Abstract
Background: Hip fracture is associated with high mortality among the
elderly. Most patients require surgery, but the timing of the operation
remains controversial. Surgery within twenty-four hours after admission has
been recommended, but evidence supporting this approach is lacking. The
objective of this study was to determine whether a delay in surgery for hip
fractures affects postoperative mortality among elderly patients.
Methods: We conducted a prospective, observational study of 2660
patients who underwent surgical treatment of a hip fracture at one university
hospital. We measured mortality rates following the surgery in relation to the
delay in the surgery and the acute medical comorbidities on admission.
Results: The mortality following the hip fracture surgery was 9%
(246 of 2660) at thirty days, 19% at ninety days, and 30% at twelve months. Of
the patients who had been declared fit for surgery, those operated on without
delay had a thirty-day mortality of 8.7% and those for whom the surgery had
been delayed between one and four days had a thirty-day mortality of 7.3%.
This difference was not significant (p = 0.51). The thirty-day mortality for
patients for whom the surgery had been delayed for more than four days was
10.7%, and this small group had significantly increased mortality at ninety
days (hazard ratio = 2.25; p = 0.001) and one year (hazard ratio = 2.4; p =
0.001). Patients who had been admitted with an acute medical comorbidity that
required treatment prior to the surgery had a thirty-day mortality of 17%,
which was nearly 2.5 times greater than that for patients who had been
initially considered fit for surgery (hazard ratio = 2.3, 95% confidence
interval = 1.6 to 3.3; p < 0.001).
Conclusions: The thirty-day mortality following surgery for a hip
fracture was 9%. Patients with medical comorbidities that delayed surgery had
2.5 times the risk of death within thirty days after the surgery compared with
patients without comorbidities that delayed surgery. Mortality was not
increased when the surgery was delayed up to four days for patients who were
otherwise fit for hip fracture surgery. However, a delay of more than four
days significantly increased mortality.
Level of Evidence: Therapeutic Level II. See Instructions
to Authors for a complete description of levels of evidence.
Hip fracture is the most common major injury in the elderly and an
important cause of mortality and morbidity. The thirty-day mortality is 10%,
and 20% to 30% of patients will die within one year after
surgery1-3.
For the majority of patients, optimum treatment requires surgical management
of the hip fracture, and several studies have demonstrated that a delay in
surgery increases morbidity and mortality in these
patients4-7.
Therefore, it is generally recommended that patients with a hip fracture
undergo surgery within twenty-four hours after
admission8.
Within many health-care systems, the demand for urgent surgery often
exceeds the available resources. This leads to a delay in the surgical
treatment of patients with a hip fracture for nonmedical reasons. The aims of
this study were to compare the effects of delays in the operation for medical
and nonmedical reasons on the mortality and morbidity following hip fracture
and, if possible, to identify an optimal time for surgery in these often frail
patients.
All patients with a hip fracture who are admitted to our university
hospital are entered into a prospective database. This hospital is the sole
provider of orthopaedic trauma care in a city with a population of 650,000.
All adult patients with a fracture of the femoral neck are included in the
database. Isolated femoral head (Pipkin-type) fractures and acetabular
fractures are excluded. An independent audit staff collects data before and
after surgery with use of a detailed data form based on the
"Standardised Audit of Hip Fractures in Europe
(SAHFE)"9 (see
Appendix). Recorded data include age, sex, type of housing, mechanism of
injury, activities of daily living, medical comorbidities, timing and type of
anesthesia and surgery, postoperative complications, length of hospital stay,
and social arrangements on discharge. Preinjury mobility was classified as the
ability to walk outside the home (either alone or accompanied), the ability to
walk inside the home (either alone or accompanied), or the inability to walk.
Cognitive ability was evaluated with use of the minimental test score. The
database is linked to the Office for National
Statistics10, which
allows 100% follow-up for mortality statistics, including those on patients
who die following hospital discharge.
At our hospital, the consultant orthopaedic surgeon recommends the type of
surgery. The majority of intertrochanteric fractures are treated with a
sliding hip screw, and, in general, undisplaced subcapital fractures are
managed with screw fixation and displaced subcapital fractures are treated
with a hemiarthroplasty. All patients are assessed for fitness for surgery by
an anesthetist. Routine preoperative investigations include a full blood-cell
count, measurement of blood urea and electrolyte levels, a radiograph of the
chest, and electrocardiography. Surgery is delayed when a patient requires
treatment for acute medical comorbidities, and the condition of such a patient
is optimized, usually in conjunction with general physicians, prior to the
operation. In this study, early surgery was defined as surgery on the day of
admission or the following day. Surgery performed after this time was
considered to be delayed.
Surgery is undertaken in a laminar flow theater under the supervision of
the consultant orthopaedic surgeon. All patients receive prophylactic
antibiotics and thromboprophylaxis with low-molecular-weight heparin. There
was no surveillance for preoperative or postoperative thromboembolism.
Patients are mobilized the day following the surgery according to their
preinjury mobility status and cognitive state.
At our hospital, one hundred hours are available each week for routine
operations for trauma, but the demand for surgery often outweighs the supply.
Surgery is scheduled according to clinical priority (with life-threatening
injuries treated first, then limb-threatening injuries, etc.) and then
according to the amount of time that the patient has waited. A trauma
coordinator reviews and prioritizes the schedule on a daily basis. This system
remained unchanged during the study period (as did the total number of
admissions per year). The trauma coordinator and the consultants (except for
the senior author [C.G.M.]) were unaware that an observational study of
surgical delay was being carried out.
Statistical Methods
Chi-square tests and the Fisher exact test (two-sided) were used to
determine basic statistical results, and these were followed by Kaplan-Meier
survival analysis to determine the association between mortality and surgical
delay. Patients were censored at the end of the follow-up period. (Survival
probabilities involve the estimation of the time until some event, usually
death or failure of some sort. Some of the patients may not experience the
event because the study ends before they die or because the investigators lose
contact with them partway through the study. Partial information is available
for these patients: it is known that the event occurred [or will occur] at
some time after the date of the last follow-up. We refer to these patients as
"censored
observations."11)
Univariate and multivariate Cox regression analysis was used to evaluate
the effect of fitness for surgery on the association between the delay until
the surgery and the postoperative mortality. This produced hazard ratios,
which are similar to odds ratios and relative risks. The hazard ratio is a
multiplication factor that expresses the comparative risk of the event
occurring between two groups (e.g., a hazard risk of 2.0 equals twice the
risk). All multivariate analyses were adjusted for age and sex. A log-rank
test for trend was used when appropriate. A number-needed-to-treat analysis of
the survival data was performed to determine how many patients would need to
move from one group to another to save one person. The SPSS statistical
program (version 12.0.1; SPSS, Chicago, Illinois) and Microsoft Excel 2000
(Redmond, Washington) were used for the statistical analysis and data storage.
The statistical analysis was undertaken with the supervision of a
statistician.
Over the four-year period from May 8, 1999, to May 7, 2003, 2903 patients
were admitted to our hospital with a hip fracture. The mean age was eighty
years (range, seventeen to 103 years). Sixty-one percent of the patients were
more than eighty years old, and 3% were less than fifty years old. There were
2219 women (76%) and 684 men (24%). Two hundred and forty-three patients were
excluded from the study: ninety-seven patients who had sequential hip
fractures within the study period had the first fracture excluded; six
patients with bilateral, simultaneous hip fracture were excluded to ensure
independent data; and 140 patients who did not have surgery at our hospital
were excluded. Of the 140 patients who did not have surgery at our hospital,
eighty-three patients were managed nonoperatively because they were bedbound
and had minimal pain, had severe chronic medical conditions, or chose not to
have surgical treatment; thirty-seven patients were admitted to our
institution for rehabilitation after surgery at another hospital; and twenty
patients were admitted to our institution with an acute medical comorbidity
that led to death before surgery could be performed. This left a cohort of
2660 patients who underwent hip fracture surgery at our institution, and they
were the subjects of this study. Two hundred and ninety-four of these patients
had a delayed presentation because the patient had ignored the symptoms or a
fracture had not been recognized initially.
Two thousand, one hundred and forty-eight patients (81%) were judged to be
fit for surgery on admission; 982 (46%) of these patients had surgery on the
day of admission or the following day (the early surgery group) and 1166 (54%)
had the operation delayed because of the unavailability of operating room time
(the delayed surgery group). The majority of these patients had surgery within
five days after admission. Table
I lists the length of the delay until surgery and the thirty-day
mortality in this group of patients who were fit for surgery.
