenectomy sites (numeric rating scale of more than 3, after three maximal coughs and dorsiflexion, respectively) alongside total morphine consumption, both measured at 24 h after surgery. Recovery from surgery was assessed in terms of sedation and nausea scores at 24 h (Likert scale of 0 = none, 1 = mild, 2 = moderate, and 3 = severe), time to extubation, times to readiness for discharge from intensive care and hospital, and safety measures of respiratory rate and arterial carbon dioxide partial pressure at 24 h after surgery and any episode of inpatient diplopia, a common transient side effect of pregabalin use in naïve patients. Changes in sensory testing after surgery were also recorded as secondary outcomes, as well as potential biomarkers of drug efficacy.

During the final study assessment at 6 months after cardiac surgery, we assessed the presence of neuropathic pain (Leeds Assessment of Neuropathic Symptoms and Signs [S-LANSS] score above 12), EQ-5D–based quality of life index, and any medication use or sleep disturbance attributable to pain over the previous 7 days. In addition to baseline anxiety and catastrophizing scores, we also collected patient age, sex, weight, baseline EQ-5D index, duration of surgery, and the need to harvest the left internal mammary artery as biologically plausible risk factors for the development of persistent postsurgical pain.

This study was powered to detect a two-thirds decrease in the percentage of patients with persistent postsurgical pain. Pilot observational data from 312 consecutive patients undergoing elective sternotomy in our center over a 6-month period revealed a persistent pain prevalence of 39.7%. Based on an α of 5% and a power of 80%, we therefore calculated a sample size per group of 43 patients to allow a detectable difference of 39.7% versus 13.2% (odds ratio [OR] = 0.231). To allow for early withdrawals and loss to follow-up, we recruited 50 patients per group.

Patients were included in the final analysis of outcomes and safety on an intention-to-treat basis, with imputation of missing data on the basis of average values for the group. Sensitivity analysis for the primary outcome was also undertaken, assuming that patients lost to follow-up had pain at 3 and 6 months, which is the most conservative possible outcome. All data was entered using two-pass verification, and analysis was performed using Stata (version 14, StataCorp, USA).

We descriptively compared the baseline characteristics across the three treatment groups. All primary and secondary outcomes are reported as frequency and percentages or medians and interquartile range. In the case of the primary outcome, this is also reported as number needed to treat (inverse of the absolute risk reduction). For the primary outcome, we used exact logistic regression to estimate ORs and CI comparing each active treatment group to the usual care group. ORs were presented using the rule of four to determine the number of decimal places with CI presented with less precision.²⁴  For ordinal score data, we used an ordinal logistic regression model, with the proportional-odds assumption tested by an approximate likelihood ratio test. If the continuous secondary outcomes were not normally distributed, we used quantile regression to compare medians between the groups, estimating the difference in medians along with bootstrap CI. Because the groups were balanced in terms of covariates and because sparse data may result in bias if too many variables are included in the model, the main analysis focused on unadjusted results. For the primary outcome we also ran a multivariable model to adjust for age, sex, weight, preoperative EQ-5D index, state anxiety, pain catastrophizing, and the duration and type of surgery as a sensitivity analysis to the main result. A Firth logistic regression model was used for this to obtain bias corrected estimates.

Identifying patients at risk of developing persistent pain is particularly important given the possibility of diplopia and sedation after initiation of pregabalin in pain-free patients, and therefore the association of each predictor variable with pain was examined using Firth logistic regression with adjustment for randomization group (treatment adjusted univariate associations). To examine whether the effect of any predictor was stronger in the treatment groups we tested for interaction. We used the numeric rating scale score as a continuous dependent variable for this to increase the power, and an ordinal logistic regression model was fitted for each predictor along with treatment group and the interaction term. The area under the receiver characteristic operating characteristic curve was calculated for each variable using the control group to allow assessment of predictive ability.

