| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the Departments of *Outcomes Research, Quantitative Health Sciences, ||Cardiothoracic Surgery, and ¶Cardiothoracic Anesthesia, The Cleveland Clinic, Cleveland, Ohio; Department of Anesthesiology and Perioperative Medicine, University of Louisville, Louisville, Kentucky; and Outcomes Research Consortium.
Address correspondence and reprint requests to Daniel I. Sessler, MD, Chair, Department of Outcomes Research, Cleveland Clinic Foundation, 9500 Euclid Ave.—P77, Cleveland, OH 44195. Address e-mail to DS{at}OR.org. On the world wide web: www.or.org.
Abstract
BACKGROUND: Sternal wound dehiscence and infection complicate 1% of cardiac surgeries. Tissue oxygen tension (PsqO2) is the primary determinant of surgical wound infection risk and is often critically low in surgical incisions. We tested the hypothesis that local transdermal delivery of oxygen improves oxygenation in sternotomy wounds after cardiac surgery. Our secondary hypothesis was that supplemental inspired oxygen improves sternal wound PsqO2.
METHODS: After undergoing cardiopulmonary bypass, 30 patients randomly received (1) 2 EpiFlo oxygen generators (Ogenix, Inc., Beachwood, OH) that provided oxygen at 6 mL/h into an occlusive wound dressing or (2) identical-appearing inactive generators. PsqO2 and temperature were measured in the wound 
5 mm below the skin surface. PsqO2 and arterial oxygen (Pao2) were measured 1 h after intensive care unit admission (Fio2 = 60%) and on the first and second postoperative mornings at Fio2 of both 30% and 50% in random order.
RESULTS: Data from four patients were excluded for technical reasons. Patient characteristics were similar in each group, as were type of surgery and perioperative management. Increasing Fio2 from 30% to 50% improved Pao2 from 99 [84–116] to 149 [128–174] mm Hg (P < 0.001, mean [95% CI]) and sternal wound PsqO2 from 23 [16–33] to 27 [19–38] mm Hg (P < 0.001). In contrast, local oxygen delivery did not improve tissue oxygenation: 24 [14–41] vs 25 [16–41] mm Hg (P = 0.88).
CONCLUSIONS: Additional inspired oxygen improved Pao2 and sternal wound PsqO2 after bypass and may, consequently, reduce infection risk. However, oxygen insufflated locally into an occlusive dressing did not improve wound PsqO2 and, therefore, does not appear to be useful clinically in cardiac surgery patients to reduce sternal wound infections.
Sternal wound infections are a major complication of median sternotomy that are associated with extended hospital stay, increased hospital costs, and higher mortality and morbidity rates.1–4 The incidence of deep mediastinal wound infection in patients undergoing median sternotomy for cardiopulmonary bypass (CPB) is at least 1%, and often higher.5 Perioperative host-defense is probably the major factor determining whether an individual surgical patient becomes infected. The primary defense against surgical pathogens is oxidative killing by neutrophils.6 Oxygen is a substrate for this process, and the reaction critically depends on tissue oxygen tension throughout the observed physiological range. Surgical wounds disrupt blood supply and cause edema, both of which impair local perfusion.7 Therefore, subcutaneous tissue oxygen tension (PsqO2) inversely correlates with the risk of surgical wound infection.8,9 Adequate tissue oxygenation is also necessary for scar formation and healing.10,11
Anemia,12 cardiac output,13 local perfusion,14,15 smoking,16 perioperative fluid management,17 perioperative management of carbon dioxide tension,18,19 uncontrolled surgical pain,20 and maintaining perioperative normothermia21 are all known to influence tissue oxygenation and perfusion. In addition, inhaled supplemental oxygen improves tissue oxygenation.22 An alternative potential approach to improving tissue oxygenation is local transdermal delivery of oxygen directly to the wound under an occlusive dressing. Transdermal oxygen administration improves tissue oxygenation in animals,23 when delivered in a sustained, low-dose (3 mL/h) method to the surgical wound model, but has yet to be tested in humans. We therefore tested the hypothesis that local transdermal delivery of oxygen increases tissue oxygenation within the superficial sternotomy wound in patients recovering from cardiac surgery. Our secondary hypothesis was that supplemental inspired oxygen improves tissue oxygenation in sternal wounds.
