- Email: [email protected]
Controlled oxygen administration in acute respiratory failure in chronic obstructive pulmonary disease
CLINICAL STUDIES
Controlled Oxygen Administration in Acute Respiratory Failure in Chronic Obstructive Pulmonary Disease A Reappraisal
ROGER C. BONE, M.D.* ALAN K. PIERCE, M.D. ROBERT L. JOHNSON, Jr., M.D. Dallas, Texas Kansas City, Kansas
From the Pauline and Adolph Weinberger Laboratory for Cardiopulmonary Research, Department of Internal Medicine, University of Texas Southwestern Medical School at Dallas, Dallas, Texas; and The University of Kansas Medical Center, Kansas City, Kansas. Requests for reprints should be addressed to Dr. Roger C. Bone. Manuscript accepted May 26, 1976. Present address: Chief, Pulmonary Division, University of Arkansas Medical Center, Liile Rock, Arkansas 72201. l
898
December 1978
Controlled oxygen therapy may aggravate carbon dioxide retention during acute exacerbations of chronic obstructive pulmonary disease (COPD). Of 50 consecutive patients with COPD and acute respiratory failure, 13 required intubation because of carbon dioxide narcosis. With discriminant analysis of their arterial oxygen tension ( Pa02) and pH on admission, a diagram separated patients into those at high risk and those at low risk for carbon dioxide narcosis. This diagram was then used to predict carbon dioxide narcosis in 73 patients with COPD and acute respiratory failure who were treated with controlled oxygen. In 16 of these patients carbon dioxide narcosis developed. Thirteen (61 per cent) were predicted by the diagram to be at high risk for this complication. Only two (4 per cent) patients judged by the diagram to be at low risk for carbon dioxide narcosis required mechanical ventilation. utilizing an oxygen tension (PO,), carbon dioxide tension (PCO,) diagram a patient’s ventilatory response was compared to that of ambulatory patients with COPD. These data suggest that hypoxemia and acidosis are more discriminatory for “carbon dioxide narcosis” than hypercapnia. Since the introduction of the concept of controlled oxygen therapy by Barach [ 11, and its practical implementation by Campbell and Gabbis [ 2,3], the administration of progressive increments of inspired oxygen has become the usual therapy for patients with chronic obstructive pulmonary disease (COPD) in acute respiratory failure [4-81. Most such patients respond to an increase in the fraction of inspired oxygen (Fro,) by a diminution in arterial hypoxemia and, in conjunction with other therapy, an improvement in clinical status. However, some patients experience progressive hypercapnia and acidosis with resultant confusion, stupor or coma even when the Fro2is only modestly increased [9]. These latter patients usually require an artificial airway and mechanical ventilation. No systematic data have been reported from which the clinician may predict which course an individual patient may take when the Fro2 is increased. We have investigated the response of patients with COPD to graded increments in F102.Four groups of patients were studied: (1) ambulatory patients without hypercapnia, (2) ambulatory patients with hypercapnia, (3) acutely ill patients managed by a known Fro2 and (4) acutely ill patients in a separate medical center managed by a fixed low flow of
The American Journal of Medicine
Volume 85
CONTROLLED
TABLE I
OXYGEN
THERAPY-BONE
ET AL.
Pulmonary Function of Stable Outpatients with COPD Without Hypercapnia Fraction of inspired Oxygen _
0.21
0.28
Patient
PaOP (mm Hg)
PaCO* (mm Hg)
1 2 3 4 5 6 7 8 9 IO Mean SD
49 73 61 79 76 68 92 84 104 64 75 16
37 38 39 35 31 39 32 40 31 37 36 4
PaO? (mm Hg)
PaOa (mm Hg)
PaC02 (mm Hg)
PaOz (mm Hg)
PaCOz (mm Hg)
35 36 57 41 33 40 39 43 34 38 38 3
109 124 118 111 142 137 120 162 166 107 130 21
38 40 40 42 35 41 39 41 33 38 39 3
135 162 146 162 180 175 185 207 221 155 173 27
39 40 42 40 35 42 35 40 34 38 39 3
73 103 81 85 104 78 91 106 124 92 94 16
NOTE: The chanse in PaCOp from F ,n, 0.21 to 0.28 was P
supplemental oxygen. Our data suggest that the carotid body response to an Increased FIo2 among anibuiatory and acutely ill patients is appropriate. Further, an assessment of the arterial carbon dioxide tension (PaOo) and pH on admission gives a reasonable index to the ultimate
occurrence
of stupor or coma among acutely
ill patients treated with oxygen. METHODS Twenty
0.40
0.35 PaCOz (mm Hg)
patients with COPD, who had been documented to have no change in symptoms, spirometry or blood gases during several weeks of observation, were selected for study from the Outpatient Clinic of Parkland Memorial Hospital. Each patient had repe@ediy demonstrated a forced expiratory volume in 1 second (FEVI.o) less than 1.5 liters, and none had an increase in FEVI,o of greater than 10 per cent following inhaled bronchodiiators. Ten of these patients had a PaGO of 140 mm Hg and IO had a PaCOp 243 mm Hg while breathing room air. After obtaining informed consent, a Teflon@ catheter was placed in a radial artery under local anesthesia. Arterial blood was obtained ior analysis while each patient successively breathed 21, 28, 35 and 40 per cent oxygen from a reservoir-two way valve system. Each oxygen-nitrogen mixture was breathed for at least 30 minutes, and exhaled nitrogen was analyzed to insure equilibration with each new test mixture. The courses of 50 consecutive patients, admitted to the Respiratory Intensive Care Unit (RICU) of Parkland Memorial Hospital with acute respiratory failure complicating COPD between October 1972 and March 1974, were reviewed. The diagnosis of acute res@ratory failure was based on clinical deterioration of steady state symptoms for hours to days preceding the admission. Patients were excluded from the study if they underwent intubation for any reason in the Emergency Room before admission to the RICU or if they had a history of tranquilizer or sedative use. The medical house staff assigned to the RICU acted as the primary physicians under the daily supervision of one of us. Each patient was given 24 per cent oxygen by a Venturi type mask in the
0.21 to 0.35,
P <0.005;
0.21 to 0.40,
P <0.0005.
All other comparisons
Emergency Room and, subsequently, intravenous aminophyilin, replacement fluids, inhaled bronchodiiators, postural drainage with chest percussion and antimicrobial agents if the sputum was purulent. if, after a short interval, the primary care physicians judged hypoxemia to remain severe, the FIo2 was increased to 0.28. Endotracheai intubation and mechanical ventilation were deemed necessary only when the patient became stuporous and unable to cooperate with therapy; in no instances were arbitrary blood gas criteria used as an indication. An additional group of 73 patients with COPD in acute distress consecutively admitted to the RICU of the University pf Kansas Medical Center from September 1974 to December 1976 was studied. These patients were under the primary care of the medical house staff assigned to that unit under the daily supervision of one of us. The precepts of treatment and the indication for intubation were the same as those for the Dallas patients except that oxygen was administered at a fixed low flow, usually 1 liter/min, initially, instead of at a fixed concentration. After the initial blood gas determination was obtained at 1 iiter/min, oxygen was adjusted to obtain a Pa02 of 50 to 60 mm Hg. The results of the Dallas series were unknown to the primary physicians and were not utilized in determining which patients required intubation. The prediction formulas of Morris et al. [lo] for spirometric studies were utilized for normal values for the ambulatory patients. Arterial blood gas and pH analyses were performed in Instrumentation Laboratories 113 pHIBlood Qas Analyzers which had been calibrated by gases of known composition. Blood obtained from outpatients was analyzed immediately after being obtained, whereas blood from patients in intensive care units was held on ice for as short an interval as practical before analysis. Venturi masks were monitored before use by a Beckman physiologic gas analyzer and were consistently within 1 per cent of the rated oxygen concentration. Statistical comparisons were made by the paired t test for values from the same patients and by the group t test among patients. Probability values of <0.05 were considered significant.
December 1978
The American Journal of Medicine
Volume 65
897
CONTROLLED
OXYGEN
TABLE II
THERAPY-BONE
ET AL
Pulmonary Function of the Stable Outpatients with COPD with Hypercapnia Fraction of Inspired Oxygen 0.21 PaOz (mm W
Patient
0.28 PaCol (mm Hg)
67 56 61 42 52 58 63 57 51 90 80 13
11 12 13 14 15 16 17 18 19 20
Pa02 (mm Hg) 88 81 87 57 81 75 97 88 71 116 84 16
45 56 52 51 50 51 47 44 43 44 48 4
0.40
0.35 PaC02 (mm W 50 54 51 50 55 53 48 48 47 44 50 3
PaO, (mm W)
PaC02 (mm W
PaO;, (mm W
PaCO* (mm W
128 121 126 77 113 97 116 116 103 154 115 21
51 61 53 54 58 53 49 50 49 42 52 5
147 168 147 81 128 130 147 146 119 176 139 27
52 56 50 54 65 50 49 53 51 42 52 6
NOTE: The changes in PaC02from one F~Qlo another were FIo2 0.21to 0.28,P <0.05; F1020.21to 0.35,P cO.0025; F1020.21t0
RESULTS Four
of 10 ambulatory patients without hypercapnia were men. Their age was 52 f 12 years (mean f SD). The FEVl,o was 0.90 f 0.59 (mean f SD). Most ambulatory outpatients without hypercapnia had mild hypoxemia breathing room air (Table I). Increasing F102 caused progressive increases in PaO,; an FIo2 of 0.28 resulted in a Pa02 of approximately that observed in normal persons breathing room air at this altitude. The PaCOp tended to increase as the Flop was increased from 0.21 to 0.28, but statistical significance was not achieved (P
or 35 per cent oxygen was not significantly different, but the PaOp of the eucapnic group was significantly higher while breathing 40 per cent oxygen (P < 0.05). As expected from the method of patient selection, the PaC02 of the hypercapnic group was significantly greater than that of the eucapnic group, and it remained so at each Flop. For the patients ifi the hypercapnic group, the hypercapnia became worse at higher F102. Thirteen of the 50 (26 per cent) acutely ill patients in the Dallas series became stuporous during treatment and ultimately required artificial airways and mechanical ventilation (Table Ill). These patients could not be distinguished from the 37 patients who were successfully managed by graded oxygen therapy by age, sex, most recent outpatient spirometric values or outpatient blood gas values. Similarly, neither group had significantly different outpatient values from the group of ambulatory outpatients with hypercapnia. However, blood gas values in the patients in acute respiratory failure were significantly worse on admission than had been observed while they were outpatients, and those patients who ultimately required mechanical ventilation had significantly worse Pa02, PaCO* and pH values on admission than the patients who were managed without mechanical ventilation. Thus, the condition in the pa-
Patient Data of 50 Acutely iii Patients with COPD (Dallas Series)
TABLE iii
Men (no.)
Age W
56 f 8
17
58f9 PNS
5 NS
FEV1.o
(liters)
Most Recent Stable Outpatient Measuremenls PaOa PaCOa (mm W (mm Hg)
Patients Not RequiringArtificial Airways 0.28 54 f 8 49 f 9 13 Patients Requiring Artificial Airways 0.66 f 0.24 55f 12 51 f 10 NS NS NS 0.67 f
PH (mm W and 7.40 and 7.40
NOTE: NS = not significant.