A two-sided Bonferroni corrected significance level of 0.0125 was used for the primary outcome to allow for multiple comparisons (two treatment groups vs. control at two time points). Secondary outcomes were considered as supportive and exploratory and were tested using a two-sided α level of 0.05. Results from all tests are shown, allowing secondary outcome results to be interpreted in light of the number of tests made.

We screened 362 patients to recruit 150 patients for the Heart Surgery and Persistent Postsurgical Pain (Heart PPPAIN) study from January 2012 to February 2014 (fig. 3). Three- and six-month assessments were completed on 148 patients. Two patients were lost to follow-up: one due to death on day 7 after surgery and one due to emigration. The data were imputed for these two patients, assuming no pain at follow-up, because this was the most likely outcome for the treatment groups.

Additional withdrawals from treatment also took place for delayed recovery from cardiac surgery, drug intolerance, or patient choice, as set out in figure 1. All patients were approached for 3 and 6 month follow-up, and any missing data for unresponsive patients was imputed on an “average for the group” basis, allowing analysis for all 150 patients. Baseline patient characteristics were similar among the three groups (table 1).

As set out in table 2, pregabalin alone and in combination with ketamine lowered the likelihood of developing pain at 3 postoperative months from 34% (17 of 50) in the control group to 6% (3 of 50) and 2% (1 of 50), respectively (OR = 0.126 [0.022 to 0.5], number needed to treat = 3.6; and 0.041 [0.0009 to 0.28], number needed to treat = 3.1, respectively) and at 6 months from 28% (14 of 50) to 6% (3 of 50) and 0% (0 of 50; OR = 0.167 [0.029 to 0.7], number needed to treat = 4.5; and 0.0381 [0 to 0.24], number needed to treat = 3.6).

These effects remained significant after adjustment for possible confounders using a multivariable model (Supplemental Digital Content, http://links.lww.com/ALN/B934). CIs for the two active arms for the primary outcome reveal no difference in efficacy. No further comparison is therefore made between the active arms.

Sensitivity analysis assuming both patients who withdrew from the treatment arms went on to experience pain at 3 and 6 months demonstrated that the findings were robust with ORs (95% CI) for pregabalin and combined pregabalin–ketamine groups, respectively, of 0.172 (0.039 to 0.6), P = 0.003; and 0.083 (0.009 to 0.39), P = 0.0002 at 3 months; and 0.227 (0.05 to 0.8), P = 0.017; and 0.054 (0.0012 to 0.38), P = 0.0004 at 6 months.

Both pregabalin alone and the combined pregabalin–ketamine group lowered the likelihood of clinically meaningful acute pain scores at 24 postoperative h, for sternotomy (OR = 0.244 [0.09 to 0.6], number needed to treat = 3.1; and 0.208 [0.08 to 0.5], number needed to treat = 2.8, respectively) and saphenectomy (OR = 0.112 [0.024 to 0.4], number needed to treat = 2.4; and 0.283 [0.09 to 0.8], number needed to treat =3.4, respectively). At this time point, these values were associated with median decreases in morphine requirement of 29 mg (interquartile range = 8 to 50) and 33 mg (interquartile range = 12 to 54), respectively, for the pregabalin and pregabalin–ketamine combined groups compared to a median of 52 mg (interquartile range = 830) for patients receiving the usual care, with improved nausea scores in both active arms (table 3).

We found an increase in pain pressure threshold in both active arms but only when tested at a site remote to the incision (table 4). By contrast, the control group showed a decrease in the pain pressure threshold. When we tested at the site of sternotomy, there were no statistically significant postoperative changes in pain pressure threshold after surgery. New temporal summation was lower in both active arms, at the incision site as well as remotely, as was the zone of hyperalgesia, when compared to usual care.

