INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There is convincing clinical and experimental evidence that diaphragm function is impaired after upper abdominal surgery (1). Typical postoperative pulmonary complications, such as basilar atelectasis and pneumonia, and postoperative disruptions in pulmonary physiology, including reductions in vital capacity and functional residual capacity, are commonly ascribed to temporary postoperative diaphragm dysfunction (6). As early as 1908, Pasteur suggested that the dysfunction after upper abdominal surgery was not an intrinsic, structural injury of the diaphragm but some type of reflex inhibition of diaphragmatic movement (7). More recently it has been suggested that such inhibition may arise from abdominal visceral afferents (8). With the arrival of direct measurements of diaphragm shortening after laparotomy in awake intact mammals, it could be seen that segmental shortening was abnormal after laparotomy (2). In some instances, the diaphragm actually lengthened with inspiration in the days following laparotomy, with 10-14 d elapsing before the diaphragm returned to normal uniphasic inspiratory shortening.

In humans and most other mammals, spontaneous breathing is periodically interrupted by sighs (8, 10, 11). These spontaneous breaths also occur after laparotomy. This fortunate coincidence of the sigh arising soon after upper abdominal surgery may provide fresh insight into the etiology of the diaphragm dysfunction. The sigh is an automatic, easily identified, centrally originating breath (12), which enjoys some independence from chemical stimulation or mechanical afferent feedback for its initiation (13, 14). We can take advantage of these characteristics during the postoperative period, employing the sigh as a natural "test signal" to examine the respiratory muscles. In an experimental preparation that allows direct measurement of diaphragm function including length, shortening, and electromyogram (EMG) activity, we can determine whether the sigh has any effect upon diaphragm function after upper abdominal surgery. If we record abnormal movement of the diaphragm after upper abdominal surgery and the abnormality persists even during a sigh, then the diaphragm deficit may represent intrinsic, even physical, injury. However, if the postoperative dysfunction of the diaphragm is truly a reflexic inhibition, then it should be possible to temporarily override the inhibition with a more primitive and potent event such as a sigh. So we present the sigh after upper abdominal surgery as a test instrument to disinhibit the diaphragm, temporarily override a reflex inhibition, restore normal diaphragm function for at least the duration of one breath, and confirm that postoperative diaphragm dysfunction is indeed a reflex inhibition. To examine the effect of sighs upon diaphragm function after upper abdominal surgery, we implanted sonomicrometer and EMG transducers into the diaphragm of dogs by laparotomy, then directly measured costal and crural diaphragm function in the postoperative period during both resting tidal breathing and sighs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Implantation of transducers and electrodes. The project was approved by the animal care committee at the University of Calgary. Each mongrel dog had pairs of sonomicrometry transducers and bipolar fine-wire EMG electrodes implanted in left costal and crural diaphragm segments, transversus abdominis, and parasternal intercostal muscles. This technique of chronic sonomicrometry and EMG implantation, and the 7-10 d progressive recovery of diaphragm segmental shortening, has been described in detail elsewhere (2, 15). Briefly, under general anesthesia, the left hemidiaphragm was exposed through a midabdominal incision. Ultrasonic transducers were then implanted between muscle fibers on the lateral portion of the costal segment corresponding roughly to the second sternocostal branch of the phrenic nerve (16), approximately midway between the central tendon and the chest wall, and in the posterior, perivertebral region of the crural segment. On each segment, immediately adjacent to each pair of transducers, a fine-wire stainless steel bipolar EMG electrode was attached. Similarly, sonomicrometry transducers and EMG electrodes were implanted in the left transversus abdominis aligned in the same cross-sectional plane, midway between inferior costal margin and iliac crest, in approximately the internal axillary line. The parasternal intercostal sonomicrometry transducers and EMG wires were placed in the left second to fourth intercostal spaces, 2-3 cm lateral to the sternum. All implants were secured by fine, synthetic, nonfibrogenic sutures (Prolene; Ethicon Ltd.), wires were externalized, and the animals were allowed to recover. Not all implants continued to function throughout the postoperative period. A few sonomicrometry transducers twisted out of alignment as they healed in the muscles; occasionally, wire breakage limited the durability of the transducers. The number of implanted transducers actually used for measurement is noted under EXPERIMENTAL PROTOCOL.