METHODS
With approval from the IRB at the Cleveland Clinic and written, informed consent, 30 patients scheduled for elective coronary artery grafting and/or valve repair with CPB were enrolled between February 2006 and March 2007. Patients undergoing emergency procedures or off-pump cardiac surgery, or who were older than 75 yr or obese (body mass index 
35 kg/m2) were excluded from the study.
IV and arterial catheters were inserted per clinical routine. After administration of 100% oxygen, anesthesia was induced with etomidate (0.2–0.4 mg/kg) or sodium thiopental (1–3 mg/kg) and fentanyl (3–5 µg/kg). Succinylcholine (1–1.5 mg/kg) or pancuronium (0.1 mg/kg) was given to facilitate endotracheal intubation. Anesthesia was maintained with fentanyl, midazolam, isoflurane (maximum 1% inhaled in 50% oxygen/50% nitrogen), and pancuronium sufficient to maintain one twitch in response to supramaximal train-of-four stimulation of the ulnar nerve at the wrist. After induction of anesthesia, prophylactic antibiotics consisted of either cefuroxime (1.5 g IV every 8 h for 3 days) or, for patients allergic to penicillin, vancomycin (1 g every 12 h for 3 days). Institutional standard myocardial protection and CPB techniques were used. Bladder and nasopharyngeal temperatures exceeding 36.5°C were achieved before separating the patients from CPB.
Upon completion of surgery and sternal closure, the surgical team positioned a Licox tissue oxygen and temperature probe (Licox, Gesellschaft für Medizinische Sondensysteme, GmBH, Kiel, Germany) subcutaneously at a depth of 0.5 cm before completing sternal wound closure. The proximal portion of the probes was secured to skin with sutures. The use of Clark-type electrode oxygen sensors to measure tissue oxygenation is well established.24
Tissue oxygen insufflation was with two Epiflo oxygen generators provided by Ogenix Inc. (Beachwood, OH). Based on computer-generated randomization codes, the devices were either both active providing oxygen at a rate of 3 mL/h each, or they were inactive and provided no oxygen or air. Each set was labeled with a sequential number. The devices were otherwise identical (Fig. 1), allowing all investigators to remain completely blinded to treatment. The peri-incisional skin was cleaned with normal saline 0.9% and dried with sterile gauze. A 4-by-4-inch Strip of DuoDERM Extra Thin CGF Sterile Dressing (ConvaTec, Princeton, NJ) was cut in half. Each half was placed on the central portion of the peri-incisional skin perpendicular to the incision 1 cm away from the midline. A cannula from each EpiFlo generator was positioned over the DuoDERM dressing with the tip of the cannulae protruding about 0.5 cm beyond the dressing. The cannulae were secured to the DuoDERM with Steri-Strip tape (3M Health Care, MN). The entire site was then sealed under a Tegaderm transparent film dressing (3M Health Care), leaving a 2-inch margin between the tip of the cannula and the edge of the transparent dressing. The transparent film dressing around the extension cannulae was pinched to insure a leak-proof seal. For further assurance, an additional Tegaderm film was applied over the tips of the cannulae. EpiFlo devices were positioned over the clavicular area on each side of the patient and secured with tape. Together, the two EpiFlo devices provided oxygen to the wound at a rate of 6 mL/h into the Tegaderm occlusive dressing.
|
Both oxygen and temperature probes of the Licox tissue oxygen system, the two extension cannulae, all Tegaderm dressings, and the DuoDERM were removed in a sterile manner on the second postoperative day after completion of the tissue oxygen measurements (described below) and after physician examination of the wound site.
Oxygen at 6 mL/h was used for transdermal therapy based on animal experiments23 and our preliminary studies in which this flow rate was able to optimally insufflate oxygen while maintaining optimal coverage of the sternotomy wound. Each device generated a maximum oxygen flow of 3 mL/h, and thus two devices were need per patient.