898
0.40,P
December
1978
The American
Journal
of Medicine
Volume
85
Mechanical f 0.03 Mechanical f 0.03 NS
PaOz (mm Hg) Ventilation 41 f9 Ventilation 32 f 5
Admission Measurements PaCO, (mm W
PH W)
51 f9
7.39 f
57 f 7 <0.03
7.32 f 0.06
0.05
CONTROLLED OXYGEN THERAPY--OONE
ET AL
0
A
AM
7.1 IO
I 20
1 40
I 30
Pa02
I 50
I 60
1 70
jmmHgi
figure 1.
Pa02 and pH on admission (Dallas patients). Patients in whom somnolence developed while they were receiving controlled oxygen therapy were, in general, severely hypoxemic or had a combination of moderate hypoxemia and acidosis. This graph o&nonstrates the importance of hypoxemia and acidosis as risk factors for carbon dioxide narcosis during controlled oxygen therapy. The line separating high from low risk patients was found by discriminant analysis (pti - 7.66 - 0.00919 PO,). A = intubated patients. l = nonintubated patients.
tients who ultimately required more aggressive therapy deteriorated more markedly from their usual status. The initial response to oxygen therapy did not clearly separate the ultimate courses of the patients. Among patients begun on 24 per cent oxygen, the PaOs increased by 11 f 7 mm Hg for conservatively-managed patients and by 8 f 5 mm Hg for ventilator-managed patients (P not significant); the change in PaCOs was i-7 f 9 mm Hg and + 10 f 7 mm Hg (P not significant), respectively. Only the change in pH suggested the ultimate outcome, -0.01 f 0.06 versus -0.06 f 0.04 (P <0.05), respectively. Among patients treated with 28 per cent oxygen, the early blood gas values did not differ between the patients managed by oxygen alone and those managed by mechanical ventilation; the values were Pa02 + 21 f 88 mm Hg versus -Ill f 11 (mean f SD) mm Hg (P not significant), PaCO* -t-4 f 9 mm Hg versus +4 f 13 mm Hg (P not significant) and pH -0.07 f 0.09 versus -0.09 versus -0.01 f 0.04 (P not significant), respectively. At their worst level of hypercapnia blood gas values in the patients who were managed by graded oxygen alone were not as severely deranged as those in the patients who became stuporous. The respective blood gas values were PaOs 55 f 10 mm Hg versus 41 f 9 mm Hg (P
versus 7.27 f 0.04 (P
December 1978
The American Journal of Medicine
Volume 65
899
CONTROLLED OXYGEN THERAPY-BONE
,.I\
10
ET AL.
20
30
40
50 PC+
a
70
80
ImmHg)
I
Figure 2. Pa02 and pti on admission (Kansas patients). The line separating high and low risr patients for carbon dioxide narcosis drawn from data from the Dallas patients also separates high and low risk patients in the Kansas patients. A = in&bated patients. 0 = nonintubated patients.
respiratory failure died. All deaths occurred in patients who received intubation. COMMENTS In patients with serious chronic airways obstruction episodes of acute deterioration may develop manifested primarily by increasing dyspnea. Although arterial blood gas values and ventilatory mechanics are usually more deranged than values obtained in each patient during the steady state, if these are known [ 111, the absolute levels of Pa02, PaCO* or pH do not clearly separate stable from acutely ill patients [ 121. Thus, the diagnosis of acute respiratory failure is based on clinical findings. We have utilized clinical criteria for admission to this series, but we believe the patients to be representative of those diagnosed as having acute respiratory failure by most experienced clinicians. Since patients with COPD and acute respiratory failure may become stuporous or comatose when oxygen is administered in an uncontrolled manner [ 2,3], therapy is usually initiated with a fixed low concentration or low flow of oxygen [ 131. Most patients show improvement with such therapy, but some become progressively hypercapnic, acidotic and somnolent. We are unable to find published data which suggest which course is likely in an individual patient. Therefore, we have investigated the response of patients with COPD, to oxygen therapy, some of these patients were in their stable state and some were in acute respiratory failure. Rahn and Otis [ 141 have reported the ventilatory response of normal persons to the hypoxemia of alti-
900
December 1978
The American Journal of Medicine
tude; both early and delayed responses were evaluated. Their data are displayed in a PC02-PO2 diagram (Figure 3). In normal persons, hypoxemia does not stimulate excess ventilation nor cause hypocapnia above a PaOp of approximately 65 mm Hg. Below this PaOp, the PaC02 is lower indicating a stimulus to hyperventilation. Normal persons acclimatized to hypoxemia are more sensitive to this stimulus and have a lower PaCO* at any Pa02 than unacclimatized persons. Thus, the normal response of PaCO* to decreases in Pa02 lies between the two extremes indicated by their data [ 141. The response to mild hypoxemia in our ambulatory outpatients without hypercapnia was in the normal range indicated by Rahn and Otis (Figure 3). When their Pa02 was increased by the administration of oxygen, their PaC02 response remained within the normal range with a slope that paralleled that of acclimatized normal persons. Our ambulatory outpatients with hypercapnia had a normal response to increasing Pa02 because the slope of increasing PaCO* was similar to that of normal persons. Thus, ambulatory outpatients with or without hypercapnia responded normally to increasing Pa02. The patients with acute respiratory failure in Dallas who did not require intubation had a more serious derangement of ventilatory response than the stable hypercapnic patients, for although their PaC02 on admission was no higher, their Pa02 was lower. The ventilatory defect was yet more marked in the acutely ill patients who required mechanical ventilation, for their admission Pa02 was significantly less than that in all other groups. Nevertheless, when the Pa02 in these groups of patients was increased by oxygen therapy,
Volume 85
CONTROLLED
_ .-I. -...-.. -.-. 120
OXYGEN
,’
0
.
