Respiratory dysfunction after uncomplicated cardiopulmonary bypass

Respiratory Dysfunction After Uncomplicated Cardiomlmonarv Bvpass I

J

J I

David P. Taggart, MD(Hons), Mohammed El-Fiky, MB, Rodger Carter, MSc, Adrian Bowman, PhD, and David J. Wheatley, FRCS Departments of Cardiac Surgery and Respiratory Medicine, Royal Infirmary, Glasgow, and Department of Statistics, Glasgow University, Glasgow, Scotland

Respiratory dysfunction is a well-recognized complication of cardiac operations. To quantify its current incidence and severity after uncomplicated cardiopulmonary bypass, serial measurements of arterial oxygen tension (Pao,), alveolar-arterial oxygen gradient (Aao,), and percentage pulmonary shunt fraction (%PSF)measured by a noninvasive technique were made in 129 patients (age, 59 2 8 years (mean f standard deviation) with good left ventricular function (left ventricular end-diastolic pressure <15 mm Hg) undergoing isolated coronary artery operations (group 1) and 30 patients undergoing general surgical procedures (group 2). Measurements were made before operation and on the first, second, and sixth postoperative days. Seven patients in group 1 who required prolonged ventilation were excluded from further study. In group 1, between the preoperative and second postoperative days, there was a marked fall in Pao, [89 2 11 versus 57 2 9 mm Hg; p < 0.0011 and a marked increase in the Aao, gradient [18 2 10 versus 50 f 11 mm Hg; p < 0.00l)l and %PSF[3 2 1% versus 19 2 6%; p < 0.00l)l with only modest improvement by the sixth 11 mm Hg; Aao,, 45 postoperative day [Pao,, 67

*

*

R

espiratory dysfunction has been a well-documented complication of cardiopulmonary bypass (CPB)since the earliest days of cardiac surgery [l-lo]. Although it remains a well-recognized complication of CPB, its incidence and severity in current practice are not clearly documented. Kirklin's group [ l l ] reported a 30% incidence of pulmonary complications after CPB, but the methods of measurement were subjective and relatively insensitive. Furthermore, the last decade has seen changes in surgical patients toward an increasingly elderly and sicker population [12-141 who may be more susceptible to respiratory dysfunction because of a significant age-related decline in respiratory reserve after the middle of the sixth decade [15, 161. To evaluate the frequency and severity of respiratory dysfunction after uncomplicated CPB, we compared 129 coronary artery surgery patients (group 1)and 30 general surgery patients. Serial measurements of arterial oxygen Accepted for publication Dec 31, 1992 Address reprint requests to MI Taggart, Department of Cardiothoracic Surgery, Royal Brompton National Heart and Lung Hospital, Sydney St, London SW3 6NP, England.

0 1993 by The Society of Thoracic Surgeons

11 mm Hg; %PSF, 15 2 41. There were similar but less severe changes in Pao, and Aao, gradients in group 2 patients, with a return to baseline values by day 6. Regression analysis in group 1 patients showed a weak correlation between postoperative respiratory dysfunction and preoperative impairment of the Aao, gradient, but no correlation with age, sex, smoking status, New York Heart Association status, bypass time, or violation of the pleural sacb). To determine the duration of respiratory dysfunction after cardiac surgery, serial Pao, and Aao, gradient measurements were continued until the sixth postoperative week in a further 30 patients (group 3). Group 3 patients demonstrated similar early impairment of respiratory function to group 1 patients but with complete resolution by the sixth postoperative week. This study demonstrates that respiratory dysfunction is both common and frequently severe even after uncomplicated cardiopulmonary bypass but resolves by the sixth postoperative week. Respiratory dysfunction is also common after a major general operation but is less severe and resolves by the sixth postoperative day. (Ann Thoruc Surg 2993;56:1123-8)

tension (Pao,), alveolar-arterial oxygen gradients (Aao,), and percentage pulmonary shunt fractions (%PSF)measured by a noninvasive technique were made before operation and on the first, second, and sixth postoperative days. In a further group of 30 patients undergoing coronary revascularization (group 3), measurement of Pao, and Aao, were repeated 6 weeks after operation. The effects of age, sex, smoking status, New York Heart Association (NYHA) status, Pao,, Aao, gradient, %PSF, bypass time, ischemic time, and violation of the pleural sac(s) on respiratory dysfunction were investigated by regression analyses.

Material and Methods Ethical permission for the studies was given by the Hospital Ethical Committee, and patients gave informed consent before inclusion.

Patients We initially studied 129 patients undergoing elective isolated coronary artery operations (group 1) and 30 patients undergoing major general surgical procedures 0003-4975/93/$6.00

1124

TAGGART ET AL RESPIRATORY DYSFUNCTION AFTER CPB

(group 2). A further 30 patients undergoing coronary revascularizationwere studied again at 6 weeks (group 3). No patient had overt clinical evidence of respiratory or cardiac impairment. No patient had suffered a myocardial infarction in the 3 months before the study. Cardiac patients requiring diuretic therapy in excess of 40 mg/day of furosemide or with a left ventricular end-diastolic pressure greater than 15 mm Hg were excluded. Valvular heart disease was excluded on clinical grounds and by left ventricular injection during coronary angiography. In the postoperative period, patients requiring ventilatory assistance for greater than 24 hours were excluded from further study.

Measurement of Respiratory Dysfunction Serial examination of Pao,, Aao, gradient, and %PSFwas performed in the preoperative period and on the first, second, and sixth postoperative days (and at 6 weeks in group 3 patients). The Pao, was measured from a blood sample obtained by direct arterial puncture of the radial artery for the preoperative, sixth-day, and 6-week samples and from an in-dwelling radial artery cannula for all other samples. All samples were collected with the patient having breathed room air for at least 10 minutes. The samples were processed immediately for Pao, and pH using a calibrated Corning 178 pH/Blood Gas Analyser (Ciba-Corning Diagnostics, Medfield, MA) and for arterial oxygen content using a Corning 2500 Co-oximeter. The Aao, gradient was measured simultaneously with arterial oxygen tension with the patient initially breathing room air and then after the patient had breathed 100% oxygen for 10 minutes. The partial pressure of alveolar oxygen was calculated with reference to the respiratory exchange ratio measured on samples of expired air collected through a nasal cannula into an anesthetic bag. These samples, obtained simultaneously with arterial blood samples, were analyzed for the fractional concentrations of oxygen and carbon dioxide using a paramagnetic and infrared analyzer, respectively (P.K. Morgan, Rainham, Kent, UK). The partial pressure of alveolar oxygen (PAo,) is given by the "ideal" alveolar air equation P71

P A O=~ Fiop(BP - 47) - PacodR + 0.209 Pacoz(1 - R)/R, where Fio, is the fractional concentration of oxygen in inspired air, BP the barometric pressure, 47 the saturated vapor pressure, 0.209 the fractional concentration of oxygen in room air, Paco, arterial carbon dioxide tension, and R the respiratory exchange ratio: R = Feco,/(Fio, Feo,), where Feo, and Feco, are the fractional expired oxygen and carbon dioxide concentrations. Measurement of the %PSF was calculated without the need for Swan-Ganz catheterization as described in the mathematical model of Riley and Permutt [MI, and which we have previously validated in cardiac surgery patients [19]. This model is essentially based on the difference in Aao, gradient present when the patient initially breathed room air and then breathed 100% oxygen [MI. To confirm its validity in postoperative cardiac patients, we previ-

Ann Thorac Surg 1993;561123-8

ously measured the %PSFusing this model and compared it with the value simultaneously obtained by Swan-Ganz catheterization [19] and demonstrated excellent correlation between the two techniques (r = 0.94, p < 0.001).

