Pulse Oximetry Overestimates Oxygen Saturation in COPD Sridhar Amalakanti MD and Mohan Rao Pentakota MD BACKGROUND: Measurement of oxygen saturation with a handheld pulse oximeter is widely practiced as a surrogate to invasive arterial blood gas analysis. Oxygen saturation is an important parameter in cases of COPD, but there are insufficient data on the role of pulse oximetry in patients with COPD, moreso in diseases across its spectrum, such as chronic bronchitis and emphysema. We assessed the performance of pulse oximetry in acute respiratory failure of patients with COPD. METHODS: This was a cross-sectional, observational study. We studied 50 subjects with COPD admitted to the Government General Hospital, a 1,000-bed tertiary referral center in Guntur, India, from June 2013 to July 2013. Simultaneous reading of S by a handheld pulse oximeter and S pO2 aO2 by an automated arterial blood gas analyzer were taken. RESULTS: Pulse oximetry was sufficiently sensitive (84.60%) to hypoxemia in respiratory failure to be used in clinical situations. The mean difference (bias) between S and S was ؊3.98 (95% CI ؊4.68 to 3.28). There was less sensitivity aO2 pO2 (82% vs 85%) and positive predictive value (69% vs 85%) of the pulse oximeter to respiratory failure in subjects with chronic bronchitis versus emphysema. CONCLUSIONS: Pulse oximetry performed poorly in comparison with the invasive arterial blood gas analysis. The variability of the readings was greater in the subjects with chronic bronchitis than in those with emphysema. Key words: blood gas analysis; transcutaneous blood gas monitoring; oximetry; COPD; chronic bronchitis; emphysema. [Respir Care 2016;61(4):423–427. © 2016 Daedalus Enterprises] Introduction tient’s blood. Studies on the performance of pulse oxim- etry show that: (1) there is a very strong correlation COPD is a relentlessly progressive debilitating disease, (r ϭ 0.91) of S readings with S values1 but only aO2 pO2 punctuated by exacerbations of acute respiratory failure. moderate correlation between the degree of change; (2) the These episodes of acute respiratory failure can be diag- pulse oximeter overestimates actual changes in S 2,3; and aO2 nosed by measuring the oxygen saturation of hemoglobin (3) high S thresholds are necessary to detect significant pO2 in the blood. Obtaining accurate data regarding the oxy- hypoxemia with pulse oximetry.4 However, the data were genation saturation of hemoglobin in blood (S ) requires aO2 gathered from studies in subjects with shock and ARDS. invasive arterial blood gas (ABG) analysis. Noninvasive We performed a prospective observational study to assess measurement of arterial S is obtained by an instrument pO2 the performance of handheld pulse oximetry in predicting called a pulse oximeter. Pulse oximetry has become a hypoxemia in subjects with an exacerbation of COPD and standard practice in monitoring oxygen levels in the pa- to compare the efficacy of handheld pulse oximetry in the diagnosis of hypoxemia in chronic bronchitis and emphy- sema. Dr Amalakanti is affiliated with the Department of Neurology, and Dr Pentakota is affiliated with the Department of General Medicine, Guntur Medical College, Guntur, India. Methods The authors have disclosed no conflicts of interest. Fifty consecutive subjects diagnosed with COPD with Correspondence: Amalakanti Sridhar MD, Department of Neurology, acute respiratory failure attending the Department of Gen- Government General Hospital, Guntur 522002, India. E-mail: [email protected]. eral Medicine, Government General Hospital, a 1,000-bed tertiary hospital in Guntur, India, over a period of 1 month DOI: 10.4187/respcare.04435 were enrolled into the study. Patients with icterus, other RESPIRATORY CARE • APRIL 2016 VOL 61 NO 4 423 PULSE OXIMETRY OVERESTIMATES OXYGEN SATURATION IN COPD respiratory disorders, and other systemic disorders com- promising or altering respiratory function were excluded. QUICK LOOK The diagnosis of COPD was considered following the Current knowledge Global Initiative for Chronic Obstructive Lung Disease Handheld pulse oximeters are used in routine clinical (GOLD) 2013 guidelines.5 Subjects with dyspnea, chronic practice to measure oxygen saturation of the blood. cough, and/or sputum production and a history of exposure They are noninvasive and easy to use. However, studies to risk factors and a post-bronchodilator FEV1/FVC of Ͻ0.70 on spirometry were included in the study. in subjects with shock and ARDS show that these in- Subjects with a chronic productive cough for 3 months struments are reliable only at high blood oxygen satu- in each of 2 successive years in whom other causes of ration levels. productive chronic cough were excluded were diagnosed What this paper contributes to our knowledge with predominantly chronic bronchitis.6 Subjects were di- agnosed with predominantly emphysema based on estab- The handheld pulse oximeter was not reliable in diag- lished guidelines for clinical, radiographic, and spirometry nosing respiratory failure on its own. It overestimated analysis.7 the oxygen saturation of blood in subjects with COPD. Subjects were diagnosed with respiratory failure when The variability of the pulse oximeter readings was greater Ն2 of the following criteria8 were met: (1) acute respira- in subjects with chronic bronchitis than in those with tory distress, including severe dyspnea at rest and a breath- emphysema. ing frequency of Ͼ35 breaths/min and/or active contrac- tion of the accessory muscles of respiration or paradoxical abdominal motion; (2) hypercapnia (P Ͼ 50 mm Hg) aCO2 with respiratory acidosis (pH Ͻ7.33); and (3) hypoxemia Results (S Ͻ 90%). aO2 Pulse oximetry was used to measure S in these COPD During the study period, 78 patients were admitted with pO2 candidates using a Nellcor N-200 (Nellcor, Hayward, Cal- COPD. Fifty subjects (41 males and 9 females) were en- ifornia) co-oximeter finger probe in a sitting position. A rolled in the study. Those excluded had associated pneu- single pulse oximeter was used for all subjects. After 1 monitis (n ϭ 8), bronchiectasis (n ϭ 3), chronic kidney min of a stable S reading, a simultaneous ABG reading disease (n ϭ 6), and heart failure (n ϭ 11). The mean age pO2 was obtained from the radial artery for ABG analysis. One of the subjects was 63 Ϯ 13 y. Twenty-seven subjects single reading of S by pulse oximeter and its corre- were diagnosed with predominant chronic bronchitis, and pO2 sponding simultaneous S were obtained per subject. 23 were diagnosed with predominant emphysema. Table 1 aO2 ABG analyzer CO-oximetry measurements included oxy- shows the data summary. Bland-Altman analysis is shown hemoglobin saturation (S ), carboxyhemoglobin, P , in Figure 1 and Table 2 and shows consistent overestima- aO2 aO2 P , and pH (ABL-520 blood gas analyzer, Radiometer, tion (mean bias ϭϪ3.9) of arterial oxygen saturation by aCO2 Copenhagen, Denmark). The study protocol conformed to pulse oximetry. Positive predictive value is 88% for COPD, the ethical guidelines for biomedical research on human 85% for emphysema, and only 69% in chronic bronchitis participants of the Indian Council of Medical Research9 (Table 3). and was approved by the institutional ethical committee at Guntur Medical College. Discussion The mean age (63 y) distribution of the subjects reflects Statistical Analysis the progressive respiratory insufficiency in COPD, which debilitates patients late in its natural history. The prepon- Data were analyzed using Microsoft Excel 2007. The derance of male population could be attributed to the habit agreement between measurements of S and S in all of smoking and also to significant exposure to pollutants pO2 aO2 subjects was analyzed using a Bland-Altman plot, in which outside of the house, both of these characteristics being the difference between S and S was plotted against more common to men. The studied subjects have largely pO2 aO2 their average.10 Bias was calculated as the mean of the been laborers in whom the exposure to air pollution might differences between the ABG analyzer and pulse oximeter be contributory to their respiratory ailments. readings (S Ϫ S ). Positive bias indicated that the Considering the agreement limits (Ϫ4.68 to 3.28), the aO2 pO2 pulse oximeter underestimated the S , whereas negative handheld pulse oximeter performs poorly in acute respi- aO2 bias indicated that the pulse oximeter was overestimating ratory failure in COPD as a whole. However, it can be the S . The limits of agreement were taken as the bias Ϯ used to aid in pointing out the trend of rise or fall in blood aO2 (1.96 ϫ SD). oxygen saturation. 424 RESPIRATORY CARE • APRIL 2016 VOL 61 NO 4 PULSE OXIMETRY OVERESTIMATES OXYGEN SATURATION IN COPD Table 1. Blood Gas Measurements in the Subgroups of Subjects With COPD Characteristic COPD (N ϭ 50) Chronic Bronchitis (n ϭ 27) Emphysema (n ϭ 23) Hypoxia (n ϭ 23) Ϯ Ϯ Ϯ Ϯ Ϯ SaO2, mean SD % 83 14 83 15 83 14 71 12 Ϯ Ϯ Ϯ Ϯ Ϯ SpO2, mean SD % 87 10 87 10 88 10 79 11 Ϯ Ϯ Ϯ Ϯ Ϯ PaO2, mean SD mm Hg 63 18 65 18 61 18 51 15 Ϯ Ϯ Ϯ Ϯ Ϯ PaCO2, mean SD mm Hg 67 12 68 12 65 12 64 8 Fig. 1. Bland-Altman plots for agreements between arterial blood gas analysis and pulse oximeter measurements in subjects with COPD (A), subjects with chronic bronchitis (B), and subjects with emphysema (C). Horizontal lines denote the mean differences. Table 2. Bland-Altman Agreement Analysis This may in part be due to the different clinical situa- tions in which the studies were performed and the fact that Diagnosis Mean Difference (Bias) 95% CI the pulse oximeter interprets the ratio of oxyhemoglobin to COPD Ϫ3.9 Ϫ4.68 to 3.28 total hemoglobin in the pulsatile arterial vessels.17 There- Chronic bronchitis Ϫ3.0 Ϫ2.25 to 3.81 fore, factors altering the detection of pulsatility of the Emphysema Ϫ2.4 Ϫ1.85 to 3.09 arteries limit the accuracy of the S values.