BACKGROUND AND PURPOSE
Setting mobility goals is an important aspect of physical therapy following spinal cord injury (SCI).1 Goal planning involves patients and therapists together identifying appropriate goals. The goals should be SMART, that is, specific, measurable, achievable, realistic, and timely.2,3 While patients need to be actively involved in the goal-planning process, they also need guidance from their therapists.4 However, the quality and appropriateness of therapists' guidance rely on their skills in accurately predicting patients' future mobility. While there are extensive data providing therapists with information about key predictors of some aspects of mobility,5–13 there is little information about predictors of all the aspects of mobility commonly addressed in therapy. For example, there is little predictive information about how quickly or how well a patient with SCI will be able to push a manual wheelchair or whether a person with paraplegia will be able to get up off the floor. Regardless of the available information to help guide therapists' predictions of patients' future mobility, very little is known about how well therapists apply this information in the clinical setting. The ability of therapists to predict patients' future mobility is important because it determines the appropriateness of goals and it ensures an efficient discharge process.14 Inaccurate predictions that either overestimate or underestimate future mobility can have deleterious implications on rehabilitation, equipment prescription, patient expectations, discharge planning, and length of stay. A Dutch study in 2000 examined the predictions made by a rehabilitation team and patients regarding functional outcomes after SCI.4 While this study provided useful information, it was a retrospective study relying on data records and it only examined whether patients could or could not independently perform a few key mobility and self-care activities. Therefore, the purpose of the current study was to progress this work and determine how accurately physical therapists can predict patients' future mobility using standardized outcome assessment scales. Accurate predictions of mobility are not only important to the success of goal planning but also important for ensuring that therapy is appropriately focused and financial resources are allocated efficiently by ordering appropriate equipment and in making the necessary modifications to homes.
A cohort observational study was undertaken involving 2 spinal injury units in Sydney, Australia. All patients (n = 93) admitted to the 2 spinal injury units between March 2009 and March 2010 were screened for inclusion. Patients were eligible for inclusion if they had a complete SCI (AIS A) at the neurological level of C5 or lower, or if they had an incomplete SCI (AIS B, C, or D) at any neurological level, provided they were receiving their first episode of rehabilitation following recent SCI. Patients were considered for inclusion only if they had attended at least 1 week of inpatient rehabilitation with 5 sessions of physical therapy. Ultimately, 50 patients were recruited.
At the time of each patient's admission to rehabilitation, treating therapists predicted the patient's likely future mobility at 3 months after admission. Three months was selected to coincide with the typical length of inpatient rehabilitation. For the purpose of this study, the commencement of rehabilitation was defined as the day when the patient first attended physical therapy in the gymnasium after injury. This day was a median (interquartile range [IQR]) of 12 days (7–24 days) since injury. All predictions were made once patients had attended at least 5 physical therapy sessions. The 5 sessions were to ensure that therapists had sufficient opportunity to assess and work with the patients before making predictions. Sometimes this timeline was not achieved because patients developed medical complications necessitating bed rest. Overall, predictions were made a median (IQR) of 45 days (31–73 days) after injury and 25 days (14–41 days) after commencement of rehabilitation.
Therapists used the following standardized assessment scales of mobility to record their predictions: the bed transfer item of the Functional Independence Measure (FIM; 1 item scored on a 7-point scale),15 the Five Additional Mobility and Locomotor (5AML) items (5 items each scored on a 7-point scale),16,17 and the Walking Index for Spinal Cord Injury (WISCI; 1 item scored on a 21-point scale).18–21 These 7 items captured patients' abilities to move about the bed, transfer onto a bed, transfer from the floor into a wheelchair, push a wheelchair on level surfaces, push a wheelchair on a ramp, push a wheelchair over a curb, and walk. The scoring of each item was based on either level of independence, time to complete the task, or the complexity of the skill. The FIM, WISCI, and 5AML all have excellent interrater reliability with the lowest reliability being for the 5AML (weighted kappa ranging from 0.82 to 0.96) and the highest reliability being for the WISCI (100% accordance between raters).16–19 At the end of the study, the therapists were asked open-ended questions designed to gain insights into the most important factors they relied upon to make their predictions.
Five physical therapists made the predictions. They had a median (IQR) of 5 years' (5–10 years) experience in SCI. The study received ethical approval from the Royal North Shore Hospital and the Prince of Wales Hospital. Informed consent was obtained from all patients and physical therapists involved in predicting outcomes. The institutional and governmental regulations concerning the ethical use of human volunteers were followed during the course of this research.
Three months after the commencement of rehabilitation, 1 of 2 independent assessors reviewed patients' mobility, using the same standardized outcome assessment scales upon which therapists' predictions were made. The assessors were blinded to therapists' predictions and not involved in the patients' rehabilitation. One assessor had 6 months and the other assessor had 3 years experience in SCI. Both were trained in all outcome measures.
