- Email: [email protected]
Quantitative assessment of cervical spondylotic my elopathy by a simple walking test
Quantitative assessment of cervical spondylotic myelopathy by a simple walking test Anoushka Singh, H Alan Crockard
Summary Background We developed a 30 m walking test as a quantifiable measure of severity of cervical spondylotic myelopathy (CSM), which will be of use in determining the effects of decompressive surgical treatment. Methods Preoperative measurements were made in 41 patients with CSM of 30 m walking times, number of steps taken over this distance, myelopathy disability index (MDI), and Nurick scores. The walking factors were compared with a similar number of age-matched and sex-matched controls. The individuals in the study were patients with CSM and no other relevant pathology consecutively referred for decompressive surgery to the National Hospital for Neurology and Neurosurgery. Findings Both walking time and the number of steps taken were significantly worse in pre-operative patients than in controls. The walking data were highly reproducible over three trials. Postoperatively, there was a significant improvement in walking time (p=0·0018) and number of steps taken (p=5·87⫻10⫺6). Only two of 41 patients were worse postoperatively. There was also a significant improvement in MDI (two-tailed Wilcoxon, related samples; p<0·0001) and Nurick scores (two-tailed Wilcoxon p<0·0001) postoperatively. The preoperative and postoperative walking scores were significantly and equally correlated with the MDI and Nurick scores. Interpretation Timed walks are an easily performed, quantitative, and valid means of assessing CSM and the effects of surgery. Lancet 1999; 354: 370–73
Department of Surgical Neurology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK (H A Crockard FRCS, A Singh MSc ) Correspondence to: Mr Alan Crockard (e-mail: A [email protected])
370
Introduction Cervical spondylotic myelopathy (CSM) is a common cause of serious morbidity in middle-aged and elderly populations,1 classically presenting with progressive spastic quadroparesis, sensory loss at or below the neck, and urinary incontinence.2 The accepted treatment for the underlying cervical-canal stenosis is some form of surgical decompression or stabilisation.3–9 However, a major difficulty with management of the condition is that its often slow and insidious course, coupled with the potential hazards of surgery, can result in uncertainty about the best operative procedure and the best stage in the disease at which to undertake the procedure. Current presurgical assessment generally involves subjective and non-standardised assessment of patients, relying on specific symptoms, such as difficulty with gait or urinary problems, together with findings on clinical examination and radiological imaging.10 However, different clinicians appear to vary greatly in their selection practices for spondylotic myelopathy surgery. It is possible that in some patients surgery is done unnecessarily, whereas in others concerns about the small but important risk of a surgical disaster, coupled with uncertainty about the benefits, means that surgery is delayed until irreversible damage has already occurred. Such difficulties are widespread in neurological and neurosurgical practice and are now being addressed by the use of more quantitative means of assessment. Ratings scales have been used to score or categorise patients quantitatively and reproducibly, thereby allowing clinicians both to conduct scientific evaluations of management and to draw on the experiences of previous evaluation studies in the management of their own patients. A few scales have already been developed to quantify neurological impairment (such as weakness, sensory impairment, and spasticity), including the ASIA/IMSOP scale11–13 for spinal cord injury and the Ranawat Scale14 for spinal cord compression. However, these and similar measures are poorly quantitative (being ordinal rather than interval) with very few and largely arbitrary categories. The sensitivity to change is likely to be poor because one category covers a huge range of actual severity. Because the information provided from a detailed clinical examination is so much greater, the scales are likely to be more applicable to statistical grouping than the management of individual patients. For example, the Ranawat scale simply distinguishes subjective weakness and tingling from objective weakness and clear signs of cord compression; such a distinction is so obvious in an individual patient as to be of limited use. Functional scales scoring disability (eg, limitation of mobility, inability to feed and dress oneself) are perhaps somewhat more helpful in quantifying and rationalising factors important to the patient that might otherwise receive less attention. Nurick first developed a functional scale for CSM, which was based solely on categories of walking difficulty.15 Other measures include the Japanese Orthopaedic Association scale,16 which measures a mixture of impairment and disability but has limited application,
THE LANCET • Vol 354 • July 31, 1999
Figure 1: Walking time and number of steps taken in normal controls and patients preoperatively and postoperatively The bars indicate SE.
and the purely functional and relatively easy to use European Myelopathy score.17 The Myelopathy Disability Index (MDI) scale18 has been validated specifically for rheumatoid cervical myelopathy and is an adaptation of the Health Assessment Questionnaire;18 the latter is a list of 20 self-assessment questions taken from the very widely used Bartel Index and the Impairment of Activities of Daily Living scale. A criticism often levelled at functional scales is that they lack the objectivity of a clinical neurological examination. For example, a scale based on patients’ reporting of their own symptoms might be coloured by their general psyche and by how urgently they perceive the need for treatment. It would be advantageous therefore to achieve further rationalisation of severity beyond these scales of functional impairment. The present study explores the use of a timed walking test (previously used for other purposes)19,20 as a more objective, quantitative, and easy-to-use measure to quantify the severity of CSM preoperatively. Clinicians already tend to question patients on walking, as a rough measure of the functional severity of spastic paraparesis and walking is of course an important aspect of the neurological examination that sensitively reflects long-tract pathology.
