Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2014;95:2367-75
ORIGINAL ARTICLE
Development and Initial Psychometric Evaluation of the Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in) Francesc Medina-Mirapeix, PT, PhD,a Mariano Gacto-Sa´nchez, PT,b Esther Navarro-Pujalte, PT, PhD,c Joaquina Montilla-Herrador, PT, PhD,a Carmen Lillo-Navarro, PT, PhD,d Pilar Escolar-Reina, PT, PhDa From the aDepartment of Physical Therapy, Regional Campus of International Excellence “Campus Mare Nostrum,” University of Murcia, Murcia, Spain; bDepartment of Physical Therapy, EUSES University School, University of Girona, Girona, Spain; cDepartment of Education, Region of Murcia, Murcia, Spain; and dDepartment of Pathology and Surgery, University “Miguel Herna´ndez,” Alicante, Spain.
Abstract Objective: To describe the development and the initial psychometric evaluation of a mobility measure for inpatient postacute rehabilitation settingsdthe Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in). Design: Self-reportebased psychometric study. Setting: Postacute rehabilitation unit of a public hospital. Participants: A consecutive sample of inpatients (NZ239) receiving postacute rehabilitation care. Interventions: Not applicable. Main Outcome Measures: We developed a 30-item mobility measure, using the Mobility Activities Measure (Mobam) framework, to assess functioning across 5 mobility activity domains classified within the International Classification of Functioning, Disability and Health. These were (1) changing and maintaining body position involving only sitting and/or lying (4 items); (2) changing and maintaining body position involving standing up (6 items); (3) carrying and moving objects using the hand and shoulder (6 items); (4) handling objects using only the hand and/or forearm (7 items); and (5) walking and moving (7 items). Psychometric analyses were conducted to test assumptions underlying the scaling and scoring of Mobam-in scales, and to test both the reliability and validity. Results: Multitrait scaling and confirmatory factor analyses (with Tucker-Lewis Index median, .99; root mean square error of approximation median, .025) supported the assumption of unidimensionality concerning each domain. Five dimensions appeared to be stable across diverse diagnostic groups (the percentage of items with discriminant validity ranged from 93% to 100%, Cronbach coefficient ranged from .859 to .966). Rasch model (Masters’ partial credit) showed that all items could be located along a continuum in each dimension, with goodness-of-fit criteria of infit and outfit mean-square values between 0.6 and 1.4. Test-retest reliability was excellent (intraclass correlation coefficients median, .98). Groups with more severe conditions and lower functional independence scored lower on Mobam-in scales, as hypothesized. Conclusions: Mobam-in covers 5 dimensions of mobility activities. The Mobam framework is an effective reference for building outcome instruments. Archives of Physical Medicine and Rehabilitation 2014;95:2367-75 ª 2014 by the American Congress of Rehabilitation Medicine
Many hospitalized patients often require early postacute rehabilitation to optimize functioning after acute care.1,2 This postacute care may be provided either in specialized rehabilitation facilities
Disclosures: none.
