American Journal of Respiratory and Critical Care Medicine

Hospitalizations for acute exacerbation in patients with chronic obstructive pulmonary disease (COPD) have a great impact on health care expenditure. The aim of this study was to look at predictive factors of hospitalization for acute exacerbation in a group of patients with moderate to severe COPD. During the year 1994, we included 64 patients with COPD in this study. At inclusion, the patients being in a stable state, we performed a complete evaluation of their clinical, spirometric, gasometric, and pulmonary hemodynamic characteristics. All patients were followed during a period of at least 2.5 yr. We recorded the intervals free of hospitalization for exacerbation and realized an analysis of the proportional hazards not to be hospitalized using the Kaplan-Meier method. Univariate analysis using the log-rank test showed that the risk of being hospitalized was significantly increased in patients with COPD with a low body mass index (BMI ⩽ 20 kg/m2, p = 0.015) and in patients with a limited 6-min walk distance ( ⩽ 367 m, p = 0.045). But above all, the risk of hospitalization for acute exacerbation was significantly increased by gas exchange impairment and pulmonary hemodynamic worsening: PaO2 ⩽ 65 mm Hg versus PaO2 > 65 mm Hg, p = 0.005; PaCO2 > 44 mm Hg versus PaCO2 ⩽ 44 mm Hg, p = 0.005; and mean pulmonary artery pressure ( Ppa) at rest > 18 mm Hg versus Ppa ⩽ 18 mm Hg, p = 0.0008. Neither age, nor the association of one or more comorbidities with COPD, nor the smoking habits had a significant impact on the risk of hospitalization in our study. Multivariate analysis showed that only PaCO2 and Ppa were independently related to the risk of hospitalization for acute exacerbation of COPD. We conclude that chronic hypercapnic respiratory insufficiency and pulmonary hypertension are predictive factors of hospitalization for acute exacerbation in COPD patients.

Acute exacerbations of chronic obstructive pulmonary disease (COPD) are defined as a worsening of COPD symptoms caused by a rapid deterioration of the underlying respiratory function (1). Moderate to severe exacerbations represent a major cause of hospital admissions. Indications for hospitalization are the inadequate response to outpatient management, the inability to perform activities of daily living owing to increased dyspnea, the development of respiratory failure, the association of comorbidities, or inadequate home care resources (1). For these patients, admission to an intensive care unit (ICU) is common and average hospital lengths of stay are long, generating high costs (2). Some studies of hospitalization for acute exacerbation were devoted to establish predictors of outcome. Seneff and coworkers (3) reported in their large cohort study of patients with COPD admitted to ICU with acute exacerbation, a length of hospital stay of 25 ± 42 d (mean ± SD). For patients with an exacerbation of COPD and a PaCO2 > 50 mm Hg, Connors and coworkers (4) calculated a median cost of hospital stay of $7,400. Some studies evaluated the effect of treatments on the reduction of the frequency of hospital admissions. For example, hospital needs are reduced in patients with COPD receiving long-term oxygen therapy (5). More recently, chronic inhaled corticosteroids have been shown to reduce the number of hospitalizations for acute exacerbation (6). To our knowledge, only one study evaluated factors predicting hospitalization for acute exacerbation in patients with COPD (7). The basal body weight, the decline of FEV1, and the rate of deterioration of arterial blood gases were related to the necessity of ICU admission. But this study had numerous limitations because it was a retrospective analysis including a small number of patients who were all chronically hypercapnic.

The aim of the present study was to assess predictive factors of hospitalization for acute exacerbation in a cohort of patients with stable COPD.


Sixty-four consecutive patients were included in this study from January 1994 to December 1994. The patients were referred to our department for a complete investigation of moderate to severe COPD (n = 99). We used the American Thoracic Society (ATS) criteria to define COPD (1): a history of productive cough for 3 mo in each of two successive years and FEV1/vital capacity (VC) ratio of less than 60%, the total lung capacity (TLC) being more than 80% of the predicted value. Furthermore, all the patients were smokers or ex-smokers with a history of smoking equivalent to at least 20 pack-years. None of the patients had a history of atopy or a significant reversibility of airflow obstruction (> 15%) after inhalation of 400 μg salbutamol via a metered-dose inhaler. The patients were in a stable state, at distance (minimum 6 wk) from any acute exacerbation of the disease, i.e., symptoms of COPD, especially the grade of dyspnea, were unchanged, the gasometric values had not worsened compared with the best previous values, and pH was ⩾ 7.36. Seventy-five patients met these inclusion criteria.

