None of the participants was current- or ex-smoker to avoid the impact of smoking on the lungs (chronic obstructive pulmonary disease or emphysema). None of the subjects had peripheral neuropathy.
Overall, our mean and normative values of DCMAPs parameters were comparable to those reported in the literature. Differences are mostly related to variations among studies in the site of stimulation or signal recording or how the parameter was measured.
The mean latency values of sternal potentials [7.08 ms (inspiration) and 7.14 ms (expiration)] are higher than values reported by Chen et al (6.5 ms) , Maranhão et al [6.12 (inspiration) and 6.42 ms (expiration)] , Resman-Găspěrsč and Podnar [6.55 (inspiration) and 6.59 ms (expiration)] , Vincent et al (6.59 ms) , and Dionne et al (6.6 ms) , They all stimulated the nerve at a lower level (supraclavicular fossa).
The mean latency values of LCW potentials [7.29 ms (inspiration) and 7.34 ms (expiration)] are close to values reported by Luo et al (7.3 (left side) and 7.6 ms (right side) , Dionne et al (7.14 ms) , and Mckenzie and Gandevia [7.68 (right side) and 7.92 ms (left side)] . They adopted the same stimulation and recording montages as the current work.
Our upper limits of the latency [8.34 (sternal/inspiration), 8.48 (sternal/expiration), 8.63 (LCW/inspiration), 8.64 ms (LCW/expiration)] are comparable to values by Vincent et al [(8.36 ms (sternal)]  and Resman- Găspěrsč and Podnar [7.92 ms (sternal/inspiration and expiration)] . They are higher than values reported by Maranhão et al (6.34 ms (sternal/inspiration) and 6.42 (sternal/expiration) . Chen et al found that the suggested normal latency limit is <8 ms (sternal) . Variation in limit values among studies is again related to variation in stimulation and recording sites.
Our mean amplitude values are higher than values reported in the literature [13-18, 20]. This is because we measured the amplitude from peak-to-peak as opposed to other studies where the amplitude was measured from onset-to-peak. Highest mean amplitude values were recorded from over the sternum (1.4 (inspiration) and 1.16 mV (expiration). Lower limits of the amplitude [0.60 (sternal/inspiration), 0.46 (sternal/expiration), 0.34 (LCW/inspiration), 0.32 mV (LCW/expiration)] are comparable to values by other studies (range: >0.3-0.7 mV) [13-15]. Swenson and Rubenstein , Johnson et al , and Vincent et al  on the other hand reported much lower normative values [0.1, 0.12, and (0.14 right and 0.11 left) mV respectively). As illustrated by Maranhão et al , the wide range of phrenic nerve amplitude creates a great problem in determining a lower normal limit.
Mean and normative values of the duration of sternal DCMAP in the current work [14.54 and 20.82 ms (inspiration) and 17.74 and 24.12 ms (expiration)] are within the range of values reported by Chen et al , Resman- Găspěrsč and Podnar , Maranhão et al , and Vinvent et al.  They all recorded the potentials from over the sternum. No data is available in the literature on the normative value of the duration of potentials recorded from over the LCW.
Mean values of the area of sternal DCMAP [7.27 mV.ms (inspiration) and 8.12 mV.ms (expiration)] are close to values in the studies by Chen et al , Resman-Găspěrsč and Podnar , and Maranhão et al.  Our normative values [2.47 mV.ms (inspiration) and 3.32 mV.ms (expiration)] are however lower than other studies (lower limit value > 4 mV.ms) [13-15]. This variation is mostly related to how the limit normal was determined (mean-2SD as in the current work compared to 5th percentile limit in other studies). Our values are higher than that by Vincent et al  (mean= 3.05 mV.ms and limit normal=0.87 mV.ms). Lower amplitude values have contributed to lower area values in their study.
The means values of the area of DCMAP from over LCW ([5.59 mV.ms (inspiration) and 6.19 mV.ms (expiration)] are comparable to value by Dionne et al (6.41 mV.ms) . Data on the normative values from costal diaphragm is scanty in the literature.