Methods Patients 41 (26 male, 15 female) patients with CSM consecutively referred to our neurosurgical unit for consideration for decompressive surgery were studied. All patients had the diagnosis corroborated by magnetic resonance imaging and none had previously undergone neck surgery or had any other pathology that might have resulted in functional impairment. Approval by the hospital ethics committee and informed consent was obtained from each patient. All patients were given an information sheet with a brief outline of the study. The patients were under the care of six different neurosurgeons. The measurement team did not influence the surgical approach. This group was compared with an equal number of controls (healthy individuals without myelopathy or other disability) and was matched for sex and, as far as possible, for age.
Procedures and data analysis Patients and controls walked on a smooth flat surface over a ward corridor for a measured 30 m distance (15 m there and back with one turn). The time taken was recorded with a stopwatch and the
THE LANCET • Vol 354 • July 31, 1999
Figure 2: A: walking time and age for controls; B: walking time and age for myelopathic patients number of steps taken was counted. The patients were requested to walk at maximum comfortable speed (with any normally used walking aids). Each trial was done three times. Each patient was assessed both preoperatively and 2 months postoperatively. MDI scores and Nurick scores were also found (by direct questioning rather than by questionnaire [both methods have been validated] on the same day as each walking assessment. Data were analysed statistically with SPSS version 7·5).
Results The mean (SD) ages of male and female patients and controls were 59·4 (11·9), 62·9 (18·2), 59·8 (12·5), and 61·9 (17·7) years. Thus a good match was achieved, justifying further comparisons between patients and controls. The median length of hospital admission for surgery was 7 days and there were no perioperative deaths. To assess the reproducibility of the walking-time data, analysis of variance was done as a two-factor repeatedmeasures test comparing the values for the three separate trials done by each individual in each of the preoperative and postoperative conditions. By far the greater variability existed between preoperative and postoperative times (p=0·001) rather than between trials (p=0·995 [interaction; p=0·998]), indicating that the walking data were highly reproducible and any major change in value was likely to represent a real change in the individual’s functional ability. Thus only the mean of the three values was used in further analysis. The source of variation for number of steps taken again lay mainly between
371
Walking time
MDI
Nurick
Preoperative Walking time MDI Nurick
·· ·· ··
0·653 ·· ··
0·61 0·66 ··
Postoperative Walking time MDI Nurick
·· ·· ··
0·57 ·· ··
0·69 0·65 ··
Correlation between three measures of myelopathy both preoperatively and postoperatively (two-tailed Spearman’s test, p<0·001 in each case)
Figure 3: Preoperative and postoperative MDI scores
preoperative and postoperative values (p=0·003) rather than between the three trials (p=0·981) and there was no interaction (p=0·959). The mean preoperative walking time over the three trials and in the controls was 24·3 (SE 0·8). Both control and patient walking data were approximately normally distributed (figure). The mean number of steps taken to walk 30 m was 46·9 (1·2) steps (figure 1). The mean preoperative walking time over all the people in each group was 85·4 (11·2) s (SE for 40 people). This was significantly worse than that for healthy controls (twotailed test [unpaired, unequal variances], p=2·4⫻10–6). The mean postoperative patient walking time was 64·7 (8·4) s (SE for 39 people) (figure 1). There was a significant improvement in the patients after operation (two-tailed paired t test, p=0·0018). One patient was unable to walk this distance preoperatively but able postoperatively. Two patients were able to walk preoperatively but not postoperatively. The mean number of steps taken to walk 30 m in each group was 74·8 (5·3) steps preoperatively, 63·5 (4·2) steps postoperatively and 46·9 (1·2) s steps for controls (figure 1). The preoperative number of steps was again significantly worse than that for control (two tailed t test, p=5·4⫻10–6) and there was again a highly significant improvement in patients after operation (two-tailed paired t test, p=5·87⫻10–6). The mean difference between preoperative and postoperative walking times was 21·7 (19·5) s and the mean difference between preoperative and postoperative steps taken was 12·5 (7·7) steps. The differences between individual patients and controls are not relevant. The mean difference for walking time between patients preoperatively and paired (matched) controls was 61·3 (22·0). The mean difference for the number of steps between patients preoperatively and paired (matched) controls was 28·1 (10·4).