The amount and type of therapy were slightly different across the 2 spinal injury units, but patients received some or all of the following each week: 1 hour of one-on-one therapy with a physical therapist 5 times a week, 1 hour of group exercise incorporating stretching, strengthening, and wheelchair skills practice 3 times a week, 1 hour of hydrotherapy once a week, 1 hour of wheelchair skills training once a week, and 1 hour of strengthening and stretching exercises with a physical therapy assistant 3 times a week (this does not include time devoted to equipment prescription, goal planning meetings, occupational therapy, recreational therapy, or community outings). All therapy was provided as part of the Australian government–funded health care system, in which all patients with a diagnosis of SCI in the state of New South Wales (population of 7 million) are provided with the same care within 1 of 2 designated SCI units (the 2 SCI units involved in this study), regardless of income or cause of injury. In addition, all patients are provided with essential equipment for mobility.
The assessments taken at 3 months were used to determine the accuracy of the therapists' predictions made on admission. Three sets of analyses were performed: 1 primary planned analysis and 2 secondary post hoc exploratory analyses. The planned analysis included all patients for whom therapists predicted mobility. The 2 post hoc exploratory analyses included patients with AIS A/B and AIS C/D lesions. These analyses were included to explore the possibility that therapists were more accurate at predicting mobility in patients with AIS A/B lesions than AIS C/D lesions. Stata software was used for all analyses. Pearson correlations (r) were used to assess the agreement between therapists' predictions on admission and patients' outcomes at 3 months for each of the 7 mobility items. While r values indicate the strength of linear relationships between therapists' predictions and patients' outcomes, they do not detect systematic differences. Therefore, cumulative percentage agreements for given levels of discrepancies were also calculated as an indicator of accuracy of predictions. For all analyses, data for items were removed if patients had already attained the highest level of mobility on admission to rehabilitation, because there was no possibility of a difference between prediction and outcome. This was the case for some of the patients with less disability on tasks such as wheelchair pushing on level surfaces, mobilizing in bed, and transferring. In addition, 2 patients were ambulating and not using a wheelchair on admission. Therapists did not predict these 2 patients' future ability to mobilize in a wheelchair because these skills were not relevant to them.
Ninety-three patients were admitted to the 2 SCI units and screened for inclusion over the 1-year study period. Thirty-six of the 93 screened patients were excluded because they were younger than 18 years (n = 2), had a diagnosis of deteriorating medical conditions (n = 3), had commenced rehabilitation elsewhere prior to admission to the SCI unit (n = 17), had sustained an AIS A injury at C4 or higher (n = 9), or had an AIS E lesion (n = 5). An additional 7 potentially eligible patients declined to be involved. The characteristics of the remaining 50 patients included in the study are shown in Table 1. The median (IQR) age was 40 years (25–56 years). Forty-five patients were male and 5 were female. Forty-six percent of subjects had AIS A (n = 16) and B (n = 7) lesions, and the remaining 54% had AIS C (n = 11) and D (n = 16) lesions. All but 7 patients remained in the hospital throughout the 3-month period. Of the 7 patients discharged early, 6 were discharged home and 1 was transferred to a rural rehabilitation hospital. These 7 patients were also assessed at 3 months and their data were included in all analyses. Three assessments were delayed because of medical complications, but the patients were assessed as soon as feasible. Assessments were completed within a median (IQR) of 103 days (93–125 days) after injury and 87 days (79–96 days) after commencement of rehabilitation.
The accuracy of therapists' predictions is shown in Table 2. The results from the primary analyses of patients with all types of AIS lesions indicate a strong positive correlation between therapists' predictions on admission and patients' outcomes at 3 months for all mobility items (r = 0.53–0.92). The highest correlation was with the WISCI (r = 0.92) and the lowest correlation was with the push on level surfaces item of the 5AML (r = 0.53). Physical therapists made perfect predictions between 37% and 66% of the time and predicted within 1 point of what patients ultimately attained 60% to 78% of the time (see Table 2). Therapists were most accurate at predicting patients' future ability to negotiate curbs in a wheelchair (within 1 point 78% of the time) and least accurate at predicting patients' future ability to push a wheelchair on a ramp (within 1 point 60% of the time). The results of the secondary analyses are also shown in Table 2. There was little difference between accuracy of therapists' predictions of patients with AIS A/B and AIS C/D lesions, with the exception that therapists were better at predicting WISCI scores for patients with AIS A/B lesions (predictions were within 1 point 91% of the time) than AIS C/D (predictions were within 1 point 63% of the time). Therapists stated that they most commonly relied upon the following factors to make predictions: neurological status, comorbidities, age, weight, and motivation.
