Pulmonary shunt as a prognostic indicator in head injury ELIZABETH A. M. FROST, M.D., CARLOS U. ARANCIBIA, M.D., AND KENNETH SHULMAN, M.D.

Transcription

1 J Neurosurg 50: , 1979 Pulmonary shunt as a prognostic indicator in head injury ELIZABETH A. M. FROST, M.D., CARLOS U. ARANCIBIA, M.D., AND KENNETH SHULMAN, M.D. Departments of Anesthesiology and Neurosurgery, Albert Einstein College of Medicine, Yeshiva University, Bronx, New York v" Severe head injury may cause momentary respiratory arrest. Resultant hypoxia would increase cerebral edema and adversely affect the quality of survival. This study examines the effect of hypoxemia on outcome. Pulmonary shunt was calculated as a convenient measurement of respiratory insufficiency in 86 severely head-injured patients who underwent surgery. All samples were taken shortly after induction into anesthesia when controlled ventilation with high inspired-oxygen concentration had been established. In 39 patients who improved, mean pulmonary shunt was 8.9%. Twelve patients who survived with deficit showed a mean shunt of 13.6%, and in 35 patients who died, the mean initial shunt was 15.6%. No significant correlation was found between abnormal chest x-ray findings or the occurrence of hypertension and shunt percentage. The American Society of Anesthesiologists at-risk classification correlated grossly with the outcome. Early pulmonary shunt is a prognostic indicator in severe head injury and should be used in conjunction with the Glasgow Coma Scale in assessing outcome. Despite an apparently adequate respiratory pattern, all patients with severe head injury must be assumed to be hypoxic until proven otherwise. While hypoxemia may prove to be refractory in overwhelming injury, patients who score low on the Glasgow Coma Scale but who have relatively normal oxygen exchange may still survive with little deficit. KEY WORDS ~ pulmonary shunt 9 prognosis 9 head injury 9 hypoxia 9 respiration A N association between respiratory insufficiency and head injury has long been recognized. Horsley and Kramer in England e,a and later Cannon in the United States 2 showed that intracranial pressure (ICP) rose immediately after head injury and the primary cause of death was arrest of respiratory center function. If artificial ventilation could be established, survival was possible. Varying degrees of hypoxia are inevitable in such situations, however, causing increased cerebral edema which might be assumed to adversely affect the quality of survival. If a correlation does exist between severity of head injury and degree of respiratory insufficiency, then proportionate cerebral hypoxia should occur almost immediately after injury. This would seriously imperil already damaged brain tissue and be reflected in prolonged or severe neurological deficit. We, and others, 7 have been impressed by an incidence of up to 65% arterial hypoxemia in spontaneously breathing patients who do not appear to be in respiratory distress on admission to the emergency room. This clinical study was undertaken 1) to assess the effect of early hypoxemia on outcome; 2) to attempt to differentiate between peripheral and central causes of respiratory insufficiency; and 3) to quantify the effect if any of a changing ventilatory pattern on initial pulmonary dysfunction. Clinical Materials and Methods Eighty-six severely head-injured adults who underwent emergency surgery at our institution between January 1, 1976, and June 30, 1977, were studied. They ranged in age from 15 to 85 years (Fig. 1). There were 45 patients (52%) under 35 years of age; 74 patients were male, and 12 were female (a ratio just over 6: 1). Patients with a history of chronic lung disease were excluded. Estimated pulmonary shunt or ventilationperfusion abnormality (Qs/Qt) was calculated as a convenient measure of respiratory insufficiency and hypoxemia according to the equation: 14 (PAO2 - PaO2) % = 3.5T(~O2---P--ff0-~) ' where PAO2 = alveolar oxygen tension, PaO2 = arterial oxygen tension, and = solubility coefficient of 02 in blood. This equation assumes a 768 J. Neurosurg. / Volume 50 / June, 1979

