Chronic fatigue syndrome (CFS) is a debilitating illness of diverse symptomatology where the diagnosis, in the absence of a definitive laboratory test, relies on a grouping of clinical criteria (9). The fatigue, for which both central psychological (20) and peripheral pathophysiological (7) mechanisms have been proposed, is often present at rest and exacerbated by the simplest of physical tasks. The restricted lifestyle of CFS patients (28) has led to the suggestion that a reduction in exercise capacity contributes to their fatigue (8,17,23) and prolongs their illness. It is for this reason that exercise-training programs have been recommended in the treatment of CFS (10,17) despite the common complaint of profound postexercise exhaustion (19) and no published evidence of physiological improvement (10). The notion of reduced exercise capacity is supported by most previous exercise studies, which describe reductions in maximal oxygen uptake (V̇O2max) of up to 38%(6,8,11,23,26), reductions in peak heart rate (11,26,27) and peak power output (8), earlier exhaustion (8,11,23,26,30), and accelerated glycolysis with increased lactate production (21,30). The two critical measures of exercise capacity are V̇O2max and the lactate threshold (LT) (5,24), yet the latter has not been measured directly in CFS patients. With regard to the former, none of the reported studies has adhered both to the strict maximal testing protocols and defined endpoints regarded as essential for the valid assessment of V̇O2max (14). Furthermore, despite women having a V̇O2max (mL·kg−1·min−1) ∼20% lower than men (22), all but two studies (6,26) have pooled gender data, this error being compounded in those studies in which the gender distribution was unequal between patient and control groups (8,11). Therefore, the purpose of the present study was to employ traditional “gold standard” maximal exercise testing methodology (14) and take account of gender in the reevaluation of exercise capacity in CFS.
Thirty-four patients (16 male and 18 female) participated in the study and were seen by the same physician (R. Burnet). To make the diagnosis of CFS, each patient was required to meet the major and at least four of the minor criteria proposed in the Centers for Disease Control (CDC) working case definition (9). The frequencies of major and minor criteria met by the patients in each gender group are illustrated in Table 1. Of the female subjects, two worked full-time and two part-time, one was a student, and 13 (72%) were unemployed, whereas within the male group, four worked full-time and one part-time, one was a student, and 10 (63%) were unemployed. The mean time interval in months from self-reported onset of symptoms to participation in the present study was almost twice as long in the female subjects (male subjects, 54.9 ± 41.3 months; female subjects, 97.7 ± 50.6 months; mean ± SD, P = 0.019). The severity of illness in the CFS patients was categorized using the Karnofsky Performance Status (KPS) scale (18). The KPS scale is commonly used for this purpose in CFS patients (29), where their ability to perform normal activity, ability to do active work, and their need for assistance are rated by using a numerical scale from 0 to 100 (no impairment). Scores for patients in the present study (male subjects, 73.6 ± 8.4; female subjects, 63.3 ± 9.0;P = 0.004) indicated that patients “Can care for themselves but are unable to carry on normal activity or to do active work”(18), this illness severity being consistent with that reported in other studies (29). Thirty-four sedentary control subjects (16 male and 18 female), matched with the CFS patients for gender, age, mass and height and with no history of medical illness in the 5 yr before the study, were recruited from the general community. Control subjects were considered sedentary if their occupation did not require physical labor and they did not perform structured physical activity, i.e., exercise, more than once a week (25). Of the female subjects, 10 worked full-time and one part-time, six were students, and one (6%) was unemployed, whereas within the male group, 12 worked full-time, four were students, and none (0%) was unemployed. The study was approved by the Human Ethics committees of the Royal Adelaide Hospital and the University of Adelaide and was conducted in accordance with the principles of the Declaration of Helsinki (Pan American Health Organization). Written consent was obtained from all subjects after provision of a written and oral explanation of the purpose of the study, the procedures to be used, and the risks entailed.
Maximal Cycle Ergometer Exercise Test
If any subject was taking medication likely to interfere with cardiorespiratory or metabolic function during exercise, this was withdrawn at least 2 wk before the maximal exercise test. All subjects attended the laboratory having fasted overnight and abstained from alcohol, caffeine, and undue physical activity in the previous 24 h. Chest electrodes were applied for monitoring heart rate (Polar Vantage NV heart rate monitor; Kempele, Finland) and the electrocardiogram (Nihon Koden Lifescope 6 Portable Patient Monitor; Nihon Koden, Tokyo, Japan), which was displayed as a precautionary measure, during and immediately after exercise testing. A Teflon catheter (Jelco 18G, 1.2 mm × 44 mm length; Jelco, Tokyo, Japan) was inserted under local anesthesia (lignocaine hydrochloride, 2%) into the deep muscle branch of an antecubital vein with its tip directed away from the heart. A 30-cm polyethylene extension (Braun Minimum Volume Extension Tubing, dead space 0.3 mL; Braun, Melsungen, Germany), filled with normal saline and sealed with a three-way stopcock (Discofix, Melsungen, Germany), was attached to the catheter and taped to the skin. A net bandage (Netelast, Seton Healthcare Group, Oldham, United Kingdom) was placed over the arm to secure the catheter and connections to the upper arm and a simple elbow splint applied (Lemmco Elbow Immobilizers, Velcropedic, Ace Surgical Supply Company, Brockton, MA). This afforded a partial restriction to forearm flexion and avoided displacement of the catheter during cycle exercise. The subject was then seated on a precalibrated Monark cycle ergometer (model 818E, Varberg, Sweden) and was fitted with a low-resistance respiratory valve (Hans Rudolph R2700, Kansas City, MO) to enable gas exchange measurements. The valve was held in place by a head support (Hans Rudolph Head-Support for Rudolph Valves, model no. 2766).
Once all monitoring equipment was in place and a 10-min rest period had elapsed, exercise began with 2 min of unloaded cycling (0 W) at 50 rev·min−1, and thereafter the power output was incremented by 25 W every 2 min until, despite strong vocal encouragement, the subject could not maintain the target power output.
Blood (2 mL of whole blood) was collected into glass syringes rinsed with sodium heparin (500 IU·mL−1), with samples being taken at rest, in the last 30 s of each 2-min workload during exercise and every minute for 10 min postexercise. Aliquots (300 μL) were analyzed for the plasma concentration of lactate ([La−], mmol·L−1) (ABL 615, Radiometer Medical, Copenhagen, Denmark).
Minute ventilation (V̇E), oxygen uptake (V̇O2), and carbon dioxide production (V̇CO2) were calculated from 30-s sampling epochs throughout each test using an open-circuit indirect calorimetry system, similar to that used previously (13). In brief, subjects breathed through a low-resistance respiratory valve with a precalibrated large flow turbine transducer (P. K. Morgan, London) attached to the inspiratory port. Expired air was directed to a 2.6-L mixing chamber (Sportech, Canberra, Australia) from which dried gas was sampled continuously (∼500 mL·min−1) and passed to oxygen (O2) and carbon dioxide (CO2) analyzers (Servomex, East Sussex, United Kingdom), calibrated before each exercise test over the physiological range of measurement with commercially produced gas mixtures of known O2 and CO2 percentages (BOC Gases, North Ryde, NSW, Australia). The ventilometer and analyzers were interfaced with an IBM-compatible computer that performed all of the necessary calculations by using standard algorithms and Labview-based software (Metabolic Analyzer) developed by ICON technologies for the Western Australian Institute of Sport. V̇E (L·min−1), V̇O2 (mL·kg−1·min−1), V̇CO2 (mL·kg−1·min−1), and the respiratory exchange ratio (RER) were calculated by averaging the values obtained from the two consecutive 30-s sampling epochs within each minute. Heart rate (HR, beats·min−1) was recorded continuously throughout each test (consecutive five beat averages) with the highest value designated as maximal heart rate (HRmax).
Maximal oxygen uptake (V̇O2max).
