A general definition of adherence to rehabilitation might be the degree to which a patient or subject ‘sticks’ to their rehabilitation program (Brewer, 1998a). Rehabilitation, however, is a voluminous construct, encompassing numerous behaviours that can be performed in a number of settings (e.g. clinic or home based) (Brewer, 1999). Adherence can relate to many aspects of the rehabilitation program such as participation in clinic based exercises and therapy, completion of home exercises, reduction of physical activity, and compliance with medication prescription (Brewer, 1998a B.W. Brewer, Fostering treatment adherence in athletic therapy, Athletic Therapy Today 3 (1998), pp. 30-32. View Record in Scopus | Cited By in Scopus (2)Brewer, 1998a and Brewer, 1999).
Many investigations of outcome after anterior cruciate ligament (ACL) reconstruction have meticulously reported the content of the rehabilitation programs that the subjects participated in postoperatively. Exercises have been described in detail and milestone periods of exercise progression recounted. Comparatively few studies, however, have addressed whether subjects were adherent to, or compliant with, their rehabilitation programs. This issue is particularly important when a study compares different treatment groups. A confounding variable in such a study is the rate of adherence of subjects in each treatment group.
Ideally, all subjects will conscientiously adhere to their prescribed rehabilitation. However, it is recognized that adherence to rehabilitation often may be less than absolute (Brewer, 1999 and Spetch and Kolt, 2001). For example, it has been previously reported that following ACL reconstruction, a proportion of patients failed to adhere to, or prematurely ceased, their participation in rehabilitation (Derscheid & Feiring, 1987). Derscheid and Feiring (1987) investigated the duration of treatment for a number of common athletic injuries treated at a sports medicine clinic. The study reported data for 59 patients who had undergone ACL reconstruction. Forty patients, although making progress with their rehabilitation, had discontinued therapy prior to achieving pre-injury levels. Of greater concern, almost 20% (11 patients) had ceased treatment without making any progress or even regressing in their condition (Derscheid & Feiring, 1987).
With regard to controlled clinical trials, if subjects in one group are less adherent to their rehabilitation than those in another group, it is possible that the disparate adherence rates among subjects may confound the findings. Unfortunately, it is difficult to control the degree of adherence to rehabilitation of all subjects who participate in a study. Yet this conundrum further highlights the importance of including some measure of adherence to determine whether the confounding variable of adherence may be potentially influencing results.
Reasoning and theoretical rationale suggest that patients who adhere to rehabilitation programs are more likely to improve when compared with non-adherent patients (Brewer, 1998a). In a review of adherence related literature, Spetch and Kolt (2001) noted that there was frequently an underlying assumption in adherence research that a direct association exists between adherence and outcome. However, such a belief remains largely unsubstantiated by scientific evidence. Recent recommendations have been made that there is a need to determine whether ‘adherence to rehabilitation is actually related to therapeutic outcome?’ and ‘does improving adherence improve recovery?’ (Brewer, 1998b, p. 78).
Several studies have attempted to determine whether adherence to rehabilitation impacts on therapeutic outcome (Brewer et al., 2004, Brewer et al., 2000b, Derscheid and Feiring, 1987 and Treacy et al., 1997). The results of such studies are conflicting and highlight the complex relationship between adherence and outcome. These studies demonstrate that measurement issues relevant to rehabilitation adherence are not clearly understood. The impact of a patient who is highly or poorly adherent, on the outcome following ACL reconstruction, has only recently been addressed. Further investigation is required for a more comprehensive understanding. Importantly, the incidence and consequence of subject adherence during clinical trials, and the implications that this may have on outcome, appears to be rarely considered in studies to date. Addressing the issue of adherence in clinical studies would permit even greater confidence in the findings of those investigations.
Although there is increased recognition that adherence is an important aspect of rehabilitation, research into characteristics and consequences of adherence is still in its infancy (Brewer, 1998b). The complex nature of adherence is reflected in the numerous ways that researchers have attempted to quantify it in the literature (e.g. appointment attendance, logbooks, patient self-reported questionnaires, therapist rating of adherence). A number of these measures have been used to evaluate orthopaedic conditions for rehabilitation following injury or surgery, making them potentially useful for studies involving the ACL population. Unfortunately, because research of adherence is a relatively new area, few of these adherence measures have been extensively tested for reliability or validity. Moreover, there appears to be no suitable ‘gold standard’ for evaluation of adherence to rehabilitation programs. This paper presents an overview of the strengths and weaknesses of a number of commonly used adherence measures and questionnaires. While adherence literature and questionnaires extends to a variety of physiotherapy related rehabilitation areas (e.g. cardiac rehabilitation), discussion will be limited to studies relating to orthopaedic and musculoskeletal conditions, which are more relevant to rehabilitation following ACL reconstruction.
2. Measures of adherence
Several measures of adherence will be reviewed in this paper. These include attendance scores, time spent exercising, Byerly’s adherence and participation score, Sluij’s questionnaire of patient compliance, the Athletic Trainers’ Perception of Athletes’ Effort Scale, the Rehabilitation Adherence Questionnaire, the Sport Injury Rehabilitation Adherence Scale, and electronic monitoring methods.
2.1. Attendance score
A popular indicator of adherence in the rehabilitation setting is measurement of a patient’s attendance of rehabilitation appointments. This is often determined by expressing the number of appointments attended as a ratio or percentage of the number of appointments scheduled during the time period of interest (Daly et al., 1995, Duda et al., 1989, Fisher et al., 1988, Lampton et al., 1993 and Udry, 1997).
The obvious advantage of using attendance as an indicator of adherence is that it is quick and easy to calculate (Brewer, 1998b). Despite its popularity, however, there are a number of shortcomings with this measure. Although attendance indices indicate the degree to which someone attends appointments, it fails to provide information regarding what they do during the sessions (Brewer, 1999 and Spetch and Kolt, 2001). Smith (1996) suggested that adherence is a multidimensional construct, of which attendance of appointments is only a single component. Smith (1996) identified that rehabilitation may also be undertaken at home, in gymnasium settings, or with personal trainers. These additional facets of rehabilitation are not assessed by the rate of attendance of appointments. Additionally, as motivated athletes are more likely to attend the majority of their appointments, attendance scores for this population are likely to be negatively skewed (Brewer, 1998b). Therefore, it has been recommended that attendance not be the sole measure of adherence (Brewer, 1998b and Brewer, 1999).
2.2. Time spent exercising as a measure of adherence
An adherence rating system similar to the attendance score was used by Rejeski, Brawley, Ettinger, Morgan, and Thompson (1997) to evaluate compliance during an exercise program for 439 patients with osteoarthritis. In this study compliance was rated using two indicators. The first was using the attendance score. In conjunction with the attendance score, the time spent exercising during each session was also measured (with the target being 40 min). As with the attendance score, although time spent exercising provides an indication of participation, limited information is obtained regarding how well patients adhered to their programs during the sessions.
2.3. Byerly’s adherence and participation score
An attempt to assess patient adherence during rehabilitation sessions was reported by Byerly, Worrell, Gahimer, and Domholdt (1994) who developed a brief measure of adherence based on attendance and participation. This score was used to evaluate adherence of 44 patients who had sustained musculoskeletal injuries and were participating in a rehabilitation program. The succinct nature of this score appears to make it suitable for raters who have limited time to devote to evaluating adherence. Patients were evaluated on a two-point scoring system by the therapist who monitored their rehabilitation program. One point was allocated for attending the rehabilitation session. Failure to attend was awarded zero points. A second point was awarded for completing 100% of prescribed exercises. If a patient failed to complete all exercises, three quarters of a point (0.75) was awarded for completing 75% of the exercises, 0.5 for completing 50% of the prescribed exercises, or 0.25 for completing 25% of the exercises. The authors made no reference to reliability testing or validation of their questionnaire. Moreover, the authors recommended that patients who scored between 1.75 and 2.0 were adherent, while those who scored less than 1.75 were non-adherent. These criteria appeared to be arbitrarily assigned, with no explanation or justification of how they were chosen.
2.4. Sluijs’ questionnaire of patient compliance
In a study to determine factors associated with exercise compliance during physiotherapy, Sluijs, Kok, and van der Zee (1993) developed a questionnaire and administered it to 1206 patients. The questionnaire consisted of a single question relating to exercise: ‘Did you manage to exercise regularly last week?’ Four responses were provided for subjects to choose from: (1) ‘not at all’, (2) ‘a little’, (3) ‘rather regularly’, and (4) ‘very regularly’. Only those with responses of ‘very regularly’ were considered compliant with their rehabilitation. The categorical nature of the responses decreased the sensitivity of detecting differences between adherent and non-adherent patients. The authors did not report whether their questionnaire had undergone any process of validation. The paper did, however, highlight that it had not been subjected to reliability testing as, ‘the questionnaires were anonymous’ (Sluijs et al., 1993, p. 43). The authors’ justification for failing to reliability test the questionnaire is unconvincing, as it could have been tested for reliability prior to (or even following) the commencement of data collection. Therefore, the results obtained using this questionnaire must be considered in light of the lack of reported reliability.
2.5. Athletic Trainers’ Perception of Athletes’ Effort Scale
A slightly more detailed assessment of a subjects’ effort during rehabilitation sessions was developed by Fields, Murphy, Horodyski, and Stopka (1995). The Athletic Trainers’ Perception of Athletes’ Effort Scale (ATPAES) score was based on the Borg Perceived Exertion Scale and used to categorize people as either ‘adherers’ or ‘non-adherers’. People are rated on a 6-20 point Likert-type scale with anchors of six corresponding to ‘no exertion at all’ and 20 denoting ‘maximal exertion’. Various numbers between the anchors were also designated verbal descriptors; for example, eight ‘extremely light’, 11 ‘light’, 15 ‘hard (heavy)’, 19 ‘extremely hard’.
The authors stated that the ATPAES system satisfied the criterion for logical validity, as the ATPAES system had been favourably evaluated by certified sports trainers. Unfortunately, no reference was made to more extensive validity assessment (such as comparing ATPAES scores with other adherence measures) or test-retest reliability evaluation of the questionnaire. Attendance was also measured in the study by Fields, Murphy, Horodyski, and Stopka et al. (1995) and subjects were considered to have adhered to their rehabilitation if they attended at least 75% of rehabilitation sessions and scored 12 or greater on the ATPAES questionnaire. However, this appeared to be an arbitrary cut-off for distinguishing adherers from non-adherers, with no explanation of how it was derived. The lack of rigorous reliability and validity testing of the ATPAES questionnaire currently limits its usefulness as an adherence measure for experimental studies.
2.6. Rehabilitation Adherence Questionnaire
Fisher et al. (1988) developed a more extensive questionnaire that has been used in a number of studies to enable self-assessment of adherence during the rehabilitation period (Byerly et al., 1994, Fields et al., 1995 and Fisher et al., 1988). The Rehabilitation Adherence Questionnaire (RAQ) consists of 40 questions pertaining to six domains expected to influence adherence. Each question requires the participant to circle a response on a four-point Likert-type scale. The title of the questionnaire, the Rehabilitation Adherence Questionnaire, may be misleading with regard to the information sought. It appears that, rather than measuring the rate of adherence to rehabilitation, the questionnaire actually requires a self-assessment of factors that might influence their rehabilitation adherence. This is illustrated by the example question for the pain domain; while a patient may strongly agree that their rehabilitation program was physically painful, their perception of the pain they experienced is not a measure of their adherence level during rehabilitation.
Fisher et al. (1988) stated that the questionnaire satisfied the requirements for face validity as the items were derived from analysis of the content of adherence literature. However, Fields et al., 1995 and Fisher et al., 1988 acknowledged that the RAQ had not been subjected to more rigorous standards of reliability or validity testing. In a psychometric evaluation of the RAQ, Brewer, Daly, Van Raalte, Petitpas, and Sklar (1999) reported that the psychometric properties were weak with the RAQ subscales failing to exhibit adequate internal consistency or criterion validity. Moreover, a low magnitude of correlation was observed between the RAQ and three alternative measures of adherence (attendance, therapist completed scores on the Sport Injury Rehabilitation Adherence Scale, and patient self-reporting of home rehabilitation adherence). Finally, when rated by several experts in the field of psychological factors in sports injury, many items of the RAQ lacked content validity. Brewer et al. (1999) advised caution when investigating psychological factors associated with adherence and recommend that alternative measures be used.
2.7. Duda’s evaluation of completion of exercises and exercise intensity
In addition to attendance and time spent exercising, participation during rehabilitation sessions has also been quantified by measuring how many of the prescribed exercises were completed (Duda et al., 1989). In a study of adherence of 40 athletes who had sustained sporting related injuries, Duda et al. (1989) calculated a ratio of exercises completed to exercises prescribed. Missed appointments were scored as zero. This study also attempted to report the amount of effort that subjects exerted during treatment sessions. The authors postulated that for a rehabilitation session to be successful, the patient must exert a sufficient level of effort. Trainers who conducted rehabilitation sessions were required to rate each subject’s intensity on a five-point scale. A score of one corresponded to minimum effort with the subject being ‘walked through exercises’ and a score of five indicated maximal effort and the subject ‘worked as hard as possible’. The authors did not report any reliability or validity findings for either of these adherence measures.
After consideration of the adherence measures discussed above, a number of common aspects are evident. There appears to be a preference for short instruments that are not time consuming for the examiner to complete. Few of the currently available adherence measures have been reliability tested. Furthermore, validity considerations were frequently limited to face (logical) validity. Therefore, there is an apparent need for the development of a reliable and valid method of evaluating rehabilitation adherence. One adherence questionnaire that has been subjected to more rigorous reliability testing and validation is the Sport Injury Rehabilitation Adherence Scale (SIRAS).
2.8. Sport Injury Rehabilitation Adherence Scale
An instrument that has been increasingly advocated as a valid measure of behaviour during rehabilitation is the SIRAS (Shaw et al., 2004 and Spetch and Kolt, 2001). The SIRAS, which consists of three items (questions) which were derived from previous adherence literature, has been used to assess adherence rates to clinically based rehabilitation sessions (Brewer et al., 2000c).
When using the SIRAS the treating therapist evaluates each patient with regard to their intensity when completing exercises, the frequency with which they follow advice from the therapist, and how receptive they are to changes in their rehabilitation program (Brewer et al., 2000c). Each item is rated on a five-point Likert scale with anchors of ‘minimum effort/maximum effort’, ‘never/always’, and ‘very unreceptive/very receptive’ provided for each item, respectively (Brewer et al., 2000a and Laubach et al., 1996). The scores for each individual component are summed to produce an overall adherence score out of 15 (Daly et al., 1995).
The SIRAS has a number of strengths that make it ideal for use in clinically based experimental research. As the questionnaire is comprised of only three questions, it is relatively quick and simple for a therapist to complete following clinical examination and treatment sessions. In addition, it has been used previously in clinically based studies. Importantly, the SIRAS has been subjected to a number of studies that have investigated its reliability and validity. Furthermore, regarding its use in studies of patients following ACL reconstruction, the SIRAS questionnaire has previously been applied to studies of ACL injured and reconstructed populations (Brewer et al., 2004, Brewer et al., 2000a, Brewer et al., 2000b, Brewer et al., 2000c, Daly et al., 1995 and Laubach et al., 1996).
The SIRAS has been found to have good test-retest reliability (ICC(2,1)=0.77) when re-administered at a one-week interval (Brewer et al., 2000c). Moderate inter-tester reliability has also been reported (r=0.57, p<0.005) (Brewer et al., 2000c). The limitation of these inter-tester reliability studies was that they assessed different therapists during different treatment sessions. This shortcoming was subsequently addressed by Brewer et al. (2002), who assessed inter-tester reliability of 43 inexperienced raters (undergraduate athletic trainer students and physical therapy students). The participants rated adherence of a standardised rehabilitation performance by viewing three segments of 9-min video footage of a patient undergoing rehabilitation. Each video segment presented minimal, moderate, and maximal adherence performance, respectively. Inter-tester agreement was found to range from rWG(i)=0.89-095 and rater-agreement index=0.84-0.90 for each video segment viewed. The same study also assessed the inter-tester reliability of two more experienced testers in a clinical environment. Twelve patients were assessed by each tester over four rehabilitation sessions producing a high rater-agreement index across the four sessions of 0.94.
The questions contained within the SIRAS were based on previous adherence literature from Duda et al., 1989 and Meichenbaum and Turk, 1987, contributing to the face or logical validity of the questionnaire. SIRAS scores have been reported to correlate significantly with attendance scores, another accepted and often used indictor of adherence (Brewer et al., 2004, Brewer et al., 2000c and Daly et al., 1995). Using Pearson correlation analysis, Brewer et al. (2000c) demonstrated a statistically significant positive correlation of the SIRAS score and rehabilitation attendance (r=0.21, p<0.05). This study was performed on 145 patients attending rehabilitation at an orthopaedic physiotherapy clinic for treatment of a variety of injuries (as identified by Derscheid & Feiring, 1987) including knee injuries and torn ACLs.
Daly et al. (1995) applied Pearson correlation coefficients to examine the relationship between attendance and SIRAS scores for 31 patients following knee surgery. A statistically significant positive correlation was found between the attendance scores and adherence scores (r=0.49, p<0.05). Furthermore, Brewer et al. (2004) demonstrated an intercorrelation of 0.25 (p<0.05) between attendance and SIRAS scores for 108 patients following ACL reconstruction.
The authors concluded that the findings of statistically significant positive correlations between adherence and attendance support the construct validity of the SIRAS score (Brewer et al., 2004, Brewer et al., 2000c and Daly et al., 1995).
With regard to questionnaire design, internal consistency indicates the degree to which the items of the questionnaire measure various aspects of the same characteristic, and nothing else (Portney & Watkins, 2000). In the case of the SIRAS, internal consistency relates to adherence. Internal consistency is frequently assessed by calculation of the correlation between the individual items that comprise the questionnaire (Portney & Watkins, 2000). The internal consistency reliability of the SIRAS score has been reported in a number of studies (Brewer et al., 2000c and Daly et al., 1995). In a study investigating the relationship between cognitive appraisal, emotional adjustment and adherence for 31 subjects who had sustained a knee injury or undergone surgery, internal consistency of the SIRAS score was found to be adequate (Î±=0.81) (Daly et al., 1995). This finding was similar to that of Brewer et al. (2000c) who reported a Cronbach’s alpha level of 0.82 for internal consistency when testing the SIRAS score on 145 patients attending rehabilitation. As the SIRAS score was likely to be applied repeatedly over the duration of a rehabilitation program, the internal consistency reliability for multiple measures has also been determined. A Cronbach’s alpha coefficient of 0.86 was found for multiple administrations of the SIRAS score (Brewer et al., 2000c) when applied to 32 subjects for 10 rehabilitation sessions. This finding supports the use of an aggregation of SIRAS questionnaires to produce an internally consistent index of adherence to a rehabilitation protocol over a number of sessions.
The construct validity of the SIRAS was demonstrated by Brewer et al. (2002). In this study, three segments of video footage (depicting performance of minimal, moderate, and maximal rehabilitation adherence, respectively) were viewed by 43 testers (undergraduate athletic trainer and physical therapy students). Bonferroni pairwise analysis, following mixed ANOVA analysis, demonstrated that SIRAS scores for the maximally adherent video footage were significantly higher than SIRAS scores for the moderately adherent video footage (p<0.001). Similarly, SIRAS scores for the moderately adherent video footage were significantly higher than SIRAS scores for the minimally adherent video footage (p<0.001). Thus, a linear relationship was demonstrated between SIRAS scores and rehabilitation adherence, supporting construct validity of the SIRAS score.
Brewer et al. (2000c) also performed analyses of variance on the adherence scores for 145 subjects to investigate the association of SIRAS scores with gender and athletic involvement. The study reported that SIRAS scores did not vary as a function of subject gender or the subjects’ level of athletic involvement (Brewer et al., 2000c). Therefore, the authors concluded that the SIRAS questionnaire can be applied to groups containing both males and females of varying athletic participation levels.
The more thorough assessment of reliability and validity of the SIRAS score, relative to other measures of adherence, and its ease of implementation support the use of this instrument for measuring adherence in clinical and experimental environments. It should be noted, however, that despite the reliability and validity strengths of the SIRAS score as an adherence indicator, this score has some inherent limitations. The multidimensional and complex construct of adherence is simplified to just three subjective criteria, perhaps limiting inferences that can be drawn. Moreover, there may be a tendency for subject populations of elite athletes or those sustaining sporting injuries to be innately inclined to greater adherence levels, potentially skewing data or producing ceiling effects.
Cite This Work
To export a reference to this article please select a referencing style below: