Advertisement
Original research| Volume 99, ISSUE 8, P1599-1608.e1, August 2018

Download started.

Ok

Quality of Life and Adaptation in People With Spinal Cord Injury: Response Shift Effects From 1 to 5 Years Postinjury

Open AccessPublished:February 23, 2018DOI:https://doi.org/10.1016/j.apmr.2018.01.028

      Highlights

      • We detected recalibration and reconceptualization response shift effects from 1 to 5 years follow-up in a robust sample of people with spinal cord injury.
      • Despite stable motor and cognitive status, people with spinal cord injury are adapting to their condition.
      • This adaptation reflects a progressive disconnection between symptoms and physical or mental health, and a real improvement in physical functioning over time.

      Abstract

      Objective

      To investigate response shift effects in spinal cord injury (SCI) over 5 years postinjury.

      Design

      Prospective cohort study observed at 1, 2, and 5 years post-SCI.

      Setting

      Specialized SCI centers.

      Participants

      Sample included 1125, 760, and 219 participants at 1, 2, and 5 years post-SCI (N = 2104). The study sample was 79% men; 39% were motor/sensory complete (mean age, 44.6±18.3y).

      Interventions

      Not applicable.

      Main Outcome Measures

      Patient-reported outcomes included the Medical Outcomes Study 36-Item Short-Form Health Survey version 2 and the Life Satisfaction-11 Questionnaire. Participant latent variable scores were adjusted for (1) potential attrition bias and (2) propensity scores reflecting risk of worse outcomes. The Oort structural equation modeling approach for detecting and accounting for response shift effects was used to test the hypothesis that people with SCI would undergo response shifts over follow-up.

      Results

      The study data comprised the time after FIM scores, an objective measure of motor and cognitive function, had improved and stabilized. Three latent variables (Physical, Mental, and Symptoms) were modeled over time. The response shift model indicated uniform recalibration and reconceptualization response shift effects over time. When adjusted for these response shift effects, Physical showed small true change improvements at 2- and 5-year follow-up, despite FIM stability.

      Conclusions

      We detected recalibration and reconceptualization response shift effects in 1- to 5-year follow-up of people with SCI. Despite stable motor and cognitive function, people with SCI are adapting to their condition. This adaptation reflects a progressive disconnection between symptoms and physical or mental health, and a real improvement in the Physical latent variable.

      Keywords

      List of abbreviations:

      AIS (ASIA Impairment Scale), CFI (comparative fit index), LiSAT (Life Satisfaction-11 Questionnaire), QOL (quality of life), RMSEA (root mean square error approximation), SCI (spinal cord injury), SEM (structural equation modeling), SF-36v2 (Medical Outcomes Study 36-Item Short-Form Health Survey version 2), TLI (Tucker-Lewis Index)
      Traumatic spinal cord injury (SCI) is a devastating and life-altering health change for an individual.
      • Dumont R.J.
      • Okonkwo D.O.
      • Verma S.
      • et al.
      Acute spinal cord injury, part I: pathophysiologic mechanisms.
      The abrupt onset of this event requires one to accept a new life reality with no preparation.
      • Yoshida K.K.
      Reshaping of self: a pendular reconstruction of self and identity among adults with traumatic spinal cord injury.
      SCI has profound physical and emotional consequences; therefore, finding a way to adapt is critical to living with SCI.
      • Hammell K.W.
      Psychological and sociological theories concerning adjustment to traumatic spinal cord injury: the implications for rehabilitation.
      Improving the quality of life (QOL) of people with SCI extends beyond the World Health Organization concept of health (ie, physical, emotional, social functioning
      World Health Organization
      Constitution of the World Health Organization.
      ) to include eudemonic well-being
      • Ryff C.D.
      Happiness is everything, or is it? Explorations on the meaning of psychological well-being.
      (ie, having a sense of meaning in one’s life, self-actualization
      • Ryan R.M.
      • Deci E.L.
      On happiness and human potentials: a review of research on hedonic and eudaimonic well-being.
      ). In addition to minimizing complications and reintegrating people with SCI into society, effective rehabilitation focused on QOL will aim to assist them to activate coping strategies to enable them to function and enjoy life to the fullest.
      Paradoxical findings have been documented regarding well-being. Bach and Tilton
      • Bach J.R.
      • Tilton M.C.
      Life satisfaction and well-being measures in ventilator assisted individuals with traumatic tetraplegia.
      documented a level of reported life satisfaction among people with SCI equivalent to or above the level reported by people with no health impairments whatsoever. It is possible that people with SCI are adapting to their condition, and that these response shift effects are hidden below the surface of standard measurement approaches. For example, as the person with SCI struggles to learn to live in his or her new body, he or she must change how he or she defines work or other activities, what normative activities are, what severe fatigue means or feels like, what matters in QOL, and what QOL even means post-SCI. All of these changes may not be apparent from standard patient-reported outcome measures because these ways of thinking about answering the evaluative questions are not usually measured.
      Response shift effects have been detected in patient populations where a health state change (catalyst) has occurred.
      • Sprangers M.A.
      • Schwartz C.E.
      Integrating response shift into health-related quality of life research: a theoretical model.
      • Schwartz C.E.
      • Sprangers M.A.
      Methodological approaches for assessing response shift in longitudinal health-related quality-of-life research.
      • Schwartz C.E.
      • Sprangers M.A.
      • Fayers P.
      You know it's there but how do you capture it? Challenges for the next phase of response shift research.
      • Schwartz C.E.
      • Ahmed S.
      • Sawatsky R.
      • et al.
      Guidelines for secondary analysis in search of response shift.
      Response shift theory
      • Sprangers M.A.
      • Schwartz C.E.
      Integrating response shift into health-related quality of life research: a theoretical model.
      • Rapkin B.D.
      • Schwartz C.E.
      Toward a theoretical model of quality-of-life appraisal: implications of findings from studies of response shift.
      postulates that stable characteristics of the individual (antecedents) interact with cognitive, affective, and behavioral strategies (mechanisms) to influence cognitive appraisal processes (appraisal), which buffer the effect of the health state change on perceived QOL (response shift effects). Research to date has documented 3 aspects of response shift: (1) recalibration (change in internal standards), (2) reprioritization (change in values), and (3) reconceptualization (change in the individual’s concept of the target construct, for example, QOL).
      • Schwartz C.E.
      • Sprangers M.A.
      • Carey A.
      • Reed G.
      Exploring response shift in longitudinal data.
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
      • Finkelstein J.A.
      • Quaranto B.R.
      • Schwartz C.E.
      Threats to the internal validity of spinal surgery outcome assessment: recalibration response shift or implicit theories of change?.
      • King-Kallimanis B.L.
      • Oort F.J.
      • Nolte S.
      • Schwartz C.E.
      • Sprangers M.A.
      Using structural equation modeling to detect response shift in disability and QOL scores of multiple sclerosis patients.
      • Li Y.
      • Rapkin B.D.
      Classification and regression tree analysis to identify complex cognitive paths underlying quality of life response shifts: a study of individuals living with HIV/AIDS.
      • Schwartz C.E.
      • Sprangers M.A.
      • Ahmed S.
      • et al.
      Response shift in patients with multiple sclerosis: an application of three statistical techniques.
      This psychological process allows the patient to maintain a personal steady state, resulting in perceived QOL scores (observed scores) that are higher than would be expected, given their physical deterioration (expected scores). Response shift effects can be direct (ie, changes in appraisal affect QOL ratings directly) or moderated (ie, changes in appraisal affect QOL ratings by attenuating the effect of catalysts).
      Response shift has been documented in a number of chronic conditions
      • Schwartz C.E.
      • Sprangers M.A.
      • Carey A.
      • Reed G.
      Exploring response shift in longitudinal data.
      • Finkelstein J.A.
      • Quaranto B.R.
      • Schwartz C.E.
      Threats to the internal validity of spinal surgery outcome assessment: recalibration response shift or implicit theories of change?.
      • Li Y.
      • Rapkin B.D.
      Classification and regression tree analysis to identify complex cognitive paths underlying quality of life response shifts: a study of individuals living with HIV/AIDS.
      • Verdam M.G.
      • Oort F.J.
      • van der Linden Y.M.
      • Sprangers M.A.
      Taking into account the impact of attrition on the assessment of response shift and true change: a multigroup structural equation modeling approach.
      and has been considered in the context of SCI-related disability over the life course.
      • Schwartz C.E.
      • Andresen E.M.
      • Nosek M.A.
      • Krahn G.L.
      Response shift theory: important implications for measuring quality of life in people with disability.
      Research has generally detected small albeit meaningful effects on QOL outcomes,
      • Schwartz C.E.
      • Bode R.
      • Repucci N.
      • Becker J.
      • Sprangers M.A.
      • Fayers P.M.
      The clinical significance of adaptation to changing health: a meta-analysis of response shift.
      often regarding deterioration but also regarding improvement
      • Finkelstein J.A.
      • Razmjou H.
      • Schwartz C.E.
      Response shift and outcome assessment in orthopedic surgery: is there is a difference between complete vs. partial treatment?.
      or stability.
      • Schwartz C.E.
      • Quaranto B.R.
      • Rapkin B.D.
      • Healy B.C.
      • Vollmer T.
      • Sprangers M.A.
      Fluctuations in appraisal over time in the context of stable and non-stable health.
      Response shift is conceptualized as an effect rather than as a construct one measures directly.
      • Schwartz C.E.
      • Ahmed S.
      • Sawatsky R.
      • et al.
      Guidelines for secondary analysis in search of response shift.
      This distinction has implications for how researchers would operationalize it and report their findings (eg, response shift is not adaptation, but rather an effect of adaptation
      • Schwartz C.E.
      • Ahmed S.
      • Sawatsky R.
      • et al.
      Guidelines for secondary analysis in search of response shift.
      ). Characterizing response shift effects in treatment outcomes can enable distinguishing adaptation effects from objective change (ie, true or unbiased change that does not violate the psychometric assumption of measurement invariance).
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      Measurement and conceptual perspectives on response shift: formal definitions of measurement bias, explanation bias, and response shift.
      SCI is different from a chronic deteriorating condition (eg, cancer, multiple sclerosis) in that it is a stable physical condition over time, not one requiring continual adaptation to a ‘moving goal post'. Therefore, the catalyst after SCI might be better characterized as adapting to the new reality, once functional improvement has stabilized (ie, 1 year postinjury
      • Hall K.M.
      • Cohen M.E.
      • Wright J.
      • Call M.
      • Werner P.
      Characteristics of the Functional Independence Measure in traumatic spinal cord injury.
      ). SCI may therefore be a good model to study response shift effects because there are no moving goal posts and response shift effects can be studied over a longer period of time. Understanding whether these response shift effects are present and how they affect a patient’s perspective on their functional health may provide useful insight to the SCI clinical team and families. When the outcomes are more subjective (eg, purpose in life), real change seems less distinctive from ‘response shift’. However, when we are looking at outcomes (eg, physical function), there are relevant differences between ‘response shift’ and ‘true change’.
      Whereas response shift effects have been suspected in patients with SCI as early as the mid-1990s,
      • Bach J.R.
      • Tilton M.C.
      Life satisfaction and well-being measures in ventilator assisted individuals with traumatic tetraplegia.
      no empirical research has been done to date evaluating response shift effects in SCI. A recent Ovid search revealed a large amount of literature on coping and SCI, and a number of articles mention response shift as a possible explanation for their findings or as relevant to their review.
      • Bruno M.A.
      • Bernheim J.L.
      • Ledoux D.
      • Pellas F.
      • Demertzi A.
      • Laureys S.
      A survey on self-assessed well-being in a cohort of chronic locked-in syndrome patients: happy majority, miserable minority.
      • Dibb B.
      • Ellis-Hill C.
      • Donovan-Hall M.
      • Burridge J.
      • Rushton D.
      Exploring positive adjustment in people with spinal cord injury.
      • Dijkers M.P.
      Quality of life of individuals with spinal cord injury: a review of conceptualization, measurement, and research findings.
      • Tate D.G.
      • Kalpakjian C.Z.
      • Forchheimer M.B.
      Quality of life issues in individuals with spinal cord injury.
      • Post M.W.
      • van Leeuwen C.M.
      • van Koppenhagen C.F.
      • de Groot S.
      Validity of the Life Satisfaction questions, the Life Satisfaction Questionnaire, and the Satisfaction With Life Scale in persons with spinal cord injury.
      Only 1 article included a putative measure of response shift: a retrospective-pretest item in a life satisfaction questionnaire.
      • van Leeuwen C.M.
      • Post M.W.
      • van der Woude L.H.
      • et al.
      Changes in life satisfaction in persons with spinal cord injury during and after inpatient rehabilitation: adaptation or measurement bias?.
      This approach uses a then-test item which asks the respondent at follow-up to rate their baseline health or QOL, and then computes a recalibration response shift score by subtracting then-minus-baseline on the then-test item(s). The limitations of this then-test method are increasingly acknowledged (eg, recall bias, implicit theories of change).
      • Schwartz C.E.
      • Sprangers M.A.
      Guidelines for improving the stringency of response shift research using the then-test.
      Structural equation modeling (SEM) is a widely used technique for investigating response shift effects in many patient populations.
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
      • King-Kallimanis B.L.
      • Oort F.J.
      • Nolte S.
      • Schwartz C.E.
      • Sprangers M.A.
      Using structural equation modeling to detect response shift in disability and QOL scores of multiple sclerosis patients.
      • Ahmed S.
      • Sawatzky R.
      • Levesque J.F.
      • Ehrmann-Feldman D.
      • Schwartz C.E.
      Minimal evidence of response shift in the absence of a catalyst.
      • Ahmed S.
      • Bourbeau J.
      • Maltais F.
      • Mansour A.
      The Oort structural equation modeling approach detected a response shift after a COPD self-management program not detected by the Schmitt technique.
      • Barclay-Goddard R.
      • Lix L.M.
      • Tatec R.
      • Weinberg L.
      • Mayo N.E.
      Response shift was identified over multiple occasions with a structural equation modeling framework.
      • Visser M.
      • Oort F.J.
      • Sprangers M.
      Assessing response shift in QL: the thentest and a structural equation modeling approach compared.
      In an SEM context, Oort et al
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
      • Oort F.J.
      Using structural equation modeling to detect response shifts and true change.
      describes response shifts as special cases of measurement invariance which are operationalized as across-occasion differences between patterns of factor correlations (higher level reconceptualization), differences between values of common factor loadings (reprioritization), differences between intercepts (uniform recalibration), and differences between residual variances (nonuniform recalibration). In other words, the presence of response shift is detected via differences in SEM model parameters over time (eg, intercepts) that need to be accounted for to properly evaluate true change in latent variables over time. We therefore seek to investigate the presence of response shift and true changes in QOL in this secondary analysis of data collected from people with SCI.

