Prove it: Titrated oxygen administration better for patients

Introduction
Medic 42 responds to a report of a man having difficulty breathing. The crew arrives to find the 69-year-old patient complaining of worsening dyspnea over the past several days. His wife says he has a ten-year history of emphysema and takes several medications including nebulized albuterol, ipratropium, and prednisolone. The patient denies chest pain but says he cannot catch his breath.

The patient is pale, cool to the touch, and sitting in a tripod position. First responders report an initial blood pressure of 156/98 mm Hg, pulse rate of 110 bpm, respiratory rate of 38 bpm, and a room-air pulse oximetry value of 89%. The patient has inspiratory and expiratory wheezes with pursed lip breathing. The first responders have the patient on high-flow oxygen by non-rebreather mask.

One medic applies a quantitative waveform capnometer under the mask and notes a reading of 47 mm Hg with a waveform that resembles a shark-fin. Another medic prepares and administers a mask with nebulized albuterol.

On the way to the hospital, the patient continues to breathe high flow oxygen through the mask and receives two additional albuterol doses mixed with ipratropium. The patient also receives a 125 mg dose of methylprednisolone. Although the wheezes have dissipated, both the patient’s oxygen saturation and end-tidal carbon dioxide (ETCO2) levels rise incrementally during the thirty-minute transport. On arrival at the emergency department (ED), the pulse oximetry value is 96% and the ETCO2 value is 55 mm Hg.

During the transfer of care, the ED physician orders the oxygen mask switched to a nasal cannula at 4 lpm and additional bronchodilators nebulized with room air instead of supplemental oxygen. He cautions the medics that high-flow oxygen administration to patients with a long history of chronic obstructive pulmonary disease (COPD) can promote hypercapnea, which could lead to respiratory failure.

On the way back to the station, the medic wonders if he should have used a nasal cannula and administered the medications with a hand-held nebulizer.

Review
Austin, M. A., Wills, K. E., Blizzard, L., Walters, E. H., & Wood-Baker, W. (2010). Effect of high flow oxygen on mortality in chronic obstructive pulmonary disease patients in prehospital setting: randomised controlled trial. British Medical Journal, 341, c5462. doi:10.1136/bmj.c5462

Researchers in the Australian State of Tasmania examined the effects on mortality that titrated oxygen had compared to high-flow oxygen administration in patients presumed by EMS to have a diagnosis of COPD.

Paramedics determined the field diagnosis of COPD based on the patient’s previous diagnosis, presenting symptoms, or the presence of risk factors, such as a smoking habit of greater than 10 packs per year. The study included all patients with a presumed diagnosis of COPD, age 35 years or older, and a complaint of breathlessness who were transported to the hospital.

Paramedics randomly placed the study patients into one of two groups. The titrated oxygen group received nasal cannula oxygen at a rate necessary to achieve a pulse oximetry reading between 88% and 92%. If the patient required an inhaled bronchodilator, medics used a nebulizer powered by compressed air instead of oxygen. The control group (high-flow oxygen group) received non-rebreather mask oxygen at 8-10 lpm regardless of the pulse oximeter reading. Nebulizers in this group ran on 100% oxygen at 6-8 lpm.

Regardless of the treatment group, initial nebulizer therapy involved salbutamol 5 mg diluted in 2.5 ml normal saline. Depending on the response, medics followed the initial bronchodilator with 0.5 mg ipratropium bromide, dexamethasone 8 mg IV, and if needed salbutamol 200-300 mg IV or 500 mg IM.

All participating medics attended a training session during which they received their treatment group assignment. If the assignment was titrated oxygen, the instruction was for the medic to provide titrated oxygen to every study patient. In cases where multiple participating medics took care of the same patient, oxygen delivery followed the assignment of the lead paramedic.

Paramedics transported all patients to a single receiving facility. The researchers instructed the ED staff to refrain from altering the prehospital oxygen administration rates until after obtaining arterial blood gas samples.

