ORIGINAL ARTICLE
Independent Preoperative Predictors of Outcomes in
Orthopedic and Vascular Surgery
The Influence of Time Interval Between an Acute Coronary Syndrome or Stroke
and the Operation
Robert D. Sanders, FRCA,∗ Alex Bottle, PhD,† Simon S. Jameson, MRCS,‡ Abdul Mozid, MRCP,§
Paul Aylin, FFPHM,† Lliam Edger, FRCA,|| Daqing Ma, MD, PhD,∗ Mike R. Reed, FRCS,‡
Matthew Walters, FRCP,¶ Kennedy R. Lees, FRCP,¶ and Mervyn Maze, FRCP, FRCA, FMedSci#
Objective: To identify independent preoperative predictors of outcome for
total hip or knee replacement (THKR) and abdominal aortic aneurysm (AAA)
repair, including the importance of the time interval between an acute coronary
syndrome (ACS) or stroke and surgery.
Background: Present guidelines do not advocate a prolonged delay after ACS
though recent data suggest delaying operations by 8 weeks. There is a lack of
data on when to schedule surgery following stroke.
Methods: The Hospital Episode Statistics database was analyzed for elective
admissions for THKR and AAA surgery between 2006–2007 and 2009–2010.
Patient factors influencing mortality, length of stay, and readmission rates were
identified by logistic regression.
Results: A total of 414,985 THKRs (mortality: 0.2%) and 14,524 AAA repairs
(mortality: 3.5%) were included. Heart failure, renal failure, liver disease,
peripheral vascular disease, and non-atrial fibrillation arrhythmia increased
the odds of mortality for both surgeries. Among other factors, previous ACS
and stroke predicted mortality after THKR but not AAA surgery. Compared
with more delayed surgery, THKR surgery performed within 6 months of an
ACS (odds ratio [OR]: 3.81; 95% confidence interval [CI]: 1.55–9.34), but
From the ∗ Magill Department of Anaesthetics, Intensive Care and Pain Medicine,
Imperial College London, Chelsea and Westminster Hospital, London; †School
of Public Health, Imperial College London; ‡Department of Orthopaedics,
Northumbria NHS Foundation Trust, Woodhorne Lane, Ashington, UK; §Department of Cardiology, The London Chest Hospital, London; ||Department of
Anaesthetics and Intensive Care Medicine, Central Middlesex Hospital, London; ¶Institute of Cardiovascular and Medical Sciences, University of Glasgow,
Lanarkshire, UK; and #Department of Anesthesia and Perioperative Care, University of California San Francisco.
Disclosure: No conflict of interest is reported by the authors. All authors have
completed the Unified Competing Interest form at http://www.icmje.org/coi
disclosure.pdf (available on request from the corresponding author) and declare: no financial relationships with any organizations that might have an
interest in the submitted work in the previous 3 years, no other relationships or
activities that could appear to have influenced the submitted work. This work
was funded by an AAGBI/Anaesthesia Department Project Grant administered
by the National Institute of Academic Anaesthesia and awarded to Dr Sanders.
The Dr Foster Unit at Imperial is largely funded via a research grant by Dr
Foster Intelligence, an independent health care information company and joint
venture with the NHS Information Centre. The Dr Foster Unit is affiliated with
the Centre for Patient Safety and Service Quality at Imperial College Healthcare NHS Trust that is funded by the National Institute of Health Research. The
Department of Primary Care and Public Health is grateful for support from the
NIHR Biomedical Research Centre funding scheme.
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions of
this article on the journal’s Web site (www.annalsofsurgery.com).
Reprints: Robert D. Sanders, FRCA, Magill Department of Anaesthetics, Intensive Care and Pain Medicine, Imperial College London, Chelsea and Westminster Hospital, 369 Fulham Road, London, SW10 9NH, UK. E-mail:
[email protected].
C 2012 by Lippincott Williams & Wilkins
Copyright ISSN: 0003-4932/12/25505-0901
DOI: 10.1097/SLA.0b013e31824c438d
Annals of Surgery r Volume 255, Number 5, May 2012
not stroke, increased the odds of mortality. The effect of ACS persisted up to
12 months (OR: 1.99; 95% CI: 1.02–3.88) and was not altered by exclusion of
patients who received percutaneous coronary intervention or coronary artery
bypass grafting for treatment of their ACS.
Conclusions: Previous stroke and ACS increased the odds of perioperative
mortality from THKR but not AAA surgery; THKR surgery conducted up to
12 months after an ACS was associated with increased mortality.
(Ann Surg 2012;255:901–907)
D
etermining the appropriateness and timing of surgery is dependent on many factors; however, perioperative risk is the most
important. Timing of elective surgery after vascular events such as
acute coronary syndrome (ACS) or stroke poses an interesting problem. Thirty years ago, Goldman et al1 observed that surgery conducted within 6 months of myocardial infarction was associated with
increased perioperative cardiac risk. Further studies have attempted
to clarify the relationship between timing of surgery relative to myocardial infarction and perioperative risk; however, these studies have
either been underpowered to detect temporal effects2 or have studied a heterogeneous group of scheduled and unscheduled procedures
with limited adjustment for other prognostic variables.3 Focus specifically on elective operations is of particular importance, as there is
the opportunity to delay surgery. Current guidelines state that surgery
should be delayed 4 to 6 weeks after at ACS,4 largely dismissing
the data given by Goldman et al because of changing diagnostic criteria for management of and outcomes after coronary events over
the 30-year period. We sought contemporary evidence to substantiate
whether there is a more prolonged phase (as suggested by Goldman et
al1 and recently by Livhits et al3 ) after an ACS during which elective
surgery is a high risk. The effect of recent stroke on perioperative risk
has been less studied and remains unclear, hence a paucity of information is available to guide clinicians when deciding on the timing
of elective surgery after major vascular events. We aimed to clarify
these important questions by using a large registry of routinely collected clinical data to study the influence of recent ACS and stroke on
perioperative risk in patients undergoing a series of common, elective
surgical procedures.