      Methods

      Participants and design

      Participants

      This secondary analysis investigates response shift effects in a cohort of people with SCI participating in the Rick Hansen SCI Registry
      • Noonan V.
      • Kwon B.
      • Soril L.
      • et al.
      The Rick Hansen spinal cord injury registry (RHSCIR): a national patient-registry.
      between 2004 and 2014 at 1, 2, and 5 years post-injury. Therefore, the cohort study data focused on the time after the sample’s physical functioning had improved and likely stabilized,
      • Hall K.M.
      • Cohen M.E.
      • Wright J.
      • Call M.
      • Werner P.
      Characteristics of the Functional Independence Measure in traumatic spinal cord injury.
      but there is likely still mental, social, and spiritual adaptation taking place. Table 1 shows the sample demographics. The sample included 1125 people; of these, 219 had data at all years, 540 had data at years 1 and 2 only, and 366 had data only at year 1. Based on the International Standards for the Neurological Classification of Spinal Cord Injury
      • Kirshblum S.C.
      • Burns S.P.
      • Biering-Sorensen F.
      • et al.
      International standards for neurological classification of spinal cord injury (revised 2011).
      and the ASIA Impairment Scale (AIS),
      • Kreutzer J.
      • DeLuca J.
      • Caplan B.
      (Eds.). Encyclopedia of clinical neuropsychology.
      39% of the sample had AIS grade A, 11% had AIS grade B, and 51% of the sample had AIS grades C or D (supplemental appendix S1, available online only at http://www.archives-pmr.org/). All sites obtained local institutional review board approval before recruiting, and patients provided written informed consent (Rick Hansen SCI Registry
      • Noonan V.
      • Kwon B.
      • Soril L.
      • et al.
      The Rick Hansen spinal cord injury registry (RHSCIR): a national patient-registry.
      provides methods used).
      Table 1Baseline demographic characteristics (N=1125)
      CharacteristicN%
      Age at injury
       <18280.02
       18–343670.33
       35–441570.14
       45–541930.17
       55–641860.17
       65–741390.12
       ≥75490.04
       Missing60.01
      Sex
       Male8850.79
       Female2400.21
       MissingNANA
      Race/ethnicity
       White9390.83
       Asian370.03
       Aboriginal430.04
       Caribbean220.02
       Other580.05
       Missing260.02
      Education
       Less than high school2600.23
       High school graduate4190.37
       Some college2250.20
       Four-year degree or more1490.13
       Other290.03
       Missing430.04
      AIS grade
      See supplemental appendix S1 for definition of AIS grades. Abbreviation: NA, not applicable.
       A3780.34
       B1030.09
       C1820.16
       D3170.28
       Missing1450.13
      See supplemental appendix S1 for definition of AIS grades. Abbreviation: NA, not applicable.

      Measures

      This analysis operationalized QOL to include the 8 domain scores of the Medical Outcomes Study 36-Item Short-Form Health Survey version 2 (SF-36v2) (general health, physical functioning, role physical, social functioning, role emotional, mental health, pain, and vitality)
      • Ware J.E.
      • Kosinski M.
      • Dewey J.E.
      How to Score Version 2 of the SF-36(R) Health Survey.
      and the Life Satisfaction-11 Questionnaire (LiSAT).
      • Fugl-Meyer A.R.
      • Melin R.
      • Fugl-Meyer K.S.
      Life satisfaction in 18-to 64-year-old Swedes: in relation to gender, age, partner and immigrant status.
      The FIM
      • Heinemann A.W.
      • Michael Linacre J.
      • Wright B.D.
      • Hamilton B.B.
      • Granger C.
      Measurement characteristics of the functional independence measure.
      motor, cognitive, and total scores were used to assess objective functioning over time because this measure is self-reported but clinician-scored in the community setting. The FIM was administered by clinicians during inpatient rehabilitation at admission and discharge. Once out in the community, the FIM was done over the phone by a clinical study coordinator on site at the respondent’s home. The scoring for the 2 methods is the same (ie, adding up the 1–7 score on each of the 18 items). FIM data were available on some participants at admission to rehabilitation and at discharge and during the study period (1, 2, and 5y postinjury). Patient-reported outcome data were collected by telephone interview or during a clinical follow-up visit by trained staff at 1, 2, and 5 years postinjury. In contrast with the SF-36 and LISAT, which measure subjective functioning (experienced health problems and satisfaction, respectively), the FIM measures objective functioning (level of independence in performing activities).

      Statistical analysis

      SEM analysis
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
      • Oort F.J.
      Using structural equation modeling to detect response shifts and true change.
      was used to investigate response shift effects in the data. In preparation for this multivariable analysis, we began by examining in the year 1 data the factor structure of the measures to be used. The SF-36v2 items fit the published 8-domain factor structure (root mean square error approximation [RMSEA]=.075, standardized root mean square residual=.028, comparative fit index [CFI]=.967, Tucker-Lewis Index [TLI]=.961; data not shown),
      • Ware J.E.
      • Kosinski M.
      • Dewey J.E.
      How to Score Version 2 of the SF-36(R) Health Survey.
      as did the FIM items (RMSEA=.047, CFI=.998, TLI=.998; data not shown).
      • Heinemann A.W.
      • Michael Linacre J.
      • Wright B.D.
      • Hamilton B.B.
      • Granger C.
      Measurement characteristics of the functional independence measure.
      In contrast, the LiSAT did not fit the published 4-factor structure.
      • Fugl-Meyer A.R.
      • Melin R.
      • Fugl-Meyer K.S.
      Life satisfaction in 18-to 64-year-old Swedes: in relation to gender, age, partner and immigrant status.
      After examining a number of models, the best-fitting model was a 1-factor model that included only 5 LiSAT items: contact with friends, psychological health, manage self-care, leisure situation, and vocational situation (RMSEA=.058, CFI=.992, TLI=.985; data not shown). According to model fit indices, this 1-factor solution more closely fits the data compared with the 2- and 3-factor models, and because none of the kept 5 items had residual correlations with other items (ie, were redundant, the item with the highest factor loading was generally kept). The LiSAT score was rescaled to range from 0 to 100, the same metric as the other measures included in the QOL measurement model. Structural equation models fit using Mplus
      • Muthen L.K.
      • Muthen B.O.
      Mplus user's guide.
      ,a treated QOL subscales as continuously distributed indicator variables.
      The response shift SEM analysis proceeds in 4 steps. Step 1 is the measurement model: an appropriate measurement model is selected on the basis of an exploratory factor analysis. This model has no across-occasion constraints, and all latent variable means are constrained to zero across time. In other words, this model tests the hypothesis that there are response shift effects (eg, factor loadings and intercepts are allowed to be different at each time point). Step 2 is the no response shift model. In this model all response shift parameters (ie, factor loadings, intercepts, residual variances) are constrained to be equal across occasions. In contrast with the model in step 1, the model in step 2 tests the hypothesis that there are not response shift effects. The fit of models 1 and 2 are then compared using a chi-square difference test. If the difference in fit in these 2 models is not significant, we conclude that there are no response shift effects and skip step 3. If the difference is significant, we proceed to step 3. Step 3 is the response-shift model. Given a significant chi-square difference test observed in step 2, response shift effects are identified using modification indices and significance testing. Modification indices indicate how modifying the model structure (eg, parameter constraints) might change model fit and therefore reveal lack of measurement invariance over time.
      • Oort F.J.
      Using structural equation modeling to detect response shifts and true change.
      • Steiger J.H.
      Structural model evaluation and modification: an interval estimation approach.
      • Schreiber J.B.
      • Nora A.
      • Stage F.K.
      • Barlow E.A.
      • King J.
      Reporting structural equation modeling and confirmatory factor analysis results: a review.
      These modification indices are akin to sensitivity analyses in regression models in that they examine the robustness of the model given specific parameter constraints (eg, intercepts at t1–tn are equal, factor loadings at t1–tn are equal). Step 4 is the final model. For the final model we test for other types of invariance after establishing partial invariance of intercepts, factor loadings, and residual variances. In this final model, we examine equality of common factor means, variances, and correlations. Differences between common factor means reflect unbiased or true changes in QOL. Based on parameter estimates in step 4, we estimate effect sizes for true change and the contribution of response shift effects and true change to observed change.
      The interested reader is referred to the Oort et al
      • Oort F.J.
      • Visser M.R.
      • Sprangers M.A.
      An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
      • Oort F.J.
      Using structural equation modeling to detect response shifts and true change.
      seminal articles describing the SEM approach for response shift detection.

