Benefit of regional anaesthesia on postoperative pain following mastectomy: the influence of catastrophising

Abstract

Background

Previous studies suggest that truncal regional anaesthesia (TRA), including techniques such as paravertebral block, may contribute significantly to analgesia after mastectomy. However, the severity and impact of postoperative pain varies markedly amongst individuals, making the identification of patients who would benefit most from TRA a potentially important step toward personalised perioperative care.

Methods

In this prospective observational study, mastectomy patients (n=122) were recruited and systematically assessed for psychosocial characteristics including pain catastrophising before surgery, and either received preoperative TRA (n=57) or no block (n=65).

Results

Age, baseline pain, and psychosocial traits did not differ between these groups. TRA was associated with lower overall pain scores and opioid consumption perioperatively, with a larger proportion of patients without block (50% vs 28%) reporting moderate-severe pain (more than three/10) on the day of surgery. Mixed model analysis of variance revealed a significant interaction between catastrophising and TRA, such that amongst patients with high baseline catastrophising, TRA was associated with substantially lower pain severity score (58% lower), while amongst patients with low baseline catastrophising, TRA was associated with only 18% lower pain severity. At 2 weeks, this interaction between baseline catastrophising and TRA was also observed when examining surgical pain burden, with higher baseline catastrophising patients who had received TRA reporting lower pain and less frequent opioid use (40% vs 70% of patients).

Conclusions

TRA provided immediate analgesic benefit for patients undergoing mastectomy on the day of surgery, but this effect appeared more pronounced and sustained amongst patients with higher baseline catastrophising.

Clinical trial registration

NCT 02329574.

Editor's key points.

• •Optimising analgesia, whilst minimising risk, is important for perioperative analgesia. Improved understanding of factors influencing analgesic requirements is needed to individualise treatment.
• •Postmastectomy pain is known to be influenced by psychosocial characteristics such as mood disturbance and catastrophising.
• •Regional anaesthesia improved pain overall compared with no regional anaesthesia. The analgesic benefit was greater in patients with high catastrophising before surgery, and included reduced opioid consumption at 2 weeks after surgery.
• •The potential to identify individuals likely to benefit most from regional anaesthesia would have clinical utility.

Breast surgery is one of the most commonly performed surgeries worldwide. A subset of patients undergoing mastectomy experience substantial acute1, 2 and persistent3, 4 postmastectomy pain (PMP) after surgery, and the effective management of this pain has been the topic of many studies.5, 6 Truncal regional anaesthesia (TRA), including techniques such as paravertebral block (PVB), proximal intercostal block (PICB), and pectoral and serratus anterior plane blocks, are often used as part of multimodal analgesia to decrease PMP.⁶ TRA provides superior analgesia and decreased opioid use in the acute postoperative period when applied to mastectomy patients as a whole.6, 7 However, TRA is time and resource consuming, and may be associated with some discomfort, failure to produce analgesia, and result in, albeit rarely, complications including pneumothorax or intravascular injection.8, 9 On the whole, there is evidence that the benefits of regional anaesthesia outweigh these aforementioned burdens and risks, especially for patients who will experience moderate to severe postsurgical pain.

The extent and severity of PMP may be impacted by the surgical type and technique (total mastectomy vs breast conserving surgery, inclusion of axillary node dissection, or reconstruction procedures), with more extensive procedures resulting in more pain.¹⁰ Importantly, even given the same surgical procedure and anaesthetic, the experience of pain after surgery is variable,1, 11 and it is therefore often unclear which patients will benefit most from the addition of regional anaesthesia. However, multiple studies have also shown an independent association between PMP and psychosocial factors such as anxiety, depression, and catastrophising,2, 12, 13 and baseline nociceptive sensitivity (measured using quantitative sensory testing [QST]).14, 15 The ability to identify patients before operation who will experience more pain based on these phenotypic traits, and who may thus benefit most from TRA, could allow more informed and personalised guidance to a given individual's anaesthetic and analgesic plan.

