DEVELOPMENT OF A PREDICTION MODEL FOR SHORT-TERM OUTCOME OF PATIENTS WITH SCIATICA, AFTER PHYSIOTHERAPY, IN THE GREEK HEALTH SYSTEM

N. Kontakiotis¹, A. Rushton², V. Sakellari¹, E. Billis³, G. Papathanasiou⁴, G. Gioftsos¹

¹University of West Attica, Department of Physiotherapy, Laboratory of Advanced Physiotherapy, Faculty of Health and Caring Sciences, Athens, Greece, ²Western University, London, School of Physical Therapy, Faculty of Health Sciences, Ontario, Canada, ³University of Patras, Department of Physiotherapy, School of Health Rehabilitation Sciences, Patras, Greece, ⁴University of West Attica, Laboratory of Neuromuscular and Cardiovascular Study of Motion, Faculty of Health and Caring Sciences, Athens, Greece

Background: Sciatica is a very common musculoskeletal condition affecting 10% to 40% of the population. Appropriate assessment and management are important to improve outcome. Physiotherapists use several measures (patient reported outcome measures and performance-based measures) to assess patients to inform selection of the most suitable treatment. There is no reliable classification algorithm for patients with sciatica based on their clinical characteristics.

Purpose: To develop a clinical prediction model to categorize patients with sciatica, in terms of predicted outcome, based on their clinical characteristics.

Methods: A prospective observational multicenter design recruited consecutive patients (n=259) with sciatica referred for physiotherapy from multiple sites in Greece. Patients’ characteristics, patient reported outcome measures and performance-based measures were collected at baseline using a standardized protocol, prior to commencement of treatment. Upon completion of treatment at 3 weeks post baseline, participants’ outcomes were collected by telephone using the Global Perceived Effect Scale (GPES). Results were dichotomized such that ‘completely recovered’ and ‘much improved’ were referred as ‘improved’ and were considered as positive. For the descriptive statistical analysis, the continuous candidate predictors expressed in the form of ‘mean’ and ‘SD’. In order to create a predictive tool for categorizing patients with sciatica, regarding clinical outcome, based on their initial clinical characteristics, the independent factors that can predict the value of the GPES scale were investigated through Multiple Linear Regression.

Results: 132 women and 127 men participated in the study, with mean (SD) age 33.7 (12.3) years and BMI 25.8 kg/m²(SD = 3.2). Positive outcome was reported by n=188 participants. The summary of the independent factors of the Multiple Linear Regression model was investigated by the STEPWISE method. The method evolved in 9 phases. The result was a model of 9 predictors (Age, Duration of Pain, ODI, Neurological Examination [Muscle Test], SBI, Location of Pain, VAS, S-LANSS, SLR). Regarding the predictive power of the final regression model, the Adjusted R2 index was calculated (93.7%). The model had a particularly high adjustment to the data of the present study, as well as a very high accuracy regarding the estimation in the general population. The assumption of independence was tested with the contribution of the Durbin-Watson (2.361). The importance of the independent variables in the final regression model, in terms of their predictive ability, were investigated with their t values (0.034 to < 0.001).

Conclusions: Analysis showed that some clinical characteristics might be useful to predict the short term outcome of patients with sciatica. Ranking these clinical characteristics, from the most important to the least important, it emerged a model with the following order: Age >ODI> SBI> Neurological examination (Muscle test)> Location of pain> Duration of pain> S-LANSS> VAS> SLR.

Implications: The algorithm(model) will enable initial decision making regarding the classification of sciatica patients in primary care of the Greek health system.

Funding acknowledgements: The authors have not declared a specific grant for this research.

Keywords:
Sciatica
Prediction model
Classification

Topics:
Primary health care
Pain & pain management
Musculoskeletal

Did this work require ethics approval? Yes

Institution: University of West Attica, Athens, Greece

Committee: Ethics and Deontology Committee

Ethics number: 38313-09/06/2020, 10226-10/02/2021.