Key words: anemia; COPD; erythropoietin; hematocrit; long-term oxygen therapy; survival

Transcription

1 Prognostic Value of the Hematocrit in Patients With Severe COPD Receiving Long-term Oxygen Therapy* Arnaud Chambellan, MD; Edmond Chailleux, MD; Thomas Similowski, MD, PhD; and the ANTADIR Observatory Group Background: Although traditionally associated with polycythemia, COPD has a systemic inflammatory component that could interfere with erythropoiesis. This study describes the distribution and prognostic value of the hematocrit in patients with severe COPD receiving long-term oxygen therapy (LTOT). Methods: A total of 2,524 patients with COPD, FEV 1 /vital capacity (VC) < 70%, FEV 1 < 80% of predicted, and PaO 2 < 7.3 kpa in whom a hematocrit was available at entry was identified between 1980 and 1999 in the French Association Nationale pour le Traitement à Domicile de l Insuffisance Respiratoire chronic respiratory insufficiency and home-care database (male/ female ratio, 5/1; mean SD age, years for men, and years for women). Correlations between hematocrit, demographic data, and pulmonary function data were examined. A multivariate Cox proportional hazard regression was performed to identify prognostic factors. Results: Mean hematocrit was % in men and % in women (< 39% in 12.6% of men, and < 36% in 8.2% of women) according to the World Health Organization definition of anemia. Hematocrit was negatively correlated with age (r 0.245) and FEV 1 /VC (r 0.068) and was positively correlated with PaCO 2 (r 0.161) and body mass index (r 0.127). Multivariate analysis found hematocrit to be an independent predictor of survival, hospital admission rate, and cumulative duration of hospitalization. The 3-year survival was 24% (95% confidence interval, 16 to 33%) when the hematocrit was < 35%, and 70% (63 to 76%) when the hematocrit was > 55%. Conclusions: A low hematocrit is not uncommon in LTOT/COPD patients. Hematocrit is negatively associated with mortality and morbidity. Whether the association is causative or not and whether or not corrective measures are warranted remain to be determined. (CHEST 2005; 128: ) Key words: anemia; COPD; erythropoietin; hematocrit; long-term oxygen therapy; survival Abbreviations: ANTADIR Association Nationale pour le Traitement à Domicile de l Insuffisance Respiratoire; BMI body mass index; BODE body mass index, airflow obstruction, dyspnea, and exercise capacity; CI confidence interval; IL interleukin; LTOT long-term oxygen therapy; TNF tumor necrosis factor; VC vital capacity *From the Service de Pneumologie (Dr. Chailleux) and Laboratoire des Explorations Fonctionnelles (Dr. Chambellan), CHU de Nantes, Nantes; and the Service de Pneumologie (Dr. Similowski), Groupe Hospitalier Pitié-Salpêtriere, Assistance Publique-Hôpitaux de Paris, Paris, France. Fédération Antadir, 66 Bd St-Michel, Paris, France. Dr. Chambellan was supported by a grant from the Association Regionale de l Insuffisance Respiratoire des Pays de la Loire. COPD is, according to a widely disseminated definition, 1 a disease state characterized by airflow limitation that is not fully reversible, where the airflow limitation is usually both progressive and associated with an abnormal inflammatory response of the lungs to noxious particles or gases (the main source of which is tobacco smoke). Besides spirometric abnormalities, bronchopulmonary lesions lead to ventilation-perfusion imbalance and the ensuing gas exchange abnormalities. FEV 1 and hypoxemia are strong prognostic markers, 2,3 and long-term oxygen therapy (LTOT) has been shown to improve survival. 4,5 The deleterious effects of hypoxemia proceed from the effects of tissue hypoxia on cell metabolism Manuscript received January 13, 2005; revision accepted February 14, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Edmond Chailleux, MD, Service de Pneumologie, CHU de Nantes, Boulevard Jacques Monod, Nantes Cedex 1, France; edmond.chailleux@chu-nantes.fr CHEST / 128 / 3/ SEPTEMBER,

