Hypermetabolism and Efficiency of CO2 Removal in Acute Respiratory Failure

Hypermetabolism and Efficiency of C02 Removal in Acute Respiratory Failure* Ritva Kiiski, M.D.; and ]ukka Takala, M.D., Ph.D., F.C.C.P. Objective: To assess the effect of hypermetabolism, dead-space ventilation, and parenteral nutrition on the minute ventilation requirement in mechanically ventilated patients. Design: A retrospective analysis of data collected in study protocols unrelated to the present study. Setting: A medical-surgical intensive care unit in a tertiary care center. Patients: One hundred eleven mechanically ventilated patients were studied during volume-controlled ventilation. Measurements: Gas exchange measurement by indirect calorimetry and arterial blood gas analysis. Methods: Minute ventilation (VE), carbon dioxide production (VcO!), and respiratory exchange ratio (REB) were measured with indirect calorimetry. Arterial C02 tension was sampled at the end of the measurement, and alveolar ventilation (VA), deadspace to tidal volume ratio (Vo/VT) and predicted resting VC02 were calculated. The VE demand at a standard PaCOz was calculated and the contribution of the observed hypermetabolism and increased VDI VT was identified. In a subgroup of patients, the effect of initiating parenteral nutrition on the VE demand was assessed. There were four study groups: multiple trauma, sepsis, ARDS, and postoperative

open-heart surgery patients. Main results: A combination of hypermetabolism and increased dead-space was observed in 67 of the III patients. Increased VC02 accounted for 69 percent of the excess VE demand in trauma, 67 percent in sepsis, 58 percent in postoperative patients, and 56 percent in ARDS. Parenteral nutrition with a caloric intake matching measured resting energy expenditure (BEE) did not increase VCO2 or the demand for VE. ConclUBiofl8: Increased VC02 is the main cause of increased VE demand in the majority of mechanically ventilated ICU patients. Parenteral nutrition at energy intakes close to actual BEE does not increase the ventilatory demand. (Chest 1994; 105:1198-1208)

ill patients frequently need ventilatory C ritically assistance because the respiratory muscle pump

and may impose a large increase in VE demand. 7 The demand for VE is inversely related to PaC02, which can be modified by the clinician during controlled mechanical ventilation. This study was designed to test the hypothesis that hypermetabolism is a major contributing factor to the ventilatory demand in ARF.

is unable to meet the increased demand for minute ventilation (VE). The VE demand is determined by three factors: metabolic C02 production (VC02), the arterial C02 tension (PaC02), and the efficiency of C02 removal. l Increased dead-space-to-tidal volume ratio (VD/VT) is traditionally regarded as an important determinant of VE,2 and it is common in several disorders, including the adult respiratory distress syndrome (ARDS),3 sepsis, and obstructive lung disease. 4,s Most of the conditions leading to acute respiratory failure (ARF) are characterized by a higher energy expenditure than estimated by predictive formulas (hypermetabolism), and the effect of an increased VC02 to the VE demand may be overlooked. VC02 is also affected by nutritional support. 6 Hypercaloric glucose infusions increase VC02

·From the Critical Care Research Pr~ram, Department of Intensive Care, Kuopio University Hospital, Kuopio, Finland Manuscript received January 12, 1993; revision accepted August

20. Reprint requests: Dr. Takala, Kuopio University Hospital, DepJJrtment 0/ Intensive Caf'e, PO Box 1777, SF-70211 Kuopio, Finland

1198

ARDS=adult respiratory distress syndrome; ARF= acute respiratory failure; EE=energy expenditure; F1ol=oxygen fraction in inspired air; OHS=openheart surgery; PEEP=positive end-expiratory pressure; REE=resting energy expenditure; RER=respiratory exchange ratio;. TPN=parenteral nutrition; VA=alveolar ventilation; Vco=carbon dioxide production; VD=del!d-space; VD /VT=dead-sp!lc~ to tidal volume ratio; VE=minute ventilation; V/Q:=distribution of ventilation and perfusion; VT=tidal volume

