Across all disciplines of medicine, patients’ perspectives
are being formally included in the process of selecting among treatment
options and in assessing results of care1-4.
Patients’ perspectives can be determined by measuring patient-reported
functional status, the importance of symptoms to patients, and their
concerns about treatments5-9.
In addition, patients’ perspectives are ascertained by
measuring their expectations of treatments. Patients’ expectations
are particularly important because they are linked to requests for
elective and possibly costly treatments and are strongly related to
patients’ assessments of outcome2,3,10-12.
However, while experts from diverse medical disciplines advocate
ascertaining patients’ expectations2,3,11-13,
few studies have systematically measured patients’ expectations
of orthopaedic procedures1,10,14-16.
In addition, most instruments that are currently available to measure
expectations of orthopaedic procedures are primarily physician-derived—that is,
based on clinical knowledge and expertise13,15,16.
Instruments that are patient-derived—that is, based
on patients’ perspectives—capture a broader array
of expectations by including those that may not be attainable or
realistic but are still important to patients. In addition, we are
not aware of any currently available standardized instruments that measure
expectations for many of the most commonly performed operations.
The goals of this study were to measure patients’ expectations
of knee surgery and to develop and test the reliability of patient-derived
knee surgery expectations surveys in a large sample of patients undergoing
various types of operations on the knee.
Phase 1: Developing Draft Surveys
Obtaining Baseline Information and Ascertaining Patients’ Expectations
Patients undergoing knee surgery by participating orthopaedists
were eligible if they were at least eighteen years old and were
fluent in English. Patients were excluded if they had cognitive
deficits or were undergoing revision total knee arthroplasty. A
total of 377 patients who were scheduled to undergo surgery by one
of twelve orthopaedists were enrolled during 1998 and 1999 (see
Appendix). Most patients were scheduled to undergo total knee arthroplasty
(161; 43%), repair of the anterior or posterior cruciate
ligament (grouped together for these analyses) (seventy-five; 20%),
or meniscal surgery (eighty-five; 23%). An additional fifty-six patients
(15%) were scheduled to have an operation for another condition,
including multiligament disorders (one patient), injury of the medial
or lateral collateral ligament (two), complete knee dislocation
(two), patellofemoral chondromalacia (nine), and osteoarthritis
and débridement procedures (forty-two). These 377 patients
were interviewed several days before the surgery and asked the open-ended
question: "What are your expectations of the surgery you
are going to have on your knee?" The patients were also
asked: "How important is each expectation?" with
possible response options ranging from very important to not important. Open-ended
responses were analyzed with standard qualitative techniques to
generate categories of expectations17-20.
Categories were named to indicate the major themes that they represented.
All open-ended responses from each patient were then reviewed again
and were coded according to category. Frequencies of categories
were calculated, and a series of multivariate logistic regression
equations were set up with categories as dependent variables and with
demographic characteristics and functional status scores as independent
variables. Demographic information, the ICD-9 (International
Classification of Diseases, Ninth Revision) diagnosis, and the patient-reported
functional status as measured by the American Academy of Orthopaedic Surgeons
Lower Limb Core* (AAOS-LLC) and the Medical Outcomes Study
Short-Form General Health Survey (SF-36) were obtained
from computerized databases maintained at the Hospital for Special
Surgery Outcomes Unit21-23.
Assembly of Draft Surveys
Two draft surveys, one for patients undergoing total knee arthroplasty
and one for patients undergoing other operations on the knee, were
developed. Items included in the draft surveys were selected on the
basis of how frequently they were cited and their clinical relevance
as determined by a panel of orthopaedists. Specifically, items were
selected if they were cited by 5% of the patients, if they
represented important functional changes resulting from surgery,
or if they represented potentially inappropriate or unrealistic
expectations. Items were phrased as questions with terminology typically
used by patients when talking with orthopaedists.
Phase 2: Testing Draft Surveys for Reliability
Test-Retest Reliability
The two draft surveys were then tested to establish test-retest
reliability in a new sample of 163 patients who were to undergo
surgery by one of fourteen orthopaedists. Patients were identified
for this testing phase in the same manner and fulfilled the same
eligibility criteria as described above for patients in the survey-development
sample.
Reliability was measured by giving draft surveys to these new
patients on two separate occasions before the surgery and comparing
their responses. Specifically, the patients were contacted three
to twelve days before the surgery, given the survey during a telephone
interview, and then given the same survey several days later, either
by telephone or in person in the hospital at the time of the surgery.