Three hundred and eighty-nine patients (15%) were deemed unfit for surgery
within twenty-four hours after admission, as determined by the trauma,
anesthesiology, and medical teams. The most frequent reason for delaying
surgery was an acute medical comorbidity that required treatment prior to the
operation (206 patients; Table
II). Other reasons for delay included investigation for chronic
medical comorbidities and staging of patients in whom a pathological fracture
was suspected because of a previous diagnosis of a malignant lesion. All of
these patients were included in the analysis.
The remaining 123 patients were excluded from the study as there was a
delay in the diagnosis of the fracture, they were admitted with a medical
condition to a medical ward and were later diagnosed as having a fracture of
the neck of the femur, or they were waiting to have an investigation for a
fracture (e.g., magnetic resonance imaging or bone scan).
Complications
The most common postoperative complications were chest infection (223
patients, 8.4%), cardiac failure (124 patients, 4.7%), and urinary tract
infection (103 patients, 3.9%). A deep wound infection developed in
twenty-eight patients (1.1%). There was no significant difference in the rate
or type of complications, including decubitus ulcers, between patients who had
early surgery and those for whom it was delayed for one to four days. There
was also no significant difference in the prevalence of clinically detected
deep vein thrombosis and pulmonary embolism among patients who were fit for
surgery and had no delay (fifteen [1.5%] of 982), patients who were fit for
surgery and had a delay (nineteen [1.6%] of 1166), and those who had a delay
for medical reasons (two [1.0%] of 206) (p = 0.75). Similarly, the
rehabilitation period following surgery, in both the acute ward and the
rehabilitation ward, did not differ significantly among groups. The only
difference in the total length of stay was in the preoperative wait for the
surgery.
Mortality
The thirty-day postoperative mortality for all patients was 9% (246
deaths). The cumulative mortality with time is shown in
Figure 1, which demonstrates a
linear relationship between mortality and time for the first twenty days
following surgery. This relationship is described by the equation y = 0.3x.
After twenty days, the mortality tails off. The mortality was 19% at ninety
days, and it rose to 30% by one year (Fig.
2).
Patients Who Were Fit for Surgery
Eighty-five (8.7%) of the 982 patients who were fit for surgery and had the
operation early died within thirty days, and eighty-five (7.3%) of the 1166
patients who were fit for surgery but had a delay before the surgery for
logistical reasons died within thirty days. This difference was not
significant (chi square = 0.43; p = 0.51). We also found no significant trend
in mortality when we compared patients who were fit for surgery and had a
delay of one to four days (log-rank test for trend statistic = 1.70; p = 0.19)
and those who were fit for surgery and had a delay of more than four days
(log-rank test for trend statistic = 0.52; p = 0.47).
A delay in surgery of one to four days had no adverse effect on the
ninety-day mortality (hazard ratio = 0.99, 95% confidence interval = 0.8 to
1.2; p = 0.91) (Fig. 3) or the
one-year mortality (hazard ratio = 1.1, 95% confidence interval = 0.9 to 1.25;
p = 0.47). However, multivariate analysis of patients with a delay of more
than four days indicated a significant associated increase in mortality at
ninety days (hazard ratio = 2.25, 95% confidence interval = 1.2 to 4.3; p =
0.01) and at one year (hazard ratio = 2.4, 95% confidence interval = 1.45 to
3.99; p = 0.001). It should be noted that this analysis had only 6% power as
only twenty-eight patients had this length of delay. Multivariate analysis
demonstrated that a delay between the fracture and the presentation of more
than one day was not a confounding variable in patients who were fit for
surgery on admission (hazard ratio = 0.9, 95% confidence interval = 0.6 to
1.3; p = 0.57). An inhospital delay of more than four days remained
significant (hazard ratio = 2.2, 95% confidence interval = 1.2 to 4.2; p =
0.01).
A number-needed-to-treat analysis was undertaken with use of the ninety-day
mortality statistics. The analysis indicated that, for there to be one
additional survivor, five patients would have to have had surgery within one
to four days rather than more than four days after admission (95% confidence
interval = 2.7 to 67.0).