METHODS Pattents

We studied 25 patients with multiple trauma, 21 patients with severe sepsis, 15 patients with ARDS, and 50 postoperative open-heart surgery patients (OHS) with previously normal lungs. Acute respiratory failure was defined as ventilator dependency due to multiple injuries (trauma), severe infection (sepsis), abnormal arterial oxygenation (ARDS), or recovery from fentanyl anesthesia and cardiopulmonary bypass (cardiac surgery). Indirect calorimetry and arterial blood gases were measured according to study protocols unrelated to the present study. The patients were selected according to the following criteria: Trauma

Multiple injuries included major thoracoabdominal and/or intracranial injuries requiring ventilatory support for at least 24 h. The mean injury severity score8 was 22 ± 4 (mean ± SE) and the Pa02/FI02 ratio was 266 ± 25. HypermetaboIsm and Efficiency of CO2 Removal in ARF (Kiiski, TakBIa)

Sepsis The criteria for sepsis were as follows: (1) bacteremia or a verified focus of infection; (2) at least two of the following clinical signs of infection: fever over 38.5°C, blood leukocyte count over 13X109 /L, c-reactive protein over 40 mg/L; and (3) one or more of the following: respiratory failure or tachypnea; hypoxemia with normal or low PaC02, agitation or confusion, decreased level of consciousness, metabolic acidosis, and urine output less than 50 ml/h. The severity of the sepsis was determined according to Elebute and Stoner. 9 The mean sepsis score on the day of the measurement was 12.2 ± 0.9, and the Pa02/Flo2 ratio was 261 ± 20. ARDS

All ARDS patients fulfilled the following clinical diagnostic criteria for ARDS: a triggering event known to be associated with the development of ARDS, bilateral diffuse lung infiltrations, Pa02 less than 7.5 kPa (56 mm Hg) on an oxygen fraction in inspired air (FI02) of at least 0.40, and positive end-expiratory pressure (PEEP) higher than 5 cm H20, and no cardiogenic cause as defined by pulmonary artery catheterization. The severity of the lung injury was assessed according to Murray and coworkers. lo Since static lung compliance was not available for all the patients, we used PEEP level, Pa02/Flo2, and the chest radiographic findings, each scored 0 to 4. The lung injury score in ARDS was 2.5 ± 0.1 (range, 1.5 to 3.7), and the Pa02/Flo2 ratio was 164±41 on the day of the study. Eight ARDS patients had a lung injury score of at least 2.5 (severe lung injury). OHS Cardiac surgery patients had elective coronary artery bypass and/or valvular operation with uncomplicated recovery, and a history of normal lung function. They were chosen as a control group because they are a homogeneous group of ICU patients who are all mechanically ventilated and who are normally weaned and extubated without difficullies. The Pa02/FI02 ratio was 295±14.

Table I-Patient Characteriltic. and Ventilator Settings During Controlled Mechanical Ventilation (Mean±SE)*