METHODS
Following local research ethics committee and Section 251
(formerly Section 60) National Information Governance Board for
Health and Social Care approval, we extracted all nonduplicate elective (planned) admissions in England and Wales for the financial years
April 1, 2006, to March 31, 2010, with valid age, sex, length of stay
(LOS), and patient ID from the Hospital Episode Statistics database.
This is an administrative database covering all admissions to NHS
(public) hospitals in England, including private patients treated in
www.annalsofsurgery.com | 901
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 255, Number 5, May 2012
Sanders et al
these hospitals. Diagnostic information is coded using the International Classification of Diseases, system version 10 (ICD-10); the 13
secondary diagnoses record comorbidities and complications. Its 12
procedure fields use the classification of interventions and procedures
of the Office of Population Censuses and Surveys (OPCS), unique to
the UK.5 Admissions (“spells”) ending in transfer to another hospital
were linked together to form “superspells” (we will refer to the continuous inpatient period, including any transfers, as one admission). The
principal procedure of interest in this study was a total hip replacement [OPCS codes W371, W381, and W391; resurfacing (W581 with
accompanying Z843) procedures were also included], total knee replacement (W401, W411 and W421), or abdominal aortic aneurysm
(AAA) repair (L183–6, L193–6, L203–6, L213–6, L271–2, L275–6,
L281–2, L285–6). The first operation per patient was selected and
the rest discarded because the postoperative risk may be higher for
reoperations.
We hypothesized that preoperative factors would differentially
influence intermediate- and high-risk surgery2,4,6 and hence clinicians should judge a procedure-specific risk.4 For intermediate risk
surgery, we selected elective total hip or knee replacements (THKRs).
Preliminary analysis suggested that the population of patients undergoing these procedures had similar demographics and incurred similar perioperative mortality, supporting their grouping. For high-risk
surgery, elective AAA repair was selected.
Preoperative “Vascular events,” including stroke, myocardial
infarction, and unstable angina, were defined as conditions necessitating hospital admission and therefore generating a primary diagnosis
code with known timing. We therefore looked retrospectively for admissions within the previous 10 years with the primary diagnosis of
stroke (I61–I64 or I66), myocardial infarction (I21, I22), and unstable
angina (I200) before their index procedure until 1996. We looked in
every spell in the preoperative admissions in case the stroke, myocardial infarction, or unstable angina occurred after transfer (ie, was in
spell >1). The time in days from the stroke, myocardial infarction,
or unstable angina admission to the index operation was noted. If a
patient had more than 1 “vascular event,” then the date of the most
recent was noted.
Comorbidities not meeting the criteria for a “vascular event,”
such as diabetes mellitus, liver disease, or renal failure, were identified from the secondary diagnosis codes on admission for the surgical
procedure (supplementary Table 1, available at http://links.lww.com/
SLA/A231). Age was analyzed as a continuous variable (“risk per
additional single year”), though in Table 1 a threshold is displayed
for illustrative purposes. The reference gender was male. We derived
a set of comorbidity variables representing risk factors for postoperative death using an ICD-10 version of the Charlson comorbidity
index7 and other risk indices2,8,9 as a starting point and augmenting their components where necessary. However, rather than using
the Charlson comorbidity index to adjust for perioperative risk, we
adjusted for the impact of each preoperative factor by logistic regression to assign appropriate weights to the comorbidities in our study
population. Our primary endpoint was the adjusted odds ratio (OR)
of perioperative mortality (defined as inpatient death within 30 days
of surgery); secondary endpoints were 2 surrogates of perioperative
morbidity: prolonged LOS (above the national upper quartile for the
4 years of index procedures combined: 12 days for AAA repair and
7 days for THR and TKR) and emergency (unplanned) readmission
for any reason within 28 days of discharge.