GROUP
I
-Stable
cold,
GROUP
II
-Stable
10 patients
retention1
cold,
10
!COzretentIonI A
1
GROUP
III-Not
mtubated, [no co2
.
GROUP
IV-Eventually [CO2
0
L-r--20
ET AL
..- .- -.-.-_....-____.___. __._.__._._~..__ _.
[no CO2
100
THERAPY--RONE
1
/
I
37 patents
narcosIsI mtubated.
13 patients
narcorlrl
I-1-I 60
40 Pa02
paWntr
80
ImmHy)
3. Mean values for PaO,-PaCO, on admission plotted with Pa4 and PaC02 at point of greatest hypercapnia after oxygen administration in four se&rate groups of patients with COPD. The hashed area is a plot for normal volunteer subjects and was taken from reference [ 141. Horizontal and vertical bars indicate standard error of the mean for the Pa4 and PaCO,, respectively.
xgure
the PaCO* increased at a slope roughly parallel to the slope of the increase in PaCO* in normal persons over the same range of Pa02, as judged by the Rahn-Otis diagram. Thus, the acutely ill patients responded with an appropriate increase in PaC02 when given oxygen, although this increase occurred from an abnormally high base line PaCOp. We interpret these findings to indicate that the carotid body response to hypoxia is normal in patients with COPD with or without acute respiratory failure. Despite the fact that the response of the carotid bodies in patients with COPD and acute respiratory failure is apparently normal, the sudden increase in PaC02 with oxygen therapy may lead to sufficient cerebral acidosis to cause coma [15-l 71. We have found that the PaOa and pH on admission are reasonable predictors of this event; the lower these values are at the outset of therapy, the more likely oxygen therapy is to result in coma. The pH-PaOa diagram indicated in Figure 1 indicates this relationship in a convenient graph that may prove useful clinically. In the final analysis, the decision that an individual
patient with acute respiratory failure from COPD is sufficiently stuporous to require an artificial airway and mechanical ventilation is a subjective value judgment on the part of the attending physician. We thought it important, therefore, to validate our findings in a different medical environment in which the subtle factors leading to this judgment might be different. The primary care physicians who managed the patients in the Kansas series utilized clinical criteria for intubation of their patients without access to the specific concepts developed in the Dallas series. Nevertheless, the pH-PaOs diagram was similarly predictive of the ultimate course of their patients following oxygen therapy. The pH-Pa02 diagram should not be used to perform intubation, prophylactically, in all patients who are in the high risk group for narcosis following oxygen therapy. The hazards of aggressive intubation in acute respiratory failure from COPD have been previously stressed [2,3]. The pH and PaOp on admission, however, will identify those patients at greatest risk and should assist the physician in planning a therapeutic regimen.
December 1978
The American Journal of Medicine
Vofume 85
991
CONTROLLED OXYGEN THERAPY-BONE ET AL.
REFERENCES 1.
2. 3. 4.
5.
6.
7. 8. 9.
902
Barach AL: Physiological methods in diagnosis and treatment of asthma and emphysema. Ann Intern Med 12: 454, 1938. Campbell EJM, Gabbis T: Mask and tent for providing controlled oxygen concentration. Lancet 1: 468, 1966. Campbell EJM: Management of respiratory failure. Br Med J 2: 1328, 1964. Smith, JP, Stone W, Muschenheim C: Acute respiratory failure in chronic lung disease. Am Rev Respir Dis 97: 791, 1968. Bedon GA, Block J, Ball WC: The 28% Venturi mask in obstructive airway disease. Arch Intern Med 125: 106, 1970. Schiff MM, Massaro D: Effect of oxygen administration on arterial blood gas values in patients with respiratory failure. N Engl J Med 277: 950, 1967. King TKC. Ali N, Briscoe WT: Treatment of hypoxia with 24% oxygen. Am Rev Respir Dis 108: 19, 1973. Eldridge F, Gherman C: Studies of oxygen administration in respiratory failure. Ann Intern Med 68: 569, 1968. Moser KM, Luschsinger PC, Adamson JS, et al.: Respiratory stimulation with intravenous doxapram in respiratory failure. A double-blind cooperative study. New Engl J Med 288: 427,
December 1978
The American Journal of Medicine
10.
11.
12. 13.
14. 15. 16.
17.
Volume 85
1973. Morris JF, Koshi A, Johnson LC: Spirometric standards for healthy nonsmoking adults. Am Rev Respir Dis 103: 57, 1971. Sukumalchandra Y, Park SS, Williams Ml-t. Jr: The effect of intermittent positive pressure breathing (IPPB) in acute ventilatory failure. Am Rev Respir Dis 92: 885, 1965. Lal S: Blood gases in respiratory failure. State on admission to hospital and management. Lancet 1:339, 1965. Campbell EJM: The J. Burns Amberson Lecture. The management of acute respiratory failure in chronic bronchitis and emphysema. Am Rev Respir Dis 96: 626, 1967. Rahn H, Otis AB: Man’s response during and after acclimatization to high altitude. Am J Physiol 157: 445, 1949. Posner JB, Swanson AG. Plum F: Acid base balance in cerebral spinal fluid. Arch Neurol 12: 479, 1965. Penman RWB: Blood lactate levels and some acid-base changes in respiratory failure and their significance in oxygen induced respiratory depression. Clin Sci 23: 5, 1962. Rudolf M, Banks RA, Semple SJG: Hypercapnia during oxygen therapy in acute exacerbations of chronic respiratory failure. Lancet 2: 483, 1977.
