KELVIN JORDAN, KIRSTIE L. HAYWOOD, KRYSIA DZIEDZIC, ANDREW M. GARRATT, PETER W. JONES, BIE NIO ONG, and PETER T. DAWES

ABSTRACT.

Objective. To assess the repeatability and validity of the electromagnetic 3-dimensional tracking system, Fastrak, in measuring cervical spine and shoulder movement in patients with ankylosing spondylitis (AS).

Methods. Fifty patients with AS had their cervical spine and shoulder movements measured on up to 3 occasions with the Fastrak. Patients also completed disease-specific and generic patient assessed health instruments, and their spinal mobility was assessed by tape measure methods. Repeatability over 2 weeks was assessed using intraclass correlation coefficients (ICC). Fastrak measurements were compared between patients with different self-ratings of AS related health. Comparisons between the Fastrak measurements and patient assessed health instruments and tape measurements were made using Spearman correlations and multilevel modeling.

Results. Patients with AS tended to be limited in both cervical spine and shoulder movements. ICC were all > 0.80 (except shoulder extension, 0.75), indicating substantial reliability. Fastrak was able to differentiate between patients with a high self-rating of AS related health and those with a poorer rating. Cervical spine flexion and shoulder flexion and abduction were most strongly related to the patient assessed health instruments, although the shoulder movements had limited relationships with the tape measurements of spinal mobility.

Conclusion. The Fastrak appears to be reliable and valid in an AS population. Shoulder movements tended to have a stronger relationship with the patient assessed health instruments than cervical spine movements. Shoulder movement may be more related to everyday function measured by these instruments, which indicates the importance of this joint in assessment of AS. (J Rheumatol 2004;31:2207-15)

Key Indexing Terms:

CERVICAL SPINE
SHOULDER
ANKYLOSING SPONDYLITIS
RANGE OF MOTION
FASTRAK

Supported by a NHS Executive (West Midlands) New Blood Research Training Fellowship, the Arthritis Research Campaign, and the Staffordshire Rheumatology Centre.

K. Jordan, PhD, Research Fellow in Biostatistics, Keele University; K.L. Haywood, DPhil, Research Officer, University of Oxford; K. Dziedzic, PhD, Arthritis Research Campaign, Senior Lecturer in Physiotherapy, Keele University; A.M. Garratt, PhD, University Research Lecturer, University of Oxford; P.W. Jones, PhD, Professor of Statistics, Keele University; B.N. Ong, PhD, Professor of Health Services Research, Keele University; P.T. Dawes, MB, ChB, FRCP(UK), Consultant Rheumatologist, Staffordshire Rheumatology Centre.

Address reprint requests to Dr. K. Jordan, Primary Care Sciences Research Centre, Keele University, Keele, Staffordshire ST5 5BG, UK. E-mail: k.p.jordan@cphc.keele.ac.uk

Submitted January 16, 2004; revision accepted May 8, 2004.

Assessment of range of motion (ROM) is used to aid diagnosis of ankylosing spondylitis (AS) and contributes to assessment of change over time in those with the disease¹. However, conventional measurement tools such as the tape measure or the inclinometer can only measure the maximum ROM in one plane (the primary plane) of movement. They cannot detail the full 3-dimensional pattern of movement, including movement in different planes, and the velocity of that movement. Thus the full picture of dynamic movement is not obtained as its speed and associated movement outside the primary plane cannot be recorded. This may be particularly important for complex joints, such as the shoulder, where movement is 3-dimensional, and when measuring functional activity.

The Fastrak (Polhemus Incorporated, Colchester, VT, USA) is an electromagnetic 3-dimensional tracking system. The position and orientation of up to 4 small remote sensors placed on appropriate parts of the body are computed relative to the source transmitter. Data from each sensor, therefore, can be measured in 3 planes (e.g., sagittal, frontal, and transverse) of joint motion. For example, the amount of lateral flexion and rotation in the primary movement of cervical spine flexion can be recorded, as can the amount of trunk movement in these 3 planes. Software to read this 3-dimensional movement and to derive velocity and acceleration has been designed in North Staffordshire, UK. Data can be displayed as a graph (Figure 1) or as raw data and can be stored on the computer and retrieved at followup visits to allow assessment of change.

Before the full potential of the 3-dimensional properties of the Fastrak and similar 3-dimensional systems can be realized, it is important that they are shown to be reliable and valid measures of ROM in the primary plane in both healthy and symptomatic subjects. Studies assessing the reliability of the Fastrak have examined lumbar spine movement^2,3, thoracic spine movement⁴, and movement at the ankle joint complex⁵. Another electromagnetic system, the Flock of Birds (Ascension Technology Corp., Burlington, VT, USA), has been evaluated on the cervical spine⁶ and the shoulder^7-9. However, these studies concentrated on a limited number of often healthy subjects. Larger studies evaluated the Isotrak, the predecessor to the Fastrak, on lumbar spine movement¹⁰ and cervical spine movement¹¹ in healthy subjects, and a smaller study assessed its reliability in shoulder movement, again in asymptomatic subjects¹². The Isotrak, however, has only a single sensor and, unlike the Fastrak, the electromagnetic source has to be attached to the subject, which may hinder movement.

The interobserver (40 subjects measured by 2 observers) and intraobserver (32 subjects measured on 3 occasions) reliability of the cervical spine and shoulder primary movement measurements of the Fastrak were found to be good in asymptomatic subjects¹³. Our objective was to validate the primary plane measurements in patients with AS: specifically (1) to compare ROM data for cervical spine and shoulder movement for patients with diagnosed AS to asymptomatic subjects; (2) to assess the repeatability of the Fastrak in AS; and (3) to assess validity by comparing the Fastrak measurements to results from widely used methods of assessing AS. These methods included disease-specific and generic patient assessed health instruments and tape measure assessed ROM.

MATERIALS AND METHODS

Subjects. Patients with a confirmed clinical diagnosis of AS (modified New York criteria¹⁴) registered with a specialist rheumatology center in England, aged 18 years and over, and who regularly attended the center for physiotherapy and participation in a self-help group were invited to participate in the study. Informed consent was obtained from all participants. While these patients were likely to be older, perhaps more severely restricted by AS, and be more male dominated than a routine clinic population, they are representative of patients regularly seen in rheumatology clinics and by physiotherapists¹⁵. Subjects were measured at baseline, 2 weeks, and 3 months. There was no change in the patient's usual care. The study was approved by the local research ethics committee.

Sample size calculation was based on detecting differences in ROM between patients with AS and the 72 healthy subjects measured in a previous reliability study¹³. Guidelines by the American Medical Association¹⁶ suggest that repeated measurements within 10% or 5° of total motion, whichever is greater, should not be considered a clinically significant change. There is little published information about shoulder ROM in patients with AS so sample size calculations were based on available AS cervical spine ROM data¹⁷. Based on a power of 90%, a significance level of 5% and detecting at least a 10% or 5° (whichever is greater) difference between ROM of AS patients and collected ROM data for the asymptomatic subjects, a sample size of approximately 60 at baseline was needed. This sample size was also sufficient to allow examination of the repeatability of the Fastrak in AS and compare the Fastrak data to the other assessment tools used.

Ninety-one patients with diagnosed AS registered with the rheumatology center were eligible to take part in the study. Three patients agreed to take part in a pilot study of the procedure for Fastrak measurement. Eighty-eight patients were sent letters and invited to take part in the main study. A patient information leaflet and consent form were also sent. Reminders were sent after 2 and 4 weeks.

Nonresponders and those who declined to take part were compared to patients taking part in the study with respect to age and sex. Those attending on all 3 occasions were compared with those missing one or 2 occasions on age, sex, and their baseline ROM.

Protocol for movement and measurement with the Fastrak. The protocol was identical to that used in the study of asymptomatic subjects¹³. Movements in each plane were performed consecutively without stopping and repeated so that there were 3 measurements of each movement on each occasion. The Fastrak was centered at zero degrees before movements in a new plane began. Movements were always performed in the same order. Sensors remained fixed to the subject until all movements in all planes were completed.

For the cervical spine, the planes of movement were those relating to the movements of flexion and extension, lateral flexion to right and left, and rotation to right and left. Patients were seated in a wooden chair. One sensor was fixed to a pair of safety spectacles and situated on the forehead. A second sensor, fixed on the sternum using a Velcro strap, measured secondary trunk movement. The analysis reported here relates to the forehead sensor, and the maximum of the 3 repetitions in the primary plane of movement was taken for the analysis. Patients were asked to move as far as they could without moving their shoulders.

For the shoulder, the movements measured were flexion and extension, abduction, and external rotation from the neutral position with elbow flexed to 90°. One sensor was placed just above the elbow using a Velcro strap, one on the acromion process using double-sided tape, and, again, one on the sternum. In the analysis reported here, only the elbow sensor was analyzed, and the maximum of the 3 repetitions in the primary plane only was taken. The dominant shoulder was used for each patient. For flexion/extension and abduction, patients were instructed to stand with their arm by their side and thumb pointing forward. For external rotation, patients were told to keep their elbow into their side. Patients were asked to move as far as they could in the requisite plane of movement.

The position and orientation of each sensor was computed relative to the source transmitter, which was placed on a wooden pedestal behind the subject.

Other assessment procedures. At each timepoint patients also completed a package of disease-specific and generic patient assessed health instruments, and had spinal ROM assessed by tape measure. The instruments and tape measure assessments of ROM reported here were those previously found to have good evidence of reliability, validity, and responsiveness in AS^18-21.

Three disease-specific and 2 generic measures of health status were included. The Ankylosing Spondylitis Quality of Life instrument (ASQoL) is an AS-specific measure of quality of life²². Responses to the 18 items are summed to produce a 0–18 scale: 0 represents the best and 18 the worst QoL. The Bath AS Disease Activity Index (BASDAI) is an AS-specific measure of disease activity comprising 6 items, each with a 10 cm visual analog scale²³. The summed item score is converted to a 0–10 scale; a lower score indicates less disease activity. The Revised Leeds Disability Questionnaire (RLDQ)²⁴ assesses AS-specific functional disability. It has 16 items that describe 4 areas: mobility, bending down, neck movement, and posture. Item summation produces a final score from 0 to 48; a higher score represents greater functional disability. The EuroQol (EQ-5D)²⁵ is a generic measure of health status. Scores range from –0.59 to 1.00, where 1.00 is perfect health. The EuroQol "thermometer" visual analog scale (EQ-VAS) rates individual health on the day of completion from 0, "worst imaginable health state," to 100, "best imaginable health state." The Short Form-12 (SF-12) is a reduced version of the Short Form-36²⁶. Two summary scores are calculated, a physical (PCS) and a mental (MCS) component summary score. The higher the score, the better the health status.

A question on the baseline questionnaire asked subjects to rate their AS-specific health from poor to excellent (5 categories in total). Health transition questions included in the 2-week and 3-month questionnaires asked patients whether their AS related health was better, the same, or worse than at baseline.

The tape-measured ROM were selected following a structured review of published evidence and a study of the reliability, validity, and responsiveness of tape-measured spinal mobility measures in AS^18,21. The selected measurements were the modified 15 cm Schober Index (MSI)^21,27 measured with a plastic tape, fingertip to floor distance following trunk-forward flexion (from the tip of the middle finger of right hand) measured with a retractable steel tape measure²¹, and cervical spine rotation with a plastic tape measured left and right^17,28. Cervical rotation was measured as a reduction in the distance between the tip of nose (a more fixed landmark than the chin) and the acromioclavicular joint²¹. One measurement was taken of each tape-measured ROM. All assessments took place in the afternoon or early evening with repeat measurements scheduled for the same time of day as far as possible.

Analysis. For the Fastrak evaluation, the maximum of 3 consecutive performances of each primary movement was calculated on each occasion. First, comparisons were made on each movement at baseline of the AS patients to that of the 72 asymptomatic subjects in the study as described¹³. As the asymptomatic subjects were not age-sex matched to the patients with AS, analysis of covariance was used to adjust in the comparisons for age, sex, and body mass index (BMI) differences.

Repeatability of the Fastrak was assessed using the intraclass correlation coefficient based on the 2-way random effects ANOVA for a single measurement [denoted by ICC(2,1)²⁹] for the baseline and 2-week measurements. This was performed on all patients measured at those 2 timepoints and then only for those patients indicating no change in AS-specific health at 2 weeks on the health transition question. The ICC ranges from 0 to 1, 1 indicating perfect reliability. It assesses the proportion of total variability explained by the variation between subjects. While there will be variability within subjects when measured on different occasions, it should be far less than that between subjects. Values of 0.80 and above indicate substantial reliability³⁰. To allow assessment of the lowest level of reliability for each measure, one-sided lower 95% confidence limits were also calculated.

The validity of the Fastrak measurements was assessed by comparing patients reporting their baseline AS related health as excellent or very good, to those rating it good, fair, or poor. As the patients with AS were now split into 2 smaller groups, this was performed using the nonparametric Mann-Whitney test. To assess relationships between the Fastrak measurements and the patient assessed health instruments and the tape measure assessments, correlation coefficients were calculated. This was performed on the baseline data. As some of the results for the patient assessed health instruments and tape measurements were highly skewed, Spearman's correlations were used. A strong relationship was defined as a correlation ≥ 0.7, moderate as 0.5–0.7, and minimal as < 0.3³¹.

The final analysis explored how much limitation in cervical spine and/or shoulder ROM affected different areas of health, for example, functioning, disability, quality of life, and mental health. It assessed how much of the variation in AS related health between subjects, and between occasions (baseline, 2 weeks, and 3 months) within subjects, could be explained by the different (Fastrak-measured) movements. Analysis was performed using 2-level repeated measures multilevel modeling³² with the response variable being each of the patient assessed health instrument scores in separate models. Multilevel modeling takes into account the hierarchical nature of the data where the occasions (level 1) are nested within subjects (level 2). Patients who do not participate on all 3 occasions can still be included in the analysis. The amount of total unexplained variation (total variance: level 2 + level 1) in scores that was between subjects (variance at level 2) rather than within subjects (variance at level 1) was determined. The initial pool of explanatory variables included the maximum Fastrak-measured ROM for each primary movement at each occasion, together with age, body mass, sex, disease duration, marital status (married/not married), and occupational status (employed/not employed). As the objective was to assess which of the explanatory variables were most associated with the patient assessed health instrument scores, a forward stepwise procedure was used. Due to the skewness of their distribution, the scores were converted to normal scores, this being the most appropriate transformation for data of this type, where the scores have no intrinsic meaning³³. This transformation preserves the monotonic ordering of the data. The exception to this was the BASDAI, which appeared to have an approximate normal distribution. Explanatory variables were added as fixed coefficients only (i.e., the relationships between instrument scores and each explanatory variable were assumed to be uniform across subjects).

Analyses were performed using SPSS 11.0 for Windows³⁴ and MLwiN³⁵.

RESULTS

Fifty of the 88 patients were measured at baseline (Table 1). After exclusion of the 9 patients who could not be contacted or were found to be ineligible, this gave an adjusted response rate of 63%. This was less than the desired 60 subjects, but this still allowed a power of 80% in the sample size calculations. Forty-four of the 50 patients were measured at 2 weeks and 40 (including 4 not measured at 2 weeks) were measured at 3 months.

Table 1. Background of patients with AS (n = 50).

The 29 nonresponders did not differ from the 50 measured at baseline on sex (p = 0.49) or age (p = 0.14). Compared to patients who attended one or 2 measurement occasions those who attended all 3 tended to be older (mean difference in ages = 7 years; p = 0.05), but did not differ significantly on baseline-measured ROM.

Forty-five (90%) patients with AS were male. This compares to 23 (32%) of the asymptomatic subjects in our earlier study. The AS patients were also older (mean difference 13.2 yrs; 95% CI 9.4, 17.0). Fastrak-measured ROM for the AS patients at baseline is given in Table 2. AS patients had a significantly reduced ROM compared to the asymptomatic subjects on all movements (shoulder external rotation p = 0.02; p < 0.01 for all other movements) except cervical spine right rotation (p = 0.10) and shoulder extension (p = 0.77). This was after adjustment for age, sex, and body mass differences. However, there was a wide range of limitation within the AS patients, varying from very little movement to "normal" movement. Table 2 also shows the estimated reduction in ROM for the AS patients compared to the asymptomatic subjects after adjustment for age, sex, and BMI.

Table 2. Summary of range of motion (degrees) of patients with AS at baseline (n = 50).

The repeatability over 2 weeks of the cervical spine and shoulder movements is shown in Table 3. Five patients felt their health had improved at 2 weeks, and 13 felt it had deteriorated. All ICC were above 0.80 except for shoulder extension (0.75), and all had one-sided confidence limits ≥ 0.60.

Table 3. Reliability of cervical spine and shoulder movements over 2 weeks.

At baseline, 14 patients rated their AS-specific health as excellent or very good and 36 as good, fair, or poor. A better rating of individual AS health was related to better ROM for all movements (Table 4).

Table 4. Range of motion (degrees) by rating of patient assessed AS-specific health*.

Spearman correlations between the Fastrak measurements at baseline and the patient assessed health instruments and tape measurements are given in Tables 5 and 6. The RLDQ and BASDAI had moderate to strong associations with both the Fastrak-measured cervical spine (RLDQ 0.58–0.77; BASDAI 0.52–0.66) and shoulder (RLDQ 0.56–0.72; BASDAI 0.48–0.72) movements. The other instruments tended to have stronger associations with the shoulder than with the cervical spine, although generally less than the RLDQ and BASDAI. The SF-12 mental component summary (MCS) score had minimal association with cervical spine movement and small associations with the shoulder. Cervical spine flexion and shoulder flexion, flexion plus extension, and abduction had the strongest associations with the patient assessed health instrument scores. As expected, the cervical spine Fastrak-measured ROM were most strongly related to the tape-measured cervical spine rotation (all correlations ≥ 0.80). All the tape-measured movements had only small to moderate relationships with the shoulder movements (0.29–0.60).

Table 5. Associations between Fastrak-measured movements and patient assessed health instruments (Spearman correlation coefficient)*.

Table 6. Associations between Fastrak-measured movements and tape measure assessed movements (Spearman correlation coefficient)*.

The final multilevel models of the AS patient assessed health instrument scores are given in Table 7. The null models (i.e., with no explanatory variables included) showed that the majority of the variation was between subjects rather than between the 3 occasions within subjects. The percentage of the variation in instrument scores between subjects (level 2) rather than between occasions within subjects ranged from 73% (EQ-5D) to 90% (RLDQ) and 92% (ASQoL). Inclusion of explanatory variables to form the final models saw the unexplained variation in scores between subjects fall by 68% for the RLDQ and by 30% or more for all instruments except the MCS (14%). However, the greatest reduction at level 1 (between occasions within subjects) was just 8% (ASQoL). The shoulder ROM had the best explanatory power for all instruments. This was shoulder flexion for the PCS, abduction for the RLDQ, BASDAI, ASQoL and MCS, and flexion plus extension for the EQ-5D. Flexion was the only cervical spine movement to further improve any of the models (RLDQ, BASDAI, and PCS). Age did not give any additional improvement to the models, although a larger BMI was related to a poorer RLDQ score.

Table 7. Final multilevel models of patient assessed health instruments*.

DISCUSSION

We performed an assessment of a new ROM measuring tool on a disease population. Patients with AS tended to have major limitations in all movements except shoulder extension. All movements had substantial reliability (> 0.80) except for shoulder extension (0.75). All had higher ICC than in the study of asymptomatic subjects¹³, where all ICC were greater than 0.60 except for cervical spine left rotation (0.54). The higher ICC may be explained partly by the greater variability in movement between AS patients. The reliability values are high even when including patients reporting improvement or deterioration in health. In the 13 patients reporting a deterioration in their AS-specific health at 2 weeks, there was no evidence of poorer ROM. Intraobserver reliability of the tape measure ROM on 26 patients at 2 weeks gave similar levels of ICC²¹.

While there were clear Fastrak cervical spine and shoulder ROM differences between patients giving different ratings at baseline of their own AS-specific health, responsiveness of these Fastrak measurements still needs to be investigated. This should be over a period longer than 3 months, where, in those patients with stable disease, little change in AS-specific health is expected. For example, in our study, at 3 months, 30 patients stated their AS-specific health was the same as baseline, only 4 stated it was better, and 6 worse.

The stronger relationship found with the patient assessed health instrument scores for some of the shoulder measurements compared to the cervical spine is surprising, given that shoulder ROM is seldom assessed in AS patients. However, in a postal study of members of the UK National Ankylosing Spondylitis Society (NASS), 43% declared moderate or severe shoulder pain, and 44% moderate or severe stiffness³⁶. The shoulder movements may be more related to work and physical activities measured by the patient assessed health instruments than the cervical spine movements, and suggests the importance of this joint in AS. Both are most highly correlated with the RLDQ. Three of the 4 items in the RLDQ neck mobility section relate to activities that would involve movement in the flexion/extension plane: looking at things on a high shelf, opening high windows, and drinking. The latter 2 would also involve shoulder flexion. The correlations and the multilevel models suggest it is limitation in shoulder flexion and abduction and cervical spine flexion that most reflect problems in everyday life. Limited cervical spine lateral flexion may have little influence on function. Limited cervical spine rotation may restrict a range of functional activities including sleeping position and intimate contact with a partner³⁷. However, patients may learn to adapt to the disease and adjust their functional expectations accordingly. Consequently, whole-body movements may be a substitute for cervical rotation. However, this study adds to previous reports that show cervical spine rotation to be a valid, reliable, responsive, and practical method of assessment of AS when measured by tape measure^17,21,28.

It is still evident that there is not a substantial relationship between shoulder and cervical spine ROM and patient assessed health. This further illustrates the different roles described by ROM assessment and patient assessed health instruments in the evaluation of AS, and the need to address both these aspects of evaluation to reflect the multidimensional nature of AS.

The participation rate of 63% for this study was acceptable, and comparisons between participants and nonparticipants suggested there were no major differences. While those who participated in all 3 measurement occasions were older than those just participating in one or 2, there was no difference in baseline ROM measurements. Patients invited to participate in the study were regular attendees to the rheumatology center, which gave easier access than a community study of AS patients. Additionally, these patients presented with a wide variety of ROM limitation. This suggests the results are generalizable to patients with diagnosed AS regularly consulting in secondary care rather than all patients with AS. However, those who do not attend exercise classes or clinics may have milder disease³⁸.

The Fastrak showed good repeatability and validity in both cervical spine and shoulder movement in the primary plane in this group of AS patients. Benefit would now be gained from research in further assessing the value of the Fastrak in measuring, for example, lumbar spine and hip movement in AS and in exploring the movements in the secondary planes. A possible method of analyzing the 3-dimensional measuring capabilities of the Fastrak, and the use of sensors placed on different parts of the body, for both patients and healthy subjects has been described³⁹. User-friendly software and standardized protocols mean Fastrak operators in clinical research require limited training and only a basic level of computer skills. However, research on the Fastrak and other 3-dimensional measuring tools is in its infancy, and there are still practical issues to be overcome before they could be readily used in clinical practice⁴⁰. Once this occurs, the Fastrak could usefully aid the clinical assessment of AS.

ACKNOWLEDGMENT

We are grateful to all patients who gave their time to participate in the study. We also thank the Consultant Rheumatologists at the Staffordshire Rheumatology Centre for allowing access to the AS database.

REFERENCES

Search PubMed for:

2. Barrett CJ, Singer KP, Day R. Assessment of combined movements of the lumbar spine in asymptomatic and low back pain subjects using a three-dimensional electromagnetic tracking system. Man Ther 1999;4:94-9. [MEDLINE]

3. Mannion A, Troke M. A comparison of two motion analysis devices used in the measurement of lumbar spinal mobility. Clin Biomech 1999;14:612-9. [MEDLINE]

4. Willems JM, Jull GA, Ng JK-F. An in vivo study of the primary and coupled rotations of the thoracic spine. Clin Biomech 1996;11:311-6. [MEDLINE]

5. Woodburn J, Turner DE, Helliwell PS, Barker S. A preliminary study determining the feasibility of electromagnetic tracking for kinematics at the ankle joint complex. Rheumatology 1999;38:1260-8. [MEDLINE]

6. Koerhuis CL, Winters JC, van der Helm FCT, Hof AL. Neck mobility measurement by means of the 'Flock of Birds' electromagnetic tracking system. Clin Biomech 2003;18:14-8. [MEDLINE]

7. Meskers CGM, Vermeulen HM, de Groot JH, van der Helm FCT, Rozing PM. 3D shoulder position measurements using a six-degree-of-freedom electromagnetic tracking device. Clin Biomech 1998;13:280-92. [MEDLINE]

8. Meskers CGM, Fraterman H, van der Helm FCT, Vermeulen HM, Rozing PM. Calibration of the "Flock of Birds" electromagnetic tracking device and its application in shoulder motion studies. J Biomech 1999;32:629-33. [MEDLINE]

9. Vermeulen HM, Stokdijk M, Eilers PHC, Meskers CGM, Rozing PM, Vlieland TPMV. Measurement of three dimensional shoulder movement patterns with an electromagnetic tracking device in patients with a frozen shoulder. Ann Rheum Dis 2002;61:115-20. [MEDLINE]

10. Van Herp G, Rowe P, Salter P, Paul JP. Three-dimensional lumbar spinal kinematics: a study of range of movement in 100 healthy subjects aged 20 to 60+ years. Rheumatology 2000;39:1337-40. [MEDLINE]

11. Trott PH, Pearcy MJ, Ruston SA, Fulton I, Brien C. Three-dimensional analysis of active cervical motion: the effect of age and gender. Clin Biomech 1996;11:201-6. [MEDLINE]

12. Johnson GR, Fyfe NCM, Heward M. Ranges of movement at the shoulder complex using an electromagnetic movement sensor. Ann Rheum Dis 1991;50:824-7. [MEDLINE]

13. Jordan K, Dziedzic K, Jones PW, Ong BN, Dawes PT. The reliability of the three-dimensional FASTRAK measurement system in measuring cervical spine and shoulder range of motion in healthy subjects. Rheumatology 2000;39:382-8. [MEDLINE]

14. van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis. A proposal for modification of the New York criteria. Arthritis Rheum 1984;27:361-8. [MEDLINE]

15. Calin A, Elswood J, Rigg S, Skevington SM. Ankylosing spondylitis — an analytical review of 1500 patients: the changing pattern of disease. J Rheumatol 1988;15:1234-8. [MEDLINE]

16. American Medical Association. Guides to the evaluation of permanent impairment. 4th ed. Chicago: American Medical Association; 1993.

17. Viitanen JV, Kokko ML, Heikkila S, Kautiainen H. Neck mobility assessment in ankylosing spondylitis: a clinical study of nine measurements including new tape methods for cervical rotation and lateral flexion. Br J Rheumatol 1998;37:377-81. [MEDLINE]

18. Haywood KL. Health outcomes in ankylosing spondylitis: an evaluation of patient-based and anthropometric measures [DPhil thesis]. York: University of York; 2000.

