| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Division of General Internal Medicine, and Department of Anaesthesiology and Pain Medicine, University of Alberta, Edmonton, Canada.
Address correspondence and reprint requests to Bruce W. Fisher, MD, MSc, 2E3.31 WCM Health Science Centre, University of Alberta, Edmonton, Canada T6G 2R7. Address e-mail to bruce.fisher{at}ualberta.ca.
Abstract
BACKGROUND: Peripheral arterial disease, as detected by a reduced ankle-to-arm blood pressure index (AAI), has been shown to predict future cardiac events. However, the utility of measuring the AAI to predict postoperative cardiac complications in patients undergoing noncardiac surgery is unknown.
METHODS: We prospectively studied 242 consecutive patients aged 50 yr or older presenting to a university hospital preadmission clinic before elective noncardiac surgery. We performed a standardized clinical evaluation that included calculation of the revised cardiac risk index (rCRI) and measurement of the AAI using both palpation and Doppler techniques. Independent observers, blinded to preoperative assessment and AAI results, ascertained cardiac complications in the first 7 days after surgery. We assessed the ability of an abnormal AAI (
0.9 or absence of all four pedal pulses) to predict postoperative cardiac complications using likelihood ratios (LR), area under the ROC curves (AUC), and multivariable logistic regression in which we adjusted for the rCRI result.
RESULTS: The cohort had a median age of 67 yr, 60% were male, 19% had diabetes, 14% had ischemic heart disease, and 35% underwent intraperitoneal or intrathoracic surgery. Postoperatively, 14 of 242 (6%) patients suffered cardiac complications, but no patients died. An abnormal AAI was present in 44 patients, 10 (23%) of whom had postoperative cardiac complications: positive LR 4.79 (95% CI: 3.04–7.54), negative LR 0.34 (95% CI: 0.15–0.77), AUC = 0.80. The AAI compared favorably with the rCRI, which had positive LR 4.22 (95% CI: 2.24–7.95), negative LR 0.57 (95% CI: 0.34–0.96), and AUC = 0.74. In multivariate analysis, the adjusted odds ratio for having a cardiac complication was 10.16 (95% CI: 2.90–36.02) for those patients with an abnormal AAI, even after adjusting for rCRI results.
CONCLUSIONS: An abnormally low AAI, indicative of underlying peripheral arterial disease, is an independent risk factor for postoperative cardiac complications. The accuracy of the AAI is similar to the rCRI, and it provides additional independent predictive value for preoperative cardiac risk stratification.
Despite improvements in surgical and anesthetic techniques, cardiac complications in the perioperative period remain common, with an incidence of 2% or more in patients undergoing nonemergent noncardiac surgery.1,2 A number of prediction rules have been published that incorporate the results of clinical evaluation to derive estimates for the risk of perioperative cardiac events.3–9 However, their estimates are not always concordant10 and frequently require subsequent expensive testing such as radionuclide scans or exercise stress tests to further stratify risk. Although currently favored for preoperative evaluation of patients undergoing a mix of elective surgical procedures, the revised cardiac risk index’s (rCRI) predictive ability is imperfect2 and it is clear that additional measures and modifications need to be developed to better predict postoperative cardiac complications.
Numerous studies have demonstrated that peripheral arterial disease (PAD) is a marker for underlying asymptomatic coronary heart disease, and PAD is independently associated with increased cardiovascular morbidity and mortality.1–15,16 However, history-taking for symptoms of intermittent claudication to detect PAD is insensitive, even with the use of validated questionnaires.16–21 Conversely, the ankle-to-arm blood pressure index (AAI; mostly commonly defined as the ratio of the systolic blood pressure values from either the dorsalis pedis or posterior tibial arteries divided by the brachial artery value, of less than or equal to 0.914,15,22,23) is a far more reliable marker for the presence of PAD.15,22–26,27
We designed this study to test whether otherwise clinically unrecognized PAD, detected by an abnormal AAI (defined a priori as 0.9 or the absence of any palpable pedal pulses) is associated with an increased risk of perioperative cardiac complications. Although there is a lack of consensus about how to best measure the AAI,28 calculation of the AAI by Doppler ultrasound is often considered the reference standard. Because few clinicians have ready access to Doppler, the accuracy of palpation of the pedal pulses also requires further investigation if the AAI is to be useful in the clinical setting. Therefore, we conducted this prospective cohort study to determine whether the presence of PAD (as defined by an AAI 0.9 or the absence of any palpable pedal pulses) is associated with perioperative cardiac complications and to determine the independent and additional predictive value of AAI over a commonly used, well-validated perioperative risk assessment scoring system, the rCRI.2
METHODS
Subjects and Setting
This prospective cohort study was conducted over 11 mo in a university hospital surgery preadmission clinic (Edmonton, Alberta, Canada). We approached consecutive patients who were older than 50 yr, scheduled to undergo elective noncardiac surgery, and had a planned postoperative stay of 48 h or longer. We obtained written informed consent from each patient, and institutional approval was obtained from the University of Alberta Health Research Ethics Board.
