Clinimetric Evaluation of the Physical Mobility Scale Supports Clinicians and Researchers in Residential Aged Care

Article Outline

• Abstract
• Methods
  • Participants and Setting
  • Procedure
  • Outcome Measure
  • Statistical Analysis
    • Interrater agreement
    • Internal construct validity testing
• Results
• Discussion
  • Study Limitations
• Conclusions
• Appendix 1
• References
• Copyright

Abstract 

Barker AL, Nitz JC, Low Choy NL, Haines TP. Clinimetric evaluation of the Physical Mobility Scale supports clinicians and researchers in residential aged care.

Objective

To investigate the interrater agreement and the internal construct validity of the Physical Mobility Scale, a tool routinely used to assess mobility of people living in residential aged care.

Design

Prospective, multicenter, external validation study.

Setting

Nine residential aged care facilities in Australia.

Participants

Residents (N=186). Phase 1 cohort (99 residents; mean age, 85.22±5.1y); phase 2 cohort (87 residents; mean age, 81.59±10.69y).

Interventions

Not applicable.

Main Outcome Measures

Kappa statistics, minimal detectable change (MDC₉₀) scores, and Bland-Altman plots were used to assess interrater agreement. Scale unidimensionality, item hierarchy, and person separation were examined with Rasch analysis for both cohorts.

Results

Agreement between raters on 6 of the 9 Physical Mobility Scale items was high (κ>.60). The MDC₉₀ value was 4.39 points, and no systematic differences in scores between raters were found. The Physical Mobility Scale showed a unidimensional structure demonstrated by fit to the Rasch model in both cohorts (phase 1: χ²=23.90, P=.16, person separation index=0.96; phase 2: χ²=22.00, P=.23, person separation index=0.96). Standing balance was the most difficult item in both cohorts (phase 1: logit=2.48, SE, 0.16; phase 2: logit=2.53, SE, 0.15). The person-item threshold map indicated no floor or ceiling effects in either cohort.

Conclusions

The Physical Mobility Scale demonstrated good interrater agreement and internal construct validity with good fit to the Rasch model in both cohorts. The comparative results across the 2 cohorts indicate generality of the findings. The Physical Mobility Scale total raw scores can be converted to Rasch transformed scores, providing an interval measure of mobility. The Physical Mobility Scale may be suited to a range of clinical and research applications in residential aged care.

Key Words: Aged, Nursing homes, Outcome assessment (health care), Rehabilitation

List of Abbreviations: BBS, Berg Balance Scale, MDC, minimal detectable change, MDC₉₀, minimal detectable change at 90% confidence interval, PSI, person separation index, RMI, Rivermead Mobility Index

THE PHYSICAL MOBILITY Scale is an ordinal measure developed by physiotherapists to assess the mobility of people living in residential aged care. It is a physical performance test that requires the user to observe and score residents' independence and equipment requirements for 9 mobility tasks including bed mobility, sitting and standing balance, ambulation, and transfers.¹ Assessment and management of mobility impairment is an integral part of clinical management of residents in residential aged care.² In this population, mobility is an important risk factor for adverse health events such as respiratory tract infections,³ falls,⁴, ⁵, ⁶ and fractures.⁷ Beyond adverse health outcomes for residents, mobility impairment increases the need for physical assistance, which increases risk of staff injury.⁸ Accurate mobility assessment informs treatment and promotes the use of appropriate equipment and assistance, which may reduce the risk of injuries sustained by residents and staff. Having accurate, reliable, and valid tools to measure resident mobility is also essential for clinicians working in residential aged care to enable resident outcomes to be monitored, and to identify residents at risk of adverse health events. There have been few comprehensive investigations into the clinimetric properties of mobility assessments used in residential aged care.