Acute Medical Comorbidities
Thirty-six (17%) of the 206 patients with acute medical comorbidities on
admission died within thirty days after the surgery. In contrast, the
thirty-day mortality of the patients who had been declared fit for surgery on
admission was 8% (170 of the 2148 patients died). This difference in mortality
rates is highly significant (hazard ratio = 2.3, 95% confidence interval =
1.62 to 3.33; p < 0.001) (Fig.
4). Multivariate analysis confirmed that this increase in
mortality persisted at ninety days (hazard ratio = 2.1, 95% confidence
interval = 1.6 to 2.7; p < 0.001) and one year (28% compared with 43%;
hazard ratio = 1.72, 95% confidence interval = 1.38 to 2.15; p < 0.001). In
this group of patients with acute medical comorbidities, there was no
significant relationship between the timing of the surgery and the mortality
at thirty days (hazard ratio = 0.68, 95% confidence interval = 0.34 to 1.39; p
= 0.29), ninety days (hazard ratio = 1.16, 95% confidence interval = 0.72 to
1.86; p = 0.54), or one year (hazard ratio = 1.03, 95% confidence interval =
0.68 to 1.58; p = 0.88). Thus, there is no clear optimal time to operate on
these so-called high-risk patients (Fig.
5). Multivariate analysis showed that a delay from the fracture to
presentation of more than one day (294 patients) was a borderline confounding
variable (hazard ratio = 2.1, 95% confidence interval = 1.01 to 4.2; p =
0.048). This means that a delay before presentation was associated with a
significant increase in mortality in patients who were medically unfit for
surgery.
Patients who sustain a hip fracture have a significantly higher mortality
rate in the year following the fracture compared with age-matched controls
without a fracture1.
Whether a delay in surgery contributes to this increased mortality remains
controversial. Several studies have demonstrated that delaying surgery for
more than twenty-four hours increases
mortality4-7.
Other studies have demonstrated no significant difference in the mortality of
patients in whom surgery was delayed by up to three
days9. There have
even been reports suggesting that operating within twenty-four hours may
increase
mortality10. The
majority of these studies have been based on retrospective reviews of patient
records or
databases12,13,
which often did not include the reasons for the delay in surgery. Most studies
involved relatively small numbers of patients, whereas, in the larger studies,
the data were often collected from several different centers over a long
period of time.
The current study has several advantages over previous reports. The data
were collected prospectively by an audit staff; the database was linked to the
British Office for National Statistics, which allowed a 100% follow-up for
mortality; the study included patients treated at a single center over a
relatively short period of time; and data on more than 2500 patients were
collected. Therefore, the accuracy and statistical power of the study allowed
us to draw some firm conclusions.
We demonstrated that a delay of up to four days before hip fracture surgery
in patients who are considered fit for anesthesia and an operation (i.e.,
those with no acute medical comorbidities) does not affect mortality,
morbidity, or the length of the postoperative hospital stay. A small number of
patients had a delay of more than four days before the surgery. Those delays
occurred during severe peaks of admissions (during winter) or after patients
had initially declined surgery but later changed their minds. Our analysis
indicated that patients with a delay of more than four days have a
significantly increased risk of dying by ninety days (hazard ratio = 2.25) and
one year (hazard ratio = 2.4). It should be recognized that only twenty-eight
patients had the surgery delayed beyond four days so this aspect of the study
has limited statistical power. However, the number-needed-to-treat analysis
indicated that only five patients needed to undergo surgery within four days
after admission to save one additional life. We believe that patients with a
hip fracture must be given priority to avoid this length of delay. Health-care
systems vary from country to country, but responsible public health bodies
should be aware of this problem and ensure that adequate facilities are
available for the timely management of these elderly, often frail, patients.
In addition, patients who decline surgery should be informed of the potential
risks associated with a delay in surgery of more than four days.
The treatment of patients admitted with a hip fracture and an acute medical
comorbidity, such as a chest infection, is a challenge. The current study
confirmed that this group has a high risk of mortality within thirty days
(17%; hazard ratio = 2.3) and at one year (43%; hazard ratio = 1.72). In
general, surgery should proceed when the patient's condition is optimal;
however, this can be a difficult decision that needs to be individualized and
should involve the anesthetist, general (or specialist) physician, and
surgeon. Our study gives no indication of the optimal time for surgery in this
group of patients but indicates that a delay from the fracture to the hospital
presentation is associated with increased mortality.
