- Email: [email protected]
Prediction of cardiac and pulmonary complications related to elective abdominal and noncardiac thoracic surgery in geriatric patients
Prediction of Cardiac and Pulmonary Complications Related to Elective Abdominal and Noncardiac Thoracic Surgery in Geriatric Patients MYRONC. GERSON,M.D., JAMESM. HURST,M.D., VICKIS. HERTZBERG, Ph.D., ROBERT BAUGHMAN, M.D., GREGORY W. ROUAN,M.D., KARENELLIS,R.N., with the technical assistance of: ESTELLE E. FISCHER,B.A., MELISSA S. COLTHAR, AS., PEGGY M. BURWINKLE, C.N.M.T. Cincinnati, ohio
PURPOSE:Major cardiac and pulmonary complications associated with abdominal and noncardiac thoracic surgery are a common cause of mortality and serious morbidity in elderly patients. We postulated that a simple, inexpensive bicycle exercise test could provide objective documentation of cardiopulmonary reserve and, therefore, predict perioperative pulmonary as well as cardiac complications. PATIENTsANDMETHODs:hiOr to elective surgery, 177 patients aged 65 years or older had assessment of the clinical history, results of physical examination, electrocardiogram, chest radiograph, blood chemistries, pulmonary function test fiidings, supine exercise test results, Dripps classification, and Goldman cardiac risk factors. Observations in patients with and without major perioperative cardiac and/or pulmonary complications were compared using univariate analysis followed by a multivariate logistic regression procedure. RESULTS:Major perioperative complications were pulmonary in 24 patients, cardiac in 25 patients, and either cardiac or pulmonary in 39 patients. By multivariate analysis, inability to perform two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was the best predictor of perioperative pulmonary, cardiac, and combined cardiopulmonary complications (p <0.0005). Among 108 patients who were able to achieve these exercise criteria, cardiac or pulmonary complications occurred in 10 patients (9.3%), with one death (0.9%). Among 69 patients unable to exercise satisfactorily, cardiac or pulmonary complications occurred in 29 patients (42%), with five total deaths (7.2%). CONCLUSION: Objective measurement of exercise capacity by supine bicycle ergometry appears to be of clinical value for preoperative risk stratification for both pulmonary and cardiac complications prior to maior elective abdominal or noncardiac thoracic surgery in elderly patients.
From the Departments of Internal Medicine (Cardiology [MCG, KE, EEF, MSCl. Pulmonarv IRBl. and General Internal Medicine lGWR1 Divisions). Surg&y (JMH). ‘E&dnmental Health (Epidemiology ‘and eiostattstlds [VSH]). and Radiology (Eugene L. Saenger Radtoisotope Laboratory [MCG, PMB]), University of Cincinnati Medical Center and the Veterans Admlnistration Medical Center, Cincinnati, Ohio. This workwas supported in part by a grant from the Southwestern Ohio Chapter of the Amencan Heart ASSOCIation, Clnclnnatl. Ohio. Requests for reprints should be addressed to Myron C. Gerson, M.D., University of Cincinnati, Division of Cardiology, Mail Location #542, Clncinnatl. Ohio 45267. Manuscript submitted May 1.1989, and accepted in revised form December 6, 1989.
T
he elderly constitute an increasingly large segment of the population of the United States. More than 600,000 individuals 65 years of age and older require major abdominal or noncardiac thoracic surgery each year [l]. Approximately 10% of these patients develop a major cardiac complication related to surgery [2], and the number of patients developing major perioperative pulmonary complications may be even greater. In a previous study of elderly patients preparing for major abdominal and noncardiac thoracic surgery [3], it was observed that a simple, inexpensive bicycle exercise test provided information that was more predictive for perioperative cardiac complications than data available from the clinical history, physical examination, chest radiograph, and electrocardiogram. In these patients, objective measurement of exercise capacity has been a better predictive screening procedure for major perioperative cardiac complications compared to more costly and more complex tests, including rest and exercise radionuclide ventriculography [3] and dipyridamole thallium imaging [4]. The purpose of the present study was to determine whether previously established exercise test criteria [3] for predicting perioperative cardiac complications could also provide information predictive of perioperative pulmonary complications.
PATIENTS AND METHODS The study population consisted of 87 men and 90 women aged 65 years or older (mean age, 73 years) scheduled for major elective abdominal or noncardiac thoracic surgery. Patients were consecutively entered into a study of cardiac risk provided that (1) there was sufficient time between scheduling and performing of surgery to permit preoperative testing, (2) surgery involved entry into the pleural or peritoneal cavity (minor procedures such as simple appendectomy, feeding tube placement, and inguinal herniorrhaphy were excluded), and (3) the patient consented to study participation. Surgical procedures were abdominal in 161 patients and thoracic in 16 patients. During the enrollment period for the study, 53% of patients aged 65 years or older who were having major abdominal or noncardiac thoracic surgery at the University of Cincinnati Hospitals were studied. Eligible patients not enrolled in the study had a mean age of 72 years, included 51% females, and had abdominal surgery in 96% of cases (all p = NS compared to study patients). In a previous study [3] of 100 elderly patients preparing for noncardiac surgery, it was found that inability both to perform two minutes of supine bicycle exercise and to raise the heart rate above 99 beats/minute during bicycle exercise was a powerful predictor of February
1990
The American
Journal
of Medicine
Volume
88
101
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
“poor general medical condition,” and clinically important valvular aortic stenosis [2]. All study patients were of an advanced age (in this case greater than 64 years) and underwent abdominal or noncardiac thoracic surgery. No study patient underwent emergency surgery. Exercise electrocardiographic testing was performed with the supine patient pedaling a bicycle ergometer at 50 revolutions/minute, beginning at a workload of 25 watts and, if tolerated, increasing the workload by 12.5 watts every minute until there was evidence of limiting dyspnea, chest pain, or fatigue. Exercise ST-segment levels were monitored in leads I, II, III, and Vg, A positive ST-segment response required a normal baseline ST-segment at rest and during hyperventilation with at least 1 mm of horizontal or downsloping ST-segment depression at 0.08 second after the J point, during or following exercise. On the basis of a previous study of 100 separate geriatric patients preparing for abdominal or noncardiac thoracic surgery, “inability to exercise” was defined as inability to pedal a supine bicycle for at least two minutes and raise the heart rate above 99 beats/minute [3]. Postoperative rhythm strips, chest radiographs, results of physical examinations, and catheterization data were reviewed up to the time of hospital discharge for evidence of cardiac or pulmonary complications. Serum creatine kinase and lactic dehydrogenase with isoenzyme determinations were measured at 0,6, 12,24,48, and 72 hours following surgery. Electrocardiograms were obtained on the first, second, and third postoperative days. Major perioperative cardiac complications included myocardial infarction, documented by new infarction Q waves or the presence of creatine kinase MB fraction by electrophoresis; cardiac death; ventricular tachycardia, defined as three or more consecutive ventricular beats at a heart rate greater than 140 beats/minute, or ventricular fibrillation; and left heart failure, defined as bilateral rales and new evidence of left heart failure documented by typical radiographic changes, or by a pulmonary capillary wedge pressure greater than 25 mm Hg. Major perioperative pulmonary complications were defined clinically or at autopsy as: (1) pneumonia (defined clinically by a new pulmonary infiltrate on chest radiogram plus at least two of the following: [a] fever 37.7’C or higher; [b] white blood cell count greater than 10,50O/mms; [c] initiation of antibiotic treatment; [d] demonstration of pathogenic organisms on a sputum Gram stain or culture); (2) noncardiogenic pulmonary edema (defined clinically by chest radiogram documentation of interstitial edema at a time when the pulmonary capillary wedge pressure is 16 mm Hg or less and the patient has not been treated for left heart failure); or (3) pulmonary embolism (documented by pulmonary arteriogram or autopsy, or supported by a ventilation/perfusion radioisotope scan showing “high probability” [7] of pulmonary embolism and followed by institution of heparin therapy). Observations in patients with and without perioperative cardiac and/or pulmonary complications were compared using the Statistical Analysis System [8]. Continuous variables were presented using median values and were tested for univariate significance using a two-tailed Wilcoxon rank sum statistic. Discrete variables were tested using a two-tailed Fisher’s exact test. Preoperative variables that were univariately