Measurement techniques. All measurements of ventilation and respiratory muscle function were performed with the animals awake and breathing quietly while lying in the right lateral decubitus position, which placed the implanted hemidiaphragm in a nondependent position. A laboratory temperature of 18-20° C was maintained during the experiments. The animals were familiar with the location, routine, and personnel doing the recordings. Because the animals were still recovering postoperatively towards fully normal diaphragm function, the recording sessions were briefabout 23-30 min each. The animals' breathing was recorded every 2-4 d, selected at random, in the postoperative period. The animals breathed spontaneously through a snout mask connected through a one-way valve to a low-resistance open breathing circuit (< 1 cm H₂O/L/s) and a pneumotachograph (Fleisch #2) to measure airflow. On the expiratory limb, CO₂ was sampled and analyzed continuously (Model CD-3A CO₂ analyzer; AMETEK/Thermox Instruments Division, Pittsburgh, PA). Dynamic measurement within the respiratory muscles of the changing distance between the sonomicrometer transducers of each pair was provided by measuring the speed of transmission of ultrasonic waves using a sonomicrometer (Model 120, Triton Technology, San Diego, CA) (2, 17). Electromyogram signals from each muscle were amplified (AM Systems, Seattle, WA), band-pass-filtered to 100-700 Hz (Frequency Devices, MA), then rectified and moving-averaged with a time constant of 100 ms (CWE Systems, Aardmore, PA).

Using computer software for data acquisition (DataSponge; Bioscience Analysis Software, Calgary, Canada), all signals were monitored in real time on the computer display and simultaneously collected at 100 Hz to a microcomputer hard disk (PS/2 Model 80; IBM, White Plains, NY), equipped with a single-board A/D system (Model MIO-16-H-9; National Instruments, Galveston, TX).

Experimental protocol. Six animals with sonomicrometer signals of good quality in both costal and crural diaphragm segments were studied during the first 2 wk after implantation. The sonomicrometry signals were of good quality, showing a smooth continuous record of diaphragm length change without any evidence of slippage or ultrasound interference. Recordings were made in the right lateral decubitus position during relaxed, quiet breathing to gather direct measurements of length and moving average EMG of diaphragm, transversus abdominis, and parasternal intercostal during both resting tidal breaths and spontaneous sighs (breaths greater than twice the average tidal volume). Additional measurements were made of ventilatory and respiratory muscle responses to progressive hypercapnea in two animals. Hypercapnea was elicited by a modification of the Read technique (18), rebreathing 6% CO₂ 93% O₂ from a 5-L bag. These two animals tolerated the CO₂ stimulus to a lower maximum level than is typical in our laboratory. Because all these postoperative animals had sonomicrometry evidence of incomplete diaphragm recovery, we limited our CO₂ challenge to only two animals.

Analysis of Respiration

After acquisition and storage on disk, data were analyzed using software programs written by one of the authors (PAE). The flow signal was evaluated for respiratory timing and digitally integrated; inspiratory time (TI), expiratory time (TE), total breath time (Ttot), respiratory frequency (FREQ), tidal volume (VT), minute ventilation (VI), mean inspiratory flow (VT/TI), and inspiratory fraction of respiration (TI/Ttot) were calculated breath by breath.