      SEM model covariates

      Two types of covariates were used in the SEM model. The first adjusted for missingness to investigate whether missing data biased the model. Missing values were treated as response factors and each patient-reported outcome score variable was treated as categorical. For example, for the year 1 latent variables, we test the difference in scores between those who complete only year 1 of the survey (group 0) versus those who complete years 1 and 2 of the survey (group 1). A similar covariate was completed for year 1 versus years 1, 2, and 5, and for years 1 and 2 versus years 1, 2, and 5. This covariate therefore effectively adjusts for latent mean differences in people who completed only year 1, only year 1 and 2, or all 3 time points.
      The second type of covariate adjusted for possible risk factors for worse QOL outcomes using propensity scores. Propensity scores were modeled as a covariate predictor of each latent variable for each year: the linear models used to generate the propensity scores specified the observed scores of the domains of interest as the dependent variables, and the independent variables were sex, race/ethnicity, AIS grade,
      • Kreutzer J.
      • DeLuca J.
      • Caplan B.
      (Eds.). Encyclopedia of clinical neuropsychology.
      education (dummy variables indicating highest degree attained), and age at injury. These variables were selected because they were expected to influence QOL change over time but were not directly related to response shift. The propensity score was the predicted score from this model.

      Response shift hypotheses

      Response shift effects are expected in evaluation-based measures, but not performance- or perception-based measures.
      • Schwartz C.E.
      • Rapkin B.D.
      Reconsidering the psychometrics of quality of life assessment in light of response shift and appraisal.
      The FIM is an observer-rated perception-based measure, whereas the SF-36v2 is a self-reported evaluative measure. Therefore, even though both measures may address aspects of physical functioning, they are not of the same nature. Our operationalization of QOL includes the SF-36v2 domain scores and the LiSAT score (based on 5 items). These are all evaluative measures, which are where response shift effects are expected to show up. We hypothesize that recalibration, reprioritization, and reconceptualization response shifts occur in response to SCI. Because the evaluative data available for analysis were collected after people with SCI had likely stabilized, we expect these response shift effects to be small.

      Results

      Objective functioning over time

      Figure 1 shows the means and 95% confidence intervals on the FIM motor, cognitive, and total scores at admission, discharge, and at 1, 2, and 5 years postinjury. It is notable that the FIM motor and total scores were much worse at admission compared with later in follow-up, improved at discharge, and then stabilized by the time of study follow-up. The only statistically significant important difference in patient-reported outcome scores over time was for role physical between years 1 and 2, and between years 1 and 5 (table 2). All other patient-reported outcome change scores over time were below the published minimally important difference thresholds.
      • Ware J.E.
      • Kosinski M.
      • Dewey J.E.
      How to Score Version 2 of the SF-36(R) Health Survey.
      • Heinemann A.W.
      • Michael Linacre J.
      • Wright B.D.
      • Hamilton B.B.
      • Granger C.
      Measurement characteristics of the functional independence measure.
      • Sekhon S.
      • Pope J.
      • Baron M.
      • Group C.S.
      The minimally important difference in clinical practice for patient-centered outcomes including health assessment questionnaire, fatigue, pain, sleep, global visual analog scale, and SF-36 in scleroderma.
      • Sumowski J.F.
      • Wylie G.R.
      • Gonnella A.
      • Chiaravalloti N.
      • Deluca J.
      Premorbid cognitive leisure independently contributes to cognitive reserve in multiple sclerosis.
      Figure thumbnail gr1
      Fig 1Mean FIM and SE of the mean over time in the study sample. FIM motor and total scores were much worse at admission and improved at discharge (prior to study period), and then stabilized at 1, 2, and 5 years postinjury (during the study period). The FIM cognitive scores did not change from admission through year 5. This stability is consistent with the natural history of SCI.
      • Hall K.M.
      • Cohen M.E.
      • Wright J.
      • Call M.
      • Werner P.
      Characteristics of the Functional Independence Measure in traumatic spinal cord injury.
      Table 2Descriptive statistics of measures over time
      Subscale ScoreYear 1 Mean (SD)Year 2

      Mean (SD)
      Year 5

      Mean (SD)
      Year 2 - Year 1

      Mean Δ (SD of Δ)
      Year 5 - Year 1

      Mean Δ (SD of Δ)
      Year 5 - Year 2

      Mean Δ (SD of Δ)
      Physical Function31.7 (30.3)34.1 (31.1)30.8 (29.0)2.2 (20.2)2.1 (20.1)0.8 (19.4)
      Role Physical40.0 (31.5)45.4 (31.1)46.6 (32.5)5.7 (28.1)5.7 (33.4)2.1 (29.8)
      Pain55.8 (25.7)57.8 (25.4)56.1 (27.3)2.5 (21.6)−2.2 (24.8)−1.1 (22.0)
      General Health61.5 (21.8)58.6 (21.7)59.2 (21.1)−2.7 (17.5)−2.7 (21.6)0.l (16.7)
      Vitality51.5 (20.0)53.0 (19.1)51.0 (20.2)2.0 (16.1)−0.5 (17.5)−0.5 (17.0)
      Social Functioning62.4 (26.7)65.7 (25.2)64.7 (26.3)3.5 (25.0)2.8 (29.0)−1.4 (25.2)
      Role Emotional69.7 (32.1)70.0 (30.4)71.5 (32.0)−0.7 (29.7)−1.1 (33.5)−1.4 (28.4)
      Mental Health69.1 (19.1)69.5 (18.6)69.1 (20.0)0.1 (14.9)−1.1 (16.6)−1.3 (16.3)
      LISAT
      LISAT score used was the one unidimensional factor based on 5 items, as described in Methods section.
      60.4 (19.9)63.0 (19.5)64.0 (19.6)1.5 (15.0)4.1 (16.4)2.4 (14.7)
      FIM Total Score106.7 (22.5)107.7 (21.7)104.0 (23.4)0.5 (8.3)0.7 (8.8)0.3 (9.0)
      FIM Cognitive Score34.5 (1.7)34.6 (1.7)34.4 (1.9)0.0 (1.8)−0.2 (2.0)−0.2 (1.9)
      FIM Motor Score72.2 (22.2)73.1 (21.3)69.5 (23.1)0.5 (8.1)0.9 (8.6)0.5 (8.7)
      NOTE. Raw scores for the SF-36v2 subscales were used (0–100 metric).
      Bolded values indicate significant change over time.
      LISAT score used was the one unidimensional factor based on 5 items, as described in Methods section.

      Covariate effect sizes

      Table 3 shows the magnitude of the effect sizes for the missingness and propensity score covariates for each latent variable (see Covariates at the bottom of the table). The missingness covariates were negligible for all time-point comparisons and latent variables (ie, all were smaller than Cohen small effect size of .20). In contrast, the propensity covariate scores were all significantly different from zero, indicating that sex, race/ethnicity, neurologic severity (AIS), education, and age at injury were meaningfully associated with the latent variables. The magnitude of the propensity effect sizes attenuated over time, suggesting that the association of the risk factors with the latent variables diminished with the passing of time.
      Table 3Parameter estimates in the final model
      Factor LoadingsPhysical (SE)Mental (SE)Symptoms (SE)
       Physical function17.1 (0.63)
       Role physical22.3 (0.71)
       Social functioning9.9 (0.63)11 (0.66)
       LiSAT8.1 (0.49)7.3 (0.52)
       Role emotional21.4 (0.67)
       Mental health16.7 (0.38)
       Pain15.8 (0.55)
       General health14.2 (0.45)
       Vitality15.7 (0.40)
      Physical FunctionRole PhysicalPainGeneral HealthVitalitySocial FunctioningRole EmotionalMental HealthLiSAT
      InterceptsY1=31.6