2 and organ function. Among these effects, the stimulation of erythropoietin production can lead to a compensatory erythrocytosis that is traditionally viewed as a typical of COPD. COPD also has a systemic inflammatory dimension. Patients with COPD often exhibit raised levels of proinflammatory cytokines, eg, interleukin (IL)-1, IL-6, tumor necrosis factor (TNF)-, chemokines (IL-8, monocyte chemotactic protein-a), and C-reactive protein. The expression of neutrophil adhesion molecules is increased, and there are changes in neutrophil functions, and many other perturbations. 6,7 In some COPD patients, this could contribute to an increased energy expenditure, promoting muscle wasting, nutritional imbalance, and weight loss. Weight loss in patients with COPD has been correlated with serum TNF- levels 8 and has a negative prognostic value. 9 COPD-related inflammation could also impair erythropoiesis, as do other chronic inflammatory processes. Hemoglobin levels in COPD patients would then reflect the balance between the stimulation of erythropoiesis by hypoxia and its depression by inflammation. Inadequate hemoglobin levels could aggravate tissue hypoxia and carry a negative prognostic impact. Hints as to the plausibility of this scenario exist in the literature. Blood cell transfusion in anemic COPD patients reduces minute ventilation and the work of breathing, 10 suggesting that correcting low hemoglobin levels could alleviate dyspnea and improve exercise capacity. In a small set of anemic ventilatordependent COPD patients, raising hemoglobin levels to 12 g/dl seemed to improve patients enough to make ventilator weaning possible. 11 A reduced hematocrit was found to be an independent predictor of poor outcomes in COPD patients following elective open abdominal aortic aneurysm resection. 12 In the COPD population from which Celli et al 13 derived the prognostic value of a multidimensional index combining four variables (body mass index [BMI], airflow obstruction, dyspnea, and exercise capacity [BODE] index), the patients who died had a significantly lower hematocrit (mean SD, 39 5%) than those who survived (42 5%). The present study was conducted to look for a putative association between hematocrit and prognosis in severe COPD patients receiving LTOT. The data were extracted, over a period of 20 years, from a large national database in France maintained by the Association Nationale pour le Traitement à Domicile de l Insuffisance Respiratoire (ANTADIR). This database has already been used, for example, to describe the main prognostic factors in various types of chronic respiratory insufficiency, 14 or to relate BMI 15 to outcome in this population. Materials and Methods Organization of the ANTADIR Observatory The ANTADIR is a French national nonprofit associative network of 23 regional associations founded in 1981 to ensure health-care support and technical follow-up to disabled patients requiring domiciliary oxygen support, mechanical ventilation, or continuous positive airway pressure. Since its foundation, ANTADIR has maintained a comprehensive demographic database or observatory, 14 providing a unique national register allowing epidemiologic surveys in large cohorts of patients with various types of chronic respiratory diseases. Individual patient records derive from the single document filled in by prescribing physicians for the reimbursement of costs by the French social security system. Available information includes demographic data (age, sex, height, and weight) and clinical data (diagnosis, room air arterial blood gases and hematocrit, pulmonary function tests, and smoking status). Prescriptions, hospitalizations, treatment withdrawals, and deaths are registered by the regional associations using common software and compiled together by ANTADIR, which manages the national database and produces an annual report. Patients Among the patients registered in the database, 11,366 had a clinical diagnosis of COPD with a FEV 1 /vital capacity (VC) ratio 70% and a FEV 1 80% of predicted. Room air Pao 2 was 7.3 kpa in 6,111 cases. The initial hematocrit was available in 2,524 of these patients (2,097 men and 427 women), who constitute the study cohort (Table 1). In this population, women were older than men and were more likely to be nonsmokers than ex-smokers or current smokers. Although they had been prescribed oxygen, 10.2% of the patients