Whole breath and intrabreath analysis. Using the recorded flow, length, and EMG signals, the software calculated whole breath ("tidal") values, as well as intrabreath development or "shape" of inspiratory airflow, respiratory muscle length, shortening, and EMG activity for each breath. These calculations have been described in detail elsewhere (2, 15). Briefly, the onset of inspiration was determined from the air flow tracing; from the sonomicrometry data, the computer algorithm then identified the resting length of the inspiratory muscles at end expiration (expressed in mm and titled L_BL, i.e., baseline length) before inspiration. From this resting end-expiratory length, the shortening for each breath was expressed as a percentage change from resting length, entitled %L_BL. Similarly, resting length of the expiratory muscle, transversus abdominis, at end inspiration was measured as L_BL and expiratory shortening was calculated and expressed as %L_BL. Moving average EMG activity was quantified arbitrarily per breath as the maximum difference in volts between baseline EMG (EMG_BL) and the peak value of the moving average signal, entitled EMG_DIFF.

These measurements defined whole-breath activity of inspiratory flow and respiratory timing, length, shortening, and EMG activity of costal and crural diaphragm, transversus abdominis, and parasternal intercostal muscles. During resting ventilation, any breath for which calculated tidal volume was greater than twice the average tidal volume for that animal was identified as a sigh, and the same whole-breath values of length, shortening, and EMG activity were calculated for each muscle. The EMG characterization of a sigh as a breath-on-a-breath (13, 19) was not employed because of concerns about the validity of moving average EMG in the postoperative period (2).

After the calculations of whole-breath values, the computer algorithm calculated the intrabreath development of "shape" of inspiratory airflow, respiratory muscle length, shortening, and EMG activity. Briefly, for each whole breath, the peak inspiratory airflow, maximum segmental shortening, and peak EMG activity were identified. Then the percent of maximum activity was determined after each successive 5% "slice" of the total time of the breath (%Ttot), producing a bin analysis of intrabreath activity based upon Ttot. This created a profile of shortening and EMG for each breath, normalized as a percentage of the peak whole-breath values and standardized over time as 20 intervals of 5% Ttot. These breath profiles were calculated for each dog during both sighs and resting tidal breaths.

Statistical analysis. Mean values were exported to spreadsheet software for review (Microsoft Excel; Microsoft, Redmond, WA), to graphic software for creating Figures 1-3 (SigmaPlot version 5.0; San Rafael, CA), and to the PC version of SAS (SAS version 6; SAS Institute, Cary, NC) for statistical analysis (22). Values for tidal breaths were averaged, then the mean values for parameters of breathing pattern, shortening (%L_BL), and EMG activity (EMG_DIFF) were compared to sighs by Student's paired t test. Because the sample size was small (n = 6 or less), all p values are reported as exact probabilities; we did not select any p value as an arbitrary threshold of significance.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Crural shortening and EMG during tidal breathing and sigh after laparotomy. Raw traces of crural diaphragm length, moving average EMG, and inspiratory airflow (upper, middle, and lower panels, respectively) during tidal breathing and sigh. Recorded from one animal 9 d after laparotomy. The y axis shows length in mm, EMG in volts, and flow in L/s for upper, middle, and lower panels, respectively. Thick horizontal line under crural EMG trace shows the time to peak EMG for tidal breath preceding the sigh. Dotted vertical line separates the initial breath (b) and second, breath-on-breath (B) portions of the sigh.

View larger version (19K): [in this window] [in a new window]   Figure 2.   Crural shortening during tidal breath and sigh after laparotomy. Normalized intrabreath shortening of crural diaphragm during tidal breathing and sigh from single animal recorded 5 d after laparotomy. The y axis shows mean values of crural shortening expressed as %LBL. The x axis shows mean values per 5% of total breath time (%Ttot). Inspiration (INSP) and expiration (EXP) were determined by airflow. Upper trace shows tidal breath; the lower trace shows a sigh.