The Physical Mobility Scale has previously been reported to have high levels of interrater agreement, test-retest reliability, and concurrent validity with the Clinical Outcome Variable Scale and the RMI.¹ To date, reliability analysis including MDC and internal construct validity testing through Rasch analysis for the Physical Mobility Scale have not been reported. The MDC provides clinicians with a value of change scoring that must be achieved to overcome measurement error.⁹ Rasch analysis is based on item response theory and permits examination of the measurement and scaling properties of an instrument. It provides information about the appropriateness of the summing of item scores to yield a total score (ie, representing overall mobility),¹⁰ ceiling and floor effects that can limit ability to measure change over time, and conversion of ordinal to interval-level scores.¹¹ Ordinal measures, such as the Physical Mobility Scale, provide information about the relative ordering of scores, but not the magnitude of separation between scores. As such, an increase in scoring on an ordinal measure indicates that a resident's mobility has improved, but it does not provide information about how much improvement has occurred. Interval measures give information on the magnitude of scoring separation, providing information such as percentage change scores.¹⁰ Interval-based measures also offer greater research potential than ordinal measures because parametric statistics can be used. While Rasch analysis has previously been used in the development and validation of several mobility outcome measures, no study has been conduced in the residential aged care setting.¹⁰, ¹², ¹³, ¹⁴, ¹⁵, ¹⁶, ¹⁷, ¹⁸, ¹⁹ This study aimed to extend the clinimetric evaluation of the Physical Mobility Scale by examining the properties of interrater agreement as measured by the MDC and the ability of the Physical Mobility Scale to fit the Rasch model.

Back to Article Outline

Methods 

This study used data from a retrospective chart audit of 4 facilities (phase 1) and a prospective validation study conducted in 6 facilities (phase 2). While 1 facility was included in both studies, different people were included in each sample so that the phase 1 cohort was external to the phase 2 cohort. The study was approved by the University of Queensland Medical Research Ethics Committee.

Participants and Setting 

Permanent high-care (nursing home) and low-care (hostel) residents were eligible for inclusion in the study if they had lived at the facility for longer than 12 months. Ninety-nine charts were received for auditing in the phase 1 cohort. Participants in phase 2 were recruited by written informed consent, and 87 residents agreed to participate. Where residents were unable to provide consent, consent was sought from a family member or guardian.

Procedure 

Physical Mobility Scale assessments for the phase 1 cohort were completed by physiotherapists as part of usual care. Physiotherapists independent from the residential aged care facilities (rater 1 and rater 2) completed the assessments for the phase 2 cohort. For the interrater agreement testing, rater 1 conducted the assessment while rater 2 observed and scored the participant on performance. Raters were blind to each other's scoring.

Outcome Measure 

The Physical Mobility Scale is composed of 9 mobility tasks that are scored on a 6-point ordinal scale from least independent (score=0) to independent (score=5) to yield a total score out of 45. A copy of the Physical Mobility Scale tool is available in the article by Nitz et al.¹

Statistical Analysis 

Data were analyzed using Stata (version 9.2)^a for all analyses except the Rasch analysis, which was completed using RUMM2020.^b

Analyses were performed on the first 28 participants recruited in the phase 2 cohort. The κ statistic was used to assess agreement between raters' item scoring of participants. Values of .60 or above were considered evidence of high agreement.²⁰ The MDC₉₀ was calculated to assess the error in measurement of the total scores between raters.²¹, ²², ²³ Bland-Altman plots with 95% limits of agreement were constructed to detect systematic bias between rater scores.²⁴