Sex, M/F Age, yr BSA, m2 Energy, kcal/d VT RR FI02 PEEP

OHS

Trauma

Sepsis

ARDS

29/21 57±1 1.77 ± 0.00** 246±20

17/8 37±4** 1.92±0.00 426±14

16/5 54±4 1.87±0.00 406±33

11/4 49±3 1.91 ±0.05 514±35

10.8±0.4 11.5±0.3 42±1 5±O.1

10.7±0.4 12.2±O.3 41±1 5±O.6

10.6±0.5 12.5±O.3 41±1 6±O.5

11.7±1.0 12.1 ±O.6 45±2* 8±0.7**

*BSA=body surface area; VT=ml/kg; RR=L/min; FI02=percent; PEEP=cm H20. Statistics: age=trauma vs others, p
nitrogen balance. 11 The TPN consisted of 1.5 g/kg/d of amino acids and the nonprotein energy content was equal to the measured energy expenditure (EE), with 40 to 60 percent given as lipids. Glucose and protein were infused over 24 h and the fat emulsion for the first 12 h of the study day. Gas exchange was measured 3 h after initiating the TPN, while the rate of infused calories was approximately 1.5 times EE. The open-heart surgery patients (Table 1) were evaluated 6 to 8 h after the operation. They were slightly hyperventilated (PaC02, 4.36 ± 0.08 kPa) according to departmental policy to avoid an increase in the pulmonary .vascular resistance due to C02 retention. To compare the patient groups, we applied a mathemati~al model based on the Bohr equation for the standardization of VE (Fig 1). According to the formu~a, ~aC02=0.863XVC02/~A where VA can also be expressed as VA=VEX(l-VD/VT). The VE demand corresponding to a standard PaC02 was calculated for each individual. We chose the mean PaC02 of aU the groups,

Exclusion Criteria

Patients younger than 17 years receiving parenteral nutrition (TPN) or enteral nutrition or requiring resuscitative measures due to hemodynamic instability at the time of the study were excluded. The first gas exchange measurement was performed within 72 h of admission to the ICU, or, in ARDS, within 72 h of fulfilling the diagnostic criteria. To assure a steady state in gas exchange, the patients were hemodynamically stable, well sedated, and diagnostic and nursing interventions were avoided during the measurement, and the stability of the gas exchange measurement was inspected from the minute-to-minute variation of the results. All patients were receiving volume-controlled mechanical ventilation set by the attending clinician according to individual needs (Table 1), and the settings were not changed within 30 min of the measurement. Oxycodone and diazepam were used for sedation, and, in ARDS, pancuronium or vecuronium were used for muscle relaxation, if deemed necessary. None of the patients was triggering the ventilator during the study, as observed by the triggering indicator and airway pressure gauges. To avoid a nutrient-induced increase in C02 production, the patients received an infusion of 5 percent dextrose in water during the study day. The second evaluation of gas exchange was done in a subgroup of patients (6 trauma patients, 12 patients with sepsis, and 10 ARDS patients) within 48 h of the baseline measurements. They received a standard regimen of parenteral nutrition (TPN group) according to another study protocol assessing the effect of nutritional support on the energy and

VD/V T

c Increased

B

..•...•.•• ,

A

..........

normal

~

normal

Increased

FIGURE 1. The e.ffect of increased dead-space (VD/VT). and metabolic rate (VC02) on the ventilatory de~and (VE)..VE is obtained from the following equation:. VE=O.863XVC02/ [(I-VD/VT)XPaC02].The graphs represent .VE at two levels !If VD/VT. A iUustrates VE when VD/VT and VC02 are normal. VE demand increases by (B-A) ~hen VD/VT is high~r than normal. The additional increase in VE is (C-B) when VC02 is also increased. (C-t\) is the VE excess due to a combination of increased VO/VT and VC02. CHEST /105/4/ APRIL, 1994