Anesthetic Regimen and Cardiopulmonary Bypass A standard anesthetic regimen was followed. Anesthesia was induced with midazolam and fentanyl, and intubation was performed after administration of atracurium or pancuronium. Anesthesia was maintained with morphine, fentanyl, midazolam, and atracurium or pancuronium. The lungs were not ventilated during CPB. Cardiopulmonary bypass was performed with pulsatile perfusion, bubble oxygenation, moderate systemic hypothermia (28"to 30"C), a 40+m arterial line filter, and 2 L of crystalloid prime. Flow rates were based on the formula that at normothermia flow was equal to 2.4 x Body Surface Area and was reduced to two-thirds at 28°C. Pulsatile flow, achieved with a Stockert pump, was defined as 72 pulsedmin, a 50% run time at 130%base flow. Mean arterial pressure was maintained between 40 and 60 mm Hg, and vasopressors or vasodilators were administered as necessary to maintain this. The Pao, in the arterial line was continuously monitored using a Polystan Po, monitor (Polystan UK, Nottingham, UK) to maintain Pao, between 100 and 200 mm Hg. Arterial Paco, was maintained between 27 and 35 mm Hg, and the pH between 7.4 and 7.53. Packed cell volume was maintained between 20% and 28%.

Postoperative Management All cardiac patients were transferred to the intensive care unit receiving ventilatory assistance, paralyzed, and monitored. Ventilation was in a controlled mandatory ventilation mode (Erica Ventilator) with a tidal volume of 10 to 12 mL/kg and a respiratory rate of 10 to 12 breathdmin. Fractional concentration of oxygen in inspired air was adjusted to maintain the Pao, between 80 and 110 mm Hg, and the respiratory rate was adjusted to maintain the Paco, between 35 and 45 mm Hg. Positive end-expiratorypressure (5 cm H,O) was administered routinely during assisted ventilation. Extubation was undertaken with the patient fully rewarmed, mentally alert, and hernodynamically stable, usually 10 to 12 hours after operation. All general surgical patients were extubated within a few hours of operation.

Data Presentation and Statistical Analysis Statistical analysis was performed using the S-PLUS statistical package [20]. Data presented as means and standard deviations are summarized in Table 1 and presented graphically, as means and standard errors, in Figures 1 to 3. Changes in respiratory function over time were assessed by paired t tests between measurements obtained before operation and those on the first, second, and sixth postoperative days. Certain preoperative and intraoperative covariates considered to be possible predictors or determinants of respiratory injury were examined. In particular, the effects of age, sex, and NYHA status, preoperative measurements of forced expiratory volume

TAGGART ET AL RESPIRATORY DYSFUNCTION AFTER CPB

Ann Thorac Surg 1993;56:1123-8

1125

Table 1. Serial Changes in Arterial Oxygen Tension, Alveolar-Arterial Oxygen Gradient, and Pulmonary Shunt Fraction” Alveolar-Arterial Oxygen Gradient

Arterial Oxygen Tension (mm Hg)

Preop

1 (n = 122) 89 (11) 59 (9)b 57 (9)b 67 (ll)b . . . 18 (10) 48 (13)b 50 (ll)b 45 (ll)b 2 (n = 30) 86 (7) 69 (11)’ 67 (9)’ 79 (6) . . . 14 (4) 21 (4)’ 19 (6) 15 (6) 3 (n = 30) 91 (6) 60 (5)b 58 (5)b 67 (5)b 93 (6) 17 (6) 43 (12)b 45 (12)b 39 (ll)b 17 (8)

3 (1)

a

24 h

48 h

6days

Numbers within parentheses are standard deviations.

6wk

Preop

Pulmonary Shunt Fraction (%)

6wk

Group

Preop

(mm Hg) 24 h 48 h 6days

p < 0.001 versus preoperative value.

in 1 second, Pao,, Aao, gradient, and %PSF, together with bypass and ischemic time and violation of one or both pleural sacs, were studied. The potential effects of these variables were investigated by including them as covariates in multiple regression analyses, using measurements of Pao,, Aao, gradient, and %PSFon days 1,2, and 6 as separate response variables. In the case of %PSF, the average of measurements on days 1 and 2 was used as a response, as there was no evidence of change over this period.

