Polanczyk CA, Rohde LE, Goldman L, Cook EF, Thomas EJ, Marcantonio ER, Mangione CM, Lee TH. Right Heart Catheterization and Cardiac Complications in Patients Undergoing Noncardiac SurgeryAn Observational Study. JAMA. 2001;286(3):309-314. doi:10.1001/jama.286.3.309
Author Affiliations: Section for Clinical Epidemiology, Division of General Medicine (Drs Polanczyk, Cook, Thomas, Marcantonio, and Lee), and Cardiovascular Division (Drs Polanczyk, Rohde, and Lee), Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Mass; Department of Medicine, University of California, San Francisco, School of Medicine, San Francisco (Dr Goldman); and Department of Medicine, University of California, Los Angeles, School of Medicine, Los Angeles (Dr Mangione).
Context Right heart catheterization (RHC) is commonly performed before high-risk
noncardiac surgery, but the benefit of this strategy remains unproven.
Objective To evaluate the relationship between use of perioperative RHC and postoperative
cardiac complication rates in patients undergoing major noncardiac surgery.
Design Prospective, observational cohort study.
Setting Tertiary care teaching hospital in the United States.
Patients Patients (n = 4059 aged ≥50 years) who underwent major elective noncardiac
procedures with an expected length of stay of 2 or more days between July
18, 1989, and February 28, 1994. Two hundred twenty one patients had RHC and
3838 did not.
Main Outcome Measure Combined end point of major postoperative cardiac events, including
myocardial infarction, unstable angina, cardiogenic pulmonary edema, ventricular
fibrillation, documented ventricular tachycardia or primary cardiac arrest,
and sustained complete heart block, classified by a reviewer blinded to preoperative
Results Major cardiac events occurred in 171 patients (4.2%). Patients who underwent
perioperative RHC had a 3-fold increase in incidence of major postoperative
cardiac events (34 [15.4%] vs 137 [3.6%]; P<.001).
In multivariate analyses, the adjusted odds ratios (ORs) for postoperative
major cardiac and noncardiac events in patients undergoing RHC were 2.0 (95%
confidence interval [CI], 1.3-3.2) and 2.1 (95% CI, 1.2-3.5), respectively.
In a case-control analysis of a subset of 215 matched pairs of patients who
did and did not undergo RHC, adjusted for propensity of RHC and type of procedure,
patients who underwent perioperative RHC also had increased risk of postoperative
congestive heart failure (OR, 2.9; 95% CI, 1.4-6.2) and major noncardiac events
(OR, 2.2; 95% CI, 1.4-4.9).
Conclusions No evidence was found of reduction in complication rates associated
with use of perioperative RHC in this population. Because of the morbidity
and the high costs associated with RHC, the impact of this intervention in
perioperative care should be evaluated in randomized trials.
Cardiac complications are the most common cause of death among patients
undergoing elective noncardiac surgery.1,2
In the last 2 decades, multifactorial cardiac risk indexes to assist clinicians
in preoperative risk stratification have been developed and prospectively
In addition, research has shown that intraoperative hemodynamic changes are
associated with increased complication rates.7
These findings logically lead clinicians to consider hemodynamic monitoring
via right heart catheterization (RHC) for selected patients undergoing high-risk
procedures, with a goal of reducing perioperative complications.8
The benefit of this strategy is, however, unproven.
The purpose of this study was to evaluate the relationship between use
of perioperative RHC and occurrence of major postoperative cardiac events
among a large cohort of patients who underwent major nonemergent noncardiac
Patients aged 50 years or older who underwent major elective noncardiac
surgeries at Brigham and Women's Hospital, Boston, Mass, between July 18,
1989, and February 28, 1994, were eligible for the study. As described in
prior reports,6,9 major noncardiac
surgeries were defined as those with an expected length of stay of 2 or more
days. For the purpose of this study, only procedures electively scheduled
and performed in the same or next day following hospital admission were included
in the analyses. Because 91% of patients (182/200) who underwent abdominal
aortic aneurysm repair also had RHC, patients undergoing this procedure were
excluded from the analysis. The Brigham and Women's Hospital Human Research
Committee approved the enrollment and clinical data collection protocol.