The limitations of this study must be recognized. It was an observational
study, not a randomized controlled trial, so it lacks the accuracy that could
be achieved with a protocoldriven study. There was no a priori protocol for
determining which patients were unfit for surgery and anesthesia, a judgment
that will always vary between clinicians. However, the striking difference in
mortality between the patients deemed to be fit for surgery and those
considered to be unfit for surgery indicates that current clinical methods are
successful in identifying patients who are at increased risk of death
following hip fracture surgery. This important subgroup of patients requires
separate analysis in future studies, and any national audit should also
include a separate analysis of these patients, who could have an important
effect on the observed mortality in individual trauma units.
In conclusion, we performed a large, prospective, observational study of
patients treated surgically for a hip fracture, with complete follow-up for
mortality statistics. The study indicated that a delay in surgery of up to
four days in patients without an acute medical comorbidity does not increase
postoperative mortality, morbidity, or duration of the rehabilitation
following surgery. We do not advocate such delays, as we believe that they
cannot be justified on humanitarian grounds; however, when surgical facilities
are unable to cope with demand, a short delay before surgery will not
significantly increase morbidity and mortality in this group of patients. A
delay of more than four days in patients who are fit for surgery significantly
increases mortality and must be avoided.
The Standardised Audit of Hip Fractures in Europe (SAHFE) form is available
with the electronic versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?
Note: The authors thank the following clinicians for allowing
their patients to be included in this study: Mr. N. Badhe, Mr. N.D. Downing,
Mr. D.M. Hahn, Mr. M. Hatton, Mr. B.J. Holdsworth, Mr. C.J. Howell, Mr. J.B.
Hunter, Mr. P.J. James, Mr. A.R. Manktelow, Mr. J.A. Oni, Mr. P.J. Radford,
Ms. B.E. Scammell, and Mr. E.P. Szypryt. They also thank Dr. Sarah Armstrong,
Statistician (University of Nottingham), and Christopher White, Audit Clerk in
Trauma and Orthopaedics, for their assistance with this study.
Roberts SE, Goldacre MJ. Time trends and
demography of mortality after fractured neck of femur in an English
population, 1968-98: database study. BMJ.2003;327:
771-5.327771
2003
[PubMed][CrossRef]
Goldacre MJ, Roberts SE, Yeates D.
Mortality after admission to hospital with fractured neck of femur: database
study. BMJ.2002;325:
868-9.325868
2002
[PubMed][CrossRef]
Miller CW. Survival and ambulation
following hip fracture. J Bone Joint Surg Am.1978;60:
930-4.60930
1978
[PubMed]
Hamlet WP, Lieberman JR, Freedman EL,
Dorey FJ, Fletcher A, Johnson EE. Influence of health status and the timing of
surgery on mortality in hip fracture patients. Am J Orthop.1997;26:
621-7.26621
1997
[PubMed]
Villar RN, Allen SM, Barnes SJ. Hip
fractures in healthy patients: operative delay versus prognosis. Br Med
J (Clin Res Ed).1986;293:
1203-4.2931203
1986
[CrossRef]
Bredahl C, Nyholm B, Hindsholm FB,
Mortensen JS, Olesen AS. Mortality after hip fracture: results of operation
within 12 h of admission. Injury.1992;23:
83-6.2383
1992
[PubMed][CrossRef]
Zuckerman JD, Skovron ML, Koval KJ,
Aharonoff G, Frankel VH. Postoperative complications and mortality associated
with operative delay in older patients who have a fracture of the hip.
J Bone Joint Surg Am.1995;77:
1551-6.771551
1995
[PubMed]
Fractured neck of femur. Prevention and management. Summary and
recommendations of a report of the Royal College of Physicians. J R
Coll Physicians Lond.1989;23:
8-12.238
1989
Grimes JP, Gregory PM, Noveck H, Butler
MS, Carson JL. The effects of time-to-surgery on mortality and morbidity in
patients following hip fracture. Am J Med.2002;112:
702-9.112702
2002
[PubMed][CrossRef]
Kenzora JE, McCarthy RE, Lowell JD,
Sledge CB. Hip fracture mortality. Relation to age, treatment, preoperative
illness, time of surgery, and complications. Clin Orthop.1984;186:
45-56.18645
1984
[PubMed]