Most previous studies of the analgesic efficacy of TRA have not discerned potentially important phenotypic differences between individual patients. Prospective studies that systematically measure relevant patient characteristics, surgical, medical, psychosocial, and psychophysical differences between patients at baseline are needed to allow discernment of differential effectiveness of analgesic therapies amongst individuals, and whether regional anaesthesia, including TRA, is more effective amongst patients with certain traits identifiable before operation.

In this prospective observational study, we used systematic preoperative measurement of psychosocial phenotypic variables, including pain catastrophising, in order to assess their influence on pain in the perioperative period. We aimed to identify characteristics of patients that could predict greater analgesic benefit from TRA, as measured by postoperative pain ratings and opioid utilisation. Specifically, we investigated the influence of baseline pain catastrophising as a potential modulator of the effect of TRA on reported pain severity scores at rest and with movement, and opioid administration in the perioperative period (pre-, intra-, and immediate postoperative in recovery room). We also explored the influence of catastrophising and TRA, and interactions between the two, on surgery-related pain and opioid use at 2 weeks after surgery.

Methods

Enrolment

This prospective, observational study was approved by the Partners Healthcare/Brigham and Women's Institutional Review Board (Boston, MA). Female patients age 18–85 yr undergoing total mastectomy with or without reconstruction, with or without axillary evaluation, were recruited from the Weiner Center for Preoperative Evaluation (anaesthesia preoperative clinic) between September 2014 and March 2017. After patients had consented to be in the study, a 5–10 min QST was performed, and they were sent initial baseline questionnaires (patient characteristics, medical, validated psychosocial, and validated baseline pain) via secure email link to an online, password secured, data entry system (Research Electronic Data Capture; REDCap, Vanderbilt University, Nashville, TN), with the study data collected, managed, and hosted at Partners HealthCare, Boston, MA, USA. Responses for initial and subsequent questionnaires were solicited via e-mail, with one additional reminder email request occurring for initial non-completion. Patients were included in subsequent follow-up if these questionnaires were completed before the time of surgery. This study was part of another larger observational study which included patients having any breast surgery, including those having breast conserving surgery (lumpectomy), who typically do not require, and are not routinely offered, regional anaesthesia. Thus, only patients who underwent total mastectomy were selected for this analysis.

Surgical procedures

Surgical indication in the majority of patients who enrolled was invasive breast cancer (82%), while 11% had ductal carcinoma in situ, and 7% were having prophylactic surgery. Surgical procedure was total mastectomy with (70%) or without (30%) reconstruction, performed by surgeons with specialty in breast (13) or plastic (12) surgery. The majority of those having reconstruction had tissue expander placement (79%), while the rest underwent flap reconstruction (n=14), and one had implants immediately placed. Sentinel node procedures were performed in 52% of patients, and axillary dissection in 16% of patients.

Truncal regional anaesthesia

Ultrasound-guided TRA (PVB or PICB, with or without pectoralis nerve block) was offered to all patients undergoing mastectomy or mastectomy with reconstruction, based on the availability of regional anaesthesia specialists, who performed or supervised all blocks, and other stochastic factors such as timing of the surgery, discretion of the anaesthesiologist, and variable degree of endorsement by breast and plastic surgeons. Among the five most common breast surgeons in the study, rates of patients choosing to have a block ranged from 17% to 85%, although no differences in postoperative pain were observed between surgeons.

The potential analgesic benefit and risk of TRA were explained during the routine informed consent process for anaesthesia for the surgery. Anaesthesiologists and surgeons providing care on the day of surgery were unaware of patients' psychosocial scores. All blocks were performed in the preoperative area before the surgery, with ASA monitoring, and under light procedural sedation with midazolam and fentanyl. Patients were positioned in a sitting position for two variants of TRA: PVB (n=36) or PICB (n=21) to provide multi-segmental trunk/chest wall analgesia. An ultrasound probe was placed parasagitally and used to determine the appropriate thoracic transverse process or rib levels, after which the block needle was advanced in-plane toward the paravertebral (PVB) or proximal intercostal space, just lateral from the tip of the transverse process (PICB). The choice of segmental levels of the blocks (most commonly thoracic [T] levels T2–3 and T4–5), number of injections (from one to five levels), and concentration and amount of local anaesthetic (LA) agents (ropivacaine or bupivacaine 0.25–0.5%, 7–20 ml per level) were determined by the regional anaesthesia team, with the goal of providing optimal coverage of the surgical area, while not exceeding the maximum recommended dose of LA agents. The patients with planned tissue expander placement also received an ultrasound-guided Pectoralis nerve block (Pecs1 block)⁶ at the fascial plane between the pectoralis major and minor muscles, with ropivacaine or bupivacaine 0.25–0.5%, 5–10 ml in the supine position.

