A Comparison Between Self-Reported and Observed Activity Limitations in Adults With Neuromuscular Disorders

Participants 

This study was approved by the medical ethics committee of the Université catholique de Louvain. The patients were recruited through the Neuromuscular Reference Center of the Cliniques Universitaires Saint-Luc in Brussels. The patients gave written informed consent before the evaluation. Fifty-seven adult patients with a diagnosed NMD were assessed by 4 physical therapists. Patient descriptions are given in table 1.

Table 1. Patient Sample Description (N=57)

	Characteristics	Values
	Age (y)⁎	50(19–80)
	Sex (n)
	Male	25
	Female	32
	Diagnosis (n)
	DMD/BMD or LGMD	7
	Hereditary neuropathy	12
	Myotonic dystrophy	9
	SMA	4
	FSHD	8
	Others (eg, ALS, PPS)	17
	Mobility level (n)
	Walking	48
	Wheelchair-bound	9

Abbreviations: ALS, amyotrophic lateral sclerosis; BMD, Becker's muscular dystrophy; DMD, Duchenne's muscular dystrophy; FSHD, facioscapulohumeral dystrophy; LGMD, limb-girdle muscular dystrophy; PPS, postpoliomyelitis syndrome; SMA, spinal muscular atrophy.

⁎ Values are mean (range).

Self-Reported and Observed Measures 

Both self-reported and observed measures were obtained with the ACTIVLIM questionnaire so that they were standardized in the item and scoring procedure. This questionnaire consists of 22 daily activities presented in the questionnaire development study4 and was originally developed by using the Rasch model, which allows the conversion of ordinal scores into linear measures located on a unidimensional scale.18 These linear measures are expressed in logits (ie, log-odd units), the constant measurement unit of activity scale; the higher the value is in logits, the higher the activity level is of the patient.

The self-reported measures were obtained by asking patients to provide their perceived difficulty in performing each activity on the ACTIVLIM questionnaire, without using technical or human help, on a 3-level ordinal scale (0=impossible, 1=difficult, 2=easy). The patients were observed at their homes so that they could be in their usual environment when performing the activities to be assessed.19 The physical therapists observed the patients performing the activities and assessed the difficulty experienced by the patients in completing these activities. The scoring procedure of the observed measures was identical to the self-reported measures (0=impossible [the patient is unable to achieve the activity without using technical or human help]; 1=difficult [the patient is able to achieve the activity without any help but experiences some difficulty]; 2=easy [the patient is able to achieve the activity without any help and experiences no difficulty]).

Procedure 

The evaluation comprised 3 stages. The first self-reported measure was collected by mail about 2 weeks before a home visit by the physical therapists. The patients' responses were not disclosed to the therapists to prevent influencing their observations. Second, the patients performed each activity of the ACTIVLIM questionnaire within the limits of their abilities during the evaluation at home. The physical therapists observed the patients performing the activities and rated each of them on the ACTIVLIM ordinal scale. For ethical reasons, some activities such as taking a bath, taking a shower, or washing one's upper body were mimicked, and the patients explained how they usually performed them. Finally, a second self-reported measure was collected about a month after the home visit. In each questionnaire, the activities were randomly presented to avoid any effects caused by the item order. Moreover, the objective of the study was not revealed to the patients to prevent any influence on their responses.

The 57 patients were evaluated by 4 examiners (E1, E2, E3, E4). First, 26 patients were simultaneously assessed by E1 and E2, with an excellent interrater reliability (ICC=.98). Second, 5 patients were assessed by E1, E3, and E4, also with an excellent interrater reliability (ICC=.99). Hence, the last 26 patients were evaluated by 1 examiner (E3 or E4). One observed measure was retained for the 31 patients evaluated by 2 or 3 examiners, and these measures were put together with the 26 measures observed by E3 or E4. As a result, each patient had only 1 observed measure for further data analyses.

Statistical Analysis 

The self-reported and observed responses were first fitted to the Rasch model by using the Rasch Unidimensional Measurement Models computer program.a This software reported overall fit statistics that were close to standardized normative distribution for item and person fit residuals (mean ± SD, −.036±.400 and −.266±.937, respectively, for the first self-reported measure; mean ± SD, −.322±.863 and −.177±.66, respectively, for observed measure; mean ± SD, −.031±.566 and −.144±.946, respectively, for the second self-reported measure) and were not significant for the chi-square item-trait interaction (P=.56, P=.28, and P=.42, respectively, for first self-reported measure, observed measure, and the second self-reported measure). As a result, the ordinal total scores obtained from the self-reported and observed ACTIVLIM questionnaires could be transformed into interval-level measures of activity limitations. Three linear measures of the patients' activity level (first self-reported measure, observed measure, second self-reported measure) were reported in such a way that they could be quantitatively compared and treated as a continuous variable.

The invariance of item hierarchy across the 3 evaluations was checked by using DIF test. A DIF can be detected by a 2-way analysis of variance for each item by comparing scores across each level of patient-related factor (first self-reported measure, observed measure, and second self-reported measure in this case) and across levels of the activity construct (eg, the class intervals).20 A significant main effect for the patient-related factor shows the presence of a uniform DIF, whereas a nonuniform DIF corresponds to a significant interaction effect (patient-related factor by class interval).

The agreement between the self-reported and observed measures was calculated by using the ICC model 2,1 (ICC2,1),21 The ICC2,1 was computed as a 2-way analysis of variance in which the targets (ie, the patients) were 1 source of variability and the raters or measures (the self-perceived and observed activity levels) were the second source of variability.22 This coefficient is preferred to the Pearson correlation coefficient because it takes into account the extent of the relative discrepancies between the evaluations, whereas the Pearson coefficient measures the degree of linearity between 2 measures.22, 23

The κ coefficient was used to quantify item-by-item agreement between both administration methods of the ACTIVLIM questionnaire.24

Study Limitations 

The present results only apply to adult patients with NMDs. Self-reported measures could be overestimated or underestimated in other pathologies such as those with cognitive impact. Further researches are, however, required to verify this assumption by adding personality and depression tests to know the affective and personality characteristics of the patients at the time of the different evaluations. Moreover, because the ACTIVLIM questionnaire is designed for adults as well as for children, this study could also be performed in children with NMDs by comparing self-reported measures reported by their parents with observed measures by external examiners.