V̇O2max was designated as the mean V̇O2 of the minute in which the highest 30-s epoch value was recorded (14) and was expressed in terms of total body mass (mL·kg−1·min−1). Although a plateau in V̇O2 (i.e., a change in V̇O2 of < 150 mL·min−1 or < 2 mL·kg−1·min−1 between successive increments in work load) is regarded as the absolute criterion of V̇O2max, this is difficult to obtain in sedentary subjects (14). Therefore, in the present study, if a plateau was not observed, a value for V̇O2 was still considered as maximal if subjects fulfilled two or more of the following secondary criteria: 1) achievement of their age-predicted HRmax ± 11 beats·min−1; 2) an RER greater than 1.10; and 3) a postexercise plasma [La−] of ≥ 8 mmol·L−1 (14). On this basis, one female CFS patient and two control subjects, initially recruited into the study, were excluded such that the final subject numbers analyzed were 16 male and 17 female CFS patients and their matched controls. Each of the groups derived from this remaining subject cohort achieved more than three of the four criteria listed in Table 2 (male subjects: CFS, 3.2 ± 0.5; controls, 3.3 ± 0.6; female subjects: CFS, 3.1 ± 0.7; controls, 3.0 ± 0.6), with no difference between groups or between the percentage of subjects in each group achieving any given criterion (Table 2). Age-based predicted values for V̇O2max were calculated from regression equations derived from maximal testing in a cohort of healthy sedentary male subjects (V̇O2max in mL·kg−1·min−1 = 57.8 − [0.445 × age in yr], N = 94) and female subjects (V̇O2max in mL·kg−1·min−1 = 42.3 − [0.356 × age in yr, N = 113]) (3). The degree of functional aerobic impairment (FAI) was determined in each subject by calculating the percentage difference between their measured and age- and gender-predicted values for V̇O2max, with a value of 0% indicating no FAI (3).
Changes in plasma [La.] during exercise.
These were modeled for each subject as a continuous exponential function of V̇O2 (mL·kg−1·min−1) given by the equation:MATHwhere a, b, and c are mathematical parameters estimated by minimizing the residual sum of squares between the plasma [La−] data and the curve fit (15).
Lactate threshold (LT) determination.
This was determined according to the method of Beaver et al. (2) by using a purpose-designed computer program in Basic (Microsoft Version 3.2) to transform the plasma [La−] and V̇O2 data into logarithms. The log-log relationships were plotted and regression lines fitted through the upper and lower segments of the resultant plots while minimizing the residual sum of squares. The LT was designated as the V̇O2 corresponding with the point of intersection of the two regression lines and expressed either in mL·kg−1·min−1 or as a percentage of V̇O2max. The power output and HR corresponding to the V̇O2 at the LT were also determined from linear regression equations of V̇O2 versus power output and HR.
Given the well-known differences in physical status between male and female subjects, all comparisons between CFS patients and controls were made within each gender group. Group differences were analyzed using Student’s independent paired t-test where appropriate. A two-way analysis of variance (ANOVA) with repeated measures was used to compare measured and predicted values for V̇O2max and HRmax. Post hoc analyses were conducted using Tukey’s test where the two-way ANOVA showed a significant main effect. Relationships between V̇O2max and both illness duration and severity in CFS patients of both genders were tested by linear regression analysis and reported as the coefficient of determination (r2). A power analysis was performed using data from a previous study in which the V̇O2max results were stratified according to gender (6). The calculation indicated that a sample size of 10 male subjects and seven female subjects would be adequate to detect a significant difference in V̇O2max. Unless otherwise stated, the level of significance was set at P ≤ 0.05 for all statistical tests. All data are expressed as the mean ± SD.
The physical characteristics of all subjects, stratified on the basis of gender, are contained in Table 3. Mean age, height, mass, and body mass index were not different between CFS patients and sedentary controls within each gender group, and in each subject cohort these parameters were not different from Australian normative data for a gender- and age-matched healthy population (12). Within CFS patients, there was no correlation between illness duration (months) and severity (KPS scale) (male subjects, r2 = 0.000; female subjects, r2 = 0.147).
Maximal Values for V̇O2, V̇E, HR, and Plasma [La−]
V̇O2max was not different between CFS patients and controls (Table 4), and these values were, respectively, 96.3 ± 17.9% and 103.6 ± 16.7% of age-predicted values (CFS, 42.5 ± 4.7; controls, 41.9 ± 4.8 mL·kg−1·min−1), indicating no functional aerobic impairment (%FAI: CFS, 3.7 ± 17.9; controls, −3.6 ± 16.7). There was no correlation between V̇O2max and either illness duration (r2 = 0.012) or severity (r2 = 0.052) in the CFS patients. HRmax was lower in CFS patients compared with controls (P = 0.016), but these values (Table 4) were, respectively, 99.1 ± 5.5% and 104.2 ± 5.8% of their age-predicted values (CFS, 186 ± 10; controls, 184 ± 11, beats·min−1). V̇Emax was not different between subject groups (Table 4). Peak plasma [La−] and the time to peak plasma [La−] were not different between CFS patients and controls (Table 4).