      Y2 and Y5=28.6
      39.5057.00Y1=62.4

      Y2 and Y5=58.6
      52.4062.069.068.560.1
      Physical FunctionRole PhysicalPainGeneral HealthVitalitySocial FunctioningRole EmotionalMental HealthLiSAT
      Residual variances574.80370.50400.50258.40129.60309.3502.175.8182.9
      Role emotional with role physical
      Residual correlation0.33
      Factor correlationsPhysicalMentalSymptoms
       Physical1.000.57Y1=0.78
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
      Y2=0.85
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
      Y5=0.73
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
       Mental1.00Y1 and Y2=0.85
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
      Y5=0.79
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
       Symptoms1.00
      Factor meansPhysical (SE)Mental (SE)Symptoms (SE)
       Y10 (---)0 (---)0 (---)
       Y2.29
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      (0.09)
      0.02 (0.08)0.03 (0.08)
       Y5.29
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      (0.11)
      0.05 (0.10)–0.04 (0.10)
      Covariate typeCovariatesPhysicalMentalSymptoms
       MissingnessY1 vs Y12–0.020.01–0.13
       MissingnessY1 vs Y1250.050.100.00
       MissingnessY12 vs Y1250.000.130.00
       Risk factorsY1 Propensity.46
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .14
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .20
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
       Risk factorsY2 Propensity.23
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .06
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .11
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
       Risk factorsY5 Propensity.09
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .04
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      .07
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.
      NOTE. All parameters constrained to be equal across time unless otherwise noted. Spaces in table are intentionally blank.
      Abbreviations: Y1, year 1; Y2, year 2; Y5, year 5. Y12: years 1 and 2; Y125, years 1, 2, and 5.
      Indicates that there was a trend difference in the factor correlations Y1, Y2, and Y5 (P=.058).
      Indicates that factor means and covariate coefficients were significantly different from zero at P<.05.

      Response shift SEM analysis

      Step 1: measurement model

      Figure 2 shows the measurement model used for the SEM analysis. Three latent factors were revealed, reflecting Physical (comprised of physical function, role physical, social function, and life satisfaction), Mental (comprised of social function, life satisfaction, role emotional, and mental health), and Symptoms (comprised of pain, general health, and vitality). Whereas social functioning and LiSAT scores loaded on both Physical and Mental factors, all other domains loaded only on 1 latent factor. This model, where all parameters (eg, factor loadings, intercepts) are freely estimated and allowed to vary over time, closely fit the data with no response shift constraints across time (table 4).
      Figure thumbnail gr2
      Fig 2Measurement model for SEM analysis. The measurement model used in response shift detection. Significant response shift effects are indicated by parameter estimate differences at each follow-up time point. Factor loadings are unstandardized, and factor variances equal 1.0. All coefficients are constrained to be equal across time unless noted. Factor covariates not shown include dummy variables for later dropout (eg, dummy variates at year 1 factors for participants who completed only year 1 [dropout] or year 1 and year 2). Additional covariates include propensity scores which adjust for age at injury, sex, race/ethnicity, education, and AIS scores. Abbreviations: GH, general health; MH, mental health; PF, physical functioning; RE, role emotional; RP, role physical; SF, social functioning; VT, vitality; Y1, year 1; Y2, year 2; Y5, year 5.
      Table 4Model fit indices
      Model Descriptionχ2 (df)SRMRRMSEA (95% CI)CFI/TLIχ2 Difference Test, P
      Step 1: measurement model906.2 (369).033.047 (.043–.051).933/.922NA
      Step 2: no response shift model984.9 (421).036.045 (.042–.049).930/.928χ252=78.75, .010
      Step 3: response shift model958.7 (419).036.045 (.041–.048).932/.931χ250=52.52, .377
      Step 4: final model (additional constraints)969.4 (424).036.044 (.041–.048).932/.931χ25=10.66, .058
      Abbreviations: CI, confidence interval; NA, not applicable; SRMR, standardized root mean square residual.

      Step 2: no response shift model

      The no response shift model constrains the factor loadings, intercepts, and residual variances to be equal across time. Forcing the model parameters to be equal across time is testing the hypothesis that there is no response shift occurring anywhere in the model. Factor means at year 2 and year 5 are freely estimated. This model had worse fit statistics than the measurement model, and a chi-square difference test of the models (step 2–step 1) indicated significantly worse fit in the no response shift model (see table 4). These results suggest some degree of response shift is present.

      Step 3: response shift model

      The response shift model retains the structure of the step 2 model but removes response shift constraints that contribute to poor model fit. When we forced the model to constrain the intercepts for general health and physical function to be equal over time (ie, unchanging or no response shift), the model fit poorly. To improve the fit of the model, we needed to allow the intercepts for general health and physical function to vary over time, specifically to be different (higher) in year 1 than in years 2 and 5 (lower). Because the intercept reflects the sample mean for that variable, allowing intercepts to vary acknowledges that it may change (decrease) over time and is evidence of response shift. The year 1 intercepts for general health and physical function were therefore freely estimated and resulted in a significantly better-fitting model than in step 1 (see table 4).

      Step 4: final model

      The final model imposed onto the step 3 model all constraints to the factor correlations and residual correlations. The initial step 4 model with all possible equality constraints did not fit significantly better than the step 3 model (see table 4). To achieve a fit better than in step 3, the final model required the factor correlations between Symptoms and Physical to be freed among all 3 time points, and to free correlations between the year 5 Symptoms and Mental factors (see table 4).

      Evaluation of response shift effects and true change

      A comparison of the change in intercepts over time (step 3) suggests uniform recalibration response shift effects for general health and physical function (see Intercepts in table 3). In other words, individuals’ intercepts (ie, means) for 2 subscales that comprise aspects of the Physical and Symptoms latent variables changed. They decreased over time, suggesting perhaps a lowering of expectations for physical function (part of physical) and general health (part of symptoms) as the individual adapted to SCI.
      A comparison of the relations between latent variables (Step 4) suggests small higher-level reconceptualization response shift effects (see Factor correlations in table 3). In other words, Symptoms and Physical are more closely linked in individuals’ QOL conceptualization at year 1, even more at year 2, and then much less at year 3. The changes in factor correlations reflect a shifting conceptualization of QOL over time.
      After adjusting for covariates and response shift effects, the Physical latent variable improved at years 2 and 5 compared with year 1 (d=.29) (see Factor means in table 3).
      Table 5 shows a comparison of response shift and observed change contributions to the true score change. True change contribution is conceptualized as the difference between the observed change value and the value of the response shift. All observed change effect sizes, with the exception for the improvement in role physical between years 1 and 5, are smaller than the Cohen criteria for small effect sizes. Adjusting for the response shift effects did not influence the true change estimate of change over time (ie, the estimates remain smaller than the Cohen threshold for a small effect size).
      Table 5Significance tests of response shift effects and effect sizes of observed change, response shift, and true change in the final model
      All statistics shown are Cohen d.
      Effect Sizes
      Subscale ScoreResponse ShiftObserved ChangeResponse Shift ContributionTrue Change Contribution
      χ21Prob.Year 2–Year 1Year 5–Year 1Year 5–Year 2Year 2–Year 1Year 5–Year 1Year 5–Year 2Year 2–Year 1Year 5–Year 1Year 5–Year 2
      Physical functionUniform recalibration5.84.0160.08–0.03–0.11–0.10–0.100.170.07–0.11
      Role physical0.170.210.040.170.210.04
      Pain0.080.01–0.070.080.01–0.07
      General healthUniform recalibration20.4<.001–0.13–0.100.03–0.17–0.180.040.070.03
      Vitality0.08–0.02–0.100.08–0.02–0.10
      Social functioning0.130.09–0.040.130.09–0.04
      Role emotional0.010.060.050.010.060.05
      Mental health0.020.00–0.020.020.00–0.02
      LiSAT0.130.180.050.130.180.05
      NOTE. Spaces in table are intentionally blank. Abbreviation: Prob., probability.
      Bolded values are small effect-sizes using Cohen's criteria.
      All statistics shown are Cohen d.