It would be expected that age as well as the severity of disease might have an effect on the walking factors. In the age-matched controls, this relation was not found to be particularly strong for number of steps taken (correlation coefficient=0·42, not illustrated) or for walking time (correlation coefficient=0·34, figure 2, A). The corresponding coefficients for preoperative patients (figure 2, B) are lower than for controls, as expected because there is also the influence of severity of disease (number of steps; correlation coefficient=0·32 time; correlation coefficient=0·26). The MDI and Nurick scale measures were not normally distributed and so medians and quartiles were used. MDI scores had a median of 14 preoperatively and 5 postoperatively (figure 3). Before operation, out of 41 patients, none was normal (ie, zero score) and one patient had 30, the worst possible score. Postoperatively, seven patients improved to zero score. However, some of these had only minimum deficit originally; four of these postoperatively normal patients had preoperative scores of two, whereas the other three had preoperative scores of 12, 13, and 14. Two patients had worse scores postoperatively. These were the same two patients who became unable to walk postoperatively on the walking tests. They were both male and not of great age. Although their scores on all three scales preoperatively were not at any extreme, their MDI scores (reflecting overall disability) were worse than their walking times, suggesting to us that some other factors not related to walking could possibly have predicted a bad outcome. There was overall a clearly significant improvement in MDI scores after surgery (two-tailed Wilcoxon test, related samples; p<0·0001). Rating on the Nurick Scale also showed a significant improvement after surgery (two-tailed Wilcoxon test, p<0·0001, figure 4). The preoperative and postoperative walking scores were compared with the respective preoperative and postoperative MDI and Nurick scores (table). The three tests were all significantly and fairly equally correlated, but not perfectly so.
Discussion
Figure 4: Preoperative and postoperative Nurick scores
372
This study shows that the easily performed walking test may be a suitable measure of severity of CSM. The test is reproducible and reliable,21 because there was a very low variability between trials. The distance of 30 m is long enough to measure time fairly accurately and the nature of such an objective quantitative measure guarantees reliability. The walking test is also a truly quantitative measure—ie, a ratio measure in which the numbers are real values rather than abstractions.22 True correlations between factors, comparisons with control data, and correction for normal control variability are therefore possible. The ability to correct for the effect of age on normal walking time could
THE LANCET • Vol 354 • July 31, 1999
allow extra sensitivity for low levels of disability that other scales would simply categorise as normal. A high degree of variability between patients is desirable for a scale that will be sensitive to perioperative change.21 The ratio of preoperative SD/mean (ie, coefficient of variation) for walking time (0·88) was greater than for steps (0·44) or for the MDI (0·48), suggesting to us that walking time is potentially the best measure. However, sensitivity is more than just variability; the variation must reflect real differences in disability. Thus comparison should also be made with the coefficients of variation of the controls that are all normal. The values were 0·21 for walking time and 0·15 for number of steps. Thus walking time still shows better variability even when accounting for the greater normal walking time variability. In addition, it shows less variability with age, a potential confounding factor. The overall sensitivity to change is well demonstrated for the walking-time data, which show an improvement after surgery and goes beyond the patient’s self-assessment of “being able to walk outdoors” or “able to climb five steps”, as is tested in the MDI. Furthermore, the more quantitative nature of walking data allows more detailed comparison with control data (on the MDI all controls are simply scored zero). It becomes clear that, despite the significant improvement after the operation, walking did not reach normal levels. Patients both preoperatively and postoperatively each took on average about 1 s for a step. By contrast, normal age-matched individuals took steps about twice the speed. It was this extra increase in step speed that seemed to characterise normal walking. Finally, the walking test shows significant correlation with previously used scales, indicating validity and relevance,23 and yet is not so strongly correlated to suggest that the test has no worth. Relevance is an important factor that will give clinicians confidence in the use of an assessment scale. Thus timed walks may be an ideal means of bridging the gap between disability and impairment rating scales and, because clinicians are strongly swayed by their own clinical experience, a measure that may be readily accepted and used by practising clinicians. Because walking time (incorporating maximum speed and endurance) is both an important aspect of patients’ function and a sensitive measure of long-tract dysfunction as part of neurological examination, the test appears to us to be empirically highly relevant to CSM severity. Correlation between walking and the Nurick scale (a functional rating of just walking) was no stronger than with the MDI (which measures overall function). Possibly, then, the walking test can be used as a valid stand-alone marker for overall myelopathic severity. However, one must be alert to the fact that walking can be impaired for other reasons. The similar correlations between the three scales does not imply that the walking test is redundant. The MDI, Nurick, and other scales are fundamentaly different from the walking test because they are determined by questioning the patient. Such methods will suffer in reliability compared with direct recordings of patient activity. The MDI and Nurick scales are essentially ordinal measures, where qualitative measures are assigned numerical values even though an effort has been made to attribute some level of quantitativity to such scales (making them interval scales).