Test-retest reliability was analyzed with the kappa statistic24,25. Kappa is a measure of concordance
or agreement greater than that due to chance24,25.
Kappa = (observed proportion of agreements - chance-expected
proportion of agreements)/(1 - chance-expected
proportion of agreements).
It is also possible to calculate a weighted kappa value, which
accounts for how far apart responses are, in a frequency table for
example. In our study, agreement about the importance of expectations (very
important, important, or a little important) was measured with the
weighted kappa statistic. Kappa and weighted kappa values can range
from -1 (perfect disagreement) to +1 (perfect agreement), and
a value of 0 corresponds to agreement that is no better than chance.
By convention, values of £0.40 are considered to indicate
slight to fair agreement; 0.41 to 0.60, moderate agreement; and >0.60,
substantial agreement25.
Phase 3: Generating Final Surveys
The selection of items for the final surveys was determined by
kappa values and potential clinical relevance. Specifically, an
item was retained in the survey if its kappa or weighted kappa value
was 0.4. However, items with lower kappa values were assessed individually
for possible clinical relevance and prior performance. For example,
for items present in both surveys, if the kappa value met the threshold
of 0.4 in either survey then that item was retained in both surveys.
In addition, a panel of orthopaedists reviewed all items and kappa
values to ensure that items with particular clinical relevance (for
example, those that possibly represented inappropriate expectations)
were retained. Items retained from these analyses formed the questions in
the final surveys.
Statistical analyses were done with use of the Statistical Analysis
System (SAS, Cary, North Carolina) and included means and standard
deviations for all continuous variables, frequencies for all ordinal
and nominal variables, comparisons of frequencies with the chi-square
test, and comparisons of means with the t test26.
In addition, Pearson product-moment correlation coefficients were
calculated between pairs of continuous variables, and Spearman rank-order
coefficients were calculated between pairs in which at least one
variable was ordinal and the other was ordinal or continuous. PROC
LOGIST was used for multivariate logistic regression models26.
This study was approved by the Institutional Review Board at
the Hospital for Special Surgery. Informed consent for participation
was obtained from patients.
Phase 1: Developing Draft Surveys
Patient Characteristics
In addition to patient characteristics (see Appendix), mean Lower
Limb Core (AAOS-LLC) scores and mean SF-36 Physical
Component Summary (PCS) scores were measured (Fig. 1). Patients undergoing
total knee arthroplasty scored worse on both scales than did patients
in the other groups (p = 0.0001). Patients in all groups,
especially those undergoing total knee arthroplasty, had worse PCS scores
compared with the mean score of 50 in the general United States
population (p < 0.001). This finding was anticipated as
the PCS is heavily weighted by lower-extremity function.
Expectations
In total, 1161 expectations were cited, with a mean of 3 1 per
patient, and they were grouped into fifty-two categories. The most
frequently cited categories in each diagnostic group were determined (see
Appendix). Return to sports was a major category for all conditions,
with patients mentioning forty-five different sports, the most frequently being
tennis, skiing, golfing, jogging, and swimming. Patients undergoing
surgery on the anterior or posterior cruciate ligament also listed
high-demand sports, such as racquetball (4%), marathon running
(3%), rugby (4%), basketball (9%), squash (3%),
roller-blading (3%), volleyball (5%), and soccer
(5%). Also, 7% of patients undergoing surgery
on the anterior or posterior cruciate ligament expected to be able
to return to professional sports. Nearly one-third of patients undergoing
surgery on the anterior or posterior cruciate ligament expected that,
as a result of the surgery, the knee would "be back to
the way it was" before symptoms started. Many of these
patients had sustained a specific injury and expected the surgery
to reverse the damage caused by the injury.
Logistic regression analysis was used to measure relationships
between expectations and patient characteristics. Age, gender, education,
the SF-36 PCS score, and the AAOS-LLC score were
considered to be independent variables. In the entire sample, more
women than men expected improvement in walking ability (57% compared
with 29%, p = 0.001) and more men than women expected improvement
in sports performance (51% compared with 40%,
p = 0.03). Younger patients were more likely to expect
improvement in sports performance and for the knee to "be
back to the way it was" before symptoms started (p = 0.0001), whereas
older patients were more likely to expect pain relief (p = 0.04)
and improved walking ability (p = 0.0001). Patients with
less education were more likely to expect psychological improvement (p = 0.01)
and pain relief (p = 0.003), and patients with more education
were more likely to expect improvement in sports performance (p = 0.0001). Patients
with a worse PCS score were more likely to expect a psychological
benefit, improvement in walking ability, and to be able to return
to work (p £ 0.03). Patients with better PCS and AAOS-LLC scores
were more likely to expect improvement in sports performance, whereas
patients with a worse AAOS-LLC score were more likely to
expect improvement in walking ability (p = 0.0001).