Our mean and normative values of the interside differences in latency, amplitude, duration, and area were presented. These values are useful in assessment of patients with a unilateral phrenic nerve lesion as those caused by trauma or surgery especially when the absolute values of both sides are within the normal range. Our values are consistent with those in the studies by Chen et al  and Resman-Găspěrsč and Podnar.  Vincent et al  reported higher interside differences in the latency, amplitude, and duration of potentials recorded from over the sternum. Swenson and Rubenstein  also reported higher mean interside amplitude differences of potentials recorded from over the sternum and LCW which may be related to differences in the placement of recording electrodes.
Differences between sternal and LCW DCMAPs:
Comparative studies showed that the latency, amplitude, and area varied significantly between sternal and LCW potentials. The duration did not differ between the two recording sites.
The latency of sternal potentials was significantly shorter than LCW potentials. Swenson and Rubenstein  reported similar finding. This can be expected considering the shorter length of the anterior (sternal) branch. McKenzie and Gandevia  in their illustrative work provided the distance from the branching points of phrenic nerve to different recording sites, which should correspond to the length of different branches. Distance corresponding to the anterior branch was shorter than that for lateral branch on the right (6.7 and 13.8 cm) and left sides (2.8 and 8.8 cm respectively).
The amplitude of sternal potentials was significantly higher than LCW potentials. The finding is consistent with that by Dionne et al  who recorded diaphragmatic potentials from over six different sites. The highest amplitude was obtained from over the sternum. They related this finding to the orientation of the electrical dipole of the diaphragm (different from limb muscles). They assumed that G1 and G2 are both active, with the DCMAP representing out-of-phase summation of opposite polarity activity at the two electrodes, and this in part has accounted for larger amplitude at the sternum as it was the only technique with G1 positioned above the xiphoid.
Amplitude difference between sternal and LCW potentials can also be due to variation between the two sites in the distance between the active and reference recording electrodes  with higher amplitude of sternal potentials being related to longer distance between recording electrodes compared to LCW recording site (16 and 3.5-5 cm respectively).
Recording of DCMAP from over the sternum was recommended by Chen et al  and Dionne et al  as this electrode position gives the maximum amplitude. It also allows the easiest and most rapid study to perform as it does not involve rib counting or multiple electrodes repositioning especially in obese subjects and in patients in intensive care unit where chest tubes and catheters are frequently encountered.
Contrary to our finding, Swenson and Rubenstein  found smaller but easily recorded potentials at the sternum. Maximum peak was recorded over the anterolateral chest at the intersection of the axillary line with transverse plane through the xiphoid. In their study both xiphoid and costal recordings had their reference electrodes placed below the umbilicus which may have accounted for their results.
The duration did not differ significantly between sternal and LCW potentials. This parameter reflects the range of conduction velocities of conducting nerve fibers  and is not expected to differ between DCMAPs recorded from over the two sites. Significantly larger mean area of sternal potentials is consistent with significantly higher amplitude values of sternal compared to LCW potentials with no difference in the duration between them.
Despite the given advantages of recording from over the sternum (higher amplitude and easier study with no need for rib counting or repositioning of the electrodes), [13, 18] it cannot be concluded that this montage should be the standard one for recording DCMAPs.
The diaphragm is a specialized muscle that demonstrates distinct muscular subvolumes (neuromuscular compartments) in which the intramuscular phrenic nerve distribution (branching) is confined . Recording electrodes at each recording site are relatively selective for the subjacent portion and record activity from the underlying portion of the diaphragm  that is innervated by separate nerve branches .
The presence of significant differences between potentials recorded from over the sternum and lateral chest wall in the current work and in other studies [17, 18, 20] as well as the evidence by some electrophysiological studies that one branch but not the other can be affected in different medical conditions [33, 34] highlights the importance of assessing the conduction along different nerve branches in every case referred for electrophysiological evaluation. It also highlights the need to provide normative data for potentials recorded from different diaphragm subvolumes innervated by different branches of the nerve.
Differences between DCMAPs recorded during inspiration and expiration:
We found variation in DCMAP parameters with respiratory cycle. The mean latency, duration and area were significantly lower, and the mean amplitude was significantly higher during inspiration compared to expiration. Findings are mostly related to changing lung volumes during movement of the diaphragm in the respiratory cycle and changing physiological properties of the diaphragm during contraction [13, 35].
Similar to our finding, Maranhão et al  found shorter latency of DCMAPs during inspiration. Resman-Găspěrsč and Podnar  found significant difference in peak but not the onset latency between inspiration and expiration. DCMAPs were recorded from over the sternum in both studies. Latency of CMAP represents the summated durations of impulse propagation along the nerve fiber, time delay across the neuromuscular junction, and depolarization time across the muscle . Increased muscle fibers conduction velocity during contraction (inspiration) therefore decreases the overall time needed for impulse propagation and hence decreases the response latency.
Significantly higher amplitude, shorter duration and smaller area during inspiration come in agreement with the results of different studies. Resman-Găspěrsč and Pondar  found similar changes in the amplitude and duration during inspiration. Changes in the area were however insignificant. Chen et al  reported increase in the amplitude and decrease in the duration with inspiration. Maranhão et al  demonstrated the same results although they did not conduct a statistical comparison. On the other hand, Swenson and Rubenstein  reported higher DCMAP amplitude during expiration. They recorded the potentials at the level of the 7th intercostal space with the reference electrode below the umbilicus.
The amplitude value of diaphragmatic potentials depends on the orientation of recording electrodes and the distance between recording electrodes and the diaphragm which change during the respiratory cycle . Chen et al suggested that during inspiration the diaphragm flattens which changes the angle, the moving dipole meets at the recording electrodes  which according to the theory of volume conduction would result in increased CMAP amplitude . Resman-Găspěrsč and Pondar added that shortening (shorter path) and thickening (faster conduction) of diaphragmatic muscle during contraction make larger amplitude and shorter conduction time and shorter duration .
In a similar context, it was proved by musculoskeletal ultrasound that thickness of the diaphragm in healthy individual increases by 28-96% during inspiration relative to expiration [37, 38]. This in turn increases conduction velocity of muscle fibers due to diminished resistance of thick fibers and improves synchronization of the electrical potentials. In addition, muscle fibers shortening bring them closer to the recording electrodes, thus muscle contraction increases temporal and spatial summation of muscle fibers action potentials leading to increased CMAP amplitude and decreased duration .
We found significantly smaller area during inspiration due to significant decrease in the duration that overweighed the increase in amplitude during inspiration. Maranhão et al  demonstrated decrease in the area of DCMAPs during inspiration. They did not however conduct a statistical comparison. Resman-Găspěrsč and Pondar  found that difference in the area between inspiration and expiration is not significant, still smaller during inspiration. They considered the area to be more useful than the amplitude due in part to its insensitivity to respiratory cycle. The current study was conducted on larger number of healthy individuals (70 subjects) compared to that by Resman-Găspěrsč and Pondar  (29 healthy subjects). It is evident that there is an assent among studies that the area of DCMAPs is smaller during inspiration. The variation is about the extent of the difference whether it is significant or not. The exact change of the area with respiration is to be further evaluated.
Differences between DCMAPs recorded on right and left sides:
We found significantly longer latency of diaphragmatic potentials on the right side. Mier et al  reported similar finding but did not mention an explanation. Delhez et al  and Katayama et al  on the other hand showed longer latency on the left side. They attributed their finding to longer anatomical course of the left phrenic nerve. Maranhão et al , Mckenzie and Gandevia , Vincent et al (16) and Chen et al  did not find significant differences between both sides.
The course as well as the length of the right and left phrenic nerves varies. Jiang et al measured the full length and the length of thoracic part of the phrenic nerve on both sides and both were shorter on the right side. The intramuscular branches were however longer on the right side in higher number of their studied corpses . McKenzie and Gandevia measured the conduction distance from entry point of phrenic nerve branches into the diaphragm to motor points adjacent to different recording sites, which should correspond to length of intramuscular branches. All distances were longer on the right side . This can be expected given that the right phrenic nerve has almost as straight course. It enters the diaphragm close to the esophageal hiatus [27, 42], thus at a point at longer distances from the anterior and lateral chest walls than the left phrenic nerve. The conduction along these branches (thin) is slower than conduction along the nerve trunk . Both factors (the length and conduction velocity along the nerve trunk and the branches) may have accounted for the longer latency of the recorded potentials on the right side.