THE LANCET • Vol 354 • July 31, 1999
These preliminary data show that the timed walk may be highly suited to the assessment of severity of CSM. Walking tests may prove useful as a practical aid for clinicians in preoperative assessment and a measurement tool in future trials on CSM operative techniques and management strategies. Contributors Alan Crockard was the consultant in charge of the research project and Anoushka Singh was the research nurse, who designed, collected, and analysed the data. Both investigators wrote the paper.
Acknowledgments We are grateful to William Harkness, Neil Kitchen, James Palmer, Michael Powell, and David Thomas, consultant neurosurgeons at the National Hospital for Neurology and Neurosurgery, London, for allowing us to study their patients.
References 1
2 3
4
5 6
7
8
9
10
11 12 13 14 15 16
17
18
19
20
21
22 23
Lindsay KW, Bone I, Callander R. Localised neurological disease and its management. Spinal cord and roots. In Lindsay KW, Bone I, Callander R. eds. Neurology and neurosurgery illustrated. Edinburgh: Churchill Livingstone, 1991: 373–410. Clarke E, Robinson PK. Cervical myelopathy: a complication of cervical spondylosis. Brain 1956; 79: 483–510. Rogers L. The surgical treatment of cervical spondylotic myelopathy by mobilisation of the cervical cord into an enlarged canal. J Bone Joint Surg Am 1961; 43: 3–6. Rogers L. The surgical treatment cervical spondylotic myelopathy by mobilisation of the cervical cord into an enlarged spinal canal. J Neurosurg 1961; 18: 490–92. Scoville WB. Cervical spondylosis treated by bilateral facetectomy and laminectomy. J Neurosurg 1961; 18: 423–28. Sypert GW. Anterior compression and fusion for cervical myelopathy. In Camins MB, O’Leary PF, eds. Disorders of the cervical spine. Baltimore: Williams and Wilkins, 1992: 407–16. Allen K, Neuropathies caused by bony spurs in the cervical spine with special reference to surgical treatment. J Neurol Neurosurg Psychiatry 1952; 15: 20–23. Bakay L, Cares H, Smith R. Ossification in the region of the posterior longitudinal ligament as a cause of cervical myelopathy. J Neurol Neurosurg Psychiatry 1970; 33: 20–26. Gonzalez-Feria L. The effect of surgical immobilization after laminectomy in the treatment of advanced cases of cervical spondylotic myelopathy. Acta Neurochir 1975; 31: 185–93. Hadley MN, Sonntag VKH. Cervical disk herniations: the anterior operative approach without interbody fusion. In: Menezes AH, Sonntag VKH, eds. Principles of spinal surgery. New York: McGraw-Hill, 1995: 531–38. Donovan WH, Wilkerson MA, Rossi D, et al. A test of ASIA guidelines for classification of spinal cord injuries. J Neurol Rehab 1990; 4: 39–53. Waters RL, Adkins RH, Yakura JS. Definition of complete spinal cord injury. Paraplegia 1991; 29: 573–81. Ditunno JF Jr. New spinal cord injury standards. Paraplegia 1992; 30: 90–91. Ranawat C, O’Leary P, Pellici P, et al. Cervical fusion in rheumatoid arthritis. Bone Joint Surg Am 1979; 61: 1003–10. Nurick S. The pathogenesis of the spinal cord disorder associated with cervical spondylosis. Brain 1972; 95: 87–100. Hirabayashi K, Wantanabe K, Wankano K, et al. Expansive open-door laminoplasty for cervical spinal stenotic myelopathy. Spine 1983; ??: 693–99. Hermann J, Linzbach M, Krzan M, Dvorak J, Bock WJ. The European Myelopathy Score. In: Bauer BL, Brock M, Klinger M, eds. Advances in neurosurgery. Berlin: Springer, 1994; 266–68. Casey ATH, Bland JM, Crockard HA. Development of a functional scoring system for rheumatoid arthritis patients with cervical myelopathy. Ann Rheum Dis 1996; 55: 901–06. Holden MK, Gill KM, Magliozzi MR, et al. Clinical gait assessment in the neurologically impaired: reliability and meaningfulness. Phys Ther 1984; 64: 35–40. Wade DT, Wood VA, Heeler A, et al. Walking after stroke: measurement and recovery over the first three months. Scand J Rehab Med 1987; 19: 25–30. Streiner DI, Norman GR, Measuring change in health measurement scales: a practicel guide to their development and use. Oxford: Oxford Univerity Press, 1996: 163–80. Dun-Rankin P. Scaling methods. New Jersey: Erlbaum Hillsdale; 1983. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; i: 307–80.
