Predictors of chronic neuropathic pain after scoliosis surgery in children

2.1 Inclusion and exclusion criteria

Inclusion criteria were: age <18 years, single stage posterior fixation spinal surgery, idiopathic scoliosis, absence of contraindication to the use of analgesics used by protocol in our institution for scoliosis surgery: paracetamol, NSAIDs, nefopam, opioids, gabapentin. Exclusion criteria were: secondary surgical intervention for infection or other surgical complications, thoracoplasty, anterior release or fixation, sacral fusion, inability to understand French or English languages and absence of consent to participate to the study.

Inclusion was performed during the preoperative consultation (a preoperative consultation is performed systematically in the 5 week preceding the scheduled date of surgery) and consent of patients and parents were obtained at this time.

2.2 Perioperative anaesthesia

Anaesthesia was standardized and follows our protocol for the management of scoliosis surgery, the anaesthesia protocol published in a previous article [19]. Preoperative anaesthesia and surgical preparation included standardized recombinant Erythropoietin administration and iron supplementation, intraoperative antifibrinolytic agent administration (tranexamic acid) and intraoperative red blood cell salvage. In addition, all patients were given oral Gabapentin 1200 mg before surgery that was continued at a daily dose of 600 mg until day 5. Anaesthesia protocol was as follows: preoxygenation for 3 min, then induction using sevoflurane (6% in a 50% mixture of O₂/N₂O) or intravenous propofol (5–7 mg kg^-1). Maintenance involved sevoflurane at 0.8 minimal alveolar concentration. Intraoperative analgesia was in the form of sufentanil boluses (0.2 μgkg^-1 when heart rate or mean arterial pressure increased by more than 20% of preoperative values) and intrathecal morphine (0.5 μgkg^-1 after incision performed under visual control by the surgeon). All patients were intubated with a non-depolarizing muscle relaxant. Muscle relaxation was systematically reversed at the end of surgery. Patients were operated upon in the prone position on a Jackson frame. A minimum of two venous cannula were inserted and Ringer’s Lactate solution (RL) administered at 2 ml kg^-1 h^-1. Further vascular filling was administered according to oesophageal Doppler measured stroke volume and intravenous fluid load response. Filling solutions were initially RL, then hydroxyethylstarch (130/0.4, Voluven®, FRESENIUS KABI FRANCE, Sevres, France) following consecutive failure of two 10 ml kg^-1 crystalloid boluses. Arterial pressure and heart rate were maintained within 20% ranges of preoperative values using vasopressor agents for low values (phenylephrine 50 μg boluses), and vascular filling or sufentanil boluses for high values. All patients received intrathecal morphine (5 μgkg^-1) following surgical dissection.

Postoperative analgesia was standardized (protocol established in 2014), and included morphine, paracetamol, non-steroidal anti-inflammatory drugs (NSAIDs) and nefopam. Morphine was administered intrathecally as above, then intravenously by titration in the PACU. Patient controlled analgesia (PCA) was initiated before leaving PACU and continued on the ward (Protocol 1). Pain assessment employed the numerical rating scale – NRS. Pain assessments were performed every 5 min during morphine titration and every 60min during PCA (patient-controlled analgesia) or NCA (nurse-controlled analgesia) morphine administration. The pain team, a pain specialist and a nurse, were responsible for analgesia following discharge from PACU. All patients were assessed at least daily with respect to the continuation and/or modification of morphine therapy. Intravenous opioids were prescribed for a maximum of three days and then either discontinued or replaced by oral morphine. Non-opioid analgesic therapy was also standardized and given systematically 30 min before the end of surgery or in the PACU. It consisted of intravenous or oral paracetamol 15 mg kg^-1 6 hourly, and intravenous ketoprofen, 1 mg kg^-1 8 hourly or oral ibuprofen 10 mg kg^-1 6 hourly where there was no contra-indication. Nefopam was also administered to patients 16 years or older at a dose of 0.25 mg kg^-1 6 hourly. Postoperative fluid management consisted a crystalloid solution with 4% glucose and 4g of NaCl per litre (B26®, APHP, Paris, France), at 2 ml kg^-1 h^-1. Oral analgesics were administered from postoperative day 2: paracetamol (15 mg kg^-1 6 hourly), ibuprofen (10 mg kg^-1 6 hourly) and oral morphine. Oral morphine was administered in both long acting (skenan LP® 0.5 mg kg^-1 -12 h^-1) and short acting (actiskenan® 0.1 mg kg^-1 every 4h as needed with a maximum daily dose of I mg kg^-1) preparations. All patients undergone rapid postoperative rehabilitation: sitting position (or at least a 45° head-up) 6 h after surgery, rapid recovery of oral intakes at day 1 and mobilization at day1.