or in dedicated units of an acute care hospital.1 A wide variety of rehabilitation outcome instruments for mobility and self-care activities are currently in use in postacute rehabilitation settings in order to evaluate the effectiveness of postacute programs across diagnostic groups.3-6 However, many of these clinical measures are neither comprehensive assessments for these activity domains
0003-9993/14/$36 - see front matter ª 2014 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.07.407
2368
F. Medina-Mirapeix et al
nor etiologically neutral to allow comparisons across diagnostic groups.7,8 Most importantly, most of the current measures use raw scores from ordinal scales and are therefore not qualified for parametric statistics such as calculating means or SDs.9 Thus, when considering inpatient rehabilitation settings, there is an important need to develop interval-scaled clinical measures that provide a reliable estimation of specific inpatient problems across diagnostic groups. The International Classification of Functioning, Disability and Health (ICF)10 provides a conceptual framework and classification system for developing comprehensive outcome instruments based on activities.3,6,11 The ICF principle of grouping activities with similar purposes (eg, caring for body parts) is theoretically understandable. However, it has been suggested that unidimensionality, a fundamental aspect of measurement theory, is difficult to achieve for ICF self-care activities because of the varying underlying sequences of movement (which are completely different, for instance, for brushing teeth as compared with cutting toenails).6 In a previous study,11 we showed that the unidimensionality of mobility activities depends on the movement sequences (eg, use of the hand, forearm, or both). Five separate mobility dimensions were identified from the Mobility Activities chapter of the ICF: (1) changing and maintaining body position involving only sitting and/or lying (eg, turning in bed); (2) changing and maintaining body position involving standing up (eg, standing 15min); (3) carrying and moving objects using the hand and shoulder (eg, hanging a hanger in a closet); (4) handling objects using only the hand and/or forearm (eg, turning a key); and (5) walking and moving (eg, climbing stairs). The 5 aforementioned dimensions were defined as a conceptual framework, which we refer to as the Mobility Activities Measure (Mobam).11 Within the context of postacute care, there is a strong consensus for the need to develop instruments for monitoring functional recovery across the continuum of care settings.3 Several solutions have been suggested, including paired instruments created for outcome assessment across inpatients and community settings.12 Our research group has developed a set of 5 short mobility measures, called Mobam, that can be used to measure outcomes in outpatient rehabilitation settings.11 The next phase of our research, which is the focus of the present article, examined the ability of the Mobam framework to develop a paired tool for inpatients. Most of the currently available clinical measures generally provide only limited coverage of the ICF mobility domain.6,13 A significant advantage of developing Mobam-based clinical measures is that these provide an ICF-oriented description of patients’ mobility problems, therefore offering an overall picture of their functioning in these activities.11 Thus, patients’ rehabilitation goals can be specifically defined according to the level of functioning on each 1 of the 5 components, and patterns of
List of abbreviations: CFA CFI ICC ICF IRT Mobam Mobam-in RMSEA TLI
confirmatory factor analysis comparative fit index intraclass correlation coefficient International Classification of Functioning, Disability and Health item response theory Mobility Activities Measure Mobility Activities Measure for Inpatient Rehabilitation Settings root mean square error of approximation Tucker-Lewis Index
these functioning components can be compared across clinical groups, either at admission or at discharge.14 The main objective of this article is to describe the development and initial psychometric evaluation of a mobility measure for inpatient postacute rehabilitation settingsdthe Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in)dby applying the Mobam framework. The specific aims are to examine (1) the dimensionality of the 5 Mobam components; (2) whether the selected items can form ordered interval scales; (3) test-retest reliability; and (4) known-groups validity.
Methods Study sample The sample included 239 inpatients who were receiving postacute care in the rehabilitation unit of a French public hospital. Patients were eligible if they were following a postacute rehabilitation program and were 18 years or older. Visual impairment or the inability to understand the simple instructions required for completing the questionnaires were defined as exclusion criteria. For descriptive purposes and to evaluate whether psychometric properties were consistent across diverse clinical groups, participants were classified into 3 major diagnostic groups: (1) musculoskeletal injuries (eg, fractures); (2) cardiopulmonary conditions (eg, pulmonary edema); and (3) medically complex pathologies (eg, stroke or other complex neurologic conditions).
Development of the tool Initial development was guided by integrating items from a wide variety of patient-oriented instruments to ICF categories of mobility activities. A literature review was conducted to identify items from health status measures, as described elsewhere.11 We included only mobility activities that could be reasonably tested in inpatient settings. For instance, the item “lying down” was considered relevant because inpatients were expected to be able to respond to different levels of difficulty concerning that item, whereas the item “walking long distances” was not relevant to patients in these settings and therefore was excluded. With regard to assigning items to ICF categories of mobility activities, this was performed by 2 health professionals working in coordination (mean of 6.8y of clinical experience plus ICF training) using established linkage rules.13,15 According to these rules, health professionals linked each concept identified in the qualitative analysis to the ICF category that represented this concept most precisely. A subset of items was selected for rewriting within the Mobam framework and, in order to cover those ICF categories not addressed by existing items, 12 new items were added. The 38 items included in this initial database were reviewed for clarity, usefulness, and nonrepetitive content by 2 measurement and content experts (mean of 10.2y of survey research experience). The selected items were subsequently pilot tested with professionals from the field of rehabilitation and inpatients. A total of 34 inpatients with mobility disabilities associated with musculoskeletal injuries, cardiopulmonary conditions, or medically complex pathologies were randomly selected from those receiving postacute care in a French hospital. Both professionals and inpatients were asked to report on the relevance and understanding of each item. Furthermore, patients were asked to think aloud as they worked with the items in order to come up with ideas for improving the content. Three items were discarded, as these proved to be unclear or poorly understood by patients.
