Radiographic evaluation of positional atelectasis in sedated dogs breathing room air versus 100% oxygen

Abstract

This study documents the degree of positional atelectasis in sedated dogs receiving 100% oxygen (O₂) versus room air. Initial lateral recumbency was determined by an orthopedic study and initial treatment (O₂ or room air) was randomized. Each dog was maintained in lateral recumbency for 15 min, at which time ventrodorsal (VD) and opposite lateral thoracic radiographs were obtained. Each dog was then maintained in the opposite lateral recumbency and received the other treatment for 15 min, followed by a VD and opposite lateral radiograph. Radiographs were scored for severity of pulmonary pattern and mediastinal shift by 3 radiologists. Dogs breathing O₂ had significantly higher scores than dogs breathing room air. If radiographically detectable dependent atelectasis is present, repeat thoracic images following manual positive ventilation and/or position change to the opposite lateral recumbency should be made to rule out the effect of O₂ positional atelectasis and avoid misdiagnosis.

Introduction

Veterinary patients receive 100% oxygen for various reasons including inhalation anesthesia, supplementation for respiratory compromise, and before/during induction of general anesthesia to minimize the chance of hypoxemia. Oxygen supplementation is also recommended for patients receiving dexmedetomidine, an alpha-2 adrenergic agonist used to sedate patients for minimally invasive and diagnostic procedures (1–4). Supplemental oxygen may be administered to dyspneic patients by mask or flow-by during radiographic evaluation of the lungs. While most thoracic radiographs are taken of non-sedated patients breathing room air, they are sometimes obtained in anesthetized or sedated patients receiving oxygen supplementation.

Atelectasis occurs in the dependent lung during lateral recumbency in awake dogs breathing room air (5,6). Research in humans and animals has documented that the degree of atelectasis in dependent portions of the lung is worsened by the administration of 100% oxygen (7–11). This atelectasis can manifest as an abnormal lung pattern (increased or asymmetrical interstitial to alveolar) with ipsilateral mediastinal shift, which may be misinterpreted as lung disease such as pneumonia or pulmonary edema (5) and can lead to an incorrect diagnosis and inappropriate or unnecessary treatment.

Computed tomography (CT) is considered the gold standard for assessing the degree of atelectasis (12). However, low availability and high cost often prohibit use of CT in most private practice settings. Radiography is the most common diagnostic imaging modality used in private practice to image the lungs.

Currently, there is no study that documents radiographically perceptible positional atelectasis in animals breathing 100% oxygen over a short time period. The purpose of this study was to evaluate the degree of positional atelectasis in a 15-minute time period in healthy dogs breathing 100% oxygen via a tightly fitted mask, versus breathing room air. The hypothesis is that radiographically perceptible positional atelectasis is evident in normal dogs breathing 100% oxygen via a mask over a 15-minute time period, and is not evident or not significant when breathing room air over the same time period using conventional vertical beam radiography.

Materials and methods

Animals and study design

The study was a prospective randomized blinded controlled crossover clinical trial. The study population (40 dogs) was derived from canine patients presented to the orthopedic service of the University of Georgia Veterinary Teaching Hospital for elective procedures, which required appendicular/pelvic radiographs to be performed under sedation. Owners signed a consent form prior to enrollment of their dog in the study, and the study was approved by the Hospital Clinical Research Committee. Patients were excluded by physical examination and pre-surgical bloodwork if they had any illness aside from their presenting complaint and if they had any evidence of clinical or radiographic pulmonary disease based on three-view thoracic radiographs performed prior to sedation for the study. Information recorded for each patient included age, gender, breed, weight, body condition score (BCS), presenting complaint, orthopedic study requested, radiographic technique used (kVp, mAs), sedation protocol and dose administered, and order of recumbency and treatment.

Thoracic radiography

All radiographs were exposed on digital radiography systems (Sound-Eklin EDR6; Sound-Eklin Medical Systems, Carlsbad, California, USA). Three-view (right lateral, left lateral, and VD) thoracic radiographs at peak inspiration and optimal technique were obtained prior to sedation to rule out underlying pulmonary disease. Sedation drugs and doses were determined by the attending clinician and were not standardized. The premedication drugs were mixed in 1 syringe and administered intravenously. Craniocaudal, caudocranial, or VD orthopedic images were made immediately after sedation. The patient was maintained in initial lateral recumbency breathing either i) 100% oxygen via a tightly fitted facemask connected to a stand-alone oxygen flowmeter at a preset flow rate of 4 L/min, or ii) room air with no mask. The total time each animal was maintained in the initial lateral recumbency was standardized. This 15-minute period included the time required to acquire orthopedic radiographs. Ventrodorsal and opposite lateral thoracic views were exposed (if in right lateral recumbency, a left lateral radiograph was made) after 15 min. The patient was then maintained in the second lateral recumbency breathing the other gas for 15 min (i.e., if administered room air for the first 15 min, the patient was administered 100% oxygen for the next 15 min). A VD and opposite lateral radiographs were then obtained after the additional 15 min, thus completing the study.