View larger version (19K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 3.   Costal shortening during tidal breath and sigh after laparotomy (left); normalized intrabreath shortening of costal diaphragm during tidal breathing and sigh from one animal recorded 6 d after laparotomy. (The y axis shows costal shortening expressed as %LBL.) Costal EMG during tidal breath and sigh after laparotomy (right); normalized intrabreath moving average EMG values of the costal diaphragm. (The y axis shows mean EMG values in volts.) Inspiration (INSP) and expiration (EXP) were determined by airflow. Upper traces show tidal breaths; lower traces, sighs.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Diaphragm Dysfunction after Laparotomy

Laparotomy and implantation of the six animals was completed without operative or postoperative complication. The dogs were awake and ambulatory within 3-6 h of the operation and freely active and feeding normally within 1-2 d, before any laboratory recordings. In the six dogs (mean weight, 29.5 kg; range, 25.5-37.0 kg), recordings were made during periods of tidal breathing, which included sighs. Because the time to recover normal diaphragm segmental function after laparotomy is typically 10-14 d or longer in some animals (2, 15), these dogs were studied at an average of 8.7 d (range, 1-16 d) after laparotomy implantation of sonomicrometry transducers and EMG electrodes, before the diaphragm had recovered. Length change of the diaphragm segments during inspiration on these early study days showed the same recognizable patterns of motion that have been described in a previous study (2). Rather than simple inspiratory shortening, during tidal breathing two of the six dogs showed biphasic motion: lengthening then shortening of the costal diaphragm segment. Two others had the same biphasic lengthening movement of their crural diaphragm segment, while a single animal revealed simultaneous biphasic movement of both costal and crural segments during inspiration. The sixth animal had uniphasic lengthening of the crural segment and shortening of the costal segment during tidal inspirations. We confirmed that both costal and crural diaphragm segments gradually recovered to normal uniphasic shortening during tidal inspiration in all these animals when measured 3-4 wk after laparotomy, as noted in the previous study (2).

Identification of Sighs

Using a working definition of sigh as a breath greater than twice the average tidal volume, spontaneous sighs were observed in each animal in the days following laparotomy; in some animals, even on the first postoperative day after laparotomy. During the early recordings, sighs were observed infrequently (approximately one sigh per 20 min of recording time), compared with the number of sighs seen by 3-4 wk after implantation. In total, approximately 15-20 sighs were measured from each animal. Among these, we were able to record in each animal 1-4 sighs that occurred shortly after laparotomy, at a time when costal or crural diaphragm motion was obviously abnormal, showing lengthening or biphasic lengthening then shortening. For each animal, we selected one of these dysfunctional sighs for analysis. We used these criteria: (1) sonomicrometry recordings of diaphragm length were absolutely stable and free of interference; (2) tidal breathing before and after the sigh was regular for at least 20 breaths; and (3) there was no body movement or any change in position of the animal. Thus, in each animal we were able to identify and analyze a sequence of 20 tidal breaths followed by a sigh within a few days after laparotomy, when diaphragm function was abnormal as shown by a failure of the diaphragm to shorten during inspiration.

Breathing pattern for the postlaparotomy sighs and the preceding resting tidal breaths is summarized in Table 1. The postlaparotomy sighs had a significant larger tidal volume than tidal breaths (p = 0.002). Mean postlaparotomy sigh VT was 1.21 L, an increase of 324% from resting VT. Mean Ttot lengthened (p = 0.049) from 2.48 s to 6.99 s, with a wide range from 3.4 to 14.7 s. Although TI lengthened significantly (p = 0.003) for the sighs, not all sighs were followed by long pauses, so the lengthening of TE was less significant (p = 0.107).

Diaphragm Shortening during Sighs after Laparotomy

Because costal and crural diaphragm segments showed both lengthening and shortening during inspiration in the postoperative recovery period, we calculated both shortening, expressed as %L_BL, and lengthening, expressed as %L_BL (i.e., negative shortening). Table 2 summarizes diaphragm length and shortening during tidal breathing and sighs for the costal and crural segments. Costal diaphragm shortening during sighs increased dramatically, going from only 0.85 %L_BL to 12.11 %L_BL, an increase of 1,425% above tidal breathing (p = 0.07). This change occurred in the space of a single breath, then reverted back to tidal-breath values immediately after the sigh. There were similar huge increases in crural segmental shortening, which increased suddenly by 777% above tidal levels, from 1.43 %L_BL to 11.11 %L_BL (p = 0.093). Meanwhile, both costal and crural segments retained a small amount of lengthening during sighs, comparable to the amount seen during tidal breathing. Besides these changes in amplitude of shortening, the shape and pattern of shortening changed abruptly with the sigh.

Although diaphragm muscle segments showed abnormal biphasic or lengthening motion, transversus abdominis and parasternal intercostal showed only uniphasic shortening in all dogs where the signals were present (5 and 3 dogs, respectively). During sighs, parasternal intercostal shortening increased 131% from tidal breathing (p = 0.082). Transversus abdominis increased shortening by 219% during sighs, from 2.34 to 5.16 %L_BL.

Diaphragm Electromyographic Activity after Laparotomy

Of the six animals reported here with measurable diaphragm dysfunction after laparotomy, EMG recordings of activity in costal and crural segments were available in five. We recorded and analyzed costal EMG in each of the three dogs noted earlier to have abnormal biphasic motionlengthening then shortening of their costal diaphragm. For the crural diaphragm, of the four animals with abnormal biphasic or complete lengthening of their crural segment, we were able to record crural EMG from three. The durability of signals from implants in other muscles was not perfect; transversus abdominis and parasternal intercostal EMG activity was recorded successfully from 3 to 4 animals, respectively.

Even though the corresponding diaphragm segment moved abnormally after laparotomy during resting tidal breathing, with either biphasic lengthening and shortening or phasic lengthening, we still recorded regular phasic inspiratory EMG activity from each of the dysfunctional costal and crural segments. These values are listed in Table 2. With the appearance of a sigh in the postlaparotomy period, the difference in peak EMG (EMG_DIFF) of the abnormal costal segment then jumped 271% from tidal breath values, increasing from 2.70 to 7.31 volts for the sigh (p = 0.011). Similarly, the crural diaphragm EMG_DIFF jumped 216% from tidal breathing when the animal sighed in the postoperative period, from 2.89 to 6.24 volts, although statistically the result was less (p = 0.067) than for costal, and the change did not quite reach usual levels of statistical significance. From the other inspiratory muscle, parasternal intercostal, EMG_DIFF increased 178% during a sigh compared with tidal breathing (p = 0.002), while for the expiratory muscle, transversus abdominis, EMG_DIFF changed nearly an identical amount, 177% (p = 0.126).

We did observe a compound breath or "breath-on-breath" pattern in the EMG from many of these postoperative sighs. Raw traces of inspiratory airflow, crural diaphragm length, and moving average EMG are shown for one typical sigh in Figure 1. This EMG was recorded 9 d after laparotomy. During tidal breathing the crural diaphragm moved very abnormally; during inspiratory airflow the primary motion of the crural diaphragm lengthened about 2%, with almost no shortening. That motion changed abruptly to a crural shortening of about 6% during the sigh, then reverted immediately to abnormal lengthening with the next tidal breath. As noted in the figure, it appears that the latter, normal shortening portion of the sigh corresponds to the "breath-on-breath": the second, compound breath that completes a sigh. This sigh appears to be an initial, abnormal tidal breath, combined with a second, normal "breath-on-breath."

Diaphragm Motion during Sighs after Laparotomy

Normalized intrabreath profiles of shortening and moving average EMG showed the relative activity of each diaphragm segment dynamically throughout each breath. These profiles provided additional information about muscle performance in the postoperative period that could not be deduced from simple peak breath values.

An intrabreath profile of segmental shortening of the crural diaphragm segment from one representative sigh is shown in Figure 2. That sigh was recorded 5 d after laparotomy, at a time when the crural segment showed phasic lengthening during tidal breathing. During the sigh, the crural diaphragm switched suddenly to significant shortening compared with tidal breathing. This intrabreath difference was not seen just with shortening; intrabreath profiles of segmental shortening and corresponding EMG activity for a costal diaphragm segment from another canine, recorded 6 d after laparotomy, are shown in Figure 3. Despite costal biphasic lengthening and shortening phasic inspiratory EMG activity was recorded during tidal inspiration (Figure 3, top). During the sigh, both shortening and moving average EMG of the costal diaphragm increased significantly compared with tidal breathing (Figure 3, lower panels).

Sighs Compared with CO₂-stimulated Breaths of Equivalent Tidal Volume

Sighs were compared with large breaths of equivalent tidal volume that were elicited by CO₂-stimulated breathing in two animals. Because all these postoperative animals had sonomicrometry evidence of incomplete diaphragm recovery, we did not attempt the CO₂ challenge in the other animals. We present this limited CO₂ challenge data as a case study. From the recording of CO₂ rebreathing, we identified 4-10 breaths of similar tidal volume to the sighs that had been recorded in that animal, then calculated breathing pattern, peak shortening, EMG_DIFF, and intrabreath profiles for the large CO₂-stimulated breaths.

The first of these two animals showed biphasic lengthening-shortening of the costal segment recorded on postoperative Day 16. Tidal volumes of whole breath, sigh, and equivalent CO₂-stimulated breath were 0.36, 1.20, 0.90 L, respectively. In this dog, phasic inspiratory EMG activity of the costal segment was present during all three. Costal EMG_DIFF increased 395% from tidal breathing to sigh and 314.1% to the CO₂-stimulated breath. Maximum lengthening of the costal segment was 0.47, 0.38, and 0.04 %L_BL during tidal breath, sigh, and stimulated breath, while peak shortening was 0.34, 7.97, and 3.80 %L_BL, respectively.

The other animal was recorded on postoperative Day 9 and showed uniphasic lengthening of the crural segment during tidal breathing. Tidal volumes of tidal breath, sigh, and stimulated breath were 0.26, 0.77, 0.50 L, respectively. Crural EMG_DIFF increased 281% from tidal breathing to sigh and 158% from tidal breath to stimulated breath. In this dog, lengthening of crural segment was 1.43, 1.10, and 2.27 %L_BL during tidal breath, sigh, and stimulated breath, respectively. Shortening during the sigh increased to 5.80 %L_BL, much greater than the 3.04 %L_BL of the stimulated breath.

Intrabreath profiles of shortening of the costal diaphragm of the dog recorded on postoperative Day 16 during tidal breathing, a sigh, and a large CO₂-stimulated breath, are shown in Figure 4. Because the tidal volume of the sigh was 1.2 L, we selected a breath with the largest tidal volume (0.9 L) recorded during CO₂ stimulation for comparison. The dysfunctional postoperative costal diaphragm showed biphasic lengthening-shortening of small amplitude during tidal breathing (upper panel). A similar abnormal biphasic motion with small-amplitude shortening persisted during CO₂-stimulated breathing (middle panel). In contrast to the CO₂-stimulated breath, the sigh showed a sudden, completely different, and apparently normal uniphasic shortening movement of large amplitude. As demonstrated in this figure, increasing tidal volume with CO₂ stimulation did not eliminate abnormal diaphragm lengthening-shortening movement in the postoperative period, although this abnormal movement disappeared during the sigh.

View larger version (26K): [in this window] [in a new window]   Figure 4.   Crural shortening during tidal breath, CO2-stimulated breath, and sigh. Normalized intrabreath shortening values during tidal, CO2-stimulated breaths, and a sigh, in upper, middle, and lower panels, respectively. Traces were recorded from one animal 16 d after laparotomy. Costal shortening is expressed as %LBL. Inspiration (INSP) and expiration (EXP) were determined by airflow.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Spontaneous sighs occurred in the days following upper abdominal surgery when diaphragm function had not yet returned to normal (2). Direct measurement of diaphragm length revealed that shortening increased abruptly with the appearance of a sigh, compared with preceding tidal breaths. Despite the evidence of abnormal diaphragm contractility in this postoperative period, moving average EMG showed phasic inspiratory activity at the same time as the diaphragm was contracting abnormally.

