Questionnaire Development
Item Generation
We developed a questionnaire to examine surgeons' preferences and practices
in the management of femoral neck fractures using consulting surgeons,
epidemiologists, and the previous literature. Ten orthopaedic surgeons in
Canada and the United States and two epidemiologists participated in the
development of the questionnaire.
We initially asked five orthopaedic traumatologists to generate items for
the questionnaire based upon four broad domains: surgeon training and
experience, fracture classification, surgical treatment options, and factors
influencing patient prognosis.
The items generated from discussion with the surgeons and epidemiologists
were augmented by data from a MEDLINE search of articles published from 1969
through 2003.
Orthopaedic traumatologists who were authors of published articles on the
topic provided additional input into potential items for the questionnaire. We
employed a "sample to redundancy" strategy by contacting new
surgeons until no new items for the questionnaire
emerged9.
Ultimately, two additional surgeons provided input.
Pretesting and Validity Assessments
We pretested the questionnaire with an independent group of three
orthopaedic surgeons (with experience in research and trauma) and two
epidemiologists to evaluate the following: (1) does the questionnaire as a
whole appear to adequately address the question of current practice in
treating femoral neck fractures (face validity), and (2) do the individual
questions adequately reflect the four broad domains of surgeon training and
experience, classification of fractures, surgical options for treating hip
fractures, and prognostic factors (content
validity)10. The
surgeons also commented on the clarity and comprehensiveness of the
questionnaire10.
The final questionnaire framed the response options in one of two ways:
either with 5-point Likert scales or nominal (yes-no) scales. A previous
report has shown that closed-ended questions result in fewer incomplete
responses than open-ended
formats11. The
respondents indicated their age and gender, the number of femoral neck
fractures they treated per year, whether they supervised resident trainees,
continent of practice, whether they had fellowship training in trauma or hip
reconstruction, and type of practice (community or academic). Academic
practice was defined as a formal affiliation with a university center.
Respondents expressed their preference for an implant according to patient
age-groups (less than sixty years old, sixty to eighty years old, and more
than eighty years old), general condition of the patients (active or frail),
and fracture classification (Garden type I/II, III, and
IV)12. We provided
the surgeons with several options for internal fixation (compression screw and
side-plate, multiple screws, and intramedullary hip-screw device) and
arthroplasty (unipolar hemiarthroplasty, bipolar hemiarthroplasty, total hip
arthroplasty with cement, and total hip arthroplasty without cement). Surgeons
were asked to rate the technical difficulty of each possible treatment option
using a 5-point scale, with 1 point indicating "not difficult" and
5 points indicating "extremely difficult."
Respondents also provided estimates of the relative impact of implants on
outcomes (the likelihood of revision surgery, mortality, quality of life,
functional recovery, infection, pain, blood loss, and surgical time). Surgeons
responded by noting whether "internal fixation was better,"
"hemiarthroplasty was better," "total hip arthroplasty was
better," "no difference," or "unsure." Finally,
we asked surgeons to rate the difficulty of revising a failed initial implant,
with use of a 5-point Likert scale, with 1 point indicating "not
difficult" and 5 points indicating "extremely
difficult."
Questionnaire Administration
We identified all surgeons who were members of the Orthopaedic Trauma
Association (active members, associate members, international members, senior
members, honorary members, and emeritus members) and European clinics
affiliated with AO International (Davos, Switzerland) through Internet-based
web sites and membership listings in annual meeting proceedings handbooks.
Each surgeon received either a mailed package (a seven-page survey, a
personalized cover letter, and a stamped return envelope) or an e-mail with a
link to the same survey on the Internet with an identification code. At six,
twelve, and eighteen weeks following the initial mailing, we remailed the
questionnaire to all nonresponders. Individual responses remained
confidential, and questionnaire completion was voluntary. Our local ethics
review board approved the study.
Statistical Analysis
We summarized categorical and dichotomous variables with percentages.
Continuous variables were summarized with means and standard deviations.
Whenever the distribution of responses for a particular item in the
questionnaire had multiple empty cells, we collapsed the categories in that
particular item to achieve a more uniform distribution of responses.
To explore whether surgeons who responded in the first mailing had
responses that were different from the surgeons who responded later, we
plotted their preferred implant when considering rates of revision surgery and
statistically compared differences between proportions
(Fig. 1).
We developed a priori hypotheses regarding potentially important predictors
of surgeons' preferences in the treatment of hip fractures. In order of
importance, these included geographic location, surgeon age, academic
practice, fellowship training, and trauma volume.
We performed multivariable linear regression analysis to evaluate
associations between surgeon age, type of practice, fellowship training,
trauma volume, and geographic location and outcome variables including
surgeons' perceptions about the difficulty of alternative procedures and
revision surgeries. We plotted residuals from the regression analyses to
ensure that their distributions were reasonably normal. We report the beta
values and 95% confidence intervals for each independent variable in the
analysis.
Multiple means were compared with analysis of variance with correction for
post hoc testing for significance with use of the least-squares-difference
approach. Proportions were compared with chi-square tests. Significance was
considered to be a p value of <0.01 to account for multiple testing. All
tests were two-tailed.
Characteristics of the Respondents
Of the 442 surgeons who received the questionnaire, 298 (67%) responded,
with the majority (51%) having responded by six weeks. Additional responses
occurred after the second mailing (18%), third mailing (11%), and fourth
mailing (17%). Figure 1 reveals
that responses did not differ across different mailing periods for one
potentially important question: "Which implant is superior in reducing
revision surgery rates?" The data in
Figure 1 are typical—we
did not find significant differences in responses among the respondents at
different mailing periods in any of the key questions. The response rates did
not differ by organization (p = 0.32). The typical respondent was a North
American man over the age of forty years who worked in an academic practice,
supervised residents, had fellowship training in trauma, and worked in a
low-volume center (<100 hip fractures per year), with a more or less 50%
chance that =30% of the patients treated had a displaced femoral neck
fracture (Table I).
Management Preferences by Patient Age
Surgeons preferred the use of internal fixation in younger patients with a
Garden type-III fracture (89%) or Garden type-IV fracture (75%). The implant
of choice was multiple screws, with 56% to 68% of the surgeons endorsing their
use in patients who were less than sixty years old
(Table II). Regression analyses
suggested that surgeons in Europe were more likely to prefer a compression
screw and side-plate over multiple screws than were surgeons in North America
(relative risk, 3.5; 95% confidence interval, 2.1 to 5.1; p < 0.01).
For the treatment of patients between sixty and eighty years old with a
displaced (Garden type-III) femoral neck fracture, 25% of the surgeons (23% of
those in North America compared with 27% of those in Europe) preferred
internal fixation. European surgeons were more likely than were North American
surgeons to prefer a sliding hip screw for fracture fixation (relative risk,
3.4; 95% confidence interval, 1.9 to 5.0; 41% and 12%, respectively; p <
0.01). The most popular procedure among surgeons for treating displaced
(Garden type-IV) fractures in sixty to eighty-year-old patients was bipolar
hemiarthroplasty, with total hip arthroplasty being the least favored of the
arthroplasty options (Table
II).
For the patients who were more than eighty years old, surgeons preferred
arthroplasty (94% preferred it for the treatment of a Garden type-III fracture
and 96%, for a Garden type-IV fracture) to internal fixation, and the majority
(60%) chose unipolar hemiarthroplasty as the procedure of choice
(Table II).
Management Preferences According to Patient Medical Condition
Almost half (49%) of the respondents favored internal fixation for Garden
type-III fractures in active patients
(Table III). However,
arthroplasty was the dominant preference among surgeons (90% to 94% of the
respondents) for the treatment of Garden type-IV fractures regardless of the
medical condition of the patient (Table
III).
We did not identify any significant differences in the perceptions of the
surgeons when arthroplasty training was considered. Although surgeons
practicing in Europe were more likely to prefer the use of a compression screw
with a side-plate than were North American surgeons, there was variability in
the choice of implant for fracture fixation across both continents.
Perceived Difficulty in Initial and Revision Procedures
Surgeons believed intramedullary hip screws were significantly more
difficult to use than either multiple screws or a compression screw and
side-plate (p < 0.05) (Table
IV). Surgeons perceived total hip arthroplasty as a significantly
more difficult procedure than either unipolar or bipolar hemiarthroplasty (p
< 0.01) (Table IV). Across
all implants, surgeons considered revision procedures to be significantly more
difficult than the original procedure (p < 0.001). The increase in the
perceived difficulty ranged from as low as 11% (19% of the surgeons found the
initial surgery difficult compared with 30% who found revision difficult, p
< 0.001) for the revision of multiple screws to 45% (23% found the initial
surgery difficult compared with 68% who found revision difficult, p <
0.001) for the revision of a unipolar hemiarthroplasty inserted with cement;
70% of the surgeons reported that the least difficult procedure was revision
of multiple screws (Table
IV).
Regression analyses evaluating surgeon factors associated with the use of a
compression screw and side-plate or multiple screws demonstrated no
significant associations.
Management Preferences According to Outcome
Surgeons believed that arthroplasty was superior to internal fixation in
reducing the risk of revision surgery (76% and 11%, respectively), decreasing
pain (50% and 17%), improving function (44% and 19%), and improving the
quality of life (30% and 17%) (Table
V). However, most believed that internal fixation was better than
arthroplasty in reducing operating time (70% and 15%, respectively), blood
loss (79% and 10%), infection rates (61% and 4%), and mortality risk (37% and
7%). There was considerable variation among the surgeons as 9% to 45% believed
that there was no difference in the superiority of either internal fixation or
arthroplasty with respect to mortality (45%), quality of life (37%), function
(26%), infection rates (30%), and pain (22%), with 2% to 16% expressing
uncertainty (Table V).
European surgeons were more likely to believe that internal fixation was
superior to arthroplasty in terms of revision surgery (relative risk, 1.5; 95%
confidence interval, 1.1 to 2.2; p = 0.005) and mortality (relative risk, 1.2;
95% confidence interval, 1.04 to 1.4; p = 0.01) than were North American
surgeons.
Key Findings
The results of this survey demonstrated four key findings. (1) Surgeons
preferred the use of internal fixation in younger patients (those who were
less than sixty years old) to treat displaced fractures, and they believed
that older patients (those who were more than eighty years old) and those with
severely displaced (Garden type-IV) fractures should be managed with
arthroplasty. (2) Surgeons varied with respect to their preference for implant
type for both internal fixation and arthroplasty. (3) While the majority of
the respondents preferred arthroplasty for Garden type-III fractures, at least
25% chose internal fixation and this proportion increased with respect to the
treatment of active patients. (4) The opinions differed markedly with respect
to the superiority of internal fixation or arthroplasty when considering
mortality, quality of life, function, infection rates, and pain.
Strengths and Limitations
The strengths of our study include (1) the use of a rigorous process for
the development of the questionnaire items with active surgeon participation;
(2) a comprehensive sampling of surgeons with an interest in caring for trauma
patients; (3) a lack of response bias among later responders; and (4) a good
survey response rate (67%), further limiting nonresponder
bias13,14.
The results are not, however, generalizable to orthopaedic surgeons who are
not members of the Orthopaedic Trauma Association or AO-affiliated centers in
Europe. Given that the Orthopaedic Trauma Association membership is heavily
based in North America, our results also may not reflect the preferences of
the surgeons working in non-academic centers and on other continents. Although
we evaluated the effect of the medical condition of the patient (i.e., frail
compared with active patients), we did not substratify the responses by
patient age. Thus, inferences about the medical condition and surgeon
preference may be limited.
Current Evidence for the Type of Implant in Hip Fracture
Management
Two meta-analyses of randomized trials have provided insight into the
management of displaced femoral neck
fractures15,16.
A previous meta-analysis of randomized trials comparing various methods of
internal fixation of femoral neck fractures found no significant differences
between alternative
implants15;
however, when compression screw and side-plate fixation was specifically
compared with the use of three or more screws (four trials involving 414
patients) with respect to fracture-healing complications, the results showed a
trend favoring compression screw and side-plate fixation (odds ratio, 0.76;
95% confidence interval, 0.47 to
1.25)15. Another
meta-analysis of randomized
trials16 (fourteen
trials involving 1091 patients) presented more compelling evidence for a
difference in outcome across internal fixation techniques: screw and
side-plate constructs performed fivefold better than multiple screws (relative
risk of revision, 1.7 and 9.1, respectively; p < 0.01). Thus, the technical
aspects of the internal fixation may make a difference.
