Hypoxemia is a major problem in the immediate postoperative period. The need for postoperative oxygen therapy after surgery involving an incision to the thorax or upper abdomen is well known . However, Canet et al.  found that oxygen saturation increased in patients who had a peripheral site for surgery while they breathed room air during their first 10 min in the postoperative anesthesia care unit (PACU) . Thus the need for oxygen therapy in those having surgery to the face, neck, lower abdomen, or extremities, is open to question, especially in the present environment of cost containment.
In the absence of an adequate noninvasive monitorinng device, it was considered wise practice to administer oxygen therapy to all postoperative patients. But pulse oximetry is now a standard of care in anesthe siology and the immediate postoperative period [3,4]. It allows the noninvasive monitoring of oxygen saturation to indicate the need for oxygen therapy. The routine use of oxygen therapy in patients at low risk for hypoxemia who have adequate oxygen saturation may be an unwarranted expense. Clinical practice for use of oxygen varies among institutions; some require oxygen therapy for this group of patients and others do not. Of those who do use oxygen therapy, both high- and low-flow devices are used, some with high humidity and others with no humidity. Nurses in the PACU often coach patients to take deep breaths but the value of this technique in maintaining oxygen saturation has not been established.
The purpose of the present study was to compare oxygen saturation levels in selected surgical patients in the immediate postoperative period when assigned one of four oxygenation regimens. In addition, the ability of a humidification device to prevent patients from having a dry mouth, nose, or throat in the immediate postoperative period was evaluated.
This study received human subjects approval from the University of Maryland Institutional Review Board. An experimental design was used with subjects randomly assigned to one of four treatment groups using a Table ofrandom numbers.
The sample consisted of 293 adult postoperative patients admitted to the PACU in a large metropolitan teaching hospital. Those having surgery to their thorax or upper abdomen, or having neurologic surgery were excluded from the study. Also excluded were patients with preoperative hypoxemia (as evidenced by an oxygen saturation level less than 90%), documented moderate to severe restrictive or obstructive pulmonary disease, or factors interfering with proper use of the pulse oximeter. This includes those whose medical records indicate peripheral vascular disease, severe anemia, or recent use of colored dyes.
Eleven subjects were removed from the study protocol when their saturation decreased below 90% and they required more rigorous treatment. This left 282 subjects for data analysis.
In Group 1, a nasal cannula was applied to the patient immediately after the baseline oxygen saturation was recorded. The oxygen flow was set at 4 L/min with humidification not used. The cannula remained in place until the patient was discharged from the unit. Lung expansion techniques were not used with these patients. In Group 2, an oxygen face tent was connected to a nebulizer and applied to the patient with an oxygen flow of 10 L/min. The face tent remained in place until the patient was discharged from the unit. Lung expansion techniques were not used with these patients. In Group 3, lung expansion techniques were used. After routine patient assessment, the nurse instructed the patient to take a series of five slow deep breaths. Subjects in this group were also encouraged to do this on their own when possible. Every 15 min, patients were again coached to perform the five deep breaths and encouraged to perform them on their own. No supplemental oxygen was administered to this group. Group 4 had no oxygen enhancing regimen. Patients received the standard care in the PACU. Oxygen enhancing techniques or supplemental oxygen were not used.
Oxygen saturation was measured using the Criticare Pulse Oximeter (Criticare Systems, Inc., Milwaukee, WI), a noninvasive measure of the percent of saturated hemoglobin in the arterial blood. The machine is activated by the pulse wave to maximize the oxygen saturation reading. Oxygen saturation and pulse were displayed digitally. The accuracy of the pulse oximeter is reported by the manufacturer to be +/- 2% saturation at ranges from 70% to 100% saturation. Confirmation of machine accuracy was achieved by comparing pulse oximeter ratings of oxygen saturation on 10 patients with simultaneous arterial blood gas analysis. The pulse oximeter and blood gas readings varied by less than 2% saturation.
Symptoms of dry mouth, nose, or throat as well as nausea were recorded as being either present or absent as per patient report. Episodes of vomiting were also recorded.
All patients were transported from the operating room to the PACU in a semi-Fowler's position receiving oxygen via nasal cannula. Upon arrival in the PACU, patients were placed in the semi-Fowler's position per unit protocol . The Criticare Pulse Oximeter was applied. Each patient's level of consciousness was then assessed using the postanesthetic recovery score (PARS)  and the axillary temperature was taken. Subjects were randomly assigned to one of four treatment groups. Demographic and medical information was collected from the patient's record. Oxygen saturation readings were noted and recorded at the time of admission to the PACU and every 15 min until the subject was discharged from the PACU. If oxygen saturation decreased to less than 90% at any time during the study, the patient was removed from the protocol. Every 15 min subjects were asked about symptoms.
Descriptive statistics were calculated for oxygen saturations upon admission to the PACU, after 15 min, and after 30 min. Repeated measure analysis of variance (ANOVA) comparing oxygen saturations over the three time periods and across the four groups were performed and Tukey test used for post hoc analysis. The number of symptoms reported by each group was compared using chi squared analysis.
The 282 subjects in the study consisted of 178 females and 104 males ranging in age from 17 to 89 yr with a mean age of 46 yr. Subjects consisted of different races with 157 (56%) African Americans, 116 (41%) Caucasians, and 9 (3%) Asians or other races. Surgeries performed included gynecologic surgery, such as vaginal hysterectomy, tubal ligation, or dilation and curettage (n = 103, 37%); orthopedic surgery, such as knee replacement, or spinal surgery (n = 78, 28%); tonsillectomy or opthamologic surgery (n = 40, 14%); dermatology or peripheral vascular surgery (n = 51, 18%); or other surgery (n = 10, 3%). The mode of anesthesia received by most patients was general anesthesia (n = 164, 58%); 81 (29%) had regional anesthesia, and 37 (13%) had intravenous sedation. The ASA physical status score of the patients varied with 82 (29%) having a score of I, 137 (50%) having a score of II, and 59 (21%) having a score of III. No ASA physical status IV or V patients were studied. The PARS of the subjects on admission to the PACU varied from 0 to 10 with 15 (5%) being <or=to7 and 267 (95%) being >or=to8. Temperatures upon admission to the PACU ranged from 34.7 degrees to 36 degrees C with a mean of 37.7 degrees.
All demographic, health history variables, and assessments at the time of admission to the PACU were matched in each group. Of note, age was found to be significantly higher in Group 1 (mean = 51 yr old) when compared to Group 2 (mean = 41 yr old) (F = 3.75, P < 0.05). Group 3 had a mean age of 43 and Group 4, 47 yr.
At the time of admission to the PACU, oxygen saturation levels ranged from 90% to 99% with means across the four groups calculated as ranging from 96.7% to 97.4%. The mean of the total sample was 97% (SD = 1.9) Table 1. Of the 293 subjects admitted to the study 11 (4%) had saturations that decreased to less than 90%, two from those receiving O2 via nasal cannula, two receiving 40% O2 by face tent, four in the deep breathing group, and three in the group receiving no oxygen enhancing regimen. Using chi squared analysis, these frequencies were not found to be significantly different. Ten of the 11 patients who had saturation levels that decreased to less than 90% had oxygen saturation levels at the time of admission to the PACU between 92% and 90% whereas only eight of the remaining 282 subjects who did not desaturate had oxygen saturations this low at the time of admission to the PACU. No other demographic or medical characteristics, including the PARS level at time of admission to the PACU, was found to be significantly different between those who did and did not desaturate.
The remaining 282 subjects all stayed in the PACU at least 30 min allowing three data collections (at admission, at 15 min, and at 30 min) to be completed and entered into the analysis. A simple ANOVA revealed no differences in oxygen saturation across the four groups at the time of admission to the PACU. Using a two-way ANOVA to compare oxygen saturation values among the four groups over the three time intervals, significant differences were found across the groups (F = 10.33, df = 3,278, P < 0.001), across the times (F = 49.27, df = 2,556, P < 0.001) and with an interaction between groups and times (F = 19.51, df = 6,556, P < 0.001). After transforming the data to meet the assumptions of the statistical tests, the results of the analysis remained highly significant. To determine the location of the differences, a one-way ANOVA was run separately for each time period. At the time of admission, as mentioned previously, there were no differences between the groups. At 15 min, there were significant differences (F = 29.66, P < 0.01) which a Tukey test determined to be between Groups 1 and 2 compared to Groups 3 and 4. At Time 3 there were significant differences (F = 35.27, P < 0.01) which a Tukey test again determined to be between Groups 1 and 2 compared to Groups 3 and 4. The interaction occurs between Time 1 compared to Times 2 and 3 Table 1.
Few subjects reported having symptoms of nausea, dryness of nose, mouth, or throat, or vomiting. Two hundred thirty-six subjects (85%) reported no symptoms throughout the study period. There were 22 reports of dryness with chi squared analysis (chi squared = 10.37, df = 3, P < 0.05) revealing a significant difference across groups Table 2. Twenty-four reports of nausea were noted with chi squared analysis (chi squared = 8.32, df = 3, P < 0.05) revealing a significant difference across groups Table 2. Nine subjects vomited with no significant differences across the four groups.
Of the 72 subjects in the group receiving 40% oxygen by face tent, 10 complained of the mask being uncomfortable. This is in contrast to only two of the 69 subjects receiving oxygen by nasal cannula complaining of discomfort due to the cannula.
The mean oxygen saturation of 97% observed in this study at the time of arrival in the PACU is higher than oxygen saturations reported by others in the immediate postoperative period. Canet et al.  reported mean oxygen saturations of 90.4% to 93.9% upon arrival in the PACU but their sample included those having abdominal surgery in whom oxygen saturation levels were found to be lower. There is also no indication in that study of subjects receiving oxygen during transfer from the operating room (OR). All subjects in our group received oxygen via nasal cannula during transfer from the OR to the PACU as is the routine in our institution. Also, entry into our study required an initial oxygen saturation at 90% or above.
In a well designed study of 95 ASA grade I or II adult patients breathing room air during transfer from the OR to the PACU, Tyler et al.  found the changes in oxygen saturation during this early postoperative period to occur within the first 3 min. Hypoxemia, defined as a SaO2 at or below 90%, occurred in 35% of the patients having elective surgery under general anesthesia, with 12% having severe hypoxemia (SaO2 at or below 85%) during transfer. In contrast to their findings, the percentage of subjects experiencing hypoxemia in our study was only 11 of 293 or 4% of the subjects. This is also in contrast to hypoxemia rates of 21% found by Hudes et al. , and 43.8% found by Canet et al. . It must be remembered that the present study was limited to surgical patients at low risk for hypoxemia with oxygen administered during transfer to the PACU.
The O2 saturation level at the time of admission to the PACU is important in predicting those most at risk for developing hypoxemia. Of those who became hypoxemic after admission to the PACU, 10 of the 11 had oxygen saturation levels between 92% and 90% when admitted to the PACU. Only eight of the remaining 282 who did not desaturate had oxygen saturations at the time of admission to the PACU of 92% or less. This supports the recommendation of Berels and Marz  that supplemental oxygen be used for PACU patients with SaO2 values below 92%.
Although age and ASA physical status score has been found by others to be predictive of surgical patients who will have a hypoxemic episode [10,11], no such demographic or medical characteristics could be found in this study to distinguish the subjects who desaturated from those who did not.
Our findings are different from those of Canet et al.  who are the only other researchers known to report O2 saturations at more than one time period in the PACU. They found that oxygen saturation values increased in subjects breathing room air in the PACU. In our study oxygen saturation values changed very little for those breathing room air while in the PACU.
Although we found statistically significant differences in oxygen saturation in those who received oxygen compared to those who did not, these differences are between a maximum oxygen saturation of 98.7% and a minimum O2 saturation of 96.8%. One must question the clinical significance of these findings. It does not seem cost effective in this time of diminishing resources to use oxygen-enhancing therapy to increase the oxygen saturation from one normal level to another. Also, there is a ceiling effect for the therapy as oxygen saturation achieves a maximum at 100%. It would seem advisable to use the noninvasive technology now available to determine those at most risk for hypoxemia and to administer oxygen therapy only to those who need it.
Hedstrand et al.  found that lung expansion techniques increase arterial oxygen tension significantly in postoperative abdominal surgical patients. Although this is a very low cost technique, our findings did not support its benefit. Patients coached to take deep breaths did not have higher O2 saturation levels than those receiving no O2 enhancing techniques.
When oxygen therapy is needed, nasal cannula has been recommended as the most cost effective device for oxygen delivery in conscious patients. Using 20 normal subjects to compare six techniques for oxygen delivery, Kory et al.  found oxygen saturations to be slightly lower when the nasal cannula was used. However, the greater ease in administration as well as greater patient comfort led them to recommend the nasal cannula for oxygen therapy over the more costly devices. Our results are similar to those reported by Hudes et al.  who studied 101 elective surgical postoperative patients and found no differences in oxygen saturations between patients given oxygen by nasal cannula compared to those given 40% oxygen by face mask.