In addition to the primary data point (mortality), the researchers collected a number of secondary data points, including demographic information, prehospital treatment time, pretreatment oxygen saturation values, per cent predicted forced expiratory volume in one second, smoking history, length of hospital stay, and the need for invasive or non-invasive ventilation.

After completing the study and analyzing the data for the primary outcome in all patients, researchers performed a secondary analysis using a subgroup of patients with a confirmed diagnosis of COPD. Pulmonary medicine specialists confirmed the COPD diagnosis using pulmonary function testing and patient’s medical history.

One important concept in hypothesis testing involves a power calculation, which can serve two purposes. First, the calculation can determine how many people the researchers must enroll in order to detect a difference caused by the treatment, if a difference really exists.

Researchers who enroll too few patients may not detect a difference even if the treatment really is changing the outcome. On the other hand, enrolling too many patients does not significantly improve accuracy, is more expensive, and is more time consuming.

Alternatively, performing a power calculation with a predetermined number of patients provides the researcher with an estimate of the power or strength of the observed effect. Statistical power greater than 0.8 suggests enough power to detect change (Mazen, Magid, Hemmasi, & Lewis, 1985).

A previous study of oxygen administration in COPD patients demonstrated a 12% absolute reduction in mortality when healthcare personnel used titrated oxygen instead of high flow oxygen (Denniston, O’Brien, & Stableforth, 2002).

To detect a similar difference with a statistical power greater than 80%, the researchers estimated they would need to enroll 200 patients. Assuming a completely arbitrary dropout rate of 25%, the researchers would actually need to enroll 135 patients in each group in order to achieve adequate power.

During the planning stage of the study design, researchers set the criteria for statistical significance, or p value, at 0.05 — a common value in clinical research. The p value represents the probability of making an error in data interpretation, such as concluding there was a difference when no difference really existed. In this study, for example, if the calculated p value for mortality is less than the predefined value, there is less than a 5% probability that the difference is the result of chance.

Looking at it from a different point of view, there is a greater than 95% probability that the mortality difference is the result of the difference in oxygen administration rates. On the other hand, if the calculated p value is greater than 0.05, the mortality difference is not statistically significant and could be the result of chance alone.

Intent to treat analysis — all patients
The investigators in this study analyzed the primary outcome data using a research principle known as intent to treat. This principle requires that patients remain in their treatment group during the data analysis regardless of what treatment they actually receive (Sainani, 2010). In this study for example, every patient randomized to receive titrated oxygen stayed in that data analysis group regardless of whether the paramedic actually titrated the oxygen or not.

At face value, it may seem unfair to place patients who received high-flow oxygen into the titrated oxygen group for analysis, but the intention to treat approach actually serves a useful purpose.

Randomization, as was the intent of the study, attempts to distribute evenly any sources of error between the groups. In this trial, medics who originally began treating with titrated oxygen switched some patients to high flow oxygen because of a perceived distress. This type of arbitrary crossover obviously disrupts the balance provided by randomization and can skew the results.

Analyzing using the intent to treat principle attempts to restore that balance and prevent overestimation of effect for one group.

During the 13-month study period, 405 patients met the inclusion criteria. Medic assignment placed 226 patients in the high flow oxygen group and 179 in the titrated oxygen group. Baseline characteristics between the patients in each group did not differ statistically (Table 1).

Mortality in the high flow oxygen group was 9% (n = 21) versus 4% (n = 7) in the titrated oxygen group (relative risk 0.42, 95% CI: 0.20 to 0.89; p = 0.02). Since the p value is less than the predetermined value of 0.05, the difference in mortality for all patients appears to be the result of the difference in oxygen administration rates.

No deaths occurred in the prehospital environment. All deaths resulted from respiratory failure with 70% occurring within the first five days of hospitalization.

Subgroup analysis — intent to treat analysis — confirmed COPD
For the subgroup analysis, pulmonary specialists using lung function tests confirmed a COPD diagnosis in 214 patients. Baseline characteristics between the patients in each group did not differ statistically (Table 2).

Mortality in the high flow oxygen group was 9% (n = 11) versus 2% (n = 2) in the titrated oxygen group (relative risk 0.22, 95% CI: 0.05 to 0.91; P = 0.04). Since the p value is less than the predetermined value of 0.05, the difference in mortality for patients with a confirmed diagnosis of COPD appears to be the result of the difference in oxygen administration rates.

Although subgroup analysis can offer some insight into the examined problem, one must be careful about drawing conclusions based on a secondary analysis. The researchers designed this study to determine if titrated oxygen administration influences mortality in patients presumed by EMS to have COPD. Attempting to answer a second question with the same data can provide misleading results because the researchers did not design the study to answer that question.

What it means for you
Many healthcare care providers downplay the harmful effects that high-flow oxygen may have on their patients. Entrenched with the prehospital culture is a belief that oxygen administration saves lives and that if a little supplemental oxygen is good, then more oxygen must be better.

This trial is the first randomized comparison of different prehospital oxygen administration strategies for patients suspected of having COPD. Researchers demonstrated that in fact, oxygen administration titrated to pulse oximetry values reduces mortality by 78% compared to high flow oxygen administration.

Researchers and physicians question the value of high-concentration oxygen administration for a variety of patient conditions. Animal models of global brain ischemia following cardiac arrest demonstrate that administration of high oxygen concentrations interferes with normal brain recovery (Richards, Fiskum, Rosenthal, Hopkins, & McKenna, 2007; Richards, Rosenthal, Kristian, & Fiskum, 2006) and results in increased neuronal cell death (Vereczki, Martin, Rosenthal, Hof, Hoffman, & Fiskum, 2006).

A simple strategy of early lowering of oxygen administration rates based on pulse oximetry improved neurological outcome for the same condition (Balan, Fiskum, Hazelton, Cotto-Cumba, & Rosenthal, 2006). An experimental animal model mimicking the effects of global ischemia and hypoperfusion following stroke demonstrated fewer abnormal neurons and behavioral deficits with normoxic (21%) compared with hyperoxic (100%) resuscitation (Hazelton, Balan, Elmer, Kristian, Rosenthal, Krause, Sanderson, & Fiskum, 2010).

The American Heart Association (AHA) recently released new oxygen administration guidelines for patients suffering from acute cardiovascular problems. For cardiac arrest, the AHA acknowledges that although resuscitation teams have usually used 100% oxygen, there is no answer to the question about whether titrated oxygen is better for those patients (Neumar, Otto, Link, Kronick, Shuster, Callaway, Kudenchuk, Ornato, McNally, Silvers, Passman, White, Hess, Tang, Davis, Sinz, & Morrison, 2010).

They add that there is insufficient evidence to recommended routine oxygen administration to non-hypoxemic patients suffering from uncomplicated myocardial infarction (O’Connor, Brady, Brooks, Diercks, Egan, Ghaemmaghami, Menon, O’Neil, Travers, & Yannopoulos, 2010) or ischemic stroke (Jauch, Cucchiara, Adeoye, Meurer, Brice, Chan, Gentile, & Hazinski, 2010).

Other studies on the administration of high-flow oxygen to patients with uncomplicated myocardial infarction demonstrate either no effect (Wilson, & Channer, 1997) or a trend toward more serious side effects (Rawles, & Kenmure, 1976).

One limitation of this study is that researchers could not blind EMS personnel, nurses, or physicians to the treatment group. Blinding is a process whereby no one knows what treatment each patient is receiving. Blinding in this study would require a procedure whereby the medics could titrate oxygen administration without knowing they were doing so. For obvious reasons, this was not possible.

Without blinding, there is always a chance that the paramedics or emergency department personnel could unintentionally perform differently with one group, thereby creating an uncontrolled variable that could affect the outcome of the study. For example, suppose a medic believed that a COPD patient suffering from respiratory distress required high-flow oxygen for a positive outcome.

A medic assigned to provide titrated oxygen might be more aggressive with drug therapy to try to overcome any perceived treatment inadequacies. Even if the medic did not, nurses or physicians might alter management of the patient because of their bias for high-flow oxygen.