      Discussion

      Within this sample of people with SCI, our analyses identified and accounted for uniform recalibration higher-level reconceptualization response shift effects when estimating true change over follow-up. People appeared to change their internal standards regarding physical functioning (an aspect of the Physical latent variable) and general health (an aspect of the Symptoms latent variable) over time, with year 1 being significantly higher than years 2 and 5. This finding indicates that as the physical function component of the Physical latent variable decreased from year 1 to years 2 and 5, the true change in the Physical latent variable actually increased. This effect was observed even after accounting for attrition and risk factors.
      The improvement in the Physical latent variable over time was likely captured because of the inclusion of a role physical measure and might have been missed if only the FIM had been administered. This is because this objective measure of function suggested stability in motor and cognitive functioning after 1 year post-injury, which is consistent with the natural history of SCI.
      • Hall K.M.
      • Cohen M.E.
      • Wright J.
      • Call M.
      • Werner P.
      Characteristics of the Functional Independence Measure in traumatic spinal cord injury.
      In contrast, the patient-reported measures suggest that people with SCI experience improvements in role physical over time. The patient-reported role Physical domain therefore appears to be more sensitive to change than the FIM, which is consistent with past research. This improvement in role physical over time likely underlies the sample’s improvements in the Physical latent variable in years 2 and 5 compared with year 1. Although these improvements are small effect sizes, they are notable because they suggest that people can experience measurable improvements in the Physical latent variable (reflecting physical functioning, role physical, social functioning, and life satisfaction) even in the context of an irreversible disability. This is consistent with past research that suggests that QOL improves with time from injury, suggesting adaptation to SCI.
      • Westgren N.
      • Levi R.
      Quality of life and traumatic spinal cord injury.
      Other research suggests that improved adjustment and resilience in people with SCI may relate to meaning-making,
      • Davis C.G.
      • Novoa D.C.
      Meaning-making following spinal cord injury: individual differences and within-person change.
      intentional embracing of gratitude,
      • Chun S.
      • Lee Y.
      “I am just thankful”: the experience of gratitude following traumatic spinal cord injury.
      and self-efficacy.
      • Kilic S.
      • Dorstyn D.
      • Guiver N.
      Examining factors that contribute to the process of resilience following spinal cord injury.
      Our results also identified reconceptualization response shift effects. Individuals with SCI appeared to change their conceptualization of QOL over time, such that the Symptoms latent factor was less correlated with the Physical and Mental latent factors at 5 years postinjury compared with 1 and 2 years postinjury. Therefore, Symptoms (comprised of SF-36v2 domain scores for pain, general health, and vitality) become less linked to Physical and Mental functioning (latent variables) in this SCI sample. This may reflect the idea that people with SCI stop considering their SCI as part of their general health over time, and instead only consider the sequelae because of their injury as reflecting their general health.
      • Tate D.G.
      • Kalpakjian C.Z.
      • Forchheimer M.B.
      Quality of life issues in individuals with spinal cord injury.
      The study is notable not only because of its substantial sample size and follow-up, but also because it measures domains relevant to QOL. By including both observer- and patient-reported indices, the data enable an examination of catalyst effects in this response shift study. The small objective change in FIM scores over time suggests that the sample has stable motor and cognitive function during the study period. The detected response shift effects, in particular for physical function, are necessarily small because of this stability: larger response shift effects in physical function would be expected when health state changes (catalysts) are also larger.
      • Schwartz C.E.
      • Sprangers M.A.
      • Fayers P.
      You know it's there but how do you capture it? Challenges for the next phase of response shift research.
      • Schwartz C.E.
      • Sprangers M.A.
      • Carey A.
      • Reed G.
      Exploring response shift in longitudinal data.
      • Ahmed S.
      • Sawatzky R.
      • Levesque J.F.
      • Ehrmann-Feldman D.
      • Schwartz C.E.
      Minimal evidence of response shift in the absence of a catalyst.
      Although the detected response shifts were small, the true change improvement in the Physical latent variable identified in this study provides some insight into the SCI experience: although individuals with SCI may perceive declines in their physical abilities, they may not perceive these declines as limiting their daily functioning. At this time, more research is needed to understand the clinical effect of this finding.

      Study limitations and directions for future research

      Other limitations of the study should be noted. First, the data were collected after the period of greatest adjustment; therefore, we cannot address response shift immediately after injury. Nonetheless, even in the context of stable FIM motor and cognitive function, response shift effects were detectable. This is an important finding for understanding adaptation to SCI but also in expanding response shift theory to address adaptation in the context of stabilized health after trauma. Future research might replicate this study using data collected closer to SCI onset and examine response shift effects in different neurologic impairment groups (eg, complete vs incomplete tetraplegia, complete vs incomplete paraplegia). One might also include a measure of appraisal
      • Rapkin B.D.
      • Schwartz C.E.
      Toward a theoretical model of quality-of-life appraisal: implications of findings from studies of response shift.
      • Rapkin B.D.
      • Garcia I.
      • Michael W.
      • Zhang J.
      • Schwartz C.E.
      Distinguishing appraisal and personality influences on quality of life in chronic illness: introducing the quality-of-life Appraisal Profile version 2.
      to expand on earlier findings by Dibb et al
      • Dibb B.
      • Ellis-Hill C.
      • Donovan-Hall M.
      • Burridge J.
      • Rushton D.
      Exploring positive adjustment in people with spinal cord injury.
      that suggest that adjustment in SCI is related to managing goals and expectations, comparison with others, feeling useful, and acceptance (reflecting appraisal and/or mechanisms consistent with response shift theory). This information can be useful for developing rehabilitation interventions focused on goal planning
      • Byrnes M.
      • Beilby J.
      • Ray P.
      • McLennan R.
      • Ker J.
      • Schug S.
      Patient-focused goal planning process and outcome after spinal cord injury rehabilitation: quantitative and qualitative audit.
      and other facilitators of response shifts
      • Schwartz C.E.
      • Finkelstein J.A.
      • Rapkin B.D.
      Appraisal assessment in patient-reported outcome research: methods for uncovering the personal context and meaning of quality of life.
      in people with SCI.

      Conclusions

      We detected uniform recalibration and reconceptualization response shift effects over 5 years of follow-up in people with SCI. Therefore, despite stable motor and cognitive status, people learn to adapt to their condition. This adaptation appears to reflect a progressive disconnection between symptoms and physical or mental health, and a real improvement in physical functioning over time. Because the focus of rehabilitation in SCI is adaptation, these small but significant response shift effects reflect an underlying important adjustment process consistent with rehabilitation ideals. Characterizing response shift effects may therefore be a useful approach to outcome assessment in SCI.

      Supplier

      • a.
        Mplus; Muthen & Muthen.

      Acknowledgments

      We thank the Rick Hansen Spinal Cord Injury Registry Network and all of the participating sites: G.F. Strong Rehabilitation Centre, Vancouver General Hospital, Foothills Medical Centre, Glenrose Rehabilitation Hospital, Royal Alexandra Hospital, University of Alberta Hospital, Royal University Hospital, Saskatoon City Hospital, Winnipeg Health Sciences Centre, Toronto Western Hospital, Toronto Rehabilitation Institute, St. Michael’s Hospital, Sunnybrook Health Sciences Centre, Hamilton General Hospital, Hamilton Health Sciences – Regional Rehabilitation Centre, Victoria Hospital, University Hospital, Parkwood Institute, The Ottawa Hospital Rehabilitation Centre, The Ottawa Hospital – Civic Campus, Hôpital de l’Enfant Jésus, Institut de réadaptation en déficience physique de Québec, Centré de réadaptation Lucie-Bruneau, Institut de réadaptation Gingras-Lindsay-de-Montréal, Hôpital du Sacré-Coeur de Montréal, Nova Scotia Rehabilitation Centre, QEII Health Sciences Centre, Saint John Regional Hospital, Stan Cassidy Centre for Rehabilitation, St. John’s Health Sciences Centre, and Dr. Leonard A. Miller Rehabilitation Centre. We also thank Maria Orlando Edelen, PhD, Allen Heinemann, PhD, and Frans Oort, PhD, for helpful discussions; and Tian Shen for helpful comments on earlier drafts of the manuscript.