Diaphragm Dysfunction after Laparotomy

This investigation was based on the presumption that diaphragm function after upper abdominal surgery is abnormal. Indeed, postoperative diaphragm dysfunction is usually considered to underlie typical postoperative pulmonary complications such as atelectasis and basilar pneumonia (3). Ford and colleagues (3) showed that after upper abdominal surgery, indicators of reduced diaphragm function reverted toward normal within 24-48 h after surgery, while other investigators have suggested that diaphragm dysfunction persisted for approximately 1 wk and was not suppressed by postoperative pain relief (4).

This study also presumes that the process of laparotomy with implantation of these sonomicrometry transducers into the diaphragm mimics the adverse respiratory effects of human upper abdominal surgery. In animals and humans, either cholecystectomy or mechanical deformation of abdominal viscera provokes changes in breathing pattern (4, 6). In a previous animal study we saw that a sham operation with an identical incision and anesthetic, but no peridiaphragmatic contact or implantation, did not elicit any postoperative dysfunction (2).

The etiology of the diaphragm dysfunction has been the focus of considerable speculation and investigation. One recent study demonstrated that intrinsic diaphragm contractility, as reflected in maximum transdiaphragmatic pressure generation, was not different before and after upper abdominal surgery, suggesting that diaphragmatic dysfunction was secondary to some inhibition of phrenic nerve output (1). This idea is consistent with Prabhakar's observations that inspiratory stimulation of the mesenteric region in cats produced a shift from diaphragmatic to intercostal breathing, while expiratory stimulation decreased expiratory muscle activity (8). Laparotomy may also adversely affect expiratory muscles, including transversus abdominis (23). Recent human clinical studies show that the abnormal breathing pattern after surgery can be improved and diaphragm EMG increased by epidural anesthesia of the thoracic roots with local anesthetic (6, 24). Cumulatively, this evidence supports a hypothesis introduced as early as 1908 that postoperative complications including atelectasis are related to reflex inhibition, not structural injury, of the diaphragm (7).

In our first chronic implantation of sonomicrometry transducers into the costal and crural diaphragm (2), we were able to measure postoperative diaphragm contractility directly and noted in awake, nonanesthetized dogs that diaphragm shortening did not stabilize until 10 d after surgery. This finding contrasted with the breathing pattern, which stabilized by 4 d after surgery. During the recovery process, the diaphragm segments displayed a sequence of segmental motions: first lengthening, then biphasic inspiratory lengthening-shortening, and finally simple shortening. In two of the animals in this project, abnormal biphasic inspiratory lengthening-shortening was still identified 15-16 d after surgery, even though indices of breathing pattern had returned to normal. This illustrates that the duration of abnormal respiratory muscle function after upper abdominal surgery is occasionally long and shows large variation among individuals.

Postoperative Diaphragm Dysfunction and Moving Average Electromyogram

It is important to recognize in the early postoperative period that diaphragm EMG may simply not be a reliable indicator of diaphragm function. This was shown clearly in an earlier study (2) where moving average EMG of the costal diaphragm was apparently normal, despite the fact that the costal diaphragm was seen clearly to be lengthening, with no evidence of purposeful function. Again in this study, phasic inspiratory diaphragm EMG was regular, even from diaphragm segments where contractility was obviously abnormal. This inconsistency has important implications. The presence of regular phasic electrical activity does not ensure that there is any real contractility of the diaphragm. The presence of this "paradoxical" phasic EMG could be ascribed to a measurement artifact, since recent implantation may induce local edema that could change interelectrode impedance and confound EMG recordings. However, persistence of significant edema for 15-16 d after laparotomy is unlikely. This false normality of the postoperative EMG may arise from the characteristics of moving average signal processing. Moving average EMG reliably represents the amplitude of the EMG signal but does not convey information about other characteristics of the electrical signal.