1198

Table 2-Alveolar Ventilation and Determinant, of Minute Ventilation (Mean±SEJ*

VA, VE,

L/min L/min VD/VT PaC02, kPa REE pred, kcal REE meas, kcal REE, m/p, % RER

OHS

Trauma

Sepsis

ARDS

5.S±0.2 8.4±0.S*· 0.36±0.01 4.36*±0.08 1,455±33 1,670±45** 115±2 O.83±O.Ol

6.0±0.S 10.0±0.4 0.39 ± 0.02 4.68±0.lS 1,730±46 2,071 ±86 119±S O.83±O.Ol

5.7±0.S 9.8±0.4 0.42±0.02 4.89±0.13 1,574±51 1,960± 106 124±5 O.83±O.O2

5.8±0.4 10.9±0.4 O.46±0.OO*· 4.90±0.14 1,647±63 2,070±96 128±6** O.82±0.02

*REE pred=resting energy expenditure by Harris-Benedict formula; REE meas=measured energy expenditure; REE m/p=ratio of measured REE to predicted REE in percent. Statistics: PaC02=p<0.05, OHS vs others; VE, REE meas=p
4.85 kPa (36 mm Hg) as the standard. The calculation assumes a constant VO/VT when VE is changed. Major changes in VO/VT are unlikely when VE is increased or decreased by adjusting the respiratory rate with a constant VT and inspiratory Bow. We have previously shown that VE can also be changed by altering VT by ± 25 percent without a signi6cant change in VO/VT. 12 In this study, the absolute required change in VE in the calculation was 1.2±0.1 L/min (12± 1 percent). The normal VE at a PaC02 of 4.85 kPa was calculated assuming a normal VO/VT and a normal metabolic rate (VC02). For the nonnal VO/VT, we suhmtuted 0.36 which is the mean preoperative value of 39 previously studied cardiac surgery patients with normal lungs. 13 The resting VCO2 was obtained from the modified deWeir formula: REE=5.68XV02+ 1.59XVC02. (I) Resting energy expenditure (REE) was obtained from the Harris-Benedict formula. I" The observed respiratory exchange ratio (RER) was applied to express the V02 as a function of VC02 and RER: REE=5.68XVC02/RER+1.59XVC02, (2) and this equation was solved for V02. The equation for calculating the normal VE can be written as follows: VE=0.863XVcQ2CtJlc/[PaC02X(1-Vo/VT)1 (3) where Vco2Ct1lc is calculated normal VC02 (equation 3), PaC02 is

Statistics

VO/VT

~

••P.I••• ·~

10

c:

E

'" ..J

~

&AI

.>

••'

9 8 7

I

0.50

...··l.:rd. .

11

....

•••••

...

.'

......

.'

0.40

••••••

..f- ....···0.30 ..

'

••··trauma •••• •••• •• •••••• ••...•• ..•······0.20

+....

.... ..... .···oha •.••• ••••• ... ...... ....... . . '

. .

'

.'

175

.'

200

225

250

275

300

veo 2 (mL/mln) FIGURE 2. rhe mean VE requirement in the subgroups with both increased VC02 and VD/VT. The .VO/VT isopl~ths have been calculated from the equation VE=0.863XV c02/[(I- Vo/ VT)XPaC02] using a constant PaC02' OHS=open-heart surgery. Two asterisks=p
1200

36 mm Hg, and VO/VT is 0.36. The application of the resting VCO2 and the standard normal VD/VT to the Bohr equation is presented in Figure 1. Hypermetabolism was defined as EE higher than the REE predicted from the Harris-Benedict formula. Alveolar ventilation (VA) was calculated as VA (L/min, BTPS)=0.863XVC02 (ml/min, STPD)/PaC02 (mm Hg). (4) The ratio of physiologic dead-space (Vo) to tidal volume was obtained from VA as VO/VT= 1-VA/VE. 15 To calculate the VE demand with hypermetabolism and increased dead-space, observed VC02 and VO/VT were applied to equation 3, respectively. The VE excess is expressed as the difference from normal VE. The VC02, VE, REE, and RER were measured by an indirect calorimetry device (Deltatrac, Datex/Instrumentarium, Helsinki, Finland) for 30 min. The device has been described in detail 16 and previously validated in this laboratory.17 The error of VC02 and oxygen consumption (V02) measurements is less than 5 percent in the study conditions. The Vo and VE measurements have been previously validated with a pneumotachometer, and the mean differences are 8.2±4.7 percent and -0.4±6.5 percent, respectively, during controlled mechanical ventilation. 15 A sample of arterial blood was taken at the end of the measurement from an indwelling catheter to determine the C02 tension (IL 1302, Instrumentation Laboratories, Lexington, Mass). The differences between the four groups were assessed by one-way analysis of variance 18 followed by the Student-NewmanKeuls test. The results before and during TPN were compared with paired t test. All results are given as mean ± SE. RESULTS

The observed VE was higher in the patients with trauma, sepsis, and ARDS than in the OHS group (Table 2, p
.