Assembly of Draft Surveys
There were significant differences with regard to frequencies
of expectations in eleven of the twenty-one categories between patients
undergoing total knee arthroplasty and those undergoing other operations
on the knee (p £ 0.02) (see Appendix). Therefore, two draft
expectations surveys were developed, one for patients undergoing
total knee arthroplasty and one for patients undergoing other general
knee procedures. For each survey, items were selected on the basis
of their clinical relevance and how frequently they had been cited
in this study.
In addition, because there was variation in expectations related
to pain relief and improvement in walking ability, we modified these
items to assess the magnitude of improvement expected—for example,
how far patients expected to be able to walk. A distinction was
also made between recreational and professional sports in the survey
for patients undergoing general knee surgery.
The response format for each item asks patients whether they
have that expectation and, if so, how important it is. If they do
not have the expectation, patients indicate either that the item
does not apply to them or that the item does apply but is not something
that they expect.
Phase 2: Testing Draft Surveys for Reliability
Patient Characteristics
The demographic characteristics, functional status, and orthopaedic
characteristics of the 163 patients enrolled in the reliability-testing
phase were similar to those of the patients in the survey-development phase
(see Appendix). Patients undergoing total knee arthroplasty were
older than those in the other groups and were also more likely to
be women, to be retired, and to have a worse functional status (p < 0.003).
Test-Retest Reliability
The first interview was conducted at a mean of 6.0 2.0 days
(range, three to twelve days) before the surgery and the second
interview, at a mean of 5.1 1.7 days (range, three to eleven days)
after the first interview. All first interviews were done by telephone;
37% of the second interviews were done by telephone, and
the remaining were done in the hospital on or close to the day of
the surgery. Surveys took less than five minutes to complete. All
patients agreed to participate in the second interview, indicating
to us that these surveys are acceptable to patients. The first and
second interviews were done by the same interviewer for 79% of
the patients, and they were done by different interviewers for 21%.
Phase 3: Generating Final Surveys
Knee Surgery Expectations Survey
Kappa values ranged from 0.4 to 0.8 for almost all items from
the draft general knee surgery expectations survey (see Appendix).
Only one item, expecting to have improved knee strength, had kappa
and weighted kappa values of <0.4. This item was not considered
to have any clinical relevance and therefore was discarded. In addition
to the pain and walking items listed, patients were also asked how
much pain relief they expected (nearly 55% expected complete
pain relief, kappa = 0.5) and how far they expected to
be able to walk (nearly 85% expected to be able to walk >1
mi [>1.6 km], kappa = 0.5).
From these analyses we generated the final version of the Hospital
for Special Surgery Knee Surgery Expectations Survey, composed of
twenty questions and requiring less than five minutes to complete
(Fig. 2).
Knee Replacement Expectations Survey
Kappa or weighted kappa values ranged from 0.4 to 0.8 for most
items of the draft total knee arthroplasty expectations survey (see
Appendix). Expecting improvement in walking was cited by all patients
during both interviews and therefore the kappa value was numerically
uninterpretable. Because of the prevalence of this item, it was retained
in the final version. Expecting the knee to be back to the way it
was before symptoms started had kappa and weighted kappa values
of 0.3 and was eliminated. It is possible that this item did not perform
well because patients undergoing total knee arthroplasty typically
have symptoms for many years before the surgery. Expecting improvement
in daily activities had a low kappa value of 0.1 and a weighted
kappa of 0.3. We decided to retain this item because it has clinical
relevance and it performed well in the general knee surgery expectations
survey. In addition to the pain and walking items listed, nearly
52% of patients undergoing total knee arthroplasty expected
complete pain relief (kappa = 0.5) and nearly 65% expected
to be able to walk >1 mi (kappa = 0.7). From these
analyses, we generated the final version of the Hospital for Special
Surgery Knee Replacement Expectations Survey, composed of seventeen
questions and also requiring less than five minutes to complete (Fig.
3).
The results of this study demonstrate that patients have multiple
expectations of knee surgery and that these expectations vary by
diagnosis, patient characteristics, and functional status. Using
a large sample of patients, we developed and tested two knee surgery
expectations surveys—a seventeen-item survey for
patients undergoing total knee arthroplasty and a twenty-item
survey for patients undergoing other types of knee surgery. The
items in the surveys are presented in simple, brief terms and address
symptom-related, functional, and psychosocial expectations. The
surveys took less than five minutes to complete and were well accepted
by the patients.
