Work in multiple disciplines of medicine has suggested that greater
expectations regarding the results of treatment correlate with better
outcomes1-4.
Recent effort has been devoted to characterizing and quantifying preoperative
expectations in orthopaedic patient populations. Studies of patients
undergoing hip, knee, and shoulder surgery have shown significant
relationships between expectations and variables such as age, sex, diagnosis,
education, and functional
status5-8.
Positive associations between expectations and satisfaction after total hip
arthroplasty9,10
and spine
surgery11,12
have been reported. Mahomed et al. also found that greater expectations were
an independent predictor of better physical function and relief of pain after
total knee and hip
arthroplasty10.
Similarly, Iversen et al. found that patients who had greater preoperative
expectations reported greater postoperative improvements in function after
spine surgery12.
However, we are not aware of studies in which patients' expectations before
shoulder surgery were related to outcomes.
Rotator cuff tears are a common cause of disability related to the shoulder
and upper extremity. Surgical repair of a chronic tear is indicated when
nonoperative treatment fails. Rotator cuff repair is an elective procedure
that requires careful preoperative evaluation and discussion of treatment
expectations, risks, and benefits. Overall, repairs of rotator cuff tears have
been shown to lead to good-to-excellent outcomes in most patients, with
significant improvement in the mean scores on self-assessment
questionnaires4,13-23.
However, multiple factors, including
age13,15,16,21,
gender13,16,21,
smoking14, larger
tear
size16,18,20,
poor tendon
quality20, Workers'
Compensation
status15,22-24,
and healing of the rotator cuff
repair25-27,
have been shown to be associated with less favorable outcomes after rotator
cuff repair.
The purpose of this study was to evaluate the effect of patients'
preoperative expectations regarding rotator cuff repair on their actual
self-assessed outcome. We hypothesized that preoperative expectations are
predictive of the outcome of rotator cuff repair.
Our study included 125 patients in whom a chronic unilateral tear of the
rotator cuff was treated with a primary repair by one surgeon (A.G.) between
January 1998 and September 2001. Patients were symptomatic for at least three
months, which defined a chronic rotator cuff
tear8,28.
No shoulder had been treated surgically prior to the rotator cuff repair. The
indication for surgery was failure of nonoperative treatment, which included a
physical therapy program and, in some cases, a corticosteroid injection.
The study was approved by the hospital institutional review board. All of
the prospective data included in this study were obtained as part of the
routine care of the patients and were part of their medical record. The data
were then reviewed retrospectively in a de-identified fashion for the purposes
of this study. Consequently, the institutional review board did not require
that the patients provide informed consent.
The average age (and standard deviation) at the time of surgery was 56.2
± 11.4 years (range, thirty-two to eighty-four years). Seventy-two
(58%) of the patients were male. Eighty-two (66%) underwent surgery on the
dominant shoulder, and 117 (94%) were right-hand dominant. Thirty-nine (31%)
had made a Workers' Compensation claim, and twenty-six (21%) were smokers. The
mean duration of symptoms was 16.0 ± 25.9 months (range, three to 210
months). Thirty-seven (30%) of the tears were small (<1 cm), forty-two
(34%) were medium (=1 cm but <3 cm), thirty-two (26%) were large (=3
cm but <5 cm), and fourteen (11%) were massive (=5
cm)22.
Prospective preoperative evaluation included a detailed medical history, a
physical examination, and completion of a series of questionnaires as
previously
described8. The
patients completed four limb-specific instruments: the Simple Shoulder Test
(SST), the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire,
and visual analogue scales for shoulder pain and shoulder function. The visual
analogue scale for pain was anchored with "none" on one end of a
10-cm line and "disabling" on the other, whereas the visual
analogue scale for function was anchored with "comfortable" and
"can't use it." The patient placed a mark on the 10-cm line for
each visual analogue scale, and the mark was measured to the nearest
millimeter to calculate the score. General health was assessed with use of the
Short Form-36 (SF-36) and a visual analogue scale for overall quality of life
anchored with "little or no problem" and "very bad."
For ease of comparison, the scores for the SST, DASH, and visual analogue
scale instruments were converted to a percentage of a perfect score, with 0%
corresponding to the worst state of health and 100% corresponding to the ideal
state of health. The raw scores for the eight subscales of the SF-36 (Physical
Function, Role Physical, Bodily Pain, General Health, Vitality, Social
Function, Role Emotion, and Mental Health) were adjusted to a percentage of
reported normative values for age and gender-matched
controls29.
Preoperative expectations regarding the results of treatment were assessed
with use of six questions from the Musculoskeletal Outcomes Data Evaluation
and Management System (MODEMS) questionnaire. The average response to the six
questions was calculated for each patient to generate a mean expectations
score. A score of 1 corresponds to the lowest level of expectations (not at
all likely), whereas a score of 5 corresponds to the highest level of
expectations (extremely likely). Answers marked "not applicable"
were not scored or included in the calculation of the mean expectations score.