We excluded 11 patients from the study because they presented an association of COPD and another respiratory disease which could potentially increase the risk of hospitalization for acute exacerbation. In patients with symptoms and a nocturnal oxymetry suggestive of sleep apnea, a conventional polysomnographic study was performed: six patients had an association of COPD and obstructive sleep apnea syndrome. Chest X-ray allowed detection of three patients with a lung cancer at the time of initial investigation and one patient with diffuse bronchiectasis. One patient with clinical signs of pulmonary embolism had a confirmatory lung perfusion scintigram and was also excluded. Patients with known stable comorbidities such as systemic hypertension, diabetes mellitus, or ischemic heart disease were not excluded.

Protocol of the Initial Evaluation

All patients underwent a complete investigation including clinical examination, spirography, arterial blood gas measurements, right heart catheterization, and computed tomography of the thorax. The degree of dyspnea was assessed by a graded scale from 0 to 5 (0 = no breathlessness; 1 = breathlessness on heavy exercise, e.g., climbing 2 or 3 floors; 2 = breathlessness on moderate exertion, e.g., climbing one floor or walking quickly; 3 = breathlessness on mild exertion, e.g., walking at normal speed; 4 = breathlessness on minimal exertion, e.g., slow walking; 5 = breathlessness on limited exertion, e.g., getting washed). A 6-min walk test was performed, the patient being familiarized with this procedure by at least one previous walk test.

Conventional spirography was performed with a 10-L closed-circuit water-sealed spirograph. Static volumes were measured by the closed-circuit helium dilution method. Reference values were those of the European Respiratory Society (8).

Our technique of right heart catheterization has been described previously (9). It is a minimally invasive procedure which is routinely performed in our laboratory after informed consent has been obtained from the patient. Briefly, the hemodynamic measurements were always done in the morning, without premedication, 2 h after a light breakfast, while the patient was in the supine position. We used small-diameter floated Grandjean catheters (4F; Plastimed, Saint-Leu-la-Forêt, France). The catheter was introduced percutaneously into an antecubital or a femoral vein. The pulmonary artery pressures and particularly the mean pulmonary artery pressure ( Ppa) were measured. Arterial blood samples were collected during heart catheterization, the patient breathing room air and measurements of PaO2 , PaCO2 , and pH were performed (model 280 Blood Gas System; Ciba Corning, Cergy Pontoise, France). Cardiac output was calculated according to Fick's equation applied to oxygen during steady-state conditions. Oxygen uptake was measured by an open-circuit system (Oxycon, Minjardt, The Netherlands). Measurements were made at rest and during a steady-state exercise. Exercise was realized on a bicycle-ergometer during at least 6 min with a load of 40 watts or less. The quantitative assessment of emphysema was made by computed tomography (CT). High-resolution CT scans (1 mm collimation) were obtained using a SOMATOM PLUS scanner during breath holding at functional residual capacity. CT scans were viewed by a window level of −600 /−800 Hounsfield units (HU) and a width of 1,000 /1,600 HU. A total of six slices was obtained from the lung apex to the lung base at the following levels: upper margin of the aortic arch, middle of the aortic arch, carina, origin of the middle lobe bronchus, right lower pulmonary veins, edge of the right diaphragmatic dome. The scoring system for the extent of emphysema was a visual one (10). Areas of low attenuation and vascular disruption were considered to be suggestive of emphysema. A score of 0 was given if there was no abnormality. If the emphysematous lesions occupied less than 25% of the lung field, the score was 1, between 25 and 50% the score was 2, between 50 and 75% the score was 3, and over 75% the score was 4. The final score was calculated as the mean percentage of the maximal possible score obtained by two independent readers (% CT). The reliability of our method was comparable to that reported in the literature for visual scores of emphysema. Indeed, the correlations of our CT score of emphysema with lung volumes and the carbon monoxide diffusing capacity were similar (results not shown) to those reported elsewhere (11).