Measurements
Preoperatively, investigators performed a standardized history and physical examination on each patient that included the six components of the rCRI (type of surgery, history of ischemic heart disease or heart failure or stroke, serum creatinine, and insulin use; Table 1). All AAI assessors received the same 1 h training session in technique from the principal investigator, and followed the same protocol for examination. Beyond the initial formal and standardized training, inter- and intraobserver reliability for AAI measurement were not evaluated. To assess the AAI, the brachial blood pressure was measured with the patient supine, using the right arm. Sequentially, right and left posterior tibial, then right and left dorsalis pedis pulses and systolic blood pressures were measured, using palpation and 5 MHz hand-held Doppler techniques (Nicolet Elite 5 mHz vascular model 110R Doppler, Nicolet Vascular, Golden, CO). Blood pressure measurement technique was analogous to that used in the upper arm, with the cuff bladder midpoint centered over the target vessel, and the lower border of the cuff positioned 2–3 cm above the point used for measurement. The AAI for each vessel was calculated as the supine ankle systolic blood pressure divided by the supine arm systolic blood pressure and was recorded separately for each modality of measurement. No noninvasive risk stratification testing was done.
|
Independent Variables of Interest
We defined an "abnormal" AAI result as a value of 0.9 on any of the four vessels examined,14,15,22,23 or the inability to palpate all of the lower extremity pulses. On the basis of the work of Lee et al., we considered an rCRI of 
2 to be indicative of high risk of cardiac events, according to their classification scheme.2
Outcomes
Physicians blinded to the preoperative evaluation (including AAI and rCRI results) and who were not involved in the patients’ continuing care collected data on the occurrence of prespecified clinically significant postoperative cardiac complications within the first seven postoperative days through review of chart, laboratory, and radiology data using standardized definitions2 and data abstraction forms. These cardiac complications replicated those of the original rCRI study2: cardiac death (any death from a cardiac cause, including those deaths after a cardiovascular procedure such as percutaneous transluminal coronary angioplasty or deaths due to an unknown cause), noncardiac death (deaths due to a clearly documented noncardiac cause, e.g., trauma, infection, malignancy), nonfatal acute myocardial infarction (diagnosed as two or more of: characteristic ischemic chest pain lasting longer than 20 min; electrocardiogram [ECG] changes including ST elevation followed by Q waves, a new left bundle branch block, new persistent T wave inversion for at least 24 h, or new ST depression persisting for at least 24 h; or peak CK-MB or troponin I levels more than the institution’s reference limits); cardiogenic pulmonary edema (defined as new onset vascular redistribution, interstitial pulmonary edema, or frank alveolar pulmonary edema confirmed by formal reading of chest radiograph by a radiologist, and considered due to heart failure); primary cardiac arrest; ventricular fibrillation; or complete heart block. All decisions to obtain cardiac enzymes and ECGs were left to the discretion of attending clinicians rather than by routine, and in all observed cases were done in response to clinical circumstances.