Anaesthetic treatment and perioperative opioid administration

The majority of patients received general anaesthesia (119, 97.5%) with the remainder receiving sedation in addition to TRA (three, 2.5%), which was determined by the responsible anaesthesiologist, factoring in patient choice and surgeon preference. The anaesthetic technique and medication choices (inhalation or i.v. agents) were at the discretion of the intraoperative anaesthesia providers and included administration of perioperative opioid analgesics and non-opioid analgesics (celecoxib [18%], gabapentin [24%], ketamine [11.5%] and oral acetaminophen [42%]). Perioperative opioid administration refers to opioids given in pre-, intra-, and postoperative periods up until discharge from recovery, including fentanyl, hydromorphone, morphine, and oxycodone, converted to mg morphine equivalent (MME) i.v. units. Approximately 40% of patients had a primarily propofol-based anaesthetic (>800 mg).

Pain and opioid use measurement

Several instruments were used to measure pain at baseline and after surgery: 1) a simple numerical rating scale (NRS) was used to assess current pain at rest and with movement; and 2) recognising that pain severity, frequency, duration, and area are all important to the experience of pain, we also used the Breast Cancer Pain Questionnaire (BCPQ), first developed by Gartner and colleagues¹⁶ and in subsequent study,⁴ which directs questions about pain severity and frequency specifically to the surgical area, and distinguishes different anatomical areas that may be involved. A pain burden index (PBI) can be calculated from these responses by summing the pain severity scale (0–10) at each of four locations (breast, axilla, chest wall, arm) multiplied by the frequency of the pain at each site (constantly: 5 points; daily: 4 points; occasionally: 3 points; weekly: 2 points; monthly: 1 point; and never: 0 points), thus PBI=Σ (pain severity at each site x frequency). The expanded version of this questionnaire also includes questions about analgesic use, including whether patients are taking any analgesics, whether they are taking these analgesics primarily for surgery-related pain, and about the type of analgesic used (opioid and non-opioid analgesic). The prescription of PRN (‘as needed’) opioid analgesic for home use after discharge was per the standard protocol of the individual surgeon, and did not take into account whether the patient had received TRA.

Baseline patient characteristics, psychosocial factors, and psychophysical assessment

Patients reported characteristics before surgery including age, BMI, ethnicity, highest educational training, typical amount of weekly exercise, menopausal status (pre-, peri-, or post-), and pre-existing pain in the area of surgery, and chronic pain in other body areas via the electronic questionnaire.

Questionnaires assessing psychosocial factors were chosen based upon previous retrospective association with pain and opioid use, and strong psychometric validation characteristics and brevity. The Pain Catastrophizing Scale (PCS), which has been validated in pain patients and controls,¹⁷ was used to measure catastrophic thinking associated with pain. Depressive symptoms, anxiety, and sleep disturbance were assessed using short-form instruments from the National Institutes of Health roadmap initiative, such as the Patient Reported Outcome Measurement Information System (PROMIS), which have been extensively validated in studies comparing results with established scales, and have been calibrated on more than 20 000 persons.¹⁸ The Brief Symptom Index 18-Somatization Scale, also previously validated in pain patients, was used to measure somatisation, as was the Positive And Negative Affect Scale. High and low catastrophising groups were defined by the baseline PCS score, with PCS>10 as the dividing line, approximately one standard deviation (6.4) above the median score (4), comprising the top 20% of the sample.