Follow-up and Criteria for Hospitalization

After the initial evaluation, the patients were followed by quarterly visits. We recorded the time to first hospitalization for exacerbation or the date of the last control (September 1997). No patient was lost to follow-up. Our criteria of hospitalization were adapted from those of the expert consensus of the ATS (1): (1) patient has acute exacerbation characterized by increased dyspnea, cough, or sputum production, plus one or more of the following: inadequate response of symptoms to outpatient management consisting of an increase of the daily doses of inhaled bronchodilators, antibiotics, and a short course of oral steroids; inability to walk between rooms (patient previously mobile); inability to eat or sleep due to dyspnea; high-risk comorbid condition, pulmonary (e.g., pneumonia) or nonpulmonary; prolonged (> 15 d), progressive symptoms before emergency visit; altered mentation; worsening hypoxemia; new or worsening hypercarbia; (2) patient has new or worsening cor pulmonale unresponsive to outpatient management; (3) comorbid conditions have worsened pulmonary function. Noticeably, we excluded two conditions: (1) hospitalizations for preparation to or worsening after planned surgery or diagnostic procedures requiring the use of anesthetics; and (2) the absence of sufficient home care resources since all patients with moderate to severe COPD in France are insured by the national health service.

Statistical Methods

The outcome of interest was the time to first hospitalization for exacerbation. Clinical, functional, gasometric, pulmonary hemodynamic variables and the CT score were tested as predictors. Time intervals to first hospitalization for exacerbation of COPD were plotted using Kaplan-Meier estimates (12). Comparisons of predictor variables were performed by the means of the log-rank test. Subgroups for continuous variables were generally defined by dividing the whole group according to the median value. A Cox proportional-hazards model was constructed to determine independent predictors of hospitalization for acute exacerbation in patients with COPD (13). Results are expressed as means ± SD or medians with interquartile ranges (IQR) i.e., 25th to 75th percentiles.

The baseline characteristics of the patients at initial evaluation are summarized in Table 1. The majority of patients were male (84%). Thirty-four percent were current tobacco users and the smoking habits for the whole group were evaluated at 47 ± 24 pack-years. Thirty-six percent had one or more associated comorbidities: essential systemic hypertension, diabetes mellitus, or cardiovascular disease. Age at inclusion was 63.5 ± 9 yr (mean ± SD). Airways obstruction was moderate to severe because the mean FEV1 was 1.09 ± 0.59 L with an average reversibility of 6 ± 6%. According to COPD severity (1), 57% of the patients were in stage III (FEV1 < 35% predicted), 27% in stage II (FEV1, 35 to 49% predicted), and 16% in stage I (FEV1 ⩾ 50% predicted). The group as a whole showed moderate hypoxemia with a mean PaO2 of 66 ± 10 mm Hg and mild hypercapnia with a mean PaCO2 of 46 ± 8 mm Hg. Nevertheless, 10 of 64 patients had a PaO2 < 60 mm Hg and were treated with long-term home oxygen therapy (LTOT) for more than 16 h a day since 26 mo as a median time. Five further patients were treated by nocturnal oxygen therapy. No patient was treated by long-term home mechanical ventilation. Twenty-eight of 64 patients were chronically hypercapnic (PaCO2 > 45 mm Hg). Finally, 21 of 64 patients had pulmonary hypertension defined by a Ppa at rest > 20 mm Hg. Chronic medications included inhaled short-acting β2-agonists (82.5%), inhaled anticholinergics (67%), inhaled corticosteroids (20.4%), oral xanthine derivatives (62.5%), and at last, three patients were on prolonged oral corticosteroids.


nMean ± SDMedian (IQR)
Age, yr6463.5 ± 964 (57–72)
Smoking, pack-years6447 ± 2445 (30–55)
BMI, kg/m2 6425 ± 524 (21–28)
Dyspnea, grade/5643.4 ± 0.93 (3–4)
6-min walking distance, m54381 ± 126367 (292–492)
VC, ml632,915 ± 9102,785 (2,360–3,360)
VC, % pred6380 ± 2177 (66–91)
FEV1, ml631,085 ± 590925 (705–1,235)
FEV1, % pred6339 ± 2033 (25–46)
RV, ml633,290 ± 8903,330 (2,550–4,000)
RV, % pred63142 ± 36141 (113–166)
TLC, ml636,230 ± 1,3106,285 (5,420–6,890)
TLC, % pred63100 ± 16100 (89–110)
PaO2 , mm Hg6466 ± 1065 (60–73)
PaCO2 , mm Hg6446 ± 844 (40–49)
Ppa at rest, mm Hg6419 ± 718 (14–22)
Ppa at exercise, mm Hg6437 ± 1038 (28–43)
CT score of emphysema, %4633 ± 2530 (10–52)

Definition of abbreviations: BMI = body mass index; CT = computed tomography; IQR = interquartile range (25th–75th percentiles); PaO2 and PaCO2 = arterial gas tensions; Ppa = mean pulmonary artery pressure; RV = residual volume; VC = vital capacity.