2.3 Data collected

Data collection was performed during hospital stay (for all outcomes of the pre, intra and early postoperative periods) and at one year at surgical follow-up (performed by the surgeons: between II and 14 months depending on patient’s availability). Data collected included: age, weight, type of surgery, ASA status, the NRS score at admission for surgery, the presence or absence of continuous pain in the 3 months before surgery (defined as a pain in the spine region limiting daily activities such as movement, sport and schooling), preoperative opioid administration in the previous 3 months, surgical duration and the number of levels fused. Further data included non-opioid and opioid analgesic administration intraoperatively and over the first five postoperative days, maximal pain at movement around the wound scar during the first five postoperative days and at 1 year post surgery, the presence of neurological damage (hypoesthesia and/or anaesthesia) in the area of wound scar during the first five postoperative days and at 1 year post surgery and DN4 (Douleur Neuropathique 4 score) [20] score (around the wound scar) at 3 days and 1 year post surgery (the complete DN4 form obtained during the surgical follow-up). Data at one year were obtained during the systematic surgical follow- up: the NRS (maximal pain at movement) and DN4 scores (around the wound scar) were considered as positive when ≥4 [1,20] and when interesting the operative area.

2.4 Statistical analysis

The primary objective of our study was to determine factors associated with the presence of a neuropathic pain at 1 year after surgery. Neuropathic pain was defined as the presence of a pain (NRS ≥ 4), the presence of signs of neurological damage within the operated area and a DN4 score ≥4 in the area around the wound scar.

Univariate analyses were performed using ANOVA (continuous variables with sample size ≥30), Mann and Whitney tests (continuous variables with sample size <30) and the X² or Fisher’s exact test (discrete variables). Significant associated factors in univariate analysis were transformed to discrete variables by receiving- operator characteristics analysis with the cut-off defined as the value maximizing the value of J score (sensitivity + specificity). Given the main outcome, the presence of a neuropathic pain, was discrete in nature, univariate model variables with significance results of less than 0.2 were entered into a stepwise multivariate logistic model. Statistical analyses were performed using the SPSS 20.0 software (IBM Company, Chicago, Illinois, USA).

According to previous publications [10], chronic pain after surgery was ranging from 9.5% to 60% (with a mean of 25%). Accordingly, we planned to recruit at least 40 patients in order to have 10 patients with persistent pain.

4.1 Prevalence and risk factors of chronic persistent pain

Our study found a high proportion of patients developed chronic pain despite the use of extensive preventive and curative analgesia technics during the perioperative period (gabapentin, regional and systemic opioids, and non-opioid analgesics). Previous studies in children have found the prevalence of moderate to severe chronic pain ranging from 2 (inguinal herniorraphy) to 48.5% (amputation) [10,21,22,23,24]. In adults, there are more data and the overall prevalence is estimated to be around 30% with higher percentages observed for amputation (50–85%) and thoracic surgery (50%) [8]. The high prevalence of chronic postoperative pain observed in our study may to some extent be explained by cumulating risk factors for persistent pain in the study population. Firstly, female gender has been previously identified as a major risk factor of chronic pain in adult populations [8,9,10,16] and 86.1% of our study population was female. Psychological factors such as depression and anxiety have also been implicated in the development of chronic post-surgical pain [8,9,16]. Anxiety, depression and catastrophization were previously found to be frequently present in adolescent patients with idiopathic scoliosis, especially during the perioperative period [22,25,26,27,28]. Finally, the extent of surgery is another identified risk factor of chronic post-surgical pain, particularly neuropathic pain, probably as a result of increased nerve injury [5,8,9,16].

Concerning the prevalence of persistent pain, our results are similar to previous studies on the same topic. Landman and collaborators [23] reported that 80% of their patient cohort suffered preoperative pain before surgery and only 40% reported reduced pain intensity 2 years after surgery. Moreover, Chidambaran and collaborators [29] reported an incidence of persistent postoperative pain of 42% in a cohort of patients undergone spinal surgery. However, our results are quite different to those published by Bastrom [24] and Sieberg [30] that described a prevalence of persistent postoperative pain of 16% and 11% at 1 and 2 years post-surgery, respectively. The discrepancies in these three studies may be due to heterogeneity in pain assessment methodology [31], differences in available psychological support [22,25] or the fluctuation of postoperative pain over time. Pain fluctuation was present in our population: 3 patients exhibiting immediate neuropathic pain were without neuropathic pain at 1 year (Table 2). Similarly, pain trajectories described in Sieberg’s study that found some patients with an improvement of pain at 1 year having worse pain at 5 years and vice versa [30].