The results of this study show that physical therapists can predict future mobility of patients with SCI with a high level of accuracy at the time of admission to rehabilitation. This finding has important implications for the goal-setting process as well as for ensuring appropriate prescription of equipment and discharge planning.6,22
Therapists were most accurate at predicting patients' ability to negotiate curbs in a wheelchair as measured with the 5AML. The scoring of the 5AML takes into account patients' ability to negotiate small and large curbs with and without assistance or aids. Therapists may have been best at predicting future ability to negotiate curbs because ability to perform this skill is largely determined by neurological status and only slightly influenced by other factors. Physical therapists were not as accurate at predicting patients' ability to push a wheelchair on a ramp using the 5AML. The scoring of the 5AML takes into account patients' ability to push up and down steep and gentle ramps with and without assistance after taking into account time. The most likely explanation for this finding is that this skill is not strongly correlated with neurological status and instead is largely determined by fitness, which is difficult to predict.
It is possible that patients' future mobility was limited or enhanced by therapists' predictions so that the predictions the therapists made became self-fulfilling. However, for the predictions to be self-fulfilling, it would require the therapists to remember what they had predicted initially, and then deliberately either advance or hinder a patient's progress over a 3-month period. It is unlikely this happened. In addition, at least 40% of the predictions were made at the acute SCI unit just prior to patients' transfer to the rehabilitation facility. Therapists in the rehabilitation facility were not aware of what the therapists from the acute facility had predicted. Similarly, in the second SCI unit, therapists predicting patients' future mobility were often not the patients' sole therapists for the entire 3-month rehabilitation period.
The secondary post hoc analyses were included to explore the possibility that therapists are more accurate at predicting mobility in patients with AIS A/B lesions than AIS C/D lesions; this was not the case. There was little difference between therapists' accuracy at predicting patients with AIS A/B lesions and patients with AIS C/D lesions (as reflected by the overlap in all confidence intervals for the correlation values for the 2 subgroups). However, these results need to be interpreted with caution because of the small sample size and the associated imprecision of estimates. Not surprisingly, the greatest difference between therapists' ability to predict mobility in the 2 subgroups was for the WISCI item. Therapists had difficulty accurately predicting future WISCI scores for those with AIS C/D lesions. For example, only 67% of predictions were within 2 points. A 2-point difference may sometimes imply an important functional difference. However, the WISCI is scored on a 21-point scale, while the FIM and 5AML are scored on 7-point scales. Therefore, the apparent differences between abilities to predict across the 7 mobility items may solely reflect the larger range of scale values for the WISCI assessment.
The findings of the study are important because they indicate how much confidence patients, funders, policy makers, and administrators can have in therapists' ability to predict patients' future mobility. All these people make decisions early after injury about different aspects of patient care. These decisions often hinge on likely future mobility. For example, timely discharge from SCI units is dependent on the appropriate prescription of equipment, which must commence as soon as possible after injury. Yet, equipment prescription relies on accurate predictions of mobility. In the same way, planning for appropriate home modifications requires some estimation of future mobility. Resources can be wasted modifying a home for a wheelchair if the patient is able to walk by the time of discharge. The findings from this study can give those involved in these types of decisions a greater degree of confidence in therapists' judgments about patients' future mobility.
Currently, it is not clear what makes a therapist a good predictor and whether it is possible to train therapists and improve their skills in this area. Presumably, the single biggest factor that determines the skill of a therapist to predict future mobility is their clinical experience. Therapists who have treated a large numbers of patients with SCI are likely to develop good skills in predicting future mobility. In the same way, it is not clear what factors therapists predominantly rely upon when making predictions. When therapists were asked to articulate the factors that they most commonly relied upon to make predictions, they invariably stated neurological status, comorbidities, age, motivation, strength, and weight. We know that all these variables are strong predictors of outcome,7,8,22–24 and although therapists stated that they considered these variables when making their predictions, we have no way of knowing whether they actually did. It is highly likely that therapists' predictions were also based on other factors that could be gauged only through spending time with a patient. These may include factors such as the attitude of the patient, as well as his or her social support and progress to date.12 We enabled therapists to gather as much information about the patient as they wanted before making predictions by stipulating that the predictions were not made until the therapist had spent at least 5 sessions with the patient. It would be worthwhile in future studies to compare the ability of therapists to predict future mobility with predictions made by independent therapists who have access to case records but who are not familiar with the patients. This comparison would provide insight into what factors therapists use to make predictions and how strongly therapists are influenced by the general presentation of a patient.
Physical therapists can predict future mobility for people with SCI with a high level of accuracy. This skill is important because, beyond facilitating appropriate goal planning for mobility goals to be achieved in therapy, accurate predictions have implications for effective discharge planning and equipment prescription. These findings are important because they indicate that patients, funders, policy makers, and administrators can have confidence in the ability of physical therapists to accurately predict patients' future mobility.