After the completion of surgery, patients were transferred to the cardiothoracic intensive care unit where there lungs were mechanically ventilated for about 2 h (synchronized intermittent mandatory ventilation with pressure support [5 cm H2O], positive end expiratory pressure of 5 cm H2O, tidal volume of 6–8 mL/kg, and Fio2 of 0.6; respiration rate was 10–12 bpm depending on the previous blood gas values) until the first set of tissue oxygen measurements were finished. Patients were subsequently weaned from ventilator support and their tracheas extubated as tolerated per routine.
Measurements
Morphometric and demographic characteristics, ASA physical status score, the duration of CPB, aortic cross-clamp time, hemodynamic values, fluid administration, and urine output were all recorded. During surgery, core body temperature was measured via the urinary bladder catheter; postoperative temperature was measured from the bladder catheter when possible. An electronic oral thermometer was used after the catheter had been removed.
PsqO2 was measured with a polarographic electrode system, as previously described.25 Each sterile, single-patient-use oxygen electrode was autocalibrated by a unique "smart card." In vitro accuracy of the oxygen sensors is ±3 mm Hg for the range from 0 to 100 mm Hg O2, and ±5% for 100–360 mm Hg O2 (in a water bath at 37°C). Temperature sensitivity is 0.25%/°C. Thermistors are incorporated into the probes and temperature compensation is included in the PsqO2 calculations. Oxygen sensor calibration remains stable (within 8% of baseline value for room air) in vivo for at least 8 h.
Tissue oxygenation was measured 1 h after admission to the intensive care unit while the patients received an inspired oxygen concentration of 60%. Tissue oxygen measurements were repeated on the first and second postoperative mornings when the patients were receiving oxygen at an Fio2 of approximately 30% and approximately 50% in a randomly determined order. Inspired oxygen for the second and third set of measurements was provided by a PULMANEX Hi-Ox® Adult Oxygen Mask (Viasys MedSystems, Wheeling, IL) with an oxygen flow rate of 1 or 5 L/min. After 30 min of exposure to the designated inspired oxygen concentration, tissue oxygenation was recorded at 5-min intervals for 15 min. Arterial blood was sampled for gas analysis in the middle of each measurement period (i.e., after 37 min exposure to each oxygen concentration).
Statistical Methods
This parallel group study of transdermal delivery status (the oxygen generators for each patient were "on" or "off") included a crossover component since all patients received both 30% and 50% Fio2 in random order (distinct order each day) on postoperative days (POD) 1 and 2. Effects of concentration and day were thus analyzed as within-subject factors and transdermal delivery status as a between-subject factor. Response variables of interest were Licox PsqO2 and partial oxygen tension in arterial blood (Pao2), both in mm Hg. For patients to be included in the statistical analysis, extubation had to occur by the time measurements were conducted on POD one. This restriction was applied to avoid manipulation of inhaled oxygen concentration in mechanically ventilated patients for the sole purpose of this study. It is also why the day of surgery was not considered in the modeling.
Comparison between the device status groups on PsqO2 and Pao2 was based on multivariable repeated-measures analysis of variance models with adjustment for inspired oxygen concentration and POD. Multivariable comparison of tissue oxygen values between 30% and 50% Fio2 were adjusted for device status group and POD, whereas multivariable analysis of arterial oxygen (also adjusting for device status group) was restricted to the first POD because some data were missing on the second day.
The standard deviation of tissue oxygen measurements is typically about 15 mm Hg for tissue oxygenation range of 60–150 mm Hg. The minimum clinically important treatment is also about 15 mm Hg. Fifteen patients in each group provided 75% power for detecting a true difference of 15 mm Hg between active and inactive insufflation if it indeed existed. We thus studied 15 patients in each group.
Statistical models were developed with significance levels of 0.05 for main effects and 0.10 for interactions. Multiple comparisons within a hypothesis were made using the Bonferroni adjustment to control the experiment-wise error rate at 0.05. Response variables were transformed (e.g., log-transformation) as appropriate for meeting model assumptions. Only the sponsor knew which devices were active; conversely, only the investigators had access to the clinical data. The statistician was blinded to the active/inactive assignments and was given group labels "A" and "B" for transdermal delivery device status. SAS software version 9.1 (SAS Institute, Cary, NC) and R software version 2.4.1 (The R Foundation for Statistical Computing) were used for all statistical analyses. Data are presented as mean (95% confidence intervals) unless otherwise noted.