Pain sensitivity

Pain sensitivity was assessed using brief, portable quantitative sensory tests: temporal summation of pain and painful after-sensations (PAS) were assessed as in our previous studies19, 20 using standardised weighted pinprick applicators, similar to those used by Rolke and colleagues.²¹ One of three designated force (128, 256, and 512 mN) probes was selected using the lowest probe to result in a mildly (1–3 out of 10) painful sensation with a single application. After a break of at least 10 s (typically between 10 and 30 s while further explanation of the next step occurred), a train of 10 stimuli was applied at the same location, at a rate of one stimulation per second, with patients rating pain at the fifth and 10th stimuli, and any ongoing pain (PAS) remaining 15 s after cessation of the last stimulus, on a scale of 0–10. The same procedure was repeated on the right index finger and the third finger of each hand, alternating sides of testing. The pressure pain threshold and tolerance were assessed in a similar manner to our previous studies¹⁹ using a digital pressure algometer (Wagner FDX, Greenwich, CT, USA) with a flat round transducer, probe area 0.785 cm². Testing sites were bilateral on the dorsal aspect of the proximal forearm (extremity site) and over the trapezius muscle at the upper back (truncal site), with subjects indicating when pressure first became painful (threshold), and when pressure became intolerable (tolerance). Two trials were performed at each site, alternating from side to side and between extremity and truncal sites.

Statistical analysis

Patient group characteristics were summarised using frequencies and percentages for categorical variables, and mean and standard deviation for continuous variables. Missing data because of patient non-reporting on the day of surgery, or non-response to electronic surveys, are indicated in Figure 1 and Table 1, Table 2 (n listed for each outcome and time point). Group comparisons were performed using the Student's t-test, Mann–Whitney U-test, or χ², as appropriate. In order to investigate the impact of baseline catastrophising and TRA, and their interaction, on postoperative pain (NRS or PBI), mixed factorial analysis of variance (ANOVA) was used. The proportion of patients using opioids at 2 weeks was compared using the χ² test. Enrolment numbers for the larger observational study (mastectomy and breast conserving surgery) were chosen to provide adequate power for the prediction of persistent postsurgical pain at 1 yr, given effect sizes and variance of psychosocial and psychophysical variables from previous cross sectional studies.¹⁹ A P-value <0.05 was considered significant.

Fig 1.

Patients enrolled and data included in analysis. POD, postoperative day; TRA, truncal regional anaesthesia.

Table 1.

Baseline characteristics of mastectomy patients with or without truncal regional anaesthesia. Data are presented as mean (standard deviation) or n (%). BSI, brief symptom index; PANAS, positive affect negative affect scale; PROMIS SF, patient reported outcome measurement information system short form; TRA, truncal regional anaesthesia.

Patient characteristics Mean (SD) or incidence (%)	No block (n=65)	TRA (n=57)	P-value
Age (range), yr	49.7 (23–81)	52.1 (31–75)	0.27
Race, n			0.88
African American	1	0
Caucasian	55	51
Asian	1	1
Hispanic	3	1
Mixed	3	3
Other	1	1
Menopause, n (%)			0.68
Premenopausal	30 (46)	22 (39)
Perimenopausal	11 (17)	9 (16)
Postmenopausal	24 (52)	25 (44)
BMI (kg/m2)	25.4 (4.7)	27.0 (6.4)	0.21
Surgical type, n (%)			0.25
Mastectomy	17 (26)	19 (33)
Mastectomy with reconstruction	48 (74)	38 (67)
Lymph node procedure, n (%)			0.06
No nodal procedure	19 (29)	7 (12)
Sentinel node procedure	35 (54)	41 (72)
Axillary node clearance	11 (17)	9 (16)
Diagnosis of depression	16 (25)	11 (19)	0.30
Diagnosis of anxiety	22 (34)	14 (25)	0.16
Preoperative pain burden index (PBI) in surgical area (0–120)	5.6 (10.8)	5.7 (9.0)	0.90
Consuming opioids before operation, n (%)	3 (5)	4 (7)	0.71
PROMIS SF-anxiety	18.0 (5.5)	16.8 (5.1)	0.35
PROMIS SF-depression	13.2 (5.1)	12.6 (4.7)	0.43
PROMIS SF-sleep disturbance	22.1 (7.4)	19.7 (7.7)	0.06
Pain catastrophising scale	6.2 (6.9)	6.5 (7.4)	0.94
Positive affect (PANAS)	33.5 (7.5)	35.1 (7.2)	0.38
Negative affect (PANAS)	18.1 (6.4)	17.5 (6.3)	0.37
BSI-somatisation	7.6 (2.3)	7.8 (2.6)	0.84
Pressure pain threshold (forearm)	5.1 (1.8)	5.3 (2.1)	0.59
Pressure pain tolerance (forearm)	8.1 (3.0)	8.2 (3.2)	0.84
Pressure pain threshold (trapezius)	7.1 (2.6)	8.1 (3.4)	0.11
Pressure pain tolerance (trapezius)	10.4 (3.6)	11.0 (4.1)	0.35
Temporal summation of pain	2.0 (1.8)	2.3 (1.7)	0.17
Pain after sensation	0.11 (0.29)	0.15 (0.33)	0.57

