Modeling and Rehabilitation Following Anterior Cruciate Ligament Reconstruction

Ralph Maddison, Ph.D. and Harry Prapavessis, Ph.D.

Department of Sport and Exercise Science
University of Auckland

Mark Clatworthy, MBChB, FRCS

Auckland Bone and Joint Clinic
Epsom, Auckland, New Zealand

ABSTRACT agery (3–5), goal-setting (6), electromyographic biofeedback

Background/Purpose: This study sought to examine the ef- (7), and stress inoculation training (8).
fectiveness of a modeling video to reduce preoperative percep- An intervention area that has received limited attention in
tions of anxiety and pain, as well as to increase postoperative the realm of athletic injury rehabilitation is observational learn-
self-efficacy and functional outcomes after anterior cruciate lig- ing or modeling (9). Despite this, modeling has been found to be
ament reconstruction. Methods: Following baseline assessment a powerful instructional tool for the acquisition of motor skills,
of state anxiety, perceptions of expected pain, injury severity, psychological responses, and behavior change in physical activ-
and knee function (International Knee Documentation Commit- ity contexts (10,11). Moreover, a recent review of preparation
tee [IKDC] system), patients scheduled for surgical reconstruc- interventions for adult patients undergoing surgery and/or inva-
tion of the anterior cruciate ligament were randomly assigned to sive medical procedures found that modeling combined with in-
either a modeling intervention or a control group. Psychologi- struction in coping strategies is highly effective in producing
cal assessments were repeated preoperatively for expected pain positive outcomes (12). Taken together, the theoretical and em-
and anxiety. Actual pain was assessed preoperatively, prior to pirical support for modeling makes it a viable intervention strat-
discharge, and at 2 weeks postoperatively. Rehabilitation egy, particularly in the realm of athletic injury rehabilitation.
self-efficacy was assessed prior to discharge and at 2 and 6 Only one study, by Flint (13), has explored the effectiveness
weeks postoperatively. IKDC functional assessments were re- of a modeling intervention to enhance physical and psychologi-
peated at 6 weeks postoperatively, whereas range of motion was cal rehabilitation outcomes in an athletic setting. This is
assessed at 2 and 6 weeks postsurgery. Results: Compared with unfortunate, because the extension of this technique into the
the participants in the control condition, participants assigned “realm of sport injury rehabilitation affords motivation, in-
to the modeling intervention reported significantly lower per- jury-rehabilitation information, and behavioral cues for recover-
ceptions of expected pain preoperatively and significantly ing athletes” (14, p. 221). In her innovative study, Flint exam-
greater self-efficacy at predischarge to perform rehabilitation ined the role of coping models compared to no models on
tasks. Those who received the modeling intervention also expe- psychological factors and functional outcomes following a reha-
rienced significantly better IKDC objective functional outcome bilitation program for anterior cruciate ligament reconstruction
scores compared with their control counterparts. No psycholog- (ACLR) among 10 female basketball players. Results showed
ical variables mediated relations between the intervention and increased self-efficacy 3 weeks after surgery among those who
functional outcomes. Conclusions: The data suggest that watched a modeling videotape compared with the 10 matched
watching a modeling video may be an effective prophylactic controls. Differences in the expected direction were also noted
treatment to decrease perceptions of expected pain, increase re- for early attainment of functional milestones (i.e., walking, jog-
habilitation self-efficacy, and provide an early stimulus with re- ging, running, and return to full function), although these differ-
spect to early function. ences were statistically nonsignificant. Flint’s study had a mod-
est sample size and hence was underpowered, which may have
(Ann Behav Med 2006, 31(1):89–98) accounted for the nonsignificant results for functional mile-
stones. In addition, the modeling intervention was introduced
INTRODUCTION postoperatively and did not provide an indication of its benefit
The role that psychology has to play in understanding the for reducing preoperative anxiety.
recovery of sport and recreation related injuries continues to Research findings have supported the effectiveness of mod-
grow (1,2). Various psychological interventions have been ad- eling in reducing both anxiety and perceptions of pain preopera-
vocated or used in the injury recovery setting. These include im- tively (15,16). Modeling as an intervention has the potential to
help reduce a person’s perception of pain and anxiety and to in-
crease one’s rehabilitation self-efficacy early on in the rehabili-
Reprint Address: R. Maddison, Ph.D., Clinical Trials Research Unit, tation process. This is particularly important when one consid-
School of Population Health, University of Auckland, Private Bag ers the comments of DeCarlo, Sell, Shelbourne, and Klootwyk
92019, Auckland, New Zealand. E-mail: r.maddison@ (17), who suggested that if certain problems are allowed to de-
ctru.auckland.ac.nz velop early in the ACLR postoperative period, they will be very
© 2006 by The Society of Behavioral Medicine. difficult to eliminate in the long term and will ultimately have a

89
90 Maddison et al. Annals of Behavioral Medicine

detrimental effect on the patient’s outcome. DeCarlo et al. high-

lighted the following variables as important in the early postop-
erative period: (a) lack of full terminal extension equal to the
contralateral side, (b) lack of quality straight leg raise and ade-
quate leg control, and (c) significant hemarthrosis of the surgical
knee.
The population examined by Flint also provides a suitable
avenue for continued research, because acute disruption of the
ACL is one of the more common and debilitating sport- and rec-
reation-related injuries (18,19). Surgical ACLR is associated
with an extensive period of rehabilitation (6–9 months) involv-
ing home- and clinic-based strength and flexibility exercises
along with cryotherapy (icing) (17). Given the nature of rehabil-
itation associated with ACLR, opportunity exists to examine the
potential utility of psychological interventions to augment the
recovery process (20). The purpose of this study was to examine
the effectiveness of a video coping modeling intervention in
promoting early recovery following ACLR. It was predicted that
athletes who received a coping modeling video intervention
would report lower preoperative anxiety and perceptions of ex-
pected pain and would report greater self-efficacy for rehabilita-
tion compared to the nonintervention group. It was also hypoth-
esized that participants in the intervention group would show
greater improvements in functional milestones (e.g., range of
motion [ROM] and crutch use) than those in the nonintervention
group.

METHOD
Participants and General
Recruitment Procedure FIGURE 1 Participant recruitment flow chart.
Seventy-two participants scheduled for ACLR agreed to
participate and were prospectively recruited from the Auckland
Bone and Joint Surgical clinic. Eight did not have the operation ter your knee surgery” (22). Assessments occurred at baseline
as planned, 3 withdrew from the study, and 3 had postoperative and preoperatively. A similar scale was also used to assess per-
complications that disrupted the standard rehabilitation proto- ceptions of actual pain, preoperatively, predischarge and 2
col; hence, the final sample was 58 (see Figure 1). It was calcu- weeks postoperatively. Participants were asked to “write down a
lated that a sample of approximately 38 participants in each con- number on a scale ranging from 0 (no pain) to 100 (pain as bad
dition (modeling vs. control) would be needed to provide power as it could be) that best describes how much pain you have right
of 80% (α = .05) and to detect a large effect (i.e., .40) between now” (22).
conditions on the variables of interest (21). Despite comprehen-
sive attempts, we were unable to recruit the necessary 76 partici- Anxiety. The State–Trait Anxiety Inventory (STAI) (23)
pants to fully power this study. was used to measure state anxiety at baseline and preoperatively.
Participants ranged in age from 15 to 53 years (M = 30), The STAI has been shown to have both construct validity and re-
with a greater distributions of males (68%) than females (32%). liability (24). Participants were asked to respond to 20 state-
The following ethnic groups were represented: NZ Pakeha, ments that describe “how you are feeling right now” on a scale
71%, NZ Maori, 14%, Pacific Islands, 5%, and other, 10%. that ranged from 1 (almost never) to 4 (almost always). Reliabil-
Rugby was overrepresented as the major cause of injury (32%), ity was acceptable for this scale at the two time points (baseline
followed by soccer (18%), snow sports (11%), netball (8%), wa- α = .90 and preoperative α = .91).
ter sports (5%), and miscellaneous activities (26%). Descriptive
data are provided in Table 1. Rehabilitation self-efficacy. Three types of self-efficacy
were assessed in the present study. The self-efficacy scales as-
Psychological Measures sessed confidence for performing knee rehabilitation exercises,
Pain. A single item assessed “perceptions of expected as well as functional task outcomes (walking with and without
pain” and asked participants to “write down a number on a scale crutches) progressively with increasing frequency and duration
ranging from 0 (no pain) to 100 (pain as bad as it could be) that on a scale of 0% (no confidence) to 100% (complete confidence)
best describes how much pain you think you will experience af- (25).
Volume 31, Number 1, 2006 Modeling and Rehabilitation 91

TABLE 1
Descriptive Statistics of the Variables of Interest

Interventiona Controlb Total

Variable M SD M SD M SD

Baseline expected pain 62.17 17.06 64.24 22.76 63.17 19.86

Preoperative expected pain 56.33 18.04 66.79 15.73 61.38 17.62
Preoperative actual pain 13.33 12.61 23.24 24.28 18.11 19.63
Predischarge actual pain 33.51 26.02 40.32 22.46 36.80 24.40
2 weeks actual pain 27.95 18.53 27.84 18.65 27.90 18.42
Baseline state anxiety 36.93 8.85 34.25 7.80 35.64 8.40
Preoperative state anxiety 38.17 10.19 37.52 6.52 37.58 8.55
Discharge CSE 72.85 16.72 57.17 29.28 65.28 24.72
Discharge WSE 59.20 22.17 44.04 24.48 51.88 24.34
Discharge ESE 87.38 11.94 76.38 23.44 82.08 19.07
2-week WSE 69.68 23.30 72.73 23.02 71.15 23.01
2-week ESE 84.25 12.36 84.66 16.90 84.54 14.60
6-week WSE 93.09 8.66 89.42 10.81 91.32 9.84
6-week ESE 88.66 10.70 85.48 13.40 87.12 12.07
ROM baseline 130.82 10.44 129.77 10.55 131.96 10.39
ROM 2 weeks 103.17 14.29 102.23 16.94 102.71 15.50
ROM 6 weeks 124.03 8.96 121.78 8.67 122.94 8.82
IKDC (O) (baseline) 3.03 .61 3.28 .53 3.15 .58
IKDC (S) (baseline) 53.53 12.81 52.25 11.89 52.91 12.28
IKDC (O) (6 weeks) 2.24 .43 2.56 .48 2.42 .48
IKDC (S) (6 weeks) 61.18 8.04 57.02 9.58 59.17 8.98
Crutch use in days 5.54 2.31 9.34 4.03 7.38 3.75

Note. All data presented are raw and uncorrected. CSE = crutches self-efficacy; WSE = walk self-efficacy; ESE = exercise self-efficacy; ROM = range of
motion; IKDC = International Knee Documentation Committee; O = objective; S = subjective.
an = 30. bn = 28.

A crutches self-efficacy (CSE) scale assessed confidence to Cronbach’s alpha values were as follows: predischarge, α = .96;
walk with crutches for increasing periods (i.e., 10, 20, and 30 2 weeks postoperative, α = .93; and 6 weeks postoperative,
min) at two speeds (i.e., slow and moderate pace). The CSE α = .95.
scale was framed for the next 2-week period, and a key was pro-
vided to define the various intensity levels. Mean scores were
Functional Milestones
derived from the six items, with higher values indicating greater
efficacy to exercise for a longer time and at a faster pace. Reli- Crutch usage. Participants were questioned at 2 weeks
ability for the CSE scale was acceptable at the predischarge as- post ACLR to determine the length of time they required the use
sessment (α = .95). of crutches to assist with walking (in days). Higher scores repre-
A walking self-efficacy (WSE) scale assessed confidence sented longer periods of crutch-assisted walking.
to walk for increasing periods (i.e., 10, 20, and 30 min) at three
speeds (i.e., slow, moderate, and moderately fast pace). The Knee assessment. The International Knee Documentation
WSE scale was framed for the next 4-week period, and a key Committee (IKDC) system (26,27) was used to clinically evalu-
was provided to define the various intensity levels. Mean scores ate the knee at two different times (baseline and 6 weeks postop-
were derived from the nine items, with higher values indicating eratively). The IKDC Standard Evaluation Form consists of
greater efficacy to exercise for a longer time and at a faster pace. eight groups, including patient’s subjective assessment of func-
Cronbach’s alpha values were as follows: predischarge, α = .97; tion, symptoms, ROM, ligament examination, compartmental
2 weeks postoperative, α = .97; and 6 weeks postoperative, findings, harvest site pathology, x-ray findings, and functional
α = .95. tests, but only the first four groups are included in the final over-
An exercise self-efficacy (ESE) scale assessed confidence all IKDC rating (28). Each group in the IKDC assessment is
to complete rehabilitation exercise for increasing periods (i.e., graded as A (normal), B (nearly normal), C (abnormal), or D
10 and 20 min) at three different times (i.e., once, twice, and (severely abnormal). The scale incorporates an objective (sur-
three times a day). Mean scores were derived from the six items, geon) and subjective (patient) assessment. Patient symptoms are
with greater values indicating greater efficacy to perform reha- evaluated at the following four activity levels: (a) strenuous ac-
bilitation exercises more frequently and for a longer time. tivity (jumping, pivoting, and hard cutting), (b) moderate activ-
92 Maddison et al. Annals of Behavioral Medicine

ity (heavy manual work, skiing, and tennis), (c) light activity relevant information and cues specific to their own stage of pro-
(light manual work, jogging), and (d) sedentary (housework and gression.
daily living activities). Overall evaluation is determined by the
worst grade in the following four categories: (a) patient subjec-
tive assessment, (b) symptoms (pain, swelling, and giving way), Procedure
(c) ROM, and (d) ligament evaluation (Lachman test, pivot shift The present study incorporated a randomized, controlled,
test, and anterior draw). Adequate construct and concurrent va- prospective, repeated measure design. Ethical approval was se-
lidity of the IKDC were reported by Irrgang et al. (28). In the cured before proceeding. Participants were approached during
present study, numerical values were allocated to each of the their initial surgical consultation and received verbal and written
IKDC grades (e.g., A = 1, B = 2, etc.) to permit analysis. information. Once written consent was obtained, participants
provided demographic information (age, gender, ethnicity, and
injury information). All participants were randomized to either
ROM. ROM was considered independently as a functional an intervention (modeling video) or a control condition (Figure
milestone and was assessed at baseline and 2 and 6 weeks after 1) using SPSS 11.5 software. Comparison of baseline group
ACLR using standardized goniometry procedures (29). To en- means for age, injury severity, and psychological variables re-
hance the reliability for ROM assessment, the following steps vealed no differences, suggesting pretreatment group equiva-
were taken: (a) standardized assessment of the joint and (b) the lence on these variables.
use of a single trained assessor to ensure consistency between In our study, all participants underwent arthroscopic auto-
successive measurements of ROM. Values for knee extension graft hamstring-tendon ACL reconstruction. The procedure was
and flexion were obtained. Difference values (i.e., extension mi- performed as a day case, with participants admitted the morning
nus flexion) were used in final analyses, with greater scores rep- of the procedure and then discharged later that evening. The
resenting greater ROM. same consultant orthopedic surgeon performed all of the
ACLRs in the study.

The Intervention
Two coping model videos (DVDs) were developed by Maddison Baseline data collection. Participants completed the IKDC
to represent the first 6 weeks of rehabilitation post-ACLR. The subjective (S) form, STAI, and the expected pain assessment.
first video (9 min) detailed the preoperative through to the ROM measurements and the IKDC objective (O) assessment
2-week postoperative period. This video was viewed preopera- were also obtained.
tively and prior to discharge from hospital (to reinforce the key
points). The second video (7 min) addressed the 2- to 6-week Preoperative period. On the day prior to their operation,
postoperative period and was viewed at these time points, re- participants in the intervention condition watched the modeling
spectively. The video consisted of edited interviews and various video before completing the STAI and actual and expected post-
action shots of the models performing a number of tasks (e.g., operative pain assessments, whereas the control group com-
stair climbing and walking). During the videos, models were pleted the psychological measures only.
shown performing appropriate time-matched rehabilitation ex-
ercises. For example, in the first video (early stages), models
were shown performing rehabilitation exercises to improve ex- Predischarge. Prior to discharge from hospital, the inter-
tension and flexion and walking with and without crutches. The vention group watched the modeling video before completing
second video showed the models demonstrating further im- the psychological inventories (actual pain, CSE, WSE, and ESE
provement in ROM, cycling on a stationary bicycle, stair climb- scales), whereas the control group completed the psychological
ing, and so on. Consistent with social learning theory, and to en- inventories only.
hance the attentional and motivational properties of the videos,
four models (two men and two women) between 20 and 40 years Two weeks postoperative. At the 2-week postoperative as-
of age, were filmed to ensure that observers would identify with sessment, participants in the intervention group watched the
at least one model with respect to age, gender, and so forth (30). modeling video before completing the psychological invento-
During the edited interviews, models detailed how they had sus- ries (actual pain, WSE, and ESE scales). The control group
tained their original injury, their thoughts and feelings associ- completed only the psychological inventories. Functional mile-
ated with the injury, and the surgery they underwent. Models stones (ROM and crutch use) were also assessed.
then detailed the types of problems they had faced during the
various stages of the rehabilitation process (e.g., pain, frustra-
tion, transport, motivation, etc.) and provided strategies they Six weeks postoperative. At the 6-week postoperative as-
used to overcome these issues (e.g., use of appropriate analge- sessment, those in the intervention group watched the modeling
sia, using the cryocuff, setting targets or goals, and having ade- video before completing the psychological inventories (WSE
quate social support). Finally, the models detailed their original and ESE scales). The control group completed the psychologi-
expectations and actual progress regarding functional outcomes cal inventories only. Functional milestones (ROM, IKDC sub-
(13). It was anticipated that observers of the video would glean jective and objective) were also assessed.
Volume 31, Number 1, 2006 Modeling and Rehabilitation 93

RESULTS meet the conditions of mediation. Mediation requires a demon-

Treatment of the Data stration that (a) the independent, mediating, and dependent
variables are significantly related and (b) the direct effects of the
To address the main hypotheses, two forms of analyses
independent variables are significantly reduced when the medi-
were conducted. When baseline data were available to serve as
ator is introduced into the analysis (33).
a covariate, one-way (intervention vs. control) analyses of
covariance (ANCOVAs) were performed on each of the psycho-
logical measures (actual and expected pain and anxiety). Prior to Psychological Variables
conducting these analyses, the assumptions underlying the use Descriptive data. As can be seen in Table 1, participants
of ANCOVA (i.e., reliability of covariates, linear relationship appeared to have relatively low levels of state anxiety. In gen-
between dependent variable and covariates, homogeneity of re- eral, walking and jogging self-efficacy increased across time.
gression slopes) were tested and satisfied (31). The alpha level Correlations between the variables of interest revealed a number
for the ANCOVA analyses was .05, with effect sizes reported of patterns between the psychological variables (see Table 2).
(´η2). All skewed data were subjected to logarithmic transforma- State anxiety at baseline was inversely related to exercise
tion to reduce a potential spurious influence of extreme scores self-efficacy at predischarge (r = –.28, p < .05). Baseline percep-
(31). tions of expected pain was inversely related to walking
When baseline data were not available, a series of repeated self-efficacy at 6 weeks (r = –.36, p < .01). Actual pain (preoper-
measure analyses of variance (ANOVAs) were conducted (reha- ative and discharge) was inversely related to exercise self effi-
bilitation self-efficacy). The alpha value for these ANOVA anal- cacy at discharge (rs = –.27 and –.26, ps < .05, respectively),
yses was .05. Any significant interactions were examined with whereas actual pain (preoperative) was inversely related to
planned multiple comparison Bonferroni tests. Corresponding walking self-efficacy at 6 weeks (r = –.35, p < .01).
measures of effect sizes, ´η2 and Cohen’s d, are reported. All data
were assessed for the various requirements of ANOVA (32). All Anxiety and pain. No condition effect (control vs. inter-
skewed data were subjected to logarithmic transformation to re- vention) was found for preoperative anxiety, F(1, 56) = 85, p =
duce a potential spurious influence of extreme scores. .36, ´η2 = .02. A general increase in state anxiety from baseline to
Path analysis was also conducted to elucidate the degree to the preoperative assessment was observed in both groups. With
which the psychological variables mediated the relationship be- respect to expected pain, a significant condition effect was
tween the modeling intervention and functional outcome vari- found, F(1, 56) = 5.42, p < .05, ´η2 = .10. As can be seen in Fig-
ables. For this form of analysis to take place, variables had to ure 2, participants in the modeling condition reported fewer per-

TABLE 2
Correlations for the Variables of Interest

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

1. Anxiety (B) — .74** .07 .03 .03 .21 .41** –.18 .01 –.28* –.22 –.13 –.19 –.06 .03 –.20 .08 –.11 –.03 –.13
2. Anxiety (P) — .22 .13 –.10 .18 .32* –.19 –.10 –.23 –.09 –.08 –.12 .01 .12 –.14 .06 .05 .17 –.03
3. Ex Pain (B) — .34** .37** .37** .21 –.05 –.11 –.14 –.13 –.04 –.36** .18 .25 –.35** .08 –.07 .23 –.12
4. Ex Pain (P) — .40** .39** .11 –.21 –.24 –.25 –.00 –.14 –.11 .05 .04 –.12 .06 .03 .02 –.01
5. Pain (P) — .42** .45** –.06 –.07 –.27* –.16 –.12 –.35** –.01 –.35 –.12 –.01 –.25* .05 .10
6. Pain (D) — .36** –.21 –.18 –.26* –.13 –.02 –.15 .03 –.08 .08 .18 .03 .10 .01
7. Pain (2W) — –.06 .01 –.18 –.24 –.17 –.17 –.16 .01 –.11 –.04 –.17 –.08 .02
8. CSE (D) — .70** .60** .24 .17 .08 .13 –.15 .18 –.18 .11 .06 –.10
9. WSE (D) — .51** .16 .11 .02 .06 –.01 .13 –.43** .12 .13 –.08
10. ESE (D) — .33* .36** .30* .14 –.17 .19 –.18 .21 –.08 –.13
11. WSE (2W) — .62** .39** .37** –.07 .02 –.00 .06 .14 .05
12. ESE (2W) — .29* .63** –.10 .06 –.07 .07 .10 .04
13. WSE (6W) — .13 –.17 .11 .03 –.01 –.26* .11
14. ESE (6W) — –.15 –.02 –.22 .05 –.01 .02
15. IKDC (O) (6W) — –.67** .09 .02 –.04 –.51**
16. IKDC (S) (6W) — –.15 .19 .18 .34**
17. Crutch use — .05 –.16 –.24
18. ROM (B) — .10 –.06
19. ROM (2W) — .18
20. ROM (6W) —

Note. B = baseline; P = pre-operation; Ex Pain = expected pain; Pain = actual pain; D = predischarge; 2W = 2 weeks; CSE = crutches self-efficacy; WSE =
walking self-efficacy; ESE = exercise self-efficacy; 6W = 6 weeks; IKDC = ; International Knee Documentation Committee; O = objective; S = subjective;
ROM = range of motion.
*p < .05. **p < .01.
94 Maddison et al. Annals of Behavioral Medicine

2.47, p = .01, d = .05, but not at the 2-week, t(56) = –0.50, p > .05,
d = .13, and 6-week assessments, t(56) = 1.43, p > .05, d = .33.

Exercise self-efficacy. No significant time effect, F(2, 55)

= 2.18, p = .12, ´η2 = .07, was seen for exercise self-efficacy, but
a significant Time × Condition interaction effect, F(2, 55) =
3.07, p = .05, ´η2 = .10, was evident. Follow-up analyses re-
vealed significant differences at predischarge, t(56) = 2.27, p <
.05, d = .47, but not at the 2-week, t(56) = –0.12, p = .90, d = .03,
and 6-week, t(56) = –0.10, p > .05, d = .26, assessments. The
FIGURE 2 Perception of preoperative expected pain condition effect modeling group reported greater efficacy to perform rehabilita-
(adjusted means and standard error).
tion exercises after watching the video compared to the control
group at predischarge only (Figure 4).
ceptions of expected pain compared to those in the control con-
dition. Of interest, those in the modeling condition showed a
Functional Outcomes
decrease from baseline to preoperative, whereas the control par-
ticipants showed an increase from baseline to preoperative (Ta- Descriptive data. As can be seen in Table 1, IKDC objec-
ble 1). No condition effect was found for actual pain, F(2, 54) = tive and subjective scores improved, whereas ROM scores be-
0.66, p = .52,´ η2 = .02. For both groups, actual reported pain gen- gan to return to baseline values for all participants across respec-
erally increased from the preoperative to the predischarge pe- tive assessment periods. Correlations between the functional
riod, before decreasing at 2 weeks (Table 1). outcome variables showed the following pattern of relationships
(see Table 2). Range of motion at 6 weeks was related to IKDC
objective and subjective scores (r = –.51, p < .01, and r = .34, p <
Rehabilitation Self-Efficacy
.01, respectively).
Crutches self-efficacy. Because crutch self-efficacy was
assessed only at predischarge, a one-way ANOVA was per-
formed. Results revealed significant group differences, F(1, 56) IKDC measures. The ANCOVA results revealed a signifi-
= 6.38, p < .01, d = .53. The modeling group reported greater cant condition effect for IKDC objective scores, F(2, 55) = 6.53,
confidence to walk with crutches compared to the control group p = .01, ´η2 = .11. The modeling condition scored significantly
at the predischarge assessment (Table 1). lower (better function) at 6 weeks compared to the control con-
dition (Figure 5). The condition effect for IKDC subjective
scores approached significance, F(2, 55) = 3.01, p = .08, ´η2 =
Walking self-efficacy. Repeated measures ANOVA results
.05. The modeling group reported higher scores on the IKDC (S)
revealed a significant time effect, F(2, 55) = 79.50, p < .01, ´η2 =
scale (i.e., less disability) at 6 weeks compared to the control
.74, and a nonsignificant “trend” toward a Time × Condition in-
group (Figure 6).
teraction for self-efficacy to walk without crutches, F(2, 55) =
2.70, p = .07, ´η2 = .08. As can be seen in Figure 3, the modeling
group reported greater self-efficacy after viewing the video at ROM. The ANCOVA results showed no significant condi-
predischarge than did the control group. Follow-up analyses re- tion effects for ROM at 2 weeks, F(1, 56) = 0.09, p < .76, ´η2 =
vealed significant group differences at predischarge, t(56) = .01, or at 6 weeks, F(1, 56) = 0.85, p = .36, ´η2 = .02.

FIGURE 3 Walking self-efficacy condition effect (raw means and standard error).
Volume 31, Number 1, 2006 Modeling and Rehabilitation 95

FIGURE 4 Exercise self-efficacy condition effect (raw means and standard error).

Testing for Mediation

Walking self-efficacy (discharge) and crutch use were the
only variables that met the criteria for mediation (33). Results
showed that the intervention was related to both discharge walk-
ing self-efficacy (path coefficient, .35) and crutch use (.43).
Walking self-efficacy was also related to crutch use (.31). How-
ever, the indirect effect of the intervention to crutch use through
walking self-efficacy (.11) was less than the direct effect of the
intervention to crutch use. Hence, no support for mediation was
found.

FIGURE 5 International Knee Documentation Committee (IKDC) DISCUSSION

objective condition effect (adjusted means and standard error).
The present study sought to examine the effectiveness of a
coping modeling video intervention to reduce preoperative anx-
iety and perceptions of pain (expected and actual) as well as to
increase self-efficacy to rehabilitate after ACLR. We also exam-
ined the effectiveness of the modeling video to facilitate im-
provements in functional outcomes post-ACLR. Overall, results
provide support for these propositions. Beyond these general
observations, a number of issues related to specific results need
to be highlighted.
First, the results did not support a condition (modeling) ef-
fect on preoperative anxiety. This is surprising given the existing
body of knowledge that has found modeling to be effective in
this area (15,16). Possible explanations for this lack of effect are
that the video did not highlight any preoperative procedural as-
pects, whereas previous research (34,35) for reducing anxiety
has focused specifically on the procedural and coping aspects of
FIGURE 6 International Knee Documentation Committee (IKDC) distressful hospital procedures. The proximity of the state anxi-
subjective condition effect (p = .08; adjusted means and standard error). ety assessment may also account for the lack of effect. Partici-
pants completed the STAI the day before their operation, rather
than in a more proximal time frame, which may have elicited a
Crutch use. Time spent walking on crutches was assessed different response.
at one point only, so a one-way ANOVA was performed. Signifi- Second, a significant condition effect was found for percep-
cant group differences, F(2, 56) = 19.65, p < .01, d = .94, were tion of expected pain but not for actual pain. With respect to ex-
found, suggesting that the modeling group spent significantly pected pain, results support an immediate effect for participants
less time on crutches compared to the control group (Table 1). who watched the modeling video (Figure 2) and provide some
96 Maddison et al. Annals of Behavioral Medicine

insight into the utility of modeling for altering perceptions of ex- frame examined. In general, 6 weeks is a modest period in which
pected pain preoperatively. Although a number of studies have to assess ACLR outcomes (28).
supported the effectiveness of modeling in reducing discomfort Fifth, unfortunately, our path analysis failed to show that
and responses to stressful medical stimuli (12), few have shown the psychological factors mediated relations between the mod-
support for modeling in reducing pain (34) . This is surprising eling intervention and functional outcome variables. Other psy-
because perceptions of pain have been highlighted as one of the chological processes that were not assessed in this study (e.g.,
most stressful medical aspects for hospital patients (36). Al- rehabilitation motivation) may have proven to be more powerful
though the present results offer support that vicarious informa- mediators. The mechanisms for why modeling works warrants
tion provided through a modeling video can help to ameliorate future investigation.
perception of expected pain, other research is warranted to ex- Sixth, as with all empirical research, the present study
plore whether the nature of pain (i.e., intensity and frequency) is not without limitations. Maddison developed and imple-
can be altered through modeling. For example, nonpharma- mented the intervention, which may have influenced the re-
cological pain management techniques used during pain focus- sults through experimenter bias (expectancy effect). Although
ing (association and disassociation) and pain reduction (relax- our design cannot rule this out, it is highly unlikely that the
ation training and meditation) (37,38) might be presented using differences found between groups were not a result of the
a modeling format. modeling intervention. The same investigator met with the
Third, the modeling video was effective in increasing early both the intervention and control participants to assess the re-
rehabilitation self-efficacy (i.e., crutches and predischarge spective variables as well as to show the video. Apart from
walking and exercise self-efficacy). Despite viewing a video at viewing the video, these sessions did not differ, which sug-
different times, no differences in later self-efficacy (2- and gests that the modeling video was the overriding factor. More-
6-week walking self-efficacy) were found. Taken together, these over, a person is less likely to rehabilitate quicker because his
findings suggest that although the vicarious experience of the interventionist expects him to. This logic is not as robust as the
modeling video was valuable in providing early sources of effi- traditional “Pygmalion in the classroom phenomenon.” Fur-
cacy, the enactive mastery experience gained over time was a thermore, there is evidence from recent psychologically based
more powerful source of efficacious beliefs, thus diminishing intervention studies (20) that the inclusion of a placebo group
the effect of the modeling video. These findings are consistent to control for nonspecific treatment factors, such as attention,
with Bandura’s (9) suggestion that enactive mastery experience caring, and support, does not produce the same positive effects
is the most powerful source of self-efficacy. Of importance, in- on psychological processes and functional outcome as the in-
spection of the descriptive data suggests that a ceiling effect was tervention group’s receipt of imagery and relaxation training.
present for the respective measures of self-efficacy. Thus, once a In short, we believe the supporting information is sufficient to
person’s confidence to perform a given task is high, whether by argue that the intervention was the defining factor in the pres-
viewing the modeling video or from past experience, further in- ent study. In addition, despite considerable attempts to recruit
creases in efficacy are difficult to achieve. a sufficient sample, and within the constraints of this project,
Fourth, with respect to functional outcomes, the modeling our study was not sufficiently powered to detect small effects,
group reported significantly less time using crutches and better like those observed for walking and exercise self-efficacy at 2
scores on the IKDC assessments. These improvements in out- and 6 weeks postsurgery, or for IKDC subjective functional
comes support the use of a modeling video in the first 6 weeks outcome. Finally, the present findings represent data from a
after ACLR. The obvious question is whether these differences group of patients with ACL injury and may therefore not be
in function would persist across time. It is plausible that the reproducible in patients with differing types of orthopedic
early differences in functional outcomes found at 6 weeks injuries.
might provide an early stimulus to improved strength and Seventh, the role of modeling in the athletic rehabilitation
functional outcomes later in the rehabilitation process. This setting is a fertile area for future research, with opportunities
possibility should be examined in future studies. We urge available to examine self-modeling techniques. For example,
some caution when interpreting the IKDC objective results. the use of self-modeling during a specific rehabilitation exercise
The IKDC objective assessment is designed as a rating scale (knee extension) might be associated with improved functional
(A–D) and is not a scoring system for knee function. However, outcomes (i.e., knee strength) post-ACLR. Another area that has
in our study a nominal scale was created to represent varying not been examined is the use of modeling on behavior such as
degrees of function. adherence to rehabilitation programs. It is plausible that compli-
Results did not support a condition effect for ROM. An a ance to rehabilitation could be improved by altering psychologi-
priori proposition was that increased confidence to perform re- cal variables previously shown to affect behavior (i.e., intention
habilitation exercises and to walk with and without crutches and perceived behavioral control).
would be reflected in ROM differences. It is possible that the Results from this study suggest that there may be temporal
current sample size was not large enough to reflect small-group limitations to the effectiveness of modeling interventions for
differences in ROM. In addition, it is possible that noticeable ACL rehabilitation. Future research might look to strengthen
differences in ROM may not present themselves in the time this intervention modality to increase its impact. For example, as
Volume 31, Number 1, 2006 Modeling and Rehabilitation 97