Each subject served as its own control. Initial lateral recumbency was determined by the orthopedic study requested. For pelvic and bilateral orthopedic studies (stifles, elbows, shoulders, etc.), the order of initial lateral recumbency was alternated. Treatment order (room air or 100% oxygen) was determined by restrained randomization via coin toss for the first patient; the subsequent patient was administered the opposite treatment for the same lateral recumbency (i.e., if the first subject had room air on the left side, the second subject had oxygen on the left side). The coin toss was repeated for every other subject. This allowed randomization of treatment order, but ensured balance of treatments for left and right recumbency.

Thoracic radiographs were evaluated by 3 board-certified radiologists blinded to treatment. Reviewers viewed pre-sedation images and images under both treatments for each patient, and the order of patients was randomized. Images were viewed on medical grade monitors (Totoku, Totoku Electric, Ueda, Nagano, Japan). Evaluators were asked to score for pulmonary pattern and mediastinal position using the criteria described in Figures 1 and 2. Pulmonary pattern and mediastinal shift scores were summed for each dog, evaluator, and treatment to calculate a summary score. Evaluators also wrote comments regarding the images on the individual score sheets.

Figure 1.

Radiographic examples of pulmonary pattern scores (right lung was dependent prior to acquisition). 0 — no change in pulmonary appearance from pre-sedation images; 1 — mild interstitial pattern in dependent lung; 2 — moderate-to-heavy interstitial pattern in dependent lung; 3 — alveolar pattern in dependent lung.

Figure 2.

Radiographic examples of mediastinal position scores (right lung was dependent prior to acquisition). 0 — no mediastinal shift present; 1 — slight ipsilateral (towards the dependent lung) mediastinal shift; 2 — clear evidence of ipsilateral mediastinal shift.

Statistical analysis

A power analysis was performed on a pilot study of 9 dogs using statistical software nQuery 4.0 (Statistical Solutions, Cork, Ireland). A paired t-test was performed on summary scores using a mean difference of 1 ± 1.9 and assuming 80% power. Based on these assumptions, a sample size of 31 dogs for the summary score would be needed to have an 80% chance of seeing a significant effect at the difference levels measured. The sample size was increased to 40 to increase the power of the study.

Values for pulmonary score, mediastinal score, and summary score were calculated by taking the average value of the 3 observers. A repeated measures analysis of variance was used to test for differences in pulmonary pattern, mediastinal shift, and summary scores between treatments (room air versus oxygen). Known or potential confounding variables were included in the model. The full model included fixed factors for treatment, evaluator, recumbency (left or right), order (1st or 2nd), sedation method (3 methods used), body condition score (4 to 8), and a continuous factor of weight. Treatment by order, treatment by recumbency, order by recumbency two-way interaction terms and a treatment by recumbency by order three-way interaction term and a random dog factor were also included. Tukey’s test was used to adjust for multiple comparisons.

All hypothesis tests were 2-sided and the significance level was α = 0.05. The repeated measures analysis of variance was implemented using PROC MIXED.

Results

Forty dogs were included in the study. Breeds included were Labrador retriever (n = 13), mixed breed (n = 6), golden retriever (n = 3), Australian shepherd (n = 3), boxer (n = 2), German short-haired pointer (n = 2), and 1 each of American bulldog, Australian cattle dog, chow, collie, corgi, great Pyrenees, German shepherd, Jack Russell terrier, pit bull, rottweiler, and Weimaraner. There were 20 spayed females, 2 intact males, and 18 neutered males. Mean ± standard deviation (SD) age was 4.9 ± 2.2 y (range: 1 to 9 y). Mean ± SD body weight was 30.9 ± 10.3 kg (range: 10 to 57 kg). Mean ± SD BCS was 6 ± 1 (range: 4 to 8 out of 9). Three different sedation protocols were used. Medetomidine and dexmedetomidine were considered the same protocol as their formulations and clinical effects are similar (13,14). Five dogs received acepromazine (0.01 to 0.02 mg/kg BW) and butorphanol (dose 0.2 mg/kg BW). Six dogs received dexmedetomidine (0.001 to 0.006 mg/kg BW) and butorphanol (dose 0.2 mg/kg BW). Twenty-nine dogs received medetomidine or dexmedetomidine alone; 3 received medetomidine (0.005 to 0.017 mg/kg BW) and the other 26 received dexmedetomidine (0.004 to 0.014 mg/kg BW).

Higher scores indicate more radiographically perceptible atelectasis, and statistically significant results are listed in Table 1. When receiving 100% oxygen via facemask, dogs had significantly higher summary, pulmonary pattern, and mediastinal shift scores (P < 0.003, P < 0.02, P < 0.004, respectively) than when breathing room air. In right lateral recumbency, summary and pulmonary pattern scores were significantly higher (P < 0.01 and P < 0.003, respectively), compared with left lateral recumbency. Dogs with a BCS of 7 had significantly higher scores than dogs with a BCS of 5 for summary, pulmonary pattern, and mediastinal position (P < 0.002, P < 0.01, and P < 0.01, respectively). Dogs with a BCS of 7 also had significantly higher scores than dogs with a BCS of 6 for pulmonary pattern (P < 0.001). The first treatment and recumbency had significantly higher summary, pulmonary pattern, and mediastinal position scores than the second (P < 0.0001, P < 0.0002, and P < 0.0001, respectively).

Table 1.

Pulmonary pattern, mediastinal shift, and summary scores in sedated dogs given room air or 100% oxygen

	Pulmonary pattern	Mediastinal position	Summary score
Treatment
100% oxygen	1.4 ± 0.9a	1.0 ± 0.7a	2.4 ± 1.6a
Room air	1.2 ± 0.7a	0.8 ± 0.7a	2.1 ± 1.3a
Body condition score
5/9	1.0 ± 0.4a	0.6 ± 0.3a	1.6 ± 0.7a
6/9	1.1 ± 0.7b	0.9 +/− 0.62	2.1 +/− 1.2
7/9	2.0 ± 0.3a,b	1.4 ± 0.4a	3.4 ± 0.7a
Recumbency
Right lateral	1.5 ± 0.8a	1.0 +/− 0.8	2.5 ± 1.5a
Left lateral	1.2 ± 0.8a	0.8 +/− 0.8	2.0 ± 1.5a
Treatment order
1st	1.5 ± 0.8a	1.2 ± 0.8a	2.7 ± 1.5a
2nd	1.1 ± 0.7a	0.6 ± 0.7a	1.8 ± 1.3a
Sedation protocol
Medetomidine or dexmedetomidine	1.5 ± 0.6a,b	1.0 +/− 0.6	2.4 +/− 1.1
Acepromazine with butorphanol	0.9 ± 0.5a	0.8 +/− 0.3	1.7 +/− 0.8
Dexmedetomidine with butorphanol	0.8 ± 0.6b	0.7 +/− 0.6	1.5 +/− 1.2
Evaluators
Evaluator 1	1.3 ± 0.8b	1.0 +/− 0.6	2.2 +/− 1.3
Evaluator 2	1.6 ± 0.8a,b	0.9 +/− 0.6	2.5 ± 1.2a
Evaluator 3	1.1 ± 0.6a	0.8 +/− 0.6	2.0 ± 1.1a

All values are reported as mean +/− standard deviation. Values with the same superscript indicate significant difference within a column for that category.

Sedation protocol resulted in significant differences in pulmonary pattern scores only. Dogs that received medetomidine/ dexmeditomidine had higher scores than dogs administered acepromazine and butorphanol (P < 0.03), and dexmedetomidine and butorphanol (P < 0.03).

Evaluator 2 had higher scores than evaluator 3 for summary and pulmonary pattern scores (P < 0.007 and P < 0.0001, respectively). Evaluator 2 also had higher scores than evaluator 1 for pulmonary pattern score (P < 0.01).

Comments made by the evaluators on the scoresheets included: post-treatment films expiratory/lungs underinflated compared to pre-treatment (n = 79), pre-treatment films more expiratory than post-treatment (n = 3), curvature of spine noted on pre or post-treatment films (n = 13), pulmonary pattern or mediastinal shift observed in contralateral lung (n = 3), patient rotated (n = 1), patient not centered on post-treatment film the same as pre-treatment film (n = 1).

A change (score greater than 0 by at least 1 evaluator) was detected in post-treatment images 77 of the 80 times a treatment was administered to a patient (40 dogs, each received both, room air and 100% oxygen via a tightly fitted facemask). No change was detected (i.e., scored all zeros for pulmonary pattern and mediastinal shift score) 3 times. Dog number 13 had no detectable change after breathing oxygen, and dog number 38 had no detectable change after breathing oxygen, or after breathing room air.

Discussion

Room air is composed of 21% oxygen and about 78% nitrogen. Oxygen is two times more soluble than nitrogen in plasma, and will diffuse across cell membranes twice as fast as nitrogen. Nitrogen will remain in terminal air spaces longer, maintaining open alveoli and delaying the onset and lessening the severity of atelectasis (15). Prior studies have demonstrated that administration of 100% oxygen will increase the severity and shorten the onset of dependent positional atelectasis (7–12,6).

To deliver 100% oxygen, the gas needs to be administered via an endotracheal tube. This requires induction and maintenance of general anesthesia, which was not the purpose of this study. We wanted to investigate the degree of atelectasis in sedated dogs receiving room air versus oxygen for a short time period (15 min). Clinical studies have shown that a tightly fitted face-mask is the most efficient method of oxygen administration in dogs compared with intranasal catheter, Elizabethan collar canopy, and flow-by techniques (17–19).

Loukopoulos and Reynolds (19) showed that when a tightly fitted mask was used in sedated dogs, the FiO₂ at the tracheal bifurcation reached a plateau of 89% when the oxygen flow rate was 6 L/min and 80% when the flow was 4 L/min. In our study we decided that 4 L/min was a good compromise. A flow rate greater than 4 L/min would have probably stimulated the sedated patients, which would have affected our results. We also believe that a FiO₂ of 80%, which corresponds to 5 times the FiO₂ at the bifurcation when room air is inhaled (19), should be sufficient to make a difference in the degree of atelectasis if this was present.

Positional atelectasis was more severe when dogs were administered oxygen than when breathing room air, as evidenced by significantly higher scores for summary, pulmonary pattern, and mediastinal shift. This is supportive of previous studies, and not surprising given the differences in solubility and diffusion rate of oxygen when compared with room air containing nitrogen. This is in partial agreement with our hypothesis, which stated that radiographically perceptible atelectasis would only be evident or significant in dogs receiving oxygen, and not in dogs breathing room air. Scores indicated that all except 1 dog breathing room air did have radiographic evidence of atelectasis, which was occasionally severe, and 2 dogs in the oxygen group had no radiographic evidence of atelectasis. The mean difference in summary score was only 0.21 when comparing the 2 treatment groups. The original power analysis assumed a mean difference of 1, as the difference between groups should be perceptible to all practitioners, generalists, and specialists alike. Although we were able to demonstrate statistical significance, it is uncertain whether a difference of 0.21 is clinically perceptible to practitioners.

Dogs had more radiographic evidence of atelectasis in right lateral recumbency than in left (for summary and pulmonary pattern scores). It has been demonstrated that right lateral recumbency decreases functional residual capacity in healthy deep-chested dogs by approximately 20 mL/kg. Decrease in functional residual capacity will often lead to peripheral airway collapse (20). The total lung volume of the right lung is larger than that of the left, which may make changes in the right lung more perceptible (16).

Dogs with a BCS of 7/9 showed more atelectasis than dogs with a BCS of 5/9 (summary, pulmonary pattern, and mediastinal shift scores), and 6/9 (pulmonary pattern). Given these results, there appears to be a trend towards observing increased atelectasis with increasing obesity. Most of our study population had a BCS of 5/9, 6/9, and 7/9. Only 5 dogs had a BCS of 8/9, and 1 dog had a BCS of 4/9; statistical significance could not be shown for these groups (BCS 8/9 and 4/9). Specific obesity related pulmonary dysfunction is not reported in dogs. However, it is known that obesity decreases thoracic compliance, restricts expansion of the thorax, and can worsen clinical signs and gas exchange in dogs with airway disease (21). In humans, “Pickwickian syndrome” or obesity hypoventilation syndrome is the specific condition of obesity, lethargy, hypoventilation, and erythrocytosis. It is defined as an increase in PaCO₂ in the absence of pulmonary disease in extremely obese patients (22). There is anecdotal speculation that this syndrome may exist in dogs, which may also explain increased atelectasis in more obese patients.

