Gastric Intramural pH as a Predictor of Success or Failure in Weaning Patients from Mechanical Ventilation

Many different criteria have been used to predict the outcome of trials [1-4] to wean patients from mechanical ventilation. The ideal test would be one that could predict the outcome rapidly and accurately. A well-publicized study [4] found that rapid, shallow breathing, as shown by a frequency/tidal volume ratio greater than 105, was the most accurate predictor of weaning failure, whereas a value less than 105 was an accurate predictor of success. Most of the commonly used indices, reported to be reliable, have good sensitivity and negative predictive value; that is, they are good predictors of successful extubation. However, none of these indices reliably predict weaning failure [1-4]. Gastric intramural pH (pHi) has recently been reported to predict the risk for massive gastrointestinal bleeding, sepsis, and multiple organ failure, as well as outcome, in critically ill patients [5-9]. Gastrointestinal mucosal ischemia ensues early after either hemodynamic compromise in critically ill patients or after blood-flow redistribution to vital organs in certain conditions, despite hemodynamic stability [10, 11]. For example, Magder and coworkers [12] evaluated respiratory muscle blood flow after acute lung injury in dogs. Respiratory muscle blood flow increased almost three times in these animals, whereas renal and splanchnic blood flow decreased substantially despite the maintenance of normal cardiac output and blood pressure [12]. In the presence of reduced splanchnic blood flow or increased tissue demands, ischemia, hypoxemia, and acidosis of the gut wall develops. We postulate that significant gastric mucosal ischemia may develop in patients during unsuccessful attempts to wean them from mechanical ventilation if blood flow from nonvital organs is diverted to meet the increased demands of the respiratory muscles, particularly if oxygen delivery is inadequate. Although the respiratory muscles may initially cope with the increased loads imposed on them during the weaning trial, they may eventually fail as pressure generators, resulting in ventilatory (task) failure. We tested the hypothesis that gastric pHi can be used as a rapid indicator of blood-flow diversion from the splanchnic bed in patients in whom the demands of the respiratory pump during weaning trials are excessive or who have inadequate oxygen delivery to meet these demands. Measurement of the gastric pHi may assist clinicians in predicting outcomes of the weaning trials. Methods Patients We recruited 29 consecutive patients from the respiratory intensive care unit (16 men, 13 women; mean age [SD], 66 17) who were thought by their primary physicians to be weanable from mechanical ventilation. Criteria for inclusion in the study were as follows: stable hemodynamics, adequate gas exchange, a vital capacity greater than 10 mL/kg, a maximum inspiratory pressure of 20 to 30 cm H2O, a resting minute ventilation of less than 10 L/min with the ability to double, and a dead-space/tidal volume ratio of less than 0.60. These patients had received mechanical ventilation because of respiratory failure (chronic obstructive pulmonary disease [n = 24], neuromuscular weakness (n = 4), and aspiration pneumonia [n = 1]) for at least 48 hours before entry into the study. All patients were ventilated on Puritan-Bennett 7200a ventilators (Carlsbad, California). Before weaning trials, all patients had adequate gas exchange, and none had dyssynchronous movements of the chest wall and abdomen during mechanical ventilation. All patients had nasogastric tubes in place and were receiving ranitidine. Intraluminal production of carbon dioxide is enhanced by the titration of gastric bicarbonate by hydrogen ion, which can result in an underestimation of gastric pHi. This can be eliminated by the use of histamine-2-receptor blockade [13]. Study Protocol The study was approved by our institutional review board. After receiving mechanical ventilation on an assist-control mode overnight, patients were placed on pressure support at levels judged to overcome the resistance of the endotracheal tube and ventilatory circuit (about 7 to 8 cm H2O). Samples of gastric juice were obtained 1 hour after enteral feeds. Before the institution of pressure support, 3 mL of gastric juice was obtained from the nasogastric tube. The first 1 mL of fluid was discarded to account for the dead space of the nasogastric tube. Samples of arterial blood were obtained simultaneously and taken to the blood gas laboratory immediately for measurement of Pco 2 and pH. After the patients had been placed on pressure-support weaning trials for approximately 20 to 30 minutes, an additional 3 mL of gastric juice and a sample of arterial blood were obtained simultaneously. Physicians attending to the patients were blinded to the nature and results of the study and were not told about the measurements being done. However, the results of blood gas studies and conventional