perioperative cardiac complications. Because the “inability to exercise” criterion was retrospectively derived in those 100 patients and because patients in that group who had perioperative cardiac complications also had an increased incidence of perioperative pulmonary complications, those 100 patients were excluded from the present study population. The “inability to exercise” criterion was prospectively tested for prediction of perioperative cardiac complications in 55 additional patients in that study [3]. These 55 additional patients together with 122 previously unreported patients form the present study population. Thus, the present study is not prospective because data had been collected (but not analyzed) from those 55 patients before the present study hypothesis regarding prediction of perioperative pulmonary complications was generated. Prior to surgery, the patient’s age, sex, ambulatory status, type of operation, duration of anesthesia, and the presence of chest pain or myocardial infarction in the preceding six months were noted from the medical history. Each patient was examined by a cardiologist for an S3 gallop; jugular vein distention, as indicated by a column of blood in the internal jugular vein extending more than 4 cm above the manubrium with the patient lying at a 45” angle; a systolic murmur and carotid pulse suggestive of clinically important valvular aortic stenosis; and a liver span greater than 12 cm, or other findings of chronic liver disease. Height and weight were recorded for each patient. A Quetelet index was calculated as 100 times the patient’s weight in pounds divided by the square of the height in inches [5]. The resting 12-lead electrocardiogram was analyzed in a blind manner for infarction Q waves (characterized by a duration greater than 0.04 second and a depth greater than or equal to 1 mV) and for “rhythm other than sinus” [2]. All preoperative rhythm strips including a one-minute strip obtained in the Exercise Laboratory were reviewed for the presence of greater than five premature ventricular contractions per minute [2]. Without reference to the clinical data, a staff radiologist reviewed the preoperative chest radiographs for cardiomegaly, pulmonary congestion, or pulmonary infiltrate. Each study patient had a preoperative measurement of serum creatinine, blood urea nitrogen, and serum electrolytes; and a serum creatine kinase and lactic dehydrogenase with isoenzyme measurements. Other preoperative laboratory data including arterial blood gases, pulmonary function tests, and hepatic enzyme levels were ordered by the surgeon if deemed appropriate. Pulmonary function measurements of forced vital capacity (FVC) and forced expiratory volume in the first second (FEVI) were repeated until two maximal efforts reproduced the FEVi and FVC with a difference within 15%. FEVi and FVC were also expressed as the percent of the predicted value for the patient’s height and weight. A ratio of observed FEVi to FVC was also calculated. The Dripps (American Society of Anesthesiologists) classification [6] was independently determined by the anesthesiologist. “Poor general medical condition” and the Goldman cardiac index were defined as previously described by Goldman and colleagues [2]. “Goldman indicators” included an Ss gallop or jugular vein distention, myocardial infarction in the prior six months, cardiac rhythm other than sinus, more than five premature ventricular contractions per minute, 102
February
1990
The American
Journal
of Medicine
Volume
88
CARDIOPULMONARY
SURGICAL
/ GERSON
ET AL
I
predictive for one or more perioperative complications were entered in a stepwise fashion into a multivariate logistic regression procedure [9] to examine the relationships of these variables to perioperative cardiac and pulmonary complications. A p value less than 0.05 was considered to represent statistical significance.
TABLE I Univariate Predictors of Perioperative Major Pulmonary Complications in Geriatric Patients Undergoing Abdominal or Noncardiac Thoracic Surgery* I
RESULTS Perioperative Pulmonary Complications Major perioperative pulmonary complications occurred in 24 study patients (14%). Pneumonia was documented in 23 patients, adult respiratory distress syndrome occurred with pneumonia in two of these patients, and pulmonary embolism was documented in one patient. There were three respiratory deaths. Eight variables were univariately predictive of major perioperative pulmonary complications (Table I). Elderly patients with a major perioperative pulmonary complication had a lower body weight and lower Quetelet index compared to patients without complications. Patients with perioperative pulmonary complications also had a lower preoperative percent of predicted FEVI. Preoperatively, the anesthesiologist assigned 22 of the 24 patients who subsequently had a perioperative pulmonary complication to Dripps class III or IV. However, 85 of the 153 patients (56%) who did not have a perioperative pulmonary complication were also assigned to Dripps class III or IV. The remaining four variables predictive of major perioperative pulmonary complications appear to reflect the preoperative functional capacity of the study patients. Eleven patients were nonambulatory prior to surgery and had more than a threefold increase in postoperative pulmonary complications. The median peak heart rate with supine bicycle exercise was 93 beats/minute in those patients with postoperative pulmonary complications and 112 beats/minute in patients without postoperative pulmonary complications. Inability to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was predictive of major perioperative pulmonary complications in 19 of 24 patients (79%) and was present in only one third of patients who had an uncomplicated course from the pulmonary standpoint. The presence of either “inability to exercise” or one or more “Goldman indicators” identified 22 of 24 patients (92%) with subsequent perioperative pulmonary complications and was also noted in 80 of the 153 patients (52%) who had no pulmonary complication (Figure 1). The latter
Patients with Complications (n = 24)
Patients without Complications (n = 153)
p Value
134 3.0 4(17%) 22(92%) 45.6 93
160 3.7 7(5%) 85 $5;’ 112
0.0026 0.0008 0.045 0.001 0.0420 0.0037
19 (79%)
50(33%)
0.0005
22(92%)
80(52%)
0.0005
Weight (pounds) Quetelet index Bedridden Dripps class Ill or IV Percent of predicted FEVl Peak heart rate (beats/ minute) Inability to bicycle 12 minutes to >99 beats/minute Inability to bicycle t2 minutes to >99 beats/ minute or presence of t 1 “Goldman indicator”
f Lontmuous varlaDles are expressea as mealans. ulscrete varlaOles are expressea as me number of Datlents with the vanable. The oercent of Datients with the variable ISelven in .& parenthesds.
combination performed in a similar manner to the Dripps classification with high sensitivity for prediction of perioperative pulmonary complications but limited specificity. By multiple logistic regression, inability to perform at least two minutes of bicycle exercise raising the heart rate above 99 beats/minute was the only predictor of major perioperative pulmonary complications (p <0.003) (Figure 2). The risk ratio for a major perioperative pulmonary complication was 7.10 for patients who were unable to exercise relative to patients able to exercise (95% confidence levels, 5.10 to 9.89). The subgroup of 54 patients in whom the surgeon ordered preoperative pulmonary function tests was analyzed separately. Major perioperative pulmonary complications occurred in seven of these 54 patients (13.0%) and in 17 of 141 patients in whom pulmonary function testing was not ordered (12.1%, p = NS). Of the data from pulmonary function testing, only the observed percent of predicted FEVl was univariately predictive of major perioperative pulmonary complications. However, when analyzed along with clinical and exercise variables, no variable from pulmonary function testing was independently predictive of ma-
UNABLE TO EXERCISE @ GOLDMAN INDICATOR PRESENT
Figure 1. Perioperative outcome based on a previously proposed combination [3] of preoperative ability to pedal a supine bicycle for at least two minutes and raise the heart rate above 99 beats/ minute and the presence or absence of Goldman cardiac indicators [Z]. PNEUM = pneumonia; ARDS = adult respiratory distress syndrome; PE = pulmonary embolism; CHF = congestive heart failure; VT = ventricular tachycardia; AMI = acute myocardial infarction; Ml = myocardial infarction.
COMPLICATIONS
102 PATIENTS 22 WlTH PULMONARYCOMPLIC*TloNSc 22%) ( 19PNEUM,2 PNE”MIARDS.1PEI 22 WiTH CARDIACCOMPLlCATlONS( 22% I ( 1, CHF,5 VI. 4 AM,. 1 VT + CHF.1 FATALMl + CHF I 6 DEATHS(6%)
I-----P < 0.001
177 PATIENTS
I
February
1990
The American
Journal
of Medicine
Volume
88
103
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
( PC 0.0005
UNABLE TO EXERCISE
/ GERSON
ET AL
PC om4
X-RAY CARD0 tVEGALY
RHYTHM OTHER THAN SINUS
jor perioperative pulmonary complications. Additionally, no data on the arterial blood gas values predicted major perioperative pulmonary complications. Perioperative Cardiac Complications Major perioperative cardiac complications occurred in 25 of 177 study patients (14%). Congestive heart failure was documented in 14 patients, including a patient with a fatal myocardial infarction documented at autopsy. Nonfatal acute myocardial infarction was documented in six patients. Isolated ventricular tachycardia occurred in five additional patients. No study patient developed ventricular fibrillation. Univariate predictors of major perioperative cardiac complications could be categorized into three groups: (1) clinical indicators previously described by Goldman and colleagues [2], (2) high-risk Dripps classification, and (3) indicators of functional capacity including maximal supine bicycle workload and inability to exercise for at least two minutes raising the heart rate above 99 beats/minute [3]. The prevalence of these variables in patients with and without subsequent major perioperative cardiac complications is shown in Table II. By multiple logistic regression, inability to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was the strongest predictor of major perioperative cardiac complications (risk ratio 6.11,95% confidence interval 4.66 to 8.00, p <0.0005) (Figure 2). Additional independent predictive power was provided by the presence of a cardiac rhythm other than sinus (risk ratio 6.96, 95% confidence interval 4.71 to 10.28, p
February
1990
The American
Journal
of Medicine
Volume
S3G/JVD
Figure 2. Independent predictors of major pulmonary, cardiac, and cardiopulmonary perioperative complications. Only inability to pedal a supine bicycle for two minutes and raise the heart rate above 99 beats/minute was predictive of both pulmonary and cardiac complications by multivariate analysis. The sensitivity and specificity of each independently predictive discrete variable are shown. S3G = third heart sound gallop: JVD = jugular venous distention.