Controlled Oxygen Administration in Acute Respiratory Failure in Chronic Obstructive Pulmonary Disease A Reappraisal
ROGER C. BONE, M.D.* ALAN K. PIERCE, M.D. ROBERT L. JOHNSON, Jr., M.D. Dallas, Texas Kansas City, Kansas
From the Pauline and Adolph Weinberger Laboratory for Cardiopulmonary Research, Department of Internal Medicine, University of Texas Southwestern Medical School at Dallas, Dallas, Texas; and The University of Kansas Medical Center, Kansas City, Kansas. Requests for reprints should be addressed to Dr. Roger C. Bone. Manuscript accepted May 26, 1976. Present address: Chief, Pulmonary Division, University of Arkansas Medical Center, Liile Rock, Arkansas 72201. l
898
December 1978
Controlled oxygen therapy may aggravate carbon dioxide retention during acute exacerbations of chronic obstructive pulmonary disease (COPD). Of 50 consecutive patients with COPD and acute respiratory failure, 13 required intubation because of carbon dioxide narcosis. With discriminant analysis of their arterial oxygen tension ( Pa02) and pH on admission, a diagram separated patients into those at high risk and those at low risk for carbon dioxide narcosis. This diagram was then used to predict carbon dioxide narcosis in 73 patients with COPD and acute respiratory failure who were treated with controlled oxygen. In 16 of these patients carbon dioxide narcosis developed. Thirteen (61 per cent) were predicted by the diagram to be at high risk for this complication. Only two (4 per cent) patients judged by the diagram to be at low risk for carbon dioxide narcosis required mechanical ventilation. utilizing an oxygen tension (PO,), carbon dioxide tension (PCO,) diagram a patient’s ventilatory response was compared to that of ambulatory patients with COPD. These data suggest that hypoxemia and acidosis are more discriminatory for “carbon dioxide narcosis” than hypercapnia. Since the introduction of the concept of controlled oxygen therapy by Barach [ 11, and its practical implementation by Campbell and Gabbis [ 2,3], the administration of progressive increments of inspired oxygen has become the usual therapy for patients with chronic obstructive pulmonary disease (COPD) in acute respiratory failure [4-81. Most such patients respond to an increase in the fraction of inspired oxygen (Fro,) by a diminution in arterial hypoxemia and, in conjunction with other therapy, an improvement in clinical status. However, some patients experience progressive hypercapnia and acidosis with resultant confusion, stupor or coma even when the Fro2is only modestly increased [9]. These latter patients usually require an artificial airway and mechanical ventilation. No systematic data have been reported from which the clinician may predict which course an individual patient may take when the Fro2 is increased. We have investigated the response of patients with COPD to graded increments in F102.Four groups of patients were studied: (1) ambulatory patients without hypercapnia, (2) ambulatory patients with hypercapnia, (3) acutely ill patients managed by a known Fro2 and (4) acutely ill patients in a separate medical center managed by a fixed low flow of
The American Journal of Medicine
Volume 85
CONTROLLED
TABLE I
OXYGEN
THERAPY-BONE
ET AL.
Pulmonary Function of Stable Outpatients with COPD Without Hypercapnia Fraction of inspired Oxygen _
0.21
0.28
Patient
PaOP (mm Hg)
PaCO* (mm Hg)
1 2 3 4 5 6 7 8 9 IO Mean SD
49 73 61 79 76 68 92 84 104 64 75 16
37 38 39 35 31 39 32 40 31 37 36 4
PaO? (mm Hg)
PaOa (mm Hg)
PaC02 (mm Hg)
PaOz (mm Hg)
PaCOz (mm Hg)
35 36 57 41 33 40 39 43 34 38 38 3
109 124 118 111 142 137 120 162 166 107 130 21
38 40 40 42 35 41 39 41 33 38 39 3
135 162 146 162 180 175 185 207 221 155 173 27
39 40 42 40 35 42 35 40 34 38 39 3
73 103 81 85 104 78 91 106 124 92 94 16
NOTE: The chanse in PaCOp from F ,n, 0.21 to 0.28 was P
supplemental oxygen. Our data suggest that the carotid body response to an Increased FIo2 among anibuiatory and acutely ill patients is appropriate. Further, an assessment of the arterial carbon dioxide tension (PaOo) and pH on admission gives a reasonable index to the ultimate
occurrence
of stupor or coma among acutely
ill patients treated with oxygen. METHODS Twenty
0.40
0.35 PaCOz (mm Hg)
patients with COPD, who had been documented to have no change in symptoms, spirometry or blood gases during several weeks of observation, were selected for study from the Outpatient Clinic of Parkland Memorial Hospital. Each patient had repe@ediy demonstrated a forced expiratory volume in 1 second (FEVI.o) less than 1.5 liters, and none had an increase in FEVI,o of greater than 10 per cent following inhaled bronchodiiators. Ten of these patients had a PaGO of 140 mm Hg and IO had a PaCOp 243 mm Hg while breathing room air. After obtaining informed consent, a Teflon@ catheter was placed in a radial artery under local anesthesia. Arterial blood was obtained ior analysis while each patient successively breathed 21, 28, 35 and 40 per cent oxygen from a reservoir-two way valve system. Each oxygen-nitrogen mixture was breathed for at least 30 minutes, and exhaled nitrogen was analyzed to insure equilibration with each new test mixture. The courses of 50 consecutive patients, admitted to the Respiratory Intensive Care Unit (RICU) of Parkland Memorial Hospital with acute respiratory failure complicating COPD between October 1972 and March 1974, were reviewed. The diagnosis of acute res@ratory failure was based on clinical deterioration of steady state symptoms for hours to days preceding the admission. Patients were excluded from the study if they underwent intubation for any reason in the Emergency Room before admission to the RICU or if they had a history of tranquilizer or sedative use. The medical house staff assigned to the RICU acted as the primary physicians under the daily supervision of one of us. Each patient was given 24 per cent oxygen by a Venturi type mask in the
0.21 to 0.35,
P <0.005;
0.21 to 0.40,
P <0.0005.