In terms of the potential sedating effects of the pregabalin and ketamine in the active arms, there were no statistically or clinically significant differences in time to extubation or length of stay on the intensive care unit. Safety was assessed in terms of adverse events of inpatient diplopia, revealing an increased likelihood in both active arms with numbers needed to harm of 6.3 for pregabalin alone and 4.5 for the combined PK group. This was transient in all cases and resolved with omission of a single capsule dose. Median sedation scores were 2 in all groups but with statistically significant increases in sedation in both active arms, although the respiratory rate and arterial carbon dioxide tension revealed no clinically meaningful differences.

Length of stay in the hospital was significantly shorter in the group receiving pregabalin and ketamine, as compared to usual care. CI indicated no statistically significant difference between both active arms with a median difference of 0.5 days. At the final study assessment, at 6 postoperative months, patients in both active arms demonstrated differences in EQ-5D indices of quality of life, as well as in the likelihood of developing neuropathic pain, requiring analgesics or describing sleep disturbance as a result of persistent postsurgical pain (table 3).

Analysis of risk factors for pain was carried out for all 150 patients with adjustment for randomization group to allow assessment of the relationship between patient or surgical factors and long-term outcomes (independent of treatment allocation). Preoperative measures of anxiety, catastrophizing, and poor quality of life were associated with poor persistent pain outcomes (table 5), as were lower responses to conditioned pain modulation. Surgical duration and technique seems to have less impact. Postoperative reductions in pain pressure threshold remote to the sternotomy were associated with persistent pain alongside the development of new temporal summation, either at the sternotomy site or remotely. Although the zone of hyperalgesia may indicate efficacy of preventive analgesia, it is not independently associated with persistent pain in this study. No significant interactions with treatment group were found, but the study has low power to detect these effects.

Our study demonstrates the potential to predict and prevent new postoperative pain at 3 and 6 months after cardiac surgery. We found that the use of a prolonged regimen of pregabalin during the entire perioperative period protected patients from pain persistence, although the addition of ketamine in a multimodal manner failed to confer additional advantage. Decreases in pain scores seemed to translate into improvements in quality of life.

Over and above the cost of this additional medication, not all patients will tolerate centrally acting analgesic drugs, preoperatively or for a prolonged postoperative period, and side effects such as somnolence or dizziness have the potential to limit mobilization and return to functioning.^11,25  It may be best to identify the patients most at risk of developing persistent pain, and hence we evaluated patient phenotypes, namely those with poor preoperative quality of life, state anxiety, and pain catastrophizing. In addition, it may be possible to predict susceptibility by challenging patients to experimental pain before surgery with the conditioned pain modulation platform.²⁶ 

Countering the argument for using these drugs in high-risk patients only is the potential to spare the use of high-dose opioids in all patients. Perioperative opioid use may delay recovery as well as increase the risks of long-term use and dependence in the postoperative period.^27,28  In other surgical models, such as thoracic surgery, local anesthetic techniques (e.g., thoracic epidural or paravertebral block) are frequently used as part of a multimodal analgesic regimen. In contrast, cardiac surgery has traditionally relied on large doses of intra- and postoperative opioids as the mainstay of analgesic treatment.²⁹  Given that nonsteroidal antiinflammatory drugs are also avoided in this group of patients, an alternative opioid-sparing regimen such as in this trial may decrease sensitization of the central nervous system^30,31  and confer long-term outcomes benefits.

The opioid-sparing properties of pregabalin are well known,²⁵  but the potential to reduce the length of stay in hospital and provide lasting improvements in quality of life to 6 months are new findings, requiring corroboration in large multicenter trials. The latter is particularly surprising given the large improvements in EQ-5D index of more than 0.2. This suggests more than a simple decrease in pain intensity and more likely a system-wide benefit of opioid sparing, perhaps even a neurocognitive effect.^27,32,33 

There is some debate in the noncardiac surgery literature regarding the preventive effects of pregabalin where smaller doses or shorter durations of treatment are used, especially for less painful incisions, or where treatment is started late in the postoperative period once sensitization of the nervous system has begun.^12,34  Another explanation for differing outcomes from various surgical incisions is the varying tissue injury, pain intensity, and pain mechanisms underlying these different procedures. This has led to consensus agreement on the need for procedure specific studies of pain persistence.¹⁰ 

We tested pregabalin specifically for the prevention of pain after surgical incisions on the chest, an area where persistence is highly prevalent. Pesonen et al.⁵  conducted a randomized trial of low-dose pregabalin for 5 days, to assess acute pain scores and opioid requirement. They demonstrated opioid sparing as well as less confusion on intensive care after cardiac surgery. Although not powered for long-term outcomes, they did report improved pain scores on movement at 3 months in the pregabalin arm compared to usual care but not at earlier time points.