The authors report that they could not determine what effect (if any) prehospital or hospital treatment for COPD had on the primary outcome of mortality. This suggests that the research team may have looked at treatment beyond oxygen administration and assisted ventilation, but one cannot be sure with such a vague statement. The authors do not report specific data about early bronchodilator or corticosteroid administration, which could have introduced confounding variables.

Another issue that could alter the interpretation of the study results involves protocol compliance both in the prehospital and in the hospital setting. For the patients with a confirmed diagnosis of COPD, paramedics did not follow the study protocol in over half (56%) of the patients in the titrated oxygen group and about one-fifth (21%) of the patients in the high flow group. All of the protocol deviations in the titrated oxygen group involved paramedics administering high flow oxygen.

The hospital staff also failed to adhere to the study protocol. Although the research team informed the staff of the need for obtaining blood gas measurements early in the course of the ED visit, the staff did not obtain a sample in every patent. When they did, they only drew samples within the first thirty minutes in 11% of the patients. In fact, the mean time from arrival in the ED until the staff obtained the blood gas sample was over an hour and a half.

Although this study is far from perfect, it does provide insight into the potential harm that oxygen therapy may play in the prehospital management of patients suffering from respiratory distress associated with COPD. One in-hospital study using oxygen flow rates greater than 4 lpm in hypercapnic COPD demonstrated increased length of stay, greater need for non-invasive ventilation, and severe acidosis (Joosten, Koh, Bu, Smallwood, & Irving, 2007).

High-concentration oxygen administration rates may promote carbon dioxide retention through ventilation-perfusion mismatch secondary to hypoxic vasoconstriction (Aubier, Murciano, Fournier, Milic-Emili, Pariente, & Derenne, 1980; Sassoon, Hassell, & Mahutte, 1987; Robinson, Frieberg, Regnis, & Young, 2000) and ventilation reduction due to removal of the normal hypoxic stimulus (Robinson, Frieberg, Regnis, & Young, 2000; Dunn, Nelson, & Hubmayr, 1991).

EMS personnel may be administering oxygen based on a perceived need rather than the actual presence of hypoxemia. To maximize effectiveness and minimize harmful side effects, medics should probably assess for the presence of hypoxemia by establishing baseline pulse oximetry readings before deciding on a course of oxygen administration in patients suffering from COPD. In this study, medics achieved the best results and reduced mortality by titrating oxygen administration rates to a pulse oximetry value of 88%-92%.

References
Aubier, M., Murciano, D., Fournier, M., Milic-Emili, J., Pariente, R., & Derenne, J. P. (1980). Central respiratory drive in acute respiratory failure of patients with chronic obstructive pulmonary disease [abstract]. American Review of Respiratory Disease, 122(2), 191-199.

Balan, I. S., Fiskum, G., Hazelton, J., Cotto-Cumba, C., & Rosenthal, R. E. (2006). Oximetry-guided reoxygenation improves neurological outcome after experimental cardiac arrest. Stroke, 37, 3008-3013. doi: 10.1161/01.STR.0000248455.73785.b1

Denniston, A. K., O’Brien, C., & Stableforth, D. (2002). The use of oxygen in acute exacerbations of chronic obstructive pulmonary disease: a prospective audit of pre-hospital and hospital emergency management. Clinical Medicine, 2(5), 449-451.

Dunn, W. F., Nelson, S. B., & Hubmayr, R. D. (1991). Oxygen-induced hypercarbia in obstructive pulmonary disease [abstract]. American Review of Respiratory Diseases, 144(3 Pt 1), 526-530.