As described previously,6 patients who
provided written informed consent to the full study protocol underwent preoperative
evaluations by study personnel using a structured data form. These evaluations
included detailed medical histories, physical examinations, and laboratory
test data collection. For patients who did not undergo this evaluation because
they could not be approached or because they refused participation in the
interview part of the study, clinical data were obtained by the anesthesiologist
from the structured evaluation in the medical record. This data source was
also used to obtain American Society of Anesthesiologists (ASA) classification
for all patients. The ASA classification system uses the general clinical
severity of a patient's systemic illnesses to predict perioperative mortality.10 Consenting patients agreed to postoperative sampling
of creatine kinase (CK) and, if total CK levels were elevated, CK-MB sampling
according to a protocol in which samples were collected immediately after
surgery, at 8 PM on the day of surgery, and on the next 2 mornings. For other
enrolled patients, samples were obtained according to physician order. Electrocardiograms
were performed in the recovery room and on the first, third, and fifth postoperative
days if patients remained hospitalized. Functional status was measured by
the Specific Activity Scale, a 4-level classification system that uses activities
of daily living to estimate a patient's functional capacity.11
Occurrence of major postoperative events was classified by a single
reviewer (L.G.) who was blinded to preoperative clinical data and use of RHC
and used postoperative clinical information including cardiac enzyme measurements,
electrocardiograms, and clinical events. Diagnosis of myocardial infarction
was made on the basis of CK-MB levels and electrocardiographic findings.6 Diagnosis of postoperative congestive heart failure
was obtained from progress notes recorded by clinicians involved in the patients'
care. Major cardiac-related events included myocardial infarction, unstable
angina, cardiogenic pulmonary edema, ventricular fibrillation, documented
ventricular tachycardia or primary cardiac arrest, and sustained complete
heart block. Major noncardiac events included pulmonary embolism documented
on autopsy, angiography, or high-probability ventilation-perfusion scan, respiratory
failure requiring intubation for at least 2 days or reintubation, noncardiogenic
pulmonary edema, lobar pneumonia confirmed by chest radiograph and requiring
antibiotic therapy, acute renal failure requiring dialysis, cerebrovascular
accident with new neurologic deficit, and gastrointestinal bleeding.
Intraoperative variables recorded by the anesthesiologist included initial
systolic and diastolic blood pressure, initial heart rate, type of anesthesia,
invasive monitoring procedures, surgery time, estimated blood loss, lowest
intraoperative systolic blood pressure, maximum heart rate, and episodes of
hypertension (systolic blood pressure >200 mm Hg or diastolic blood pressure
>120 mm Hg) and hypotension (systolic blood pressure <90 mm Hg or decrease
of >13 from baseline level).
All records were reviewed to evaluate whether RHC was indicated perioperatively
to monitor hemodynamic parameters during the perioperative period. In only
1 patient, the chart information suggested that RHC was performed to manage
other critical illness or clinical instability preoperatively; this patient
was excluded from all analyses for the purposes of this study.
Preoperative clinical variables that would relate to the decision to
use RHC during a major procedure were considered and included in a multivariate
logistic regression analysis. The significant independent variables identified
in this cohort were sex, myocardial infarction in the 6 months prior to the
procedure, history of chronic ischemic heart disease, hypertension, poor general
medical status, type of surgical procedure (intrathoracic, vascular, or abdominal),
presence of S3 or jugular venous distension, frequent premature
ventricular contractions or cardiac rhythm other than sinus on resting electrocardiogram,
preoperative oxygen saturation of less than 94%, use of digoxin, and ASA classification.
To represent as completely as possible factors that might influence physicians'
decisions to use perioperative RHC, other preoperative variables were also
included in the propensity score if they were deemed clinically relevant,
even if they were not statistically significant correlates. These variables
included age, previous coronary artery bypass graft surgery or percutaneous
coronary angioplasty, significant valvular disease, peripheral vascular disease,
chronic obstructive pulmonary disease, history of chronic renal failure, diabetes,
and preoperative creatinine level of more than 2.0 mg/dL (177 µmol/L).
The multivariate regression model of propensity for using perioperative RHC
had a c statistic of 0.85, which represents the area
under the receiver operating characteristic curve.
Patients who did not undergo perioperative RHC were matched to patients
who had RHC on the basis of propensity score and type of surgical procedure.
Patients were randomly selected from those who underwent perioperative RHC,
then all 3838 patients who did not have RHC were searched to find those who
had the closest propensity score for undergoing RHC (within 0.03 on a scale
Univariate correlation between intraoperative parameters and occurrence
of perioperative cardiac complications were performed using the χ2 test and the Fisher exact test for categorical variables and the t test or Wilcoxon rank sum test for continuous variables.