Identifying the patient at risk of developing persistent pain is particularly important given the possibility of diplopia and sedation after initiation of pregabalin in pain-free patients.^25,35  Our study suggests that preoperative assessments of quality of life, state anxiety, and pain catastrophizing, combined with conditioned pain-modulation response, could potentially predict high-risk patients. With this in mind, simplified scoring and tests suitable for routine care should be evaluated. Another putative risk factor for pain in other studies is young age,^36,37  but we failed to corroborate this, likely as a result of the small numbers of younger patients in this relatively elderly population (median age of 66 yr).

The relationship between preexisting anxiety and catastrophizing and both acute and chronic pain is established for other surgical incisions.^17,18  Further study is required to determine whether these risk factors are fixed or whether some (e.g., state anxiety) can be modified, for example, by pregabalin.

A unique opportunity exists to translate the findings of this study into clinical practice, because cardiac surgery is the only adult discipline that still routinely administers anxiolytic premedication to patients before transfer to the operating room. Currently these are benzodiazepine or opioid-based, but with pregabalin ranked first in terms of patient tolerability in a recent meta-analysis of anxiolytics,³⁸  our study could strengthen the case for such a change in clinical practice.

This study was pragmatic and relatively unrestrictive in terms of the types of cardiac surgeries included and therefore generalizable to other institutions. However, it does reflect the processes and challenges unique to surgery in central London teaching hospitals in the United Kingdom.³⁹ 

Our work has several further limitations. Identical-dose oral pregabalin was administered to all patients rather than titrating for body weight. Although the multivariable model (Supplemental Digital Content, http://links.lww.com/ALN/B934) included weight, there remains a possibility of under- or overdosing in some subjects. This may explain the sedation and diplopia in both active arms and justifies a dose-finding study with smaller doses than the 300 mg/day of pregabalin used in this trial.

Although there was a statistically significant difference in sedation between usual care and both active arms, it could be argued that is unlikely to be of clinical significance (median differences of 0.38 and 0.26). The use of a limited four-point Likert scale may have lowered the sensitivity of measurement for this important side effect. Blinding questionnaires for participant and research assistants would have determined the effectiveness of blinding to treatment allocation and is a limitation of our study design.⁴⁰ 

The powerful effect of pregabalin in preventing persistent pain made the combined pregabalin–ketamine arm of this trial redundant. A ketamine-only arm might have proved useful in determining the contribution of ketamine to this effect.¹² 

We provide evidence for the protective effect of a prolonged regimen of perioperative pregabalin on pain at 3 and 6 months after cardiac surgery. We also present data suggesting that patients with state anxiety, feature of pain catastrophizing, low preoperative quality of life, or decreased response to conditioned pain modulation may be at increased risk of persistent pain. This may warrant focused discussion during informed consent as well as the use of pregabalin, in spite of its potential for short-term side effects.

In addition, postoperative de novo signs of new temporal summation and decreased pain thresholds at a site remote to incision may be early warning signs of hyperalgesia and central nervous system sensitization. These signs could also trigger early intervention to prevent pain persistence. Further studies should screen and target these features as a potential means of risk identification and mitigation in all surgical patients.

Supported by the European Association of Cardiothoracic Anaesthesiologists (Rome, Italy).

The authors declare no competing interests.

Full protocol available at: s.anwar@qmul.ac.uk. Raw data available at: s.anwar@qmul.ac.uk.