or both, occurred in 39 of 177 study patients (22%). Twelve variables were univariately predictive of major cardiopulmonary complications (Table III). Patients who had a major perioperative cardiopulmonary complication were older, had a lower body weight, and had a lower Quetelet index. No clinically applicable cutoff for these variables could be identified that accurately separated high-risk from low-risk patients. Cardiomegaly on the preoperative chest radiograph identified 10 of 39 patients (26%) who had major perioperative cardiopulmonary complications and also identified 17 of 138 patients (12%) who did not have a major perioperative cardiopulmonary complication. An Ss gallop or jugular vein distention was identified in seven of 39 patients (18%) with subsequent cardiopulmonary complications and in only six of 138 patients (4%) who remained free of cardiopulmonary complications. The Dripps classification was III or IV, correctly predicting perioperative cardiopulmonary complications in 33 of 39 patients (84%), but a highrisk Dripps classification was also assigned to the majority of patients who had no perioperative cardiopulmonary complication. The remaining univariate predictors of major postoperative cardiac or pulmonary complications were based on Goldman cardiac risk indicators or variables from bicycle exercise testing, or both. Patients assigned to Goldman cardiac risk index class III or IV had more than twice as many cardiopulmonary complications compared to patients in class I or II; however, fewer than one third of patients who had a major cardiopulmonary complication had a high-risk (III or IV) cardiac risk index. One or more “Goldman indicators” identified 59% of patients who had a cardiopulmonary complication and 36% of patients who did not. Supine bicycle peak heart rate and the maximum bicycle workload were significantly lower in patients who had subsequent cardiopulmonary complications compared to patients who did not. Inability to perform two minutes of bicycle exercise raising the heart rate above 88
CARDIOPULMONARY
99 beats/minute correctly identified high surgical risk in 74% of patients who developed complications while suggesting high risk in only 29% of patients without subsequent complications. Prediction by inability to exercise of subsequent perioperative cardiac and pulmonary complications could not be explained by a bradycardiac effect of cardiac drugs. Of the 29 patients who were unable to exercise and developed a perioperative cardiac or pulmonary complication, only four patients (14%) were receiving a P-blocker or a calcium antagonist. The presence of either a “Goldman indicator” or inability to exercise provided the highest sensitivity for predicting perioperative cardiopulmonary complications of the variables tested but suggested high risk in nearly one half of the patients with an uncomplicated postoperative cardiopulmonary course. By multiple logistic regression, inability to exercise best predicted major perioperative cardiopulmonary complications (risk ratio 6.96,95% confidence interval 5.26 to 9.21, p <0.0005). Additional independent predictive information was provided by the presence of cardiomegaly on chest radiograph (risk ratio 6.62,95% confidence interval 4.86 to 9.03, p <0.004). Patient age also provided independent predictive information for perioperative cardiopulmonary complications (risk ratio 1.09, 95% confidence interval 1.05 to 1.14, p <0.042). However, the difference in median age (74 versus 71 years) between patients with and patients without a subsequent perioperative cardiopulmonary complication provided no clinically useful cutoff for prediction of perioperative risk.
SURGICAL
COMPLICATIONS
ET AL
TABLE II Univariate Predictors of Perioperative Major Cardiac Complications in Geriatric Patients Undergoing Abdominal or Noncardisc Thoracic Surgery Patients with Complications (n = 25) Quetelet Index Sa gallop or jugular vein distention >5 premature ventricular beats/minute Rhythm other than sinus Dripps class Ill or IV Goldman class III or IV 11 “Goldman indicator” Maximum bicycle workload (watts) Inability to bicycle 22 minutes to >99 beats/minute Inability to bicycle 22 minutes to >99 beats/minute or 11 “Goldman indicator”
Patients without Complications (n = 152)
p Value
3.1 6 (24%)
3.7 7 (5%)
0.018 0.004
5 (20%)
10 (7%)
0.042
4(16%)
6 87 20 56
(4%) (57%) (13%) (37%) 38
0.036 0.045 0.001 0.015 0.0319
19 (76%)
50 (33%)
0.0005
22 (88%)
80 (53%)
0.001
;; {g 16 (64;) 0
TABLE Ill Univariate Predictors of Either Perioperative Major Pulmonary or Cardiac Complications in Geriatric Patients Undergoing Abdominal or Noncardiac Thoracic Surgery Patients with Complications (n = 39)
Perioperative Death Six of the 177 study patients died perioperatively (3.4%). The cause of death was established at autopsy in four patients and clinically in two. There was one cardiac death, three pulmonary deaths, and two noncardiac deaths related to bowel infarction or perforation. Five of the six patients who died were unable to exercise for two minutes on the bicycle and raise the heart rate above 99 beats/minute. The one patient who died but was able to exercise died from an aspiration pneumonia. Perioperative mortality occurred in five of 69 patients who were unable to exercise (7.2%) compared to one of 108 patients who were able to exercise (0.9%). COMMENTS Perioperative cardiac and pulmonary complications related to elective surgery are a major cause of morbidity and mortality in the United States [l]. These problems are particularly common in elderly patients and in patients who undergo major abdominal or noncardisc thoracic procedures [2,10]. In addition to excess morbidity and mortality related to cardiac and pulmonary complications, these complications are very costly. Perioperative cardiopulmonary complications commonly result in prolonged stays in the surgical intensive care unit, costly supplemental diagnostic procedures, intensive therapeutic measures, and prolonged hospitalization. Numerous clinical variables and diagnostic test procedures have been evaluated in an attempt to identify preoperatively those patients who are likely to have a major perioperative cardiac or pulmonary complication. In general, these studies have separately evaluat-
/ GERSON
Age (years) Weight (pounds) Quetelet index S3 gallop or jugular vein distention Drrpps class Ill or IV Radiographic cardiomegaly Goldman class Ill or IV 21 “Goldman indicator” Peak heart rate (beats/ minute) Maximum bicycle workload (watts) Inability to bicycle 12 minutes to >99 beats/minute Inability to bicyle 12 minutes to >99 beats/minute or 21 “goldman indrcator”
Patients without Complications (n = 138)
74 137 3.2 7 (18%) 33 10 12 23
(84%) (26%) (31%) (59%) 99
1;; 3.7 6 (4%) 74 17 19 49
0
p Value 0.0285 0.0062 0.0014 0.009
(54%) (12%) (14%) (36%) 112
0.0005 0.043 0.018 0.010 0.0196
50
0.0488
29 (74%)
40 (29%)
0.0005
34 (87%)
68 (49%)
0.0005
ed either preoperative predictors of cardiac complications or preoperative predictors of pulmonary complications. A screening test for identifying patients at high risk for perioperative cardiopulmonary complications should ideally provide a combination of high predictive value for both cardiac and pulmonary complications, low cost, and wide availability, and should provide predictive information not already available from clinical variables. In two previous studies of elderly patients undergoing major elective abdominal or noncardiac thoracic surgery [3,4], it was demonstrated
February
1990
The American
Journal
of Medicine
Volume
88
105
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
