Content uploaded by James Wright
Author content
All content in this area was uploaded by James Wright on Dec 06, 2017
Content may be subject to copyright.
SPINE Volume 25, Number 24, pp 3167–3177
©2000, Lippincott Williams & Wilkins, Inc.
The Characteristics of Patient Satisfaction Measures
Pamela L. Hudak, MSc, BScPT,* and James G. Wright, MD, MPH, FRCSC†‡
The words “patient” and “customer satisfaction” are
widely used in health care contexts. Because this concept
is so much a part of current rhetoric in health care, it is
easy to accept unquestioningly its inclusion as an integral
part of the evaluation of health care quality
11
and to
assume that its measurement and interpretation are evi-
dent. Many clinicians, however, appear skeptical of the
usefulness of satisfaction measures. This may, in part,
reflect a belief that as an indicator of health care quality,
satisfaction is an administrative issue and not something
that is of interest or use to clinicians who are often more
immediately concerned with the effectiveness of their
treatment interventions than with how patients feel
about the cost or accessibility of services, the parking and
food at the hospital, or their interpersonal relationships
with health care professionals. There also seems to be a
sense that these measures are concerned with aspects of
patients’ health care experience over which clinicians
have no control.
It is known that “satisfied and dissatisfied patients
behave differently.”
20
Satisfied patients seem more likely
to cooperate with their treatments, continue using med-
ical care services, maintain a relationship with a specific
provider, participate in their own treatment,
3,7
and co-
operate with their health care providers by disclosing
important medical information.
3
Conversely, if dissatis-
fied, “patients may make services less effective, either by
neglecting to seek care when needed or refusing to com-
ply with the prescribed course of treatment.”
48
It is as-
sumed that outcomes such as patient’s health status, fre-
quency and length of hospital stay, and continuity of care
are influenced by patient compliance and satisfaction.
3,38
Because patient satisfaction can influence the quality of
care provided and the outcomes of treatment, this infor-
mation can be of direct relevance and value to clinicians.
Having given the “right” treatment, it is reasonable to
believe that clinicians should strive to satisfy their pa-
tients as well.
The heterogeneity of patient satisfaction measures
makes choosing a measure confusing. It is noteworthy
that the word “satisfaction” does not even appear in
many measures considered reflective of this concept.
Lack of clarity concerning the meaning of satisfaction
and its relation to other measures has been highlighted as
a major weakness in this field of inquiry.
1,22,39,55,73,74,76
For example, how does health care quality relate to sat-
isfaction, and can it be assumed that high ratings of
health care quality imply high levels of satisfaction? The
correlation among various measures or concepts is be-
yond the scope of this article. In this article, a satisfaction
measure is considered any measure that seeks patients’
evaluations or affective responses to distinct dimensions
of the health care experience.
35,66
Clarity about how satisfaction measures differ is nec-
essary, because conclusions about satisfaction vary de-
pending on the characteristics of the measures
used.
13,49,65,68
This article is intended to orient clinicians
to the characteristics of patient satisfaction measures and
should assist them in selecting and administering a mea-
sure appropriate for their needs. Three measures will be
discussed in detail: the Patient Satisfaction Question-
naire (PSQ),
71
the Patient Satisfaction Scale (PSS),
8
and
the Client Satisfaction Questionnaire (CSQ).
32
These
particular measures were selected because they vary
across many of the characteristics of satisfaction mea-
sures. Although the PSS is intended specifically for pa-
tients with low back pain, the PSQ and CSQ are com-
monly used for any patient group. These are by no means
the only satisfaction measures available, nor are they
always the best choice depending on the users’ purpose.
Comprehensive reviews of satisfaction measures
are available.
20,47,66
Characteristics of Satisfaction Measures
Satisfaction measures differ in a variety of ways.
42
The
characteristics of satisfaction measures fall into two ma-
jor categories: content and method (Figure 1). Content
refers to the focus or substance of the measure, and
method refers to how the measure is administered and
presented. Within content, four axes representing the
fundamental ways that any satisfaction measure can be
classified can be identified. First, measures may be
global, consisting of one or two questions about overall
or general satisfaction, or multidimensional, consisting
of multiple items probing satisfaction with different as-
pects. Second, the focus of the measure may be satisfac-
tion with either care or treatment outcome. Third, mea-
sures may be generic and intended to be used in any
patient population or disease-specific and designed to be
used in particular patient populations. Finally, direct
measures ask about the patient’s personal experiences
with health care, and indirect measures ask about the
patient’s attitudes toward health care and the health care
system in general.
From the *Institute for Medical Sciences and the Departments of †Sur-
gery and the ‡Public Health Sciences, Clinical Epidemiology, and
Health Care Research Program, University of Toronto, and The Hos-
pital for Sick Children, Toronto, Canada.
Supported in part by the Physical Medicine Rehabilitation Foundation
through a Woodbridge Research Grant and by the Physiotherapy
Foundation of Canada through an Ann Collins Whitmore Memorial
Award (PLH). JGW is the R. B. Salter Chair of Surgical Research and is
supported as a Scientist of the Medical Research Council of Canada.
3167
A number of axes within method can also be identi-
fied. Measures may focus on factual aspects of care (or
outcome), or on the patient’s affective responses to that
care (or outcome), may use closed- or open-ended ques-
tions, and may be self- or interviewer-administered. Fi-
nally, measures vary according to the type of response
format used (Figure 1).
Content Axes
The axes within content describe the focus or substance
of patient satisfaction measures.
Global–Multidimensional
Global measures consist of one or two questions about
overall or general satisfaction. Examples of global rat-
ings include a single item rating of satisfaction on a
10-cm visual analog scale with anchors of “completely
dissatisfied” and “very satisfied” at the extreme ends of
the horizontal line
10
or a single question, “All things
considered, how satisfied are you with the care you re-
ceived?” rated on a 4-point rating scale (very satisfied,
somewhat satisfied, somewhat dissatisfied, very dissatis-
fied). In contrast, the PSQ and CSQ are examples of
multidimensional measures. The PSQ
71
consists of 43
items reflecting six dimensions of service quality related
to hospital care: access to care, availability of services,
technical quality of care, interpersonal care, communi-
cation, and financial. Originally a 36-item measure, the
CSQ has been divided into two parallel 18-item mea-
sures
34
and an 8-item measure.
32
The CSQ examines
dimensions of care, such as physical surroundings, gen-
eral satisfaction, and interpersonal and technical aspects
of care.
With global measures, the dimensions to be consid-
ered are not provided. It is not known what the patient is
thinking about in rating satisfaction with health care,
including what the patient included or excluded in mak-
ing decisions about satisfaction, why these things re-
ceived high or low ratings, or how they were combined
or weighted.
19
Therefore, while global measures are sim-
ple, direct, easy to construct, and easy to use,
19
not
knowing what patients have in mind when they form the
global rating means the measure may not be used by
different people in a reproducible manner.
19
Multidi-
mensional measures, however, ask patients explicitly to
think about particular aspects or dimensions of satisfac-
tion, and the multiple items probe satisfaction with dif-
ferent dimensions. All patients rate the same items. This
approach assumes that various dimensions within the
concept of satisfaction are distinct, which indeed appears
to be the case.
5,12,40,47,70
Various authors have suggested
that patients recognize distinct dimensions of care when
forming an opinion about its quality.