Variables with a P value of less than .10 were entered
into the multiple regression analysis. Stepwise logistic regression analysis
was used to determine independent (P<.05) correlates
of cardiac complications after adjustment for preoperative clinical variables.
Differences between matched pairs were evaluated using the Wilcoxon
signed rank test for continuous variables. The association between RHC and
postoperative complications (after adjustment for propensity score alone and
after simultaneous adjustment for propensity and each significant variable)
was determined by conditional logistic regression analysis. A 2-sided P value of less than .05 was considered statistically significant.
Statistical analyses were carried out using SAS software (SAS Institute Inc,
Table 1 shows the clinical
characteristics of 4059 patients who underwent major noncardiac surgery (other
than aortic aneurysm repair) with (n = 221) and without (n = 3838) RHC. Overall,
patients who underwent noncardiac procedures were relatively young, with a
high prevalence of hypertension and a low prevalence of cardiovascular comorbid
conditions. Patients who underwent perioperative RHC were significantly older;
had a significantly higher prevalence of hypertension, diabetes, congestive
heart failure, previous myocardial infarction, and coronary artery bypass
graft surgery; and had a worse functional status as measured by the Specific
Activity Scale.11 Right heart catheterization
was performed more frequently in patients who underwent high-risk procedures,
such as vascular and intrathoracic surgeries, than lower-risk procedures.
In univariate analysis, patients who underwent perioperative RHC were
more likely to have postoperative myocardial infarction (2.3% vs 0.8%; P = .04) and congestive heart failure (13.6% vs 2.4%; P<.001) than patients who did not undergo RHC. In addition,
patients who underwent RHC had a 3-fold increase in the prevalence of the
combined end point of postoperative major cardiac events (15.4% vs 3.6%; P<.001) (Table 2).
Major noncardiac complications were also more common in patients who underwent
perioperative RHC (10.0% vs 2.7%; P<.001).
The multivariate regression model of propensity for using perioperative
RHC had a c statistic of 0.85, indicating good discrimination
between patients who did and did not undergo RHC. The overall cohort had a
propensity score of 0.06 (median, 0.02; range, 0.003-0.74) for RHC, reflecting
a very low average likelihood of receiving this intervention. Patients who
underwent RHC had a mean propensity score of 0.19 (median, 0.12; range, 0.004-0.74).
After adjustment for the propensity for RHC as well as additional adjustment
for type of surgical procedure and all clinical characteristics described
in Table 3, the odds ratio (OR)
of major postoperative cardiac events for patients who underwent RHC was 2.0
(95% confidence interval [CI], 1.3-3.2). After similar adjustments, the OR
of postoperative congestive heart failure for patients who underwent RHC was
2.9 (95% CI, 1.7-4.9), while the OR of postoperative acute ischemic syndromes
(myocardial infarction or unstable angina) was 1.3 (95% CI, 0.7-2.6). Similar
findings were observed for the combined end point of major postoperative noncardiac
events (OR, 2.1; 95% CI, 1.2-3.5).
In this cohort, there were 221 patients (5%) who underwent perioperative
RHC. We were unable to adequately match (within 0.03 on a scale from 0-1.00)
the propensity score for RHC in 6 patients; hence, the case-matching analysis
included 215 pairs of patients with and without RHC. Among these pairs, there
were no statistically significant differences between patients with and without
RHC with regard to age, cardiovascular risk factors, previous cardiovascular
morbidity, use of drugs, functional status, preoperative laboratory profile,
and electrocardiographic findings (Table
3). The mean propensity to receive the procedure was nearly identical
in the 2 groups (0.17 vs 0.17; P = .99).
For matched pairs, the risk of developing major postoperative cardiac
events was higher but nonsignificant in patients who underwent RHC (OR, 1.6;
95% CI, 0.9-2.8) (Table 4). In
analyses for specific types of cardiac complications, postoperative congestive
heart failure was significantly associated with perioperative RHC (OR, 2.9;
95% CI, 1.4-6.2). In this matched-pairs analysis, RHC was also associated
with increased risk of postoperative noncardiac complications (Table 4).
Several multivariate analyses were performed to adjust for potential
treatment selection bias. In the case-matched population, logistic regression
models were performed including each variable individually and in combinations
to evaluate their impact on the associations between RHC and cardiac complications.
The ORs for major cardiac and noncardiac complications did not change substantially
(<10%) from the baseline analysis for all 25 variables evaluated.