the individual progresses through rehabilitation, separate mod- (16) Robinson RJ, Kobayashi K: Development and evaluation of a
eling videos might be employed that detail specific aspects of presurgical preparation program. Journal of Pediatric Psychol-
that process (e.g., model is shown performing rehabilitation ex- ogy. 1991, 16:193–212.
ercises with the physiotherapist while verbalizing concerns, (17) DeCarlo MS, Sell DE, Shelbourne KD, Klootwyk TE: Current
thoughts, etc.). Alternatively, more interactive media formats concepts on accelerated ACL rehabilitation. Journal of Sport
Rehabilitation. 1994, 3:304–318.
might be used, in which the observer chooses from a selection of
(18) Derscheid GL, Feiring DC: A statistical analysis to characterize
models with whom they best associate, thereby maximizing the
treatment adherence of the 18 most common diagnoses seen at
model similarity relationship. sport medicine clinics. Journal of Orthopaedic and Sports
Physical Therapy. 1987, 9:40–46.
(19) Roos H, Ornell M, Gardsell P, Lohmander LS, Lindstrand A:
REFERENCES
Soccer after anterior cruciate ligament injury—An incompati-
(1) Brewer BW: Psychology of sport injury rehabilitation. In ble combination? A national survey of incidence and risk fac-
Singer RN, Hausenblas HA, Janelle CM (eds), Handbook of tors and a 7-year follow-up of 310 players. Acta Orthopaedica
Sport Psychology. New York: Wiley, 2001, 787–809. Scandinavia. 1995, 66:107–112.
(2) Williams JM, Rotella RJ, Scherzer CB: Injury risk and rehabili- (20) Cupal DD, Brewer BW: Effects of relaxation and guided imag-
tation: Psychological considerations. In Williams JM (ed), Ap- ery on knee strength, reinjury anxiety, and pain following ante-
plied Sport Psychology: Personal Growth to Peak Performance. rior cruciate ligament reconstruction. Rehabilitation Psychol-
Mountainview, CA: Mayfield, 2001, 456–479. ogy. 2001, 46:28–43.
(3) Ievleva L, Orlick T: Mental links to enhanced recovery: An ex- (21) Cohen J: Statistical Power Analysis for the Behavioral Sci-
ploratory analysis. The Sport Psychologist. 1991, 5:25–40.
ences. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc., 1988.
(4) Sordoni C, Hall C, Forwell L: The use of imagery by athletes
(22) Jensen MP, Karoly P, Braver S: The measurement of clinical pain
during rehabilitation. Journal of Sports Rehabilitation. 2000,
intensity: A comparison of six methods. Pain. 1986, 27:117–126.
9:329–338.
(23) Spielberger CD, Gorsuch RL, Lushene RF: The State–Trait
(5) Sordoni C, Hall C, Forwell L: The use of imagery in athletic in-
Anxiety Inventory (STAI). Palo Alto, CA: Consulting Psycholo-
jury rehabilitation and its relationship to self-efficacy. Physio-
gists Press, 1970.
therapy Canada. 2002, Summer:177–185.
(24) Smith RE, Smoll FL, Wiechman SA: Measurement of trait anx-
(6) Theodorakis Y, Beneca A, Malliou P, Goudas M: Examining
iety in sport. In Duda JL (ed), Advances in Sport and Exercise
psychological factors during injury rehabilitation. Journal of
Psychology Measurement. Morgantown, WV: Fitness Informa-
Sport Rehabilitation. 1997, 6:355–363.
tion Technology, 1998, 105–128.
(7) Levitt R, Deisinger JA, Remondet Wall J, Ford L, Cassisi JE:
(25) McAuley E, Mihalko SL: Measuring exercise related
EMG feedback-assisted postoperative rehabilitation of minor
self-efficacy. In Duda JL (ed), Advances in Sport and Exercise
arthroscopic knee surgeries. Journal of Sports Medicine and
Psychology Measurement. Morgantown, WV: Fitness Informa-
Physical Fitness. 1995, 35:218–223.
(8) Ross MJ, Berger RS: Effects of stress inoculation training on tion Technology, 1998, 371–389.
athletes’ postsurgical pain and rehabilitation after orthopedic (26) Anderson AF: Rating Scales: Knee Surgery. Baltimore: Wil-
injury. Journal of Consulting and Clinical Psychology. 1996, liams & Wilkins, 1994.
64:406–410. (27) Hefti F, Muller W, Jakob RP, Staubli HU: Evaluation of knee
(9) Bandura A: Self-Efficacy: The Exercise of Control. New York: ligament injuries with the IKDC form. Knee Surgery, Sports
Freeman, 1997. Traumatology, Arthoscopy. 1993, 1:226–234.
(10) McCullagh P, Weiss MR: Modeling: Considerations for motor (28) Irrgang JJ, Ho H, Harner CD, Fu FH: Use of the International
skill performance and psychological responses. In Singer RN, Knee Documentation guidelines to assess outcome following
Hausenblas HA, Janelle CM (eds), Handbook of Sport Psychol- anterior cruciate ligament reconstruction. Knee Surgery, Sports
ogy. New York: Wiley, 2001, 205–238. Traumatology, Arthoscopy. 1998, 6:107–114.
(11) McCullagh P, Weiss MR: Observational learning: The forgotten (29) Gerhardt JL, Cocchiarella, Lea R: The Practical Guide to
psychological method in sport psychology. In Van Raalte JL, Range of Motion Assessment. Chicago: American Medical As-
Brewer BW (eds), Exploring Sport and Exercise Psychology. sociation, 2002.
Washington, DC: American Psychological Association, 2002, (30) McCullagh P, Ram N: A comparison of imagery and modeling
131–149. [Abstract]. Journal of Sport and Exercise Psychology. 2000,
(12) O’Halloran CM, Altmaier EM: The efficacy of preparation for Suppl. 1:S9.
surgery and invasive medical procedures. Patient Education (31) Tabachnick BG, Fidell LS: Using Multivariate Statistics (4th
and Counseling. 1995, 25:9–16. Ed.). New York: Allyn & Bacon, 2001.
(13) Flint FA: The Psychological Effects of Modeling in Athletic In- (32) Hair JF, Anderson RE, Tatham RL, Black WC: Multivariate
jury Rehabilitation. Eugene: University of Oregon, 1991. Data Analysis With Readings. New York: Macmillan, 1992.
(14) Flint FA: Modeling in injury rehabilitation. In Pargman D (ed), (33) Baron RM, Kenny DA: The moderator–mediator variable dis-
Psychological Bases of Sport Injuries. Morgantown, WV: Fit- tinction in social psychological research: Conceptual, strategic,
ness Information Technology, 1999, 221–234. and statistical considerations. Journal of Personality and Social
(15) Melamed BG, Siegel LJ: Reduction of anxiety in children Psychology. 1986, 251:1173–1182.
facing hospitalization and surgery by use of filmed model- (34) Padilla GV, Grant MM, Rains BL, et al.: Distress reduction and
ing. Journal of Consulting and Clinical Psychology. 1975, the effects or preparatory teaching films and patient control. Re-
43:511–521. search in Nursing and Health. 1981, 4:375–387.
98 Maddison et al. Annals of Behavioral Medicine

(35) Shipley RH, Butt JH, Horwitz B, Farbry JE: Preparation for a (37) Heil J: Mental training in injury management. In Heil J (ed),
stressful medical procedure: Effect of amount of stimulus Psychology of Sport Injury. Champaign, IL: Human Kinetics,
pre-exposure and coping style. Journal of Consulting and Clini- 1993, 151–174.
cal Psychology. 1978, 46:499–507. (38) Taylor J, Taylor, S: Pain education and management in the reha-
(36) Van der Ploeg HM: Stressful medical events: A survey of pa- bilitation from sports injury. The Sport Psychologist. 1998,
tient’s perceptions. In Maes S, Spielberger CD, Defares PB, 12:68–88.
Sarason IG (eds), Topics in Health Psychology. New York:
Wiley, 1988, 193–203.