51,55
The challenge in developing multidimensional mea-
sures is to ensure that the measure has enough items and
adequately covers the important dimensions of satisfac-
tion. Ware et al
71
reviewed the content of existing satis-
faction measures and concluded that the dimensions
commonly included were interpersonal manner, techni-
cal quality, accessibility and convenience, finances, effi-
cacy and outcomes, continuity, physical environment,
and availability. These dimensions represent the major
characteristics of providers and health and medical care
services found in the published literature. Similar dimen-
sions (except finances) have also been proposed for the
United Kingdom.
23,55
In a meta-analysis,
27
the research-
ers reported that 25% of the studies reviewed measured
only one dimension (although often using multiple items
all referring to that dimension), whereas 46% used mea-
sures tapping two to four dimensions, 23% tapped five
to seven dimensions, and only 6% tapped eight or
more dimensions.
The main potential disadvantage of multidimensional
measures is failure to consider all aspects of satisfaction
important to patients.
66
Although patients have had little
input into defining the dimensions covered in patient
satisfaction measures,
11
information is available about
what patients consider important when it comes to
health care. Findings in recent work indicate that the
focus and emphasis of patient satisfaction should be re-
considered: “amenities (e.g., food, parking, cleanliness)
that are so prominent in many satisfaction question-
naires are seen as very distinct from quality of care by
most patients, and being treated with respect and being
involved in treatment decisions, aspects of care not in-
cluded in many satisfaction surveys, are paramount is-
sues for patients.”
11
Having someone show interest and
concern and take the time to listen and explain is a pri-
ority for patients.
48
The failure to communicate infor-
mation about the condition and treatment options is the
most frequent source of patient dissatisfaction.
7,39
Two specific limitations of global measures for under-
standing patient satisfaction have been described, both
of which are potentially minimized by the use of a mul-
tidimensional measure. First, because global measures
can mask specific dissatisfactions
33,39
and do not provide
detailed information indicating where changes or im-
Figure 1. Characteristics of patient satisfaction measures.
3168 Spine •Volume 25 •Number 24 •2000
provements are needed to increase patient satisfac-
tion,
5,39
they are considered less informative than multi-
dimensional measures, which assess different dimensions
of care. In addition, global measures of satisfaction tend
to produce scores that are highly skewed, with a large
proportion of respondents reporting high levels of satis-
faction.
5,20,39,40
This tendency could be minimized by
use of a multi-item measure, because these usually have
greater score variability then global measures.
67
Multi-
item measures also typically have higher reliability and
validity than global measures,
55,66,67
although it is not
clear which approach is actually more valid in the con-
text of patient satisfaction.
Which type of measure should clinicians choose? The
most satisfactory solution is to use a multidimensional
measure when possible, supplemented by a global mea-
sure. This approach combines the reproducible consis-
tency, higher reliability, greater score variability, and ad-
ditional information provided by a multidimensional
measure, with the simplicity and practicality of a global
measure. When a multidimensional measure is selected,
the clinician should check to see that the dimensions in
which he or she is interested are actually represented,
because these vary among the different measures. At min-
imum, questions regarding interpersonal behavior and
caring, involvement in decision making, and adequacy
and accuracy of information should be included.
Care–Outcome
Although the association between these concepts is com-
plex and not well understood, there is evidence to indi-
cate that many patients evaluate care and outcome sep-
arately. In a study of individuals attending outpatient
neurology clinics in the United Kingdom for the diagno-
sis and treatment of migraine,
22
“respondents’ accounts
made it clear that attitudinal responses to the role of the
physician mattered less to them and were independent of
judgments of the success of the visit.” This indicates that
patients can and do make separate judgments of care and
treatment outcome, and for many, the outcome is most
important. Treatment outcome is also of fundamental
importance to clinicians.
As a measure of treatment outcome, satisfaction is the
patient’s rating of a particular treatment intervention; as
a measure of care, satisfaction is a patient’s rating of the
quality of the medical care process.
71
For example,
“How satisfied are you with the overall results of your
back surgery?”
44
is a measure of treatment outcome,
whereas “In an overall, general sense, how satisfied are
you with the service you received?”
32
is a measure
of care.
Most existing measures emphasize satisfaction with
care. Satisfaction with treatment outcome was included
in only 4% of the studies included in a meta-analysis of
patient satisfaction with medical care
27
. This inattention
to outcome has been highlighted by others.
7
Some mul-
tidimensional measures concerned primarily with care
include an item(s) on treatment outcome; for example,
although the PSS focuses predominantly on satisfaction
with care for low back pain, it includes a question on the
effectiveness of prescribed treatment for low back pain.
8
Although satisfaction with the treatment outcome
may contribute to satisfaction with care (and vice versa),
these concepts should be assessed separately, because
they represent distinct phenomena that are more validly
conceptualized from different perspectives.
30
Existing
satisfaction measures are built on viewing patients as
customers and the body as a broken or damaged object
or machine. This view may be appropriate to evaluate
satisfaction with care—i.e., those parts of a health care
experience external to the person and his or her body
(e.g., friendliness of staff, surroundings, and food) but
becomes more complicated when satisfaction with treat-
ment outcome is measured. This is because asking pa-
tients to rate their satisfaction with the outcome of
treatment assumes the treatments are measurable in iso-
lation. It requires patients to rate treatments as though
they were something “separate” from their bodies. How-
ever, because the body itself is always implicated in treat-
ment interventions, patients can never truly detach them-
selves to evaluate outcomes in the same way that they can
rate a consumer product or service. Thus, the challenge
when measuring satisfaction with treatment outcome is
to consider the unique relationship of patients to their
bodies and to particular treatment interventions and to
include questions about the experiential, emotional, and
social aspects of treatment outcome.
Independent of this argument, there is good reason to
separate satisfaction with care and treatment outcome
depending on the intended purpose of measurement. A
measure of satisfaction with care is unlikely to be as
capable of detecting differential treatment effects in a
randomized control trial or pre-postintervention study
than a measure of satisfaction with outcome,
9
because
the correlation between outcomes and the quality of care
is probably weak.
11,70
Measures of satisfaction with care can be further dis-
tinguished according to their focus on health plans, hos-
pitals, or physicians (or care givers) and aspects of the
medical care process.
17
Particularly in the United States,
clinicians have probably heard of different initiatives to
develop standardized measures of health care quality.
The Health Plan Employer Data and Information Set
(HEDIS) was developed by the National Committee on
Quality Assurance (NCQA) to measure health care qual-
ity in managed-care health plans.
63
The Consumer As-
sessment of Health Plans (CAHPS) Project, funded by
the Agency for Health Care Policy Research (AHCPR), is
a standardized set of surveys designed to collect patients’
assessments of their health plans.
2,6,14,28,65
These two
initiatives were recently unified to produce the
CAHPS2.0H survey.
4,42
Other standardized systems be-
ing used or developed for health care performance are
described by Di Palo.
17
Items about aspects of a health plan are designed for
the patient as a member of that plan and reflect satisfac-
3169Patient Satisfaction Measures •Hudak and Wright
tion with the managed-care experience.
17
The CAHPS
surveys actually contain a blend of items, including items
about interactions with a personal physician during the
previous 6 months and aspects of the health plan. Items
about a physician and aspects of the medical care process
(including access to care, thoroughness of care, and so
on) reflect patients’ actual experiences while in care and
are probably of most immediate concern to both clini-
cians and patients when considering an episode of care
for a particular health concern.
In summary, clinicians should measure satisfaction
with care and satisfaction with treatment outcome sepa-
rately. Measuring care and treatment outcome indepen-
dently allows clinicians to look at their relationship to
each other, as well as to other variables. There are many
standardized multidimensional measures of satisfaction
with care from which to choose,
47,71
including measures
of satisfaction with a health plan,
14,42
with the hospi-
tal,
41
and with caregivers and with aspects of the medical
care process.
31,52,53,71,72
When it comes to satisfaction
with treatment outcome, however, standardized multidi-
mensional measures have yet to be developed for many
clinical conditions, including low back pain. As men-
tioned previously, the PSS combines satisfaction with
care and treatment.
8
However, although one item in the
effectiveness subscale asks about effectiveness of pre-
scribed treatment for low back pain, the measure does
not inquire about satisfaction with different dimensions
of treatment outcome. In the spine literature, satisfaction
with outcome has been most often assessed using a
global measure.
29,37,62
Generic–Disease–Specific
Generic scales can be used to assess satisfaction in any
patient population, whereas disease-specific scales have
been designed for use in particular patient populations
and include items that are appropriate for that group.
The PSQ and the CSQ are generic measures that can
be used to assess patient satisfaction in any clinical set-
ting, although the PSQ was originally designed as a pop-
ulation survey instrument and the CSQ for mental health
and general health care settings. The PSS is a multidimen-
sional, disease-specific measure of patient satisfaction in-
tended specifically for patients with low back pain. This
measure covers three dimensions: information received
from the provider, caring, and effectiveness of treatment.
The pros and cons of generic and disease-specific mea-
sures of patient satisfaction are the same as those for any
concept. Because the content of generic measures is gen-
eral or broad enough to be applicable to any situation in
which patient satisfaction information is collected, these
measures allow for comparisons across conditions and
settings. They cannot, however, reflect sources of satis-
faction (or dissatisfaction) that are unique to a particular
disease or health care setting. Disease-specific measures,
on the contrary, are very sensitive to differences in satis-
faction associated with a particular context or condition
but may miss unexpected sources of satisfaction. The
usual recommendation when measuring health status is
to administer both a generic and disease-specific mea-
sure, because they capture different aspects of health.
Although the Foundation for Accountability endorses
both general and disease-specific satisfaction mea-
sures,
17,24
comparative studies providing evidence to use
when deciding to recommend one type of measure over
another are not available. Given the focus of satisfaction
with care, it may well be that a generic measure is ade-
quate. Satisfaction with treatment outcome however, is
likely to be captured more precisely by a disease-
specific measure.
Direct–Indirect
Direct measures ask patients about their own experi-
ences with health care, and indirect scales ask about pa-
tient’s attitudes toward health care and the health care
system in general.
20
When patients are asked about spe-
cific providers or services, or about specific events or
episodes, the measure is considered direct. When asked
about doctors or health care in general or about their
own general experience over a given time frame (e.g., the
past 12 months), the measure is indirect.
20
Direct mea-
sures are alternatively referred to as specific or micro
measures, whereas indirect ones are classified as general
or macro measures.
45
The PSQ is an indirect measure that asks patients
about their attitudes toward the health system or about
care in general. Patients are asked to keep in mind the
medical care they are currently receiving (which can be
from any source) when responding to questions such as
“I feel confident that I can get the medical care I need
without being set back financially.” The CSQ, in con-
trast, is a direct measure in which patients are asked to
think about the services they have received from a par-
ticular program or kind of care (in-patient, adult ambu-
latory, pediatric) when responding to questions such as
“Have the services you received helped you to deal more
effectively with your problems?” Finally, the PSS asks
patients directly about care received specifically for low
back pain: “The physician did not understand the con-
cerns I had about my back problem.”
Direct and indirect measures of satisfaction appear to
assess different dimensions of satisfaction.
40,46,57
Direct
inquiries about care received have been associated with
greater satisfaction than indirect inquiries,
27,33
and di-
rect measures assess services received, or not received,
more precisely.
26,44
Because of their specificity, direct
measures are considered by clinicians to be more appro-
priate and informative for probing satisfaction with spe-
cific medical encounters,
65
specific settings
45
or particu-
lar treatment interventions. In addition, direct measures
can provide useful information for continual quality-
improvement programs or program evaluation.
20
These four axes, global–multidimensional, care–
treatment outcome, generic–disease specific, and direct–
indirect represent fundamental ways in which the con-
tent of patient satisfaction measures can be classified.
3170 Spine •Volume 25 •Number 24 •2000
Sample items reflecting each of these content character-
istics are shown in Table 1.
Method Axes
The axes within method describe the different ways that
measures of patient satisfaction can be administered
and presented.
Factual–Affective
Factual measures focus on the facts of what actually oc-
curred, whereas affective measures emphasize the pa-
tient’s perception of what occurred. This has also been
described as a distinction between reports and ratings (or
evaluations).
11,70
Although reports are considered more
objective, ratings are subjective and presumed to “cap-
ture a personal evaluation of care that cannot be shown
by observing care directly.”
71
For example, “How long
did you spend with your physician today?” is a factual
question asking for an objective response in number of
minutes, whereas “Did your physician spend enough
time with you today?” is an affective question emphasiz-
ing not the relative period of time spent with the physi-
cian, but rather the patient’s perception of the adequacy
of that time.
20
Ratings are considered both a measure of
care and a reflection of the respondent, because they are
influenced not only by the care received, but also by
patients’ preferences and expectations.
71
The PSQ includes a mix of factual and affective ques-
tions and statements: “I have to pay for more of my
medical care than I can afford” is a factual statement,
and “Sometimes doctors make me feel foolish” is an
affective one. The CSQ has affective questions, such as
“How would you rate the quality of service you re-
ceived?” The PSS also has a mix of statements and ques-
tions, including “The physician gave me a clear idea of
how long it might take for my back to get better” (fac-
tual) and “The physician seemed to believe that my pain
was real” (affective).
Whether clinicians choose a factual or affective mea-
sure depends on the type of information desired and how
it is to be used. If the goal is complete understanding of
how patients perceive the care they receive, subjective
perceptions or ratings are appropriate.
20
As part of a
quality assurance program however, facts or reports can
be more readily interpreted and acted on and are consid-
ered by some to better reflect the quality of care than
ratings of satisfaction or excellence.
11
Open–Closed-Ended Questions
Open-ended or unstructured questions ask the patient to
talk about or comment in writing on the services re-
ceived. There are no answer choices from which a re-
sponse must be selected.
15,76
Closed-ended questions
force patients to select a response from among a limited
number of possible responses generated by the developer
of the measure. The PSQ, CSQ, and PSS all use closed-
ended questions, although the CSQ has a single open-
ended section in which additional comments and sugges-
tions can be provided. “The physician gave me enough
information about the cause of my back pain” is a
closed-ended statement from the PSS. “How satisfied are
you with the care provided in this hospital’s orthopedic
clinic?” is an open-ended question when no response
options are given; rather, space is allowed for patients to
answer as they choose.
Although individuals provide responses about satis-
faction or dissatisfaction with different dimensions when
asked specific questions, they may not consider particu-
lar dimensions important enough for them to describe
spontaneously in response to an open-ended ques-
tion.
7,39
This is consistent with the finding that individ-
uals are more likely to report health events when the
event is perceived as important or as having a significant
impact.
39
Specific questions (which are usually closed
ended, although not necessarily so) appear to act as
probes to elicit dissatisfaction with dimensions of care or
outcome that have less importance for the respondent
than those mentioned in response to open-
ended questions.
Because closed-ended questions may help to avoid un-
derreporting, and open-ended ones may be useful for
assessing patients’ priorities, the concurrent use of both
closed- and open-ended questions should provide a more
complete understanding of satisfaction.
7
A simple exam-
ple is provided by Nork et al
44
who combined the closed-
ended question, “How satisfied are you with the overall
Table 1. Sample Items According to Content Characteristics
Global “In an overall, general sense, how satisfied are you with the service you received?” (CSQ)
Multidimensional “Doctors examine their patients carefully before deciding what is wrong.” (thoroughness) (PSQ)
“I can reach a doctor to ask him questions at any time.” (accessibility) (PSQ)
“I think my doctor forgets to ask about problems I’ve had in the past.” (follow-up) (PSQ)
Care “All things considered, how satisfied are you with your practice?” (GPAS)
Treatment outcome “How satisfied are you with the overall results of your back surgery?” (FOQ)
“The treatment the doctor prescribed for my back was effective.” (PSS)
Generic “Have the services you received helped you to deal more effectively with your problems?” (CSQ)
Disease-specific “The doctor told me what to do to prevent future back problems.” (PSS)
Direct “The doctor did not understand the concerns I had about my back problem.” (PSS)
Indirect “Doctors always try to explain about an operation so the patient won’t worry.” (PSQ)
PSS ⫽Patient Satisfaction Scales (Cherkin et al, 1991); PSQ ⫽Patient Satisfaction Questionnaire (Ware et al, 1983); CSQ ⫽Client Satisfaction Questionnaire
(Larsen et al, 1979); FOQ ⫽Functional Outcomes Questionnaire (Nork et al, 1999); GPAS ⫽General Practice Assessment Survey (National Primary Care Research
& Development Centre).
3171Patient Satisfaction Measures •Hudak and Wright
results of your back surgery?” (responses: very satisfied,
fairly satisfied, not too satisfied, not at all satisfied) with
the open-ended question “If not too satisfied or not at all
satisfied, what are the reasons for your dissatisfaction?”
Self-Administered–Interviewer-Administered
The format of most satisfaction measures is written and
self-report, although interviewer-administered question-
naires are also available (e.g., The Evaluation Rating
Scale).
46
Interviewer-administered measures have been
shown to yield higher satisfaction scores than self-
administered ones.
20,34
This potential bias toward
higher scores, as well as the additional cost and time
required to use interviewer-administered measures,
should be considered. There are, however, situations in
which an interviewer-administered method is preferable.
For example, little is known about nonrespondents, and
yet these individuals potentially have much to teach us
about context-specific sources of dissatisfaction. Ac-
tively seeking out and focusing on reasons for dissatis-
faction may be most effectively accomplished in an inter-
view, rather than by self-report questionnaire, because a
well-trained interviewer can effectively probe for sources
of dissatisfaction not easily elicited using a standard-
ized measure.
7
Response Formats
The response formats for closed-ended questions vary.
All items in the PSQ and PSS are rated using a 5-point
“agree–disagree” response format. Items in the CSQ are
rated using a variety of different 4-point response for-
mats (including excellent–poor, very satisfied–quite dis-
satisfied, and no, definitely not–yes, definitely).
Although studies comparing different response for-
mats have been reported,
49,65
it remains unclear which
format(s) are superior. Ware and Hays
65
demonstrated
that a 5-point evaluation rating scale (excellent, very
good, good, fair, poor) yields lower mean scores, greater
response variability, and higher correlations with mea-
sures of behavioral intention than a 6-point satisfaction
scale (extremely satisfied, very satisfied, somewhat satis-
fied, neither satisfied nor dissatisfied, somewhat dissatis-
fied, very dissatisfied). Based on a comparative study of
seven measures of patient satisfaction, Ross et al
49
also
recommended use of response formats such as the eval-
uation rating scale as a way to deal with the problem of
acquiescence in satisfaction measurement. Acquiescence,
or “yea-saying”
59
is the general tendency to agree with
items no matter what the content of the items.
49
Response formats are not independent of the ques-
tions they accompany, and part of the issue is therefore
related to the framing of the questions themselves. The
evaluation rating scale is accompanied by such questions
as “How would you rate the technical quality of this
visit?”
65
whereas the satisfaction rating scale has ques-
tions such as “How satisfied are you with the technical
quality of this visit?” The first question asks for an “eval-
uation” of technical quality, whereas the second asks
directly about “satisfaction” with the technical quality.
A high score on the evaluation rating scale is assumed to
indicate high satisfaction, because patients are assumed
to value technical quality, for example.
27
Because these
two formats are highly correlated, it may be argued that
they can be used interchangeably.
27,50
Various authors,
however, have highlighted the need to clearly distinguish
between an evaluation of care received and satisfaction
with that care,
9,27,36,54,74
because asking about the qual-
ity or success of care is not equivalent to asking about
satisfaction with that care.
54
Asking patients whether
they are happy is also not the same as asking whether
they are satisfied.
25
Two recent qualitative studies highlight the complex-
ity of patient satisfaction and add to this debate concern-
ing evaluations of care and satisfaction with care. Results
in these two studies showed incongruity between what
patients actually expressed in interviews about their
health care experiences and their reports of satisfaction
when standardized satisfaction measures were used. Wil-
liams et al
74
conducted both unstructured interviews and
structured discussions around responses to the CSQ-18
with mental health services users. They found that ex-
pressions of satisfaction on the CSQ-18 hid a variety of
negative experiences. Experiences described in positive
or negative terms did not necessarily correlate with pa-
tients’ evaluations of the service. Dougall et al
18
also
found that although patients verbally and on standard-
ized patient satisfaction measures expressed high levels
of satisfaction, in-depth interviews showed many nega-
tive perceptions and experiences. Based on these find-
ings, Dougall et al argued that standardized patient sat-
isfaction measures alone are inadequate indicators of
patients’ experiences in health care delivery.
These findings reflect a certain amount of ambiguity
about the concept of satisfaction itself and emphasize the
need for clarification regarding the association between
perceptions and experiences and expressions of satisfac-
tion. Differences in response format terminology proba-
bly stem from this ambiguity as well, in addition to being
attempts to find methods that minimize bias and improve
the distribution of responses along the satisfac-
tion continuum.
Measures of Behavioral Intention
Behavioral intentions refer to a person’s intentions to
behave in a certain manner.
21
Intention to purchase a
product is often measured as an indicator of product
satisfaction in marketing research.
49
Measures of behav-
ioral intentions used as indicators of patient satisfaction
include questions about willingness to return for care,
willingness to recommend the physician (or program) to
a friend, and intention to follow the doctors’ recommen-
dations.
49
Although these measures have face validity,
there is some evidence that they do not perform as well as
other measures of satisfaction. In a comparative study of
seven different measures of patient satisfaction, the mea-
sure of behavioral intention was judged to be inade-
quate, because the distribution of responses was highly
3172 Spine •Volume 25 •Number 24 •2000
skewed (even after transformation) and reliability in
highly acquiescent patients was poor.
49
Measurement Issues
The measurement of patient satisfaction suffers from
some specific methodologic problems, including selec-
tion bias, high undifferentiated levels of satisfaction, and
acquiescence bias.
Selection Bias
Satisfaction is usually measured at the end of a program
or phase of treatment.
5
Unfortunately, dissatisfied pa-
tients are more likely to withdraw from treatment before
completion, particularly when care extends over
time,
5,32,45
and therefore they are underrepresented in
the sample of respondents. This systematic bias toward
representation of individuals who have completed a pro-
gram or treatment in studies of patient satisfaction may
yield high satisfaction scores, but these scores can be
misleading and show little about where improvements in
care are needed
7
or when less than desirable treatment
outcomes are obtained. It is unlikely that the roots of
dissatisfaction will be discovered by consulting only
those patients who have completed a program or episode
of treatment.
Strategies have been proposed for minimizing selec-
tion bias
20
: In continuing care settings (e.g., family med-
icine, general pediatrics), a cross-section of patients
should be sampled to obtain information from patients
who have varying levels of exposure to the physician or
institution. In episodic care settings (e.g., specialty con-
sultations for specific management problems or proce-
dures), patients should be sampled when care is com-
pleted. Patients tend to describe their experiences in
terms of episodes of care rather than single visits, partic-
ularly when seeking care for a complicated condition.
11
When possible, satisfaction questionnaires should be
given (rather than mailed) to patients known to be leav-
ing the program or practice.
Clinicians should make a point to follow up a small
sample of nonrespondents and dropouts.
7
Blais
5
re-
ported that although patients who discontinue treatment
reported lower satisfaction with some dimensions of a
health promotion program, there were no significant dif-
ferences in other dimensions between those who with-
drew and those who completed treatment. For this rea-
son, he suggests that rather than ask only about
satisfaction, it is important to explore the withdrawing
patient’s actual reasons for leaving the program. This
information can provide insight into those dimensions of
care that have been influential in the decision to leave the
program and should be considered to reduce the with-
drawal rate and to highlight any external factors unre-
lated to satisfaction that contribute to attrition.
5
High Undifferentiated Levels of Satisfaction
From a measurement perspective, high undifferentiated
levels of satisfaction is a constant problem
5,26
that ap-
plies irrespective of the population studied, the method
used, or the target of the ratings.
5,32,60
When the distri-
bution of satisfaction scores is highly skewed or clustered
in only a few response options at the top of the possible
range, it is almost impossible for the measure to detect
any improvement or distinguish different levels or de-
grees of satisfaction.
26,59
Although high levels of satis-
faction may look good, these results are of little use when
directions for improvement in care are sought or when
attempting to detect small but important differences in
satisfaction related to the outcome of different treatment
interventions. Possible reasons for this clustering of
scores at the extreme positive end of the scale (whether
represented by a response of extremely satisfied or
strongly agree) is the overrepresentation of patients who
have completed a program or episode of treatment (se-
lection bias) and the use of global measures. It is also
possible, however, that high scores are an accurate re-
flection of patients’ satisfaction with a particular health
care program or treatment outcome.
Because multi-item measures usually yield more score
variability, as well as higher reliability and validity, than
global or single-item measures,
67
this approach should
reduce the problem of skewed satisfaction scores.
5
In
addition, because questions of a detailed and specific
nature generate greater levels of dissatisfaction than
more general questions,
55
a multidimensional measure
should help to clarify whether the issue is that the the
right questions have not been asked or the patients truly
are satisfied. Other approaches taken to overcome this
problem include forcing the respondent to rank the dif-
ferent aspects of care
45
and increasing the number of
points at the positive end of the satisfaction scale.
64
Acquiescence Bias
Acquiescence bias, the “tendency to agree with state-
ments of opinion regardless of content,”
69
is a potential
source of bias in patient satisfaction research. When this
bias is present, the wording of the items can influence the
results obtained. Agreement with favorably worded
items results in higher levels of satisfaction (thus overes-
timating satisfaction), whereas agreement with nega-
tively worded items results in lower levels (thus underes-
timating satisfaction).
49,69
Single-item satisfaction
measures and measures containing all favorably or all
unfavorably worded items tend to be biased by acquies-
cence to a greater extent than measures that have a bal-
ance of positively and negatively worded items.
69
Al-
though the use of balanced measures has been
recommended as a way to minimize this problem,
69
re-
sults in recent work have not supported this approach as
a solution.
49
Ross et al
49
have suggested the use of neu-
trally worded items and response formats, such as the
evaluation rating scale (poor, fair, good, very good, ex-
cellent) as a way to deal with the problem of acquies-
cence. This suggestion, however, should be weighted
with the previous argument (see Methods–Response For-
mats) that if information about satisfaction is wanted,
then patients should be asked about satisfaction. The use
3173Patient Satisfaction Measures •Hudak and Wright
of neutrally worded questions is a reasonable or sensi-
ble alternative.
Because there are no specific measures for estimating
the extent to which acquiescence bias is operating in a
given situation,
59
it is difficult to assess, unless the data
collections methods have been specifically designed in
advance to do so.
49,69,75
Some investigators have esti-
mated the tendency to acquiesce by simply counting the
number of items on a measure answered favorably or
unfavorably by each subject,
59
but this approach is
crude. The most reasonable way to proceed is to assume
that acquiescence bias is likely to influence the satisfac-
tion results to some extent. Clinicians should consider
this bias as one possible explanation for high levels of
satisfaction, apparent group differences in satisfaction,
or absence of group differences when the groups differ
with respect to age, health status, education or income
(because individuals who are older, less well educated,
less affluent, and in poorer health have a greater tendency
to acquiesce).
49,69
However, group differences in satis-
faction scores of one standard deviation or more are
unlikely to be due to acquiescence bias and probably
reflect true group differences in satisfaction.
69
How Should the Clinician Proceed?
At the end of this discussion of why it is important to
measure patient satisfaction and how the different char-
acteristics of these measures should be considered when
choosing a measure, it is not possible to recommend any
particular measure, because no single measure alone is
ideal. Some investigators have actually cautioned against
the use of standardized measures because they are not
designed to capture context-specific dimensions of pa-
tient satisfaction.
7
Standardized measures are easy to
use, however, and allow for comparisons across settings
and studies
20
and have often been through the rigors of
psychometric testing for reliability and validity. What
“validity” in this context really means is not, however,
clear.
73
At least some patients’ descriptions of their
health care experiences appear inconsistent with their
reports on satisfaction measures,
18
which emphasizes the
need for clarification about the association between per-
ceptions and experiences and expressions of satisfaction.
Ambiguity about the meaning of satisfaction itself is the
dominant weakness in this field of inquiry, particularly
because of the way in which meaning influences the de-
sign and interpretation of measures.
Clinicians are faced with a difficult choice. Obviously,
it is much simpler to use an already developed measure.
A standardized measure does not, however, provide the
richness of detail and potential depth of understanding
possible with a qualitative approach. The collection and
interpretation of qualitative data are more labor-
intensive than is necessary for a standardized measure,
and some knowledge of and skill in this research ap-
proach are needed. Given these caveats, the preferable
option is to complement the use of a standardized quan-
titative measure with sections of open-ended questions
or even a qualitative method.