The amplitude of DCMAPs (except for sternal potentials during expiration) was significantly lower on the right side. Resman-Găspěrsč and Podnar similarly found lower amplitude of sternal potentials on the right side during inspiration but not expiration and attributed the finding to the higher position of right hemidiaphragm, which is lifted by the liver (longer and thinner muscle ﬁbers) . Absence of a difference between right and left sternal potentials during expiration can be expected because at the xiphoid and during expiration, anatomical and physiological differences between the two sides are kept to minimum.
The duration of diaphragmatic potentials was significantly longer on the right side. Given that electrical potentials can propagate easily through thickened muscle fibers with lower resistance, (Kim BJ) it can be speculated that the duration of DCMAPs from the right hemidiaphragm (thinner fibers) is longer than the left. The area did not differ between the right and left sides. Similar findings were found by Resman-Găspěrsč and Podnar . This is quite expected as lower amplitude and higher duration on the right side are equated by higher amplitude and lower duration on the left side.
Despite the difference in DCMAP parameters between the right and left sides, we found strong/highly strong right-to-left correlation in all parameters (P<0.001) indicating that if one side deviated from the mean value the other side tended also to deviate in the same direction. Mier et al also found significant correlation between right and left phrenic nerve conduction time (r = 0-81, p < 0-001) . Swenson and Rubenstein found constancy of only the onset latency. They did not find right-to-left amplitude correlation . Similar to their findings, Maranhão et al did not find consistent right-to-left correlation . The presence of strong/highly strong right-to-left correlation indicates that in unilateral lesion, the opposite side would serve as a reasonable standard for comparison.
Differences between DCMAPs recorded in men and women:
Measured parameters were different between men and women. The mean latency, amplitude, and area of DCMAPs were significantly increased in men. The duration did not differ significantly between both sexes. Maranhão et al reported substantial differences in amplitude, latency, and duration of DCMAPs between men and women . Resman-Găspěrsč and Pondar  and Vincent et al  found significantly higher amplitude in men. Vincent et al  found that the latency was different after adjustment for age and BMI.
Differences between men and women in parameters of DCMAPs are mostly related to variations in the anthropometric measures between both sexes (longer nerves contributing to longer latency and greater muscle mass contributing to higher amplitude in men) [14, 16]. Larger area in men is due to significantly higher amplitude and longer duration (not to significant level) in men compared to women. Only the duration of the DCMAPs did not differ significantly between men and women. This finding is expected given that the duration reflects range of conduction velocities of nerve fibers  which should not differ by gender.
The correlation and regression analyses between different parameters of phrenic nerve conduction study and subjects’ demographic and anthropometric data
Correlation and regression analyses revealed that latency of DCMAPs is significantly related to age which is consistent with data from several studies [13-16] and can be attributed to dropout of largest fibers, segmental degeneration, and reduced internodal length with aging [43, 44]. No correlation was found between age and other parameters. The same was reported by Maranhão et al  and Vincent et al  and can be explained by the exclusion of elderly individuals from enrollment in the study (oldest individual was 61 years). Prominent electrophysiological changes in amplitude, duration, and area are seen in individuals above the age of 60-65 years .
Regression analysis showed significant relation between amplitude and area of DCMAPs and gender. The finding is consistent with the presence of significant differences between men and women in the amplitude and area of DCMAPs and can be explained by larger muscle mass in men. The latency of LCW (but not sternal) potentials was also related to gender. The contribution is indirect and is in part related to longer nerves in men (more evident for the lateral branch).
The latency of DCMAPs was also found to correlate with height which can be explained by increased length of nerves with relative conduction slowing in tall individuals . Our results agree with Resman-Găspěrsč and Podnar  and Maranhão et al.  Results of the regression analysis did not however show a relation between latency of DCMAPs and height. This indicates that height is not an independent contributor to the latency of DCMAPs. McKenzie and Gandevia found that conduction distance does not directly relate to the subject’s height and conduction velocity is not uniform along the main nerve trunk . In one study, the length of the sternum was used as a surrogate for the height. Nevertheless, it was not found to correlate with the latency of DCMAPs . It is evident that the length of the phrenic nerve cannot only be represented by subjects’ height as the intramuscular branches which vary in thickness and size also contribute to the conduction distance and latency.