RESULTS
Data from four randomized patients (three from the inactive group, one from the active group) were excluded from the study for cause (reoperation, unmanipulable experimental conditions, and lack of compliance with the study protocol). Thus, 26 patients were included in the study: 12 from the inactive group and 14 from the active group. However, when PsqO2 was analyzed as an outcome, data from one additional patient (inactive group) was excluded because of a faulty Licox probe; thus, data from 25 patients (11 from the inactive group and 14 from the active group) were analyzed. Further, when Pao2 was analyzed, 2 additional patients (active group) were excluded due to missing data; thus, data from 24 patients (12 from each group) were analyzed. There were no clinically important differences between the groups in morphometric or demographic characteristics, or perioperative factors (Table 1). Postoperative inotropic and vasopressor requirements were also similar.
|
To meet the regression assumption of normality of residual errors, PsqO2 and Pao2 response values were each analyzed on the logarithmic scale. As a result of this transformation, the geometric means are estimated (rather than the usual arithmetic mean) in the mixed models. It follows that confidence intervals for geometric means are asymmetric and assessment of factor level differences is based on ratios of means instead of differences of means.
Geometric mean PsqO2 was similar under all experimental scenarios, whereas some variability in geometric mean Pao2 was observed (Figs. 2 and 3). Although the between-patient variability of both responses was quite large, the variability of the within-patient differences in these response variables was small in relation to the mean differences, as evidenced by the nearly parallel lines in Figures 4 and 5.
|
|
|
|
|
In the mixed model, no interactions were found to be statistically significant at the 0.10 level. After these considerations, the ratio (95% CI) of means, when comparing active to inactive insufflation, was estimated at 1.0 (0.5–1.9), after adjusting for Fio2 and postoperative day (Table 2 and Fig. 2) (P = 0.88). Additionally, it was estimated that mean PsqO2 increased by 20% (95% CI of 10–20% increase) when manipulating Fio2 concentration from 30% to 50% (P < 0.001). This statistical significance was, in large part, caused by the small variability of the differences in PsqO2; mean (standard error) of difference (50% Fio2 minus 30% Fio2) in POD-adjusted PsqO2 among all patients was 4.9 (1.2).
As seen in Figure 3, arterial oxygen was generally greater at 50% Fio2 than at 30%. Median Pao2 was greater among patients receiving active transdermal delivery than among patients treated with an inactive device on both PODs. However, the interquartile ranges for the 24 patients with recorded Pao2 on POD 1 (12 from each group) overlapped substantially (at each inspired oxygen concentration), whereas only 8 patients had recorded tissue oxygen pressure on POD 2. Because of these sample size issues, only data from POD 1 were considered in the modeling of Pao2.
Results of a mixed model with transdermal delivery status and inhaled O2 concentration as predictors and Pao2 as the response indicated that mean Pao2 did not differ between the transdermal delivery groups, but that Pao2 increased 50% (20% increase, 80% increase [95% CI]) when Fio2 was manipulated from 30% to 50% (P < 0.001, Table 2). The ratio (95% CI) of mean Pao2 between active and inactive insufflation, estimated at 1.1 (0.9, 1.4), was not statistically significant (P = 0.34).
DISCUSSION
The risk of developing a wound infection is highly dependent on tissue oxygenation, and tissue oxygen partial pressure is probably the best single predictor of infection risk. Consequently, interventions that improve tissue oxygenation reduce infection risk.8,9 Tissue oxygen partial pressure is a balance between supply and local demand. Demand is a function of tissue metabolism that, in turn, depends on factors such as local temperature and tissue injury. Perfusion is the primary determinant of supply which is why subcutaneous oxygenation is improved by increasing local flow by regional warming,14,15,26 aggressive hydration,17 or sympathetic nerve block.27–29 However, tissue oxygenation can also be improved by augmenting the oxygen content of blood by increasing hemoglobin, supplemental oxygen,30 or mild hypercapnia.18,19,31
Our primary result in this pilot study is that local transdermal delivery of oxygen did not improve oxygenation within sternal wounds after cardiac surgery. The most obvious reason that local transdermal oxygen delivery failed to increase tissue oxygen is that skin is relatively impervious to oxygen. We measured oxygen partial pressure about 0.5 cm below the skin surface; it remains possible that oxygen partial pressure did increase in the outermost layer of skin (perhaps the first mm). The failure to increase PsqO2 in deeper wound levels, though, would likely not provide protection against infections. Our protocol was based closely on a study in rabbits in which local transdermal delivery of oxygen markedly improved ischemic epithelial wound healing,23 a process closely linked to tissue oxygenation.10,11 It remains possible that a higher transdermal oxygen delivery rate would prove more effective or that this approach might be effective in other situations such an open wound. We did find, unsurprisingly, that increasing inspired oxygen from 30% to 50% significantly increased the tissue oxygen partial pressure. These findings are consistent with numerous previous studies of experimental wounds and abdominal incisions.30,32,33 Not all patients had pulmonary artery catheters; we were thus unable to monitor oxygen delivery in all patients participating in the study.