Table 2.

Pain scores and total opioid consumption in perioperative period (pre-, intra-, and recovery room) amongst patient subgroups with high (Pain Catastrophizing Scale [PCS] >10) vs low (PCS<10) baseline catastrophising, who underwent mastectomy with or without truncal regional anaesthesia. MME, mg morphine equivalent; POD, postoperative day; TRA, truncal regional anaesthesia.

Pain score	High catastrophising				Low catastrophising
	No block (n=11)	TRA (n=12)	Difference	P-value	No block (n=54)	TRA (n=45)	Difference	P-value
POD0, rest	3.8 (2.3)	1.6 (1.3)	2.2*	0.011	3.3 (2.2)	2.7 (1.8)	0.6	0.17
POD0, movement	5.4 (1.9)	2.6 (1.6)	2.8*	0.001	4.5 (2.7)	3.7 (2.2)	0.8	0.12
Perioperative opioid consumption (MME i.v.)	32.0 (12.7)	19.6 (10.1)	12.4*	0.016	29.8 (13.3)	25.7 (14.1)	4.4	0.17

*p<0.05.

Results

The preoperative baseline characteristics of patients are shown in Table 1. Although patients were not randomised to receive or not receive TRA, no differences were seen between these groups on baseline patient characteristics, surgical, psychosocial, or psychophysical variables. Similarly, baseline pain in the surgical area (PBI) was not different between groups.

Impact of truncal regional anaesthesia on pain and perioperative opioid consumption

Immediately after surgery (2 h after surgery end), mean pain scores at rest and with movement were lower in the group of patients receiving TRA (Fig. 2A). Similarly, 32/65 (49% of patients) without block reported moderate to severe pain (NRS≥4/10), compared with only 16/57 (28% of patients) who received TRA (χ², 5.7, P=0.025). Those receiving TRA also had significantly lower perioperative opioid (pre-, intra-, and postoperative in recovery room) use (24.5 [13.5] vs 30.4 [13.0] mg MME, P=0.015) (Fig. 2B). However, pain scores on postoperative day (POD)1 and POD14 were not different between these groups (Supplementary Table S1).

Fig 2.

Effect of truncal regional anaesthesia on perioperative pain and opioid administration. (a) Overall pain score at rest and with movement in patients with and without truncal regional anaesthesia on postoperative day 0 (POD0) and (b) total perioperative opioid consumption (pre-, intra-, and postoperative in the recovery room) in patients with and without truncal regional anaesthesia. TRA, truncal regional anaesthesia.