that objective assessment of functional capacity using a simple supine bicycle exercise test could provide improved prediction of major perioperative cardiac complications compared to the patient history, physical examination, chest radiograph, electrocardiogram, Goldman cardiac risk index, and Dripps American Society of Anesthesiologist’s classification. The present study confirms and extends those observations by demonstrating that the same criteria from supine bicycle exercise testing provide information that is independently predictive of perioperative pulmonary complications in elderly patients having abdominal or noncardiac thoracic surgery. Specifically, elderly patients who were able to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute had a fivefold reduction in major perioperative pulmonary complications, a sixfold reduction in major cardiac complications, and a fivefold reduction in combined cardiopulmonary complications compared to patients who underwent major abdominal or noncardiac thoracic surgery but who were unable to perform that level of exercise. No other combination of clinical, radiographic, or electrocardiographic variables provided comparable predictive power. The mechanism by which inability to exercise predicts major perioperative cardiopulmonary complications remains unclear but is probably multifactorial. For patients who developed a pulmonary complication, the reason for not completing at least two minutes of bicycle exercise and raising the heart rate above 99 beats/minute was weakness or amputation in 10 patients, arthritis or leg pain in two patients, claudication in three patients, inability to follow instruction in one patient, and multifactorial in three patients. No patient stopped exercising because of chest pain or limiting dyspnea. Previous studies have demonstrated relationships between perioperative pulmonary complications and preoperative variables including obesity [lo], chronic obstructive pulmonary disease [11,12], type of operation [11,13], cigarette smoking [lo], measurements of pulmonary function [13,14], and arterial oxygen tension and carbon dioxide tension measurements [10,13,14]. Previous studies have also suggested that measurement of maximum oxygen consumption during exercise may be of value in predicting perioperative pulmonary complications related to thoracotomy [15,16]. Previous studies have not examined exercise testing as a predictor of major perioperative pulmonary complications related to abdominal surgery. The present study differs from previous studies in that major perioperative pulmonary complications were more closely related to low body weight rather than to obesity. All study patients had abdominal or thoracic surgery that placed them at high risk for perioperative pulmonary complications. Cigarette smoking was not examined as a risk factor for major perioperative pulmonary complications in the present study. In this study, patients selected by the attending surgeon to undergo pulmonary function testing or arterial blood gas measurement prior to elective surgery had no more frequent perioperative pulmonary complications than those patients in whom these preoperative tests were deemed unnecessary. Although the percent of predicted FEVi was univariately predictive of perioperative pulmonary complications, it was less pre106
February
1990
The American
Journal
of Medicine
Volume
dictive compared to data from bicycle exercise testing. Arterial blood gas measurements were not predictive of perioperative pulmonary complications. In a larger study, retention of carbon dioxide on arterial blood gas measurement might prove predictive of perioperative pulmonary complications, but this was a highly infrequent observation in this series of patients preparing for elective abdominal or thoracic surgery. Importantly, the same findings from the preoperative bicycle exercise test that best predicted perioperative pulmonary complications in this study have provided independent prediction of perioperative cardiac complications in this and previous [3,4] studies. In this study, 92% of patients who had a perioperative major pulmonary complication and 88% of patients who had a major perioperative cardiac complication were either unable to exercise for two minutes and raise the heart rate above 99 beats/minute or had a Goldman cardiac risk indicator. These latter variables were more specific for cardiac and pulmonary endpoints than other combinations of variables with comparable sensitivity for predicting these endpoints. Further study is required before these findings can be recommended for routine clinical application. The findings suggest, however, that geriatric patients with no Goldman risk factors who are able to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute are at low risk for major perioperative cardiac or pulmonary complications and need not undergo extensive additional testing. In contrast, patients with Goldman cardiac risk indicators or poor exercise capacity are likely to benefit from further preoperative evaluation by an internist, cardiologist, or pulmonary specialist. Previous studies have demonstrated that appropriately selected patients with severe coronary artery disease can undergo major elective surgery without increased cardiac risk following successful coronary artery bypass surgery [17,18]. Patients with severe chronic pulmonary disease have a reduced perioperative rate of major pulmonary complications if they undergo intensive preoperative preparation including cessation of cigarette smoking and use of incentive spirometry [19-221. An important limitation of the present study requires comment. Inability to perform two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute is a predictor of perioperative cardiopulmonary complications that was identified and tested in patients aged 65 years or older. There is no evidence that this threshold for exercise capacity has any clinical relevance in younger patients. Separate studies are needed to identify specific predictors from exercise testing for perioperative complications in younger individuals. We conclude that objective measurement of exercise capacity by supine bicycle exercise testing provides information predictive of major perioperative cardiac and pulmonary complications in elderly patients who undergo major abdominal or noncardiac thoracic surgery. Therefore, exercise testing may be of clinical value for preoperative screening prior to elective surgery in elderly patients.
REFERENCES 1. Vital and Health Statistics, U.S. Department of Health and Human Serwces. Series 13, no. 88. Washington, D.C.. U.S. Government Printing Office, 1983; 1417.
68
CARDIOPULMONARY 2. Goldman L, Caldera DL, Nussbaum SR. eta/: Multrfactorral index of cardrac rrsk in noncardiac surgical procedures. N Engl J Med 1977; 297: 8455850. 3. Gerson MC, Hurst JM, Hertzberg VS, et al: Cardiac prognosis rn noncardiac genatrrc surgery. Ann Intern Med 1985; 103: 832-837. 4. Gerson MC, Hurst JM, Hertzberg VS. Ellis KS: Cardrac rusk stratrfication prror to geriatric abdomrnal or noncardiac thoracic surgery (abstr). Circulatron 1988; 78(suppl II). 11-108. 5. Benn RT: Some mathematical propertres of weight-for-height Indices used as measures of adiposity. Br J Prev Sot Med 1971; 25. 42-50. 6. House of Delegates of the American Society of Anesthesrologrsts, Inc: New classrficatron of physrcal status. Anesthesiology 1963; 24: 111. 7. Brello DR, Mattar AG, McKnight RC. Siegel BA. Ventilatron-perfusion studies rn suspected pulmonary embolism AJR 1979, 133: 1033-1037. 8. SAS Institute Inc: SAS users guide. Basics and statistrcs. versron 5 edrtion. Cary. North Carolrna: SAS Institute Inc, 1985; 402-506. 9. Breslow NE, Day NE: Statistical methods in cancer research: volume l-the analysrs of case-control studies InternatIonal Agency for Research on Cancer, Lyon, France, 1980; 192-279. 10. Tisr GM: Preoperabve evaluation of pulmonary functron. Validity. indications, and benefits. Am Rev Respir Dis 1979: 119: 293-310. 11. Tarhan S, Moffrtt EA. Sessler AD, Douglas WW. Taylor WF: Risk of anesthesra and surgery In patients with chronic bronchrtis and chronic obstructrve pulmonary drsease. Surgery 1973: 74: 720-726. 12. Wrllrams CD, Brenowrtz JB: “Prohibitrve” lung functron and major surgical procedures. Am J Surg 1976; 132: 763-766. 13. Latimer RG. Drckman M, Day WC, Gunn ML, Schmidt CD: Ventrlatory patterns
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
and pulmonary complications after upper abdomrnal surgery determrned by preoperative and postoperative computerized sprrometry and blood gas analysis. Am J Surg 1971: 122: 622-632. 14. Stein M, Koota GM, Simon M. Frank HA Pulmonary evaluatron of surgrcal patients. JAMA 1962; 181: 7655770. 15. Eugene J, Brown SE, Light RW. Mrlne NE, Stemmer EA: Maximum oxygen consumptron: a physrologic guide to pulmonary resectron. Surgtcal Forum 1982; 33: 260-262. 16. Smith TP, Krnasewrtz GT, Tucker WY, Spellers WP. George RB: Exercise capacrty as a predrctor of post-thoracotomy morbidity. Am Rev Respir DIS 1984; 129: 730-734. 17. Mahar LJ, Steen PA, Tinker JH, et al: Perroperatrve myocardial Infarctron in patients with coronary artery disease with and wrthout aorta-coronary artery bypass grafts. J Thorac Cardiovasc Surg 1978; 76: 533-537. 18. McCollum CH. Garcra-Rinaldi R, Graham JM, DeBakey ME: Myocardral revascularization prror to subsequent major surgery In patients wrth coronary artery disease. Surgery 1977; 81: 302-304. 19. Bartlett RH. Brennan ML, Gauanrga AB. Hanson EL. Studres on the pathogenesis and prevention of postoperative pulmonary complications. Surg Gynecol Obstet 1973; 137: l-9. 20. Craven JL. Evans GA, Davenport JL, Willrams RHP: The evaluation of the Incentive spirometer in the management of postoperabve pulmonary complications. Br J Surg 1974; 61: 793-797. 21. Dohr S. Gold Ml: Comparison of two methods of postoperabve respiratory care. Chest 1978; 73: 592-595. 22. Thoren L: Post-operatrve pulmonary complrcations; observatrons on therr prevention by means of physiotherapy. Acta Char Stand 1954; 107: 193-205.