were reported as current smokers (men, 9.9%; women, 11.3%). Hematocrit Hematocrit values considered for analysis were those recorded with the arterial blood gases retained to prescribe LTOT. Six hematocrit categories (21 to 34%, 35 to 39%, 40 to 44%, 45 to 49%, 50 to 54% and 55%) were defined for the purpose of statistical calculations. Table 1 Demographic and Functional Characteristics of the Patients by Sex* Characteristics Men (n 2,097) Women (n 427) Age, yr 68 (10) 70 (10) Smoking status, No. (%) Smoker 163 (9.9) 40 (11.3) Ex-smoker 1,438 (87.7) 205 (57.7) Nonsmoker 38 (2.3) 110 (31.0) FEV 1,L 0.92 (0.37) 0.68 (0.23) FEV 1, % predicted 34 (13) 37 (13) VC, L 2.26 (0.71) 1.49 (0.51) VC, % predicted 61 (19) 63 (19) FEV 1 /VC, % 41 (12) 47 (12) Pao 2, kpa 6.5 (0.7) 6.4 (0.7) Paco 2, kpa 6.4 (1.2) 6.7 (1.1) BMI, kg/m 2 24 (5) 25 (7) Hematocrit, % 45.9 (6.8) 43.9 (6.3) *Data are expressed as mean (SD) unless otherwise indicated. Available for 2,450 patients Clinical Investigations

3 Anthropometric Data and Pulmonary Function Tests The BMI was calculated as the ratio of weight to height squared from the data available in the initial patients records. Spirometric values obtained in body temperature and pressure, saturated, conditions were expressed as percentages of predicted values. 16 Outcomes The annual death rate, annual hospital admission rate, and annual number of days spent in the hospital were computed from the follow-up records. Statistical Analysis In the 2,524 studied patients, the associations between hematocrit and age, BMI, and pulmonary function tests were studied by linear correlation and represented graphically for each hematocrit category. The population survival was calculated from the onset of LTOT by the actuarial and Kaplan- Meier methods with a closing date of January 1, The study of prognostic factors was performed using the log-rank test and Cox semiparametric models as previously described. 14,15 The influence of hematocrit, age, sex, BMI, arterial blood gases, and pulmonary function tests on the annual rate of hospital admissions and on the number of days spent in the hospital were studied by univariate regression followed by a multivariate stepwise regression. These patients were also compared to the 3,587 patients with a FEV 1 /VC ratio 70%, a FEV 1 80% of predicted, and a room air Pao kpa but in whom a hematocrit was not available. Results Hematocrit Anemia can be defined by a hematocrit 39% in men and 36% in women. 17 According to this definition, the prevalence of anemia in the cohort studied was 12.6% in men and 8.2% in women (but 18.5% of the women had a hematocrit 39%). The prevalence of a hematocrit value 55% was 8.4% (men, 8.9%; women, 5.9%). Hematocrit was significantly higher in men than in women (t 5.6, p 0.001), and was negatively correlated with age (r 0.245, p 0.001) and positively correlated with Paco 2 (r 0.161, p 0.001) and BMI (r 0.127, p 0.001), but there was no significant correlation between hematocrit and Pao 2 (r 0.033, p 0.095). A weak negative correlation existed between hematocrit and FEV 1 /VC (r 0.084, p 0.001), and between hematocrit and FEV 1 percentage of predicted (r 0.054, p 0.006) [Fig 1]. Current smokers had significantly higher hematocrit values than ex-smokers or nonsmokers (Table 2). Survival At the time of analysis, 1,842 patients had died, 266 were alive, and 416 were unavailable for follow-up by the ANTADIR system (two thirds of those were known to have stopped oxygen for motives such as alleged poor tolerance or a personal wish; one third was known to have entered a long-term care facility). The mean duration of follow-up was 10 years. The overall median survival was 3.0 years. Survival was increased when the hematocrit was high and reached the longest duration in polycythemic patients (p 0.001, logrank test) [Fig 2]. The 3-year survival rate was 24% (95% confidence interval [CI], 16 to 33%) in patients with a hematocrit 35%; 36% (95% CI, 31 to 42%) in patients with a hematocrit between 35% and 39%; 47% (95% CI, 43 to 51%) in patients with a hematocrit between 40% and 44%; 51% (95% CI, 47 to 55%) in patients with a hematocrit between 45% and 49%; 59% (95% CI, 54 to 