Catastrophising, truncal regional anaesthesia, and pain

In order to investigate the impact of catastrophising on pain after mastectomy, within the context of variations in anaesthetic and surgical management, we performed a mixed model ANOVA, taking into account the following factors: block (TRA, no block), state (rest, move), and catastrophising (high, low), with surgical covariates of axillary dissection (yes, no) and reconstruction (yes, no). There was a main effect of block group (F=15.5, P<0.001), and state (F=60.2, P<0.001), such that higher pain scores occurred in those not receiving block, and higher pain scores were reported with movement, but there was no main effect of catastrophising (F=0.253, P=0.616). However, there was a significant interaction of catastrophising and block (F=5.13, P=0.026). Specifically, amongst high catastrophising patients (PCS>10), those who received TRA reported significantly less pain (58% lower pain scores, NRS: 1.6 [1.3]) than those not receiving TRA (NRS: 3.8 [2.3]) (Fig. 3a and Table 2). In contrast, amongst low catastrophising patients, those who received TRA did not have significantly less pain (18% lower pain scores, NRS: 2.7 [1.8]) than those who did not receive TRA (3.3 [2.2]) (Fig. 3a and Table 2).

Fig 3.

Effect of truncal regional anaesthesia on perioperative pain and opioid administration in patients with high and low baseline catastrophising. (a) Pain on postoperative day 0 (POD0) and (b) total perioperative opioid consumption in patients who reported high (pain catastrophising scale [PCS] >10) and low (PCS≤10) catastrophising at baseline, with and without truncal regional anaesthesia. catas, catastrophising; TRA, truncal regional anaesthesia.

Catastrophising, truncal regional anaesthesia, and perioperative opioid consumption

We also performed a mixed factorial ANOVA to determine the impact of block and catastrophising on perioperative opioid consumption (total amount administered in pre-, intra-, and postoperative in the recovery room), with the following factors: block (TRA, no block), catastrophising (high, low PCS), with surgical covariates of axillary dissection (yes, no) and reconstruction (yes, no). We again observed a main effect for block group (F=5.27, P=0.024), but no main effect for catastrophising (F=1.2, P=0.276), and the interaction for catastrophising with block was not significant (F=0.73 P=0.393). However, amongst those with higher baseline catastrophising, significantly greater opioid consumption was seen in the no block group, while amongst those with lower baseline catastrophising, this was not the case (Fig. 3b and Table 2).

Impact of catastrophising and TRA on pain and opioid use at 2 weeks

We subsequently investigated whether there was any difference in reported pain or opioid use at 2 weeks after surgery between patients who had or had not received TRA. Patients answered questions about severity in each of the surgically related body areas and frequency of pain over the 2 weeks after surgery on the BCPQ, allowing calculation of a PBI¹⁶ during this time. We subsequently performed a mixed factorial ANOVA using 2 week PBI as outcome and with the same factors: block (TRA, no block), catastrophising (high, low), with surgical covariates of axillary dissection (yes, no) and reconstruction (yes, no). This did not show a significant main effect of TRA (F=1.16, P=0.283), similar to results seen using the simple NRS (Supplementary Table S1), and also did not show a significant main effect of catastrophising (F=1.14, P=0.288). However, we again observed a significant interaction between TRA and catastrophising (F=4.4, P=0.038), such that those with higher baseline catastrophising who received TRA reported lower PBI at 2 weeks than their counterparts who did not receive TRA (Fig. 4a). Notably, there was also a main effect of axillary dissection (F=5.9, P=0.017) on PBI at 2 weeks, such that those who had axillary dissection had a higher PBI.

Fig 4.

Effect of truncal regional anaesthesia on pain burden and PRN opioid use at 2 weeks in patients with high and low baseline catastrophising. (a) Pain burden index scores and (b) opioid use at 2 weeks amongst high and low catastrophisers who did or did not receive truncal regional anaesthesia. catas, catastrophising; TRA, truncal regional anaesthesia.

Regarding opioid use at 2 weeks, patients answered questions about whether they were taking any pain medications for surgery-related pain at 2 weeks, and if so, whether these included opioids. There is no difference in overall incidence of opioid usage at 2 weeks between TRA and the no block group (38% vs 37%, P=0.52). However, amongst those who did not get TRA, high catastrophisers (seven/10, 70%) were more likely than low catastrophisers (14/47, 30%) to be taking opioids at 2 weeks (χ² 5.7, P=0.028). In contrast, amongst those who did get TRA, high catastrophisers (4/10, 40%) were not more likely than low catastrophisers (14/37, 37%) to be taking opioids at 2 weeks (χ² 0.16, P=0.589) (Fig. 4b).