19. Haywood KL, Garratt AM, Jordan K, Dziedzic K, Dawes PT. Disease-specific, patient-assessed measures of health outcome in ankylosing spondylitis: reliability, validity and responsiveness. Rheumatology 2002;41:1295-302. [MEDLINE]

20. Haywood KL, Garratt AM, Dziedzic K, Dawes PT. Generic measures of health-related quality of life in ankylosing spondylitis: reliability, validity and responsiveness. Rheumatology 2002;41:1380-7. [MEDLINE]

21. Haywood KL, Waterfield J, Jordan K, Garratt AM, Dziedzic K, Dawes PT. Spinal mobility in ankylosing spondylitis: reliability, validity and responsiveness. Rheumatology Oxford 2004;43:750-7. [MEDLINE]

22. Doward LC, Spoorenberg A, Cook SA, et al. Development of the ASQoL: a quality of life instrument specific to ankylosing spondylitis. Ann Rheum Dis 2003;62:20-6. [MEDLINE]

23. Garrett S, Jenkinson T, Kennedy LG, Whitelock H, Gaisford P, Calin A. A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286-91. [MEDLINE]

24. Abbott CA, Helliwell PS, Chamberlain MA. Functional assessment in ankylosing spondylitis: evaluation of a new self-administered questionnaire and correlation with anthropometric variables. Br J Rheumatol 1994;33:1060-6. [MEDLINE]

25. EuroQol Group. EuroQol — a new facility for the measurement of health-related quality of life. Health Policy 1990;16:199-208. [MEDLINE]

26. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996;34:220-33. [MEDLINE]

27. Macrae IF, Wright V. Measurement of back movement. Ann Rheum Dis 1969;28:584-9.

28. Viitanen JV, Heikkila S, Kokko ML, Kautiainen H. Clinical assessment of spinal mobility measurements in ankylosing spondylitis: a compact set for follow-up and trials? Clin Rheumatol 2000;19:131-7. [MEDLINE]

29. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:420-8.

30. Shrout PE. Measurement reliability and agreement in psychiatry. Stat Methods Med Res 1998;7:301-17. [MEDLINE]

31. McHorney CA, Ware JE, Raczek AE. The MOS-36-item Short-Form Health Survey (SF-36): II. Psychometric and clinic tests of validity in measuring physical and mental health constructs. Med Care 1993;3:247-63. [MEDLINE]

32. Beacon HJ, Thompson SG. Multi-level models for repeated measurement data: application to quality of life data in clinical trials. Stat Med 1996;15:2717-32. [MEDLINE]

33. Goldstein H. Multilevel statistical models. 2nd ed. London: Arnold; 1995.

34. SPSS for Windows. Release 11. Chicago: SPSS Inc.; 2001.

35. Rasbash J, Browne W, Goldstein H, et al. A user's guide to MLwiN. London: Institute of Education, University of London; 2000.

36. Will R, Kennedy G, Elswood J, et al. Ankylosing spondylitis and the shoulder: commonly involved but infrequently disabling. J Rheumatol 2000;27:177-82. [MEDLINE]

37. Nemes L. Psychosocial and physical problems in AS. NASS Newsletter 1991;Spr/Sum:8-10.

38. Jones SD, Steiner A, Garrett SL, Calin A. The Bath Ankylosing Spondylitis Patient Global Score (BAS-G). Br J Rheumatol 1996;35:66-71. [MEDLINE]

39. Jordan K, Jones PW, Dziedzic K. Describing three-dimensional cervical spine movement in a diseased and a non-diseased group using multilevel modelling. Stat Med 2003;22:2365-80. [MEDLINE]

40. Jordan K, Dziedzic K, Mullis R, Dawes PT, Jones PW. The development of three-dimensional range of motion measurement systems for clinical practice. Rheumatology 2001;40:1081-4. [MEDLINE]