373
Summary Background We developed a 30 m walking test as a quantifiable measure of severity of cervical spondylotic myelopathy (CSM), which will be of use in determining the effects of decompressive surgical treatment. Methods Preoperative measurements were made in 41 patients with CSM of 30 m walking times, number of steps taken over this distance, myelopathy disability index (MDI), and Nurick scores. The walking factors were compared with a similar number of age-matched and sex-matched controls. The individuals in the study were patients with CSM and no other relevant pathology consecutively referred for decompressive surgery to the National Hospital for Neurology and Neurosurgery. Findings Both walking time and the number of steps taken were significantly worse in pre-operative patients than in controls. The walking data were highly reproducible over three trials. Postoperatively, there was a significant improvement in walking time (p=0·0018) and number of steps taken (p=5·87⫻10⫺6). Only two of 41 patients were worse postoperatively. There was also a significant improvement in MDI (two-tailed Wilcoxon, related samples; p<0·0001) and Nurick scores (two-tailed Wilcoxon p<0·0001) postoperatively. The preoperative and postoperative walking scores were significantly and equally correlated with the MDI and Nurick scores. Interpretation Timed walks are an easily performed, quantitative, and valid means of assessing CSM and the effects of surgery. Lancet 1999; 354: 370–73
Department of Surgical Neurology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK (H A Crockard FRCS, A Singh MSc ) Correspondence to: Mr Alan Crockard (e-mail: A [email protected])
370
Introduction Cervical spondylotic myelopathy (CSM) is a common cause of serious morbidity in middle-aged and elderly populations,1 classically presenting with progressive spastic quadroparesis, sensory loss at or below the neck, and urinary incontinence.2 The accepted treatment for the underlying cervical-canal stenosis is some form of surgical decompression or stabilisation.3–9 However, a major difficulty with management of the condition is that its often slow and insidious course, coupled with the potential hazards of surgery, can result in uncertainty about the best operative procedure and the best stage in the disease at which to undertake the procedure. Current presurgical assessment generally involves subjective and non-standardised assessment of patients, relying on specific symptoms, such as difficulty with gait or urinary problems, together with findings on clinical examination and radiological imaging.10 However, different clinicians appear to vary greatly in their selection practices for spondylotic myelopathy surgery. It is possible that in some patients surgery is done unnecessarily, whereas in others concerns about the small but important risk of a surgical disaster, coupled with uncertainty about the benefits, means that surgery is delayed until irreversible damage has already occurred. Such difficulties are widespread in neurological and neurosurgical practice and are now being addressed by the use of more quantitative means of assessment. Ratings scales have been used to score or categorise patients quantitatively and reproducibly, thereby allowing clinicians both to conduct scientific evaluations of management and to draw on the experiences of previous evaluation studies in the management of their own patients. A few scales have already been developed to quantify neurological impairment (such as weakness, sensory impairment, and spasticity), including the ASIA/IMSOP scale11–13 for spinal cord injury and the Ranawat Scale14 for spinal cord compression. However, these and similar measures are poorly quantitative (being ordinal rather than interval) with very few and largely arbitrary categories. The sensitivity to change is likely to be poor because one category covers a huge range of actual severity. Because the information provided from a detailed clinical examination is so much greater, the scales are likely to be more applicable to statistical grouping than the management of individual patients. For example, the Ranawat scale simply distinguishes subjective weakness and tingling from objective weakness and clear signs of cord compression; such a distinction is so obvious in an individual patient as to be of limited use. Functional scales scoring disability (eg, limitation of mobility, inability to feed and dress oneself) are perhaps somewhat more helpful in quantifying and rationalising factors important to the patient that might otherwise receive less attention. Nurick first developed a functional scale for CSM, which was based solely on categories of walking difficulty.15 Other measures include the Japanese Orthopaedic Association scale,16 which measures a mixture of impairment and disability but has limited application,
THE LANCET • Vol 354 • July 31, 1999
Figure 1: Walking time and number of steps taken in normal controls and patients preoperatively and postoperatively The bars indicate SE.
and the purely functional and relatively easy to use European Myelopathy score.17 The Myelopathy Disability Index (MDI) scale18 has been validated specifically for rheumatoid cervical myelopathy and is an adaptation of the Health Assessment Questionnaire;18 the latter is a list of 20 self-assessment questions taken from the very widely used Bartel Index and the Impairment of Activities of Daily Living scale. A criticism often levelled at functional scales is that they lack the objectivity of a clinical neurological examination. For example, a scale based on patients’ reporting of their own symptoms might be coloured by their general psyche and by how urgently they perceive the need for treatment. It would be advantageous therefore to achieve further rationalisation of severity beyond these scales of functional impairment. The present study explores the use of a timed walking test (previously used for other purposes)19,20 as a more objective, quantitative, and easy-to-use measure to quantify the severity of CSM preoperatively. Clinicians already tend to question patients on walking, as a rough measure of the functional severity of spastic paraparesis and walking is of course an important aspect of the neurological examination that sensitively reflects long-tract pathology.