Specificity of effects

We examined correlations between catastrophising and measures of anxiety, depression, and sleep disruption. PCS scores were correlated with PROMIS anxiety (Spearman R=0.34, P<0.001), PROMIS depression (Spearman R=0.45, P<0.001), and PROMIS sleep disturbance at (Spearman R=0.19, P=0.038). These correlations suggest modest inter-relationships, and are consistent with prior work situating catastrophising in a biopsychosocial context. However, ANOVA testing for interactions between these other variables and TRA on pain and opioid outcomes found no significant effects, highlighting the specificity of this interaction to catastrophising.

Discussion

Truncal regional anaesthesia before mastectomy provided an overall analgesic benefit for patients in the form of reduced pain scores and analgesic consumption on the day of surgery. However, patients with higher baseline catastrophising showed a more pronounced benefit from TRA compared with patients with lower baseline catastrophising. Moreover, this interaction between catastrophising and TRA extended beyond the day of surgery, demonstrated by the lower pain burden and lower incidence of opioid use at 2 weeks amongst higher catastrophising patients who had received TRA.

Truncal regional anaesthetic techniques such as PVB, PICB, and pectoral block have been increasingly applied for anaesthetic and analgesic purposes in breast surgery.⁶ Similar to prior studies, we observed lower rest and movement pain scores right after the surgery, and lower perioperative opioid consumption, in mastectomy patients who received TRA. Importantly, half of the patients (49%) who did not receive TRA reported moderate to severe pain (NRS≥4/10) at rest at 2 h after surgery, whereas the other half of the patients (51%) who did not receive TRA experienced only mild or no pain. Given this fifty-fifty chance of no/mild pain, together with a potential for higher cost, procedural pain, and rare potential complications of TRA,8, 9 identifying patients who will benefit most from TRA carries greater significance.

Multiple studies have shown an important association between PMP and higher pre- and post-treatment levels of anxiety, depression, and catastrophising.2, 12, 13 Catastrophising is defined as an exaggerated negative mindset including rumination, amplification, and helplessness,²² which is brought to bear during or after an anticipated painful experience, and it has proved to be one of the most robust psychosocial predictors of pain-related outcomes after surgery.²³ Many studies have shown an important link between higher catastrophising and pain severity and impact in both acute²⁴ and chronic pain settings,20, 22, 25 including acute postoperative pain²⁴ and chronic postsurgical pain. However, the mechanism of effect of catastrophising on pain and disability is still unclear. One theory is that catastrophising augments negative affect including depressive symptoms.²² Another is that catastrophising increases attention to pain symptoms, effectively augmenting the influence of sensory pain input to the brain.²² Catastrophising may also prime or preactivate pain-related areas in the brain even before a painful stimulus,²⁶ or may impede activation of endogenous pain-inhibitory mechanisms.²⁷ Of note, we observed lower median PCS scores in this patient sample, compared with a previous retrospective cross section cohort.¹⁹

The degree of difference in pain score and opioid consumption amongst high catastrophisers was both statistically and clinically significant (difference in pain score of 2–3 points on a 10-point scale and difference of 12.35 mg of i.v. MME between TRA/no TRA), despite this subgroup consisting of a relatively small number of patients. A much smaller difference in the same direction was observed amongst low catastrophisers (<1 point on a 10-point scale, and only 4.39 mg of i.v. MME between TRA/no TRA). A possible mechanism underlying this greater effect of TRA in higher catastrophisers could include a higher overall pain susceptibility. TRA, by decreasing transmission to the CNS, has a relatively greater impact, and therefore the difference is easier to observe, even in a smaller group of patients (i.e. enrichment). An alternative hypothesis is that high catastrophisers gain more from the increased attention of additional care providers (regional anaesthesia team) who come to rally around them in their most anxious time (preoperative area, immediately before being taken into surgery) providing both psychological and pharmacological anxiolysis, and effectively decreasing concern about pain in this crucial moment. The idea that catastrophising usefully engages care providers in the perioperative environment is plausible, but not well studied. Although catastrophising is typically viewed as a dysfunctional process because it is associated with greater reported pain severity, it may also importantly elicit such support or empathy from others in the social environment.²⁸