Analyses were performed on the phase 1 (n=99) and phase 2 (n=87) cohorts separately to provide information about the generality of the results. Rater 1 scores were used in the Rasch analysis of phase 2 data. Rasch analysis was used to investigate the scaling properties of the Physical Mobility Scale, through probabilistic modelling of item difficulty and person ability on a common linear continuum.²⁵, ²⁶ The properties of unidimensionality (measurement of 1 underlying construct), item hierarchy (the relative difficulty of the items compared with one another), and person separation (the extent to which items distinguish between different levels of mobility) were examined.²⁷ Overall fit of the Physical Mobility Scale to the Rasch model was determined by the item-fit and person-fit statistics. Item-trait interaction quantifies the fit of the observed data to the predicted model; for instance, a mean of 0 and SD of 1 indicate perfect fit to the model.²⁸ The chi-square statistics for the overall and individual item-trait interactions should show no significant deviations from the model expectation, confirming scale unidimensionality.²⁸ Confirming unidimensionality validates the summing of item scores to yield an overall score to measure the underlying construct.¹⁰ Residual fit statistics provide additional information about the unidimensionality of the measure. The residuals are the standardized person-item differences between the observed data and what is expected by the model for every person's response to every item.²⁷ Item residual statistics should range between –2.5 and 2.5.²⁷, ²⁸ Item fit residuals of greater than ±2.5 may indicate the measurement of another construct by that item. If several items are found to have fit residuals that exceeded ±2.5, these items may be redundant. Item logit (log odds of the probability) scores were calculated to examine the hierarchy of the scale items. More difficult items have higher logit values, while easier items have lower values. This facilitates the identification of items that are easy enough for residents with low levels of mobility to perform and items that are challenging for residents with high levels of mobility. Person-item threshold maps, displaying person ability and item difficulty on a common logit scale, were constructed to detect floor or ceiling effects. The PSI was calculated as a measure of the scale agreement. The PSI indicates how well the items of the scale separate the participants in the sample. The PSI should be above .70 to allow the scale to differentiate at least 2 strata of resident mobility.²⁸ Cronbach α was used to measure the correlation between items within the Physical Mobility Scale, representing a measure of internal consistency. Alpha values greater than .70 were deemed acceptable. Once fit to the Rasch model was confirmed, raw Physical Mobility Scale total scores were transformed into interval-level scores.

Back to Article Outline

Results 

Table 1 shows the demographic characteristics of participants in the study. Residents in the phase 1 cohort were more likely to be women (Pearson χ²=5.480; P=.019) and nonambulant (Pearson χ²=7.430; P=.006). No other significant differences between cohorts were found (P>.05). No participants withdrew from the study, and no adverse events attributable to the study assessments were reported.

Table 1. Demographics and Characteristics by Cohort

Characteristics	Phase 1 n=99	Phase 2 n=87
Age	85.22±5.1	81.59±10.69
Sex⁎
Women	72(72.73)	49(58.33)
High-level (nursing home) care⁎	79(79.80)	77(88.51)
Diagnosis⁎†
Dementia	38(38.38)	44(50.57)
Osteoporosis	16(16.16)	20(22.99)
Depression	16(16.16)	16(18.39)
Diabetes mellitus	14(14.14)	10(11.49)
Previous fractured neck of femur	14(14.14)	10(11.49)
Mobility
Nonambulant	36(36.36)	16(18.39)
Independent ambulation	33(33.33)	18(20.69)
Duration of observation period (y)	3.50±2.3	0.51±0.06

NOTE. Values are n (%) or mean ± SD.

The MDC₉₀ value was 4.39 points. Agreement between raters on 6 of the 9 items was high (κ≥.60) (table 2). The Bland-Altman plot (fig 1) shows that the mean differences between rates scores was small (mean difference, –.214; 95% confidence interval, –1.28 to 0.85), indicating no systematic differences.

Table 2. Interrater Agreement and Rasch Model Fit Statistics by Physical Mobility Scale Item and Cohort

PMS Item	κ (SE)	Item Difficulty Logits (SE)		Item Fit χ2 (P)		Item Fit Residuals
Phase	2	1	2	1	2	1	2
	n=28	n=99	n=87	n=99	n=87	n=99	n=87
Supine to left side-lie	0.60(0.10)	−0.45(0.15)	−0.44(0.14)	1.88(0.39)	4.24(0.12)	0.72	0.45
Supine to left side-lie	0.75(0.11)	−0.40(0.14)	−0.87(0.15)	3.66(0.16)	1.69(0.43)	0.98	0.49
Supine to sit	0.80(0.10)	−1.13(0.16)	−2.46(0.16)	2.35(0.31)	3.33(0.19)	−0.81	1.41
Sitting balance	0.47(0.12)	−1.43(0.16)	−1.61(0.17)	1.87(0.39)	1.75(0.42)	−0.36	0.16
Sit to stand	0.67(0.10)	0.82(0.18)	0.96(0.14)	4.43(0.11)	3.71(0.16)	−2.83	−0.59
Stand to sit	0.46(0.09)	0.91(0.17)	0.75(0.17)	2.49(0.29)	3.12(0.21)	−1.51	−0.37
Standing balance	0.61(0.09)	2.48(0.16)	2.53(0.15)	3.83(0.15)	0.44(0.80)	−0.35	−0.09
Transfers	0.62(0.10)	−0.47(0.15)	0.15(0.16)	1.24(0.54)	0.50(0.78)	−1.21	−0.40
Mobility	0.59(0.09)	−0.34(0.14)	0.98(0.15)	2.14(0.34)	3.21(0.20)	−0.12	−0.01

NOTE. High positive logit values represent difficult items, while low negative values represent easy items. Nonsignificant (P>.05) item fit χ² values indicate fit to the Rasch model and confirm unidimensionality.

Abbreviations: PMS, Physical Mobility Scale.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Bland-Altman plot of interrater agreement. Abbreviations: LOA, limits of agreement; PMS, Physical Mobility Scale.

Table 2 summarizes the results of the Rasch analysis. Analysis of individual item fit statistics reveal that all 9 Physical Mobility Scale items did not deviate significantly (P>.05) from the Rasch model, confirming unidimensionality and the appropriateness of summing of item scores to provide an overall measure of mobility. Further supporting unidimensionality, item-trait interaction statistics identified no significant deviations from the Rasch model in both cohorts (phase 1: χ²=23.90, P=.16; and phase 2: χ²=22.00, P=.23). These results indicate that resident performance scores were consistent with the order of item difficulty—higher scoring residents were able to perform more difficult items. The residual mean value for the items was 0 in both cohorts, with an SD of 1.21 in the phase 1, and 1.52 in the phase 2 cohorts, further providing support of unidimensionality. The PSI in both cohorts (.96) indicated that the participants were well targeted to the items,²⁹ and that the Physical Mobility Scale was very effective in discriminating differing levels of mobility across the sample. Standing balance was the most difficult task in both cohorts, and supine to sit and sitting balance the easiest (see table 2). There was a lack of consistency between the item difficulty parameters across cohorts for the supine to sit, transfers, and mobility items.

The person-item threshold maps (fig 2) indicated that the range of item difficulty was comprehensive and well matched to person ability, showing no floor or ceiling effects. Cronbach α was .98 in the phase 1 cohort and .97 in the phase 2 cohort, indicating high internal consistency. Interitem correlations were also high in phase 1 (mean, .80±.09; range, .62–.96) and phase 2 (mean, .76±.07; range, .68–.85).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  Person-item threshold maps by cohort. Abbreviation: Freq, frequency.

Raw ordinal-level Physical Mobility Scale total scores were transformed to interval-level Rasch transformed scores (appendix 1).

Back to Article Outline

Discussion 

This study provides new evidence of the Physical Mobility Scale MDC₉₀ scores and internal construct validity. The Physical Mobility Scale was found to have a small error in measurement and good individual item and overall fit to the Rasch model. These results support continued clinical and research use of the Physical Mobility Scale, and validate the Physical Mobility Scale as an interval measure of mobility in residential aged care. The Physical Mobility Scale is the first mobility measure to be validated by Rasch analysis in the residential aged care setting. The Rasch transformed interval scores yield a total Physical Mobility Scale score out of 100, because such change scores can be interpreted as a percentage change in mobility.¹⁰ The interval Physical Mobility Scale score also offers advanced research application because parametric statistical analyses can be employed.