      Supplemental Appendix S1. AIS grades

      Based on the International Standards for the Neurological Classification of Spinal Cord Injury,
      • Kirshblum S.C.
      • Burns S.P.
      • Biering-Sorensen F.
      • et al.
      International standards for neurological classification of spinal cord injury (revised 2011).
      the AIS
      • Kreutzer J.
      • DeLuca J.
      • Caplan B.
      (Eds.). Encyclopedia of clinical neuropsychology.
      classifies the level of impairment as follows:
      • Grade A: complete. No sensory or motor function is preserved in the sacral segments S4-5.
      • Grade B: sensory incomplete. Sensory but not motor function is preserved below the neurologic level and includes the sacral segments S4-5 (light touch or pin prick at S4-5 or deep anal pressure), and no motor function is preserved >3 levels below the motor level on either side of the body.
      • Grade C: motor incomplete. Motor function is preserved at the most caudal sacral segments for voluntary anal contraction or the patient meets the criteria for sensory incomplete status (sensory function preserved at the most caudal sacral segments [S4-5] by light touch, pin prick, or dermatome) and has some sparing of motor function >3 levels below the ipsilateral motor level on either side of the body. (This includes key or nonkey muscle functions to determine motor incomplete status.) For AIS grade C, less than half of key muscle functions below the single neurological level of injury have a muscle grade ≥3.
      • Grade D: motor incomplete. Motor incomplete status as previously defined, with at least half (half or more) of key muscle functions below the single neurological level of injury having a muscle grade ≥3.
      More than 95% of the sample showed during the study period a change of less than the published minimally clinically important difference of 22 points on the FIM total score, 17 points on the FIM motor score, and 3 points on the FIM cognitive score
      • Beninato M.
      • Gill-Body K.M.
      • Salles S.
      • Stark P.C.
      • Black-Schaffer R.M.
      • Stein J.
      Determination of the minimal clinically important difference in the FIM instrument in patientswith stroke.
      (fig 2 and table 2).