Results The study comprised 129 patients undergoing elective coronary artery operations (group l), 30 patients undergoing major general surgical procedures (group 2), and a further 30 coronary artery surgery patients followed up until 6 weeks after operation (group 3). The general surgical operations comprised 25 bowel resections and 5 portacaval shunts. As the aim of the study was to examine the effects of uncomplicated CPB on respiratory function, 7 patients in group 1 who required prolonged ventilatory assistance were excluded from analysis. All other cardiac surgery patients received ventilatory assistance for less than 24 hours, with a time to extubation of 11 k 4 hours (mean k standard deviation). All general surgical patients received ventilatory assistance for less than 4 hours. There were 103 men and 19 women in group 1 with an

Pre-op.

Day 1

Day 2

Day 6

Time

Fig 1. Serial changes in arterial oxygen tension (pa02) (mean 2 standard error) at various time points. (* p < 0.002 versus preoperative value.)

24 h

48 h

6days

19 (5)b 19 (6)b 15 (4)b

p < 0.05 versus preoperative value

*

age (mean standard deviation) of 59 f 8 years, and 26 (21%)were more than 65 years old. Twenty-five patients were in NYHA class I (20%), 75 in NYHA class I1 (61%), and 22 in NYHA class I11 (18%).Twenty-one patients had never smoked (17%),78 were ex-smokers (64%)of at least 1-year duration, and 23 still occasionally smoked (19%). The group 2 patients comprised 18 men and 12 women with an age (mean k standard deviation) of 61 f 8 years, and 10 (33%)were more than 65 years old. All group 2 patients were smokers up to the time of operation. All cardiac patients underwent isolated coronary artery bypass grafting; 101 (83%)received one left internal mammary artery and a mean of 2.4 vein grafts, 20 (16%) received a mean of 3.1 vein grafts without an internal mammary artery, and 1 patient received two internal mammary artery grafts. Pleurae were maintained intact in 22 patients (18%), one or both pleural sacs were opened in 98 patients (82%), and this information was not recorded in 2 patients. Serial postoperative changes in Pao,, Aao, gradient, and %PSF for the three groups are summarized in Table 1 and graphically illustrated for group 1in Figures 1to 3. In group 1 between the preoperative and second postoperative days there was a highly significant decrease in Pao, [89 f 11 versus 57 f 9 mm Hg; p < O.OOl)] accompanied by a highly significant increase in the Aao, gradient [18 f 10 versus 50 f 11 mm Hg; p < O.OOl)], and %PSF [3 f 1% versus 19 f 6%; p < O.OOl)]. There was only modest

Pre-op.

Day 1

Day 2

Day 6

Time

Fig 2. Serial changes in alveolar-arterial oxygen gradient (Aa02) (mean & standard error) at various time points. (* p < 0.002 versus preoperative value.)

1126

TAGGART ET AL RESPIRATORY DYSFUNCTION AFTER CPB

Ann Thorac Surg 1993;5611234

grade, smoking status, bypass time, ischemic time, opening of one or both pleural sacs, and preoperative measurements of forced expiratory volume in 1 second, Pao,, Aao, gradient, and %PSF. The only consistent correlation was with impairment of the preoperative Aao, gradient, which was significant for all three postoperative measurements at all time points ( p value between 0.03 and 0.01). Age, sex, smoking status, NYHA status, duration of CPB and the ischemic time, and inadvertent pleurotomy (Table 3) did not influence the degree of postoperative respiratory impairment.

/ / /

/

Comment Pre-op.