64%) in patients with a hematocrit between 50% and 54%; and 70% (95% CI, 63 to 76%) in patients with a hematocrit 55%. Multivariate analysis ranked age, hematocrit, BMI, Pao 2, sex, and FEV 1 percentage of predicted as prognostic factors, in descending order (Table 3). The prognostic influence of the hematocrit was significant in both men and women. The interactions between correlated variables (hematocrit and BMI, hematocrit and FEV 1, hematocrit and Pao 2 ) were not significant predictors of survival. Admission Rate and Hospital Length of Stay For the 1,799 patients with a follow-up of at least 1 year, the annual hospitalization rate was (median, 0.86) and the average number of days in the hospital was (median, 13.3). Univariate analysis showed that hematocrit was negatively correlated with the rate (r 0.091, p 0.001) and duration (r 0.095, p 0.001) of hospitalizations (Fig 3). This correlation, although weak, was the second strongest found in the data set after the correlation with between BMI and hospitalizations. Patients with a hematocrit 35% were hospitalized (mean SD) times per year, with days in the hospital per year. However, patients with a hematocrit 55% were hospitalized times a year on average, with a mean annual time spent in the hospital of days. Stepwise multivariate regression selected BMI (p 0.001), hematocrit (p 0.002), Paco 2 (p 0.002), and age (p 0.027) as predictive factors of the annual admission rate, whereas hematocrit (p 0.001), BMI (p 0.008), and age (p 0.017) were predictive of the annual duration of hospitalization. Comparison of the Patients With and Without a Known Hematocrit Value These two groups of patients were not significantly different regarding the sex ratio, age, tobacco smoking CHEST / 128 / 3/ SEPTEMBER,

4 Figure 1. Links between hematocrit, age, BMI, and respiratory function. The values shown are the mean age, BMI, and spirometric and arterial blood gas values for each interval of hematocrit (SEM). pred predicted. status, Pao 2, VC, and FEV 1. The patients with a known hematocrit value had a marginally but significantly higher Paco 2 ( mm Hg vs mm Hg) and lower BMI ( kg/m 2 vs kg/m 2 ). The outcomes were slightly better in the group with an unknown hematocrit than in the group with a known hematocrit (median survival, 1,154 days vs 1,086 days [p 0.016, log-rank]; admission rate, Table 2 Hematocrit by Sex and Smoking Habits* Hematocrit, % Men Women All Smoker 49.0 (7.1) 45.9 (7.5) 48.4 (7.3) Ex-smoker 45.5 (6.6) 44.2 (6.4) 45.3 (6.6) Non-smoker 44.3 (5.7) 42.8 (5.9) 43.2 (5.9) *Data are expressed as mean (SD). Hematocrit is higher in smokers vs ex-smokers and nonsmokers. p p vs [p 0.003]; length of stay, days vs days [p 0.001]). Discussion This study shows that in a large population of COPD patients treated with LTOT, the prevalence of a low hematocrit at the time the prescription of oxygen is far from negligible (12.6% of the men and 18.5% of the women meeting the World Health Organization definition of anemia [hematocrit 39%] 17 ). In this particular population, the study also points to a strong association between hematocrit and long-term survival in this population. These findings seem to contrast with commonly held views. Although an apparent lack of polycythemic response to hypoxia has been noted in COPD patients in early studies, 18,19 red cell mass has consistently been found increased 20,21 and anemia never 1204 Clinical Investigations

5 Figure 2. Ten-year survival analysis of the influence of hematocrit calculated by the actuarial method. The initial numbers in each group were 120 ( 35 years old), 323 (35 to 39 years old), 675 (40 to 44 years old), 739 (45 to 49 years old), 456 (50 to 54 years old), and 211 ( 55 years old); log-rank, 171; degree of freedom 5(p 0.001). reported with a prevalence such as the one we found. Polycythemia in COPD has been assigned no effect on survival 22 or a negative effect. 4,5 Cohort Characteristics, Prognostic Impact, and Shape of Risk The patients in our cohort were mostly elderly people of the male gender (83%). As expected, they had severe COPD (mean FEV 1 in men, L/s; in women, L/s; median survival, 3.0 years). Hematocrit was negatively correlated with age, a common finding in geriatric medicine. 