The final follow-up checking was completed through September 1997 with a median length of follow-up of 30 mo. During the study period, 29 patients were admitted to the hospital for acute exacerbation (45%). All patients had increased dyspnea, cough, or sputum production and one or more of the criteria listed in Table 2, top. The status of the patients on the day of hospitalization is reported in Table 2, bottom. Most of these patients had severe hypoxemia and hypercarbia on admission. The average ratio of arterial oxygen tension to fraction of inspired oxygen (PaO2 /Fi O2 ) was 220 ± 58 mm Hg and the average PaCO2 was 61 ± 22 mm Hg. The pH was lower than 7.35 in 50% of the patients, suggesting acute respiratory failure. The median time to hospital admission for acute exacerbation was 29 mo (95% confidence interval [CI]: 24 to 33). Kaplan-Meier estimates for time to hospitalization are shown in Figure 1A. The results of univariate analysis are shown in Table 3. Neither older age nor smoking nor the association of comorbidities nor lung volumes (expressed as a percentage of the predicted values) were associated with an increased risk of hospital admission. On the contrary, low (< 20 kg/m2) body mass index (BMI), decreased PaO2 , increased PaCO2 , and increased Ppa were associated with significantly increasing rates of hospital admissions at 1 yr. Patients with dyspnea of grade 2 never experienced hospital admissions for acute exacerbation. On the contrary, the patients with a grade of dyspnea of 3, 4, or 5 had a similar risk of hospitalization at 1 yr. LTOT indeed significantly increased the risk of hospitalization for acute exacerbation. It must be emphazised that no significant changes occurred either in oxygen therapy or in chronic medications between the initial assessment and the final evaluation at first hospitalization or at last follow-up. Conversely, during the study period, six patients quit smoking: the number of smokers at hospitalization or last follow-up was significantly lower than the number of smokers at initial evaluation (p = 0.03).


Indications for Hospitalization (n = 29)n (%)
Inadequate response to outpatient management17 (59)
Severe dyspnea (inability to walk, sleep, or eat)15 (52)
High-risk comorbid condition 4* (14)
Prolonged symptoms (> 15 d) 6 (21)
Altered mentation 4 (14)
Worsening hypoxemia18 (62)
New or worsening hypercarbia11 (38)
New or worsening cor pulmonale11 (38)
Worsening pulmonary function due to comorbid conditions 3 (10)
1 or 2 of the above listed criteria10 (34)
3 or 4 of the above listed criteria13 (45)
5 or 6 of the above listed criteria 6 (21)
Physiologic status at hospitalizationnMean ± SDMedian (IQR)
APACHE II score27     20 ± 5 19 (15–23)
pH267.34 ± 0.097.34 (7.30–7.41)
PaO2 , mm Hg 2755 ± 1550 (45–63)
PaCO2 , mm Hg2761 ± 2255 (43–72)
PaO2 /Fi O2 , mm Hg27220 ± 58212 (187–276)

Definition of abbreviations: APACHE = acute physiology and chronic health evaluation; Fi O2 = fraction of inspired oxygen, Fi O2 = 20% + (4 × oxygen liter flow) in case of oxygen delivery by nasal prongs (1); IQR = interquartile range (25th–75th percentiles).

*Pneumonia in all four cases.

With or without oxygen therapy.


Predictor VariableSubgroups* % of Patients Free of Hospitalization for Exacerbation at 1 yr (cumulative proportion ± SE  )p Value
Age, yr⩾ 64 57 ± 9
< 64 76 ± 80.1
Tobacco useSmoking 64 ± 8
Not smoking 71 ± 80.99
Nonrespiratory comorbiditiesWith  61 ± 10
Without 70 ± 70.25
BMI, kg/m2 ⩽ 20  36 ± 14
20 < ⩽ 26 68 ± 8
> 26 84 ± 90.015
Dyspnea, grade/54 or 5 61 ± 9
2 or 3 74 ± 70.14
6-min walking distance, m⩽ 367  64 ± 10
> 367 81 ± 80.045
VC, % pred⩽ 77 57 ± 9
> 77 76 ± 80.5
FEV1, % pred⩽ 33 63 ± 9
> 33 71 ± 80.5
PaO2 , mm Hg⩽ 65 53 ± 9
> 65 81 ± 70.005
PaCO2 , mm Hg> 44 50 ± 9
⩽ 44 80 ± 70.005
Ppa at rest, mm Hg> 18 47 ± 9
⩽ 18 85 ± 60.0008
Ppa at exercise, mm Hg> 38  57 ± 10
⩽ 38 93 ± 50.0001
CT score of emphysema, %> 30  49 ± 10
⩽ 30 77 ± 90.025
Home oxygenotherapyWith38.5 ± 13
Without 77 ± 60.01