Methods Patients 41 (26 male, 15 female) patients with CSM consecutively referred to our neurosurgical unit for consideration for decompressive surgery were studied. All patients had the diagnosis corroborated by magnetic resonance imaging and none had previously undergone neck surgery or had any other pathology that might have resulted in functional impairment. Approval by the hospital ethics committee and informed consent was obtained from each patient. All patients were given an information sheet with a brief outline of the study. The patients were under the care of six different neurosurgeons. The measurement team did not influence the surgical approach. This group was compared with an equal number of controls (healthy individuals without myelopathy or other disability) and was matched for sex and, as far as possible, for age.
Procedures and data analysis Patients and controls walked on a smooth flat surface over a ward corridor for a measured 30 m distance (15 m there and back with one turn). The time taken was recorded with a stopwatch and the
THE LANCET • Vol 354 • July 31, 1999
Figure 2: A: walking time and age for controls; B: walking time and age for myelopathic patients number of steps taken was counted. The patients were requested to walk at maximum comfortable speed (with any normally used walking aids). Each trial was done three times. Each patient was assessed both preoperatively and 2 months postoperatively. MDI scores and Nurick scores were also found (by direct questioning rather than by questionnaire [both methods have been validated] on the same day as each walking assessment. Data were analysed statistically with SPSS version 7·5).
Results The mean (SD) ages of male and female patients and controls were 59·4 (11·9), 62·9 (18·2), 59·8 (12·5), and 61·9 (17·7) years. Thus a good match was achieved, justifying further comparisons between patients and controls. The median length of hospital admission for surgery was 7 days and there were no perioperative deaths. To assess the reproducibility of the walking-time data, analysis of variance was done as a two-factor repeatedmeasures test comparing the values for the three separate trials done by each individual in each of the preoperative and postoperative conditions. By far the greater variability existed between preoperative and postoperative times (p=0·001) rather than between trials (p=0·995 [interaction; p=0·998]), indicating that the walking data were highly reproducible and any major change in value was likely to represent a real change in the individual’s functional ability. Thus only the mean of the three values was used in further analysis. The source of variation for number of steps taken again lay mainly between
371
Walking time
MDI
Nurick
Preoperative Walking time MDI Nurick
·· ·· ··
0·653 ·· ··
0·61 0·66 ··
Postoperative Walking time MDI Nurick
·· ·· ··
0·57 ·· ··
0·69 0·65 ··
Correlation between three measures of myelopathy both preoperatively and postoperatively (two-tailed Spearman’s test, p<0·001 in each case)
Figure 3: Preoperative and postoperative MDI scores
preoperative and postoperative values (p=0·003) rather than between the three trials (p=0·981) and there was no interaction (p=0·959). The mean preoperative walking time over the three trials and in the controls was 24·3 (SE 0·8). Both control and patient walking data were approximately normally distributed (figure). The mean number of steps taken to walk 30 m was 46·9 (1·2) steps (figure 1). The mean preoperative walking time over all the people in each group was 85·4 (11·2) s (SE for 40 people). This was significantly worse than that for healthy controls (twotailed test [unpaired, unequal variances], p=2·4⫻10–6). The mean postoperative patient walking time was 64·7 (8·4) s (SE for 39 people) (figure 1). There was a significant improvement in the patients after operation (two-tailed paired t test, p=0·0018). One patient was unable to walk this distance preoperatively but able postoperatively. Two patients were able to walk preoperatively but not postoperatively. The mean number of steps taken to walk 30 m in each group was 74·8 (5·3) steps preoperatively, 63·5 (4·2) steps postoperatively and 46·9 (1·2) s steps for controls (figure 1). The preoperative number of steps was again significantly worse than that for control (two tailed t test, p=5·4⫻10–6) and there was again a highly significant improvement in patients after operation (two-tailed paired t test, p=5·87⫻10–6). The mean difference between preoperative and postoperative walking times was 21·7 (19·5) s and the mean difference between preoperative and postoperative steps taken was 12·5 (7·7) steps. The differences between individual patients and controls are not relevant. The mean difference for walking time between patients preoperatively and paired (matched) controls was 61·3 (22·0). The mean difference for the number of steps between patients preoperatively and paired (matched) controls was 28·1 (10·4).