No attempts were made to influence patients' expectations aside from the usual
preoperative counseling practices of the senior author (A.G.). This routine
counseling included a discussion of the pathological condition, natural
history, prognosis, treatment options, surgical risks and benefits, and
rehabilitation.
Follow-up evaluation at a mean of 54.1 ± 7.6 weeks (range, 32.7 to
88.7 weeks) consisted of a physical examination and completion of the same
series of outcome instruments. The improvement compared with the baseline
score on each outcome instrument was calculated for each patient by
subtracting the baseline score from the follow-up score.
Surgical Technique
Three different repair techniques were used. Twenty-six (21%) of the
patients had open repair, sixty-two (50%) had miniopen repair, and
thirty-seven (30%) had arthroscopic repair. Larger tears were repaired with
open techniques, whereas smaller tears were repaired with mini-open or
arthroscopic techniques. All of the patients had a subacromial decompression
with acromioplasty at the time of the rotator cuff repair. Adjunctive
procedures included thirty-four distal clavicular resections for the treatment
of arthritis of the acromioclavicular joint, five biceps tenodeses, five
biceps tendon relocations, and two biceps tenotomies. All of the repairs were
performed by the senior author, who is a fellowship-trained shoulder surgeon.
The size of the rotator cuff tear was measured intraoperatively prior to
repair.
Postoperative Management and Rehabilitation
The affected upper extremity was placed in a sling after the surgery. Most
of the patients were discharged home on the day of the surgery. No analgesic
catheters were used. Patients were admitted for overnight observation if they
were of advanced age or had substantial medical comorbidities that required
observation. No patients required readmission for pain control. All of the
outpatients were seen in the office by the attending surgeon within one to
four days.
All of the patients were referred for physical therapy. Passive
range-of-motion exercises were initiated on the day after the surgery. During
the first five weeks postoperatively, active use of the upper extremity was
discouraged. For patients who had had a mini-open or open repair, the passive
range-of-motion exercises commenced with pendulum circumduction, supine
forward elevation, supine external rotation, supine horizontal adduction, and
standing internal rotation. For patients who had had an arthroscopic repair,
the passive motion was limited to pendulum circumduction and supine passive
external rotation for the first four weeks. After the fifth week, passive
stretching in all directions was commenced for the patients who had had an
arthroscopic repair. Use of the sling was discontinued after five weeks in all
treatment groups. Patients began isometric deltoid, rotator cuff, and
periscapular strengthening exercises six to eight weeks postoperatively.
Resisted exercises with a Thera-Band (Hygenic, Akron, Ohio) commenced ten to
twelve weeks postoperatively, with patients with larger tears starting later
than those with smaller tears. Formal physical therapy was usually
discontinued after three months, and the patients continued a home exercise
program.
Statistical Methods
Statistical analysis was performed with the aid of a statistician using
Stata software (version 7.0; Stata, College Station, Texas). Two-tailed t
tests were utilized to compare baseline and follow-up data for each outcome
parameter. Bivariate relationships were evaluated with use of Spearman
correlations for continuous variables, and both the correlation coefficient
and the p value are reported for each test. The Spearman correlation
coefficient ranges from —1 to 1 and indicates both the direction and
strength of the linear relationship between the independent variable (mean
expectations) and the dependent variable (outcome parameter). The significance
of the correlation is indicated by the p value, which is the probability that
the correlation is due to chance alone. P values of <0.05 were considered
significant.
Multivariate analyses were performed for each outcome measure with use of
linear regression analysis to control for potential confounding variables.
Independent variables entered initially in each model included mean
expectations, age, gender, smoking, Workers' Compensation status, symptom
duration, number of previous operations, number of comorbidities, size of the
rotator cuff tear, and repair technique. A backward-elimination stepwise
technique was employed to remove noncontributory independent variables. The
significance of each independent variable in the model was evaluated after
calculation, and, at each run, the single least significant variable was
removed and the model was recalculated. This process of removing insignificant
variables was repeated in a stepwise fashion until a model including only
variables with significance was achieved. The regression coefficients and
associated p values for the final models are provided in the Appendix.
Overall, the patients had high expectations regarding the surgical
treatment of the rotator cuff tear, with an average score (and standard
deviation) of 4.44 ± 0.69 (range, 2 to 5)
(Table I). More than 85% of the
respondents to each question expected that the surgery was "very
likely" or "extremely likely" to result in improvement in
the parameter dealt with by that question. The number of respondents to each
question varied, with patients who were retired tending to indicate that
question 4, which related to going back to the patient's usual job, was not
applicable. Nevertheless, mean expectations were strongly associated with each
individual expectations question (p < 0.0001). Since no single question was
a statistical outlier, mean expectations are a reasonable measure of a
patient's overall preoperative expectations.
The average postoperative scores on the SST, DASH, and all three visual
analogue scale scales were each significantly (p < 0.0001) improved
compared with the preoperative scores (Fig.
1). Patients demonstrated similar improvement (p < 0.0001) on
the SF-36 subscales for Role Physical and Bodily Pain
(Fig. 2). Rotator cuff repair
also resulted in average postoperative scores of nearly 100% of the normative
values on the six other SF-36 subscales.