V̇O2max was lower in CFS patients compared with controls (P = 0.002), but these values (Table 4) were, respectively, 101.2 ± 20.4% and 112.6 ± 15.4% of age-predicted values (CFS, 30.1 ± 4.0; controls, 30.6 ± 4.5 mL·kg−1·min−1), indicating no functional aerobic impairment (%FAI: CFS, −1.2 ± 20.4; controls, −12.6 ± 15.4). The value in control subjects was significantly higher than their age-predicted value (P = 0.008). In CFS patients, there was no correlation between V̇O2max and either illness duration (r2 = 0.073) or severity (r2 = 0.011). HRmax was not different between CFS patients and controls, and these values were, respectively, 98.9 ± 5.1% and 99.5 ± 4.0% of their age-predicted values (CFS, 186 ± 11; controls, 187 ± 13 beats·min−1). V̇Emax was not different between subject groups (Table 4). Peak plasma [La−] and the time to peak plasma [La−] were not different between CFS patients and controls (Table 4).
Plasma Lactate Accumulation and the Lactate Threshold (LT)
Within each gender group, the rate constants describing the increase in plasma [La−] relative to workload were not different between CFS patients and controls (male subjects: CFS, 0.030 ± 0.011; controls, 0.029 ± 0.011; female subjects: CFS, 0.046 ± 0.035; controls, 0.036 ± 0.020). The V̇O2 at the LT in each gender group, whether expressed in mL·kg−1·min−1 or as a percentage of V̇O2max, was not different between CFS patients and controls (Table 5). Similarly, the HR and power output at the LT were not different between CFS patients and controls within each gender group.
The results from the present study strongly suggest that both maximal oxygen uptake and lactate metabolism are normal in CFS patients of both genders and cannot explain the undue fatigue they experience during and after exercise. It should be emphasized that these two variables are not the only determinants of exercise performance (1), which, on clinical grounds, is well recognized to be impaired in CFS patients, with a reduced exercise time to exhaustion being a common finding (8,11,23,26,30). Unfortunately, a valid assessment of performance in CFS patients is difficult to obtain because laboratory tests, where subjects exercise at a given intensity until exhaustion, lack sufficient precision to discriminate between sedentary subject groups (16).
With regard to V̇O2max, the variance between the present results and the common finding in previous studies of reductions in V̇O2max of up to 38% could have two explanations. First, it is well known that V̇O2max (mL·kg−1·min−1) is ∼20% lower in female subjects (22), a finding confirmed in both CFS patients and control subjects in the present study. Therefore, the pooling of male and female data makes any reported differences from sedentary controls difficult to interpret, and even more so where the gender balance has been unequal between the patient and control groups (8,11). Only two studies have analyzed their V̇O2max data on a gender basis and both reported lower values in CFS patients (6,26). Second, most previous studies have not employed test protocols and methods of interpretation regarded as essential for the valid assessment of V̇O2max (4). In particular, the well-established endpoints for a maximal effort during incremental exercise and acceptance of a given V̇O2 as a maximal value have not been met in most studies (6,8,11,23). Furthermore, because CFS patients do not have defined or life-threatening cardiorespiratory pathology, the employment of symptom-limited protocols (23,26) and avoidance of vocal exhortation as maximal effort is approached (26) seem unwarranted and will make it less likely for a given subject to reach true V̇O2max. This was reinforced in the present study where the criteria identifying a maximal effort were achieved to an equal degree by CFS patients and their healthy sedentary controls.
The finding of similar values for V̇O2max in CFS patients and healthy sedentary controls does not necessarily imply that exercise performance will be similarly matched (24). This is acknowledged in most exercise testing protocols by obtaining an additional exercise capacity index in the companion measurement of the LT. This measurement defines the percentage of V̇O2max that can be accessed before the onset of progressive lactic acid accumulation and is regarded as a better predictor of capacity for endurance performance than V̇O2max (5,24). Although a leftward shift of the LT with an earlier onset of lactic acidosis has been suggested in CFS patients (21), the gas exchange threshold, an indirect estimate of the onset of metabolic acidosis, was found to be normal in a study by Sisto et al. (26). This latter finding was supported in the present study where the LT was measured directly and found to be not different between CFS patients and controls. Furthermore, within each gender group, the rate of plasma lactate accumulation during incremental exercise, the postexercise peak plasma [La−], and the time to peak also were not different, indicating that excessive lactic acidosis is not a mechanism of fatigue in these patients despite suggestions to the contrary (21,30).