The most important independent factor given its statistical weight was morphine consumption at day 1. Opioid-related hyperalgesia with neuropathic sensitization and progress to chronic neuropathic pain was previously described [32,33]. In contrast with our results, previous studies in adults and children have identified postoperative pain as a major predictive factor for postoperative chronic pain [8,10]. This was considered a marker of pain sensitivity and predisposition to nervous system pain sensitization. Given that postoperative pain was statistically linked to opioid consumption, both factors are probably involved as risk factors of chronic pain. Continuous preoperative pain in the three months preceding surgery was also an independent factor associated with the development of postoperative neuropathic pain in our cohort. This factor has been previously identified in paediatric patients, particularly in scoliosis surgery [23,24,29] and in adults [8,11,16,18]. Preoperative pain is thought to sensitize central and peripheral pain pathways, favouring the development of chronic neuropathic pain post-surgery [16]. Finally, day 3 neuropathic pain was not found to be independently associated with postoperative neuropathic pain at one year. Immediate neuropathic pain has been identified as a risk factor of persistent postoperative pain during abdominal surgery in adults [34,35]. Hyperalgesia at day 3 was probably not found to be an independent predictor of chronic postoperative pain in our study due to its strong association with other predictors identified in the multivariate analysis (preoperative pain, postoperative pain and opioid consumption).

Two patients described persistent non-neuropathic post-surgical pain, as described elsewhere in adult populations [8]. Early studies in the field of persistent post-surgical pain only described neuropathic pain. More recent studies describe both chronic neuropathic and less frequently non-neuropathic pain [8,10,18], as described in our population. This suggests that the development of chronic post-surgical pain is a complex and multifactorial process.

It appears that adults and children share the same mechanisms and risk factors for neuropathic post-surgical pain [8]. Preoperative pain, the extent of surgery and therefore of nerve damage, and central and peripheral nervous system sensitization by either pain or opioid agents are described as contributing factors [8,9,16].

4.2 Implication for prevention of chronic persistent pain

Given that many risk factors of chronic persistent pain (the extend of surgery and neurological damage and the female patients) are not preventable, our results strongly suggests the management of patients scheduled for scoliosis surgery repair with preoperative pain by either a chronic or acute pain team in order to decrease the preoperative sensitization. In addition, every effort should be made in order to decrease the postoperative opioid consumption. This could be achieved using a multimodal analgesia and by applying regional analgesia techniques such as epidural catheter (placed by surgeon) [36]. Finally, a better control of postoperative pain and a psychological support might also provide a substantial help in preventing the occurrence of chronic pain after surgery.

4.3 Study limitations

Our study suffers some limitations. First, the sample size was limited and some predictive factors might have not been identified as a result. However, a recent study performed on a similar type of surgically corrected idiopathic scoliosis patients identified the preoperative pain and the opioid consumption as predictors of chronic pain [29]. This strongly suggests a good external validation of our study. Another limitation of our study is the lack of reporting of the nature of preoperative pain (nociceptive versus neuropathic). Moreover, preoperative pain was evaluated before surgery and might not reflect the overall period before surgery. Further data may have further refined the predictive value of persistent postoperative pain as a function of its nature. Finally, we used the DN4 questionnaire in our study while it was not yet validated in the paediatric population. However given the age of patients included in this study, they were expected to respond as young adults. In addition, the psychological consequences of this pain and its influence on daily activities were not assessed using the DN4 questionnaire.

4.4 Conclusions and implications

Our findings indicate that despite an adequate management of postoperative pain, chronic neuropathic persistent pain occurred in 48.2% of patients. The current study identified preoperative pain and postoperative opioid consumption as independent predictors of chronic neuropathic pain after posterior fixation surgery for correction of idiopathic scoliosis in children. Patients scheduled for scoliosis surgery correction and exhibiting persistent preoperative pain should be systematically assigned to dedicated pain teams; postoperative opioid administration should be reduced as possible. We hope our results may assist those implementing preventive strategies to reduce the incidence of chronic neuropathic pain in this population.