Day 1

Day 2

Day 6

Time

Fig 3. Serial changes in pulmonary shunt fraction (mean f standard error) at various time points. (“ p < 0.001 versus preoperative value.)

improvement in these parameters by the sixth postoperative day [Pao,, 67 t 11 mm Hg; Aao,, 45 t 11 mm Hg; %PSF, 15 t 41. Group 2 patients also showed a significant but less marked decrease in Pao, over the first and second postoperative days but had essentially returned to preoperative values by the sixth postoperative day. There was a significant increase in the Aao, gradient in group 2 patients only on the first postoperative day, with a return to preoperative values by the sixth postoperative day. As shown in Table 2, in the preoperative period 35% of group 1 patients had a Pao, greater than 90 mm Hg, 65% of patients a Pao, between 60 and 90 mm Hg, and no patient had a Pao, less than 60 mm Hg (the conventional definition of respiratory failure). By the second postoperative day the percentage of patients in the same groups were 0%, 34%, and 66%, and by the sixth postoperative day 0%, 74%, and 26%, respectively. Thus one-quarter of the cardiac surgical patients had significant respiratory impairment by the end of the first postoperative week. The 30 coronary artery surgery patients in group 3 demonstrated identical changes in Pao, and Aao, to group 1 patients over the first postoperative week. In group 3 patients, Pao, and Aao, gradient returned to baseline values by the sixth postoperative week. The following variables were included as potential covariates for respiratory dysfunction: age, sex, NYHA

Respiratory dysfunction is a familiar complication of CPB, but its exact prevalence and severity in current surgical practice is not precisely documented. Almost a decade ago, approximately one-third of patients were reported to experience pulmonary dysfunction after CPB [l l ]. Improvements in medical, anesthetic, and surgical practice as well as extracorporeal perfusion technology, which might have been expected to reduce this incidence, may have been offset by the less favorable features of the current surgical population. Patients undergoing coronary operations today are older and have poorer left ventricular function and a higher prevalence of other diseases than those operated on even 5 years ago [12-141. The number of patients more than 65 years of age now exceeds 40% in some current series [12-141, and this is of particular relevance as there is an age-related decline in respiratory function particularly marked after 65 years of age [15, 161. Both these factors are likely to increase the incidence and severity of respiratory dysfunction in cardiac surgical patients. Kirklin’s group [ l l ] reported a 30% incidence of pulmonary dysfunction after CPB, but the methods of quantification, such as measurement of tracheal secretions, were relatively insensitive and nonspecific. At the most severe end of the spectrum, Hammermeister and colleagues [21] reported that prolonged ventilation (>48 hours) was required in 8% of more than 8,000 patients undergoing coronary operations (with a 25% mortality), similar to the 5% incidence observed in our series (7 of 129 patients). Consequently, to assess the current incidence and severity of pulmonary dysfunction in patients after uncomplicated CPB, we measured three sensitive and objective parameters of functional gas exchange. Although hypoxia

Table 2. Number of Patients in Group 1 With Varying Degrees of Respiratory Dysfunction at Different Timesa Po, (mm Hg) Aao, gradient (mm Hg) %PSF Time

>90

60-90

<60

Preop

43 (35) 0 (0) 0 (0) 0 (0)

79 (65) 47 (40) 38 (34) 73 (74)

0 (0) 71 (60) 75 (66) 25 (26)

24 hours 44 hours 6 days a

Numbers within parentheses are percentages. Po, = alveolar-arterial oxygen gradient;

Aao,

=

arterial oxygen tension;

<20

78 (64) 4 (3) 2 (2) 0 (0) PSF

=

>20

<3%

>3%

44 (36) 114 (97) 111 (98) 97 (100)

77 (68) 0 (0) 2 (2) 0 (0)

36 (32) 110 (100) 102 (98) 87 (100)

pulmonary shunt fraction.