Notably, this more pronounced analgesic effect of TRA amongst higher catastrophisers appeared to extend beyond the day of surgery, as these patients reported lower PBI and use of opioids at POD14. Previous studies have shown that intense and prolonged nociceptive activation by surgical injury can result in central sensitisation, hyperalgesia, and pain persistence, which can be prevented by aggressive and early pain relief.²⁹ Regional anaesthesia, which blocks or decreases transmission of nociceptive impulses from the injury site to the CNS, may prevent central sensitisation,⁷ potentially explaining an extension of the benefit beyond the pharmacologic duration of the LAs used. Although several studies support this hypothesis,5, 7 there is also at least one study failing to demonstrate the long-term effect of single shot TRA for persistent PMP.³⁰ Indeed, when taking the group as a whole, we also failed to observe a significant difference at 2 weeks. Only when separately examining patients with higher baseline catastrophising scores was a difference in pain burden and opioid at 2 weeks observed.

Pain is a complex interaction between nociception and psychology, but few studies have differentiated patients’ response to analgesic therapy based on their psychosocial reaction to pain. The integration of psychosocial traits, such as catastrophising, into studies of regional anaesthesia, may provide an important insight into the mechanisms of postoperative pain and allow better prevention. The present work suggests that the analgesic benefit of TRA is present and especially pronounced for individuals who have higher preoperative catastrophising. Conversely, this may suggest that patients with lower catastrophising scores benefit less from TRA. The ability to discern phenotypic differences between individuals may be an important step toward personalising the perioperative analgesic plan and allow stratification and rational direction of scarce resources to those who are most likely to benefit, while minimising the risk amongst those who are less likely to benefit. To the best of our knowledge, this is the first study demonstrating an interaction between regional anaesthesia, catastrophising, and postoperative pain.

There are several limitations to this study. First, as the design was a prospective observational study, patients were not randomised for TRA. Although we did not observe any significant baseline differences between these groups (e.g. in age, baseline pain, opioid use, or baseline catastrophising), it is still possible that there is another unmeasured third factor that explains the lower pain scores amongst high catastrophising individuals who received TRA. Second, there was no placebo control, and the degree to which any of the analgesic effects of TRA, or a differential distribution of benefit amongst subjects which may be because of the varying extent of placebo response, is unknown. A randomised, placebo-controlled study using the same detailed baseline phenotypic assessment is needed to confirm these findings, and investigate the question of placebo. Third, being a pragmatic observational study that occurred in the natural clinical setting of a busy academic centre, the regional anaesthetic technique and protocol for analgesic use were not controlled or standardised over the study course. Fourth, because the primary collection method for pain assessment after the day of surgery was an electronic questionnaire, a selection bias may have occurred, more frequently including individuals who were more comfortable using e-mail and a computer. Finally, some attrition occurred at later time points, although equally amongst groups receiving or not receiving TRA, and between groups with high and low baseline catastrophising.

Conclusion

In summary, TRA provided some analgesic benefit overall for patients undergoing mastectomy, decreasing pain scores and opioid utilisation on the day of surgery. But for patients with a higher baseline catastrophising score (PCS>10), the benefit of TRA was more pronounced and prolonged, with decreased PBI and use of opioid analgesics also at 2 weeks after operation. Future studies of novel and known preventive analgesic strategies should include preoperative definition of phenotypic variables such as catastrophising, so that the differential efficacy between individuals may be systematically studied, and inform evidence-based personalisation of care.

Authors' contributions

Design: NZ, LD, TK, MG, GS, RE, KS

Analysis: NZ, KV, RE, KS

Recruitment: NZ, KS

Manuscript preparation: all authors

Data collection: TS

Handling editor: L. Colvin

Editorial decision date: 14 January 2019

Declarations of interest

The authors declare that they have no conflicts of interest.

Funding

National Institutes of Health (NIH K23 GM110540 to KLS).

Appendix A. Supplementary data

The following is the Supplementary data to this article: