JSUR Journal

Introduction

Traumatic spinal injury, or spinal trauma, is defined as acquired damage to the spinal cord and/or spinal column [1]. Spinal column trauma has an incidence up to 45 times greater than traumatic Spinal Cord Injuries (SCI) [2]. The worldwide incidence of traumatic SCI is estimated at 23 cases per million population in 2007, i.e., nearly 180,000 new traumatic SCI cases annually [3]. According to the Australian Institute of Health and Welfare, [4] land transport crashes constituted the majority (46%) of traumatic SCI in 2017-2018, followed by falls (35.8%). Other causes of SCI include water-related injury, heavy falling object, and horserelated and football-related injuries [4]. SCI has been considered to be one of the most detrimental conditions of acquired disabilities resulting in adverse outcomes to the individual, [5] with negative impacts on families and on society at large [6,7]. According to current evidence, sociodemographic and trauma-related factors associated with higher functional outcome following a SCI include: decreased severity of the SCI, younger age, lower neurological level of injury, and the higher initial American Spinal Injury Association (ASIA) Impairment Scale motor score [8]. The main modifiable factors associated with higher functional outcome following a SCI are: availability of surgical management, higher functional status at discharge from acute care, shorter acute care length of stay, and presence of a specialised multidisciplinary functional rehabilitation process [8]. Other potential modifiable factors include: prevention of medical complications, and higher intensity and patient participation level in functional rehabilitation therapies [8].

Quality of Life (QoL) can be affected after spinal trauma. A study identified that suboptimal physical health status following spinal trauma was predicted by tachycardia and hyperglycaemia (reversible factors), as well as multiple comorbidities and thoracic spine injuries (irreversible factors) [2]. However, none of the early predictors were associated with mental health-related QoL outcomes. This study noted that the proportion of patients with satisfactory mental health well-being (53.4%) was higher than the proportion of patients with satisfactory physical well-being (32.6%) [2]. Another study reported that younger age, return to work, and non-hospitalisation were associated with better overall QoL ratings [9]. Injury severity was related to physical domains of QoL, but was not associated with mental health after SCI [9]. Changes in physical function after SCI often affects psychosocial function thus affecting patients’ health-related QoL [10]. Psychological changes after SCI include depression and anxiety, loss of independence and social isolation [11]. Secondary conditions or complications related to the SCI can further affect QoL and incur further costs to the healthcare system [11]. Not all aspects of QoL are necessarily affected following a SCI. While some studies have shown that individuals with SCI typically have rated their physical functioning, pain and general health lower than the general population, [12] their ratings on emotional and mental health are varied and may not be lower [13,14]. To build on the current understanding, our study aimed to assess changes in QoL in the first three months following traumatic spinal injuries in Western Australia, taking into account the potential impact of sociodemographic-, injury- and treatment-related factors (including surgery). We hypothesised that QoL ratings would improve over time following surgery.

Method

Study Sample

A prospective cohort of patients with traumatic spinal injuries who were admitted to either the Western Australia State Spinal Unit or State Trauma Centre between October 2022 and January 2023 were recruited. The care of patients admitted via the State Spinal Unit (i.e., typically patients with only spinal injuries) was managed by spinal orthopaedic surgeons, while the care of patients admitted via the State Trauma Unit (i.e., typically patients with multiple injuries) was managed by trauma surgeons. These patients received multidisciplinary care, and surgical care if required. Based on the patients’ post-intervention function and eligibility, they were either discharged home, to a local rehabilitation facility, or to the State Rehabilitation Service. Patients were included if there was evidence of spinal injury on CT or MRI. Patients were excluded if they were not able to consent due to moderate cognitive impairment (i.e., Mini Mental State Examination score less than 20 out of 28) or if they had spinal injuries due to non-traumatic causes (e.g., ischaemic, cancer related causes).

Data Collection

Patients were approached within the first 4 days post-admission to obtain consent and were followed for three further assessments (i.e., 7-9 days post-admission, 4-5 weeks post-admission, 3 months post-admission). Patients were assessed in person if they were inpatients, or via telephone if they were discharged home.

Variables

Measurements: Age, gender and whether individuals identified as Aboriginal or Torres Strait Islander were recorded. Baseline function (i.e., independent or supported) and living status (i.e., living alone or with others) were recorded. The Clinical Frailty Scale (CFS) [15,16] and the Charlson Comorbidity Index (CCI) [17] were also recorded. The Mini Mental State Examination (MMSE) [18] was used to determine baseline cognition, and was modified. The two motor-related items were removed from the 30item MMSE to derive a total score out of 28. This adaptation was implemented as many patients were in strict spinal precautions or had lost hand function. Surgical intervention status (i.e., with or without surgery), mechanism of injury (i.e., motor vehicle accident [MVA] or non-MVA), highest level of spinal injury (i.e., lumbar/ sacral or cervical/thoracic) and the ASIA Impairment Scale [15] were recorded.

Outcome Variables: EuroQoL-5D (EQ-5D). This was used as recommended by the International Consortium for Health Outcomes Measurement forum to allow benchmarking across different countries, cultures, and patient demographic groups [19-21]. The five dimensions of the scale include: mobility, selfcare, usual activity, pain/discomfort and anxiety/depression. The EQ-5D-3L version rates each domain at three levels (1 = “no problem”, 2 = “some problem” or 3 = “extreme problem”)[19]. An overall QoL score is created by summing scores from each of the five domains, with scores ranging from 5 to 15. The EQ-5D also consists of a visual analogue scale (VAS) that assesses perceived health status from 0 (“worst health you can imagine”) to 100 (“best health you can imagine”)[19].

Analysis

Descriptive statistics were performed. Generalised linear mixed models with a cumulative logit link (i.e., an ordinal logistic regression with mixed effects) were performed for each of the five EQ-5D outcome domains separately. To analyse the overall QoL score (5 to 15) and the VAS, general linear mixed models were performed. Initially, univariate models were conducted, with fixed effects of time, each of the patient factors (age, gender, baseline function, Clinical Frailty Scale, living alone and Charlson Comorbidity Index) and each of the injury factors (surgical intervention, baseline ASIA, mechanism of injury, highest level of injury) for each of the of the five EQ-5D domains. Subsequently, a multivariate model was run for each outcome where time, and all patient and injury factors were included, as well as interactions between time and each injury factor so as to assess whether the relationships between each of the injury factors with each QoL outcome changed over time. All models contained a random effect of patient ID. Backwards hierarchical model selection was performed, where factors/interactions were sequentially removed (p > .05). All analyses were conducted in the R environment for statistical computing [22]. 

Ethics

The research project was assessed by the East Metropolitan Health Service Health Research and Ethics Committee. Approval was granted (Quality Activity reference number 47361) to recruit patients with spinal injuries and access medical records.

Results

(Table 1) shows the demographic and injury-related information of the 104 participants at admission. The mean age of participants was 55.8 (SD =±21.2), and 39.4% were aged above 65 years and 3% of the sample identified as Aboriginal or Torres Strait Islander. The majority of participants (84.6%) did not have a spinal cord injury. The most common cause of spinal injury was due to falls (54.8%) followed by motor vehicle related accidents (31.7%). Univariate and multivariate results are reported in (Supplementary Tables A.1 and A.2) respectively. Thirty-two participants (30.8%) underwent surgery, of which three participants required a second surgery.

Table 1: Demographic and Injury-related Variable Table at Hospital Admission.

Note. OR = Odds Ratio; MD = Mean Difference; p < .01 **, p < .001 ***; n.s. = not significant

Table A.1: Univariate results analysing the seven outcome variables.

Note. OR = Odds Ratio; MD = Mean Difference; p < .01 **, p < .001 ***; n.s. = not significant. For the multivariate model, time, and all demographic and injury factors were included, as well as interactions between time and each injury factor.

Table A.2: Multivariate results analysing the seven outcome variables.

Three participants (2.9%) passed away, but none of these were post-surgical patients. There were no missing data for predictor variables (as the information was obtained during the first assessment). (Figure 1) shows how the QoL scores changed over the four assessment timepoints (Baseline, 7-9 days, 4-5 weeks and 3 months) across the five EQ-5D outcome subdomains (Mobility, SelfCare, Usual Activities, Pain/Discomfort and Anxiety/Depression).

Figure 1: Distribution of QoL ratings over time across the five EQ-5D outcome subdomains

Mobility

Mobility QoL improved over time (see Figure 1 and Table A.2). Specifically, patients had significantly lower mobility QoL scores (i.e., better mobility) at every time point compared to any earlier time point (p < 0.001). Comparing each time point interval, the rate of improvement of mobility scores was the largest between the admission and 7-9 days interval (OR 0.02, 95% CI 0.01-0.04, p < .001; Table A.2). Other than baseline functional status and baseline ASIA, there were no statistically significant relationships between any of the other patient or injury factors with the mobility QoL domain (see Table A.2).

Self-Care

Age, baseline function, baseline ASIA, and an interaction between time and mechanism of injury were associated with the self-care QoL domain. The interaction between time and mechanism of injury indicated that how the self-care QoL domain changed over time depended on the mechanism of injury (see Figure 2 and Table A.3). Those with a non-MVA mechanism of injury were less likely to have higher self-care QoL (i.e., more likely to have better self-care) at admission than those with a MVA, although this did not reach statistical significance (OR 0.30, 95% CI 0.08-1.14, p = 0.078). Conversely, at 3-months post admission, those with a non-MVA mechanism of injury were more likely to have higher self-care QoL domain scores (more likely to have poorer self-care) than those with an MVA mechanism of injury (OR 6.02, 95% CI 1.06-34.1, p = 0.043; Table A.3). For those with an MVA mechanism of injury, self-care QoL scores were significantly lower (i.e., better self-care) at each time point than at all previous time points.

Figure 2: Proportions of patients with each Self-Care QoL score broken down by mechanism of injury and time

Table A.3: Interaction results between time and mechanism of injury from the multivariate model (also adjusted for age, baseline function and baseline ASIA) on Self-Care.

Usual Activities

Usual activities improved over time (Figure 1 and Table A.2). Specifically, patients had significantly lower usual activities QoL domain scores (i.e., better usual activities) at every time point compared to any earlier time point (p < 0.001). Comparing each time point interval, the rate of improvement of usual activities scores was the largest between the 7-9 days and 4-5 weeks interval (OR 0.021, 95% CI 0.10-0.44, p < .001; Table A.2). Other than baseline ASIA, there were no statistically significant relationships between any of the other patient or injury factors with the mobility QoL domain (see Table A.2).

Pain/Discomfort

Patients had lower pain and discomfort QoL domain scores at 4-5 weeks and 3 months when compared to admission and 7-9 days (Figure 1 and Table A.2). The rate of improvement of pain/discomfort scores was the largest between the 7-9 days and 4-5 weeks interval (OR 0.44, 95% CI 0.24-0.82, p = .009; Table A.2). There were no associations between any of the patient or injury factors with the pain and discomfort QoL domain (Table A.2).

Anxiety/Depression

Anxiety/depression QoL domain scores were lower (i.e., better QoL) at 4-5 weeks and 3 months when compared to admission, and when comparing 3 months to 7-9 weeks (Figure 1 and Table A.2). Multivariately, time, gender, clinical frailty scale, living alone, and highest level of injury were associated with the anxiety/depression QoL domain (see Table A.2).

Overall QoL

(Figure 3) shows the overall QoL scores over time. On multivariate analysis, baseline function and an interaction between baseline ASIA and time (p = 0.025) were associated with the overall QoL domain (see Figure 4 and Table A.4). Those participants independent at baseline had a lower overall QoL score (i.e., better overall QoL) than those that were supported at baseline (MD -1.59, 95% CI -2.30 to -0.89, p < .0001; Table A.2).

Figure 3: Overall QoL Scores over time

Figure 4: Means and 95% confidence intervals of overall QoL score over time, categorised by baseline ASIA estimated from the model (i.e., these values are averaged over baseline function levels).

Table A.4: Interaction results between time and baseline ASIA from the multivariate model (also adjusted for baseline function) on Overall QoL.

Overall QoL score was significantly greater (worse overall QoL) in those with an SCI (ASIA A-D) compared to non-SCI (ASIA E) patients at all time points except at admission (i.e., this difference was greater at 7-9 days, 4-5 weeks and 3 months). Comparing each time point interval, the rate of improvement of overall QoL scores in SCI patients was the largest between the 7-9 days and 4-5 weeks interval (MD 1.38, 95% CI 0.48-2.27, p < .003; Table A.4) while the largest rate of improvement of overall QoL scores in non-SCI patients was between the admission and 7-9 days interval (MD 2.19, 95% CI 1.8-2.58, p < .0001; Table A.4).

Visual Analogue Scale (VAS)

(Figure 5) shows the VAS ratings over time. Following multivariate analysis, time, baseline function and baseline ASIA were associated with VAS. Specifically, VAS increased over time, where each time point had significantly greater VAS than at all previous time points (p < .0001; Table A.2). Comparing each time point interval, the rate of improvement of VAS ratings was the largest between the 7-9 days and 4-5 weeks interval (MD 13.78, 95% CI 8.99-18.57, p < .0001; Table A.2). Other than baseline functional status and baseline ASIA, there were no statistically significant relationships between any of the other patient or injury factors with VAS ratings (see Table A.2).