Few of our subjects experienced dryness demonstrating no benefit in the use of the high-flow, high-humidity face tent. In a study involving 185 patients ordered to receive oxygen therapy at flow rates of 5 L/min or more, Campbell et al.  found patient complaints of dry nose and throat in approximately 43% of the subjects, regardless of whether they were receiving dry oxygen or oxygen with a high-humidity device. In our study, we asked subjects specifically about dryness in an effort to have them distinguish between dryness and soreness, which subjects in all four groups reported in their throat.
Others have reported that patients are dissatisfied with high-humidity devices and often complain of nausea and intolerable noise . In our study, in the group who used the face tent, 10 (14%) of the 72 subjects reported the face tent to be uncomfortable and requested that it be removed. There were also 10 reports of nausea in this group. Since this form of oxygen delivery is more expensive than other delivery devices, and does not result in higher oxygen saturation levels or in fewer complaints of dry mouth or throat, there does not appear to be any benefit to the use of this system for this group of patients.
This study was confined to surgical procedures excluding the thorax, upper abdomen, or neurologic system in patients who did not have moderate or severe respiratory or cardiac disease. Although oxygen therapy was found to increase oxygen saturation levels in these patients, the difference in saturation between those who did and did not receive oxygen was not clinically significant. Those who desaturated were those who arrived in the PACU with oxygen saturations of 92% or less. We recommend that oxygen saturation levels be noted at the time of admission to the PACU and those patients with oxygen saturations of 92% or less be given oxygen therapy. Oxygen at 4 L per minute via nasal cannula maintains adequate oxygen saturation levels. The use of a humidification device with nasal cannula needs to be further tested to determine whether it results in fewer complaints of dryness in older adults.
1. Nunn JF, Payne JP. Hypoxaemia after general anaesthesia. Lancet 1962:631-2.
2. Canet J, Ricos M, Vidal F. Early postoperative arterial oxygen desaturation determining factors and response to oxygen therapy. Anesth Analg 1989;69:207-12.
3. Ehrenwerth J, Donielson D. Pulse oximetry in the postanesthesia care unit. J Post Anesth Nurs 1987;2:9-11.
4. Eichhorn JH. Prevention of intraoperative anesthesia accidents and related severe injury through safety monitoring. Anaesthesia 1989;70:572-7.
5. Vaughan RW, Wise L. Postoperative arterial blood gas measurement in obese patients: effects of position on gas exchange. Ann Surg 1975;182:705-9.
6. Aldrete JA, Kroulik D. A postanesthetic recovery score. Anesth Analg 1970;49:924-34.
7. Tyler IL, Tantisira B, Winter PM, Matoyama EK. Continuous monitoring of arterial oxygen saturation with pulse oximetry during transfer to the recovery room. Anesth Analg 1985;64:1108-12.
8. Hudes ET, Marans IIJ, Hirano GM, et al. Recovery room oxygenation: a comparison of nasal catheters and 40 per cent oxygen masks. Can J Anaesth 1989;36:20-4.
9. Berels DA, Marz MS. SaO2
monitoring in the postanesthesia care unit. J Post Anesth Nurs 1991;6:394-401.
10. Cullen DJ, Nemeskel AR, Cooper JB, et al. Effect of pulse oximetry, age, and ASA physical status on the frequency of patients admitted unexpectedly to a postoperative intensive care unit and the severity of their anesthesia-related complications. Anesth Analg 1992;74:181-4.
11. Moller JT, Johannessen NW, Berg H, et al. Hypoxaemia during anaesthesia--an observer study. Br J Anaesth 1991;66:437-44.
12. Hedstrand U, Liw G, Rooth G, Ogren CH. Effect of respiratory physiotherapy on arterial oxygen tension. Acta Anaesthesiol Scand 1978;22:349-52.
13. Kory RC, Bergman JC, Sweet RD, Smith JR. Comparative evaluation of oxygen therapy techniques. JAMA 1962;179:767-72.
14. Campbell EJ, Baker D, Crites-Silver P. Subjective effects of humidification of oxygen for delivery by nasal cannula. A prospective study. Chest 1988;93:289-93.
15. Leigh JM. Audible noise levels of oxygen masks operating on venturi principle. Br Med J 1973;15:652.