measurements of lung mechanics were available to them. The investigators did not interfere with the primary caretakers' decisions regarding extubation or reinstitution of mechanical ventilation. Additional data, obtained at the time of gastric juice and arterial blood sampling, included respiratory frequency, tidal volume, minute ventilation, blood pressure, and heart rate. Negative inspiratory pressure during spontaneous ventilation was measured through a unidirectional valve attached to the airway, and the most negative pressure recorded during airway occlusion was considered the maximum negative inspiratory pressure [14]. Patients were considered to be successfully weaned if they were able to sustain spontaneous ventilation for more than 24 hours after extubation [1-4]. Calculation of Gastric Intramural pH Gastric pHi was calculated using the Henderson-Hasselbalch Equation as follows: 6.1 + log bicarbonate/(gastric Pco 2 0.0307) where bicarbonate is the bicarbonate concentration obtained from arterial blood and gastric Pco 2 is the value determined in a gastric juice specimen. The reliability of this method has been validated by previous investigators [15-17]. Two important assumptions are made in calculating gastric pHi: 1) Pco 2 in the lumen of the stomach is the same as that in the surrounding gastric tissue wall, and 2) bicarbonate in tissue is equal to that in arterial blood. As other investigators have suggested, because of the titration of protons by bicarbonate, the tissue CO2 concentration increases [15-17]. Cunningham and colleagues [18] found no differences between the Pco 2 measured in the lumina of rat ilea and that measured in the walls of the rat ilea during conditions of ischemia. In addition, Fiddian-Green and colleagues [6] found that pHi calculated from Pco 2 in gastric juice and arterial bicarbonate was linearly related to pHi measured directly with a pH probe. Data Analysis Paired t-tests were used to compare the values obtained during mechanical ventilation with those obtained during weaning trials. Unpaired t-tests were used to compare values from the group that was successfully weaned with those from the group that were not. Ninety-five percent CIs, sensitivity, specificity, positive predictive value, and negative predictive value were also calculated for all variables. A stepwise discriminant analysis, including all predictor variables, was used to determine which variables were useful in distinguishing between groups and to identify the best predictor of weaning success. These variables included nine potential predictors of successful weaning from mechanical ventilation: respiratory rate, tidal volume, systolic blood pressure, heart rate, arterial Pco 2, arterial Po 2, arterial pH, gastric Pco 2, and gastric intramural pH. Each variable was measured before and during attempted weaning. Results Respiratory and gas exchange data obtained during mechanical ventilation for patients who were successfully weaned and those who were not are shown in Table 1. No statistical differences in any of these variables measured during assisted mechanical ventilation were found between patients who were successfully weaned and those who were not. Respiratory rate increased and tidal volume decreased statistically during weaning trials in both groups (Table 2). Similarly, neither group showed changes in blood pressure, arterial Pco 2, arterial Po 2, and arterial pH within the time frame that these variables were measured (Table 2). However, in contrast to the patients who were successfully weaned from mechanical ventilation, patients in whom the weaning trial failed showed substantial changes in gastric Pco 2 and gastric pHi. Three of the patients in the group in whom the weaning failed were extubated; however, they had to be reintubated within 3 to 20 hours. The remaining eight patients were placed back on the ventilator after 60 to 120 minutes of weaning because of increasing effort, excessive diaphoresis, hypertension, tachycardia, arrhythmias, and a subjective sense of dyspnea. In three of these patients, the gastric Pco 2 was measured 20 minutes after they were placed back on mechanical ventilation; their gastric Pco 2 returned to levels observed before weaning. The mean gastric Pco 2 in these patients before the weaning trials (on mechanical ventilation) was 39 13 mm Hg and, after weaning trials (again, on mechanical ventilation), the mean value was 39 12 mm Hg. Table 1. Characteristics of Patients While on Assisted Mechanical Ventilation Who Were Later Successfully Weaned or in Whom Weaning Failed* Table 2. Change in Physiologic Variables during Weaning for Patients Who Were Successfully Weaned and Patients Who Failed Weaning* Using discriminant analysis, we found that gastric pHi during weaning was the best single predictor of weaning outcome. A decrement in pHi of more than 0.09 would have classified 9 of 11 unsuccessful cases correctly and all 18 successful cases correctly. An increase in gastric Pco 2 of 10 mm Hg or more was the next