These surveys have several possible uses in daily practice, especially
if patients complete them at the time of their evaluation by the
orthopaedist. First, they provide a simple way to obtain a more
comprehensive evaluation that would otherwise require a lengthy
interview. Second, they provide a way for patients to specifically
state what they anticipate from surgery; this is particularly useful
for patients who would otherwise express their goals in vague, nonspecific
terms. Third, they provide the orthopaedist with a written template
to guide a formal discussion about what are realistic and unrealistic goals
for each patient. In addition, the orthopaedist can add comments
to the survey to reflect specific discussion and recommendations.
Fourth, if the survey is included in the patient’s chart,
it can be referred to at the time of long-term postoperative follow-up
as a way for patients and orthopaedists to jointly assess fulfillment
of expectations and for patients to remember that these items were addressed
preoperatively. This may be particularly helpful for patients who
are dissatisfied with the outcome.
A major strength of these instruments is that they were patient-derived—that
is, all items were determined by patients, not set a priori by
physicians. This is important not only because of the intrinsic
face validity of patient-derived instruments but also because
patient-derived instruments can include a broad spectrum
of items that might not be readily apparent to physicians, such
as expecting psychological improvement. In addition, physician-derived
instruments usually do not include expectations that are probably
not attainable but are still held by patients. For example, in our
study many patients expected the knee to recover to its preinjury
state and many patients who were to undergo total knee arthroplasty
expected complete pain relief.
Another strength of these instruments is that a panel of orthopaedists
reviewed the surveys to ensure content validity and to phrase questions
in words typically used by patients. The orthopaedic panel also
separated certain knee-function expectations into distinct survey
items, which had initially been considered together. Squatting after
total knee arthroplasty is an example of a knee function that was
selected to be a distinct item because a patient’s response
that it was expected would generate specific discussions with the
patient regarding its likelihood postoperatively.
Valid and reliable questionnaires in a written format, such as
the surveys developed in this study, facilitate the important task
of obtaining and recording patients’ expectations3. Measuring patients’ expectations
is necessary for various reasons. For example, knowing these expectations
helps physicians to provide more focused clinical care, highlights
areas for patient education, and promotes shared decision-making when
several treatment options are available2,3.
Involving patients in their care by discussing expectations has
also been shown to increase patients’ adherence to recommendations1,27,28. In addition, fulfillment of
expectations is associated with patients’ assessment of
outcome and satisfaction, two measures that may justify elective orthopaedic
procedures even if they are costly1-3.
Within orthopaedics, several types of expectations have been
studied. For example, studies of total hip arthroplasty have assessed
the relationship between fulfillment of expectations and satisfaction
with the outcome10,15,16,29. Other
investigators have considered the influence of orthopaedists’ expectations
on patients with a hip fracture, the "ambitiousness" of
the expectations of patients undergoing surgery for the treatment
of lumbar spinal stenosis, and the level of concern about not knowing
what to expect among patients undergoing total hip or total knee
arthroplasty1,9,28.
There are several limitations to this study. First, the participants
were all patients in a tertiary-care orthopaedic institution and
therefore may not be representative of other patient populations.
Second, some patients were interviewed in the hospital on the day
of the surgery, when their responses may have been affected by apprehension
and anxiety. Third, although the surveys were intended to be self-administered,
they were tested during telephone and in-person interviews.
This was done to maximize response rates during the test-retest phase.
In conclusion, two valid and reliable instruments have been created
for patients undergoing knee surgery: the seventeen-item Hospital
for Special Surgery Knee Replacement Expectations Survey and the
twenty-item Hospital for Special Surgery Knee Surgery Expectations
Survey. These instruments have several possible uses in both clinical
practice and research, and they should enhance shared decision-making
and our ability to assess outcomes and patient satisfaction following
these procedures.
Note: The authors thank the orthopaedic surgeons in the Adult
Arthroplasty Service and the Sports Medicine Service at the Hospital
for Special Surgery for their participation.
Tables showing demographic characteristics of the patients in
the survey-development sample, the most frequently cited expectations
by diagnosis, demographic and orthopaedic characteristics of the patients
in the survey-testing sample, and the test-retest reliability of
the draft knee surgery expectations survey and the draft total knee
arthoplasty expectations survey are available with the electronic
versions of this article, on our web site (www.jbjs.org) and on
our CD-ROM (call 781-449-9780, ext. 140, to order).