Background & Aims

Complex regional pain syndrome (CRPS) is a condition elicited by an injury to the limbs. characterized by a combination of pain and sensory, autonomic, trophic and motor abnormalities. CRPS usually affects a single limb following surgery or limb fracture and its symptoms are not explained by the original injury [1, 2]. The course of illness varies considerably between patients, with some of them having spontaneous recovery, persistence of symptoms or having progressive symptoms. It can affect the patients’ ability to work and function in daily life [3]. Patients’ definition of recovery is more complex than previously thought, and it likely encompasses multiple aspects [4, 5]. This makes it important to understand which outcome measures reflect patients’ impression of change, so that they can be targeted in future rehabilitation programs. Additionally, we tried to predict patients’ impression of change before treatment initiation based on physical and biological factors. Several studies tried to predict the outcome of CRPS, but they are characterized by low sample size and vary in the factors they used to determine CRPS resolution[6]. The predictive approach holds promise in allowing physicians to tailor treatments to cases where positive responses are less likely, thereby considering alternative or combinational therapeutic strategies. Furthermore, our investigation seeks to contribute valuable insights into the pathophysiology and trajectory of CRPS.

Methods

A total of 54 CRPS patients participated in our study, and we followed their progress for 12 months. All patients completed the following on baseline and 12-month sessions: pain score (current, mean, min and max) on 0-10 numeric rating scales, the Graded Chronic Pain Scale[7], The State-Trait Anxiety Inventory (STAI) [8], and Beck Depression Inventory (BDI) [9]. Trained physicians assessed CRPS severity scores (CSS) and Quantitative sensory testing (QST) [10]. Laboratory measures of inflammatory markers as well as lipid profiles were measured at both sessions.

To understand the factors that contribute the most to patients’ impression of change, the PGIC score was correlated with the 12-month score of pain, disability, CSS and BDI. We also correlated PGIC with the difference in score (12 months – baseline) of the same four variables.

To evaluate patients’ impression of change, we utilized the Patient’s Global Impression of Change (PGIC) questionnaire[11]. Patients were categorized as improvers if they reported improvement by 1 point or more, while the rest were categorized as non-improvers. We used the following predictors at baseline: pain scores, BDI, STAI, QST and lab values. Subsequently, the data were split into training (40) and testing (14) sets. Factor analysis was performed on the given data after splitting, and a stepwise logistic regression model was used to predict patients’ impression of change after 12.

Results

To understand which factors correlate the most with PGIC scores, we performed a Pearson’s correlation and found that PGIC scores at 12 months correlate mostly with reported pain and disability scores at 12 months with moderately strong correlation coefficients. Interestingly, we found that PGIC correlations were lower for the score differences (Follow up – baseline) in general.

To predict future responses based on PGIC scores, we performed a factor analysis on the input data, resulting in a four-factor solution encompassing pain, QST, and two factors related to lab results. Age and gender were included separately without undergoing factor analysis. Using these six variables, we conducted a stepwise logistic regression, which significantly predicted PGIC scores in the training data. When applied to the testing data, the model predicted the impression of change with 78.5% accuracy and an 80% area under the curve (AUC).

Conclusions

Our results indicate that patients’ reported impression of change cannot be reduced to one outcome measure only. This shows that considering more than outcome measure when determining treatment success, such as reduction in pain, disability and CSS Scores, is more in line with what the patient’s consider as treatment success. This insight could pave the way for developing new assessment models for treatment response, centered on these three factors, offering a more accurate reflection of what patients consider as improvement in CRPS.

Our machine learning approach provides reliable prediction of patients’ impression of change as a result of CRPS treatment. The model uses information collected at baseline to provide prediction of impression of change at 1 year. This predictive ability will have valuable clinical and pathophysiological implications. This can be the seed for further studies designed to assess the effect of each of these variables on the pathophysiology of CRPS.

References

  1. Birklein, F., et al., Complex regional pain syndrome-phenotypic characteristics and potential biomarkers, in Nature Reviews Neurology. 2018, Nature Publishing Group. p. 272-284.
  2. Marinus, J., et al., Clinical features and pathophysiology of complex regional pain syndrome, in Lancet Neurol. 2011. p. 637-685.
  3. de Mos, M., et al., Outcome of the complex regional pain syndrome. The Clinical journal of pain, 2009. 25(7): p. 590-597.
  4. Llewellyn, A., et al., Are you better? A multi-centre study of patient-defined recovery from Complex Regional Pain Syndrome. European Journal of Pain (United Kingdom), 2018. 22(3): p. 551-564.
  5. Hush, J.M., et al., Recovery: What does this mean to patients with low back pain? Arthritis Care and Research, 2009. 61(1): p. 124-131.
  6. Wertli, M., et al., Prognostic factors in complex regional pain syndrome 1: A systematic review, in Journal of Rehabilitation Medicine. 2013. p. 225-231.
  7. Von Korff, M., et al., Grading the severity of chronic pain. Pain, 1992. 50(2): p. 133-149.
  8. Spielberger, C.D., et al., Manual for the State-Trait Anxiety Inventory; Palo Alto, CA, Ed. Palo Alto: Spielberger, 1983.
  9. Beck, A.T., R.A. Steer, and G.K. Brown, Beck Depression Inventory Manual. The Psychological Corporation. San Antonio, TX, 1996.
  10. Rolke, R., et al., Quantitative sensory testing: A comprehensive protocol for clinical trials. European Journal of Pain, 2006. 10(1): p. 77-77.
  11. Hurst, H. and J. Bolton, Assessing the clinical significance of change scores recorded on subjective outcome measures. Journal of Manipulative and Physiological Therapeutics, 2004. 27(1): p. 26-35.

Presenting Author

Abdelrahman Sawalma

Poster Authors

Abdelrahman Sawalma

University Hospital Würzburg, Department of Anesthesiology, Intensive Care, Emergency and Pain Medicine, Center for Interdisciplinary Pain Medicine, Würzburg, Germany

Lead Author

Juliane Becker

University Hospital Würzburg, Department of Anesthesiology, Intensive Care, Emergency and Pain Medicine, Center for Interdisciplinary Pain Medicine, Würzburg, Germany

Lead Author

Ann-Krisitn

University Hospital Würzburg, Department of Anesthesiology, Intensive Care, Emergency and Pain Medicine, Center for Interdisciplinary Pain Medicine, Würzburg, Germany

Lead Author

Heike Rittner

University Hospital Würzburg, Department of Anesthesiology, Intensive Care, Emergency and Pain Medicine, Center for Interdisciplinary Pain Medicine, Würzburg, Germany

Lead Author

Topics

  • Specific Pain Conditions/Pain in Specific Populations: Complex Regional Pain Syndrome (CRPS)