*In the case of continuous variables, the median was used to create two equal subgroups except BMI for which the sample was divided into three equal-sized subgroups.

The log-rank test was used to compare the hospital-free intervals of the different subgroups.

The results of the Cox proportional-hazards analysis of factors associated with hospitalization for acute exacerbation are shown in Table 4. We included the subsequent predictor variables as dichotomous variables in a stepwise analysis: BMI (< 24 versus ⩾ 24 kg/m2), percentage of predicted FEV1 (⩽ 33 versus > 33%), PaO2 (⩽ 65 versus > 65 mm Hg), PaCO2 (> 44 versus ⩽ 44 mm Hg), Ppa at rest (> 18 mm Hg versus ⩽ 18 mm Hg), and LTOT. As can be seen, only two variables were independent predictors of hospitalization, i.e., PaCO2 (relative risk [RR]: 2.1; CI: 1.4 to 3.1) and Ppa at rest (RR: 2; CI: 1.3 to 3.1). Similarly, when entering all predictor factors as continuous variables except LTOT, only two variables, PaCO2 and Ppa at rest, were independently associated with an increased risk of hospitalization. The predictive equation of the time (in days) to hospitalization was h(t) = h0 (t)e(0.0486 · PaCO2 + 0.0589 · Ppa ). The Kaplan-Meier estimates of time free of hospitalization for acute exacerbation in patients with COPD stratified for PaCO2 > or ⩽ 44 mm Hg and for Ppa at rest > or ⩽ 18 mm Hg are shown in Figures 1B and 1C. Comparisons between strata by the log-rank test were highly significant (p = 0.0054 and p = 0.0008, respectively).


Predictor VariableRelative Risk95% CIp Value
 > 44 mm Hg compared with ⩽ 44 mm Hg2.11.4–3.10.0003
Ppa at rest
 > 18 mm Hg compared with ⩽ 18 mm Hg21.3–3.10.0013

Definition of abbreviations: PaCO2 = arterial carbon dioxide tension; Ppa = mean pulmonary artery pressure.

The major finding of our study is that PaCO2 and Ppa are independent predictors of hospitalization for acute exacerbation in patients with moderate to severe COPD. The second important point is that markers of airflow obstruction, e.g. FEV1 and hypoxemia, were not retained as independent predictors of hospitalization in the Cox proportional-hazards model. This study is to our knowledge the first prospective one aiming to assess predictors of hospitalization for exacerbation in patients with COPD. Predicting a high risk of hospitalization for acute exacerbation is important for a number reasons: to identify patients in whom more aggressive therapeutic interventions should be performed, especially in long-term home care; and to identify factors that could lead to a specific treatment.

One important question is the potential relationships in patients with COPD between predictors of hospitalization, predictors of outcome after hospital admission and, more generally, predictors of survival. As pointed out by many studies, the best predictors of mortality in patients with COPD are the severity of airflow obstruction as indicated by the baseline postbronchodilator FEV1 and advancing age (14, 15). A Swedish study, published a few years ago, has demonstrated that, in patients under LTOT, FEV1 and performance status were the best predictors of survival in men whereas poor performance status and continuous oral corticotherapy were associated with poor survival in women (16). A recent study, published by our group, has shown that in patients with COPD receiving LTOT, Ppa and age were the only independent predictors of mortality whereas lung volumes and arterial blood gas levels had no prognostic value (17). It is noteworthy that the severity of airways obstruction as assessed by the percentage of predicted FEV1 or by the FEV1/VC ratio, was not a predictor of hospitalization for acute exacerbation, neither in univariate nor multivariate analysis. It is possible that our cohort was too small to detect any influence of the degree of airways obstruction on the risk of hospital admission. However, the dispersion of the percentage of predicted FEV1 was quite large (39 ± 20%) and comparable to that reported in other studies, as for example the large study (n = 985) of Anthonisen and colleagues (14), in which the mean percentage of predicted prebronchodilator FEV1 was 36.1 ± 11.4%. In the same way, the degree of hypoxemia was not predictive of hospitalization in our series of patients. Indeed, the range of PaO2 values was quite large from 45 mm Hg to 90 mm Hg although PaO2 less than 60 mm Hg was observed in only 15 of 64 patients. It must be emphasized that in the latter patients, the relevant PaO2 is no longer the PaO2 in ambient air conditions but the PaO2 corrected by oxygen therapy. Similarly, a lower age at the initial evaluation was not predictive of a longer hospitalization-free interval. Finally, the best predictors of mortality, i.e., the severity of airways obstruction, advancing age, and hypoxemia, seemed to be, in the present study, of little value as predictors of hospital admission for acute exacerbation.

