The Associated Reaction Rating Scale: a clinical tool ...

Clinical Rehabilitation http://cre.sagepub.com/

The Associated Reaction Rating Scale: a clinical tool to measure associated reactions in the hemiplegic upper limb Anne Macfarlane, Lynne Turner-Stokes and Lorraine De Souza Clin Rehabil 2002 16: 726 DOI: 10.1191/0269215502cr546oa The online version of this article can be found at: http://cre.sagepub.com/content/16/7/726

Published by: http://www.sagepublications.com

Additional services and information for Clinical Rehabilitation can be found at: Email Alerts: http://cre.sagepub.com/cgi/alerts Subscriptions: http://cre.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://cre.sagepub.com/content/16/7/726.refs.html

>> Version of Record - Jul 1, 2002 What is This?

Downloaded from cre.sagepub.com at DUKE UNIV on January 16, 2013

Clinical Rehabilitation 2002; 1 6: 726–735

The Associated Reaction Rating Scale: a clinical tool to measure associated reactions in the hemiplegic upper limb Anne Macfarlane, Lynne Turner-Stokes Regional Rehabilitation Unit, Northwick Park Hospital, London and Lorraine De Souza Department of Health Studies, Brunel University, Middlesex, UK Received 28th April 2000; returned for revisions 26th January 2001; revised manuscript accepted 17th July 2001.

Objective: To determine what items should be included in a clinical assessment tool developed to measure associated reactions in the hemiplegic upper limb and to assess the reliability of the resultant measure. Design: Development through a structured consultative process using focus group methodology. Evaluation of inter- and intra-rater reliability between two independent observers. Subjects: Nineteen consecutive hemiplegic patients admitted to Northwick Park Hospital for rehabilitation following stroke. Methods: Focus groups were conducted in two centres, comprising physiotherapists experienced in the clinical management of brain injury. The groups identied four key characteristics related to severity of associated reactions, which became the items of the rating scale. Evaluation of inter- and intra-rater reliability was undertaken by comparison of agreement between ratings of associated reactions occurring during a single standardized task (sit-to-stand), by two senior physiotherapists in 19 subjects. Results: There were good correlations between the two raters in total (rho 0.89 p < 0.005) and modal scores (rho 0.88 p < 0.005). Reliability testing of each item revealed moderate to very good inter-rater agreement (weighted kappa values 0.43–0.85) and good to very good intra-rater agreement (weighted kappa values 0.61–0.87). A slight tendency for one rater to score more severely than the other only reached signicance for one item (excursion). Overall (modal) severity scores showed a good level of agreement (kappa 0.76–0.81) both between and within raters. Conclusion: Items to be included in a clinical assessment tool to measure associated reactions in the hemiplegic upper limb were determined. Reliability of the resultant measure was found to be encouraging. These results however apply only to observations made during a specic standardized task (sitto-stand) and further study of sensitivity to change and reproducibility in different tasks is required before the ndings can be extrapolated into routine practice.

Address for correspondence: Anne Macfarlane, Physiotheraphy Department, Royal Free Hospital, Pond Street, Hampstead NW3 2QG, UK. © Arnold 2002

10.1191/0269215502cr546oa Downloaded from cre.sagepub.com at DUKE UNIV on January 16, 2013

The Associated Reaction Rating Scale Introduction and background Associated reactions are abnormal postural reactions frequently seen in the affected side of hemiplegic patients and most easily observed in the upper limb.1 Although the term was rst coined by neurologists, it has been adopted primarily by physiotherapists and is now widely used and taught.2 In a survey of physiotherapists’ methods of grading spasticity, over half the respondents incorporated associated reactions within a variety of grading scales.3 Recent studies have demonstrated that associated reactions in the upper limb are purposeless and do not contribute posturally to the realization of voluntary movement.4 Functionally, they are considered to be the product of effort 5,6 and a reection of underlying postural instability.7 Their clinical impact is wide-reaching. They have been implicated in precluding the return of selective movement in the hemiplegic limb8 increasing the likelihood of contracture formation 9 and interfering with balance and function. 10 Clinicians have consequently targeted associated reactions in treatment and have viewed their reduction as a measure of patient progression.11 Although subjective clinical grading scales are widely used by physiotherapists to measure the severity of associated reactions, closer examination reveals that there are many different versions.3 These various scales have tended to be passed down through generations of physiotherapists by post-graduate courses, but have not been published or subjected to formal validity and reliability testing. This has led to a call for a uniform measure. In a multidisciplinary setting, the whole team is involved in carrying over movement learnt in a therapy setting into everyday functional activity. An assessment tool that can be applied by physiotherapy staff, but is nevertheless understandable by the rest of the team, would be helpful for describing associated reactions, heightening awareness, monitoring progress towards the return of normal movement, and ultimately assisting accurate communication between clinicians. The aims of this study were (a) to identify characteristics of severity to be included as items in a clinical tool to measure associated reactions