Surgeon variability with regard to optimal approaches to reduce mortality
risk and improve patient function and quality of life also parallels a current
lack of evidence. Arthroplasty is associated with a potentially increased risk
of mortality compared with internal fixation (relative risk, 1.27; 95%
confidence interval, 0.84 to 1.92; p = 0.25) although the confidence interval
is wide16. Pain
relief and function are also similar in patients treated with arthroplasty or
internal fixation (relative risk of no or little pain, 1.12; 95% confidence
interval, 0.88 to 1.35; and relative risk of good function, 0.99; 95%
confidence interval, 0.90 to
1.10)16. Many of
the arguments in favor of arthroplasty are based upon its association with a
decreased risk of revision surgery within the first year compared with that
after internal fixation (77% reduction in risk; 95% confidence interval, 58%
to 87%; p =
0.0003)16. However,
this benefit with respect to the risk of revision occurs at an increased risk
of infection (relative risk, 1.81; 95% confidence interval, 1.16 to 2.85; p =
0.009), greater blood loss (weighted mean difference, 176.4 mL; 95% confidence
interval, 132.4 to 220.4; p < 0.05), and longer surgical time (weighted
mean difference, 29.0 minutes; 95% confidence interval, 23.2 to 34.8 minutes;
p < 0.05)16.
Surgeons who preferred arthroplasty for displaced femoral neck fractures in
sixty to eighty-year-old patients remained discordant in their choice of
procedure. Hemiarthroplasty was more popular than total hip arthroplasty. This
parallels the current lack of evidence with regard to the optimal arthroplasty
strategy in patients with a displaced femoral neck
fracture16.
Current Practice in the Management of Hip Fractures
Our study furthers the generalizability of previous reports by identifying
surgeon opinion across North America and Europe. The variability in the choice
of implant for internal fixation may relate to the perceived difficulty of
different methods of internal fixation as well as to a current lack of
evidence. Surgeons in the current survey preferred fixation with multiple
screws and believed that it was the least difficult type of internal fixation
to perform.
We also found a near consensus (94% to 96%) among respondents in favor of
arthroplasty for the treatment of displaced femoral neck fractures in patients
who were more than eighty years of age. This finding is consistent with
previous
surveys6,8.
Crossman et al. conducted a survey of 233 hospitals in the United Kingdom and
reported that 94% used arthroplasty for the treatment of frail elderly
patients8. Chua et
al. surveyed ninety-nine Canadian surgeons at the Annual Meeting of the
Canadian Orthopaedic Association about their preferences in the management of
hip fractures6. The
great majority of surgeons (96%) preferred arthroplasty in managing elderly
women (those more than eighty years old) with a displaced femoral neck
fracture. Our findings suggest that 92% to 93% of the respondents believed
that hemiarthroplasty (either a unipolar or bipolar prosthesis) is the
treatment of choice in elderly patients who sustain a displaced fracture of
the femoral neck. The infrequent use of total hip arthroplasty as an option
for this patient group may be related to surgeon experience, patient health
status, and geographic variation.
Deciding when evidence is sufficient to conclude that one management
approach is superior to another is a subjective matter and, inevitably,
somewhat arbitrary. Differences in management may reflect ignorance of the
evidence, inadequate evidence of the superiority of one approach over another,
disagreement about the interpretation of the evidence, or differing expertise,
experience, values, and preferences (for instance, differing views of the
importance of early mortality compared with reoperation).
If evidence has established the superiority of one approach, and practice
variability remains, education is necessary. Evidence has not definitively
established the relative merits of the optimal device for either internal
fixation or arthroplasty in the management of displaced femoral neck
fractures, and this lack of evidence is reflected in a variability in
surgeons' views and preferences. Given the limited resources for research, the
most compelling questions are those that both remain unresolved and are a
source of uncertainty or controversy among practicing clinicians. Thus, the
magnitude of the variability in belief and practice concerning the optimal
approaches to internal fixation and arthroplasty recommends this issue as a
research question of international interest.
Note: For a complete list of all members of the International
Hip Fracture Research Collaborative Membership, please see our web site
().
We are grateful to Sheila Sprague for her assistance with the survey
administration.