February
1990
The American
Journal
of Medicine
Volume
88
107
PURPOSE:Major cardiac and pulmonary complications associated with abdominal and noncardiac thoracic surgery are a common cause of mortality and serious morbidity in elderly patients. We postulated that a simple, inexpensive bicycle exercise test could provide objective documentation of cardiopulmonary reserve and, therefore, predict perioperative pulmonary as well as cardiac complications. PATIENTsANDMETHODs:hiOr to elective surgery, 177 patients aged 65 years or older had assessment of the clinical history, results of physical examination, electrocardiogram, chest radiograph, blood chemistries, pulmonary function test fiidings, supine exercise test results, Dripps classification, and Goldman cardiac risk factors. Observations in patients with and without major perioperative cardiac and/or pulmonary complications were compared using univariate analysis followed by a multivariate logistic regression procedure. RESULTS:Major perioperative complications were pulmonary in 24 patients, cardiac in 25 patients, and either cardiac or pulmonary in 39 patients. By multivariate analysis, inability to perform two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was the best predictor of perioperative pulmonary, cardiac, and combined cardiopulmonary complications (p <0.0005). Among 108 patients who were able to achieve these exercise criteria, cardiac or pulmonary complications occurred in 10 patients (9.3%), with one death (0.9%). Among 69 patients unable to exercise satisfactorily, cardiac or pulmonary complications occurred in 29 patients (42%), with five total deaths (7.2%). CONCLUSION: Objective measurement of exercise capacity by supine bicycle ergometry appears to be of clinical value for preoperative risk stratification for both pulmonary and cardiac complications prior to maior elective abdominal or noncardiac thoracic surgery in elderly patients.
From the Departments of Internal Medicine (Cardiology [MCG, KE, EEF, MSCl. Pulmonarv IRBl. and General Internal Medicine lGWR1 Divisions). Surg&y (JMH). ‘E&dnmental Health (Epidemiology ‘and eiostattstlds [VSH]). and Radiology (Eugene L. Saenger Radtoisotope Laboratory [MCG, PMB]), University of Cincinnati Medical Center and the Veterans Admlnistration Medical Center, Cincinnati, Ohio. This workwas supported in part by a grant from the Southwestern Ohio Chapter of the Amencan Heart ASSOCIation, Clnclnnatl. Ohio. Requests for reprints should be addressed to Myron C. Gerson, M.D., University of Cincinnati, Division of Cardiology, Mail Location #542, Clncinnatl. Ohio 45267. Manuscript submitted May 1.1989, and accepted in revised form December 6, 1989.
T
he elderly constitute an increasingly large segment of the population of the United States. More than 600,000 individuals 65 years of age and older require major abdominal or noncardiac thoracic surgery each year [l]. Approximately 10% of these patients develop a major cardiac complication related to surgery [2], and the number of patients developing major perioperative pulmonary complications may be even greater. In a previous study of elderly patients preparing for major abdominal and noncardiac thoracic surgery [3], it was observed that a simple, inexpensive bicycle exercise test provided information that was more predictive for perioperative cardiac complications than data available from the clinical history, physical examination, chest radiograph, and electrocardiogram. In these patients, objective measurement of exercise capacity has been a better predictive screening procedure for major perioperative cardiac complications compared to more costly and more complex tests, including rest and exercise radionuclide ventriculography [3] and dipyridamole thallium imaging [4]. The purpose of the present study was to determine whether previously established exercise test criteria [3] for predicting perioperative cardiac complications could also provide information predictive of perioperative pulmonary complications.
PATIENTS AND METHODS The study population consisted of 87 men and 90 women aged 65 years or older (mean age, 73 years) scheduled for major elective abdominal or noncardiac thoracic surgery. Patients were consecutively entered into a study of cardiac risk provided that (1) there was sufficient time between scheduling and performing of surgery to permit preoperative testing, (2) surgery involved entry into the pleural or peritoneal cavity (minor procedures such as simple appendectomy, feeding tube placement, and inguinal herniorrhaphy were excluded), and (3) the patient consented to study participation. Surgical procedures were abdominal in 161 patients and thoracic in 16 patients. During the enrollment period for the study, 53% of patients aged 65 years or older who were having major abdominal or noncardiac thoracic surgery at the University of Cincinnati Hospitals were studied. Eligible patients not enrolled in the study had a mean age of 72 years, included 51% females, and had abdominal surgery in 96% of cases (all p = NS compared to study patients). In a previous study [3] of 100 elderly patients preparing for noncardiac surgery, it was found that inability both to perform two minutes of supine bicycle exercise and to raise the heart rate above 99 beats/minute during bicycle exercise was a powerful predictor of February
1990
The American
Journal
of Medicine
Volume
88
101
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
“poor general medical condition,” and clinically important valvular aortic stenosis [2]. All study patients were of an advanced age (in this case greater than 64 years) and underwent abdominal or noncardiac thoracic surgery. No study patient underwent emergency surgery. Exercise electrocardiographic testing was performed with the supine patient pedaling a bicycle ergometer at 50 revolutions/minute, beginning at a workload of 25 watts and, if tolerated, increasing the workload by 12.5 watts every minute until there was evidence of limiting dyspnea, chest pain, or fatigue. Exercise ST-segment levels were monitored in leads I, II, III, and Vg, A positive ST-segment response required a normal baseline ST-segment at rest and during hyperventilation with at least 1 mm of horizontal or downsloping ST-segment depression at 0.08 second after the J point, during or following exercise. On the basis of a previous study of 100 separate geriatric patients preparing for abdominal or noncardiac thoracic surgery, “inability to exercise” was defined as inability to pedal a supine bicycle for at least two minutes and raise the heart rate above 99 beats/minute [3]. Postoperative rhythm strips, chest radiographs, results of physical examinations, and catheterization data were reviewed up to the time of hospital discharge for evidence of cardiac or pulmonary complications. Serum creatine kinase and lactic dehydrogenase with isoenzyme determinations were measured at 0,6, 12,24,48, and 72 hours following surgery. Electrocardiograms were obtained on the first, second, and third postoperative days. Major perioperative cardiac complications included myocardial infarction, documented by new infarction Q waves or the presence of creatine kinase MB fraction by electrophoresis; cardiac death; ventricular tachycardia, defined as three or more consecutive ventricular beats at a heart rate greater than 140 beats/minute, or ventricular fibrillation; and left heart failure, defined as bilateral rales and new evidence of left heart failure documented by typical radiographic changes, or by a pulmonary capillary wedge pressure greater than 25 mm Hg. Major perioperative pulmonary complications were defined clinically or at autopsy as: (1) pneumonia (defined clinically by a new pulmonary infiltrate on chest radiogram plus at least two of the following: [a] fever 37.7’C or higher; [b] white blood cell count greater than 10,50O/mms; [c] initiation of antibiotic treatment; [d] demonstration of pathogenic organisms on a sputum Gram stain or culture); (2) noncardiogenic pulmonary edema (defined clinically by chest radiogram documentation of interstitial edema at a time when the pulmonary capillary wedge pressure is 16 mm Hg or less and the patient has not been treated for left heart failure); or (3) pulmonary embolism (documented by pulmonary arteriogram or autopsy, or supported by a ventilation/perfusion radioisotope scan showing “high probability” [7] of pulmonary embolism and followed by institution of heparin therapy). Observations in patients with and without perioperative cardiac and/or pulmonary complications were compared using the Statistical Analysis System [8]. Continuous variables were presented using median values and were tested for univariate significance using a two-tailed Wilcoxon rank sum statistic. Discrete variables were tested using a two-tailed Fisher’s exact test. Preoperative variables that were univariately
perioperative cardiac complications. Because the “inability to exercise” criterion was retrospectively derived in those 100 patients and because patients in that group who had perioperative cardiac complications also had an increased incidence of perioperative pulmonary complications, those 100 patients were excluded from the present study population. The “inability to exercise” criterion was prospectively tested for prediction of perioperative cardiac complications in 55 additional patients in that study [3]. These 55 additional patients together with 122 previously unreported patients form the present study population. Thus, the present study is not prospective because data had been collected (but not analyzed) from those 55 patients before the present study hypothesis regarding prediction of perioperative pulmonary complications was generated. Prior to surgery, the patient’s age, sex, ambulatory status, type of operation, duration of anesthesia, and the presence of chest pain or myocardial infarction in the preceding six months were noted from the medical history. Each patient was examined by a cardiologist for an S3 gallop; jugular vein distention, as indicated by a column of blood in the internal jugular vein extending more than 4 cm above the manubrium with the patient lying at a 45” angle; a systolic murmur and carotid pulse suggestive of clinically important valvular aortic stenosis; and a liver span greater than 12 cm, or other findings of chronic liver disease. Height and weight were recorded for each patient. A Quetelet index was calculated as 100 times the patient’s weight in pounds divided by the square of the height in inches [5]. The resting 12-lead electrocardiogram was analyzed in a blind manner for infarction Q waves (characterized by a duration greater than 0.04 second and a depth greater than or equal to 1 mV) and for “rhythm other than sinus” [2]. All preoperative rhythm strips including a one-minute strip obtained in the Exercise Laboratory were reviewed for the presence of greater than five premature ventricular contractions per minute [2]. Without reference to the clinical data, a staff radiologist reviewed the preoperative chest radiographs for cardiomegaly, pulmonary congestion, or pulmonary infiltrate. Each study patient had a preoperative measurement of serum creatinine, blood urea nitrogen, and serum electrolytes; and a serum creatine kinase and lactic dehydrogenase with isoenzyme measurements. Other preoperative laboratory data including arterial blood gases, pulmonary function tests, and hepatic enzyme levels were ordered by the surgeon if deemed appropriate. Pulmonary function measurements of forced vital capacity (FVC) and forced expiratory volume in the first second (FEVI) were repeated until two maximal efforts reproduced the FEVi and FVC with a difference within 15%. FEVi and FVC were also expressed as the percent of the predicted value for the patient’s height and weight. A ratio of observed FEVi to FVC was also calculated. The Dripps (American Society of Anesthesiologists) classification [6] was independently determined by the anesthesiologist. “Poor general medical condition” and the Goldman cardiac index were defined as previously described by Goldman and colleagues [2]. “Goldman indicators” included an Ss gallop or jugular vein distention, myocardial infarction in the prior six months, cardiac rhythm other than sinus, more than five premature ventricular contractions per minute, 102
February
1990
The American
Journal
of Medicine
Volume
88
CARDIOPULMONARY
SURGICAL
/ GERSON
ET AL
I
predictive for one or more perioperative complications were entered in a stepwise fashion into a multivariate logistic regression procedure [9] to examine the relationships of these variables to perioperative cardiac and pulmonary complications. A p value less than 0.05 was considered to represent statistical significance.