All other comparisons
Emergency Room and, subsequently, intravenous aminophyilin, replacement fluids, inhaled bronchodiiators, postural drainage with chest percussion and antimicrobial agents if the sputum was purulent. if, after a short interval, the primary care physicians judged hypoxemia to remain severe, the FIo2 was increased to 0.28. Endotracheai intubation and mechanical ventilation were deemed necessary only when the patient became stuporous and unable to cooperate with therapy; in no instances were arbitrary blood gas criteria used as an indication. An additional group of 73 patients with COPD in acute distress consecutively admitted to the RICU of the University pf Kansas Medical Center from September 1974 to December 1976 was studied. These patients were under the primary care of the medical house staff assigned to that unit under the daily supervision of one of us. The precepts of treatment and the indication for intubation were the same as those for the Dallas patients except that oxygen was administered at a fixed low flow, usually 1 liter/min, initially, instead of at a fixed concentration. After the initial blood gas determination was obtained at 1 iiter/min, oxygen was adjusted to obtain a Pa02 of 50 to 60 mm Hg. The results of the Dallas series were unknown to the primary physicians and were not utilized in determining which patients required intubation. The prediction formulas of Morris et al. [lo] for spirometric studies were utilized for normal values for the ambulatory patients. Arterial blood gas and pH analyses were performed in Instrumentation Laboratories 113 pHIBlood Qas Analyzers which had been calibrated by gases of known composition. Blood obtained from outpatients was analyzed immediately after being obtained, whereas blood from patients in intensive care units was held on ice for as short an interval as practical before analysis. Venturi masks were monitored before use by a Beckman physiologic gas analyzer and were consistently within 1 per cent of the rated oxygen concentration. Statistical comparisons were made by the paired t test for values from the same patients and by the group t test among patients. Probability values of <0.05 were considered significant.
December 1978
The American Journal of Medicine
Volume 65
897
CONTROLLED
OXYGEN
TABLE II
THERAPY-BONE
ET AL
Pulmonary Function of the Stable Outpatients with COPD with Hypercapnia Fraction of Inspired Oxygen 0.21 PaOz (mm W
Patient
0.28 PaCol (mm Hg)
67 56 61 42 52 58 63 57 51 90 80 13
11 12 13 14 15 16 17 18 19 20
Pa02 (mm Hg) 88 81 87 57 81 75 97 88 71 116 84 16
45 56 52 51 50 51 47 44 43 44 48 4
0.40
0.35 PaC02 (mm W 50 54 51 50 55 53 48 48 47 44 50 3
PaO, (mm W)
PaC02 (mm W
PaO;, (mm W
PaCO* (mm W
128 121 126 77 113 97 116 116 103 154 115 21
51 61 53 54 58 53 49 50 49 42 52 5
147 168 147 81 128 130 147 146 119 176 139 27
52 56 50 54 65 50 49 53 51 42 52 6
NOTE: The changes in PaC02from one F~Qlo another were FIo2 0.21to 0.28,P <0.05; F1020.21to 0.35,P cO.0025; F1020.21t0
RESULTS Four
of 10 ambulatory patients without hypercapnia were men. Their age was 52 f 12 years (mean f SD). The FEVl,o was 0.90 f 0.59 (mean f SD). Most ambulatory outpatients without hypercapnia had mild hypoxemia breathing room air (Table I). Increasing F102 caused progressive increases in PaO,; an FIo2 of 0.28 resulted in a Pa02 of approximately that observed in normal persons breathing room air at this altitude. The PaCOp tended to increase as the Flop was increased from 0.21 to 0.28, but statistical significance was not achieved (P
or 35 per cent oxygen was not significantly different, but the PaOp of the eucapnic group was significantly higher while breathing 40 per cent oxygen (P < 0.05). As expected from the method of patient selection, the PaC02 of the hypercapnic group was significantly greater than that of the eucapnic group, and it remained so at each Flop. For the patients ifi the hypercapnic group, the hypercapnia became worse at higher F102. Thirteen of the 50 (26 per cent) acutely ill patients in the Dallas series became stuporous during treatment and ultimately required artificial airways and mechanical ventilation (Table Ill). These patients could not be distinguished from the 37 patients who were successfully managed by graded oxygen therapy by age, sex, most recent outpatient spirometric values or outpatient blood gas values. Similarly, neither group had significantly different outpatient values from the group of ambulatory outpatients with hypercapnia. However, blood gas values in the patients in acute respiratory failure were significantly worse on admission than had been observed while they were outpatients, and those patients who ultimately required mechanical ventilation had significantly worse Pa02, PaCO* and pH values on admission than the patients who were managed without mechanical ventilation. Thus, the condition in the pa-
Patient Data of 50 Acutely iii Patients with COPD (Dallas Series)
TABLE iii
Men (no.)
Age W
56 f 8
17
58f9 PNS
5 NS
FEV1.o
(liters)
Most Recent Stable Outpatient Measuremenls PaOa PaCOa (mm W (mm Hg)
Patients Not RequiringArtificial Airways 0.28 54 f 8 49 f 9 13 Patients Requiring Artificial Airways 0.66 f 0.24 55f 12 51 f 10 NS NS NS 0.67 f
PH (mm W and 7.40 and 7.40
NOTE: NS = not significant.