More atelectasis was evident in dogs during the first recumbency and treatment than the second for summary, pulmonary pattern, and mediastinal shift scores. Nineteen out of 40 dogs received oxygen first, and 23 were in right lateral recumbency first. These groups were almost equally balanced making it unlikely that these variables affected this result. We are uncertain why the first recumbency and treatment showed more atelectasis than the second, and there were no significant interactions between order, treatment, and recumbency.

Dogs sedated with medetomidine/dexmedetomidine developed slightly more atelectasis than dogs sedated with acepromazine and butorphanol or dexmedetomidine and butorphanol, for pulmonary pattern scores only. Medetomidine is a specific α² adrenergic agonist that produces dose-dependent sedation in dogs. Dexmedetomidine is the active dextroisomer of medetomidine, and has similar clinical effects at half the dose. Both have analgesic, muscle relaxing, anxiolytic, and anesthesia sparing effects (13). Butorphanol is a kappa-agonist and mu-antagonist synthetic opioid, having 3 to 5 times the potency of morphine (23). Acepromazine is a neuroleptic phenothiazine derivative that interferes with dopamine transmission in the central nervous system, and produces mild to moderate sedation (24). These 3 drugs produce similar degrees of respiratory depression and decrease in respiratory rate, so it is unclear why pulmonary scores were slightly higher in dogs receiving medetomidine/dexmedetomidine. Clinically all dogs, irrespective of the drugs administered, showed a similar degree of sedation with minimal resistance to handling. Respiratory rate was monitored throughout the study, but other respiratory parameters were not measured (i.e., minute volume and end-tidal carbon dioxide). Due to the lack of these data we cannot exclude that dogs that received medetomidine/dexmedetomidine experienced more respiratory depression resulting in a higher degree of atelectasis. Since butorphanol was administered to both groups of dogs that showed less atelectasis, it is possible that butorphanol may alter alveolar stability or pulmonary blood flow. Unfortunately, neither of these variables has been studied in dogs.

We encountered a few limitations during this study. First, thoracic CT was not performed, although it is considered the gold standard for assessment of these pulmonary changes (12). While CT would have been useful to corroborate radiographic findings, it is not available in most general practices, is more expensive than radiography, and it may require general anesthesia when the thoracic region is scanned. Second, a perfect seal is not achievable while using mask administration of 100% oxygen in dogs. However, it has been demonstrated that there is no significant difference in inspired oxygen concentration or PaO₂ in dogs breathing oxygen via a tightly fitted mask compared with intubated dogs (25). We chose this method of oxygen administration because it is noninvasive, easy to perform, and what is often done in clinical circumstances. Thirdly, only two 15-minute time periods were utilized due to the constraints of using client-owned patients presented for orthopedic studies in a busy academic hospital setting. Potentially, if each patient had been maintained in lateral recumbency longer, the effects may have been greater. Lastly, the exact positioning and degree of inspiration on pre- and post-treatment radiographs were not consistently repeatable, as indicated by the evaluator comments. This was our most significant limitation, as it complicated interpretation of the radiograph studies. In many dogs, the lungs were underinflated or expiratory, especially on post-treatment images. This would have created an artifactual pulmonary pattern, which may confound evaluation, resulting in an artificially increased pulmonary pattern score. Some inconsistencies were also seen in patient positioning in which the patient was mildly rotated or there was mild curvature in the vertebral column. This may create an artificial mediastinal shift, which would increase the score for mediastinal position. While interobserver variability is inherent to any study, this limitation may also contribute to the discrepancy in evaluator scoring. However, given the relatively large sample size, use of random allocation to treatment group, and use of patients as their own control, these effects should be distributed evenly through the treatments, allowing accurate assessment of any observed differences.

Radiographically detectable dependent positional atelectasis is slightly more severe in sedated dogs breathing 100% oxygen via a tightly fitted facemask compared with room air over a 15-minute time period. Dogs also showed more radiographic evidence of dependent positional atelectasis in right lateral recumbency, and with increasing BCS (obesity). This emphasizes the importance of recognition of dependent positional atelectasis in dogs breathing oxygen. If radiographic changes are present, this finding should be questioned and follow-up images after ventilation or position change should be made before interpreting the changes as pulmonary disease.