Statistical Analysis
Our analyses are primarily descriptive in nature, and we did not have prespecified power calculations. Our intent was to enroll about 250 elective surgery patients older than age 50 yr, with the anticipation that this would provide approximately 20 postoperative cardiac events to permit adequate multivariate analysis for the 2 factors of interest (AAI and rCRI). Blood pressure and AAI values obtained using different sites and techniques (palpation vs. Doppler) were compared using t-tests. We calculated the sensitivity, specificity, positive and negative likelihood ratios (LR), and area under the receiver operating characteristic curve (AUC) of the AAI and rCRI for predicting postoperative cardiac complications.29,30 The unadjusted (univariate) association between abnormal AAI and elevated rCRI scores and postoperative outcomes were evaluated using 
2 tests and the strength of these associations quantified using odds ratios (OR). Last, given that we had fewer than 20 events and that the rCRI was derived from multivariable regression of an exhaustive list of covariates, we conducted parsimonious multivariate logistic regression analyses that included only the rCRI (elevated or not) and the AAI (abnormal or not) to determine if the latter provided independent additional information with respect to predicting postoperative cardiac complications.
RESULTS
Of those approached, 278 of 315 patients (88%) agreed to participate. Of the consenting patients, we excluded from enrollment those who could not complete the entire assessment (n = 10), those in whom assessment of pedal pulses and blood pressure was not possible due to obesity, peripheral edema, or extensive skin lesions (n = 7), and those patients with rapid irregular heart rates (n = 2). Of the 259 consecutively enrolled patients, 11 had not undergone surgery by 1 yr after assessment, and a further 6 patients had postoperative stays <48 h; these patients were excluded from further analysis. The median age of those who did not agree to participate or who were excluded from the study was 66 yr and 52% were men. These values, the types of surgery undergone, and other characteristics were not significantly different from those included in the study (data not shown). In our final study sample of 242 patients, the median age was 67 yr, 145 (60%) were men, 34 (14%) had rCRI scores of 2 or higher, and only 5 had a previous diagnosis of PAD. The two most common types of surgery were orthopedic and intraperitoneal (Table 1).
Postoperative Cardiac Complications
Fourteen patients (6%) suffered cardiac complications within 7 days of surgery: 6 patients with nonfatal myocardial infarctions, 7 patients with episodes of cardiogenic pulmonary edema, and one patient suffering both complications (counted as a single event). These complications occurred in 6 of 72 (8%) patients undergoing intraperitoneal surgery, 3 of 113 (3%) of patients undergoing orthopedic surgery, and 5 of 41 (12%) patients undergoing breast, head and neck, plastic, or neurosurgeries. No patients died.
Measurements of AAI
Overall, 44 (18%) patients had PAD detected by an abnormal palpated AAI, because of either a palpated AAI <0.9 (n = 30) or by the absence of all four pedal pulses (n = 14). Doppler techniques systematically detected higher systolic blood pressures at each site than were detected by palpation (Table 2), and an AAI calculated using Doppler technique categorized fewer patients as abnormal (Table 3). Of these 44 patients, 29 had rCRI scores that were <2 (Table 1).
|
|
Association Between PAD and Outcomes
An abnormal palpated AAI was strongly associated with postoperative cardiac complications, sensitivity 71%, specificity 85%, LR positive 4.79 (95% CI: 3.04–7.54), and LR negative 0.34 (95% CI: 0.15–0.77) (Table 3). The AUC for abnormal palpated AAI predicting cardiac complications was 0.80 (95% CI: 0.75–0.84). An abnormal AAI calculated by Doppler technique lacked sensitivity but had somewhat greater specificity and positive LR (Table 3).
Association Between rCRI and Outcomes
A rCRI score of 2 was also associated with postoperative cardiac complications, specificity was similar to the palpated AAI, although it was less sensitive (Table 2). The AUC for the rCRI was 0.74 (95% CI: 0.69–0.78); this value is not larger than the AUC for the palpated AAI in our study and is of almost the same magnitude (AUC = 0.76) reported by Lee et al. in the derivation cohort of the original rCRI studies. An abnormal palpated AAI predicted postoperative cardiac complications in three patients who had rCRI scores <2; conversely, the elevated rCRI predicted complications in one patient who had a normal palpated AAI.
Univariate and Multivariate Analyses Comparing the AAI and rCRI
In unadjusted analyses, an abnormal palpated AAI was associated with postoperative cardiac complications (OR: 14.26, 95% CI: 4.23–48.10, P < 0.001). After adjustment for the rCRI scores, the AAI still had substantial independent value in predicting cardiac complications (adjusted OR: 10.16, 95% CI: 2.90–36.02, P = 0.003).