In addition to these attributes, the Physical Mobility Scale has several benefits over existing mobility measures used in residential aged care. Over half of high-care (nursing home) residents are nonambulant³⁰ and consequently are unable to perform assessments such as the Timed Up & Go test³⁰ and the 10-foot walk test,⁵ and are unable to complete many items on the BBS³⁰ and RMI,³¹ resulting in floor effects in these measures. The Physical Mobility Scale (see fig 2) was found not to have floor or ceiling effects. The prevalence of dementia in residential aged care is high, generally greater than 30%.³² Over 50% of participants in the phase 2 of this study had a documented diagnosis of dementia (see table 1). The BBS³⁰ and the RMI³¹ require residents to follow instructions to perform the assessed mobility tasks. People with cognitive impairment may have difficulty following these instructions and so may be unable to complete the full assessment. The Physical Mobility Scale includes mobility tasks that are commonly performed by residents in everyday activities, as opposed to more clinically oriented tests such as the Timed Up & Go test, BBS, and RMI. Accordingly, the Physical Mobility Scale assessment can be performed through observation of the residents moving in their everyday activities without the need for the residents to follow multistep instructions. Measures such as timed chair stands and the 10-foot walk test are more limited assessment tasks because they provide information about only 1 component of mobility, omitting information on performance of activities such as bed mobility, ambulation (for timed chair stands), or chair mobility and sitting balance (for the 10-foot walk test). Further, they do not provide information about the level of assistance and equipment that a resident requires to perform mobility tasks safely. These tools therefore do not facilitate the prevention of handling injuries that could be sustained by residents and staff. In contrast, the Physical Mobility Scale includes a broad range of commonly performed resident mobility tasks, and identifies staff and equipment assistance requirements. Consequently, the Physical Mobility Scale offers a valid and reliable method for measuring and communicating information to staff about the level of assistance and equipment that a resident requires to perform mobility tasks safely.

The MDC₉₀ score obtained means that a change in Physical Mobility Scale raw score of greater than 5 points is required before the change in scoring can be interpreted as not being error in measurement. These results suggest that the Physical Mobility Scale is sufficiently sensitive to detect change in a resident's mobility. The items sitting balance, stand to sit, and mobility demonstrated only moderate interrater agreement, while the remaining 6 items were found to have high levels of interrater agreement. Lower agreement may be reflective of the subjective wording of these items. However, the κ statistics of .46 through .59 obtained in this study, and intraclass correlation coefficients of .68 through .88 from the past study for these 3 items,¹ represent clinically useful levels of agreement, so review and modification of these items is not of crucial importance.

Study Limitations 

Several limitations of the study need to be acknowledged. First, although the concordance of estimates across cohorts provides promising evidence of the generality of the results, the sample sizes used were relatively small for Rasch analysis. This may affect the precision of the estimates of the Rasch model, particularly the logit scores of item difficulty. As such, the item difficulty results should not be overinterpreted. A large-scale study is still required to establish normative values for Physical Mobility Scale scores and to confirm the findings of the current study. Second, the use of concurrent observational assessment by the 2 raters means that the reliability of application of the tools by the raters was not assessed.

Back to Article Outline

Conclusions 

This study provides important insights into the clinimetric properties of a commonly used mobility assessment tool in residential aged care. The Physical Mobility Scale demonstrated good interrater agreement and internal construct validity including good individual item and overall fit to the Rasch model. The Physical Mobility Scale total raw scores can be converted to Rasch transformed scores, providing an interval measure of mobility. This study therefore provides evidence that the Physical Mobility Scale is a valid and reliable measure of mobility for people living in residential aged care, and is suited to a range of clinical and research applications.