Choosing a measure simply because it is standardized
is not sufficient. To ensure that a measure is appropriate
for specific purposes, consideration of the different char-
acteristics of the measure discussed in this article is rec-
ommended. In summary:
1. A multidimensional measure should be used when
possible, supplemented by a global measure. Global
measures are easy to use, whereas multidimensional
measures have many desirable measurement proper-
ties and provide a wider breadth of information (in-
cluding possible areas for improvement).
2. The actual items in the measure should be exam-
ined carefully and regarded in relation to the clin-
ical setting and patient population. If there are di-
mensions specific to a particular context or patient
that are not covered, addition of questions should
be considered. The dimensions of interpersonal be-
havior and caring, involvement in decision making
and adequacy and accuracy of information are im-
portant to patients and should be included. If ques-
tions are added, advice about the construction of
the question and analysis should be sought from a
statistician (closed-ended questions) or a qualita-
tive researcher (open-ended questions).
3. Satisfaction with care should be assessed sepa-
rately from satisfaction with treatment outcome.
There are different multidimensional measures of
satisfaction with care, but measures of satisfaction
with treatment outcome currently do not exist for
many clinical conditions.
4. Direct measures are preferable. In general, the speci-
ficity of direct measures makes them more informa-
tive than indirect measures for clinicians interested in
probing satisfaction with specific medical encounters
or particular treatment interventions.
5. Both closed- and open-ended questions should be in-
cluded. The concurrent use of both types of question
should lead to a fuller understanding of satisfaction.
6. A sample of nonrespondents should be included.
Patients who have left a program or phase of treat-
ment should be interviewed in greater detail about
their perceptions and experiences, as well as their
satisfaction with care and treatment outcome. The
interviewer should carefully probe for sources of
dissatisfaction. Patients’ actual reasons for leaving
the program or choosing not to cooperate with
recommended treatments should also be explored.
Although these interviews should be open ended
and unstructured, consistent lines of inquiry can be
developed and pursued. The development of rec-
ommendations for such an interview guide would
be useful.
3174 Spine •Volume 25 •Number 24 •2000
Recommendations for Measuring Satisfaction in
Patients With Lumbar Spine Problems
Multidimensional Measures
Satisfaction with Care. Particularly when measuring sat-
isfaction with care, generic measures
52,71
or measures
intended for use in primary care settings,
43,53
general
internal medicine,
72
or other orthopedic practices
56
may
also perform well in low back pain patients. Because data
are unavailable to support this practice, however, only
those measures designed specifically for use in patients
with low back pain will be reviewed. Measures of satis-
faction with caregivers and aspects of the medical care
process are the focus, rather than measures of satisfac-
tion with health plans or hospital care, because these
measures may not be specific enough to capture the ac-
tual experience of patients who receive care for an epi-
sode of low back pain. The CAHPS surveys,
2
for exam-
ple, ask patients to comment on care received during the
past 6 months for all health care needs received from a
personal physician and specialists. It seems reasonable to
assume that for clinicians providing care for an episode
of low back pain, it is the patient’s experience in that
particular intervention that is of most concern.
Two multidimensional measures of satisfaction with
physicians and aspects of the medical care process are
intended for use in outpatient settings: the 9-item mea-
sure developed by Deyo and Diehl
16
and the 17-item PSS
referred to throughout this article.
8
Both measures in-
clude items about information adequacy and physician
caring. The PSS also included items about the technical
quality of care, emotional support and assurance, and
the effectiveness of prescribed treatment. Deyo and
Diehl’s measure includes items about the amount of time
spent with the physician, whether the patient would see
the same physician on a return appointment, and
whether help was sought from another health profes-
sional or hospital after the last visit. Neither measure
addresses issues of access to care, involvement in deci-
sion-making, coordination among caregivers or trust in
the clinician, which are dimensions of importance to pa-
tients. Items in the PSS are rated on a five-category re-
sponse scale (1 ⫽strongly agree to 5 ⫽strongly dis-
agree), whereas those in the measure developed by Deyo
and Diehl are rated on a three-category scale (1 ⫽dis-
satisfaction, 2 ⫽don’t know, 3 ⫽satisfaction). The
greater number of items and response options in the PSS
should allow detection of finer gradations in satisfaction,
and this is therefore the preferred measure. The PSS is
reproduced in Appendix 1.
Development of the PSS involved 305 patients who
were attending a primary care clinic for treatment of
mechanical low back pain.
8
Results of a factor analysis
of patients’ responses to the 17-item PSS indicated that
10 of the 17 items reflect three distinct dimensions of
care: information (three items), caring (four items), and
effectiveness (three items). Subscales scores are calcu-
lated as the unweighted mean of all items in the subscale,
and an overall scale score is created by adding the nu-
merical value of the 10 subscale items. The mean ⫾stan-
dard deviation (Cronbach’s
␣
) for each subscale are in-
formation 2.72 ⫾0.92 (0.75), caring 2.09 ⫾0.67 (0.84),
and effectiveness 2.59 ⫾0.76 (0.71), respectively. The
Cronbach’s
␣
of the overall score was 0.87. Evidence of
construct validity was provided by correlating the sub-
scale scores with other items expected to be associated
with patient satisfaction. All correlation coefficients of
construct validity were statistically significant and in the
expected directions. Although clinicians may choose to
use the 10 subscale items only if length and respondent
burden are issues, use of all 17 items provides addi-
tional information.
Satisfaction With Treatment Outcome. Stocki et al
58
pub-
lished the only multidimensional measure of satisfaction
with treatment outcome used in patients with lumbar
spine problems. Originally used in patients with spinal
stenosis, this six-item measure includes five dimensions
of outcome (pain relief; ability to walk; ability to do
housework, yard work, or job; strength in thighs, legs,
and feet; and balance and steadiness on feet), as well as a
global question about the overall result of surgery. Items
are rated on a four-category response scale (1 ⫽very
satisfied, 2 ⫽somewhat satisfied, 3 ⫽somewhat dissat-
isfied, 4 ⫽very dissatisfied), and a satisfaction scale score
is calculated as the unweighted mean of all six items if the
number of responses exceed four (score range: 1–4). Un-
fortunately, no mention is made of the conceptual basis
used to select items to represent treatment outcome. Im-
portant dimensions, in particular the emotional and so-
cial dimensions of treatment outcome, are not covered.
For clinicians who choose to use this measure, it is rec-
ommended that the overall, or global, item be reported
separately from the unweighted mean of the other five
dimensions of outcome.
Global Measures
There are many variations of global questions for both
satisfaction with care and treatment outcome, including
different measures of behavioral intent (such as “Would
you recommend this treatment to a friend with a similar
condition?”) The authors prefer the following question,
because it is a direct rating of satisfaction and not of
quality or another related concept: “All things consid-
ered, how satisfied are you with the results of your recent
back surgery (or treatment for low back pain or what-
ever is more appropriate for the purpose)?” (1 ⫽ex-
tremely satisfied, 2 ⫽very satisfied, 3 ⫽somewhat sat-
isfied, 4 ⫽mixed [approximately equal satisfaction and
dissatisfaction], 5 ⫽somewhat dissatisfied, 6 ⫽very dis-
satisfied, 7 ⫽extremely dissatisfied, 8 ⫽not sure/
no opinion).
Open-Ended Questions
An open-ended question should be added to both the
multidimensional and global measures. The following is
one example of how this question might be framed when
3175Patient Satisfaction Measures •Hudak and Wright
asking about treatment outcome: “How do you feel
about your recent back surgery and how it turned out?