Sensitivity Analyses Around the Definition of PAD and Abnormal AAI
Five patients gave a self-reported history that fulfilled the Edinburgh Claudication Questionnaire criteria for claudication and three patients had a history of amputation or revascularization for the treatment of PAD. Only 1 of the 5 patients with self-reported intermittent claudication had a palpated AAI 0.9, and none had postoperative cardiac complications. The palpated AAI was 0.9 in 2 of the 3 patients with previous amputation or revascularization; 1 of these 2 patients suffered a postoperative cardiac complication. Nevertheless, adding PAD diagnosis by history of claudication with a validated instrument or previous PAD surgery provided no additional predictive value over measurement of the palpated AAI alone in sensitivity analyses (the positive LR for this combination variable was 4.8; identical to that for abnormal AAI alone).
In addition, as some authors have suggested that an AAI >1.2 should be considered abnormal, particularly in those with diabetes mellitus, we reanalyzed our data using this cutpoint to define an abnormal AAI. Seven patients had AAI >1.2, and 2 had type 2 diabetes. None of these seven patients had cardiac complications. This did not significantly affect any of the results, except for a slight deterioration in specificity (0.82 vs 0.85 with our original definition of abnormal AAI).
DISCUSSION
We found that the presence of lower limb PAD (defined as either the absence of all four pedal pulses or any palpated AAI 0.9) was independently correlated with increased risk of postoperative cardiac complications, and that it added predictive value over the commonly used and previously validated rCRI.
The determination of the AAI by palpation alone is a simple maneuver that is easily introduced into routine clinical practice without requiring additional equipment. It is noteworthy that in the rCRI no physical examination measures were included in the final models,2 although an elevated jugular venous pressure, a third heart sound, and valvular aortic stenosis were all part of the original cardiac index.6 We believe that the patients in our study with nonpalpable pulses or AAI 0.9 had extensive but "subclinical" atherosclerosis present in multiple beds that was not signaled by the historical and laboratory criteria included in the rCRI, and that recognition of this increased burden of atherosclerosis may have been responsible for the AAI ’s better sensitivity for predicting cardiovascular complications.
To our knowledge, this is the first prospective study with blinded ascertainment of outcomes to examine the potential role of using the AAI to detect PAD for the purpose of perioperative cardiac risk stratification in patients undergoing noncardiac, nonperipheral vascular surgeries. Before clinicians consider adopting our findings, they should consider the following potential limitations. This was a study with a small number of events, although there were sufficient events that we were able to replicate and validate the usefulness of the rCRI in our cohort. Second, all of the patients were drawn from a single center, perhaps limiting generalizability. That said, the rCRI was also developed and then validated in a single center of size similar to ours. Third, we did not perform formal measurements of inter- and intrarater reliability for detecting an abnormal AAI. Fourth, the final sample of patients consisted of 242 of 315 (77%) of the patients approached. Although this raises the possibility of selection bias, the age and sex distribution, and types of surgery of the nonparticipating patients were similar to that of the study group. Fifth, postoperative cardiovascular complications were diagnosed by attending clinicians, in response to clinical circumstances, rather than by any study protocols that dictated routine ascertainment of clinical signs, ECG, or cardiac biomarkers. Last, because of the small number of patients who experienced events (n = 14), we were precluded from conducting multivariate analyses on more variables than the AAI and rCRI. However, we specifically chose the rCRI as a comparator because it is a validated model from a large population that incorporated a number of these variables into one score.
In conclusion, detection of PAD by determining the palpated AAI is simple, rapid, and noninvasive. Our study suggests it is also an independent predictor of perioperative cardiac complication, with independent and additive predictive performance to that of the commonly used rCRI. On the basis of our results, we believe that measurement of the AAI has potential added value as part of routine preoperative cardiac risk stratification in addition to the rCRI.
Footnotes
Accepted for publication February 25, 2008.
Supported by the FAM and SRM hold career salary support awards from the Alberta Heritage Foundation for Medical Research and the Canadian Institutes of Health Research. FAM also holds the Merck Frosst/Aventis Chair in Patient Health Management at the University of Alberta. Sanofi-Synthelabo Inc. provided 3 hand held Doppler devices for use in the study but did not have input into the design, conduct, analysis, interpretation, or reporting of this study.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