We then grouped myocardial infarction and unstable angina to
form ACS and analyzed how the preoperative timing of ACS or stroke
influenced perioperative outcomes. Patients with a defined preoperative event were extracted and grouped according to the preoperative
event type (stroke or ACS). Each cohort (based on the defined preoperative event) was then analyzed to define the adjusted risk imposed
by the timing of the condition preoperatively using the outcomes
mentioned earlier. The ACS group was analyzed with and without
exclusion of all percutaneous coronary intervention (PCI) for treatment of the ACS (OPCS codes K49, K50, K75), PCI and/or coronary
TABLE 1. Patient Demographics for THKR and AAA Surgery
THKR
Absent
Operation Risk Factor
Age (>70 years)
Male
Hypertension
Atrial fibrillation
Non-AF arrhythmia
Valvular heart disease
Heart failure
Respiratory disease
Diabetes mellitus
Renal failure
Cancer
Liver disease
Peripheral vascular disease
Stroke
Stable angina
Myocardial infarction
Unstable angina
ACS
AAA
Present
Case Total
(%)
Mortality
Rate:
Number
(%)
216740 (52.2)
243598 (58.7)
250605 (60.4)
396084 (95.4)
412746 (99.5)
411977 (99.3)
411852 (99.2)
374278 (90.2)
377354 (90.9)
411340 (99.1)
411647 (99.2)
414128 (99.8)
412045 (99.3)
411924 (99.3)
414881 (99.8)
408525 (98.4)
409494 (98.7)
403753 (97.3)
122 (0.1)
374 (0.2)
388 (0.2)
679 (0.2)
802 (0.2)
796 (0.2)
689 (0.2)
699 (0.2)
698 (0.2)
673 (0.2)
800 (0.2)
810 (0.2)
798 (0.2)
811(0.2)
829 (0.2)
791 (0.2)
792 (0.2)
761 (0.2)
Absent
Case Total
(%)
Mortality
Rate:
Number
(%)
198245 (47.8)
171387 (41.3)
164380 (39.6)
18901 (4.6)
2239 (0.5)
3008 (0.7)
3133 (0.8)
40707 (9.8)
37631 (9.1)
3645 (0.9)
3338 (0.8)
857 (0.2)
2940 (0.7)
3061 (0.7)
104 (0.0)
6460 (1.6)
5491 (1.3)
11232 (2.7)
707 (0.4)
455 (0.3)
441 (0.3)
150 (0.8)
27 (1.2)
33 (1.1)
140 (4.5)
130 (0.3)
131 (0.3)
156 (4.3)
29 (0.9)
19 (2.2)
31 (1.1)
18 (0.6)
0 (0.0)
38 (0.6)
37 (0.7)
68 (0.6)
Present
Case Total
(%)
Mortality
Rate:
Number
(%)
Case Total
(%)
Mortality
Rate:
Number
(%)
4862 (33.5)
2043 (14.1)
6554 (45.1)
12789 (88.1)
14348 (98.8)
14274 (98.3)
14030 (96.6)
12298 (84.7)
12873 (88.6)
13791 (9.5)
13867 (95.5)
14413 (99.2)
13017 (89.6)
14228 (98.0)
14520 (100.0)
13699 (94.3)
13976 (96.2)
13255 (91.3)
114 (2.3)
84 (4.1)
225 (3.4)
412 (3.2)
498 (3.5)
497 (3.5)
435 (3.1)
416 (3.4)
453 (3.5)
384 (2.8)
482 (3.5)
489 (3.4)
406 (3.1)
495 (3.5)
511 (3.5)
484 (3.5)
488 (3.5)
464 (3.5)
9662 (66.5)
12481 (85.9)
7970 (54.9)
1735 (11.9)
176 (1.2)
250 (1.7)
494 (3.4)
2226 (15.3)
1651 (11.4)
733 (5.0)
657 (4.5)
111 (0.8)
1507 (10.4)
296 (2.0)
4 (0.0)
825 (5.7)
548 (3.8)
1269 (8.7)
397 (4.1)
427 (3.4)
286 (3.6)
99 (5.7)
13 (7.4)
14 (5.6)
76 (15.4)
95 (4.3)
58 (3.5)
127 (17.3)
29 (4.4)
22 (19.8)
105 (7.0)
16 (5.4)
0 (0.0)
27 (3.3)
23 (4.2)
47 (3.7)
ACS indicates acute coronary syndrome.
902 | www.annalsofsurgery.com
C 2012 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 255, Number 5, May 2012
artery bypass grafting (CABG) (OPCS codes K49, K50, K75, K40–
46), and PCI with coronary stent placement (OPCS code K75).
STATISTICAL ANALYSIS
Logistic regression models were fitted for mortality with the
following patient factors: age, sex, Carstairs deprivation populationweighted fifth (the population divided into 20 percentiles),10 comorbidity and a “time” flag for previous stroke, unstable angina or myocardial infarction. Unstable angina and myocardial infarction were
also grouped to form ACS. Other comorbidities, which were derived
using the secondary diagnoses in the index admission, were included
as indicator variables if they were recorded in at least 30 patients and
sometimes had to be dropped even with less than 30 to enable the
model to converge. To avoid the problems of stepwise model selection
procedures, we retained all candidate variables even if not significant
and only removed them because of nonconvergence. To elicit any
possible effect of the timing of the stroke and ACS, we identified only
those patients with a prior event (in the previous 10 years) and fitted
the lag time as a continuous variable. We also looked for a nonlinear
effect by categorizing the lag. To identify a clinically important “window” during which surgery should be avoided, we set thresholds of 6
(based on previous studies1–3 ) and 12 months (based on advised termination of dual antiplatelet therapy for ACS11 ) to the index surgical
procedure. We tried fitting 2-level models to account for the clustering
of patients within hospitals but found minimal clustering and therefore present the results from the single-level regression models. All
analyses were performed using SAS v9.2 (SAS Institute Inc, Cary,
NC). Data are tabulated and reported as adjusted ORs. Where of interest data are presented as “number needed to harm” (NNH) that is
the inverse of the attributable risk (risk in the people exposed minus
the risk in the people not exposed). In this context, exposure refers to
previous ACS. The data are presented in accordance with STROBE
(STrengthening Reporting of Observational studies in Epidemiology)
guidance.
RESULTS
Cohort Demographics
A total of 414,985 THKRs and 14,524 AAA repairs were
performed between 2006–2007 and 2009–2010. Summary data are
available in Table 1. Mortality was 0.2% for THKR with 829 deaths
(0.21% for total hip replacement with 401 deaths and 0.19% for
total knee replacement with 428 deaths) and 3.5% for AAA repair
with 511 deaths. For THKR, there were 431 causes for readmission
(3-character ICD-10 codes); 38.9% were attributable to surgical reasons and 11.1% were for cardiovascular causes (ACS 1.3% and stroke
0.5%). For AAA surgery, there were 185 causes of readmission;
34.9% of readmissions were for surgical reasons while 14.6% were
secondary to cardiovascular disorders (ACS 3.9% and stroke 1.0%).
