Spirometric predictors for the exclusion of severe hypoxemia in chronic obstructive pulmonary disease

BACKGROUND: Controversy has existed over the need for routine arterial blood gas (ABG) analysis in patients with chronic obstructive pulmonary disease (COPD). Some authors recommend it in all patients with COPD, but others find it unnecessary if the forced expiratory volume in 1 s (FEV1) is 50% of predicted or greater. OBJECTIVES: To clarify this controversy, and to investigate correlations between severe hypoxemia and multiple spirometric parameters in patients with COPD with FEV1 50% of predicted or greater. PATIENTS AND METHODS: In 103 consecutive patients with COPD with FEV1 50% of predicted or greater, and without any other cardiopulmonary disorder, the incidence of severe hypoxemia (partial pressure of arterial oxygen less than 60 mmHg) was established by ABG analysis. Positive and negative predictive values (PPVs and NPVs, respectively) for severe hypoxemia for multiple spirometric parameters (FEV1, FEV1/forced vital capacity [FVC], peak expiratory flow [PEF], maximal midexpiratory flow rate [FEF25-75]) were evaluated in a stepwise manner. RESULTS: Twenty-two patients (21%) were found to be severely hypoxemic. In the severely hypoxemic group, the mean values for FEV1, FEV1/FVC, PEF and FEF25-75 were 59.0±8.19%, 53.6±11.3, 50.6±9.3 and 34.4±14.2% of predicted, respectively. The mean values for the same parameters in the other patients were 58.0±4.6%, 52.7±7.8, 51.5±7.5 and 39.1±7.7% of predicted, respectively. Comparing these parameters between the two groups, only the difference in FEF25-75 was statistically significant (P<0.01). Valid PPVs and NPVs could not be established for any of the parameters at any level, except for the NPV for FEF25-75 50% of predicted or greater, which was 92%. This threshold value resulted in a false negative finding in less than 5% of the patients with hypoxemia. CONCLUSIONS: The results of the present study showed that one in five patients with COPD with FEV1 50% of predicted or greater was severely hypoxemic. In such patients, hypoxemia may be excluded, and ABG analysis may not be needed when the FEF25-75 is also 50% of predicted or greater. The FEV1, FEV1/FVC and PEF parameters failed to predict or exclude severe hypoxemia.

C hronic obstructive pulmonary disease (COPD) is one of the leading causes of morbidity and mortality in the world, and currently affects approximately 15% of smokers in advanced age (1).The natural course of the disease is progressive and characterized by the limitation of expiratory air flow.
In patients with COPD, bronchodilators are the mainstay of treatment, unless chronic severe hypoxemia is present.Once the stage of severe hypoxemia ensues, continuous oxygen treatment becomes the principal therapeutic approach, assuming that the patient has already stopped smoking.
In patients with COPD, the American Thoracic Society (ATS) (2) recommends routine blood gas analysis only in patients with COPD with forced expiratory volume in 1 s (FEV 1 ) less than 50% of predicted.On the other hand, the European Respiratory Society (ERS) (3) recommends it in all patients without reference to FEV 1 measurements.Especially in North America, it is commonly held that patients with COPD with minor spirometric abnormalities are rarely hypoxemic, and that arterial blood gas (ABG) analysis will not provide any additional information; thus, it should not be performed unless there is other clinical evidence of hypoxemia.Defenders of the contrary idea believe that performing ABG analysis in such patients is important because some may be suffering from chronic severe hypoxemia; early recognition of these patients and treatment of their disease may prevent future complications, such as pulmonary hypertension and cor pulmonale.Although these different recommendations imply that different approaches are used for patients with COPD around the world, interestingly no study has been conducted to resolve the discrepancy.
In the present study, we aimed to investigate the incidence of severe hypoxemia, and to define correlations between severe hypoxemia and multiple spirometric parameters at different thresholds in patients with COPD with FEV 1 50% of predicted or greater, and who have no other cardiopulmonary disorder.

PATIENTS AND METHODS
One hundred forty-four patients with COPD were enrolled in the study who had been consecutively referred to the chest medicine clinics of Turgut Ozal Medical Center Research Hospital, Malatya, Turkey, between February and November 1999, for routine management of COPD or because of increased symptoms.The FEV 1 values of all patients were 50% of predicted or greater, and the previous diagnosis of COPD met the diagnostic criteria for COPD set by the ATS and ERS in 1995 (2,3).The study protocol was approved by the ethics committee of the Research Hospital.Patients using supplemental oxygen or those who had had an acute exacerbation in the previous month were excluded.
Standard spirometric evaluations for FEV 1 , FEV 1 /forced vital capacity (FVC), peak expiratory flow (PEF) and maximal midexpiratory flow rate (FEF 25-75 ) parameters were performed by a technician using Vmax 20c (SensorMedics, USA); blood samples for arterial gas components (partial pressure of arterial oxygen [PaO 2 ], partial pressure of arterial carbon dioxide [PaCO 2 ], pH, oxygen saturation [SaO 2 ]) were drawn from the radial artery while patients were in the supine position and breathing room air at rest.According to the PaO 2 values, the patients were divided into two groups: group 1 -patients with severe hypoxemia (PaO 2 less than 60 mmHg); group 2 -others (PaO 2 60 mmHg or greater).Other coexisting cardiopulmonary disorders that might cause hypoxemia were excluded by detailed radiologic and physical examination, medical history, 12-lead electrocardiography and echocardiography.The data from patients with hemoglobin levels lower than 100 g/L were not evaluated further.
The prediction equations were those of Quanjer (4).Data were expressed as percentage mean ± SD.In severely hypoxe- mic and other patients, the spirometric parameters and ABG values were compared using the Mann-Whitney U test.The correlation between appropriate variables was assessed using the Pearson correlation coefficient.P<0.05 was considered statistically significant.According to the results of spirometric testing, the patients were categorized first into intervals for each parameter, and then according to the number of severely hypoxemic patients in those intervals.Positive and negative predictive values (PPVs and NPVs, respectively) for severe hypoxemia were calculated for each parameter at different thresholds in a stepwise manner (Tables 1 and 2).Open forms of the equations that were used to calculate PPVs and NPVs in the present study are as follows: PPV = (number of true positive cases/total number of true positive and false positive cases) ´100% NPV = (number of true negative cases/total number of true negative and false negative cases) ´100% RESULTS One-hundred three patients completed the study (10 women, 93 men).An additional 41 patients with other coexisting cardiopulmonary disorders (n=37) and anemia (n=6) were excluded from the study after further investigations.Twentytwo patients were found to be severely hypoxemic (21%) (group 1).Eighty-one patients were either mildly hypoxemic (PaO 2 of 60 mmHg or greater, and less than 80 mmHg) or normoxemic (PaO 2 of 80 mmHg or greater) (group 2).The mean ages of the groups were 65.0±7.7 years and 63.7±6.4 years, respectively (P>0.05).
For groups 1 and 2, the mean values were as follows: for PaO 2 , they were 54.4±2.8 mmHg and 73.3±9.2 mmHg, Severe hypoxemia and its correlation with spirometry  Comparing these parameters between the groups, only the difference in FEF 25-75 was statistically significant (P<0.01).Correlation analysis revealed a positive correlation between PaO 2 and FEF 25-75 (r=0.22,P=0.024) (Figure 1).Regarding the distribution of severely hypoxemic patients in the described intervals, FEV 1 , FEV 1 /FVC and PEF distribution appeared almost homogenous (Tables 1 and 2).On the other hand, for the FEF 25-75 parameter, more than 95% of patients with severe hypoxemia were found to be below the threshold level of 50% of predicted.Any considerable PPV (the highest being 42% and missing 50% of patients with severe hypoxemia [Table 2]) could not be established for any parameter at any threshold level.The NPVs for FEF 25-75 50% of predicted or greater became 92%; FEF 25-75 values above that threshold showed false negative results in less than 5% of patients (one of 22) with severe hypoxemia.