www.archives-pmr.org
Mobility Activities Measure for Inpatient Rehabilitation Settings A core set of 35 items was chosen for inclusion within the Mobam-in measure for this field study. Core items included the following elements: 6 items relative to changing and maintaining body position involving only sitting and/or lying; 8 corresponding to changing and maintaining body position involving standing up; 6 concerning carrying and moving objects using the hand and shoulder; 8 aspects related to handling objects using only the hand and/or forearm; and 7 items corresponding to the walking and moving dimension. All the aforementioned items were comprised in a questionnaire, which also included an overall question that was phrased, “How much difficulty do you currently experience (without any help from another person or device) when pursuing the following activities?” A 5-point Likert scale was used to measure patients’ perceptions of difficulty with the following response options (scores range from 4 to 0, respectively): none, mild, moderate, severe, and unable to do it.
Data collection The institutional review board of the hospital approved the study protocol. Recruitment was carried out at admission to the inpatient rehabilitation unit. Patients were recruited and assessed by physical therapists trained in determining eligibility and data collection. All patients provided written informed consent before data were collected. A total of 246 eligible participants were invited to participate, among whom 7 (2.85%) declined participation. When patients were unable to complete the Mobam-in questionnaire themselves (nZ31), 2 forms of assistance were provided: (1) the questionnaire was administrated by a trained rehabilitation clinician, and (2) participants were given a response card with the 5 response options. At the time of admission, patients completed the Mobam-in questionnaire, while clinicians rated the severity of each patient’s disability and level of independence via the modified Rankin scale16 and the FIM,4 respectively. To examine knowngroups validity, participants were stratified into 3 distinct severitybased levels of disability (slight, moderate, severe), based on the Rankin scores, and 2 levels of independence (higher, lower) using the median value of FIM scores as the cutoff point. Medical records and self-reported methods were used to provide demographic data (age, sex, educational level). To evaluate test-retest reliability, a subgroup of 38 participants was randomly selected from our study of 239 inpatients. These patients completed the Mobam-in questionnaire a second time 2 to 3 days after their initial self-report.
Analyses A set of analyses was conducted with the aim of evaluating the multi-item scales of the Mobam-in. Multitrait scaling analysis and confirmatory factor analysis (CFA) were used for item reduction and to evaluate the unidimensionality of each scale.3,17 Item response theory (IRT) was subsequently used (1-parameter IRT model), as this methodological theory assumes unidimensionality. The adequacy of items to IRT models was tested, before their integration into an interval scale, for each 1 of the 5 different dimensions. Afterward, scoring metrics of Mobam-in scales were established. Test-retest reliability of the scales was estimated, and the “known-groups” method was also used to test validity. Sample size was based on common rules of thumb for determining the adequate N to apply CFA,18 Rasch polytomous models,19 and multitrait analysis.20 A sample size of N200 was required because it was the higher minimal number determined from these rules. www.archives-pmr.org
2369 Unidimensionality Multitrait scaling analysis17 was used initially to evaluate the unidimensionality of scales. This analysis uses a correlation matrix of all items and scales to test the extent to which items converge and diverge from other scales. Scales are therefore scored using the method of summated ratings, and correlations between an item and its hypothesized scale are corrected for overlap by not including the item in the scale when estimating the item-scale correlation. In this matrix, the statistical package SPSS version 19.0a was used to examine whether items were linearly related to the hypothesized scale (item internal consistency) and whether items correlated significantly with the hypothesized scale rather than with scales measuring other dimensions (item discriminant validity). Item internal consistency was assessed by corrected itemhypothesized scale correlations. A correlation of .40 was considered satisfactory; items with correlations
ORIGINAL ARTICLE
Development and Initial Psychometric Evaluation of the Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in) Francesc Medina-Mirapeix, PT, PhD,a Mariano Gacto-Sa´nchez, PT,b Esther Navarro-Pujalte, PT, PhD,c Joaquina Montilla-Herrador, PT, PhD,a Carmen Lillo-Navarro, PT, PhD,d Pilar Escolar-Reina, PT, PhDa From the aDepartment of Physical Therapy, Regional Campus of International Excellence “Campus Mare Nostrum,” University of Murcia, Murcia, Spain; bDepartment of Physical Therapy, EUSES University School, University of Girona, Girona, Spain; cDepartment of Education, Region of Murcia, Murcia, Spain; and dDepartment of Pathology and Surgery, University “Miguel Herna´ndez,” Alicante, Spain.
