In this study of 80 never-smoking pre-ban FAs with significant history of exposure to SHS, we confirmed our previous findings in a smaller cohort that about half of the pre-ban FAs had decreased resting diffusing capacity (DcoSB) below the 95% prediction limit as well as curvilinear flow-volume curves (concave to the volume axis) and decreased air flow at mid and low lung volumes. More importantly, we found that those FAs with reduced resting DcoSB had lower exercise capacity as indicated by their maximum predicted work (watts) and maximum oxygen uptake (
). This lower exercise capacity was associated with decreased oxygen uptake per breath (O2
pulse) and decreased anaerobic threshold (anaerobic threshold was reached at lower absolute
and lower % predicted value of maximum
). In addition, the exercise capacity was directly associated with their FEV1
at baseline. Furthermore, we found that while the two groups of FAs had similar pulmonary blood flow (
), those with reduced DcoSB at rest increased their diffusing capacity with exercise (DcoWB) less with increasing workload and pulmonary blood flow, which indicates that despite exercise-induced increase in pulmonary blood flow, they had reduced pulmonary capillary recruitment. Finally, although no association between the resting diffusing capacity and SHS exposure was observed in our cohort, the diffusing capacity during exercise showed a trend association with the years of SHS exposure with the association reaching a significant level in those FAs with more than 10 years of exposure.
Our finding of decreased exercise capacity in the group of FAs with reduced resting diffusing capacity is interesting as none of FAs had any history of cardiopulmonary diseases and all reported exercising regularly 3 to 5 times weekly. Although the average maximum watts and average peak
achieved by all the FAs were within the normal reference range, the FAs with reduced resting diffusing capacity achieved significantly lower levels compared to those with higher resting diffusing capacity, and many within the former group achieved maximum work levels well below the normal reference range (25% only achieved levels below the 77% maximum predicted
). The lower exercise capacity of this group of FAs is likely due to decreased aerobic capacity as reflected by decreased
at AT and decreased O2
pulse. Since the stroke volume was similar between the two groups of FAs, the decreased aerobic capacity thus appears to be due to decreased arteriovenous (A-V) oxygen difference, which in turn suggests an abnormal distribution of oxygenated blood to non-essential tissues and reduced distribution of oxygenated blood to exercising muscle 
. We also demonstrated that FEV1
was a predictor of exercise capacity in a linear fashion. The correlation coefficient of 0.33 between exercise capacity and FEV1
reflects a medium effect size according to the criteria of Cohen 
, and thus is consistent with a physiologically meaningful association. Given the presence of curvilinearity in flow-volume loops and decreased flow at mid and low lung volumes in the cohort, this association suggests that dynamic airflow obstruction may play a role in decreased exercise capacity of the FAs.
The decreased exercise capacity and lower exercise-induced increase in diffusing capacity in the FAs with reduced resting diffusing capacity further substantiate our previous findings of abnormal resting pulmonary function tests (i.e. decreased flow in low- and mid-lung volumes, air trapping, and reduced resting diffusing capacity) in this cohort with significant history of SHS exposure. The lower increase in exercise diffusing capacity, its incremental nature based on resting diffusing capacity, and lower ratio of diffusing capacity to pulmonary blood flow during exercise suggest that these FAs have reduced pulmonary capillary bed recruitment. Further anatomic or physiologic studies such as CT imaging or assessment of the inspiratory flow-volume loop assessment during exercise could help elucidate the underlying cause.
The finding that the FAs with reduced resting diffusing capacity had a lesser exercise-induced increase in their diffusing capacity with various stratification confirms our hypothesis that even some of the FAs with resting diffusing capacity within the 95% “normal” prediction limit had reduced pulmonary capillary recruitment. In addition, it shows that the diffusing capacity during exercise is a more sensitive measure of underlying abnormalities in pulmonary capillary bed than resting diffusing capacity. Together, these results show that the observation of a normal diffusing capacity at rest may not indicate that the pulmonary capillary recruitment during exercise will be normal.