Perioperative Risk Factors
Heart failure, liver failure and renal failure, peripheral vascular
disease, and non-atrial fibrillation arrhythmia were common factors
that increased the odds of perioperative death, prolonged LOS, and
readmission for both THKR and AAA surgery (Table 2). For THKR,
but not AAA surgery, the odds of perioperative mortality were increased by ACS and stroke, as well as diabetes mellitus, hypertension, atrial fibrillation, valvular heart disease, cancer, and respiratory
disease (Table 2). Similar factors, and also age, affected risk of prolonged LOS and readmission rates from THKR (Table 2). In addition
to the common factors, perioperative mortality after AAA repair was
predicted by increased age (Table 2). Although mortality after AAA
repair was not affected by prior ACS or stroke, the odds of pro
C 2012 Lippincott Williams & Wilkins
Timing Surgery Postvascular Event
longed LOS were increased by previous stroke, whereas the odds of
readmission were affected by ACS (Table 2).
Perioperative Risk Imposed by the Timing of Prior
ACS and Stroke
We analyzed the effect of timing of ACS or stroke for either type of surgery based on continuous methods but did not find a
linear relationship between the timing of vascular event and perioperative mortality, prolonged LOS, and readmission rates. Analysis by
dichotomous methods revealed that in patients undergoing THKR,
ACS 6 months before an operation increased the risk of perioperative mortality (Table 3). This effect persisted even after removing all
patients who had a PCI (OR: 3.14, P = 0.032), PCI and/or CABG
(OR: 3.35, P = 0.024), or just those with coronary stent placement
(OR: 4.1, P = 0.003) for treatment of their ACS. If the threshold
was reset at 12 months [11 of 967 (1.13%) patients died], recent ACS
still increased the odds of perioperative death. Removal of all patients
who had PCI (OR: 2.02, P = 0.056), PCI and/or CABG (OR: 1.6,
P = 0.255), or just a coronary stent (OR: 2.20, P = 0.022) as treatment of their ACS did little to affect the point estimate of the results.
To quantify the effect of prior ACS on postoperative mortality, we
calculated the NNH. Overall prior ACS exerted a modest effect on
mortality from THKR (NNH 1:700); however, if surgery was conducted within 6 months (NNH 1:64) or 12 months (NNH 1:183) of
an ACS, the risk was augmented. Patients who had an ACS in the
6 months before AAA repair had a non-significantly increased odds
of prolonged LOS (P = 0.08) and readmission (P = 0.25) suggesting that this may increase the odds of perioperative morbidity, but
there was no discernable effect on perioperative mortality (Table 3).
ACS within the 12 months before AAA repair was associated with
a significantly higher postoperative readmission rate, supporting an
effect on morbidity (Table 3). Stroke in the 6 months before THKR or
AAA surgery did not influence mortality, LOS, or readmission rates
(Table 4).
DISCUSSION
The common perioperative risk factors for THKR and AAA
surgery include heart, renal, and liver failure; non-atrial fibrillation
arrhythmia; and peripheral vascular disease. Some factors, such as
preoperative ACS and stroke, relate in a more procedure-specific
manner only affecting mortality from THKR surgery. Furthermore,
for THKR surgery, the timing of ACS influences the odds of perioperative death; thresholds of 6 and 12 months showed a prolonged
increase in risk after an ACS. This effect persisted even after removal
of patients whose ACS treatment involved placing a coronary stent
or any PCI or any PCI and/or CABG. Although previous ACS and
stroke did not predict mortality from AAA repair, they influenced perioperative morbidity. In particular, patients undergoing AAA repair
within 12 months of an ACS were at an increased risk of emergency
readmission to hospital (Table 3).
COMMON PERIOPERATIVE RISK FACTORS
The common factors that influenced perioperative mortality
are consistent with current knowledge1,2,8 though liver disease, a
potent contributor to perioperative mortality,9,12 has been excluded
from recent perioperative risk scoring systems.1,2,8 Although liver
disease only affected 0.8% patients for AAA surgery and 0.3% for
THKR (Table 1), it is important to identify these individuals as high
risk to allow provision of suitable perioperative care.
PERIOPERATIVE RISK FACTORS FOR AAA SURGERY
Our data for AAA repair are consistent with the suboptimal predictive value of the revised cardiac risk index for predicting outcomes
www.annalsofsurgery.com | 903
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
904 | www.annalsofsurgery.com
2.50 (2.31–2.71)
1.44 (1.40–1.47)
2.14 (1.98–2.31)
1.73 (1.14–2.63)
1.38 (1.31–1.46)
1.60 (1.51–1.70)
1.54 (1.47–1.60)
<0.001
<0.001
0.002
<0.001
<0.001
<0.001
<0.001
0.043
0.040
<0.001
<0.001
6.79 (5.48–8.42)
1.41 (1.16–1.71)
1.37 (1.13–1.67)
8.89 (7.29–10.85)
2.43 (1.65–3.58)
8.34 (4.77–14.56)
2.18 (1.48–3.2)
1.64 (1.02–2.65)
1.44 (1.02–2.05)
1.93 (1.36–2.75)
1.73 (1.33–2.25)
1.58 (1.54–1.62)
3.40 (3.16–3.65)
1.67 (1.55–1.80)
2.32 (2.01–2.69)
1.52 (1.40–1.64)
1.57 (1.45–1.69)
0.002
1.84 (1.26–2.69)
0.85 (0.85–0.86)
1.32 (1.29–1.34)
1.13 (1.11–1.14)
2.35 (2.28–2.43)
1.60 (1.46–1.76)
OR (95% CI)
0.259
<0.001
0.027
<0.001
<0.001
P
LOS
THKR
0.95 (0.87–1.04)
0.58 (0.50–0.67)
1.18 (1.02–1.36)
1.54 (1.26–1.87)
3.44 (2.29–5.16)
OR (95% CI)
AF indicates atrial fibrillation.