We further compared intraoperative clinical and hemodynamic parameters
between matched pairs (Table 5).
Although there was no difference in perioperative hypotensive or hypertensive
episodes between cases and controls, patients who underwent RHC had higher
maximal perioperative heart rate, surgery time, and net positive fluid balance
(P<.001 for all). Finally, patients managed with
RHC had a significantly prolonged mean (SD) length of hospital stay compared
with patients who did not have RHC (11.3 [10.2] days [median, 9 days] vs 8.1
[5.6] days [median, 7 days]; P<.001).
In this observational study from a large cohort of patients undergoing
elective major surgical procedures, perioperative RHC was not associated with
improved postoperative outcomes and was associated with prolonged hospitalization.
These results were observed despite adjustment for a wide variety of potentially
confounding variables using case-matched analysis and multivariate models
to adjust for the type of surgical procedures and propensity to use RHC. Indeed,
after such adjustments, patients who underwent RHC remained more likely to
develop major postoperative cardiac complications compared with patients who
did not undergo RHC. Most of this increment in postoperative events was related
to higher rates of development of postoperative cardiogenic pulmonary edema
and was associated with a greater perioperative net fluid intake.
The rationale for monitoring intraoperative parameters guided by RHC
is based in part on the concept that deliberate increases in oxygen delivery
may overcome the increased metabolic demand associated with major surgery
and, ultimately, improve postoperative outcomes. A pivotal study by Shoemaker
et al12 in the late 1980s suggested that achievement
of supranormal values of oxygen transport measures could be associated with
decreased mortality in very high-risk patients undergoing noncardiac surgery
(with multiple organ dysfunction, massive blood loss, severe trauma, or extensive
surgery). However, this subgroup of patients represents a small percentage
of the general surgical caseload, even at tertiary care hospitals.13
Since this initial report, few randomized trials (with relatively small
sample sizes) have addressed the impact of perioperative RHC on clinical outcomes.
Most of these studies failed to demonstrate significant differences in cardiac
morbidity between groups of patients who did and did not receive RHC.14- 16 These findings resonated
in a recent report that questioned the benefit of RHC in critically ill medical
Several possible explanations for our findings can be postulated. First,
physicians may not have properly used the information obtained from RHC monitoring
during surgery and the immediate postoperative period. Iberti et al18 demonstrated in a multicenter survey that physicians'
understanding of pulmonary artery catheterization data is extremely variable.
Shoemaker et al12 suggested that the overwhelming
majority of RHC performed in general surgery patients is not used to obtain
crucial data to adequately evaluate the systemic oxygen profile. Several recent
reports, however, also demonstrated that deliberate boosting of cardiac index
and oxygen delivery fail to improve outcomes and even may be detrimental in
critically ill patients.19
Second, as suggested by the report from the Study to Understand Prognoses
and Preferences for Outcomes and Risks of Treatment (SUPPORT),17
RHC may be a marker for an aggressive style of care that contributes to worse
outcomes. Similar findings were also reported in studies of the impact of
RHC on patients with myocardial infarction.20,21
Moreover, our findings raise the question of whether some of the increased
cardiac morbidity associated with perioperative RHC may be related to an increased
risk of pulmonary cardiogenic edema due to greater perioperative fluid loading.
Third, any invasive procedure carries some risk of adverse outcomes.
Right heart catheterization has been associated with ventricular arrhythmias
and heart block,22- 24
catheter-related sepsis,25 pulmonary embolism,26 central venous access complications,27
and death.28,29 It is possible
that the potential benefits of RHC are offset in part by the physiological
burden of indwelling instrumentation.30
The findings of this study must be interpreted in the context of the
limitations of the study design. Generalizability of our findings may be limited
to low- and moderate-risk populations who undergo major noncardiac procedures.
Because these data represent the practice in only 1 tertiary hospital, these
results should be confirmed in other settings. Exclusion of aortic aneurysm
repair from our analysis precludes any inference about the usefulness of RHC
for this procedure. Because of the observational nature of our data, we cannot
entirely exclude the possibility that important confounding variables might
have been neglected in the analysis or that adjustment for confounders was
However, in this cohort of patients undergoing major elective noncardiac
surgery, we were unable to demonstrate evidence of benefit associated with
use of perioperative RHC. Because of the morbidity and the high costs associated
with RHC, the impact of this intervention in perioperative care should be
carefully reevaluated. We believe that the results from this observational
study should foster new attempts to address this important question in randomized