Predicting the pain continuum after adolescent idiopathic scoliosis surgery: A prospective cohort study

Abstract

Background: Chronic postsurgical pain (CPSP) affects half a million children annually in the United States, with dire socioeconomic consequences, including long-term disability into adulthood. The few studies of CPSP in children are limited by sample size, follow-up duration, non-homogeneity of surgical procedure and factors evaluated.

Methods: In a prospective study of 144 adolescents undergoing a single major surgery (spine fusion), we evaluated demographic, perioperative, surgical and psychosocial factors as predictors of a continuum of postsurgical pain: immediate, pain maintenance at 2-3 months (chronic pain/CP) and persistence of pain a year (persistent pain/PP) after surgery.

Results: We found an incidence of 37.8% and 41.8% for CP and PP. CP and acute pain were both significant predictors for developing PP (p-value <0.001 and 0.003). Preoperative pain and higher postoperative opioid requirement was significantly associated with CP (p = 0.015, p = 0.002), while Childhood Anxiety Sensitivity Index (p = 0.002) and surgical duration (p = 0.014) predicted PP. The final regression models had reasonable predictive accuracy (c-statistic of 0.73 and 0.83 for CP and PP, respectively). Anxiety scores and catastrophizing for child and parent were found to be significantly correlated (p = 0.005, p = 0.013 respectively). Pain trajectories revealed that 65% of patients who developed PP reported CP and high pain trends; however, 33% of those who developed PP could not be identified using solely pain criteria.

Conclusion: Persistent postsurgical pain in children is a significant problem. It can be predicted in part by combinations of psychological and clinical variables, which may provide evidence-based measures to prevent development of CPSP in the future.

Significance: In a homogeneous cohort of adolescents undergoing spine fusion, we report a high incidence of persistent postsurgical pain (41.8%) predicted by child anxiety, perioperative pain, and surgical duration. Our results stress timely preventive and therapeutic strategies.

Trial registration: ClinicalTrials.gov NCT01839461 NCT01731873.

Conflict of interest statement

Conflicts of interest

None declared.

© 2017 European Pain Federation - EFIC®.

Figures

Figure 1

This figure depicts the recruitment and follow-up timeline for the study. Reasons for not recruiting eligible screened/approached patients, reasons for withdrawal and follow-up losses are described in detail.

Figure 2

The correlation between anxiety (Visual Analogue Scale) and catastrophizing scores (based on Patient Catastrophizing Scale-child questionnaire) of child with corresponding scores of parent are presented in the upper panel and correlation with acute pain outcome (area under curve of pain scores over postoperative days 1 and 2) are presented in the lower panel. Spearman’s correlation coefficient r between anxiety scores for child and parent was 0.27 (p = 0.005) and between overall PCS scores for child and parent was 0.28 (p = 0.013). Anxiety of child correlates with pain AUC (Spearman’s correlation coefficient r = 0.23, p = 0.021), but the child overall catastrophizing score does not.

Figure 3

This figure shows the correlation of the acute pain outcome (area under curve of pain scores over postoperative days 1 and 2 or AUC) with the predictors significant by multiple regression. We see that patients who have higher preoperative pain scores have higher postoperative pain AUC (Spearman’s correlation coefficient r of 0.32, p

Figure 4

Linear pain trajectories and their 95% prediction intervals of four combinatorial patient cohorts with and without chronic and persistent pain are plotted, based on the linear trajectory models. The numbers of patients in each group are mentioned (N) – of note, pain trajectories of total of 107 patients who had both CP and PP outcomes reported are presented. Although the PP outcome was reported for 110 patients, there were three of these patients for whom CP was not reported, leaving an intersection of 107 patients. Time point ‘0’ on the x-axis is the preoperative pain score. While 75% of the patients had clear cut pain trajectories (either low or high pain scores all along from preoperative to years later), the most worrisome group is the CP = No, PP = Yes group. Patients in this category would likely not be identified as at risk for developing pain in the future if they were followed up for only 1 year after surgery. It is likely that a patient with pain score<2/10 at 1 year after surgery is highly unlikely to go on to develop pain later, or the converse. Coloured bands represent 95% prediction intervals and the lines represent fitted linear regression lines of the pain trajectory.