727

in the hemiplegic upper limb and (b) to undertake a preliminary evaluation of its reliability. Part 1: Development of the instrument Associated reactions may occur in a range of different patterns but retain some common characteristic features. In order to develop a tool to describe and measure them, it was rst necessary to determine which of these characteristic features are most representative of the phenomenon and how these characteristics relate to level of severity. In the absence of consensus within the literature on this topic, focus group methodology12 was used to gather the opinions of physiotherapists with experience in management of hemiplegia, and hence to generate descriptors to form the basis of a grading scale. Focus group methodology Focus groups were held in two hospitals with differing neurological services: one representing a regional specialist service and the other a more general district service. Qualied physiotherapists working in neurology, who were familiar with the phenomenon of associated reaction, were invited to take part. A topic guide was prepared to structure the discussion and was piloted with three physiotherapists not involved in the subsequent focus groups. The two groups comprised 10 and 7 participants respectively. Videotapes of patients carrying out functional tasks such as sit-to-stand and walking were shown as stimulus material for debate over ratings of severity of observed associated reactions. All discussion was audio-taped and transcribed. From the transcriptions, the terms most commonly used to describe associated reactions were identied and the frequency of their occurrence within the group recorded. Selection of items Frequencies were totalled for each group and corroborated against terms used in the literature. Table 1 shows the six most frequently occurring terms. Of these, three (excursion, duration and ‘stereo-typical posturing’) were considered to be too closely interlinked to be separable in their relationship to severity. Thus four distinct char-


728 A Macfarlane et al. Table 1 Most frequently mentioned terms indicating the severity of associated reactions: frequency of use in focus groups and citation within the literature Terms

Frequency

Citation in literature

Hospital A

Hospital B

Total

10

8

18

Dickstein et al. (1995),19 (1996)20 Dvir et al. (1993),11 (1996)21

Impact on function

5

8

13

BBTA (1998) (L Fletcher, personal communication) Davies (1985)10

Number of joints involved

8

3

11

BBTA (1998) (L Fletcher, personal communication)

After-contraction (duration)

6

5

11

Walshe (1923)1 Bobath (1990)2 Edwards (1997)7

Amount of stereotypical posturing

8

3

11

Edwards (1997)7

Physical input by patient/therapist needed to release reaction

6

4

10

Bobath (1990)2

Excursion of limb

acteristics related to severity of associated reactions were identied:

Excursion of the associated reaction and duration after effort ceases The extent of the reaction in the affected limb – dened as the number of joints involved The ability of the subject/therapist to release the associated reaction The affect of the associated reaction on function.

These four key characteristics became the items of the Associated Reaction Rating Scale (Appendix). Item 3 was modied to exclude therapist handling, to reduce confounding variables (see Discussion). To dene levels of severity within each item, the transcript was further analysed for additional words or phrases which were used to denote severity. This information was again corroborated with terms used in the literature and in existing clinical scales from post-graduate courses. From the above investigations, a preliminary tool and guidelines for scoring were developed for assessment of associated reaction during a single standardized task of sit-to-stand (see Discussion). The tool was piloted by two senior neurological physiotherapists on a small group of patients not involved in the reliability study and,

on the basis of these results, was adjusted to form the Associated Reaction Rating Scale (ARRS). In this scale, each of the four items is rated on four hierarchical levels from 0 to 3, and the rater is directed to score the worst level observed during the test. The items are rated individually and can be summated to a total score (range 0–12). However, for reasons that will be discussed, it is currently considered appropriate to take the most commonly occurring score (mode) across the four items as the overall rating of severity. Part 2: Evaluation A preliminary evaluation study was undertaken to assess the inter- and intra-reliability of the scale. Ethical permission was obtained from the Harrow Research Ethics Committee. Associated reaction occurring during a single standardized task (sit-to-stand) was assessed on two separate occasions by two raters who were blinded to each other’s ratings. Method Raters Two senior neurological physiotherapists who were not involved in the focus groups or piloting of the tool were selected as raters (rater L and rater S) and remained constant throughout the