Please comment in the space below on the results of your
back surgery and your satisfaction or dissatisfaction
with this result. We are interested in your comments,
both favorable and critical, about the surgery.”
Summary
Information on patient satisfaction is considered a way
of including patients’ perspectives in the planning and
assessment of services.
61
The study of patient satisfaction
is a relatively new field, and despite the surge in popular-
ity and use of satisfaction measures during the past three
decades, different issues remain to be explored. This is
not meant to dissuade clinicians from using satisfaction
measures, but rather to allow them to proceed in a
thoughtful way, recognizing what these measures can
reasonably show us about patients’ perceptions of the
care and treatment interventions they receive. The pro-
posed approach to classifying the characteristics of pa-
tient satisfaction measures should help to highlight po-
tential reasons for variation in results when satisfaction
measures perform differently and will be of value if it
increases the specificity with which clinicians select mea-
sures to achieve their purposes.
Appendix
Patient Satisfaction Subscales (reprinted,
with permission
8
)
Questions included in “information” subscale (3)
The physician gave me enough information about the
cause of my back pain.
The physician did not give me a clear explanation of
the cause of my pain.
The physician told me what to do to prevent future
back problems.
Questions included in “caring” subscale (4)
The physician seemed to believe that my pain
was real.
The physician did not understand the concerns I had
about my back problem.
The physician did not seem comfortable dealing with
my back pain.
The physician was not concerned about what hap-
pened with my pain after I left the office.
Questions included in “effectiveness” subscale (3)
The treatment the physician prescribed for my back
was effective.
The physician seemed confident that the treatment
she/he recommended would work.
The physician gave me a clear idea of how long it
might take for my back to get better.
Questions Not Included in Any Subscale (7)
After seeing the physician I did not know what I
needed to do for my back pain.
The physician did not listen carefully to my descrip-
tion of my back problem.
The physician made me feel less worried about my
back problem.
The physician performed a thorough examination of
my back.
The physician did not understand what was wrong
with my back.
The physician should have ordered more tests or ra-
diographs.
The physician should have referred me to a back spe-
cialist.
Response options: 1 ⫽strongly agree, 2 ⫽agree, 3 ⫽
neither agree nor disagree, 4 ⫽disagree, 5 ⫽strongly dis-
agree. For analysis, strongly favorable responses are re-
corded “1” and strongly unfavorable responses are re-
corded “5” (see original article
8
).
Note: Subscales based on results of factor analyses.
References
1. Abramovwitz S, Cote AA, Berry E. Analyzing patient satisfaction: A multi-
analytic approach. Qual Rev Bull 1987;13:122–130.
2. Agency for Health Care Policy Research. Consumer Assessments of Health
Plans Study: Survey and Reporting Kit. Rockville, MD: Agency for Health Care
Policy and CAHPS娂1.0 Research, 1998.
3. Aharony L, Strasser S. Patient satisfaction: what we know about and what
we still need to explore. Med Care Rev 1993;50:49–79.
4. Allen H. Anticipating market demand: tracking enrollee satisfaction and
health over time. Int J Qual Health Care 1998;10:521–30.
5. Blais R. Assessing patient satisfaction with health care: did you drop some-
body? Can J Prog Eval 1990;5:1–13.
6. CAHPS Adult Core Questionnaire—English. Med Care 1999;37:MS106–9.
7. Carr-Hill RA. The measurement of patient satisfaction. J Public Health
Med 1992;14:236–49.
8. Cherkin D, Deyo RA, Berg AO. Evaluation of a physician education inter-
vention to improve primary care for low-back pain: II. Impact on patients.
Spine 1991;16:1173–8.
9. Cherkin DC. Patient satisfaction as an outcome measure. Chiropract
Tech 1990;2:138–42.
10. Cinotti G, Gumina S, Giannicola G, et al. Contralateral recurrent lumbar
disc herniation. Spine 1999;24:800–6.
11. Cleary PD, Edgman-Levitan S. Health care quality: incorporating consumer
perspectives. JAMA 1997;278:1608–12.
12. Cleary PD, McNeil BJ. Patient satisfaction as an indicator of quality care.
Inquiry 1988;25:25–36.
13. Counte MA. An examination of the convergent validity of three measures of
patient satisfaction in an outpatient treatment center. J Chronic Dis 1979;32:
583–8.
14. Crofton C, Lubalin JS, Darby C. Foreward. Med Care 1999;37:MS1–MS9.
15. Del Greco L, Walop W. Questionnaire development: 1. Formulation. Can
Med Assoc J 1987;136:583–5.
16. Deyo R, Diehl AK. Patient satisfaction with medical care for low-back pain.
Spine 1986;11:28–30.
17. Di Palo MT. Rating satisfaction research: is it poor, fair, good, very good or
excellent? Arthritis Care Res 1997;10:422–30.
18. Dougall A, Russel A, Rubin G, et al. Rethinking patient satisfaction: patient
experiences of an open access flexible sigmoidoscopy service. Soc Sci Med 2000;
50:53–62.
19. Feinstein AR. Clinimetrics. New Haven, CT: Yale University Press; 1987.
20. Ferris LE and the Health Services Research Group. A guide to direct mea-
sures of patient satisfaction in clinical practice. Can Med Assoc J 1992;146:
1727–31.
21. Fishbein M, Ajzen I. Belief, attitude, intention and behavior: an introduction
to theory and research. Don Mills, Ontario: Addison-Wesley; 1975.
22. Fitzpatrick R, Hopkins A. Problems in the conceptual framework of patient
3176 Spine •Volume 25 •Number 24 •2000
satisfaction research: an empirical exploration. Sociol Health Illness 1983;5:297–
311.
23. Fitzpatrick R. Measurement of patient satisfaction. In: Measuring the Out-
comes of Medical Care. London: Royal College of Physicians and King’s Fund
Center; 1990.
24. Foundation for Accountability. Accountability! Helping Americans make better
healthcare decisions. Portland, OR: Foundation for Accountability; 1996.
25. Gutek BA. Strategies for studying client satisfaction. J Soc Issues 1978;34:
44–56.
26. Guyatt GH, Mitchell A, Molloy DW, et al. Measuring patient and relative
satisfaction with level or aggressiveness of care and involvement in care decisions
in the context of life threatening illness. J Clin Epidemiol 1995;48:1215–24.
27. Hall JA, Dornan MC. Meta-analysis of satisfaction with medical care: de-
scription of research domain and analysis of overall satisfaction levels. Soc Sci
Med 1988;27:637–44.
28. Hays RD, Shaul JA, Williams VSL, et al. Psychometric properties of the
CAHPS 1.0 survey measures. Med Care 1999;37:MS22–31.
29. Hazard RG, Haugh LD, Green PA, et al. Chronic low back pain the rela-
tionship between patient satisfaction and pain, impairment and disability out-
comes. Spine 1994;19:881–7.
30. Hudak PL, McKeever P, Wright JG. The metaphor of “patient as customer”:
implications for measuring patient satisfaction. J Clin Epidemiol 1999;52:485.
31. Hulka, BS, Zyzanski SJ, Cassel JC, et al. Scale for the measurement of atti-
tudes toward physicians and primary medical care. Med Care 1970;8:429–36.
