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

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.

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 best predictor and, again, would have correctly identified 9 of 11 unsuccessful cases and all 18 of the successful cases. The following prediction rule using two variables was developed after studying the first l3 patients: Classify the patient as a potential failure if the initial gastric pH was less than 7.30 or if it decreased by 0.09 or more during the weaning attempt. It was 100% successful in classifying the original 13 patients, as well as the remaining 16 patients that followed.

Data on pHi and other variables commonly used to predict weaning success are summarized in Table 3. The sensitivity, specificity, and positive predictive value, negative predictive value, and accuracy variables for pHi and other commonly measured indices are compared in Table 4. We specifically used the threshold values from a recent report [4]. As shown, the sensitivity of predicting weaning success was similar for most of the indices; however, specificity of the gastric pHi was clearly superior to the specificities of the other indices. The positive predictive value for gastric pHi was 100% (CI, 81% to 100%), whereas for the remaining four commonly used indices, the CI ranged from 60% to 69%. The gastric pHi during mechanical ventilation and during weaning is shown in Figure 1. Using both pHi indices (that is, pHi < 7.3 and a decrement in pHi > 0.09), a clear demarcation with respect to outcome is evident. All patients that had a decrease in pHi of more than 0.09 or had a pHi of less than 7.30 during weaning failed the weaning trial. Two patients had a pHi of less than 7.30 and had an increase in pHi during weaning. We believe that the lower pHi in these patients may in itself predict a poor outcome in weaning.

View larger version:

• In this window
• In a new window

Figure 1. Among patients who were successfully weaned ( = 18), the mean gastric intramural pH (pHi) level did not change (7.45 during mechanical ventilation compared with 7.46 during weaning; = 0.29). Among patients in whom weaning failed ( = 11), pHi decreased from 7.36 to 7.09 ( = 0.003). Gastric intramural pH during mechanical ventilation and during weaning.nPnP

Discussion

The main finding of our study was that gastric pHi decreased significantly after approximately 20 to 30 minutes in patients who failed weaning from mechanical ventilation, whereas pHi was unchanged in patients who were successfully extubated. The second important finding of our study was that commonly used weaning indices (respiratory rate, tidal volume, maximum negative inspiratory pressure, and the ratio of frequency/tidal volume) were not good predictors of weaning failure in our study group. Yang and Tobin [4] measured frequency and tidal volume after discontinuation of mechanical ventilation, whereas we measured these variables during low-level pressure support, which may, in part, explain the lower specificity for frequency, tidal volume, and frequency/tidal volume ratio in our study.

Weaning from mechanical ventilation adds a substantial amount of work to the respiratory pump and cardiovascular system. By some estimates, oxygen consumption increases by 25% during weaning, which must be met by a similar increase in oxygen delivery [19, 20]. In general, the cardiovascular status of the patients is ignored during weaning, and most reports have emphasized respiratory indices as variables to predict weaning success or failure. However, it is apparent that if the increase in oxygen consumption during weaning is not met by an increase in cardiac output and oxygen transport, the increased blood-flow demands of the respiratory muscles will result in a commensurate reduction in blood flow to the splanchnic bed. Thus, we studied gastric pHi, which is a good index of splanchnic ischemia.

The premise of our study was that splanchnic blood flow diminishes early during periods of low flow or during high demand for blood flow in other areas. Using an ultrasonic flow probe around the superior mesenteric vein, Friesen and colleagues [21] evaluated gut blood-flow measurements during hemorrhagic shock in animals, showing that the gut became ischemic early (before the whole body) during progressive hemorrhage. In addition, they concluded that gastrointestinal pHi, based on measurements of intraluminal PCO ₂ and arterial bicarbonate, was a useful indicator of the adequacy of tissue oxygenation during these experiments [21]. Similarly, Gutierrez and coworkers [5] evaluated gastric pHi as a therapeutic index of tissue oxygenation in critically ill patients. In their study, patients admitted to the intensive care unit with a normal pHi had higher survival rates than controls [5]. These same investigators did another study [7] in which they measured gastric pHi at admission to the intensive care unit and 12 hours later, and they found that the index was useful in predicting patient survival. They advocated using gastric pHi as a useful addition to patient monitoring in the intensive care unit setting.

Most studies investigating gastric pHi have used luminal gastric tonometry, which requires approximately 60 to 90 minutes for equilibration between gastric juice CO₂ and the tonometer (saline) CO₂. Sun and colleagues [22] directly measured gastric intramural P_co2 and pH in a model of anaphylactic shock. The response time of the gastric PCO ₂ sensors was rapid (52 seconds). In their study, gastric CO₂ increased from 48 6 mm Hg to 133 5 mm Hg, and pHi decreased from 7.35 0.01 to 6.96 0.01 for a period of 30 minutes after induction of anaphylaxis. Thus, one of the main reasons we used direct measurement of gastric juice CO₂, as opposed to tonometry, was that changes in pHi are likely to occur within 20 to 30 minutes and that the longer periods required for tonometry (that is, 60 to 90 minutes) [9] are not practical in critically ill patients, because respiratory distress or ventilatory failure may be clinically obvious by this time. Hussain and colleagues [23] studied respiratory muscle blood flow after the induction of endotoxemia in spontaneously breathing dogs. Significant blood-flow diversion to the respiratory muscles was evident, whereas blood flow to the splanchnic bed and other nonvital organ systems decreased significantly. After mechanical ventilation, respiratory muscle blood flow decreased to control values, with a commensurate increase in perfusion to other vital and nonvital organs (for example, the splanchnic bed). Our findings support the hypothesis that during periods of high demand by the respiratory muscles, as might occur during weaning, blood-flow diversion away from the splanchnic bed might be reflected by a decrease in gastric pHi. We hypothesize that patients who failed weaning trials had an inadequate ability to increase their oxygen delivery; thus, they could not meet the excessive metabolic demands of the respiratory pump and had to resort to redistribution of blood flow and, therefore, splanchnic ischemia occurred. Ordinarily, this phenomenon is not detected until overt pump failure develops and patients require reintubation.

Weaning is essentially a form of endurance exercise testing during which both the ventilatory and cardiovascular pumps are stressed. Blood-flow diversion to the respiratory muscles and other vital organs that may be indirectly inferred from the changes in gastric pHi may also reflect undue loading of the respiratory pump during the weaning process. Although the respiratory muscles may initially cope with the increased demands placed on them, they may eventually fail as effective pressure generators, and task failure (ventilatory failure) will follow [24]. Thus, changes in gastric pHi may herald weaning ventilatory failure before any change occurs in arterial blood gas values. Our results suggest that this simple and rapid measurement may be of great value in predicting the likelihood of weaning success or failure during weaning trials.

Abbreviations

pHi = intramural pH