The Associated Reaction Rating Scale study. They had ve and seven years’ postqualication experience respectively. Subjects Ratings were undertaken on consecutive inpatients admitted to Northwick Park Hospital for rehabilitation, who have hemiplegia resulting from stroke, and who met the criteria for inclusion. Inclusion criteria included the ability to understand and follow simple commands, and to rise independently from sit-to-stand. Subjects were excluded if they had any pathological disorder of the nonaffected side, or any additional disorder affecting the hemiplegic upper limb, e.g. orthopaedic injury or signicant contracture of the upper limb (>30 loss of range at the elbow, wrist or metacarpo-phalangeal joints). Nineteen subjects completed the requirements of the study. These were 11 males and 8 females with a mean age of 53 years (range 22–79 years). Ten patients had right hemiplegia and nine had left hemiplegia. The mean time between onset of stroke and testing was 5.4 months (range of 1–27 months). The test procedure was standardized according to a previously published protocol for ‘sit-tostand’ studies on stroke patients.13 (The full testing protocol is available on request from the authors.) No specic training was given other than the opportunity to read the scoring guidelines and practise on one patient prior to commencing the study. Three ‘sit-to-stand’ trials were carried out for each test and raters were directed by the guidelines to score the worst performance (most severe associated reaction) observed over the three tests. All handling of the upper limb (necessary to achieve the start position and rate the release of associated reaction item) was carried out by the patients themselves. Raters were, however, told the degree of any xed contracture present in the upper limb prior to rating to allow them to differentiate between mechanical loss in range and associated reaction. This ‘non hands-on’ nature of the testing allowed the two raters to rate the patient simultaneously, but independently. Each patient was rated by both raters on two consecutive days. Ratings were made at the same time of day and before physiotherapy treatment.

729

Data handling and analysis Ratings were recorded for each item individually and summated to a total score. The most commonly rated (modal) score was also recorded. In instances where there was an even split between numbers, the higher of the scores (most severe rating) was taken to represent the modal score – for example, if the item scores for a particular patient were 1,1,2,2, the modal score would be 2 (moderate). Statistical analysis was undertaken using SPSS for Windows version 7.5.1. Weighted kappa values14 within 95% condence intervals were calculated using a specically designed Excel module. Because the Associated Reaction Rating Scale is an ordinal measure, nonparametric statistical tests were applied throughout. On a basic level, association between raters was assessed by examining the degree of correlation (Spearman rank test) for total and for overall (modal) scores. Wilcoxon matched pairs signed rank tests were used to test for systematic bias between raters. On a more detailed level, item by item analysis of agreement was undertaken by recording percentage of absolute agreement (where the two observers gave an identical rating) and weighted kappa values with 95% condence intervals. Results The Associated Reaction Rating Scale took approximately 10 minutes to complete including standardization of the test position and was described by the raters as ‘easy’ to use. Descriptive statistics for the Associated Reaction Rating Scale ratings for each rater on both occasions are presented in Table 2. Scores were more or less symmetrically distributed across the full range of the scale. There was no strong evidence of a oor or ceiling effect. Although the mean scores suggested a slight tendency for rater S to rate more severely than rater L, this only reached signicance for ‘excursion’ (Wilcoxon signed rank test p < 0.005) and was reected in the modal score (for day 1 only), but not the total summated score. Figure 1 shows scattergrams of the total scores between and within raters. There were good cor-


730 A Macfarlane et al. Table 2

Descriptive statistics of item scores for each rater on each occasions (19 paired observations) Excursion

Joints

Release

Function

Total

Overall severity (mode)

Day 1

Day 2

Day 1 Day 2

Day 1

Day 2

Day 1

Day 2

Day 1

Day 2

Day 1

Day 2

Rater L Range Median IQR

0–2 2 1–2

0–2 2 1–2

0–3 2 1–3

0–3 2 1–3

0–3 2 1–2

0–3 2 1–2

0–3 1 1–2

0–3 2 1–2

0–10 6 4–9

0–10 7 4–9

0–3 2 1–2

0–3 2 1–2

Rater S Range Median IQR

0–3 2 1–2

0–3 2 1–2

0–3 2 1–2

0–3 2 1–2

0–3 1 1–3

0–3 1 1–3

0–3 1 1–2

0–2 1 1–2

0–11 7 4–9

0–11 7 4–9

0–3 2 1–2

0–3 2 1–2

Wilcoxon z Signicance (p)

–2.0