Age
Sex
Hypertension
Atrial fibrillation
Non-AF
arrhythmia
Valvular heart
disease
Heart failure
Respiratory
Disease
Diabetes mellitus
Renal failure
Cancer
Liver disease
Peripheral
vascular disease
Stroke
Stable angina
Myocardial
infarction
Unstable angina
ACS
Operation Risk
Factor
Mortality
1.61 (1.48–1.76)
1.48 (1.38–1.58)
1.52 (1.35–1.71)
1.19 (0.58–2.46)
1.28 (1.17–1.40)
<0.001
0.011
<0.001
<0.001
<0.001
1.16 (1.11–1.20)
1.22 (1.08–1.37)
1.28 (1.13–1.45)
1.65 (1.31–2.09)
1.35 (1.19–1.53)
<0.001
<0.001
<0.001
<0.001
<0.001
1.07 (0.93–1.23)
<0.001
1.23 (1.09–1.39)
1.33 (1.28–1.38)
0.95 (0.94–0.96)
0.83 (0.81–0.85)
1.04 (1.01–1.07)
1.21 (1.15–1.28)
1.32 (1.13–1.53)
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
OR (95% CI)
P
P
<0.001
<0.001
<0.001
0.638
<0.001
<0.001
0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.323
<0.001
<0.001
0.005
<0.001
<0.001
Readmission
TABLE 2. Effect of Preoperative Risk Factors on Outcomes From THKR and AAA Surgery
1.22 (0.78–1.91)
0.98 (0.71–1.34)
1.52 (0.90–2.58)
0.82 (0.50–1.24)
1.00 (0.75–1.33)
5.43 (4.31–6.83)
1.12 (0.75–1.67)
5.69 (3.32–9.73)
2.13 (1.68–2.69)
3.54 (2.65–4.73)
1.15 (0.91–1.46)
0.96 (0.54–1.73)
1.05 (1.03–1.06)
1.05 (0.81–1.35)
1.04 (0.87–1.26)
1.20 (0.94–1.53)
2.02 (1.11–3.7)
OR (95% CI)
Mortality
0.376
0.878
0.118
0.348
0.993
<0.001
0.592
<0.001
<0.001
<0.001
0.249
0.894
<0.001
0.720
0.656
0.143
0.022
P
AAA
1.05 (0.85–1.30)
0.97 (0.84–1.12)
1.37 (1.05–1.78)
0.96 (0.80–1.15)
1.01 (0.88–1.14)
3.26 (2.79–3.81)
1.07 (0.88–1.30)
1.62 (1.06–2.46)
1.75 (1.55–1.97)
1.78 (1.46–2.16)
1.19 (1.06–1.32)
1.08 (0.80–1.44)
1.02 (1.02–1.03)
1.80 (1.61–2.00)
1.11 (1.02–1.21)
1.80 (1.61–2.02)
2.21 (1.61–3.03)
OR (95% CI)
LOS
0.660
0.682
0.021
0.639
0.932
<0.001
0.496
0.025
<0.001
<0.001
0.002
0.629
<0.001
<0.001
0.013
<0.001
<0.001
P
1.35 (1.03–1.76)
1.28 (1.07–1.54)
0.89 (0.59–1.34)
3.55 (0.37–34.52)
1.12 (0.89–1.40)
1.04 (0.87–1.24)
1.17 (0.90–1.52)
1.07 (0.82–1.39)
1.06 (0.53–2.12)
1.25 (1.05–1.49)
1.47 (1.11–1.94)
1.32 (1.14–1.52)
1.47 (1.02–2.12)
0.93 (0.87–0.99)
1.09 (0.93–1.27)
1.17 (1.04–1.31)
1.36 (1.15–1.59)
0.95 (0.57–1.58)
OR (95% CI)
P
0.027
0.008
0.583
0.275
0.352
0.670
0.238
0.624
0.876
0.011
0.006
<0.001
0.039
0.026
0.311
0.007
<0.001
0.835
Readmission
Sanders et al
Annals of Surgery r Volume 255, Number 5, May 2012
C 2012 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
C 2012 Lippincott Williams & Wilkins
0.780
0.003
0.045
3.81 (1.55–9.34)
1.99 (1.02–3.88)
P
1 (0.99–1.01)
OR (95% CI)
1.18 (1.02–1.35)
1.22 (0.93–1.59)
1 (1–1)
OR (95% CI)
LOS
THKR
0.024
1.21 (0.98–1.49)
1.28 (0.87–1.89)
1 (1–1)
<0.001
0.152
OR (95% CI)
P
Readmission
0.082
0.207
0.032
P
1.18 (0.54–2.61)
0.85 (0.26–2.84)
1 (0.99–1.01)
OR (95% CI)
Mortality
Timing of operation
after stroke:
continuous
Timing of operation
after stroke:
6-month threshold
Operation Risk
Factor
0.197
0.900
0.87 (0.1–7.57)
P
1.01 (1–1.02)
OR (95% CI)
Mortality
1.35 (0.91–1.99)
1 (1–1)
OR (95% CI)
LOS
THKR
0.133
0.290
P
0.86 (0.46–1.63)
1 (0.99–1)
OR (95% CI)
Readmission
0.647
0.315
P
0.59 (0.07–4.75)
1 (0.98–1.01)
OR (95% CI)
Mortality
TABLE 4. The Effect of the Timing of Surgery After Stroke on Outcomes From THKR and AAA Surgery
Timing of operation
after ACS:
continuous
Timing of operation
after ACS: 6-month
threshold
Timing of operation
after ACS:
12-month threshold
Operation Risk
Factor
Mortality
TABLE 3. The Effect of Timing of ACS on Outcomes From THKR and AAA Surgery
0.617
0.789
P
0.675
0.800
0.679
P
0.69 (0.28–1.68)
1 (1–1.01)
OR (95% CI)
LOS
AAA
1.05 (0.73–1.51)
1.52 (0.95–2.43)
1 (1–1.01)
OR (95% CI)
LOS
AAA
0.411
0.301
P
0.804
0.083
0.132
P
0.62 (0.13–2.97)
1 (0.99–1.01)
OR (95% CI)
Readmission
2.07 (1.37–3.13)
1.41 (0.79–2.54)
0.99 (0.99–1)
OR (95% CI)
Readmission
0.554
0.914
P
0.001
0.250
0.017
P
Annals of Surgery r Volume 255, Number 5, May 2012
Timing Surgery Postvascular Event
www.annalsofsurgery.com | 905
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 255, Number 5, May 2012
Sanders et al
from vascular surgery or for predicting perioperative mortality.2,13 Indeed, we have found that neither previous stroke nor ACS increases
the risk of perioperative mortality from AAA surgery though the wide
confidence intervals (CIs) for stroke suggest we should be cautious in
suggesting that all strokes do not influence outcome. The lack of influence of previous ACS on perioperative mortality is consistent with
the lack of benefit accrued by cardiac revascularization before AAA
surgery.14,15 We suspect that the lack of effect of previous stroke and
ACS on mortality may be related to the widespread vascular disease