TABLE I Univariate Predictors of Perioperative Major Pulmonary Complications in Geriatric Patients Undergoing Abdominal or Noncardiac Thoracic Surgery* I
RESULTS Perioperative Pulmonary Complications Major perioperative pulmonary complications occurred in 24 study patients (14%). Pneumonia was documented in 23 patients, adult respiratory distress syndrome occurred with pneumonia in two of these patients, and pulmonary embolism was documented in one patient. There were three respiratory deaths. Eight variables were univariately predictive of major perioperative pulmonary complications (Table I). Elderly patients with a major perioperative pulmonary complication had a lower body weight and lower Quetelet index compared to patients without complications. Patients with perioperative pulmonary complications also had a lower preoperative percent of predicted FEVI. Preoperatively, the anesthesiologist assigned 22 of the 24 patients who subsequently had a perioperative pulmonary complication to Dripps class III or IV. However, 85 of the 153 patients (56%) who did not have a perioperative pulmonary complication were also assigned to Dripps class III or IV. The remaining four variables predictive of major perioperative pulmonary complications appear to reflect the preoperative functional capacity of the study patients. Eleven patients were nonambulatory prior to surgery and had more than a threefold increase in postoperative pulmonary complications. The median peak heart rate with supine bicycle exercise was 93 beats/minute in those patients with postoperative pulmonary complications and 112 beats/minute in patients without postoperative pulmonary complications. Inability to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was predictive of major perioperative pulmonary complications in 19 of 24 patients (79%) and was present in only one third of patients who had an uncomplicated course from the pulmonary standpoint. The presence of either “inability to exercise” or one or more “Goldman indicators” identified 22 of 24 patients (92%) with subsequent perioperative pulmonary complications and was also noted in 80 of the 153 patients (52%) who had no pulmonary complication (Figure 1). The latter
Patients with Complications (n = 24)
Patients without Complications (n = 153)
p Value
134 3.0 4(17%) 22(92%) 45.6 93
160 3.7 7(5%) 85 $5;’ 112
0.0026 0.0008 0.045 0.001 0.0420 0.0037
19 (79%)
50(33%)
0.0005
22(92%)
80(52%)
0.0005
Weight (pounds) Quetelet index Bedridden Dripps class Ill or IV Percent of predicted FEVl Peak heart rate (beats/ minute) Inability to bicycle 12 minutes to >99 beats/minute Inability to bicycle t2 minutes to >99 beats/ minute or presence of t 1 “Goldman indicator”
f Lontmuous varlaDles are expressea as mealans. ulscrete varlaOles are expressea as me number of Datlents with the vanable. The oercent of Datients with the variable ISelven in .& parenthesds.
combination performed in a similar manner to the Dripps classification with high sensitivity for prediction of perioperative pulmonary complications but limited specificity. By multiple logistic regression, inability to perform at least two minutes of bicycle exercise raising the heart rate above 99 beats/minute was the only predictor of major perioperative pulmonary complications (p <0.003) (Figure 2). The risk ratio for a major perioperative pulmonary complication was 7.10 for patients who were unable to exercise relative to patients able to exercise (95% confidence levels, 5.10 to 9.89). The subgroup of 54 patients in whom the surgeon ordered preoperative pulmonary function tests was analyzed separately. Major perioperative pulmonary complications occurred in seven of these 54 patients (13.0%) and in 17 of 141 patients in whom pulmonary function testing was not ordered (12.1%, p = NS). Of the data from pulmonary function testing, only the observed percent of predicted FEVl was univariately predictive of major perioperative pulmonary complications. However, when analyzed along with clinical and exercise variables, no variable from pulmonary function testing was independently predictive of ma-
UNABLE TO EXERCISE @ GOLDMAN INDICATOR PRESENT
Figure 1. Perioperative outcome based on a previously proposed combination [3] of preoperative ability to pedal a supine bicycle for at least two minutes and raise the heart rate above 99 beats/ minute and the presence or absence of Goldman cardiac indicators [Z]. PNEUM = pneumonia; ARDS = adult respiratory distress syndrome; PE = pulmonary embolism; CHF = congestive heart failure; VT = ventricular tachycardia; AMI = acute myocardial infarction; Ml = myocardial infarction.
COMPLICATIONS
102 PATIENTS 22 WlTH PULMONARYCOMPLIC*TloNSc 22%) ( 19PNEUM,2 PNE”MIARDS.1PEI 22 WiTH CARDIACCOMPLlCATlONS( 22% I ( 1, CHF,5 VI. 4 AM,. 1 VT + CHF.1 FATALMl + CHF I 6 DEATHS(6%)
I-----P < 0.001
177 PATIENTS
I
February
1990
The American
Journal
of Medicine
Volume
88
103
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
( PC 0.0005
UNABLE TO EXERCISE
/ GERSON
ET AL
PC om4
X-RAY CARD0 tVEGALY
RHYTHM OTHER THAN SINUS
jor perioperative pulmonary complications. Additionally, no data on the arterial blood gas values predicted major perioperative pulmonary complications. Perioperative Cardiac Complications Major perioperative cardiac complications occurred in 25 of 177 study patients (14%). Congestive heart failure was documented in 14 patients, including a patient with a fatal myocardial infarction documented at autopsy. Nonfatal acute myocardial infarction was documented in six patients. Isolated ventricular tachycardia occurred in five additional patients. No study patient developed ventricular fibrillation. Univariate predictors of major perioperative cardiac complications could be categorized into three groups: (1) clinical indicators previously described by Goldman and colleagues [2], (2) high-risk Dripps classification, and (3) indicators of functional capacity including maximal supine bicycle workload and inability to exercise for at least two minutes raising the heart rate above 99 beats/minute [3]. The prevalence of these variables in patients with and without subsequent major perioperative cardiac complications is shown in Table II. By multiple logistic regression, inability to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute was the strongest predictor of major perioperative cardiac complications (risk ratio 6.11,95% confidence interval 4.66 to 8.00, p <0.0005) (Figure 2). Additional independent predictive power was provided by the presence of a cardiac rhythm other than sinus (risk ratio 6.96, 95% confidence interval 4.71 to 10.28, p
February
1990
The American
Journal
of Medicine
Volume
S3G/JVD
Figure 2. Independent predictors of major pulmonary, cardiac, and cardiopulmonary perioperative complications. Only inability to pedal a supine bicycle for two minutes and raise the heart rate above 99 beats/minute was predictive of both pulmonary and cardiac complications by multivariate analysis. The sensitivity and specificity of each independently predictive discrete variable are shown. S3G = third heart sound gallop: JVD = jugular venous distention.