TAGGART ET AL RESPIRATORY DYSFUNCTION AFTER CPB

Ann Thorac Surg 1993;561123-8

1127

Table 3. Effect of Pleurotomy on Respiratoy Dysfunction" Aao, Gradient (mm Hg)

Po, (mm Hg)

Pleurae intact (22 patients) Pleura open (98 patients) a

Pre

Day 2

Day 6

Pre

Day 2

88 (13)

57 (9) 58 (8)

65 (9) 67 (11)

17 (10) 17 (10)

49 (9) 50 (12)

89

(11)

PSF (%)

Day 6

Pre

Day 2

Day 6

45

3 (2) 3 (1)

18 (5) 19 (6)

14 (4) 15 (4)

(12)

45 (11)

Numbers in parentheses are standard deviations.

may reflect poor ventilation, the Aao, gradient remains relatively independent of ventilatory effort. Whereas an increase in %PSF invariably leads to an increase in the Aao, gradient, an increase in this gradient does not necessarily produce an increase in the %PSF. We measured the %PSF without the need for Swan-Ganz catheterization based on the difference in Aao, gradient with the patient breathing room air and 100% oxygen as described by Riley and Permutt [MI. We have previously confirmed the validity of this method in postoperative cardiac surgical patients by comparing the %PSFobtained by this technique with that simultaneously obtained by Swan-Ganz catheterization [191, demonstrating excellent correlation between the two techniques (r = 0.94, p < 0.001). Our study demonstrates that respiratory dysfunction is both common and frequently severe even after uncomplicated CPB. The degree of impairment is significant (Pao, < 60 mm Hg breathing room air) in 66%of patients on the second postoperative day and 26% of patients on the sixth postoperative day. Nevertheless, most patients merely required supplemental oxygen in the early postoperative period and were discharged on the seventh or eighth postoperative day. Furthermore, follow-up studies at 6 weeks showed complete resolution of this respiratory dysfunction as witnessed by a complete return to normal of arterial oxygen tension and alveolar-arterial oxygen gradients. The pathophysiology of hypoxia in patients after CPB is different from that in general surgical patients. The latter patients showed hypoxia without a proportional increase in the Aao, gradient, implying that hypoxia is due to a decrease in alveolar oxygen from hypoventilation probably owing to morphine analgesia. In contrast, the increase in the Aao, gradient and %PSF that accompanies hypoxia in the CPB patients reflects ventilation-perfusion inequality (ie, blood is delivered to nonventilated alveoli), resulting in a more severe and longer lasting degree of dysfunction. The continuing Aao, gradient in patients after CPB while breathing 100%oxygen is due to true shunting and, therefore, atelectasis. The precise pathologic mechanism producing such dramatic atelectasis in patients after CPB is not clear but is probably multifactorial. Postoperative changes in respiratory function are influenced by numerous factors including preexisting cardiac or respiratory impairment, general anesthesia, and the effects of CPB itself. Median sternotomy may impair pulmonary function tests [22, 231 by

reducing chest wall movement, but this does not explain changes in Pao,, Aao, gradient, or %PSF. Nonventilation of the lungs during CPB is probably one mechanism contributing to atelectasis. Stimulation of humoral and cellular immune systems during CPB results in activation of a myriad of inflammatory mediators including complement and white blood cells. This inflammatory cascade system [ll, 241 is at least partially responsible for increased capillary permeability after CPB, producing flooding of the pulmonary interstitium [25, 261 and leading to intrapulmonary shunting [lo]. Diaphgramaticparalysis, probably as a result of phrenic nerve cold injury, is reported to occur in up to 30% of patients after CPB and to persist in one-third at 1 year [27]. Although topical hypothermia was used as an adjunct to cardioplegic myocardial protection in our patients, diaphgramatic paralysis neither explains the frequency nor resolution within 6 weeks of the pulmonary dysfunction witnessed in our study. Left-sided pleural effusions can be detected in 40% of patients after a coronary artery operation regardless of whether the pleura is opened or remains intact [28]. In our patients, there was no difference in the degree of respiratory dysfunction between those whose pleural sacs were opened during harvesting of the internal mammary artery and those in whom they remained intact (patients receiving only vein grafts). The only consistently significant predictor of postoperative respiratory dysfunction was impairment of the preoperative Aao, gradient, although the correlation was relatively weak. Age did not appear as a covariate of respiratory impairment after CPB. Although there is a sharp decline in respiratory function after the age of 65 years [12, 131, only 26 (21%) of our post-CPB patients exceeded this age. We may therefore have missed an age-related effect that would be apparent in an older population. Likewise we could not demonstrate a significant correlation between respiratory dysfunction after CPB and deteriorating NYHA status between grades I and I11 (those in NYHA grade IV were excluded) or with current cigarette smoking (although most such patients only smoked a few cigarettes per day). This implies that the important determinants of respiratory dysfunction are intraoperative factors. Nevertheless, we failed to demonstrate an effect of the duration of CPB (up to 160 minutes), ischemic times (up to 65 minutes), or violation of one or more pleural sacs on postoperative respiratory dysfunction. It is probable that respiratory injury may be an