PaCO2 appeared in our study as a good predictor of hospitalization. However, hypercapnia was moderate in most patients because the values of PaCO2 ranged from 32 mm Hg to 70 mm Hg in 13 patients with a PaCO2 higher than 50 mm Hg. Curiously, PaCO2 has been found a predictor of mortality in many studies but only in univariate analysis and particularly in patients with severe chronic respiratory insufficiency (18, 19). On the other hand, Seneff and coworkers (3) demonstrated that the PaCO2 at ICU admission of patients age 65 yr or older, with acute exacerbation of COPD, was a good predictor of mortality after hospital discharge. Indeed, patients with PaCO2 at ICU admission higher than 50 mm Hg had a 1-yr mortality of 70%, significantly higher than the 1-yr mortality of 54% of patients with an initial PaCO2 lower than 50 mm Hg. Vitacca and coworkers (7) showed that the rate of deterioration over a period of 2 yr in lung volumes and blood gas values was related to the necessity of ICU admission for acute exacerbation. Their study was retrospective and compared a group of 16 patients with chronic hypercapnic COPD admitted to an ICU for acute exacerbation with a matched group of 15 hypercapnic patients with COPD who had not needed hopitalization over the study period. In the present study, the only predictor variables that we have tested were indicators of the severity of illness at initial evaluation; no attempt was made to assess time-related variables such as the worsening of blood gases or the rate of decline of FEV1 or smoking cessation during the study period. However, moderate hypercapnia at baseline, suggestive of chronic alveolar hypoventilation, might identify patients who will easily decompensate their respiratory conditions and develop acute respiratory failure. Interestingly, Costello and coworkers have shown that many patients with COPD hospitalized with hypercapnia associated with an acute exacerbation revert to normocapnia during recovery. But the patients who remain hypercapnic after resolution of the acute episode have a poorer long-term prognosis than those with reversible hypercapnic respiratory failure (20).

The best independent predictor of hospitalization for acute exacerbation in our study was Ppa. In our series of 64 patients, 21 had pulmonary hypertension defined by Ppa at rest > 20 mm Hg though only 10 had PaO2 lower than 60 mm Hg. It is well known from studies of our group and others that pulmonary hypertension is likely to occur even in patients with moderate hypoxemic COPD, with PaO2 values higher than 60 mm Hg (21). We have demonstrated that Ppa and age were the best predictors of mortality in patients with severe COPD requiring oxygen therapy (17) but also one of the best predictors of mortality in patients not treated with LTOT (22). In our opinion, Ppa is a marker of the deleterious effects of alveolar hypoxia on the pulmonary circulation. Ppa might reflect the individual susceptibility to hypoxemia and the variability of the pulmonary vascular alterations induced by alveolar hypoxia. This hypothesis could explain why Ppa was finally an independent predictor of hospital admissions, even though the PaO2 had a very significant predictive value in univariate analysis.

It was not surprising that malnutrition significantly increased hospital needs, when assessed in univariate analysis, because it is also a predictor of poorer survival. Similarly, patients under LTOT are at risk for hospitalizations because they are the most severely ill patients compared with patients who did not require LTOT.

