Abstract
Objectives: To evaluate the internal consistency of the Mayo-Portland Adaptability Inventory (MPAI), further refine the instrument, and provide reference data based on a large, geographically diverse sample of persons with acquired brain injury (ABI).
Subjects: 386 persons, most with moderate to severe ABI.
Settings: Outpatient, community-based, and residential rehabilitation facilities for persons with ABI located in the United States: West, Midwest, and Southeast.
Methods: Rasch, item cluster, principal components, and traditional psychometric analyses for internal consistency of MPAI data and subscales.
Results: With rescoring of rating scales for 4 items, a 29-item version of the MPAI showed satisfactory internal consistency by Rasch (Person Reliability = .88; Item Reliability = .99) and traditional psychometric indicators (Cronbach's alpha = .89). Three rationally derived subscales for Ability, Activity, and Participation demonstrated psychometric properties that were equivalent to subscales derived empirically through item cluster and factor analyses. For the 3 subscales, Person Reliability ranged from .78 to .79; Item Reliability, from .98 to .99; and Cronbach's alpha, from .76 to .83. Subscales correlated moderately (Pearson r = .49-.65) with each other and strongly with the overall scale (Pearson r = .82-.86).
Conclusions: Outcome after ABI is represented by the unitary dimension described by the MPAI. MPAI subscales further define regions of this dimension that may be useful for evaluation of clinical cases and program evaluation.
IN 1987, Lezak 1 developed the original Portland Adaptability Inventory to provide a scale for meaningful documentation of a variety of behavioral and social challenges that many persons with acquired brain injury (ABI) face. Malec and Thompson 2 subsequently attempted to further refine this instrument, adding items for rating pain and specific areas of cognitive impairment. The resulting scale, the Mayo-Portland Adaptability Inventory (MPAI),* used item rating categories that focused on current functional ability without reference to preinjury level. Reflecting World Health Organization (WHO) distinctions 3 among impairment, activity, and participation, ratings on most MPAI items are designed to indicate whether performance is (a) within normal limits, (b) mildly limited but not to a degree that interferes significantly with everyday functioning (impairment only), or (c) sufficiently limited that it does interfere with everyday functioning to varying degrees (restriction of activity and participation). While this rating scheme could be applied to most items, those items that more directly measure participation, such as employment or independent living, did not lend themselves well to this rating system and were scaled in terms of extent of participation.
Also reflecting the WHO nosology, further development of the MPAI has been based on the assumption that an accurate characterization of persons with ABI and their outcomes requires assessment of key indicators of ability, activity, and participation. Rating scale (Rasch) analyses of developing versions of the MPAI have supported this model. 4 These Rasch analyses of the MPAI demonstrate that items that are relevant to each of the domains of ability, activity, and participation form a single dimension in relationship to the level of overall severity of negative outcome after ABI. Moderately strong intercorrelations between subscales derived from a previous version (MPAI 2.3) also demonstrate this overlap between dimensions of impairment and activity/participation. The Physical/Cognitive Impairment Subscale included only impairment (or more positively stated-ability) items but was substantially correlated (r = .75) with the Social Participation Subscale that focused on items representing activity and participation. Rasch analysis of other outcome measures included in the NIDRR (National Institute on Disability and Rehabilitation Research) TBI (traumatic brain injury) Model System database has also supported a single dimension of outcome that includes indicators of ability, activity, and participation (M.V. Johnston et al, unpublished data).
Forms of the MPAI for completion by persons with ABI and their significant others have been developed and studied. 5,6 However, research reported in the current article focuses on the Staff version of the MPAI. Initial Rasch analyses of the Staff MPAI 4 demonstrated satisfactory internal consistency (Person Reliability = .82; Person Separation = 2.12; Item Reliability = .99; Item Separation = 9.33). Following this initial analysis, the MPAI was further refined as the MPAI 2.3. The MPAI 2.3 used a 6-point rating scale throughout. Some items in the original version (specifically, Psychotic Symptoms, Alcohol Use, Drug Use, Law Violations) were found not to contribute to the measurement of outcome of ABI; these items were retained at the end of the MPAI 2.3 because, when present, they offer information that is important in developing service plans. However, these items do not contribute to the MPAI score because they are not specific measures of ABI sequelae or outcome.
Rasch analyses of the MPAI 2.3 for 126 cases from the Mayo Medical Center and Bancroft Rehabilitation 7 revealed excellent Person Reliability (.92), Person Separation (3.49), Item Reliability (.95), and Item Separation (4.54). A 5-point scale appeared to best represent the range of individual item ratings. Three subscales were identified in analyses of MPAI 2.3 data: the physical/cognitive impairment subscale, the social participation subscale, and the pain/emotional disorder subscale. Although the pain/emotional disorder subscale appeared to contribute information distinct from the other subscales, it was composed of only 4 items and consequently lacked sufficient internal consistency for formal scoring. These analyses also resulted in further refinements and development of another version of the measure, the MPAI-3. New items (Fatigue, Dizziness/Balance, Sensitivity to Mild Symptoms, and Managing Money and Finances) were added to this version to better represent the milder end of challenges for persons with ABI.
Concurrent and predictive validity of the Staff MPAI has been demonstrated in a number of studies. The Staff MPAI correlates moderately well with the Disability Rating Scale, Rancho Scale, neuropsychological measures, and a form of the MPAI completed independently by a significant other of the person with ABI who is being evaluated. 2,8 Malec et al 9 reported that Staff MPAI and time since injury were the best predictors (69% correct classification) of vocational placement resulting from participation in a vocational rehabilitation program designed specifically for persons with ABI. In another study 10 it was found that Staff MPAI was the best predictor of long-term vocational and independent living outcome, following a comprehensive day rehabilitation program for persons with ABI. Logistic regression analysis, including age, education, severity of injury, traumatic vs nontraumatic injury, time since injury, and Rasch-converted staff MPAI score as potential predictors, showed that the MPAI alone predicted vocational status (67% correct classification) and independent living (70% correct classification) status 1-year after program completion. Malec and Degiorgio 11 reported that logistic regression of the MPAI could be used to estimate the probability of community-based employment as a result of either comprehensive or limited outpatient rehabilitation and vocational intervention following ABI. In this study, evaluation with the MPAI was shown to be a potentially useful method for determining (in conjunction with time since injury) whether limited (a few hours per week) or extensive day treatment would be required for successful vocational reintegration. In other studies, 5,6,10 functional evaluation with the MPAI has been shown to assist in rehabilitation planning and to be useful in negotiating rehabilitation goals for people with ABI and their significant others. In these studies, results of reevaluation with the MPAI following rehabilitation intervention correlated with other outcome indicators, ie, extent of independence in living and work.
The study reported here examined the internal consistency of the MPAI-3 in a larger group of persons with ABI representing a broad range of injury severity and geographical diversity served through Learning Services Corporation, Mayo Medical Center, and Rehab Without Walls. Objectives of this study were as follows:
1. Examine the internal consistency of the MPAI-3.
2. Examine individual item fit and the appropriateness of individual item rating scales.
3. Using multiple methods (Rasch, item cluster, and factor analyses), identify viable subscales within the overall measure.
4. Provide reference data based on a large geographically diverse group of persons who represent a broad range of severity of ABI.