Suppliers

Back to Article Outline

Appendix 1 

Appendix 1. Raw Score to Rasch Score Conversion Table

Back to Article Outline

References 

1. Nitz JC, Hourigan SR, Brown A. Measuring mobility in frail older people: reliability and validity of the Physical Mobility Scale. Australas J Ageing. 2006;25:31–35
   • View In Article
   • CrossRef
2. Saliba D, Solomon D, Rubenstein L, et al. Quality indicators for the management of medical conditions in nursing home residents. J Am Med Dir Assoc. 2005;6(Suppl 3):S36–S48
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (132 KB)
   • CrossRef
3. Binder EF, Kruse RL, Sherman AK, et al. Predictors of short-term functional decline in survivors of nursing home-acquired lower respiratory tract infection. J Gerontol A Biol Sci Med Sci. 2003;58:60–67
   • View In Article
   • MEDLINE
4. Capezuti E, Boltz M, Renz S, Hoffman D, Norman RG. Nursing home involuntary relocation: clinical outcomes and perceptions of residents and families. J Am Med Dir Assoc. 2006;7:486–492
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (115 KB)
   • CrossRef
5. Thapa PB, Gideon P, Brockman KG, Fought RL, Ray WA. Clinical and biomechanical measures of balance as fall predictors in ambulatory nursing home residents. J Gerontol A Biol Sci Med Sci. 1996;51:M239–M246
   • View In Article
   • MEDLINE
6. Lord SR, March LM, Cameron ID, et al. Differing risk factors for falls in nursing home and intermediate-care residents who can and cannot stand unaided. J Am Geriatr Soc. 2003;51:1645–1650
   • View In Article
   • MEDLINE
   • CrossRef
7. Girman CJ, Chandler JM, Zimmerman SI, et al. Prediction of fracture in nursing home residents. J Am Geriatr Soc. 2002;50:1341–1347
   • View In Article
   • MEDLINE
   • CrossRef
8. Nelson A, Powell-Cope G, Gavin-Dreschnack D, et al. Technology to promote safe mobility in the elderly. Nurs Clin North Am. 2004;39:649–671
   • View In Article
   • Full Text
   • Full-Text PDF (529 KB)
   • CrossRef
9. Stratford PW, Goldsmith CH. Use of the standard error as a reliability index of interest: an applied example using elbow flexor strength data. Phys Ther. 1997;77:745–750
   • View In Article
   • MEDLINE
10. Hsieh CL, Jang Y, Yu TY, Wang WC, Sheu CF, Wang YH. A Rasch analysis of the Frenchay Activities Index in patients with spinal cord injury. Spine. 2007;32:437–442
    • View In Article
    • CrossRef
11. Smith EV. Metric development and score reporting in Rasch measurement. J Appl Meas. 2000;1:303–326
    • View In Article
    • MEDLINE
12. Engberg A, Garde B, Kreiner S. Rasch analysis in the development of a rating scale for assessment of mobility after stroke. Acta Neurol Scand. 1995;91:118–127
    • View In Article
    • MEDLINE
    • CrossRef
13. Wright BD, Linacre JM, Smith RM, Heinemann AW, Granger CV. FIM measurement properties and Rasch model details. Scand J Rehabil Med. 1997;29:267–272
    • View In Article
    • MEDLINE
14. Franchignoni F, Giordano A, Ferriero G, Orlandini D, Amoresano A, Perucca L. Measuring mobility in people with lower limb amputation: Rasch analysis of the mobility section of the prosthesis evaluation questionnaire. J Rehabil Med. 2007;39:138–144
    • View In Article
    • MEDLINE
    • CrossRef
15. Williams GP, Robertson V, Greenwood KM, Goldie PA, Morris ME. The high-level mobility assessment tool (HiMAT) for traumatic brain injury, part 2: content validity and discriminability. Brain Inj. 2005;19:833–843
    • View In Article
    • MEDLINE
    • CrossRef