In view of our results, it can be hypothesized that long-term home mechanical ventilation, if it improves alveolar ventilation, might result in a significant reduction of the frequency of hospitalizations for acute exacerbations in patients with COPD. Furthermore, hospital admissions for acute exacerbation should be a major endpoint in the evaluation of long-term care modalities as well as survival. For example, Léger and associates recorded, in an uncontrolled study, a significant reduction in the number of hospitalizations in patients with severe COPD who were treated with nasal intermittent positive pressure ventilation (23). Home supervision programs also may reduce hospital admissions of patients with chronic hypercapnic COPD, as shown by Clini and colleagues (24), who performed comparisons with historical control groups. Oxygen therapy remains the treatment of choice of pulmonary hypertension in markedly hypoxemic patients, but normalization of Ppa is rarely obtained (25). In moderately hypoxemic patients (PaO2 of 60 to 70 mm Hg) with mild pulmonary hypertension, LTOT would be very debatable because it does not improve survival (26).

Our study has several limitations. Principally, the indications for hospitalization of patients with moderate to severe acute exacerbation of COPD derive from subjective interpretation of clinical features such as the worsening of symptoms like dyspnea, the presence of cor pulmonale, the response to outpatient management, the worsening of blood gases, and the presence of comorbidities. Even with the strict criteria that we have used in our study, the decision of hospitalization for acute exacerbation remains an event with some imprecision. Nevertheless, despite this approximation concerning the decision-making of hospital admission, the main characteristics of the patients who were hospitalized for acute exacerbation during the study period were very similar to those reported in other studies. For example, the blood gas values on admission were similar to those of the patients of the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment (SUPPORT) (4). In this cohort of 1,016 patients admitted with an exacerbation of COPD and PaCO2 of 50 mm Hg or more, the median pH was 7.36, the median PaCO2 was 56 mm Hg, and the median PaO2 /Fi O2 was 211 mm Hg. Second, some important factors, which probably could be good predictors of hospitalization for acute exacerbation, were not assessed in our study, e.g., the number of previous hopitalizations, indices of quality of life, and health status scores. However, in our hands, an indicator related to daily living activities such as the grade of dyspnea was only a significant predictor in univariate analysis and was not retained in the final multivariate model. And, at last, we have not assessed the role of COPD medication in preventing hospitalizations for acute exacerbation. For example, Paggiaro and colleagues have shown that chronic inhaled corticosteroids could reduce the severity of the exacerbations compared with a placebo (6). On the other hand, a short course of oral steroids in the outpatient treatment of an acute exacerbation accelerates the recovery of PaO2 and FEV1, reduces the treatment failure rate, and possibly lessens hospital admissions (27).

In summary, our study shows that baseline hypercapnia and pulmonary hypertension, even when moderate, are independent predictors of hospitalization for acute exacerbation in patients with COPD. Factors such as advancing age and severity of airways obstruction were not predictive of a higher risk of hospital admission. The 6-min walk distance, denutrition, and hypoxemia reduced the probability of hospitalization when these variables were used in univariate analysis.

1. ATS Statement. 1995. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease: inpatient management of COPD. Am. J. Respir. Crit. Care Med. 152(Suppl.):S97–S106, S78–S83.
2. Mushlin A. I., Black E. R., Connolly C. A., Buonaccorso K. M., Eberly S. W.The necessary length of hospital stay for chronic pulmonary disease. J.A.M.A.26619918083
3. Seneff M. G., Wagner D. P., Wagner R. P., Zimmermann J. E., Knaus W. A.Hospital and 1-year survival of patients admitted to intensive care units with acute exacerbation of chronic obstructive pulmonary disease. J.A.M.A.274199518521857
4. Connors A. F., Dawson N. V., Thomas C., Harrell F. E., Desbiens N., Fulkerson W. J., Kussin P., Bellamy P., Goldman L., Knaus W. A.for the SUPPORT investigatorsOutcomes following acute exacerbation of severe chronic obstructive lung disease. Am. J. Respir. Crit. Care Med.1541996959967
5. Petty, T. L. 1985. Long-term outpatient oxygen-therapy. In T. L. Petty, editor. Chronic Obstructive Pulmonary Disease. Marcel Dekker, New York. 375–388.
6. Paggiaro P. L., Dahle R., Bakran I., Frith L., Hollingworth K., Efthimiou J.on behalf of the International COPD Study GroupMulticentre randomised placebo-controlled trial of inhaled fluticasone propionate in patients with chronic obstructive pulmonary disease. Lancet3511998773780
7. Vitacca M., Foglio K., Scalvini S., Marangoni S., Quadri A., Ambrosino N.Time course of pulmonary function before admission into ICU: a two-year retrospective study of COLD patients with hypercapnia. Chest102199217371741
8. Quanjer, P. H., G. J. Tammeling, J. E. Cotes, O. F. Pedersen, R. Peslin, and J. C. Yernault. 1993. Lung volumes and forced expiratory flows. Eur. Respir. J. 6(Suppl. 16):15–40.
9. Weitzenblum E., Loiseau A., Hirth C., Mirhom R., Rasaholinjanahary J.Course of pulmonary hemodynamics in patients with chronic obstructive pulmonary disease. Chest751979656662
10. Goddard P. R., Nicholson E. M., Laszlo G., Watt I.Computed tomography in pulmonary emphysema. Clin. Radiol.331982379387
11. Eda S., Kubo K., Fujimoto K., Matsuzawa Y., Sekiguchi M., Sakai F.The relations between expiratory chest CT using helical CT and pulmonary function tests in emphysema. Am. J. Respir. Crit. Care Med.155199712901294
12. Kaplan E. L., Meier P.Nonparametric estimation from incomplete observations. J. Am. Stat. Assoc531958457481
13. Cox, D. R. 1972. Regression models and life tables. J. R. Stat. Soc. [B] 34:187–220.
14. Anthonisen N. R., Wright E. C., Hodgkin J. E.the IPPB Trial GroupPrognosis in chronic obstructive pulmonary disease. Am. Rev. Respir. Dis.13319861420
15. Hodgkin J. E.Prognosis in chronic obstructive pulmonary disease. Clin. Chest Med.111990555569
16. Ström K.Survival of patients with chronic obstructive disease receiving long-term oxygen therapy. Am. Rev. Respir. Dis.1471993585591
17. Oswald-Mammosser M., Weitzenblum E., Quoix E., Moser G., Chaouat A., Charpentier C., Kessler R.Prognostic factors in COPD patients receiving long-term oxygen therapy: importance of pulmonary artery pressure. Chest107199511931198
18. Report of the Medical Research Council Working PartyLong-term domiciliary oxygen therapy in chronic hypoxic cor pulmonale complicating chronic bronchitis and emphysema. Lancet11981681686
19. France A. J., Prescott R. J., Biernacki W., Muir A. L., Macnee W.Does right ventricular function predict survival in patients with chronic obstructive lung disease? Thorax431988621626
20. Costello R., Deegan P., Fitzpatrick M., McNicholas W. T.Reversible hypercapnia in chronic obstructive pulmonary disease: a distinct pattern of respiratory failure with a favorable prognosis. Am. J. Med.1031997239244
21. Chaouat A., Weitzenblum E., Kessler R., Charpentier C., Ehrhart M., Levi-Valensi P., Zielinski J., Delaunois L., Cornudella R., Moutinho dos Santos J.Sleep-related O2 desaturation and daytime pulmonary haemodynamics in COPD patients with mild hypoxaemia. Eur. Respir. J.10199717301735
22. Weitzenblum E., Hirth C., Ducoloné A., Mirhom R., Rasaholinjanahary J., Ehrhart M.Prognostic value of pulmonary artery pressure in chronic obstructive pulmonary disease. Thorax361981752758
23. Léger P., Bedicam J. M., Cornette A., Reybet-Degat O., Langevin B., Polu J. M., Jeannin L., Robert D.Nasal intermittent positive pressure ventilation: long-term follow-up in patients with severe chronic respiratory insufficiency. Chest1051994100105
24. Clini E., Vitacca M., Foglio K., Simoni P., Ambrosino N.Long-term home care programmes may reduce hospital admissions in COPD with chronic hypercapnia. Eur. Respir. J.9199616051610
25. Weitzenblum E., Sautegeau A., Ehrhart M., Mammosser M., Pelletier A.Long term oxygen therapy can reverse the progression of pulmonary hypertension in patients with chronic obstructive pulmonary disease. Am. Rev. Respir. Dis.1311995494498
26. Gorecka D., Gorzelak K., Sliwinski P., Tobiasz M., Zielinski J.Effect of long-term oxygen therapy on survival in patients with chronic obstructive pulmonary disease with moderate hypoxaemia. Thorax521997674679
27. Thompson W. H., Nielson C. P., Carvalho P., Charan N. B., Crowley J.Controlled trial of oral prednisone in outpatients with acute COPD exacerbation. Am. J. Respir. Crit. Care Med.1541996407412
Correspondence and requests for reprints should be addressed to Romain Kessler, Service de Pneumologie, Hôpital de Hautepierre, 67 200 Strasbourg, France.


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record