It could be argued that the cohort of CFS patients in the present study was less incapacitated than those in previous studies where reductions in V̇O2max have been reported with consistency. Yet all patients were reassessed by the same physician (R. Burnet) within the 3 months preceding the present exercise study and continued to meet all the diagnostic criteria as defined by Fukuda et al. (9). Furthermore, subjective evaluation of their fatigue immediately before testing by a second physician (G. Scroop) using the KPS scale, which has been used widely to evaluate illness severity in CFS, provided values consistent with those reported previously in CFS patients (29). Mean illness duration was longer than in previous studies (23,26,30), implying a longer period for “deconditioning,” yet V̇O2max was normal and did not correlate with illness duration.
Taken together, the present findings indicate that the exercise capacity of CFS patients is not significantly impaired, either as a direct result of their illness or their imposed sedentary lifestyle, and that neither physical deconditioning nor cardiorespiratory dysfunction is a critical factor in the fatigue that CFS patients experience. Although the recommendation (17,27) or imposition (10) of exercise-training programs for CFS patients may have benefits in terms of maintaining flexibility and improving self-esteem and social interaction (24), such programs could well be based on a false premise if the intention is to improve patient management and well-being by correcting the effects of deconditioning.
The authors are grateful to the Chronic Fatigue Syndrome Society (SA) Inc. and the several private individuals throughout Australia who generously donated funds to support this research. We would particularly like to acknowledge the selfless support of the CFS patients and their carers who volunteered to participate in this study.
Address for correspondence: Dr. Garry Scroop, Department of Physiology, University of Adelaide, South Australia 5005, E-mail: email@example.com.
1. Astrand, P. O., and K. Rodahl. Textbook of Work Physiology, 3rd Ed. Singapore: McGraw-Hill, 1986, pp. 648.
2. Beaver, W. L., K. Wasserman, and B. J. Whipp. Improved detection of lactate threshold
using a log-log transformation. J. Appl. Physiol. 59: 1936–1940, 1985.
3. Bruce, R. A., F. Kusumi, and D. Hosmer. Maximal oxygen intake and nomographic assessment of functional aerobic impairment in cardiovascular disease. Am. Heart J. 85: 546–562, 1973.
4. Buchfuhrer, M. J., J. E. Hansen, T. E. Robinson, D. Y. Sue, K. Wasserman, and B. J. Whipp. Optimizing the exercise
protocol for cardiopulmonary assessment. J. Appl. Physiol. Respirat. Environ. Exerc. Physiol. 55: 1558–1564, 1983.
5. Coyle, E. F., A. R. Coggan, M. K. Hopper, and T. J. Walters. Determinants of endurance in well-trained cyclists. J. Appl. Physiol. 64: 2622–2630, 1988.
6. De Becker, P., I. Campine, and K. De Meirleir. Measuring physical disability in patients with chronic fatigue syndrome using exercise
testing (Abstract). Chest 116: 335s, 1999.
7. Evengard, B., R. S. Schacterle, and A. L. Komaroff. Chronic fatigue syndrome: new insights and old ignorance. J. Intern. Med. 246: 455–469, 1999.
8. Fischler, B., P. Dendale, V. Michiels, R. Cluydts, L. Kaufman, and K. De Meirleir. Physical fatigability and exercise capacity
in chronic fatigue syndrome: association with disability, somatization and psychopathology. J. Psychosom. Res. 42: 369–378, 1997.
9. Fukuda, K., S. E. Straus, I. Hickie, M. C. Sharpe, J. G. Dobbins, and A. Komaroff. The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann. Intern. Med. 121: 953–959, 1994.
10. Fulcher, K. Y., and P. D. White. Randomised controlled trial of graded exercise
in patients with the chronic fatigue syndrome. Br. Med. J. 314: 1647–1652, 1997.
11. Fulcher, K. Y., and P. D. White. Strength and physiological response to exercise
in patients with chronic fatigue syndrome. J. Neurol. Neurosurg. Psychiatry 69: 302–307, 2000.