or both, occurred in 39 of 177 study patients (22%). Twelve variables were univariately predictive of major cardiopulmonary complications (Table III). Patients who had a major perioperative cardiopulmonary complication were older, had a lower body weight, and had a lower Quetelet index. No clinically applicable cutoff for these variables could be identified that accurately separated high-risk from low-risk patients. Cardiomegaly on the preoperative chest radiograph identified 10 of 39 patients (26%) who had major perioperative cardiopulmonary complications and also identified 17 of 138 patients (12%) who did not have a major perioperative cardiopulmonary complication. An Ss gallop or jugular vein distention was identified in seven of 39 patients (18%) with subsequent cardiopulmonary complications and in only six of 138 patients (4%) who remained free of cardiopulmonary complications. The Dripps classification was III or IV, correctly predicting perioperative cardiopulmonary complications in 33 of 39 patients (84%), but a highrisk Dripps classification was also assigned to the majority of patients who had no perioperative cardiopulmonary complication. The remaining univariate predictors of major postoperative cardiac or pulmonary complications were based on Goldman cardiac risk indicators or variables from bicycle exercise testing, or both. Patients assigned to Goldman cardiac risk index class III or IV had more than twice as many cardiopulmonary complications compared to patients in class I or II; however, fewer than one third of patients who had a major cardiopulmonary complication had a high-risk (III or IV) cardiac risk index. One or more “Goldman indicators” identified 59% of patients who had a cardiopulmonary complication and 36% of patients who did not. Supine bicycle peak heart rate and the maximum bicycle workload were significantly lower in patients who had subsequent cardiopulmonary complications compared to patients who did not. Inability to perform two minutes of bicycle exercise raising the heart rate above 88
CARDIOPULMONARY
99 beats/minute correctly identified high surgical risk in 74% of patients who developed complications while suggesting high risk in only 29% of patients without subsequent complications. Prediction by inability to exercise of subsequent perioperative cardiac and pulmonary complications could not be explained by a bradycardiac effect of cardiac drugs. Of the 29 patients who were unable to exercise and developed a perioperative cardiac or pulmonary complication, only four patients (14%) were receiving a P-blocker or a calcium antagonist. The presence of either a “Goldman indicator” or inability to exercise provided the highest sensitivity for predicting perioperative cardiopulmonary complications of the variables tested but suggested high risk in nearly one half of the patients with an uncomplicated postoperative cardiopulmonary course. By multiple logistic regression, inability to exercise best predicted major perioperative cardiopulmonary complications (risk ratio 6.96,95% confidence interval 5.26 to 9.21, p <0.0005). Additional independent predictive information was provided by the presence of cardiomegaly on chest radiograph (risk ratio 6.62,95% confidence interval 4.86 to 9.03, p <0.004). Patient age also provided independent predictive information for perioperative cardiopulmonary complications (risk ratio 1.09, 95% confidence interval 1.05 to 1.14, p <0.042). However, the difference in median age (74 versus 71 years) between patients with and patients without a subsequent perioperative cardiopulmonary complication provided no clinically useful cutoff for prediction of perioperative risk.
SURGICAL
COMPLICATIONS
ET AL
TABLE II Univariate Predictors of Perioperative Major Cardiac Complications in Geriatric Patients Undergoing Abdominal or Noncardisc Thoracic Surgery Patients with Complications (n = 25) Quetelet Index Sa gallop or jugular vein distention >5 premature ventricular beats/minute Rhythm other than sinus Dripps class Ill or IV Goldman class III or IV 11 “Goldman indicator” Maximum bicycle workload (watts) Inability to bicycle 22 minutes to >99 beats/minute Inability to bicycle 22 minutes to >99 beats/minute or 11 “Goldman indicator”
Patients without Complications (n = 152)
p Value
3.1 6 (24%)
3.7 7 (5%)
0.018 0.004
5 (20%)
10 (7%)
0.042
4(16%)
6 87 20 56
(4%) (57%) (13%) (37%) 38
0.036 0.045 0.001 0.015 0.0319
19 (76%)
50 (33%)
0.0005
22 (88%)
80 (53%)
0.001
;; {g 16 (64;) 0
TABLE Ill Univariate Predictors of Either Perioperative Major Pulmonary or Cardiac Complications in Geriatric Patients Undergoing Abdominal or Noncardiac Thoracic Surgery Patients with Complications (n = 39)
Perioperative Death Six of the 177 study patients died perioperatively (3.4%). The cause of death was established at autopsy in four patients and clinically in two. There was one cardiac death, three pulmonary deaths, and two noncardiac deaths related to bowel infarction or perforation. Five of the six patients who died were unable to exercise for two minutes on the bicycle and raise the heart rate above 99 beats/minute. The one patient who died but was able to exercise died from an aspiration pneumonia. Perioperative mortality occurred in five of 69 patients who were unable to exercise (7.2%) compared to one of 108 patients who were able to exercise (0.9%). COMMENTS Perioperative cardiac and pulmonary complications related to elective surgery are a major cause of morbidity and mortality in the United States [l]. These problems are particularly common in elderly patients and in patients who undergo major abdominal or noncardisc thoracic procedures [2,10]. In addition to excess morbidity and mortality related to cardiac and pulmonary complications, these complications are very costly. Perioperative cardiopulmonary complications commonly result in prolonged stays in the surgical intensive care unit, costly supplemental diagnostic procedures, intensive therapeutic measures, and prolonged hospitalization. Numerous clinical variables and diagnostic test procedures have been evaluated in an attempt to identify preoperatively those patients who are likely to have a major perioperative cardiac or pulmonary complication. In general, these studies have separately evaluat-
/ GERSON
Age (years) Weight (pounds) Quetelet index S3 gallop or jugular vein distention Drrpps class Ill or IV Radiographic cardiomegaly Goldman class Ill or IV 21 “Goldman indicator” Peak heart rate (beats/ minute) Maximum bicycle workload (watts) Inability to bicycle 12 minutes to >99 beats/minute Inability to bicyle 12 minutes to >99 beats/minute or 21 “goldman indrcator”
Patients without Complications (n = 138)
74 137 3.2 7 (18%) 33 10 12 23
(84%) (26%) (31%) (59%) 99
1;; 3.7 6 (4%) 74 17 19 49
0
p Value 0.0285 0.0062 0.0014 0.009
(54%) (12%) (14%) (36%) 112
0.0005 0.043 0.018 0.010 0.0196
50
0.0488
29 (74%)
40 (29%)
0.0005
34 (87%)
68 (49%)
0.0005
ed either preoperative predictors of cardiac complications or preoperative predictors of pulmonary complications. A screening test for identifying patients at high risk for perioperative cardiopulmonary complications should ideally provide a combination of high predictive value for both cardiac and pulmonary complications, low cost, and wide availability, and should provide predictive information not already available from clinical variables. In two previous studies of elderly patients undergoing major elective abdominal or noncardiac thoracic surgery [3,4], it was demonstrated
February
1990
The American
Journal
of Medicine
Volume
88
105
CARDIOPULMONARY
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