23 It was Table 3 Multivariate Study of Prognostic Factors by Cox Model* Factors Relative Risk (95% CI) t Test p Value Age, by 10 yr 1.42 ( ) Hematocrit, by 5% 0.86 ( ) BMI, by 5 kg/m ( ) Pao 2,by5mmHg 0.86 ( ) Sex, female vs male 0.74 ( ) FEV 1 % predicted, by 10% 0.93 ( ) *Relative risk of death for a given increase of each significant variable in the final multivariate Cox model calculated by stepwise selection (n 2,450). A relative risk 1 indicates a poorer prognosis for higher values of the variable (age). Sex was coded 0 for men and 1 for women; a relative risk 1 indicates a better prognosis for women. also correlated, less strongly, with BMI and Paco 2. Age and the male gender were identified as negative prognostic factors by the Cox model. This model also indicated that increasing values of Pao 2,BMI,and FEV 1 percentage of predicted were significantly associated with a better prognosis. These results fit with established facts in COPD. 4,5,24,25 In addition, the Cox model showed that each 5% increase in hematocrit was associated with improved survival, and that hematocrit was the strongest prognostic factor next to age. Even though cross-correlation did exist between some of the model variables, none of their interactions had a significant effect. The 10-year actuarial survival curve (Fig 2) shows a strong correlation between each 5% stratum of hematocrit and the mortality rate, particularly within the first years. The annual hospital admission rate and the cumulative hospital length of stay were highest in the patients with the lowest hematocrit levels and decreased in a roughly linear manner with hematocrit. This observation may be viewed as surprising, as a U-shaped risk could have been expected, but it may be due to the relatively small number of patients with high or very high hematocrit values. Further studies will be necessary to determine whether the hematocrit-associated risk is gradual or if it has a threshold. Hypercapnia was positively correlated with hematocrit. If one equates hypercapnia with disease severity, the inverse correlation of hematocrit with sur- CHEST / 128 / 3/ SEPTEMBER,

6 Figure 3. Influence of hematocrit on hospitalizations. The values shown are the mean annual total of days spent in hospital and the mean annual rate of admission for six intervals of hematocrit (SEM), calculated in 1,799 patients with a follow-up of at least 1 year. vival may then seem paradoxical. However a low hemoglobin level stimulates ventilation and could thus be responsible for artificially low values of Paco Of note, our first study from the ANTADIR database showed that in COPD patients receiving LTOT, a low Paco 2 had a negative prognostic value. 14 Limitations and Strength of the Study The study has limitations. First, it is a retrospective observational study, and as such it tells nothing on the causative nature of a statistical association between a risk factor and a given outcome. Second, the only hematocrit value considered is the one noted at the time of the initiation of LTOT. As a result, the prognosis value of the dynamics of the hematocrit over time is unknown, as are on the impact of the compliance with oxygen therapy and of changes in smoking status. Third, the study pertains to a fraction of the relevant population, because a hematocrit was available in only one third of the patients meeting the inclusion criteria. A hematocrit should be gathered systematically on ANTADIR forms, and why this is not the case is only too obvious. The comparison of the population with and without known hematocrit values shows only slight differences in characteristics, but indicate differences in prognostic suggesting that somehow clinicians might note down hematocrit in more severe patients. However, we feel that this does not intrinsically undervalue our results, which are what they are in the population studied. Fourth, there is no possibility to identify, from the database, the existence of comorbidities that could either be specific etiologies for the decrease in hemoglobin (eg, chronic renal failure or GI bleeding) or intrinsically aggravate the prognosis. Finally, the database has been in operation over a long time period, during which treatments for COPD and patterns of healthcare use have changed, which may have influenced the hematocrit. These limitations must be taken into account (see Perspectives ) but also weighted against the strengths of the study. The principal of these strengths lie in the size of the cohort and in the nature of the database from which it was derived. Indeed, the ANTADIR observatory has been continuously recording data from LTOT patients 25 years, and from 23 regions of France. Therefore, we were able to study a very large population of patients characterized in a standard manner in terms of demographic data, pulmonary function tests, and intercurrent events. Although not a typical epidemiologic database (such as the Framingham database, in which all patients are followed up at fixed intervals to collect complete information), the characteristics of the ANTADIR database are such that selection and censoring biases are not likely to be major. The 10-year average follow-up adds weight to our observations. Possible Mechanisms of Anemia in COPD Anemia increases in prevalence with age, which may account in part for our observations. GI bleeding; chronic renal failure; deficiencies in vitamin B 12, folate, or iron; and myelodysplastic syndromes can coexist with COPD, and should not be overlooked in anemic COPD patients. Chronic infections, chronic 1206 Clinical Investigations

7 inflammation, and neoplastic diseases cause the socalled anemia of chronic disease, 26 also a feature of chronic heart failure where anemia is an independent risk factor for mortality. 27 These diseases can all coexist with COPD, but COPD itself might cause anemia of chronic disease. A full review of the possible mechanisms is beyond the scope of this article, but some facts warrant attention. In the anemia of chronic disease, shortened survival of RBCs is thought to result from raised levels of IL-1 and TNF. 26 These findings are common in COPD and exaggerated during exacerbations. 28 IL-1 and TNF also decrease the erythropoietin response to hypoxia, impede iron utilization, and impair marrow erythropoietic response. 26 Low hemoglobin levels in COPD patients could thus directly derive from COPD-related inflammation, and be aggravated by undernutrition when present. Dilution due to sodium retention 29 could lower hematocrit in COPD, but sodium retention is often associated with hypercapnia, 30 which does not fit with what we observed. Theophylline can blunt erythropoietin production. 31 Of note, oxygen could aggravate the negative effects of COPD on erythropoiesis by blunting erythropoietin production, which could counterbalance the positive effects of raising Pao 2. Another factor to take into account is the tobacco-smoking status and its changes overt time. In the present cohort, approximately 10% of the patients were still smokers at the initiation of LTOT, and possible changes in tobaccosmoking status during the follow-up are not known. Yet, cigarette smoking in COPD patients interferes with RBC production and with the effects of longterm oxygen on this production. 32 The Relationship Between Low Hematocrit and Mortality Anemia is associated with increased mortality in patients with end-stage renal insufficiency, 33 malignancy, 34 acute myocardial infarction and chronic heart failure, 27,35 surgical patients in general, 36 and critically ill patients. 37 In some cases, the causative nature of the association is suspected because correcting hemoglobinemia improves prognosis. 38 In heart failure patients, open studies 39 suggest that the correction of anemia with erythropoietin improves ejection fraction, reduces symptoms, decreases hospitalizations, and reduces mortality. These data however need confirmation. In other situations, correcting hemoglobinemia does not have a positive influence on the prognosis. 