Predicting 7-Year Survival in Individuals with Chronic Spinal Cord Injury: The Role of Self-Reported Functional...

Previous research investigated prognostic factors for survival in prevalent cases of chronic spinal cord injury (CSCI) based on self reported information. However, most of these studies originated in the US and relied on cohorts drawn from individual hospitals or hospital networks. This population-based cohort study aims to identify prognostic factors affecting survival within the Swiss Spinal Cord Injury (SCI) community, emphasizing self-reported data on functional independence (FI), secondary health conditions (SHC), comorbidities, and emotional well-being.

Data from the Swiss 2012 SCI community survey were linked to vital status information. Missing data were addressed using multiple imputation. FI was measured using the self-report version of the Spinal Cord Independence Measure (SCIM-III-SR). SHC and dimensions of health-related well-being were assessed using multi-categorical variables. Emotional well-being was evaluated using the Short-Form Health-Survey (SF-36). Continuous and multi-categorical predictors were dichotomized by selecting the most predictive cut-off. The prognostic model was developed using a flexible parametric survival model, with model selection guided by the Akaike Information Criterion, following the reduction of potential predictors based on univariable analyses. Basic demographic and lesion characteristics were included in all models under consideration (base model).

The study included 1,011 individuals with a median time since injury of 11 years. During the observation period, 110 deaths occurred. The model selection procedure identified the following self-reported predictors: SCIM-Respiratory-and-Sphincter-Management (HR: 1.77, p=.012), SCIM-Self-Care (HR 1.66, p=.038), contractures (HR; 1.42, p=0.125), respiratory problems (HR;2.06, p=0.002), and cancer (HR: 1.86, p=.0.084). From the base model, being male (HR: 2.15, p=0.001) and age at survey (HR: 1.09, p<0.001) were the strongest predictors. Four distinct model-based risk groups were established. The prognostic index demonstrated a good fit between observed and predicted survival, along with high discrimination between risk groups.

Self-reported FI and health conditions, assessed years after injury, provide valuable prognostic information for the survival of prevalent cases with CSCI. However, internal and external validation of the existing models is required before clinical implementation can be considered.