2 Pulmonary shunt in head injury PaO2 TABLE 1 Estimated pulmonary shunt calculated at different levels of fractional inspired oxygen concentration (FiOt) and arterial oxygen content (PaO~) Fractional Inspired Oxygen Concentration (FiO~) (mmhg) hemoglobin level of 15 gm and ph of 7.5. The denominator figure of 3.5 represents the arteriovenous difference and has been found to be a more representative average in critically ill patients than the classic 5 vol% value. 14 Alveolar oxygen tension (PAO2) was calculated according to the equation: PAO2 = (FiO2 X 713) - (PaCO2 1.). It is thus possible to calculate pulmonary shunt reasonably accurately knowing only the fractional inspired oxygen concentration (FiO2) and the arterial oxygen tension. Table 1 shows percentage shunt values at different levels of FiO2 and PaO2. All samples were initially taken shortly after induction into anesthesia when controlled hyperventilation with high inspired-oxygen concentration had been established. Samples were repeated at approximately 1- hour intervals, according to clinical indication, well into the postoperative period. It is well established that general anesthetic agents decrease functional residual capacity?,4,8,1~ Although the resultant increase in pulmonary shunting may be very variable, is it rarely exceeds 9% in a normotensive patient with otherwise healthy lungs. 5 In addition, shunting does not increase during anesthesia, 12 and is rapidly reversible in the postoperative period. The American Society of Anesthesiologists (ASA) at-risk classification is the most widely used technique J. Neurosurg. / Volume 50 / June, 1979 CASES 3O i0 i AGE FIG. 1. Age distribution of the 86 adult patients studied. 769

3 E. A. M. Frost, C. U. Arancibia and K. Shulman NO. CASES BLOOD PRESSURE BASELINE PULSE RECORDINGS li l TEMPERATURE FIG. 2. Distribution of immediate preoperative baseline recordings of blood pressure, pulse, and temperature subdivided into normal (N), high (H), and low (L) values. for preoperative assessment of surgical risk. Classifications of 4 or 5 were found to coincide with Glasgow Coma Scale values of 8 or less, and therefore either of these recordings was used as an indication of the most severely injured patients. Results Pulmonary shunting was insignificant (< 5%) in 15 patients (17.6%). Shunting (5% to 9%)that might have been caused by general anesthesia occurred in 14 patients (16.4%), while pulmonary shunting above 9% was calculated in 57 patients (66%). Thirty-nine patients (46%) improved; survival with moderate deficit and discharge was achieved in 12 patients (14%), while 35 patients (40%) died. Of the 12 patients over the age of 55 years, only two (15%) were discharged improved, three (%) survived with deficit, and seven (58%) died. In the 39 patients in the improved group (Group 1), mean initial pulmonary shunt was 8.9% (SD, 5.5). The 12 patients who survived but with neurological sequelae (Group 2) showed a mean initial shunt of 13.6% (SD, 4.5), and in the 35 who died (Group 3), mean shunt was calculated at 15.6% (SD, 6.1). Statistical analysis shows a significant difference (p = < 0.01) between the calculated shunts in Group 1 and the results obtained in Groups 2 and 3. Thirty-one of the 39 patients in Group 1 scored above 8 in the Glasgow Coma Scale (GCS) or were classified as ASA I to III. All had initial pulmonary shunts of < 14%. The calculated shunts in the more severely injured patients (GCS < 9) were 14.2%, 0.3%, 9.7%, 9%, 13.6%, 5.5%, and 7%, respectively. In one patient who scored GCS 12, the shunt was calculated at 16.8%. This patient had sustained a gunshot wound and, although the chest film and physical examination were reported normal, subsequent inquiry gave a past history of asthma. In Group 2, eight of the 12 patients scored between 9 and 14 (GCS), and had pulmonary shunts of 11% to 14%. One patient, again following a gunshot wound, scored 15 (GCS) and had a calculated shunt of 18.1%. In three patients with GCS values of 6, 3, and 7, the calculated shunts were 5.5%, 2.7%, and 0.34%, respectively. In Group 3, 27 of the 35 patients had low scores (< 9 GCS or equivalent) and initial shunts of > 13%. Two patients had suffered multiple gunshot wounds and scored 4 (GCS). Calculated shunts on these patients were 8.1% and 2.8%, respectively. In six patients who were initially accorded higher GCS values, four had shunts of > 16%. Insignificant shunt (1.5%) was calculated in one patient who scored 4 (GCS) and who had sustained an epidural hematoma. Pulmonary study of the 71 patients who had calculated initial shunts of > 5% showed that infection was already present (associated with aspiration or delay in hospital arrival in eight patients (11%). Drug or alcohol overdose contributed to respiratory depression in five patients (8%). Frank pulmonary edema was seen in three patients (4%) and disseminated intravascular coagulopathy in two (3%). Iatrogenic causes of ventilatory insufficiency (such as pneumothorax, fluid overload, and hematoma following carotid angiography) occurred in three patients (4%). In the majority of patients (50 or 71%), respiratory insufficiency appears to have been initiated by neurogenic causes. In patients whose initial shunt was calculated at < 5%, prolonged anesthesia resulted in small increases (up to 6% to 7.5%) in most instances, quickly reversed in the postoperative period. Little or no change with time was seen in patients who had initial shunts of 9% to 15%. Patients who had large ventilation-perfusion abnormalities initially, did not improve with increased minute volume ventilation, and the use of positive end-expiratory pressure during the operative period. Average stay in the intensive care unit for patients in Group 1 was 3.2 days. Patients in Group 2 spent an average of 7.6 days in the intensive care unit. All patients required initial ventilatory support, and 2 to 5 days elapsed before adequate spontaneous arterial oxygenation was achieved. In Group 3, average intensive care unit stay was 5.1 days. Despite vigorous respiratory therapy, arterial blood remained significantly desaturated for an average of 5 days or until the death of the patient. Pus cells were found in en- 770 J. Neurosurg. / Volume 50 / June, 1979

4 Pulmonary shunt in head injury dotracheal smears after 24 hours of intubation and controlled ventilation in all patients. No correlation was found between abnormal findings on chest films (10 patients) and amount of pulmonary shunt. Other baseline recordings are shown in Fig. 2. Again abnormal findings were unrelated to shunt percentage, particularly with regard to hypertension where division between the three groups was 11, 10, and 14, respectively. The ASA at-risk classification correlated well with outcome: 97% of those dying were Class III to V (Fig. 3). Discussion Venous admixture is a significant factor in the production of arterial hypoxemia. It may be produced by perfusion of poor or non-ventilated alveoli or by the opening of anatomic arteriovenous anastomoses or redistribution of flow through preferential flow channels? 1 Transient respiratory arrest at the time of head injury may cause diffuse microatelectases, resulting in large areas of inadequately ventilated lung. If the head injury is severe enough, and particularly if the hypothalamus is involved, it has been postulated that there is a sudden, centrally mediated, massive sympathetic discharge that produces intense, generalized, but transient vasoconstriction? 5 A sudden shift of blood results from the high-resistance systemic circulation to the lower-resistance pulmonary system. Increased flow through both regular channels and newly opened arteriovenous anastomoses further aggravates the hypoxic state. A primary reduction in cardiac output has been described which could further increase the pulmonary venous congestion, ~ but it is noteworthy in our patients that, whereas shunting was significant, frank pulmonary edema was observed only in three patients initially and in an additional four some 24 hours after injury. Intubation and hyperventilation should reverse an atelectatic state by re-expansion of alveoli. This effect is only partly realized as the initial arterial hypoxemia observed by us and others in the emergency room in spontaneously breathing patients is greater than that found in the operating room situationj The continued existence of significant shunt is probably due to several factors. First, total lung capacity is about 6 liters, whereas volumes used even in hyperventilation rarely exceed 900 cc. There is thus little opportunity to correct miliary collapse that may have occurred in the remaining major part of the lung. The problem is compounded as the patient is lying supine and neither coughing nor moving. Second, increases in pulmonary vascular pressures will produce some degree of pulmonary edema which further interferes with gas exchange. Pulmonary hypertension and hypervolemia injure pulmonary blood vessels and alter pulmonary capillary permeability producing microhemorrhage? 