that objective assessment of functional capacity using a simple supine bicycle exercise test could provide improved prediction of major perioperative cardiac complications compared to the patient history, physical examination, chest radiograph, electrocardiogram, Goldman cardiac risk index, and Dripps American Society of Anesthesiologist’s classification. The present study confirms and extends those observations by demonstrating that the same criteria from supine bicycle exercise testing provide information that is independently predictive of perioperative pulmonary complications in elderly patients having abdominal or noncardiac thoracic surgery. Specifically, elderly patients who were able to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute had a fivefold reduction in major perioperative pulmonary complications, a sixfold reduction in major cardiac complications, and a fivefold reduction in combined cardiopulmonary complications compared to patients who underwent major abdominal or noncardiac thoracic surgery but who were unable to perform that level of exercise. No other combination of clinical, radiographic, or electrocardiographic variables provided comparable predictive power. The mechanism by which inability to exercise predicts major perioperative cardiopulmonary complications remains unclear but is probably multifactorial. For patients who developed a pulmonary complication, the reason for not completing at least two minutes of bicycle exercise and raising the heart rate above 99 beats/minute was weakness or amputation in 10 patients, arthritis or leg pain in two patients, claudication in three patients, inability to follow instruction in one patient, and multifactorial in three patients. No patient stopped exercising because of chest pain or limiting dyspnea. Previous studies have demonstrated relationships between perioperative pulmonary complications and preoperative variables including obesity [lo], chronic obstructive pulmonary disease [11,12], type of operation [11,13], cigarette smoking [lo], measurements of pulmonary function [13,14], and arterial oxygen tension and carbon dioxide tension measurements [10,13,14]. Previous studies have also suggested that measurement of maximum oxygen consumption during exercise may be of value in predicting perioperative pulmonary complications related to thoracotomy [15,16]. Previous studies have not examined exercise testing as a predictor of major perioperative pulmonary complications related to abdominal surgery. The present study differs from previous studies in that major perioperative pulmonary complications were more closely related to low body weight rather than to obesity. All study patients had abdominal or thoracic surgery that placed them at high risk for perioperative pulmonary complications. Cigarette smoking was not examined as a risk factor for major perioperative pulmonary complications in the present study. In this study, patients selected by the attending surgeon to undergo pulmonary function testing or arterial blood gas measurement prior to elective surgery had no more frequent perioperative pulmonary complications than those patients in whom these preoperative tests were deemed unnecessary. Although the percent of predicted FEVi was univariately predictive of perioperative pulmonary complications, it was less pre106
February
1990
The American
Journal
of Medicine
Volume
dictive compared to data from bicycle exercise testing. Arterial blood gas measurements were not predictive of perioperative pulmonary complications. In a larger study, retention of carbon dioxide on arterial blood gas measurement might prove predictive of perioperative pulmonary complications, but this was a highly infrequent observation in this series of patients preparing for elective abdominal or thoracic surgery. Importantly, the same findings from the preoperative bicycle exercise test that best predicted perioperative pulmonary complications in this study have provided independent prediction of perioperative cardiac complications in this and previous [3,4] studies. In this study, 92% of patients who had a perioperative major pulmonary complication and 88% of patients who had a major perioperative cardiac complication were either unable to exercise for two minutes and raise the heart rate above 99 beats/minute or had a Goldman cardiac risk indicator. These latter variables were more specific for cardiac and pulmonary endpoints than other combinations of variables with comparable sensitivity for predicting these endpoints. Further study is required before these findings can be recommended for routine clinical application. The findings suggest, however, that geriatric patients with no Goldman risk factors who are able to perform at least two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute are at low risk for major perioperative cardiac or pulmonary complications and need not undergo extensive additional testing. In contrast, patients with Goldman cardiac risk indicators or poor exercise capacity are likely to benefit from further preoperative evaluation by an internist, cardiologist, or pulmonary specialist. Previous studies have demonstrated that appropriately selected patients with severe coronary artery disease can undergo major elective surgery without increased cardiac risk following successful coronary artery bypass surgery [17,18]. Patients with severe chronic pulmonary disease have a reduced perioperative rate of major pulmonary complications if they undergo intensive preoperative preparation including cessation of cigarette smoking and use of incentive spirometry [19-221. An important limitation of the present study requires comment. Inability to perform two minutes of supine bicycle exercise raising the heart rate above 99 beats/minute is a predictor of perioperative cardiopulmonary complications that was identified and tested in patients aged 65 years or older. There is no evidence that this threshold for exercise capacity has any clinical relevance in younger patients. Separate studies are needed to identify specific predictors from exercise testing for perioperative complications in younger individuals. We conclude that objective measurement of exercise capacity by supine bicycle exercise testing provides information predictive of major perioperative cardiac and pulmonary complications in elderly patients who undergo major abdominal or noncardiac thoracic surgery. Therefore, exercise testing may be of clinical value for preoperative screening prior to elective surgery in elderly patients.
REFERENCES 1. Vital and Health Statistics, U.S. Department of Health and Human Serwces. Series 13, no. 88. Washington, D.C.. U.S. Government Printing Office, 1983; 1417.
68
CARDIOPULMONARY 2. Goldman L, Caldera DL, Nussbaum SR. eta/: Multrfactorral index of cardrac rrsk in noncardiac surgical procedures. N Engl J Med 1977; 297: 8455850. 3. Gerson MC, Hurst JM, Hertzberg VS, et al: Cardiac prognosis rn noncardiac genatrrc surgery. Ann Intern Med 1985; 103: 832-837. 4. Gerson MC, Hurst JM, Hertzberg VS. Ellis KS: Cardrac rusk stratrfication prror to geriatric abdomrnal or noncardiac thoracic surgery (abstr). Circulatron 1988; 78(suppl II). 11-108. 5. Benn RT: Some mathematical propertres of weight-for-height Indices used as measures of adiposity. Br J Prev Sot Med 1971; 25. 42-50. 6. House of Delegates of the American Society of Anesthesrologrsts, Inc: New classrficatron of physrcal status. Anesthesiology 1963; 24: 111. 7. Brello DR, Mattar AG, McKnight RC. Siegel BA. Ventilatron-perfusion studies rn suspected pulmonary embolism AJR 1979, 133: 1033-1037. 8. SAS Institute Inc: SAS users guide. Basics and statistrcs. versron 5 edrtion. Cary. North Carolrna: SAS Institute Inc, 1985; 402-506. 9. Breslow NE, Day NE: Statistical methods in cancer research: volume l-the analysrs of case-control studies InternatIonal Agency for Research on Cancer, Lyon, France, 1980; 192-279. 10. Tisr GM: Preoperabve evaluation of pulmonary functron. Validity. indications, and benefits. Am Rev Respir Dis 1979: 119: 293-310. 11. Tarhan S, Moffrtt EA. Sessler AD, Douglas WW. Taylor WF: Risk of anesthesra and surgery In patients with chronic bronchrtis and chronic obstructrve pulmonary drsease. Surgery 1973: 74: 720-726. 12. Wrllrams CD, Brenowrtz JB: “Prohibitrve” lung functron and major surgical procedures. Am J Surg 1976; 132: 763-766. 13. Latimer RG. Drckman M, Day WC, Gunn ML, Schmidt CD: Ventrlatory patterns
SURGICAL
COMPLICATIONS
/ GERSON
ET AL
and pulmonary complications after upper abdomrnal surgery determrned by preoperative and postoperative computerized sprrometry and blood gas analysis. Am J Surg 1971: 122: 622-632. 14. Stein M, Koota GM, Simon M. Frank HA Pulmonary evaluatron of surgrcal patients. JAMA 1962; 181: 7655770. 15. Eugene J, Brown SE, Light RW. Mrlne NE, Stemmer EA: Maximum oxygen consumptron: a physrologic guide to pulmonary resectron. Surgtcal Forum 1982; 33: 260-262. 16. Smith TP, Krnasewrtz GT, Tucker WY, Spellers WP. George RB: Exercise capacrty as a predrctor of post-thoracotomy morbidity. Am Rev Respir DIS 1984; 129: 730-734. 17. Mahar LJ, Steen PA, Tinker JH, et al: Perroperatrve myocardial Infarctron in patients with coronary artery disease with and wrthout aorta-coronary artery bypass grafts. J Thorac Cardiovasc Surg 1978; 76: 533-537. 18. McCollum CH. Garcra-Rinaldi R, Graham JM, DeBakey ME: Myocardral revascularization prror to subsequent major surgery In patients wrth coronary artery disease. Surgery 1977; 81: 302-304. 19. Bartlett RH. Brennan ML, Gauanrga AB. Hanson EL. Studres on the pathogenesis and prevention of postoperative pulmonary complications. Surg Gynecol Obstet 1973; 137: l-9. 20. Craven JL. Evans GA, Davenport JL, Willrams RHP: The evaluation of the Incentive spirometer in the management of postoperabve pulmonary complications. Br J Surg 1974; 61: 793-797. 21. Dohr S. Gold Ml: Comparison of two methods of postoperabve respiratory care. Chest 1978; 73: 592-595. 22. Thoren L: Post-operatrve pulmonary complrcations; observatrons on therr prevention by means of physiotherapy. Acta Char Stand 1954; 107: 193-205.
February
1990
The American
Journal
of Medicine
Volume
88
107