1128

TAGGART ET AL RESPIRATORY DYSFUNCTION AFTER CPB

inevitable sequela of t h e systemic activation of inflammatory mediators even after a short period of extracorporeal circulation. The results of our s t u d y need to be interpreted cautiously before being applied t o t h e general cardiac surgical population. In particular, the following points should be considered. First, our study almost certainly underestimates the true incidence and severity of respiratory dysfunction after CPB. We only studied patients a t the better end of t h e surgical spectrum (mean age 59 years, majority i n NYHA class I o r I1 w i t h good left ventricular function a n d no other disease). Furthermore, we excluded from analysis 7 patients who required prolonged ventilatory s u p p o r t (>24 hours). Second, i n t h e current s t u d y a bubble rather t h a n m e m b r a n e oxygenator was used d u r i n g CPB. Although some studies have demonstrated that membrane oxygenators may reduce complement activation a n d transpulmonary sequestration of leukocytes, there has been no consistent demonstration of clinical benefit for “routine” CPB. However, we are currently comparing the pulmonary consequences of membrane a n d bubble oxygenators d u r i n g routine CPB. In summary, our s t u d y demonstrates that respiratory impairment after uncomplicated CPB even i n low-risk patients is common, frequently severe, and still marked i n at least one-quarter of t h e patients a t t h e end of t h e first postoperative week b u t resolves completely by t h e sixth postoperative week. It is likely, however, that t h e incid e n c e and severity of respiratory dysfunction would be higher i n older patients w i t h poorer cardiac function and an increased prevalence of other disease. We acknowledge the Biomedical Research Committee of Scotland for funding this study.

References 1. Dodrill FD. The effects of total body perfusion upon the lungs. In: Alle JG, ed. Extracorporeal circulation. Springfield, IL: Thomas, 1958:327-35. 2. Clowes GH. Extracorporeal maintenance of circulation and respiration. Physiol Rev 1960;40:826-919. 3. Baer DM, Osborn JJ. The postperfusion pulmonary congestion syndrome. Am J Clin Pathol 1960;34:442-5. 4. Nahas RA, Melrose DG, Sykes MK, Robinson B. Postperfusion lung svndrome: role of circulatow exclusion. Lancet 1965;2:251-4’. 5. Neville WE, Kontaxis A, Gavin T. Clowes GH. PostDerfusion pulmonary ‘vasculitis. Its relationship to blood traima. Arch Surg 1963;86:12&37. 6. Schramel R, Schmidt R, Davis F, Palmisano D, Creech 0. Pulmonary lesions produced by prolonged perfusion. Surgery 1963;5422&31. 7. Allardyce DB, Yoshida SH, Ashmore PG. The importance of microembolism in the pathogenesis of organ dysfunction caused by prolonged use of the pump oxygenator. J Thorac Cardiovasc Surg 1966;52:706-15.