in this patient cohort (so individual end-organ insults are not necessarily discriminative in this population) and the high-risk nature of the
surgery involved. Given the limitations of the patient history in this
population, risk stratification of patients undergoing AAA surgery
may be enhanced by additional preoperative biomarker monitoring
such as measurement of preoperative brain natriuretic peptide.16
There are significant differences between the risk factors we
identified and the customized probability model (CPM) for predicting
perioperative mortality from all vascular surgery. Unlike the CPM,
we found age to be predictive of mortality for AAA repair but did
not find respiratory disease to increase the odds of death.8 As the
CPM, by definition includes many types of vascular surgery, it may
not be sufficient for judging a procedure-specific risk for AAA repair.
Indeed our study highlights a problem with extrapolating findings
from one procedure to another and argues that risk prediction models
should be procedure specific.
PERIOPERATIVE RISK FACTORS
FOR THKR SURGERY
Many more preoperative factors influenced perioperative outcome in THKR than AAA surgery. It is of interest that age did not
predict perioperative mortality from THKR surgery; this is important
as greater numbers of elderly patients are presenting for orthopedic
surgery. Nonetheless, age did influence perioperative morbidity in
this surgical cohort indicating these patients still require judicious
perioperative care. Many vascular risk factors (prior stroke and ACS,
diabetes mellitus, hypertension, and heart and renal failure) proved
discriminative to predict perioperative mortality for THKR, consistent with the revised cardiac risk index (that was devised on the basis
of large orthopedic population2 ). In addition to the common factors
and vascular risk factors, cancer and respiratory disease were important predictors of perioperative mortality. Cancer may indicate general frailty, poor nutritional status, or vulnerability to postoperative
complications such as infection or thromboembolism. It is important
that respiratory disease has been isolated as an important factor for
THKR surgery. In these patients, regional rather than general anesthetic techniques may prove beneficial as it is associated with reduced
respiratory complications after surgery.17
TIMING OF ELECTIVE SURGERY AFTER
ACSs AND STROKE
Although we did not find that the timing of prior ACS influenced outcomes when analyzed as a continuous variable, our data
based on thresholds of 6 and 12 months show that ACS induces a
long-term increase in perioperative risk for THKR. We suggest that
surgery should be avoided for 6 months after an ACS and considered
high risk up to 12 months. For surgery conducted within 6 months
of a prior ACS, one might expect one additional perioperative death
in every 64 procedures. If surgery was conducted within 12 months,
one additional death may occur in every 183 procedures. Thus, while
prior ACS itself is a modest risk factor for mortality, the timing of the
procedure plays an important role in exaggerating the mortality risk.
Furthermore, we have shown this effect persists even when all patients
906 | www.annalsofsurgery.com
having PCI or PCI and/or CABG or just those with coronary stents
are excluded. Current guidelines suggest that surgery be delayed for
2 weeks after angioplasty, 6 weeks after bare metal stent placement,
or 12 months after drug eluting stent placement.4 Our data suggest
that there is a pervasive prolonged risk after ACS for THKR surgery
regardless of whether coronary intervention is performed. Indeed,
even when all patients who had PCI and/or CABG were excluded
from the analysis, THKR operation within 6 months of an ACS was
associated with increased risk. Although exclusion of patients who
had a PCI and/or CABG only slightly reduced, the point estimate
of harm at 12 months statistical significance was lost. It is likely
that exclusion of the subjects reduced statistical power to identify the
effect. In sum, our observations support further investigation of the
hypothesis that patients with ACS should, wherever possible, have
elective surgery deferred for longer than is currently the case. This
strategy could be expected to reduce complications resulting from
the ACS (by allowing uninterrupted 12-month antiplatelet therapy
as recommended by current guidelines9,18 ) and perioperative complications, particularly bleeding risk from aggressive antithrombotic
strategies,19 or increased risk of thrombotic events if dual therapy is
stopped.19 Unfortunately, data on the use of antiplatelet agents were
not available for use in our analysis.