Hazelton, J. L., Balan, I., Elmer, G. I., Kristian, T., Rosenthal, R. E., Krause, G., Sanderson, T. H., & Fiskum, G. (2010). Hyperoxic reperfusion after global cerebral ischemia promotes inflammation and long-term hippocampal neuronal death. Journal of Neurotrauma, 27, 753-762. doi: 10.1089=neu.2009.1186

Jauch, E. C., Cucchiara, B., Adeoye, O., Meurer, W., Brice, J., Chan, Y. Y., Gentile, N., & Hazinski, M. F. (2010). Part 11: Adult stroke: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation, 122(18 Suppl 3), S818-S828. doi: 10.1161/CIRCULATIONAHA.110.971044

Joosten, S. A., Koh, M. S., Bu, X., Smallwood, D., & Irving, L. B. (2007). The effects of oxygen therapy in patients presenting to an emergency department with exacerbation of chronic obstructive pulmonary disease. Medical Journal of Australia, 186(5), 235-238.

Mazen, A., Magid, M., Hemmasi, M., & Lewis, M. F. (1985). In search of power: A statistical power analysis of contemporary research in strategic management. Academy of Management Proceedings, 30(2), 30-34.

Neumar, R. W., Otto, C. W., Link, M. S., Kronick, S. L., Shuster, M., Callaway, C. W., Kudenchuk, P. J., Ornato, J. P., McNally, B., Silvers, S. M., Passman, R. D., White, R. D., Hess, E. P., Tang, W., Davis, D., Sinz, E., & Morrison, L. (2010). Part 8: Adult advanced cardiovascular life support. 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation, 122(18 Suppl 3), S729-S767. doi: 10.1161/CIRCULATIONAHA.110.970988

O’Connor, R. E., Brady, W., Brooks, S. C., Diercks, D., Egan, J., Ghaemmaghami, C., Menon, V., O’Neil, B. J., Travers, A. H., & Yannopoulos, D. (2010). Part 10: Acute coronary syndromes: 2010 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation, 122(18 Suppl 3), S787-S817. doi: 10.1161/CIRCULATIONAHA.110.971028

Rawles, J. M., & Kenmure, A. C. (1976). Controlled trial of oxygen in uncomplicated myocardial infarction. British Medical Journal, 1(6018), 1121–1123.

Richards, E. M., Fiskum, G., Rosenthal, R. E., Hopkins, I., & McKenna, M. C. (2007). Hyperoxic reperfusion after global ischemia decreases hippocampal energy metabolism. Stroke, 38, 1578-1584. doi: 10.1161/STROKEAHA.106.473967

Richards, E. M., Rosenthal, R. E., Kristian, T., & Fiskum, G. (2006). Postischemic hyperoxia reduces hippocampal pyruvate dehydrogenase activity. Free Radical Biology and Medicine, 40, 1960–1970. doi:10.1016/j.freeradbiomed.2006.01.022

Robinson, T. D., Frieberg, D. B., Regnis, J. A., & Young, I. H. (2000). The role of hypoventilation and ventilation-perfusion redistribution in oxygen-induced hypercapnia during acute exacerbations of chronic obstructive pulmonary disease. American Journal of Respiratory and Critical Care Medicine 2000, 161(5), 1524-1529.

Sainani, K. L. (2010). Making sense of intention-to-treat. Physical Medicine and Rehabilitation, 2, 209-213. doi: 10.1016/j.pmrj.2010.01.004

Sassoon, C. S., Hassell, K. T., & Mahutte, C. K. (1987). Hyperoxic-induced hypercapnia in stable chronic obstructive pulmonary disease [abstract]. American Review of Respiratory Diseases, 135(4), 907-911.

Vereczki, V., Martin, E., Rosenthal, R. E., Hof, P. R., Hoffman, G. E., & Fiskum, G. (2006). Normoxic resuscitation after cardiac arrest protects against hippocampal oxidative stress, metabolic dysfunction, and neuronal death. Journal of Cerebral Blood Flow and Metabolism, 26, 821-835. doi:10.1038/sj.jcbfm.9600234

Wilson, A. T., & Channer, K. S. (1997). Hypoxaemia and supplemental oxygen therapy in the first 24 hours after myocardial infarction: the role of pulse oximetry. Journal of the Royal College of Physicians of London, 31, 657– 661.