It would be expected that age as well as the severity of disease might have an effect on the walking factors. In the age-matched controls, this relation was not found to be particularly strong for number of steps taken (correlation coefficient=0·42, not illustrated) or for walking time (correlation coefficient=0·34, figure 2, A). The corresponding coefficients for preoperative patients (figure 2, B) are lower than for controls, as expected because there is also the influence of severity of disease (number of steps; correlation coefficient=0·32 time; correlation coefficient=0·26). The MDI and Nurick scale measures were not normally distributed and so medians and quartiles were used. MDI scores had a median of 14 preoperatively and 5 postoperatively (figure 3). Before operation, out of 41 patients, none was normal (ie, zero score) and one patient had 30, the worst possible score. Postoperatively, seven patients improved to zero score. However, some of these had only minimum deficit originally; four of these postoperatively normal patients had preoperative scores of two, whereas the other three had preoperative scores of 12, 13, and 14. Two patients had worse scores postoperatively. These were the same two patients who became unable to walk postoperatively on the walking tests. They were both male and not of great age. Although their scores on all three scales preoperatively were not at any extreme, their MDI scores (reflecting overall disability) were worse than their walking times, suggesting to us that some other factors not related to walking could possibly have predicted a bad outcome. There was overall a clearly significant improvement in MDI scores after surgery (two-tailed Wilcoxon test, related samples; p<0·0001). Rating on the Nurick Scale also showed a significant improvement after surgery (two-tailed Wilcoxon test, p<0·0001, figure 4). The preoperative and postoperative walking scores were compared with the respective preoperative and postoperative MDI and Nurick scores (table). The three tests were all significantly and fairly equally correlated, but not perfectly so.
Discussion
Figure 4: Preoperative and postoperative Nurick scores
372
This study shows that the easily performed walking test may be a suitable measure of severity of CSM. The test is reproducible and reliable,21 because there was a very low variability between trials. The distance of 30 m is long enough to measure time fairly accurately and the nature of such an objective quantitative measure guarantees reliability. The walking test is also a truly quantitative measure—ie, a ratio measure in which the numbers are real values rather than abstractions.22 True correlations between factors, comparisons with control data, and correction for normal control variability are therefore possible. The ability to correct for the effect of age on normal walking time could
THE LANCET • Vol 354 • July 31, 1999
allow extra sensitivity for low levels of disability that other scales would simply categorise as normal. A high degree of variability between patients is desirable for a scale that will be sensitive to perioperative change.21 The ratio of preoperative SD/mean (ie, coefficient of variation) for walking time (0·88) was greater than for steps (0·44) or for the MDI (0·48), suggesting to us that walking time is potentially the best measure. However, sensitivity is more than just variability; the variation must reflect real differences in disability. Thus comparison should also be made with the coefficients of variation of the controls that are all normal. The values were 0·21 for walking time and 0·15 for number of steps. Thus walking time still shows better variability even when accounting for the greater normal walking time variability. In addition, it shows less variability with age, a potential confounding factor. The overall sensitivity to change is well demonstrated for the walking-time data, which show an improvement after surgery and goes beyond the patient’s self-assessment of “being able to walk outdoors” or “able to climb five steps”, as is tested in the MDI. Furthermore, the more quantitative nature of walking data allows more detailed comparison with control data (on the MDI all controls are simply scored zero). It becomes clear that, despite the significant improvement after the operation, walking did not reach normal levels. Patients both preoperatively and postoperatively each took on average about 1 s for a step. By contrast, normal age-matched individuals took steps about twice the speed. It was this extra increase in step speed that seemed to characterise normal walking. Finally, the walking test shows significant correlation with previously used scales, indicating validity and relevance,23 and yet is not so strongly correlated to suggest that the test has no worth. Relevance is an important factor that will give clinicians confidence in the use of an assessment scale. Thus timed walks may be an ideal means of bridging the gap between disability and impairment rating scales and, because clinicians are strongly swayed by their own clinical experience, a measure that may be readily accepted and used by practising clinicians. Because walking time (incorporating maximum speed and endurance) is both an important aspect of patients’ function and a sensitive measure of long-tract dysfunction as part of neurological examination, the test appears to us to be empirically highly relevant to CSM severity. Correlation between walking and the Nurick scale (a functional rating of just walking) was no stronger than with the MDI (which measures overall function). Possibly, then, the walking test can be used as a valid stand-alone marker for overall myelopathic severity. However, one must be alert to the fact that walking can be impaired for other reasons. The similar correlations between the three scales does not imply that the walking test is redundant. The MDI, Nurick, and other scales are fundamentaly different from the walking test because they are determined by questioning the patient. Such methods will suffer in reliability compared with direct recordings of patient activity. The MDI and Nurick scales are essentially ordinal measures, where qualitative measures are assigned numerical values even though an effort has been made to attribute some level of quantitativity to such scales (making them interval scales).