898
0.40,P
December
1978
The American
Journal
of Medicine
Volume
85
Mechanical f 0.03 Mechanical f 0.03 NS
PaOz (mm Hg) Ventilation 41 f9 Ventilation 32 f 5
Admission Measurements PaCO, (mm W
PH W)
51 f9
7.39 f
57 f 7 <0.03
7.32 f 0.06
0.05
CONTROLLED OXYGEN THERAPY--OONE
ET AL
0
A
AM
7.1 IO
I 20
1 40
I 30
Pa02
I 50
I 60
1 70
jmmHgi
figure 1.
Pa02 and pH on admission (Dallas patients). Patients in whom somnolence developed while they were receiving controlled oxygen therapy were, in general, severely hypoxemic or had a combination of moderate hypoxemia and acidosis. This graph o&nonstrates the importance of hypoxemia and acidosis as risk factors for carbon dioxide narcosis during controlled oxygen therapy. The line separating high from low risk patients was found by discriminant analysis (pti - 7.66 - 0.00919 PO,). A = intubated patients. l = nonintubated patients.
tients who ultimately required more aggressive therapy deteriorated more markedly from their usual status. The initial response to oxygen therapy did not clearly separate the ultimate courses of the patients. Among patients begun on 24 per cent oxygen, the PaOs increased by 11 f 7 mm Hg for conservatively-managed patients and by 8 f 5 mm Hg for ventilator-managed patients (P not significant); the change in PaCOs was i-7 f 9 mm Hg and + 10 f 7 mm Hg (P not significant), respectively. Only the change in pH suggested the ultimate outcome, -0.01 f 0.06 versus -0.06 f 0.04 (P <0.05), respectively. Among patients treated with 28 per cent oxygen, the early blood gas values did not differ between the patients managed by oxygen alone and those managed by mechanical ventilation; the values were Pa02 + 21 f 88 mm Hg versus -Ill f 11 (mean f SD) mm Hg (P not significant), PaCO* -t-4 f 9 mm Hg versus +4 f 13 mm Hg (P not significant) and pH -0.07 f 0.09 versus -0.09 versus -0.01 f 0.04 (P not significant), respectively. At their worst level of hypercapnia blood gas values in the patients who were managed by graded oxygen alone were not as severely deranged as those in the patients who became stuporous. The respective blood gas values were PaOs 55 f 10 mm Hg versus 41 f 9 mm Hg (P
versus 7.27 f 0.04 (P
December 1978
The American Journal of Medicine
Volume 65
899
CONTROLLED OXYGEN THERAPY-BONE
,.I\
10
ET AL.
20
30
40
50 PC+
a
70
80
ImmHg)
I
Figure 2. Pa02 and pti on admission (Kansas patients). The line separating high and low risr patients for carbon dioxide narcosis drawn from data from the Dallas patients also separates high and low risk patients in the Kansas patients. A = in&bated patients. 0 = nonintubated patients.
respiratory failure died. All deaths occurred in patients who received intubation. COMMENTS In patients with serious chronic airways obstruction episodes of acute deterioration may develop manifested primarily by increasing dyspnea. Although arterial blood gas values and ventilatory mechanics are usually more deranged than values obtained in each patient during the steady state, if these are known [ 111, the absolute levels of Pa02, PaCO* or pH do not clearly separate stable from acutely ill patients [ 121. Thus, the diagnosis of acute respiratory failure is based on clinical findings. We have utilized clinical criteria for admission to this series, but we believe the patients to be representative of those diagnosed as having acute respiratory failure by most experienced clinicians. Since patients with COPD and acute respiratory failure may become stuporous or comatose when oxygen is administered in an uncontrolled manner [ 2,3], therapy is usually initiated with a fixed low concentration or low flow of oxygen [ 131. Most patients show improvement with such therapy, but some become progressively hypercapnic, acidotic and somnolent. We are unable to find published data which suggest which course is likely in an individual patient. Therefore, we have investigated the response of patients with COPD, to oxygen therapy, some of these patients were in their stable state and some were in acute respiratory failure. Rahn and Otis [ 141 have reported the ventilatory response of normal persons to the hypoxemia of alti-
900
December 1978
The American Journal of Medicine
tude; both early and delayed responses were evaluated. Their data are displayed in a PC02-PO2 diagram (Figure 3). In normal persons, hypoxemia does not stimulate excess ventilation nor cause hypocapnia above a PaOp of approximately 65 mm Hg. Below this PaOp, the PaC02 is lower indicating a stimulus to hyperventilation. Normal persons acclimatized to hypoxemia are more sensitive to this stimulus and have a lower PaCO* at any Pa02 than unacclimatized persons. Thus, the normal response of PaCO* to decreases in Pa02 lies between the two extremes indicated by their data [ 141. The response to mild hypoxemia in our ambulatory outpatients without hypercapnia was in the normal range indicated by Rahn and Otis (Figure 3). When their Pa02 was increased by the administration of oxygen, their PaC02 response remained within the normal range with a slope that paralleled that of acclimatized normal persons. Our ambulatory outpatients with hypercapnia had a normal response to increasing Pa02 because the slope of increasing PaCO* was similar to that of normal persons. Thus, ambulatory outpatients with or without hypercapnia responded normally to increasing Pa02. The patients with acute respiratory failure in Dallas who did not require intubation had a more serious derangement of ventilatory response than the stable hypercapnic patients, for although their PaC02 on admission was no higher, their Pa02 was lower. The ventilatory defect was yet more marked in the acutely ill patients who required mechanical ventilation, for their admission Pa02 was significantly less than that in all other groups. Nevertheless, when the Pa02 in these groups of patients was increased by oxygen therapy,
Volume 85
CONTROLLED
_ .-I. -...-.. -.-. 120
OXYGEN
,’
0
.