Abstract Objective: To describe the development and the initial psychometric evaluation of a mobility measure for inpatient postacute rehabilitation settingsdthe Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in). Design: Self-reportebased psychometric study. Setting: Postacute rehabilitation unit of a public hospital. Participants: A consecutive sample of inpatients (NZ239) receiving postacute rehabilitation care. Interventions: Not applicable. Main Outcome Measures: We developed a 30-item mobility measure, using the Mobility Activities Measure (Mobam) framework, to assess functioning across 5 mobility activity domains classified within the International Classification of Functioning, Disability and Health. These were (1) changing and maintaining body position involving only sitting and/or lying (4 items); (2) changing and maintaining body position involving standing up (6 items); (3) carrying and moving objects using the hand and shoulder (6 items); (4) handling objects using only the hand and/or forearm (7 items); and (5) walking and moving (7 items). Psychometric analyses were conducted to test assumptions underlying the scaling and scoring of Mobam-in scales, and to test both the reliability and validity. Results: Multitrait scaling and confirmatory factor analyses (with Tucker-Lewis Index median, .99; root mean square error of approximation median, .025) supported the assumption of unidimensionality concerning each domain. Five dimensions appeared to be stable across diverse diagnostic groups (the percentage of items with discriminant validity ranged from 93% to 100%, Cronbach coefficient ranged from .859 to .966). Rasch model (Masters’ partial credit) showed that all items could be located along a continuum in each dimension, with goodness-of-fit criteria of infit and outfit mean-square values between 0.6 and 1.4. Test-retest reliability was excellent (intraclass correlation coefficients median, .98). Groups with more severe conditions and lower functional independence scored lower on Mobam-in scales, as hypothesized. Conclusions: Mobam-in covers 5 dimensions of mobility activities. The Mobam framework is an effective reference for building outcome instruments. Archives of Physical Medicine and Rehabilitation 2014;95:2367-75 ª 2014 by the American Congress of Rehabilitation Medicine
Many hospitalized patients often require early postacute rehabilitation to optimize functioning after acute care.1,2 This postacute care may be provided either in specialized rehabilitation facilities
Disclosures: none.