The Sigh as a Test Signal

We based this study on the sigh as a relatively independent automatic breath, which might be capable of overriding some pre-existing diaphragm inhibition of local origin. The sigh has been investigated under various names by previous authors (9, 14, 25, 26). There is no absolute quantitative definition of what constitutes a sigh (9, 25, 27). Because we were unsure about the amplitude that a sigh might exhibit in the postoperative period, we designated the sigh conservatively as a breath greater than twice the average tidal volume. The frequency of spontaneous sighs has been investigated in humans and estimated at approximately 9-10 sighs/h in healthy young adults during wakefulness (9, 30, 31), and less than 2 sighs/h during sleep (28). We are less certain about sigh frequency or volume in dogs (32); sighs have been reported to occur at a rate of 1.5- 3/h during wakefulness (33).

The size and frequency of sighs imply that these events serve some important function for the respiratory system. Pulmonary compliance and functional residual capacity increase after sighs (25), so sighs may be useful for maintaining alveolar stability, preventing collapse and subsequent physiological and anatomical shunting (27, 29). Sighs persist after bilateral vagotomy (13), and they depend on both chemical stimulation and mechanical feedback (14, 26). A sigh may be viewed as a type of augmented breath, or a breath-on-a-breath with an early phase of inspiratory flow similar to normal tidal breathing, followed by a second phase or compound breath that begins where a normal inspiration would peak (10, 14). This double or compound breath is accompanied by a similar pattern of electrical activity in phrenic nerve and diaphragm (13, 19). Although the exact genesis is not certain, apparently sighs originate centrally. Extracellular recordings from several types of respiratory neurons in the ventral and dorsal respiratory groups (DRG and VRG) in the brainstem of unanesthetized cats all showed activity related to sighs (12). From all this, we can conclude that sighs are spontaneous, automatic, identifiable breath events originating centrally, which we used as a test signal.

Respiratory Muscle Activity during a Sigh

Several studies have measured EMG activity of costal and crural diaphragm segments during sighs. In anesthetized dogs, Van Lunteren and colleagues noted that the peak EMG activity of the diaphragm during spontaneous sighs was 214% of the peak values during control breaths (21). Reduced expiratory muscle activity following sighs might help increase end-expiratory lung volume and functional residual capacity. In cats (20), although diaphragm EMG increased 250% during sighs, the corresponding EMG activity of the expiratory muscle triangularis sterni showed a delayed onset and significant reduction.

Sighs after Laparotomy

During the postoperative period, we were able to identify sighs without difficulty by their characteristic flow shape, long pause, and large tidal volume. Spontaneous sighs occurred regularly despite clear evidence of diaphragm dysfunction in these animals as shown by abnormal diaphragm motion and contractility. Of course, if sighs function to increase functional residual capacity, we might anticipate plenty of sighs in a postoperative period characterized by poor diaphragm motion. The sighs in these postoperative animals were equivalent in size to typical spontaneous sighs measured in other studies, with very large VT and significant increases in TI, Ttot, VT/TI. To achieve this normal sigh size in the postoperative animal, either activity and shortening of other respiratory muscles had to increase greatly to compensate for diaphragm dysfunction that persisted during the sigh, or the diaphragm itself had to suddenly normalize contraction during each sigh. Our results confirm the latter. The diaphragm dysfunction of resting tidal breathing abruptly resolved with the appearance of the sigh; that is, the sigh momentarily relieved diaphragm inhibition. As noted in Figure 1, where a "breath-on-breath" EMG pattern was noted for postoperative sighs, it seemed to be the second, compound breath component of the sigh that was abruptly normal.

Finally, in this study, both EMG activity and shortening of the expiratory muscle transversus abdominis increased significantly during the sigh after laparotomy. This result contrasts with other investigations involving anesthetic, which showed less expiratory activity with sighs (20, 23).