Perhaps, our most striking finding was the low levels of observed sternal wound oxygen tensions averaging 23 mm Hg with 30% inspired oxygen. This partial pressure of oxygen is markedly less than that recorded in upper-arm wounds in euthermic, euvolemic, uninjured volunteers, or in anesthetized patients before the onset of surgery34 that are typically 65 mm Hg.14,35 Our observed sternal oxygen tension was also markedly lower than tissue oxygen adjacent to surgical wounds,22,33 in the mucosa of the small and large intestine,34 and in mastectomy wounds33 as described in prior studies.
There are several potential explanations for the low sternal wound oxygen tension that we observed. Sympathetically mediated peripheral vasoconstriction results from hypothermia, pain, fear, intravascular volume depletion,36,37 and exposure to medications such as β-adrenergic antagonists and 
1-adrenergic agonists that are common in cardiac surgery patients. Subcutaneous tissue is particularly vulnerable to vasoconstriction because there is little regulation of blood flow, except in response to locally applied heat.38,39 Blood loss and third space fluid shifts may further compromise subcutaneous tissue perfusion and, ultimately, tissue oxygenation.29 Regardless, this is the first report to our knowledge of tissue oxygen tension measured in sternal wounds of patients after cardiac surgery. It might be that subcutaneous tissue oxygenation of this region is lower than other subcutaneous tissue areas reported in previous studies. Sustained exposure to oxygen insufflation may have improved subcutaneous oxygen. But, because of the very preliminary nature of this trial, we sought to limit any potential risk of having a subcutaneous probe in place for prolonged periods.
Tissue oxygen tension is typically lower than arterial oxygen level by a factor of two to four. As might be expected, tissue oxygenation improves much less than arterial oxygen in response to supplemental oxygen administration. In the current study, sternal wound oxygenation increased by an average of 4 mm Hg (from 23 to 27 mm Hg) with supplemental oxygen. This increase, although small, may be clinically important because both values are on the steep part of the curve relating tissue oxygen partial pressure to oxidative killing by neutrophils.40 In this study, we used 30% and 50% inspired oxygen concentrations. A higher inspired oxygen concentration might have further increased both arterial and wound oxygenation. It is also possible that a higher dose (greater oxygen flow) might have improved wound oxygenation.
In summary, supplemental inspired oxygen improved arterial and sternal wound tissue oxygenation after cardiac surgery using CPB. However, local transdermal oxygen administration did not improve wound oxygenation and therefore does not appear to be a clinically useful strategy in adult cardiac surgery patients to improve sternal wound healing.
ACKNOWLEDGMENTS
The authors thank Edward Mascha, PhD (Departments of Quantitative Health Sciences and Outcomes Research, The Cleveland Clinic), for his assistance. Nancy Alsip, PhD, edited the manuscript (University of Louisville).
Footnotes
Accepted for publication February 20, 2008.
Supported by Ogenix, Inc. (Beachwood, OH), NIH Grant GM 061655 (Bethesda, MD), and the Joseph Drown Foundation (Los Angeles, CA). The Hi-Ox masks were generously provided by Viasys MedSystems, Wheeling, IL.
The investigators developed the protocol; the sponsors had no role in data-acquisition, analysis, or manuscript preparation. None of the authors has a personal financial interest in this research.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