The duration of DCMAPs correlated with height. This is because relative conduction slowing in tall individuals decreases the synchronization of muscle fibers potentials which increases the duration of CMAP . This is further supported by the results of the regression analysis where duration of only the LCW recordings was found to be related to height. Lateral intramuscular branch is longer than sternal (anterior) branch  which should magnify the effect of increased length on the duration of the potentials.
The amplitude and area of sternal and LCW potentials (during inspiration and expiration) correlated negatively with the BMI. This is most likely due to amplitude attenuation by the thicker subcutaneous tissue in the person with higher BMI [47, 48]. Such a relation was not shown in the results of regression analysis which indicate that BMI is not an independent contributor to amplitude and area of the DCMAPs yet stills a factor. Factors as the mass of the diaphragm (not measured in the current work) may have major contribution to the amplitude and area than the BMI.
Absence of a correlation between the latency of DCMAPs and BMI is plausible since the latency represents the conduction along the fastest fibers regardless the number of axons . Thus in clinical setting, the fastest fibers appear to conduct equally in thin and obese individuals.
Chest circumference did not correlate with most of the recorded parameters. It only correlated positively with the latency of sternal potentials during inspiration and expiration and negatively with amplitude (of LCW DCMAP) and area (of sternal DCMAP) during inspiration and duration of sternal potentials during expiration. Multiple regression analysis revealed that BMI is related to the amplitude of LCW responses and the area of sternal responses during inspiration (p=0.023 and 0.013 respectively).
Our results are contradictory to that by Resman-Găspěrsč and Podnar  who found significant positive correlation between chest circumference and the CMAP area and Chen et al  who found significant positive correlation between chest circumference and the amplitude of diaphragmatic potentials and attributed this to the greater diaphragmatic muscle mass and more flattened diaphragms in persons with larger chest circumference. These contradictory findings may be because we measured the chest circumference at the level of the 4th intercostal space while Chen et al  measured the circumference at the level of the xiphoid process. Also, the intervening breast tissue in women may have accounted for differences in the measures of chest circumference between the two levels. The rib cage cross-section area may be a better measure to correlate with DCMAPs.
Participants were tested in the supine position which is characterized by larger diaphragmatic excursion than the upright position. In establishing reference values for labs or comparing patients or in follow up studies, the same decubitus must be adopted . The current required for supramaximal stimulation was less than 50 mA in the majority of subjects with pulse duration of 0.2 ms.
We stimulated the nerve at the posterior border of the sternocleidomastoid muscle at the level of cricoid cartilage. This stimulation site is easy to locate, less painful and it is easy to avoid brachial plexus co-stimulation by moving the electrode more medial [2, 23, 50]. It is to be mentioned that, the stimulation site can vary from subject to subject and may even vary between the left and the right sides of one subject due to asymmetry .
The phrenic nerve was generally easy to stimulate. Difficulties, especially on the left side, were sometimes encountered possibly due to anatomical variation of the phrenic nerve that may pass through the anterior scalene muscle or there may be medial or lateral displacement of the phrenic nerve . This difficulty was overcome by increasing current intensity and/or medial repositioning of the stimulating electrode.
The technique was also challenging, and the nerve could not be stimulated in 3 subjects with short or obese neck (despite more rotation and/or extension of the neck was tried). In clinical practice, if percutaneous electric stimulation could not be performed due to technical problems, needle electric stimulation or magnetic stimulation of the phrenic nerve is to be tried [23, 25].
Stimulation was performed during quiet respiration (at the end of inspiration and expiration). Generally, the phases of respiration may create a technical problem during performing phrenic nerve conduction study. However, if supramaximal stimulation to the phrenic nerve is repeated several times during quiet respiration and the two highest amplitude potentials are obtained, the phases of respiration are not an influential factor .