40 In COPD patients, Cappell and Nadler 41 observed increased mortality in cases of upper-gi bleeding as compared to control subjects (odds ratio, 4.3; 95% CI, 1.22 to 14.8; p 0.01). Upchurch et al, 12 in 158 COPD patients undergoing elective abdominal aortic aneurysmectomy, found than a low preoperative hematocrit was significantly associated with a poor outcome. From these observations, it may be postulated that COPD patients are rendered more sensitive to anemia, acute or chronic, by the prevailing tissue hypoxia. In the BODE index study, 13 hematocrit was significantly lower in the patients who died than in those who survived. That this variable was not selected by the statistical process in the BODE index may be surprising in view of our observations. However, the populations of patients are very different, ours being restricted to patients with very severe COPD needing LTOT. Conclusion and Perspectives Because of the limitations outlined above, this study cannot be more than a token of the putative clinical relevance of anemia in severe COPD patients. It is, however, an important addition to the existing anecdotal evidence pointing to this common sense possibility. In our view, the observations presented are a strong incentive for confirmatory prospective studies assessing the prognostic value of low hemoglobin levels in severe COPD patients, and in other forms of respiratory insufficiency. It would be useful to assess the effects of oxygen therapy on the production of erythropoietin. Finally, the question of the correction of hemoglobinemia in COPD should be raised and the corresponding pathophysiologic and prognostic effects assessed. Meanwhile, clinicians should be aware that anemia can be an issue in COPD patients; seeking and correcting associated factors such as GI bleeding or folate deficiency is probably particularly important in this setting. ACKNOWLEDGMENT: The authors thank Dr. D. Veale for editorial assistance and Ms. F. Binet for technical assistance. The authors acknowledge all the regional associations who provide data to the observatory: AIR Angers, AVD Angoulême, DDS Besançon, AVAD Bordeaux, AIR Caen, AIRRA Clermont-Ferrand, ALIZE Dijon, ADAIR Fouquières, AGIR Grenoble, GHAHR Le Havre, ARARR La Réunion, SANTELYS Respiration Lille, ALAIR Limoges, ARARD Marseille, APARD Montpellier, AIR Mulhouse, ARAIRLOR Nancy, ARIRPLO Nantes, CARDIF Paris, ARAIRCHAR Reims, AADAIRC Rochefort, ADIR Rouen, ADIRAL Strasbourg, SADIR Toulouse, and ARAIR Tours. References 1 Pauwels RA, Buist AS, Calverley PM, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary. Am J Respir Crit Care Med 2001; 163: Burrows B, Bloom JW, Traver GA, et al. The course and prognosis of different forms of chronic airways obstruction in a sample from the general population. N Engl J Med 1987; 317: CHEST / 128 / 3/ SEPTEMBER,

8 3 Traver GA, Cline MG, Burrows B. Predictors of mortality in chronic obstructive pulmonary disease: a 15-year follow-up study. Am Rev Respir Dis 1979; 119: Continuous or nocturnal oxygen therapy in hypoxemic chronic obstructive lung disease: a clinical trial; Nocturnal Oxygen Therapy Trial Group. Ann Intern Med 1980; 93: Long term domiciliary oxygen therapy in chronic hypoxic cor pulmonale complicating chronic bronchitis and emphysema: report of the Medical Research Council Working Party. Lancet 1981; 1: MacNee W, Wiggs B, Belzberg AS, et al. The effect of cigarette smoking on neutrophil kinetics in human lungs. N Engl J Med 1989; 321: Wouters EF. Chronic obstructive pulmonary disease: 5. Systemic effects of COPD. Thorax 2002; 57: Di Francia M, Barbier D, Mege JL, et al. Tumor necrosis factor- levels and weight loss in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1994; 150: Landbo C, Prescott E, Lange P, et al. Prognostic value of nutritional status in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1999; 160: Schonhofer B, Wenzel M, Geibel M, et al. Blood transfusion and lung function in chronically anemic patients with severe chronic obstructive pulmonary disease. Crit Care Med 1998; 26: Schonhofer B, Bohrer H, Kohler D. Blood transfusion facilitating difficult weaning from the ventilator. Anaesthesia 1998; 53: Upchurch GR Jr, Proctor MC, Henke PK, et al. Predictors of severe morbidity and death after elective abdominal aortic aneurysmectomy in patients with chronic obstructive pulmonary disease. J Vasc Surg 2003; 37: Celli BR, Cote CG, Marin JM, et al. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004; 350: Chailleux E, Fauroux B, Binet F, et al. Predictors of survival in patients receiving domiciliary oxygen therapy or mechanical ventilation: a 10-year analysis of ANTADIR observatory. Chest 1996; 109: Chailleux E, Laaban JP, Veale D. Prognostic value of nutritional depletion in patients with COPD treated by long-term oxygen therapy: data from the ANTADIR observatory. Chest 2003; 123: Standardized lung function testing. Official statement of the European Respiratory Society. Eur Respir J Suppl 1993; 16: World Health Organization. Nutritional anemias: report of a WHO scientific group. WHO Technical Report Series 405. Geneva, Switzerland: World Health Organization, 1968; Baldwin EF, Cournand A, Richards DW. Pulmonary insufficiency: III. A study of 122 cases of chronic pulmonary emphysema. Medicine 1949; 28: Wilson RH, Borden C, Ebert RV. Adaptation to anoxia in chronic pulmonary emphysema. Arch Intern Med 1951; 88: Vanier T, Dulfano J, Wu C, et al. Emphysema, hypoxia and the polycthemic response. N Engl J Med 1963; 269: Cocking JB, Darke CS. Blood volume studies in chronic obstructive non-specific lung disease. Thorax 1972; 27: Renzetti AD Jr, McClement JH, Litt BD. The Veterans Administration Cooperative Study of Pulmonary Function: 3. Mortality in relation to respiratory function in chronic obstructive pulmonary disease. Am J Med 1966; 41: Freedman M, Sutin D. Blood disorders and their management in old age. In: Brocklehurst s textbook of geriatric medicine and gerontology. 5th ed. New York, NY: Churchill Livingstone, 1998; Anthonisen NR, Wright EC, Hodgkin JE. Prognosis in chronic obstructive pulmonary disease. Am Rev Respir Dis 1986; 133: Wilson DO, Rogers RM, Wright EC, et al. Body weight in chronic obstructive pulmonary disease: The National Institutes of Health Intermittent Positive-Pressure Breathing Trial. Am Rev Respir Dis 1989; 139: Means RT Jr. Advances in the anemia of chronic disease. Int J Hematol 1999; 70: Al-Ahmad A, Rand WM, Manjunath G, et al. Reduced kidney function and anemia as risk factors for mortality in patients with left ventricular dysfunction. J Am Coll Cardiol 2001; 38: Chung KF. Cytokines in chronic obstructive pulmonary disease. Eur Respir J Suppl 2001; 34:50s 59s 29 de Leeuw PW, Dees A. Fluid homeostasis in chronic obstructive lung disease. Eur Respir J Suppl 2003; 46:33s 40s 30 Farber MO, Roberts LR, Weinberger MH, et al. Abnormalities of sodium and H 2 O handling in chronic obstructive lung disease. Arch Intern Med 1982; 142: Oren R, Beeri M, Hubert A, et al. Effect of theophylline on erythrocytosis in chronic obstructive pulmonary disease. Arch Intern Med 1997; 157: Calverley PM, Leggett RJ, McElderry L, et al. Cigarette smoking and secondary polycythemia in hypoxic cor pulmonale. Am Rev Respir Dis 1982; 125: Foley RN, Parfrey PS, Harnett JD, et al. The impact of anemia on cardiomyopathy, morbidity, and mortality in endstage renal disease. Am J Kidney Dis 1996; 28: Tammemagi CM, Neslund-Dudas C, Simoff M, et al. Impact of comorbidity on lung cancer survival. Int J Cancer 2003; 103: Al Falluji N, Lawrence-Nelson J, Kostis JB, et al. Effect of anemia on 1-year mortality in patients with acute myocardial infarction. Am Heart J 2002; 144: Carson JL, Duff A, Poses RM, et al. Effect of anaemia and cardiovascular disease on surgical mortality and morbidity. Lancet 1996; 348: Vincent JL, Baron JF, Reinhart K, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002; 288: Wu WC, Rathore SS, Wang Y, et al. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med 2001; 345: Silverberg DS, Wexler D, Sheps D, et al. The effect of correction of mild anemia in severe, resistant congestive heart failure using subcutaneous erythropoietin and intravenous iron: a randomized controlled study. J Am Coll Cardiol 2001; 37: Corwin HL, Gettinger A, Pearl RG, et al. The CRIT Study: anemia and blood transfusion in the critically ill; current clinical practice in the United States. Crit Care Med 2004; 32: Cappell MS, Nadler SC. Increased mortality of acute upper gastrointestinal bleeding in patients with chronic obstructive pulmonary disease: a case controlled, multiyear study of 53 consecutive patients. Dig Dis Sci 1995; 40: Clinical Investigations