or in dedicated units of an acute care hospital.1 A wide variety of rehabilitation outcome instruments for mobility and self-care activities are currently in use in postacute rehabilitation settings in order to evaluate the effectiveness of postacute programs across diagnostic groups.3-6 However, many of these clinical measures are neither comprehensive assessments for these activity domains
0003-9993/14/$36 - see front matter ª 2014 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2014.07.407
2368
F. Medina-Mirapeix et al
nor etiologically neutral to allow comparisons across diagnostic groups.7,8 Most importantly, most of the current measures use raw scores from ordinal scales and are therefore not qualified for parametric statistics such as calculating means or SDs.9 Thus, when considering inpatient rehabilitation settings, there is an important need to develop interval-scaled clinical measures that provide a reliable estimation of specific inpatient problems across diagnostic groups. The International Classification of Functioning, Disability and Health (ICF)10 provides a conceptual framework and classification system for developing comprehensive outcome instruments based on activities.3,6,11 The ICF principle of grouping activities with similar purposes (eg, caring for body parts) is theoretically understandable. However, it has been suggested that unidimensionality, a fundamental aspect of measurement theory, is difficult to achieve for ICF self-care activities because of the varying underlying sequences of movement (which are completely different, for instance, for brushing teeth as compared with cutting toenails).6 In a previous study,11 we showed that the unidimensionality of mobility activities depends on the movement sequences (eg, use of the hand, forearm, or both). Five separate mobility dimensions were identified from the Mobility Activities chapter of the ICF: (1) changing and maintaining body position involving only sitting and/or lying (eg, turning in bed); (2) changing and maintaining body position involving standing up (eg, standing 15min); (3) carrying and moving objects using the hand and shoulder (eg, hanging a hanger in a closet); (4) handling objects using only the hand and/or forearm (eg, turning a key); and (5) walking and moving (eg, climbing stairs). The 5 aforementioned dimensions were defined as a conceptual framework, which we refer to as the Mobility Activities Measure (Mobam).11 Within the context of postacute care, there is a strong consensus for the need to develop instruments for monitoring functional recovery across the continuum of care settings.3 Several solutions have been suggested, including paired instruments created for outcome assessment across inpatients and community settings.12 Our research group has developed a set of 5 short mobility measures, called Mobam, that can be used to measure outcomes in outpatient rehabilitation settings.11 The next phase of our research, which is the focus of the present article, examined the ability of the Mobam framework to develop a paired tool for inpatients. Most of the currently available clinical measures generally provide only limited coverage of the ICF mobility domain.6,13 A significant advantage of developing Mobam-based clinical measures is that these provide an ICF-oriented description of patients’ mobility problems, therefore offering an overall picture of their functioning in these activities.11 Thus, patients’ rehabilitation goals can be specifically defined according to the level of functioning on each 1 of the 5 components, and patterns of
List of abbreviations: CFA CFI ICC ICF IRT Mobam Mobam-in RMSEA TLI
confirmatory factor analysis comparative fit index intraclass correlation coefficient International Classification of Functioning, Disability and Health item response theory Mobility Activities Measure Mobility Activities Measure for Inpatient Rehabilitation Settings root mean square error of approximation Tucker-Lewis Index
these functioning components can be compared across clinical groups, either at admission or at discharge.14 The main objective of this article is to describe the development and initial psychometric evaluation of a mobility measure for inpatient postacute rehabilitation settingsdthe Mobility Activities Measure for Inpatient Rehabilitation Settings (Mobam-in)dby applying the Mobam framework. The specific aims are to examine (1) the dimensionality of the 5 Mobam components; (2) whether the selected items can form ordered interval scales; (3) test-retest reliability; and (4) known-groups validity.
Methods Study sample The sample included 239 inpatients who were receiving postacute care in the rehabilitation unit of a French public hospital. Patients were eligible if they were following a postacute rehabilitation program and were 18 years or older. Visual impairment or the inability to understand the simple instructions required for completing the questionnaires were defined as exclusion criteria. For descriptive purposes and to evaluate whether psychometric properties were consistent across diverse clinical groups, participants were classified into 3 major diagnostic groups: (1) musculoskeletal injuries (eg, fractures); (2) cardiopulmonary conditions (eg, pulmonary edema); and (3) medically complex pathologies (eg, stroke or other complex neurologic conditions).
Development of the tool Initial development was guided by integrating items from a wide variety of patient-oriented instruments to ICF categories of mobility activities. A literature review was conducted to identify items from health status measures, as described elsewhere.11 We included only mobility activities that could be reasonably tested in inpatient settings. For instance, the item “lying down” was considered relevant because inpatients were expected to be able to respond to different levels of difficulty concerning that item, whereas the item “walking long distances” was not relevant to patients in these settings and therefore was excluded. With regard to assigning items to ICF categories of mobility activities, this was performed by 2 health professionals working in coordination (mean of 6.8y of clinical experience plus ICF training) using established linkage rules.13,15 According to these rules, health professionals linked each concept identified in the qualitative analysis to the ICF category that represented this concept most precisely. A subset of items was selected for rewriting within the Mobam framework and, in order to cover those ICF categories not addressed by existing items, 12 new items were added. The 38 items included in this initial database were reviewed for clarity, usefulness, and nonrepetitive content by 2 measurement and content experts (mean of 10.2y of survey research experience). The selected items were subsequently pilot tested with professionals from the field of rehabilitation and inpatients. A total of 34 inpatients with mobility disabilities associated with musculoskeletal injuries, cardiopulmonary conditions, or medically complex pathologies were randomly selected from those receiving postacute care in a French hospital. Both professionals and inpatients were asked to report on the relevance and understanding of each item. Furthermore, patients were asked to think aloud as they worked with the items in order to come up with ideas for improving the content. Three items were discarded, as these proved to be unclear or poorly understood by patients.