12. Gore, C. J., and D. A. Edwards. Australian Fitness Norms: A Manual for Fitness Assessors. South Australia: The Health Development Foundation, 1992, p. 34.
13. Gore, R. J., G. C. Scroop, J. D. Marker, and P. G. Catcheside. Plasma volume, osmolarity, total protein and electrolytes during treadmill running and cycle ergometer exercise
. Eur. J. Appl. Physiol. Occup. Physiol. 65: 302–310, 1992.
14. Howley, E. T., D. R. Bassett, and H. G. Welch. Criteria for maximal oxygen uptake: review and commentary. Med. Sci. Sports Exerc. 27: 1292–1301, 1995.
15. Hughson, R. L., K. H. Weisiger, and G. D. Swanson. Blood lactate concentration increases as a continuous function in progressive exercise
. J. Appl. Physiol. 62: 1975–1981, 1987.
16. Jeukendrup, A., W. H. M. Saris, F. Brouns, and A. D. M. Kester. A new validated endurance performance test. Med. Sci. Sports Exerc. 28: 266–270, 1996.
17. Joint Working Group of the Royal Colleges of Physicians, Psychiatrists and General Practitioners. Chronic Fatigue Syndrome. London: Royal College of Physicians, 1996, pp. 24–27.
18. Karnofsky, D. A., and J. H. Burchenal. The clinical evaluation of chemotherapeutic agents in cancer. In: Evaluation of Chemotherapeutic Agents, C. M. Macleod (Ed.). New York: Columbia University Press, 1949, pp. 109–205.
19. Komaroff, A. L. Clinical presentation of chronic fatigue syndrome. In: Ciba Foundation Symposium 173, Chronic Fatigue Syndrome, G. R. Bock, and J. Whelan (Eds.). New York: J. Wiley and Sons, 1993, pp. 43–61.
20. Kruesi, M. J. P., J. Dale, and S. E. Straus. Psychiatric diagnoses in patients who have chronic fatigue syndrome. J. Clin. Psychiatry 50: 53–56, 1989.
21. Lane, R. J. M., D. Woodrow, and L. C. Archard. Lactate responses to exercise
in chronic fatigue syndrome. J. Neurol. Neurosurg. Psychiatry 57: 662–663, 1994.
22. McArdle, W. D., F. I. Katch, and V. L. Katch. Exercise
Physiology: Energy, Nutrition and Human Performance, 4th Ed. Baltimore: Williams & Wilkins, 1996, pp. 204–207.
23. Riley, M. S., C. J. O’Brien, D. R. McCluskey, N. P. Bell, and D. P. Nicholls. Aerobic work capacity in patients with chronic fatigue syndrome. Br. Med. J. 301: 953–956, 1990.
24. Robergs, R. A., and S. O. Roberts. Fundamental Principles of Exercise Physiology for Fitness Performance and Health
. New York: The McGraw-Hill, 2000, pp.267, 274, 426.
25. Shephard, R. J. Semantic and physiological definitions. In: Endurance in Sport, R. J. Shephard, and P. O. Astrand (Eds.). London: Blackwell Scientific Publications, 1992, pp. 3–7.
26. Sisto, S. A., J. La Manca, D. L. Cordero, et al. Metabolic and cardiovascular effects of a progressive exercise
test in patients with chronic fatigue syndrome. Am. J. Med. 100: 634–640, 1996.
27. Stokes, M. J., R. G. Cooper, and R. H. T. Edwards. Normal muscle strength and fatigability in patients with effort syndromes. Br. Med. J. 297: 1014–1017, 1988.
28. Vercoulen, J. H., E. Bazelmans, C. M. Swanink, et al. Physical activity in chronic fatigue syndrome: assessment and its role in fatigue. J. Psychiatr. Res. 31: 661–673, 1997.
29. Vollmer-Conna, U., I. Hickie, D. Hadzi-Pavlovic, et al. Intravenous immunoglobulin is ineffective in the treatment of patients with chronic fatigue syndrome. Am. J. Med. 103: 38–43, 1997.
30. Wong, R., G. Lopaschuk, G. Zhu, et al. Skeletal muscle metabolism in the chronic fatigue syndrome. Chest 102: 1716–1722, 1992.