THE LANCET • Vol 354 • July 31, 1999
These preliminary data show that the timed walk may be highly suited to the assessment of severity of CSM. Walking tests may prove useful as a practical aid for clinicians in preoperative assessment and a measurement tool in future trials on CSM operative techniques and management strategies. Contributors Alan Crockard was the consultant in charge of the research project and Anoushka Singh was the research nurse, who designed, collected, and analysed the data. Both investigators wrote the paper.
Acknowledgments We are grateful to William Harkness, Neil Kitchen, James Palmer, Michael Powell, and David Thomas, consultant neurosurgeons at the National Hospital for Neurology and Neurosurgery, London, for allowing us to study their patients.
References 1
2 3
4
5 6
7
8
9
10
11 12 13 14 15 16
17
18
19
20
21
22 23
Lindsay KW, Bone I, Callander R. Localised neurological disease and its management. Spinal cord and roots. In Lindsay KW, Bone I, Callander R. eds. Neurology and neurosurgery illustrated. Edinburgh: Churchill Livingstone, 1991: 373–410. Clarke E, Robinson PK. Cervical myelopathy: a complication of cervical spondylosis. Brain 1956; 79: 483–510. Rogers L. The surgical treatment of cervical spondylotic myelopathy by mobilisation of the cervical cord into an enlarged canal. J Bone Joint Surg Am 1961; 43: 3–6. Rogers L. The surgical treatment cervical spondylotic myelopathy by mobilisation of the cervical cord into an enlarged spinal canal. J Neurosurg 1961; 18: 490–92. Scoville WB. Cervical spondylosis treated by bilateral facetectomy and laminectomy. J Neurosurg 1961; 18: 423–28. Sypert GW. Anterior compression and fusion for cervical myelopathy. In Camins MB, O’Leary PF, eds. Disorders of the cervical spine. Baltimore: Williams and Wilkins, 1992: 407–16. Allen K, Neuropathies caused by bony spurs in the cervical spine with special reference to surgical treatment. J Neurol Neurosurg Psychiatry 1952; 15: 20–23. Bakay L, Cares H, Smith R. Ossification in the region of the posterior longitudinal ligament as a cause of cervical myelopathy. J Neurol Neurosurg Psychiatry 1970; 33: 20–26. Gonzalez-Feria L. The effect of surgical immobilization after laminectomy in the treatment of advanced cases of cervical spondylotic myelopathy. Acta Neurochir 1975; 31: 185–93. Hadley MN, Sonntag VKH. Cervical disk herniations: the anterior operative approach without interbody fusion. In: Menezes AH, Sonntag VKH, eds. Principles of spinal surgery. New York: McGraw-Hill, 1995: 531–38. Donovan WH, Wilkerson MA, Rossi D, et al. A test of ASIA guidelines for classification of spinal cord injuries. J Neurol Rehab 1990; 4: 39–53. Waters RL, Adkins RH, Yakura JS. Definition of complete spinal cord injury. Paraplegia 1991; 29: 573–81. Ditunno JF Jr. New spinal cord injury standards. Paraplegia 1992; 30: 90–91. Ranawat C, O’Leary P, Pellici P, et al. Cervical fusion in rheumatoid arthritis. Bone Joint Surg Am 1979; 61: 1003–10. Nurick S. The pathogenesis of the spinal cord disorder associated with cervical spondylosis. Brain 1972; 95: 87–100. Hirabayashi K, Wantanabe K, Wankano K, et al. Expansive open-door laminoplasty for cervical spinal stenotic myelopathy. Spine 1983; ??: 693–99. Hermann J, Linzbach M, Krzan M, Dvorak J, Bock WJ. The European Myelopathy Score. In: Bauer BL, Brock M, Klinger M, eds. Advances in neurosurgery. Berlin: Springer, 1994; 266–68. Casey ATH, Bland JM, Crockard HA. Development of a functional scoring system for rheumatoid arthritis patients with cervical myelopathy. Ann Rheum Dis 1996; 55: 901–06. Holden MK, Gill KM, Magliozzi MR, et al. Clinical gait assessment in the neurologically impaired: reliability and meaningfulness. Phys Ther 1984; 64: 35–40. Wade DT, Wood VA, Heeler A, et al. Walking after stroke: measurement and recovery over the first three months. Scand J Rehab Med 1987; 19: 25–30. Streiner DI, Norman GR, Measuring change in health measurement scales: a practicel guide to their development and use. Oxford: Oxford Univerity Press, 1996: 163–80. Dun-Rankin P. Scaling methods. New Jersey: Erlbaum Hillsdale; 1983. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; i: 307–80.
373