www.archives-pmr.org
Mobility Activities Measure for Inpatient Rehabilitation Settings A core set of 35 items was chosen for inclusion within the Mobam-in measure for this field study. Core items included the following elements: 6 items relative to changing and maintaining body position involving only sitting and/or lying; 8 corresponding to changing and maintaining body position involving standing up; 6 concerning carrying and moving objects using the hand and shoulder; 8 aspects related to handling objects using only the hand and/or forearm; and 7 items corresponding to the walking and moving dimension. All the aforementioned items were comprised in a questionnaire, which also included an overall question that was phrased, “How much difficulty do you currently experience (without any help from another person or device) when pursuing the following activities?” A 5-point Likert scale was used to measure patients’ perceptions of difficulty with the following response options (scores range from 4 to 0, respectively): none, mild, moderate, severe, and unable to do it.
Data collection The institutional review board of the hospital approved the study protocol. Recruitment was carried out at admission to the inpatient rehabilitation unit. Patients were recruited and assessed by physical therapists trained in determining eligibility and data collection. All patients provided written informed consent before data were collected. A total of 246 eligible participants were invited to participate, among whom 7 (2.85%) declined participation. When patients were unable to complete the Mobam-in questionnaire themselves (nZ31), 2 forms of assistance were provided: (1) the questionnaire was administrated by a trained rehabilitation clinician, and (2) participants were given a response card with the 5 response options. At the time of admission, patients completed the Mobam-in questionnaire, while clinicians rated the severity of each patient’s disability and level of independence via the modified Rankin scale16 and the FIM,4 respectively. To examine knowngroups validity, participants were stratified into 3 distinct severitybased levels of disability (slight, moderate, severe), based on the Rankin scores, and 2 levels of independence (higher, lower) using the median value of FIM scores as the cutoff point. Medical records and self-reported methods were used to provide demographic data (age, sex, educational level). To evaluate test-retest reliability, a subgroup of 38 participants was randomly selected from our study of 239 inpatients. These patients completed the Mobam-in questionnaire a second time 2 to 3 days after their initial self-report.
Analyses A set of analyses was conducted with the aim of evaluating the multi-item scales of the Mobam-in. Multitrait scaling analysis and confirmatory factor analysis (CFA) were used for item reduction and to evaluate the unidimensionality of each scale.3,17 Item response theory (IRT) was subsequently used (1-parameter IRT model), as this methodological theory assumes unidimensionality. The adequacy of items to IRT models was tested, before their integration into an interval scale, for each 1 of the 5 different dimensions. Afterward, scoring metrics of Mobam-in scales were established. Test-retest reliability of the scales was estimated, and the “known-groups” method was also used to test validity. Sample size was based on common rules of thumb for determining the adequate N to apply CFA,18 Rasch polytomous models,19 and multitrait analysis.20 A sample size of N200 was required because it was the higher minimal number determined from these rules. www.archives-pmr.org
2369 Unidimensionality Multitrait scaling analysis17 was used initially to evaluate the unidimensionality of scales. This analysis uses a correlation matrix of all items and scales to test the extent to which items converge and diverge from other scales. Scales are therefore scored using the method of summated ratings, and correlations between an item and its hypothesized scale are corrected for overlap by not including the item in the scale when estimating the item-scale correlation. In this matrix, the statistical package SPSS version 19.0a was used to examine whether items were linearly related to the hypothesized scale (item internal consistency) and whether items correlated significantly with the hypothesized scale rather than with scales measuring other dimensions (item discriminant validity). Item internal consistency was assessed by corrected itemhypothesized scale correlations. A correlation of .40 was considered satisfactory; items with correlations