GROUP
I
-Stable
cold,
GROUP
II
-Stable
10 patients
retention1
cold,
10
!COzretentIonI A
1
GROUP
III-Not
mtubated, [no co2
.
GROUP
IV-Eventually [CO2
0
L-r--20
ET AL
..- .- -.-.-_....-____.___. __._.__._._~..__ _.
[no CO2
100
THERAPY--RONE
1
/
I
37 patents
narcosIsI mtubated.
13 patients
narcorlrl
I-1-I 60
40 Pa02
paWntr
80
ImmHy)
3. Mean values for PaO,-PaCO, on admission plotted with Pa4 and PaC02 at point of greatest hypercapnia after oxygen administration in four se&rate groups of patients with COPD. The hashed area is a plot for normal volunteer subjects and was taken from reference [ 141. Horizontal and vertical bars indicate standard error of the mean for the Pa4 and PaCO,, respectively.
xgure
the PaCO* increased at a slope roughly parallel to the slope of the increase in PaCO* in normal persons over the same range of Pa02, as judged by the Rahn-Otis diagram. Thus, the acutely ill patients responded with an appropriate increase in PaC02 when given oxygen, although this increase occurred from an abnormally high base line PaCOp. We interpret these findings to indicate that the carotid body response to hypoxia is normal in patients with COPD with or without acute respiratory failure. Despite the fact that the response of the carotid bodies in patients with COPD and acute respiratory failure is apparently normal, the sudden increase in PaC02 with oxygen therapy may lead to sufficient cerebral acidosis to cause coma [15-l 71. We have found that the PaOa and pH on admission are reasonable predictors of this event; the lower these values are at the outset of therapy, the more likely oxygen therapy is to result in coma. The pH-PaOa diagram indicated in Figure 1 indicates this relationship in a convenient graph that may prove useful clinically. In the final analysis, the decision that an individual
patient with acute respiratory failure from COPD is sufficiently stuporous to require an artificial airway and mechanical ventilation is a subjective value judgment on the part of the attending physician. We thought it important, therefore, to validate our findings in a different medical environment in which the subtle factors leading to this judgment might be different. The primary care physicians who managed the patients in the Kansas series utilized clinical criteria for intubation of their patients without access to the specific concepts developed in the Dallas series. Nevertheless, the pH-PaOs diagram was similarly predictive of the ultimate course of their patients following oxygen therapy. The pH-Pa02 diagram should not be used to perform intubation, prophylactically, in all patients who are in the high risk group for narcosis following oxygen therapy. The hazards of aggressive intubation in acute respiratory failure from COPD have been previously stressed [2,3]. The pH and PaOp on admission, however, will identify those patients at greatest risk and should assist the physician in planning a therapeutic regimen.
December 1978
The American Journal of Medicine
Vofume 85
991
CONTROLLED OXYGEN THERAPY-BONE ET AL.
REFERENCES 1.
2. 3. 4.
5.
6.
7. 8. 9.
902
Barach AL: Physiological methods in diagnosis and treatment of asthma and emphysema. Ann Intern Med 12: 454, 1938. Campbell EJM, Gabbis T: Mask and tent for providing controlled oxygen concentration. Lancet 1: 468, 1966. Campbell EJM: Management of respiratory failure. Br Med J 2: 1328, 1964. Smith, JP, Stone W, Muschenheim C: Acute respiratory failure in chronic lung disease. Am Rev Respir Dis 97: 791, 1968. Bedon GA, Block J, Ball WC: The 28% Venturi mask in obstructive airway disease. Arch Intern Med 125: 106, 1970. Schiff MM, Massaro D: Effect of oxygen administration on arterial blood gas values in patients with respiratory failure. N Engl J Med 277: 950, 1967. King TKC. Ali N, Briscoe WT: Treatment of hypoxia with 24% oxygen. Am Rev Respir Dis 108: 19, 1973. Eldridge F, Gherman C: Studies of oxygen administration in respiratory failure. Ann Intern Med 68: 569, 1968. Moser KM, Luschsinger PC, Adamson JS, et al.: Respiratory stimulation with intravenous doxapram in respiratory failure. A double-blind cooperative study. New Engl J Med 288: 427,
December 1978
The American Journal of Medicine
10.
11.
12. 13.
14. 15. 16.
17.
Volume 85
1973. Morris JF, Koshi A, Johnson LC: Spirometric standards for healthy nonsmoking adults. Am Rev Respir Dis 103: 57, 1971. Sukumalchandra Y, Park SS, Williams Ml-t. Jr: The effect of intermittent positive pressure breathing (IPPB) in acute ventilatory failure. Am Rev Respir Dis 92: 885, 1965. Lal S: Blood gases in respiratory failure. State on admission to hospital and management. Lancet 1:339, 1965. Campbell EJM: The J. Burns Amberson Lecture. The management of acute respiratory failure in chronic bronchitis and emphysema. Am Rev Respir Dis 96: 626, 1967. Rahn H, Otis AB: Man’s response during and after acclimatization to high altitude. Am J Physiol 157: 445, 1949. Posner JB, Swanson AG. Plum F: Acid base balance in cerebral spinal fluid. Arch Neurol 12: 479, 1965. Penman RWB: Blood lactate levels and some acid-base changes in respiratory failure and their significance in oxygen induced respiratory depression. Clin Sci 23: 5, 1962. Rudolf M, Banks RA, Semple SJG: Hypercapnia during oxygen therapy in acute exacerbations of chronic respiratory failure. Lancet 2: 483, 1977.