Elective AAA surgery conducted within 12 months of an ACS
was associated with increased odds of readmission but not death. We
speculate that this disparity may relate to the extent and severity of
vascular disease in this patient cohort as opposed to the orthopedic
population. Nonetheless, AAA repair conducted within a year of an
ACS should be considered higher risk based on the increased risk
of morbidity. After AAA surgery, cardiovascular causes account for
14.6% of readmissions, and specifically ACS accounts for 3.9% of
readmissions; therefore, it is certainly plausible that timing of ACS
may influence readmission rate.
Our findings are in agreement with Livhits et al’s recent study
that, despite showing increased perioperative risk (in mixed surgeries)
for up to 1 year after myocardial infarction, recommended that surgery
should be delayed for 2 months;3 we consider this recommendation
too permissive given the increased risk imparted by performing the
surgery too soon after an ACS. Unfortunately, extrapolation of their
findings to specific operations is difficult given the inclusion of both
elective and unscheduled procedures. Furthermore, Livhits et al’s
use of controls that never experienced a myocardial infarction likely
exaggerates the impact of the timing of myocardial infarction and
their limited adjustment for other patient variables (by use of Charlson
comorbidity index rather than weighting each condition accordingly)
reduces confidence in their findings.3 Despite these limitations, their
data supports our proposal that surgery should be delayed beyond
the currently recommended period of 4 to 6 weeks after an ACS.
Beyond delaying surgery, we must focus preoperative optimization
and vigilant perioperative care in a procedure-specific manner; the
protective potential of statin,20 alpha-2 adrenergic agonist,21 and βblocker therapy22,23 needs evaluation in THKR surgery.
We were surprised that the preoperative timing of stroke did
not influence perioperative outcome in a similar manner to ACS.
This may reflect the heterogeneity of stroke or that severe strokes
rendered the patient unsuitable for surgery. The wide CIs for the 6month threshold for stroke before THKR suggest that the study may
have been underpowered to address the timing of stroke. However,
the results did not differ if a 12-month threshold was imposed or if
the data were analyzed in a continuous manner (Table 4). However,
timing is only one of many stroke-related factors that may influence
perioperative outcomes. Indeed, “stroke” is a heterogeneous condition
including both hemorrhagic and ischemic events that we were unable
to divide into separate cohorts.
C 2012 Lippincott Williams & Wilkins
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 255, Number 5, May 2012
Timing Surgery Postvascular Event
STRENGTHS AND LIMITATIONS
Hospital Episode Statistics is a large administrative data set that
has shown good comparability with clinical risk prediction scores of
perioperative mortality despite lacking much clinical information.24
The large sample size surveyed gives us power to address the effects
of a diverse range of comorbidities and their timing (and avoiding the
use of composite endpoints2 ) on the “hard” measures mortality and
surrogates of morbidity (LOS and readmission). A further strength
of our analysis is that we have not been constrained by focus on
perioperative cardiac risk and therefore identified other influential
variables beyond vascular risk factors, such as cancer and liver disease. Nonetheless, some variables, for example, stable angina, had to
be dropped from the analysis because of limited patient numbers with
that diagnosis. We focused on short-term endpoints (ie, those within
30 days of the operation), as it was our objective to study perioperative risk; however, we acknowledge that further analyses of longer
term outcomes (such as 12 month mortality) are also warranted.
The reliability of administrative data has been questioned.
However, the accuracy of UK hospital administrative data was found
to approximate 84% and 97% for diagnostic and operation codes,
respectively, in a 2001 systematic review of coding accuracy.25 The
coding is subject to annual inspection, and accuracy has improved
since the introduction of Payment by Results system with its data
assurance framework. For example, the average clinical coding error
rate of 16.5% across primary and secondary diagnosis and procedure coding in 2007–2008 improved in 2008–2009 to 12.8% with a
concomitant reduction in the variability across acute NHS Trusts in
England.26 We have no reason to suspect that data inaccuracies are
more likely to affect records for patients with prior vascular events
than those without.
CONCLUSIONS
Although some common preoperative risk factors emerged
in our analysis, preoperative predictors of perioperative mortality
and surrogates of morbidity are procedure specific. Therefore, riskprediction systems should focus on specific procedures rather than
including multiple different types of surgery. Furthermore, our data
suggest that patients who have an ACS before elective THKR surgery
should be considered higher risk within the first year of the event.
ACKNOWLEDGMENTS
R.D.S. and L.E. developed the initial question and performed
the initial literature searches (with D.M. and M.M.). All authors contributed to study design (lead by A.B., P.A., and R.D.S.). Specifically
K.R.L., M.W., and A.M. provided expert medical opinion and discussed methodology. S.S.J. and M.R. provided similar surgical input.
A.B. analyzed the data. All authors contributed to data interpretation.
R.D.S. and A.B. wrote the manuscript with input from all authors.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
REFERENCES
1. Goldman L, Caldera DL, Nussbaum SR, et al. Multifactorial index of cardiac
risk in noncardiac surgical procedures. N Engl J Med. 1977;297:845–850.
2. Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and prospective
validation of a simple index for prediction of cardiac risk of major noncardiac
surgery. Circulation. 1999;100:1043–1049.
3. Livhits M, Ko CY, Leonardi MJ, et al. Risk of surgery following recent myocardial infarction. Ann Surg. 2011;253:857–864.
4. Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on
perioperative cardiovascular evaluation and care for noncardiac surgery: executive summary: a report of the American College of Cardiology/American Heart
Association Task Force on Practice Guidelines (writing committee to revise
C 2012 Lippincott Williams & Wilkins
24.
25.
26.
the 2002 guidelines on perioperative cardiovascular evaluation for noncardiac
surgery). J Am Coll Cardiol. 2007;50:1707–1732.
NHS Connecting for Health. OPCS classification of interventions and procedures (OPCS-4). http://www.connectingforhealth.nhs.uk/systemsandservices/
data/clinicalcoding/codingstandards/opcs4/index html. Accessed February
2010.
Boersma E, Kertai MD, Schouten O, et al. Perioperative cardiovascular mortality in noncardiac surgery: validation of the Lee cardiac risk index. Am J Med.
2005;118:1134–1141.
Sundararajan V, Henderson T, Perry C, et al. New ICD-10 version of the
Charlson comorbidity index predicted in-hospital mortality. J Clin Epidemiol.
2004;57:1288–1294.
Kertai MD, Boersma E, Klein J, et al. Optimizing the prediction of perioperative
mortality in vascular surgery by using a customized probability model. Arch
Intern Med. 2005;165:898–904.
Friedman LS. The risk of surgery in patients with liver disease. Hepatology.
1999;29:1617–1623.
Carstairs V, Morris R. Deprivation: explaining differences in mortality between
Scotland and England and Wales. BMJ. 1989;299:886–889.
Wright RS, Anderson JL, Adams CD, et al. 2011 ACCF/AHA focused update
of the guidelines for the management of patients with unstable angina/nonST-elevation myocardial infarction (updating the 2007 guideline): a report of
the American College of Cardiology Foundation/American Heart Association
Task Force on Practice Guidelines. J Am Coll Cardiol. 2011;57:1920–1959.
Shih LY, Cheng CY, Chang CH, et al. Total knee arthroplasty in patients with
liver cirrhosis. J Bone Joint Surg Am. 2004;86-A:335–341.
Ford MK, Beattie WS, Wijeysundera DN. Systematic review: prediction of
perioperative cardiac complications and mortality by the revised cardiac risk
index. Ann Intern Med. 2010;152:26–35.
McFalls EO, Ward HB, Moritz TE, et al. Coronary-artery revascularization
before elective major vascular surgery. N Engl J Med. 2004;351:2795–2804.
Poldermans D, Schouten O, Vidakovic R, et al. A clinical randomized trial to
evaluate the safety of a noninvasive approach in high-risk patients undergoing
major vascular surgery: the DECREASE-V Pilot Study. J Am Coll Cardiol.
2007;49:1763–1769.
Karthikeyan G, Moncur RA, Levine O, et al. Is a pre-operative brain natriuretic peptide or N-terminal pro-B-type natriuretic peptide measurement an
independent predictor of adverse cardiovascular outcomes within 30 days of
noncardiac surgery? A systematic review and meta-analysis of observational
studies. J Am Coll Cardiol. 2009;54:1599–1606.
Rodgers A, Walker N, Schug S, et al. Reduction of postoperative mortality
and morbidity with epidural or spinal anaesthesia: results from overview of
randomised trials. BMJ. 2000;321:1493–1497.
Yusuf S, Zhao F, Mehta SR, et al. Effects of clopidogrel in addition to aspirin
in patients with acute coronary syndromes without ST-segment elevation. N
Engl J Med. 2001;345:494–502.
Hall R, Mazer CD. Antiplatelet drugs: a review of their pharmacology and
management in the perioperative period. Anesth Anal. 2011;112:292–318.
Schouten O, Boersma E, Hoeks SE, et al. Fluvastatin and perioperative events
in patients undergoing vascular surgery. N Engl J Med. 2009;361:980–989.
Wallace AW, Galindez D, Salahieh A, et al. Effect of clonidine on cardiovascular morbidity and mortality after noncardiac surgery. Anesthesiology.
2004;101:284–293.
Noordzij PG, Poldermans D, Schouten O, et al. Beta-blockers and statins are
individually associated with reduced mortality in patients undergoing noncardiac, nonvascular surgery. Coron Artery Dis. 2007;18:67–72.
Poldermans D, Boersma E, Bax JJ, et al. The effect of bisoprolol on perioperative mortality and myocardial infarction in high-risk patients undergoing
vascular surgery. Dutch Echocardiographic Cardiac Risk Evaluation Applying
Stress Echocardiography Study Group. N Engl J Med. 1999;341:1789–1794.
Aylin P, Bottle A, Majeed A. Use of administrative data or clinical databases
as predictors of risk of death in hospital: comparison of models. BMJ.
2007;334:1044.
Campbell SE, Campbell MK, Grimshaw JM, et al. A systematic review of
discharge coding accuracy. J Public Health Med. 2001;23:205–211.
Commission A.PbR data assurance framework 2008/09. Key messages from
Year 2 of the national clinical coding audit programme. http://www.auditcommission.gov.uk/SiteCollectionDocuments/AuditCommissionReports/
NationalStudies/20090827pbrdataassuranceframework0809rep.pdf. Accessed February 2010.
www.annalsofsurgery.com | 907
Copyright © 2012 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.