We thanks the patients and physiotherapy staff from the 2 Sydney spinal injury units who participated.
1. Donnelly C, Eng J, Hall J, et al. Client-centred assessment and the identification of meaningful treatment goals for individuals with a spinal cord injury
. Spinal Cord. 2004;42(5):302–307.
2. Schut H, Stam H. Goals in rehabilitation
teamwork. Disabil Rehabil. 1994;16:223–226.
3. Bovend' Eerdt TJ, Botell RE, Wade DT. Writing SMART rehabilitation
goals and achieving goal attainment scaling: a practical guide. Clin Rehabil. 2009;23(4):352–361.
4. Schönherr M, Groothoff J, Mulder G, Eisma W. Prediction
of functional outcome
after spinal cord injury
: a task for the rehabilitation
team and the patient. Spinal Cord. 2000;38:185–191.
5. Burns S, Golding D, Rolle W, Graziani V, Ditunno J. Jr. Recovery of ambulation in motor-incomplete tetraplegia. Arch Phys Med Rehabil. 1997;78:1169–1172.
6. Lazar L, Yarkony G, Ortolano D, et al. Prediction
of functional outcome
by motor capability after spinal cord injury
. Arch Phys Med Rehabil. 1989;70:819–822.
7. Waters R, Adkins R, Yakura J, Sie I. Motor and sensory recovery following complete tetraplegia. Arch Phys Med Rehabil. 1993;74:242–247.
8. Waters R, Adkins R, Yakura J, Vigil D. Prediction
of ambulatory performance based on motor scores derived from standards of the American Spinal Injury Association. Arch Phys Med Rehabil. 1994;75:756–760.
9. Waters R, Yakura J, Adkins R, Sie I. Recovery following complete paraplegia. Arch Phys Med Rehabil. 1992;73:784–789.
10. Welch R, Lobley S, O'Sullivan S, Freed M. Functional independence in quadriplegia: critical levels. Arch Phys Med Rehabil. 1986;67:235–240.
11. Woolsey R. Rehabilitation outcome
following spinal cord injury
. Arch Neurol. 1985;42:116–119.
12. Yarkony G, Roth E, Heinemann A, Wu Y, Katz R, Lovell L. Benefits of rehabilitation
for traumatic spinal cord injury
. Multivariate analysis in 711 patients. Arch Neurol. 1987;44:93–96.
13. Saboe L, Darrah J, Pain K, Guthrie J. Early predictors of functional independence 2 years after spinal cord injury
. Arch Phys Med Rehabil. 1997;78:644–650.
14. Hillier S, Fisher P, Stiller K. The timing and achievement of mobility skills during SCI rehabilitation
. Spinal Cord. 2011;49:416–420.
15. Keith R, Granger C, Hamilton B, Sherwin F. The Functional Independence Measure: a new tool for rehabilitation
. In: Eisenberg M, Gresiak R, eds. Advances in Clinical Rehabilitation
. Vol 2. New York: Springer; 1987:6–18.
16. Harvey L, Batty J, Fahey A. Reliability of a tool for assessing mobility in wheelchair-dependent paraplegics. Spinal Cord. 1998;36:427–431.
17. Middleton J, Harvey L, Batty J, Cameron I, Quirk R, Winstanley J. Five Additional Mobility and Locomotor items to improve responsiveness of the FIM in wheelchair-dependent individuals with spinal cord injury
. Spinal Cord. 2006;44:495–504.
18. Ditunno P, Dittuno J Jr. Walking Index for Spinal Cord Injury
(WISCI II): scale revision. Spinal Cord. 2001;39:654–656.
19. Ditunno J Jr, Ditunno P, Graziani V, et al. Walking Index for Spinal Cord Injury
(WISCI): an international multicenter validity and reliability study. Spinal Cord. 2000;38:234–243.
20. Morganti B, Scivoletto G, Ditunno P, Ditunno JF, Molinari M. Walking Index for Spinal Cord Injury
(WISCI): criterion validation. Spinal Cord. 2005;43(1):27–33.
21. Ditunno J Jr, Burns A, Marino R. Neurological and functional capacity outcome
measures: essential to spinal cord injury
clinical trials. J Rehabil Res Dev. 2005;42:35–42.
22. Schönherr M, Mulder G. Functional outcome
of patients with spinal cord injury
: rehabilitation outcome
study. Clin Rehabil. 1999;13:457–463.
23. Yarkony G, Roth E, Heinemann A, Lovell L. Spinal cord injury rehabilitation outcome
: the impact of age. J Clin Epidemiol. 1988;41:173–177.
24. DeVivo M, Kartus P, Rutt R, Stover S, Fine P. The influence of age at time of spinal cord injury
on rehabilitation outcome
. Arch Neurol. 1990;47:687–691.