16. Antonucci G, Aprile T, Paolucci S. Rasch analysis of the Rivermead Mobility Index: a study using mobility measures of first-stroke inpatients. Arch Phys Med Rehabil. 2002;83:1442–1449
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (316 KB)
    • CrossRef
17. MacKnight C, Rockwood K. Rasch analysis of the hierarchical assessment of balance and mobility (HABAM). J Clin Epidemiol. 2000;53:1242–1247
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (138 KB)
    • CrossRef
18. Tesio L, Cantagallo A. The functional assessment measure (FAM) in closed traumatic brain injury outpatients: a Rasch-based psychometric study. J Outcome Meas. 1998;2:79–96
    • View In Article
    • MEDLINE
19. Avlund K, Kreiner S, Schultz-Larsen K. Construct validation and the Rasch model: functional ability of healthy elderly people. Scand J Soc Med. 1993;21:233–246
    • View In Article
    • MEDLINE
20. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174
    • View In Article
    • MEDLINE
    • CrossRef
21. Stratford PW, Binkley J, Solomon P, Finch E, Gill C, Moreland J. Defining the minimum level of detectable change for the Roland-Morris questionnaire. Phys Ther. 1996;76:359–365discussion 366-8
    • View In Article
    • MEDLINE
22. Davidson M, Keating JL. A comparison of five low back disability questionnaires: reliability and responsiveness. Phys Ther. 2002;82:8–24
    • View In Article
    • MEDLINE
23. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19:231–240
    • View In Article
    • MEDLINE
    • CrossRef
24. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8:135–160
    • View In Article
    • MEDLINE
    • CrossRef
25. Guiloff RJ. Clinical trials in neurology. London: Springer-Verlag; 2001;
    • View In Article
26. Hopman-Rock M, Van Buuren S, De Kleijn-De Vrankrijker M. Polytomous Rasch analysis as a tool for revision of the severity of disability code of the ICIDH. Disabil Rehabil. 2000;22:363–371
    • View In Article
    • MEDLINE
    • CrossRef
27. Keenan AM, Redmond AC, Horton M, Conaghan PG, Tennant A. The Foot Posture Index: Rasch analysis of a novel, foot-specific outcome measure. Arch Phys Med Rehabil. 2007;88:88–93
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (295 KB)
    • CrossRef
28. Kutlay S, Kucukdeveci AA, Elhan AH, Yavuzer G, Tennant A. Validation of the Middlesex Elderly Assessment of Mental State (MEAMS) as a cognitive screening test in patients with acquired brain injury in Turkey. Disabil Rehabil. 2007;29:315–321
    • View In Article
    • MEDLINE
    • CrossRef
29. RUMM Laboratories. Interpreting RUMM2020: polytomotous data. Duncraig, Australia: RUMM Laboratories; 2005;
    • View In Article
30. Thapa PB, Brockman KG, Gideon P, Fought RL, Ray WA. Injurious falls in nonambulatory nursing home residents: a comparative study of circumstances, incidence, and risk factors. J Am Geriatr Soc. 1996;44:273–278
    • View In Article
    • MEDLINE
31. Kose N, Cuvalci S, Ekici G, Otman AS, Karakaya MG. The risk factors of fall and their correlation with balance, depression, cognitive impairment and mobility skills in elderly nursing home residents. Saudi Med J. 2005;26:978–981
    • View In Article
    • MEDLINE
32. van Doorn C, Gruber-Baldini AL, Zimmerman S, et al. Dementia as a risk factor for falls and fall injuries among nursing home residents. J Am Geriatr Soc. 2003;51:1213–1218
    • View In Article
    • MEDLINE
    • CrossRef

• a Statacorp, 4905 Lakeway Dr, College Station, TX 77845.
• b RUMM Laboratory, 14 Dodoneaea Court, Duncraig, Western Australia 6023.