2.00

., ...0

E

c ........

•...> ••

'U

•

1.75 1.50

•

1.25

6

.•

5

-....... • c E

.

1.00

....

3

2

.0 0

0.75

N

0.50

1

0.25

0

0

U

.>

0.00 0.0

•

VE EXCESS

0.2

0."

0.6

0.1

1.0

VO/VT FIGURE 3. Hype~etabolism and increased dead-space in III patients. Normal VC02 at 1.00 (horizontal line) and normal deadspace at 0.36 (vertical line). Closed triangles=trauma; open triangles=sepsis; closed squares=ARDS; open circles=openheart surgery.

in patients with trauma, 10.1 ±0.4 L/min in patients with sepsis, 1O.9±0.4 L/min in patients with ARDS vs. 7.5±0.2 L/min in OHS, p
TRAUMA SEPSIS ARDS 4. The contribution of hypermetabolism to the VE excess (open columns) in 6 trauma patients, 10 patients with sepsis, and 12 ARDS patients during 5 percent dextrose infusion (closed columns) and parenteral nutrition (hatched columns). Asterisk= p
the change in VC02 during TPN, but the changes were not statistically significant (Fig 4). When the individual variation in VO/VT was eliminated, TPN increased the VE demand by 0.52 L (6 percent) in trauma, 0.58 L (8 percent) in sepsis, and 0.64 L (9 percent) in ARDS. DISCUSSION The main observations in this study were that an increased VCO2 was the main cause of an increased VE demand in ARF, and that an increased VO/VT without hypermetabolism was unusual. In the subgroups with TPN with energy intake close to the measured REE, VE demand did not increase significantly. Accurate measurements of gas exchange and Vo/ VT are mandatory for the data analysis in this study. We have previously validated the gas exchange measurements and shown that VE and VA can be accurately measured even in non-steady-state conditions. ls Physical activity was minimal and external stimuli were specifically avoided during the SO-min measurement period to avoid artifacts. The observed VC02 and VO/VT were first used to determine the VE required to obtain a PaC02 of 4.85 kPa for

Table 3-The Contribution of Hypermetabolism and lncrea.ed VD/VT to the VI: Demand (Mean±SD)*

VE4.85 OHS (22) T (14) S(18) ARDS (13)

8.4±A

10.S±0.6* 1O.7±0.3* 1l.3±O.7*

dYE

dVEmet

2.1 ±0.3

1.3±0.2 (58%) 1.9±0.3 (69%) 1.9±0.2 (67%)

3.0±OA

2.9±O.3 4.3±0.6*

2A±0.6* (56%)

VO/VT

REE m/p

0.43 ±0.01 0.4S ±0.02 OA7±O.03

OA9±0.02**

117±3 122±3 125±S 129±6

*VE4.85=VE at observed VO/VT and VC02 and PaC02 4.85 kPa (36 mm Hg), L/min; dYE-increase in VE due to hypermetabollsm and increased VO/VT, L/min; dVEmet=increase in VE due to hypermetabolism only, L/min; Vo/VT=ratio of dead-space to tidal volume; REE m/p=ratio of measured REE to predicted REE, percent; T=multiple trauma; S=sepsis. Statistics: VE4.85-p
1201