Greater mean preoperative expectations correlated with better postoperative
performance on the SST (p < 0.0001), DASH (p < 0.0001), visual analogue
scale for shoulder pain (p = 0.0051), visual analogue scale for shoulder
function (p = 0.0243), visual analogue scale for quality of life (p = 0.0006),
and all SF-36 subscales (p values ranging from <0.0001 to 0.0271)
(Table II). Multivariate
analysis confirmed that expectations were a significant independent predictor
of better postoperative outcome scores on all instruments
(Table II). This is shown in
the Appendix, along with the complete data from the final multivariate model
for each outcome instrument.
Greater mean preoperative expectations were associated with better
improvement compared with the baseline on the DASH (p = 0.018) and the SF-36
subscales for Physical Function (p = 0.0002), Role Physical (p = 0.002),
Bodily Pain (p < 0.0001), Vitality (p = 0.007), Social Function (p =
0.011), and Role Emotional (p = 0.005)
(Table III). Multivariate
analysis revealed even stronger positive associations between greater
expectations and greater improvement on the SST, the DASH, each visual
analogue scale, and all SF-36 subscales except for General Health
(Table III). This is shown in
the Appendix along with the complete data from the final multivariate model
for improvement from baseline for each outcome parameter.
This study demonstrates that patients' preoperative expectations have a
dramatic positive association with their self-assessed outcome after rotator
cuff repair. These findings are consistent with those of previous studies that
have shown a positive association between greater expectations and treatment
outcomes1-3,10,12;
however, the degree of association was particularly robust in our study.
We are aware of two previous studies in which the relationship between
preoperative expectations and the functional outcome of orthopaedic surgery
was evaluated. Iversen et al. found that greater expectations were associated
with better reported function and satisfaction six months after surgery for
lumbar stenosis; however, greater expectations were also associated with less
pain relief12. In
contrast, we found that greater expectations were associated with both better
function and better pain relief. Mahomed et al. found that greater
expectations were an independent predictor of better SF-36 physical function
scores and relief of pain six months after total knee and hip
arthroplasty10. We
found that greater mean expectations were an independent predictor of better
outcome on both the SF-36 and limb-specific outcome instruments after rotator
cuff repair.
A few studies have dealt with the relationship between preoperative
expectations and satisfaction after orthopaedic surgery. Mancuso et al.
retrospectively assessed the expectations of patients who had undergone total
hip arthroplasty and found a strong correlation between preoperative
expectations and satisfaction two to three years after the
surgery9. Similarly,
Lutz et al. reported that greater expectations were an independent predictor
of improved satisfaction one year after lumbar
discectomy11. We
did not address postoperative satisfaction in this study.
While the results may be specific to the population studied and not
necessarily applicable to all patients undergoing rotator cuff repair, the use
of multivariate analysis provides a statistical method for accounting for
differences in patient populations. The results of our multivariate analysis
demonstrated that greater preoperative expectations are consistently and
independently associated with significantly better performance on multiple
outcome instruments after rotator cuff repair. In contrast, the multivariate
analysis showed that gender, tear size, and repair technique were not
significant predictors of any outcome parameters in this study. While
expectations are certainly not the only determinant of outcome, the
multivariate analysis strongly suggests that, among the factors that we
analyzed, preoperative expectations are the single most important predictor of
both one-year performance and improvement from baseline after rotator cuff
repair. These findings provide substantial evidence that preoperative
expectations are an important factor to consider when evaluating the outcome
of treatment. Variability in expectations may help to explain divergent
results within and among various patient populations.
While the results of this study are compelling, there are several potential
limitations. First, the multivariate analysis that we utilized may not have
included all confounding variables that could influence the results. For
example, anatomic factors such as tendon quality and the integrity of the
repair were not taken into account in our study. In addition, the surgeon's
approach to management of his patients might have biased the expectations,
thus limiting the reproducibility of the study. While there is not a
particular aspect of the surgeon's approach to management that we can identify
as being unusual or unique, subtleties of the surgeon's experience and
approach expressed during preoperative counseling may influence expectations
greatly. Another limitation is that the reproducibility and validity of the
specific visual analogue scales utilized in this study have not been studied,
to our knowledge. Finally, patients were evaluated at only two points in time,
and the effect of preoperative expectations on outcome may change over
time.
Expectation regarding the results of treatment is an important factor that
appears to independently affect the outcome of rotator cuff repair. The
results of this observational study raise the question of whether the approach
to counseling patients prior to rotator cuff repair can affect their
expectations and thus the outcome of surgery. We have not addressed this
issue, but we believe that the preoperative discussion of expectations at
least affects a patient's decision to choose surgical treatment. Future
studies are necessary to address the limitations that we noted and to further
characterize the relationship between preoperative expectations and outcomes
of orthopaedic surgery.
Tables showing the details of the stepwise multiple regression analyses are
available with the electronic versions of this article, on our web site at
jbjs.org (go to the article citation and click on "Supplementary
Material") and on our quarterly CD-ROM (call our subscription
department, at 781-449-9780, to order the CD-ROM). ?