DISCUSSION
The results of the present study show that one in five (21%) stable patients with COPD with FEV 1 50% of predicted or greater were severely hypoxemic.Although the FEV 1 , FEV 1 /FVC and PEF parameters did not show any statistical correlation with severe hypoxemia in these patients, the FEF 25-75 parameter was found to have quite a high NPV at the threshold level of 50% of predicted or greater (Figure 1).
The finding of such a high rate of severe hypoxemia in patients with relatively mild spirometric abnormalities is surprising, and its reasons need to be enlightened.Very well known mechanisms of hypoxemia in stable patients with COPD without any other cardiopulmonary disorder involve disturbances in the ventilation-perfusion ratio (2,5-7), decreased diffusing capacity (8,9), increased dead space (10,11), respiratory muscle fatique (12,13), deterioration of performance status (14), increased closing volume (15,16), fibrosis (2,(17)(18)(19), loss of alveolar capillary bed (2,6) and genetic variations in ventilatory drive (20,21).Spirometric measurements help to grossly assess the air flow obstruction, the severity of the disease and the arterial oxygenation.However, the measurements provide no direct information about the distribution, diffusion or perfusion status of the lung.Because there are many factors that are effective in the pathophysiology of hypoxemia in COPD as explained above, previous investigators could not show a direct correlation between the level of hypoxemia and spirometric measurements (22,23).In addition, the relative weight of contributing factors to the deterioration of PaO 2 probably varies from patient to patient.Further studies are needed in this area, and detailed analysis of such patients using carbon monoxide diffusing capacity of the lungs, thorax computed tomography, ventilation and/or perfusion scans, etcetera, may help to determine the underlying pathology leading to severe hypoxemia.
The FEV 1 and FEV 1 /FVC parameters are usually accepted as the classical monitoring parameters of COPD (19,22).Unless the FEV 1 and FEV 1 /FVC values are below 50% and 45% of predicted, respectively, severe hypoxemia is generally believed to occur rarely (2,23).For an FEV 1 less than 50% of predicted, the use of ABG analysis is routinely recommended by the leading thoracic societies.However, for patients with COPD with FEV 1 50% of predicted or greater, the ATS and ERS have completely different attitudes, probably due to the rather multifactorial nature of hypoxemia, the wide personal variations in its occurrence, the conflicting results coming from different studies and the attitudes of third parties for payment.The present study is the first to be conducted to clarify this discrepancy.
Although detailed data were given about the components of ABG analysis in the Results section, we will not discuss them.All of the statistically significant values between the mean values were originating from our intentional grouping necessary for the study design (severely hypoxemic patients and others), and they can be easily presupposed.
Among the spirometric parameters that we used, the mean values showed a statistically significant difference only in FEF 25-75 between the groups.FEV 1 measures the early effort-dependent portion and enough of the relatively effort-independent midportion of the expiratory manoeuvre.FEV 1 /FVC is the denominator of obstruction level and mostly measures the obstruction in large airways, and gives better results in mild and moderate obstruction levels.In addition, PEF measures only the most effort-dependent portion of the expiratory manoeuvre, namely the large airways.Because the present study only included the patients with FEV 1 50% of predicted or greater, this prerequisite probably, although not deliberately, led to the selection of patients with grossly equal obstruction levels and a large airway status.In another way, the importance of these spirometric tests describing mainly the large airway status in these patients with already relatively well conditioned large airways seems limited in the assessment of severe hypoxemia.However, FEF 25-75 provides data about the least effort-dependent portion of the expiratory manoeuvre, and is the denominator of small airways where the pathological process of COPD and obstruction are thought to start first.As a result, FEF 25-75 representing the condition of small airways may naturally be the better indicator of severe hypoxemia than FEV 1 , FEV 1 /FVC and PEF for the present study group, but that may not be the case for patients with COPD with FEV 1 less than 50% of predicted.
Confirming the common opinion, we also could not set any valuable PPV for severe hypoxemia with regard to spirometric parameters.However, for the first time, in the present study, it was found that the FEF 25-75 parameter had a strong NPV and a very low false negativity at the 50% of predicted or greater threshold (Table 2).Above this level of FEF 25-75 , performing ABG analysis seems unnecessary.The FEV 1 , FEV 1 /FVC and PEF parameters also had some relatively high NPVs (79% to 80%), but at those levels, they missed many or the majority of the patients with severe hypoxemia (Tables 1 and 2).
The monitoring of arterial oxygenation by pulse oximetry is another important point to be mentioned.It is a simple method that may tell us when to perform ABG analysis.The application of pulse oximetry is usually preferred during continuous monitoring of hospitalized patients with COPD to see sudden falls and rises in arterial oxygenation.Even in those cases, it is vital to obtain a complete baseline ABG measurement at least once at the beginning, and to evaluate pulse oximetry results accordingly thereafter.On the other hand, if we assume that the majority of the stable patients with COPD with FEV 1 50% of predicted had never undergone ABG analysis, regarding our results, the first ABG measurement becomes justifiable when FEF 25-75 is less than 50% of predicted.If arterial oxygenation is satisfactory, then their follow-up may be done with only pulse oximetry for a period of time, while watching expected disease progression and changes in spirometric parameters.
The early recognition of severe hypoxemia is vitally important in the management of patients with COPD.The timely initiation of supplemental or continuous oxygen treatment in these patients prolongs life, controls symptoms better and prevents complications.

CONCLUSIONS
Contrary to the general belief, we think that a considerable number of stable patients with COPD with FEV 1 50% of predicted or greater have severe hypoxemia, and unless the FEF 25-75 parameter is also 50% of predicted or greater, withholding ABG analysis necessitates strong clinical reasoning during such patients' clinical evaluation.

Figure 1 )
Figure 1) The relationship between partial pressure of arterial oxygen (PaO 2 ) and maximal midexpiratory flow rate (FEF 25-75 ) in patients with chronic obstructive pulmonary disease with a forced expiratory volume in 1 s 50% of predicted or greater (r=0.22,P=0.024)

TABLE 1 Number of severely hypoxemic and other patients at different intervals of forced expiratory volume in 1 s (FEV 1 ) and FEV 1 /forced vital capacity (FVC) parameters, and the positive and negative predictive values (PPVs and NPVs, respectively) at different thresholds Spirometric intervals* Number of severely hypoxemic patients
FEV 1 and FEV 1 /FVC intervals are given as the percentage predicted *