each individual. The method has potential limitations, since the mathematical normalization of PaC~ assumes an unchangEd VO/VT when VE is changed. According to our previous studies, VE can be adjusted by changing VT without an effect on VO/VT which justifies the standardization for a constant PaC02.12 Since our calculation assumes a constant VT, a change in VO/VT is even less likely. An increase in VO/VT would lead to underestimation of the VE requirement and overestimate the importance of hypermetabolism in the VE demand. Similarly, a decrease in VO/VT would underestimate the contribution of hypermetabolism in the VE requirement. Since the standard PaC02 was the mean of the study groupi, the required changes in VE were small. Accordingly, it is unlikely that the conclusions would have been distorted by the assumption of unchanged VO/VT in this material. Although a combination of increased VC02 and VO/VT was observed in the majority of the patients, VC02 was the main contributing factor in the VE excess. Hypermetabolism and the consequent increase in VC02 can be expected in multiple trauma, and the observed rate of metabolism in the trauma patients, 19±3 percent above predicted REE, was comparable to previous findings. II The severity of the illness correlates with the hypermetabolism in the patients with sepsis. IS The metabolic rate in ARDS has not been well documented previously. Our results indicate that, in general, the metabolic rate increased by approximately 30 percent in ARDS, and that this increase in VC02 was as important as the increased VO/VT in determining the demand for VEe Ravenscraft and associates l9 reported lower rates of VC02 and a smaller contribution of VC02 to the VE demand. When compared with our study, the patient groups were not identical, and the methods of VE and VC02 measurement were different, and RER was not measured. Estimating RER instead of measuring it may cause a marked error in the calculation of the resting VC02, especially in sepsis and ARDS. Increased VO/VT was observed in 25 (50 percent) of the OHS patients, and more than one half of the other groups. The OHS patients had no history of lung disease, but the efficiency of the C02 removal may have been impaired by the cardiopulmonary bypass and atelectasis. We observed an increase in VO/VT in patients with sepsis in accordance with previous findings. 4 In ARDS, substantial mismatch in the distribution of ventilation and perfusion (V /Q) increases the physiologic deadspace. 20 Mechanical ventilation may contribute to the V /0 inequality by overdistending well-ventilated and poorly perfused alveoli. 21 Nutrition-induced increase in VC02 may result in hypercapnia. This observation was made during 1202

hypercaloric high-carbohydrate feeding in spontaneously breathing patients. 22 In patients with compromised lung function, an increased metabolic rate may indeed increase the work of breathing and further increase VC02.23 Talpers and associates7 have recently shown that the energy excess is a more important determinant of VC02 than the glucose:fat ratio in the TP~ regimen. In two groups of stable leu patients, VC02 was similar during isocaloric TPN with various carbohydrate contents. In contrast, VC02 increased significantly from fasting baseline when TPN was infused at rates 1.5 to 2.0 times REE. In our study, the energy intake was matched with the measured REE, and 40 to 60 percent of the calories were given as fat. The patients in our study were in a more acute stage of illness than the patients studied by Talpers and coworkers, but our findings are similar: the changes in VC02 and VE requirement after the initiation of TPN were not significant. We· conclude that hypermetabolism is the major cause for increased VE demand in ARF, and that increased dead-space ventilation without an accompanying increase in VC02 is uncommon. Parenteral nutrition does not increase ventilatory demand in mechanically ventilated patients if energy intake is close to the actual energy expenditure. ACKNOWLEDGMENT: The authors thank Prof. John M. Kinneyfor the idea of hypermetabolism as the major determinant of ventilatory demand, and Dr. Robert F. Lodato for helpful comments on the manuscript.