5 Thus, after the transient systemic and pulmonary NO. CAS ES 3O 20 I0 IE IIE IIIE IVE VE A.S.A CLASSIFICATION FIG. 3. Distribution of patients according to the American Society of Anesthesiologists' at-risk assessment, based on the physical status of the patient. E connotes an emergency situation. vascular hypertension subside, damaged pulmonary vasculature remains. We confirmed the finding of others that chest x-ray changes lag 12 to 14 hours behind arterial blood gas alterations, making the former a poor initial diagnostic toolj Regarding the second goal of our study (an attempt to differentiate between peripheral and central causes of respiratory insufficiency), in the absence of autopsy data, ICP measurements, and lung scans, we can only assume, by excluding other causes, that arterial hypoxemia was due to neurogenic causes in over 70% of our patients. As in previous studies, we found that, in patients who died, hypoxemia was refractory to controlled hyperventilation and increased inspired oxygen concentration for prolonged periods of time. u It was, however, noteworthy that three patients who had low scores on the Glasgow Coma Scale (< 6) but who were not hypoxic, survived with minimal neurological deficit after a period of intensive care. Conclusions Pulmonary shunt is a prognostic indicator of outcome in head-injured patients. Correlation with GCS values appears to be close enough to warrant consideration of shunt calculations in assessment of outcome in intracranial trauma. As with GCS recordings, shunt calculations are of least prognostic value following gunshot wounds. It must be assumed that all of these patients are hypoxic until proven otherwise. Respiratory dysfunc- J. Neurosurg. / Volume 50 / June,

5 E. A. M. Frost, C. U. Arancibia and K. Shulman tion appears to be due to neurogenic causes in the majority of patients and to occur at the time of injury. Our data indicate that patients over 50 years old with an initial shunt above 15% will not survive. In severe head injury, when a large pulmonary shunt is present, it does not appear that intensive respiratory care can prevent death. However, if initial pulmonary shunting ~s small, even in the event of profound coma, the prognosis may be good with maximum supportive care. References 1. Brown RS, Shoemaker WC: Sequential hemodynamic changes in patients with head injury: evidence for an early hemodynamic defect. Ann Surg 177: , Cannon WB: Cerebral pressure following trauma. Am J Physiol 6:91-121, Don HF, Wahba M, Cuadrado L, et al: The effects of anesthesia and 100 per cent oxygen on the functional residual capacity of the lungs. Anesthesiology 32: , Don HF, Wahba WM, Craig DB: Airway closure, gas trapping, and the functional residual capacity during anesthesia. Anesthesiology 36: , Frost EAM, Arancibia CU, Tabbador K: Anesthesia for intracranial aneurysm surgery. Presented at the Southern Society of Anesthesiologists Annual Meeting, South Carolina, March, Horsley V, Kramer SP: Cited in reference 9 7. Katsurado K, Yamada R, Sugimoto T: Respiratory insufficiency in patients with severe head injury. Surgery 73: , Laws AK: Effects of induction of anaesthesia and muscle paralysis on functional residual capacity of the lungs. Can Anaesth Soe J 15:3-331, Lewin W: Changing attitudes to the management of severe head injuries. Br Med J 2: , Marshall BE, Cohen P J, Klingenmaier CH, et al: Pulmonary venous admixture before, during, and after halotbane: oxygen anesthesia in man. J Appl Physiol 27: , 1969 ll. Maxwell JA, Goodwin JW: Neurogenic pulmonary shunting. J Trauma 13: , Panday J, Nunn JF: Failure to demonstrate progressive fall of arterial PO2 during anesthesia. Anaesthesia 23:38-46, Rehder K, Marsh HM, Rodarte JR, et al: Airway closure. Anesthesiology 47:40-52, Shapiro BA: Clinical Application of Blood Gases. Chicago: Year Book Medical, 1973, p Theodore J, Robin ED: Pathogenesis of neurogenic pulmonary oedema. Lancet 2: , 1975 This work was supported in part by NIH Grant NO 1 NS I. This paper was presented at the Annual Meeting of the American Association of Neurological Surgeons in New Orleans, Louisiana, April 23-27, Address reprint requests to: Elizabeth A. M. Frost, M.D., Department of Anesthesiology, Albert Einstein College of Medicine, Yeshiva University, 1300 Morris Park Avenue, Bronx, New York J. Neurosurg. / Volume 50 / June, 1979