      References

        • Dumont R.J.
        • Okonkwo D.O.
        • Verma S.
        • et al.
        Acute spinal cord injury, part I: pathophysiologic mechanisms.
        Clin Neuropharmacol. 2001; 24: 254-264
        • Yoshida K.K.
        Reshaping of self: a pendular reconstruction of self and identity among adults with traumatic spinal cord injury.
        Sociol Health Illn. 1993; 15: 217-245
        • Hammell K.W.
        Psychological and sociological theories concerning adjustment to traumatic spinal cord injury: the implications for rehabilitation.
        Spinal Cord. 1992; 30: 317-326
        • World Health Organization
        Constitution of the World Health Organization.
        Basic documents. 2006; 45: 1
        • Ryff C.D.
        Happiness is everything, or is it? Explorations on the meaning of psychological well-being.
        J Pers Soc Psychol. 1989; 57: 1069-1081
        • Ryan R.M.
        • Deci E.L.
        On happiness and human potentials: a review of research on hedonic and eudaimonic well-being.
        Annu Rev Psychol. 2001; 52: 141-166
        • Bach J.R.
        • Tilton M.C.
        Life satisfaction and well-being measures in ventilator assisted individuals with traumatic tetraplegia.
        Arch Phys Med Rehabil. 1994; 75: 626-632
        • Sprangers M.A.
        • Schwartz C.E.
        Integrating response shift into health-related quality of life research: a theoretical model.
        Soc Sci Med. 1999; 48: 1507-1515
        • Schwartz C.E.
        • Sprangers M.A.
        Methodological approaches for assessing response shift in longitudinal health-related quality-of-life research.
        Soc Sci Med. 1999; 48: 1531-1548
        • Schwartz C.E.
        • Sprangers M.A.
        • Fayers P.
        You know it's there but how do you capture it? Challenges for the next phase of response shift research.
        in: Fayers P. Hays R. Assessing quality of life in clinical trials. 2nd ed. Oxford University Press, Oxford2005: 275-290
        • Schwartz C.E.
        • Ahmed S.
        • Sawatsky R.
        • et al.
        Guidelines for secondary analysis in search of response shift.
        Qual Life Res. 2013; 22: 2663-2673
        • Rapkin B.D.
        • Schwartz C.E.
        Toward a theoretical model of quality-of-life appraisal: implications of findings from studies of response shift.
        Health Qual Life Outcomes. 2004; 2: 14
        • Schwartz C.E.
        • Sprangers M.A.
        • Carey A.
        • Reed G.
        Exploring response shift in longitudinal data.
        Psychol Health. 2004; 19: 51-69
        • Oort F.J.
        • Visser M.R.
        • Sprangers M.A.
        An application of structural equation modeling to detect response shifts and true change in quality of life data from cancer patients undergoing invasive surgery.
        Qual Life Res. 2005; 14: 599-609
        • Finkelstein J.A.
        • Quaranto B.R.
        • Schwartz C.E.
        Threats to the internal validity of spinal surgery outcome assessment: recalibration response shift or implicit theories of change?.
        Appl Res Qual Life. 2014; 9: 215-232
        • King-Kallimanis B.L.
        • Oort F.J.
        • Nolte S.
        • Schwartz C.E.
        • Sprangers M.A.
        Using structural equation modeling to detect response shift in disability and QOL scores of multiple sclerosis patients.
        Qual Life Res. 2011; 20: 1527-1540
        • Li Y.
        • Rapkin B.D.
        Classification and regression tree analysis to identify complex cognitive paths underlying quality of life response shifts: a study of individuals living with HIV/AIDS.
        J Clin Epidemiol. 2009; 62: 1138-1147
        • Schwartz C.E.
        • Sprangers M.A.
        • Ahmed S.
        • et al.
        Response shift in patients with multiple sclerosis: an application of three statistical techniques.
        Qual Life Res. 2011; 20: 1561-1572
        • Verdam M.G.
        • Oort F.J.
        • van der Linden Y.M.
        • Sprangers M.A.
        Taking into account the impact of attrition on the assessment of response shift and true change: a multigroup structural equation modeling approach.
        Qual Life Res. 2015; 24: 541-551
        • Schwartz C.E.
        • Andresen E.M.
        • Nosek M.A.
        • Krahn G.L.
        Response shift theory: important implications for measuring quality of life in people with disability.
        Arch Phys Med Rehabil. 2007; 88: 529-536
        • Schwartz C.E.
        • Bode R.
        • Repucci N.
        • Becker J.
        • Sprangers M.A.
        • Fayers P.M.
        The clinical significance of adaptation to changing health: a meta-analysis of response shift.
        Qual Life Res. 2006; 15: 1533-1550
        • Finkelstein J.A.
        • Razmjou H.
        • Schwartz C.E.
        Response shift and outcome assessment in orthopedic surgery: is there is a difference between complete vs. partial treatment?.
        J Clin Epidemiol. 2009; 82: 1189-1190
        • Schwartz C.E.
        • Quaranto B.R.
        • Rapkin B.D.
        • Healy B.C.
        • Vollmer T.
        • Sprangers M.A.
        Fluctuations in appraisal over time in the context of stable and non-stable health.
        Qual Life Res. 2014; 23: 9-19
        • Oort F.J.
        • Visser M.R.
        • Sprangers M.A.
        Measurement and conceptual perspectives on response shift: formal definitions of measurement bias, explanation bias, and response shift.
        J Clin Epidemiol. 2009; 62: 1126-1137
        • Hall K.M.
        • Cohen M.E.
        • Wright J.
        • Call M.
        • Werner P.
        Characteristics of the Functional Independence Measure in traumatic spinal cord injury.
        Arch Phys Med Rehabil. 1999; 80: 1471-1476
        • Bruno M.A.
        • Bernheim J.L.
        • Ledoux D.
        • Pellas F.
        • Demertzi A.
        • Laureys S.
        A survey on self-assessed well-being in a cohort of chronic locked-in syndrome patients: happy majority, miserable minority.
        BMJ Open. 2011; 1: e000039
        • Dibb B.
        • Ellis-Hill C.
        • Donovan-Hall M.
        • Burridge J.
        • Rushton D.
        Exploring positive adjustment in people with spinal cord injury.
        J Health Psychol. 2014; 19: 1043-1054
        • Dijkers M.P.
        Quality of life of individuals with spinal cord injury: a review of conceptualization, measurement, and research findings.
        J Rehabil Res Dev. 2005; 42: 87
        • Tate D.G.
        • Kalpakjian C.Z.
        • Forchheimer M.B.
        Quality of life issues in individuals with spinal cord injury.
        Arch Phys Med Rehabil. 2002; 83: S18-S25
        • Post M.W.
        • van Leeuwen C.M.
        • van Koppenhagen C.F.
        • de Groot S.
        Validity of the Life Satisfaction questions, the Life Satisfaction Questionnaire, and the Satisfaction With Life Scale in persons with spinal cord injury.
        Arch Phys Med Rehabil. 2012; 93: 1832-1837
        • van Leeuwen C.M.
        • Post M.W.
        • van der Woude L.H.
        • et al.
        Changes in life satisfaction in persons with spinal cord injury during and after inpatient rehabilitation: adaptation or measurement bias?.
        Qual Life Res. 2012; 21: 1499-1508
        • Schwartz C.E.
        • Sprangers M.A.
        Guidelines for improving the stringency of response shift research using the then-test.
        Qual Life Res. 2010; 19: 455-464
        • Ahmed S.
        • Sawatzky R.
        • Levesque J.F.
        • Ehrmann-Feldman D.
        • Schwartz C.E.
        Minimal evidence of response shift in the absence of a catalyst.
        Qual Life Res. 2014; 23: 2421-2430
        • Ahmed S.
        • Bourbeau J.
        • Maltais F.
        • Mansour A.
        The Oort structural equation modeling approach detected a response shift after a COPD self-management program not detected by the Schmitt technique.
        J Clin Epidemiol. 2009; 62: 1165-1172
        • Barclay-Goddard R.
        • Lix L.M.
        • Tatec R.
        • Weinberg L.
        • Mayo N.E.
        Response shift was identified over multiple occasions with a structural equation modeling framework.
        J Clin Epidemiol. 2009; 62: 1181-1188
        • Visser M.
        • Oort F.J.
        • Sprangers M.
        Assessing response shift in QL: the thentest and a structural equation modeling approach compared.
        Qual Life Res. 2003; 12: 45
        • Oort F.J.
        Using structural equation modeling to detect response shifts and true change.
        Qual Life Res. 2005; 14: 587-598
        • Noonan V.
        • Kwon B.
        • Soril L.
        • et al.
        The Rick Hansen spinal cord injury registry (RHSCIR): a national patient-registry.
        Spinal Cord. 2012; 50: 22-27
        • Kirshblum S.C.
        • Burns S.P.
        • Biering-Sorensen F.
        • et al.
        International standards for neurological classification of spinal cord injury (revised 2011).
        J Spinal Cord Med. 2011; 34: 535-546
        • Kreutzer J.
        • DeLuca J.
        • Caplan B.
        (Eds.). Encyclopedia of clinical neuropsychology.
        Springer, New York2011: 255-257
        • Ware J.E.
        • Kosinski M.
        • Dewey J.E.
        How to Score Version 2 of the SF-36(R) Health Survey.
        Quality Metric, Lincoln2000
        • Fugl-Meyer A.R.
        • Melin R.
        • Fugl-Meyer K.S.
        Life satisfaction in 18-to 64-year-old Swedes: in relation to gender, age, partner and immigrant status.
        J Rehabil Med. 2002; 34: 239-246
        • Heinemann A.W.
        • Michael Linacre J.
        • Wright B.D.
        • Hamilton B.B.
        • Granger C.
        Measurement characteristics of the functional independence measure.
        Top Stroke Rehabil. 1994; 1: 1-15
        • Muthen L.K.
        • Muthen B.O.
        Mplus user's guide.
        6th ed. Muthen & Muthen, Los Angeles2011
        • Steiger J.H.
        Structural model evaluation and modification: an interval estimation approach.
        Multivariate Behav Res. 1990; 25: 173-180
        • Schreiber J.B.
        • Nora A.
        • Stage F.K.
        • Barlow E.A.
        • King J.
        Reporting structural equation modeling and confirmatory factor analysis results: a review.
        J Educ Res. 2006; 99: 323-338
        • Schwartz C.E.
        • Rapkin B.D.
        Reconsidering the psychometrics of quality of life assessment in light of response shift and appraisal.
        Health Qual Life Outcomes. 2004; 2: 16
        • Sekhon S.
        • Pope J.
        • Baron M.
        • Group C.S.
        The minimally important difference in clinical practice for patient-centered outcomes including health assessment questionnaire, fatigue, pain, sleep, global visual analog scale, and SF-36 in scleroderma.
        J Rheumatol. 2010; 37: 591-598
        • Sumowski J.F.
        • Wylie G.R.
        • Gonnella A.
        • Chiaravalloti N.
        • Deluca J.
        Premorbid cognitive leisure independently contributes to cognitive reserve in multiple sclerosis.
        Neurology. 2010; 75: 1428-1431
        • Westgren N.
        • Levi R.
        Quality of life and traumatic spinal cord injury.
        Arch Phys Med Rehabil. 1998; 79: 1433-1439
        • Davis C.G.
        • Novoa D.C.
        Meaning-making following spinal cord injury: individual differences and within-person change.
        Rehabil Psychol. 2013; 58: 166-177
        • Chun S.
        • Lee Y.
        “I am just thankful”: the experience of gratitude following traumatic spinal cord injury.
        Disabil Rehabil. 2013; 35: 11-19
        • Kilic S.
        • Dorstyn D.
        • Guiver N.
        Examining factors that contribute to the process of resilience following spinal cord injury.
        Spinal Cord. 2013; 51: 553-557
        • Rapkin B.D.
        • Garcia I.
        • Michael W.
        • Zhang J.
        • Schwartz C.E.
        Distinguishing appraisal and personality influences on quality of life in chronic illness: introducing the quality-of-life Appraisal Profile version 2.
        Qual Life Res. 2017; 26: 2815-2829
        • Byrnes M.
        • Beilby J.
        • Ray P.
        • McLennan R.
        • Ker J.
        • Schug S.
        Patient-focused goal planning process and outcome after spinal cord injury rehabilitation: quantitative and qualitative audit.
        Clin Rehabil. 2012; 26: 1141-1149
        • Schwartz C.E.
        • Finkelstein J.A.
        • Rapkin B.D.
        Appraisal assessment in patient-reported outcome research: methods for uncovering the personal context and meaning of quality of life.
        Qual Life Res. 2017; 26: 545-554
        • Beninato M.
        • Gill-Body K.M.
        • Salles S.
        • Stark P.C.
        • Black-Schaffer R.M.
        • Stein J.
        Determination of the minimal clinically important difference in the FIM instrument in patientswith stroke.
        Arch Phys Med Rehabil. 2006; 87: 32-33