REFERENCES 1 Mille-Emili J. Recent advances in assessment of control of breathing. Lung 1982; 160:1-17 2 Giovannini I, Chiarla C, Boldrini G, Castagneto M. Impact of fat and glucose administration on metabolic and respiratory interactions in sepsis. JPEN 1989; 13:141-46 3 Lamy M, Fallat RJ, Koeniger E, Harm-Peter D, Ratliff JL, Eberhart RC, et al. Pathologic features and mechanisms of hypoxemia in adult respiratory distress syndrome. Am Rev Respir Dis 1976; 114:267-84 4 Siegel J, Giovannini I, Coleman B. Ventilation perfusion maldistribution secondary to hyperdynamic cardiovascular state as the major cause of increased pulmonary shunting in sepsis. J Trauma 1979; 29:432-60 5 Jones NL. Pulmonary gas exchange during exercise in patients with chronic airway obstruction. Clin Sci 1966; 31:3943 6 Askanazi J, Rosenbaum SH, Hyman AI, Silverberg PA, Milic-Emili J, Kinney JM. Respiratory changes induced in the large glucose loads of total parenteral nutrition. JAMA 1980; 243:1444-47 7 Talpers SS, Romberger DJ, Gunce SB, Pingleton SK. Nutritionally associated increased carbon dioxide production: excess total calories vs high proportion of carbohydrate calories. Chest 1992; 102:551-55 8 Baker SP, O'Neill B, Haddon W, Long WB. The injury severity score: a method for describing patients with multiple injury and evaluating emergency care. J Trauma 1974; Hypermetabolsm and Efficiency of CO:l Removal in ARF (Kliski, Takala)

14:187-96

9 Elebute EA, Stoner HB. The grading of sepsis. Br J Surg

1983; 70:29-31 10 Murray JF, Matthay MA, Luce JM, Flick MR. An expanded 11 12 13 14

definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138:720-23 Pitkanen 0, Takala J, Poyhonen M, Kari A. Nitrogen and energy balance in septic and injured patients: response to parenteral nutrition. Clin Nutr 1991; 10:258-65 Kiiski R, Takala J, Kari A, Mille-Emili J. The effect of tidal volume on gas exchange and oxygen transport in ARDS. Am Rev Respir Dis 1992; 146:1131-35 Tulia H, Takala J, Alhava E, Huttunen H, Kari A, Manninen H. Respiratory changes after open-heart surgery. Intensive Care Moo 1991; 17:365-69 Harris JA, Benedict FG. Standard basal metabolism constants for physiologists and clinicians: a biometric study of basal metabolism in man. Philadelphia: JB Lippincott, 1919; 223-50

15 Kiiski R, Takala J, Eissa NT. Measurement of changes in

dead space by indirect calorimetry: a clinical and laboratory validation. Crit Care Med 1991; 19:1303-09 16 MeriliHnen PT. Metabolic monitor. Int J CUn Monit Comp

1987; 4:167-77 17 Takala J, Keininen 0, Viiisanen P, Kari A. Measurement of 18 19

20

21

22

23

gas exchange in intensive care: laboratory and clinical validation of a new device. Crit Care Moo 1989; 17:1041-47 Norusis MJ. SPSS/PC+ for the IBM PC/XT/ AT. Chicago: SPSS Inc, 1986 Ravenscraft SA, McArthur CD, Path MJ, Iber C. Components of excess ventilation in patients initiated on mechanical ventilation. Crit Care Med 1991; 19:916-25 Dantzker DR, Brook CJ, Dehart P, Lynch JP, Weg JG. Ventilation-perfusion distributions in the adult respiratory distress syndrome. Am Rev Respir Dis 1979; 120:1009-52 Ralph DD, Robertson HT, Weaverly LJ, Hlastala MP, Carrico CJ, Hudson LD. Distribution of ventilation during positive end-expiratory pressure in the adult respiratory distress syndrome. Am Rev Respir Dis 1985; 131:54-60 Askanazi J, Elwyn DH, Silverberg PA, Rosenbaum SH, Kinney JM. Respiratory distress secondary to a high carbohydrate load: a case report. Surgery 1980; 87:596-68 Al-Saady NM, Blackmore CM, Bennett ED. High fat, low carbohydrate, enteral feeding lowers PaC02 and reduces the period of ventilation in artificially ventilated patients. Intensive Care Moo 1989; 15:290-95

CHEST /105/ 4/ APRI~ 1994

1203