32. Larson D, Attkisson C, Hargreaves W, et al. Assessment of client/patient satis-
faction: development of a general scale. Eval Program Plann 1979;2:197–207.
33. Lebow JL. Consumer assessments of the quality of medical care. Med
Care 1974;12:328–37.
34. LeVois M, Nguyen T, Attkisson C. Artifact in client satisfaction assessment:
experience in community mental health settings. Eval Program Plann 1981;4:
139–50.
35. Linder-Pelz S. Toward a theory of patient satisfaction. Soc Sci Med 1982;
16:577–82.
36. Linn LS, DiMatteo MR, Chang BL, et al. Consumer values and subsequent
satisfaction ratings of physician behavior. Med Care 1984;22:804–12.
37. Ljunggren AE, Weber H, Kogstad O, et al. Effect of exercise on sick leave due
to low back pain: a randomized, comparative, long-term study. Spine 1997;22:
1610–7.
38. Lochman JE. Factors related to patients’ satisfaction with their medical care.
J Commun Health 1983;9:91–109.
39. Locker D, Dunt D. Theoretical and methodologic issues in sociological
studies of consumer satisfaction with medical care. Soc Sc Med 1978;12:283–92.
40. MacDonald H, Bombardier C. Patient satisfaction with health care: review
of the literature and selection of a measure for the integrated study. Toronto,
Ontario: Institute for Work, Health; 1993. Working Paper No. 9.
41. Meterko M, Rubin H, Ware JEJ, et al. Patient judgments of hospital quality
questionnaire. Med Care 1990;9:S1–44.
42. National Committee for Quality Assurance. Health Plan Employer Data and
Information Set, 1999. Washington, DC: National Committee for Quality As-
surance, 1998.
43. National Primary Care Research and Development Center, University of
Manchester. General Practice Assessment Survey. Available at www.npcrdc.
man.ac.uk
44. Nork SE, Hu SS, Workman KL, et al. Patient outcomes after decompression
and instrumented posterior spinal fusion for degenerative spondylolisthesis.
Spine 1999;24:561–9.
45. Pascoe GC, Attkisson CC. Comparison of indirect and direct approaches to
measuring patient satisfaction. Eval Program Plann 1983;6:359–71.
46. Pascoe GC, Attkisson CC. The evaluation ranking scale: a new methodology
for assessing satisfaction. Eval Program Plann 1983;6:335–47.
47. Pascoe GC. Patient satisfaction in primary health care: a literature review
and analysis. Eval Prog Plan 1983;6:185–210.
48. Rivkin MO, Bush PJ. The satisfaction continuum in health care: consumer
and provider preferences. In S. J. Mushkin (Ed.), Consumer incentives for health
care. New York: Prodist, 1974.
49. Ross CK, Steward CA, Sinacore JM. A comparative study of seven measures
of patient satisfaction. Med Care 1995;33:392–406.
50. Rotter DL, Hall JA, Katz NR. Relations between physicians’ behaviors and
analogue patients’ satisfaction, recall and impressions. Med Care 1987;25:437–51.
51. Rubin HK. Can patients evaluate the quality of hospital care? Med Care
Rev 1990;47:267–326.
52. Rubin HR, Gandek B, Rogers WH, et al. Patients’ ratings of outpatient visits
in different practice settings: results from the medical outcomes study.
JAMA 1993;270:835–40.
53. Safran DG, Kosinski M, Tarlov AR, et al. The primary care assessment survey:
testse of data quality and measurement performance. Med Care 1998;36:728–39.
54. Shaw IF. Literature review: consumer evaluations of the personal social
services. Br J Soc Work 1984;14:277–84.
55. Sitzia J, Wood N. Patient satisfaction: a review of issues and concepts. Soc Sci
Med 1997;45:1829–43.
56. Solomon DH, Bates DW, Horsky J, et al. Development and validation of a
patient satisfaction scale for musculoskeletal care. Arthritis Care Res 1999;12:
96–100.
57. Stewart MA, Wanklin J. Direct and indirect measures of patient satisfaction
with physicians’ services. J Commun Health 1978;3:195–204.
58. Stocki G, Liang MH, Fossel AH, et al. Relative responsiveness of condition-
specific and generic health status measures in degenerative lumbar spinal stenosis.
J Clin Epidemiol 1995;48:1369–78.
59. Streiner DL, Norman GR. Health measurement scales: a practical guide to
their development and use. Oxford, UK: Oxford University Press; 1991.
60. Tanner BA. A multidimensional client satisfaction instrument. Eval Program
Plann 1982;5:161–7.
61. Thompson AG, Sunol R. Expectations as determinants of patient satisfac-
tion: concepts, theory, evidence. Int J Qual Health Care 1995;7:127–41.
62. Torstensen TA, Ljunggren AE, Meen HD, et al. Efficacy and costs of medical
exercise therapy, conventional physiotherapy, and self-exercise in patients with
chronic low back pain: a pragmatic, randomized, single-blinded, controlled trial
with 1-year follow-up. Spine 1998;23:2616–24.
63. Veroff DR, Gallagher PM, Wilson V, et al. Effective reports for health care
quality data: lessons from a cahps demonstration in Washington state. Int J Qual
Health Care 1998;10:555–60.
64. Ware JE, Davies AR. Behavioral consequences of consumer dissatisfaction
with medical care. Eval Prog Plann 1983;6:291–7.
65. Ware JE, Hays RD. Methods for measuring patient satisfaction with specific
medical encounters. Med Care 1988;26:393–402.
66. Ware JE, Davies-Avery A, Stewart AL. The measurement and meaning of
patient satisfaction. Health Med Care Serv Rev 1978; 1:1.
67. Ware JE, Harris WJ, Gandek B, et al. MAP-R for Windows: Multitrait/
Multi-Item Analysis Program: revised User’s Guide. Boston: Health Assessment
Lab; 1997.
68. Ware JE, Snyder MK, Wright WR. Some Issues in the Measurement of
Patient Satisfaction with Health Care. Santa Monica, CA: The Rand Corpora-
tion; 1977.
69. Ware JE. Effects of acquiescent response set on patient satisfaction ratings.
Med Care 1978;16:327–36.
70. Ware JE. How to survey patient satisfaction. Drug Intell Clin Pharma-
col 1981;15:892–9.
71. Ware JE, Snyder MK, Wright R, et al. Defining and measuring patient sat-
isfaction with medical care. Eval Program Plann 1983;6:247–63.
72. Webster D. Final report on the patient satisfaction questionnaire project.
Philadelphia, PA: American Board of Internal Medicine; 1989.
73. Williams B. Patient satisfaction: a valid concept? Soc Sci Med 1994;38:509 –16.
74. Williams B. The meaning of patient satisfaction: an explanation of high
reported levels. Soc Sci Med 1998;47,9:1351–9.
75. Winkler JD, Kanouse DE, Ware JE. Controlling for acquiescence response
set in scale development. J Appl Psychol 1982;67:555–61.
76. Woodward CA, Chambers LW. Guide to Questionnaire Construction and
Question Writing. The Canadian Public Health Association, Ottawa, Ontario,
Canada, 1986.
Address reprint requests to
Pamela L. Hudak
Department of Orthopaedic Surgery S-107
The Hospital for Sick Children
555 University Avenue
Toronto ON M5G 1X8, Canada
E-mail: phudak@sickkids.on.ca
3177Patient Satisfaction Measures •Hudak and Wright