Ann Thorac Surg 1993;5611-

8. Asada S, Yamaguchi M. Fine structural change in the lung following cardiopulmonary bypass. Its relationship to early postoperative course. Chest 1971;59:47&83. 9. Ratliff NB, Young WG Jr, Hackel DB, Mikat E, Wilson JW. Pulmonary injury secondary to extracorporeal circulation. An ultrastructural study. J Thorac Cardiovasc Surg 1973;65: 425-32. 10. Byrick RJ, Noble WH. Postperfusion lung syndrome. Comparison of Travenol bubble and membrane oxygenators. J Thorac Cardiovasc Surg 1978;76:685-93. 11. Kirklin JK, Westaby S, Blackstone EH, Kirklin JW, Chenoweth DE, Pacific0 AD. Complement and the damaging effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1983;86:&35-57. 12. Christakis GT, Ivanov J, Weisel RD, et al. The changing pattern of coronary artery bypass surgery. Circulation 1989; 8O(Suppl 1):151-61. 13. McGrath LB, Laub GW, Graf D, Gonzalez-Lavin L. Hospital death on a cardiac surgical service: negative influence of changing practice patterns. Ann Thorac Surg 1990;49:410-2. 14. Jones EL, Weintraub WS, Craver JM, Guyton RA, Cohen CL. Coronary bypass surgery: is the operation different today? J Thorac Cardiovasc Surg 1991;101:108-15. 15. Evans TI. The physiologic basis of geriatric general anesthesia. Anest Intensive Care 1973;1:319-22. 16. Craig DB, McLeskey CH, Mitenko PA, Thomson IR, Janis KM. Geriatric anaesthesia. Can J Anaesth 1987;34:156-9. 17. Riley RL, Cournand A. “Ideal” alveolar air and the analysis of ventilation-perfusion relationships in the lungs. J Appl Physiol 1949;1:825. 18. Riley RL, Permutt S. Venous admixture component of A-aPo, gradient. J Appl Physiol 1973;35:430-1. 19. El-Fiky MM, Taggart DP, Carter R, Stockwell MC, M a d e BH, Wheatley DJ. Respiratory dysfunction following cardiopulmonary bypass: verification of a non-invasive technique to measure shunt fraction. Resp Med 1993;8719%3. 20. Becker RA, Chambers JM, Wilkins AR. The new S language. CA: Wadsworth and Brooks Cole, 1988. 21. Hammermeister KE, Burchfiel C, Johnson R, Grover FL. Identification of patients at greatest risk for developing major complications at cardiac surgery. Circulation 1990;82(Suppl 4):380-9. 22. Shapira N, Zabatino SM, Ahmed S, Murphy DM, Sullivan D, Lemole GM. Determinants of pulmonary function in patients undergoing coronary bypass operations. Ann Thorac Surg 1990;50:268-73. 23. Berrizbeita LD, Tessler S, Jacobwitz IJ, Kaplan P, Budilowicz L, Cunningham JN. Effects of sternotomy and coronary bypass surgery on postoperative pulmonary mechanics. Chest 1989;96873-6. 24. Kirklin JW. The science of cardiac surgery. Eur J Cardiothorac Surg 1990;46%71. 25. Smith EE, Naftel DC, Blackstone EH, Kirklin JW. Microvascular permeability after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1987;94:225-33. 26. Royston D, Minty BD, Higenbottam TW, Wallwork J, Jones GJ. The effect of surgery with cardiopulmonary bypass on alveolar-capillary barrier function in human beings. Ann Thorac Surg 1985;40:139-43. 27. Efthimiou J, Butler J, Woodham C, Benson MK, Westaby S. Diaphragm paralysis following cardiac surgery: role of phrenic nerve cold injury. Ann Thorac Surg 1991;52:1005-8. 28. Peng MJ, Vargas FS, Cukier A, Terra-Filho M, Teixeira LR, Light RW. Postoperative pleural changes after coronary revascularization. Comparison between saphenous vein and internal mammary artery grafting. Chest 1992;101:327-30.