The association between diffusing capacity during exercise and cabin years of SHS exposure had a two-tailed p-value of 0.095; however, if we assume that the adverse event on lung function would only be seen with increased SHS exposure (a one-tail assumption), the association then becomes significant (one-tail p-value of 0.048). In addition, the LOWESS smoother analysis showed a significant lung function and SHS exposure association (two-tailed p-value of 0.032) in the subgroup of FAs who had a longer SHS exposure history, a finding that is supported by the plausible biologic mechanism that higher exposure could cause higher adverse events. The correlation coefficient (r) of 0.32 between diffusing capacity during exercise and cabin years of SHS exposure represents a medium effect size according to the criteria of Cohen 
, and thus is not only statistically significant, but also is consistent with a meaningful association. Although most FAs did report additional SHS exposure apart from cabin exposure, it is expected that their non-cabin related SHS exposure was relatively insignificant compared to the intensity of their exposure while aboard aircraft 
. In a previous study of this cohort, we investigated potential contribution of non-cabin related SHS exposure to subjects' lung function via comparison of the subjects with only cabin SHS exposure and those reported additional non-airline SHS exposure (i.e. exposure during childhood, adulthood, and/or non-airline employments), and did not find any significant differences in lung function between the two groups 
, which also suggest that non-cabin SHS exposure in FAs was dwarfed by their cabin SHS exposure.
Overall, our findings provide strong physiologic evidence consistent with presence of emphysema and/or COPD in this cohort of never-smoking FAs who were exposed to high concentrations of SHS for extended periods of time in the aircraft cabin. It remains unclear whether the pulmonary function abnormalities seen in this cohort are stable or progressive.
Our study has several possible limitations. First, and from a technical standpoint, performing the within breath diffusing capacity (DcoWB) measurement during exercise requires excellent self-control of breathing by the subject to maintain a low and constant expiratory flow, even if the subject is able to see and monitor flow on a screen during the maneuver. The measurement becomes increasingly difficult to perform at high workloads. To improve distribution of blood flow and ventilation-perfusion matching, we elected to perform our studies in the supine posture 
. We aimed to attain maximum recruitment of the pulmonary capillary bed and diffusing capacity at a lower work rate during supine exercise rather than in the erect posture, thus easing the demands on the subject to control breathing during the test maneuver. The FAs who participated in our study were able to perform the DcoWB technique reliably during exercise from rest to 60% maximum observed work rate, but few were able to reach 80% let alone near maximum work rates. However, as shown in the results
section, DcoWB increased with workloads and
in a linear manner with no evidence of a plateau up to the maximum levels measured in FAs. Reproducibility of the DcoWB measurements was within 5%, which was considered satisfactory, and linear relationships were observed between DcoWB and
and were similar to those reported previously by other investigators 
. Second, in contrast to previous studies, we felt carbon monoxide (CO) backpressure on DcoWB measurement had to be considered. Stokes 
and Huang 
concluded the effect was negligible in their studies. Charloux et al found carboxyhemoglobin levels increased from 0.3% to 5.2% after 12 measurements in 4 subjects. In our study, carboxyhemoglobin levels increased from 0.3% to 10% after 12 measurements, and thus, we adjusted DcoWB for changes in hemoglobin and carboxyhemoglobin. Third, our study, which was a cross-sectional study of a group of FAs with a remote and unique high SHS exposure, did not include a similar control population for comparison. However, we used the available established prediction formulas to determine whether the obtained physiologic measurements were within the “normal” range. The appropriate control groups for our pre-ban FAs would be post-smoking ban flight attendants, who have been exposed to aircraft cabin 
except for SHS, and “ground-level” controls with no history of significant SHS exposure, both of which are challenging control groups as the post-ban flight attendants are in general younger and the “ground-level” controls do not have the specific exposure to non-SHS cabin factors that may contribute to lung function abnormalities. Of note, the normal predicted values include a wide range, and if one could show a longitudinal decline for an individual with time, even a value within the normal range may be considered to be abnormal. While our study was a cross-sectional one with measurements at a single time point, we were able to show that some FAs with normal resting diffusing capacity had abnormal physiologic responses to exercise, which suggest that despite having resting values within the “normal” range, they had abnormal pulmonary function. Fourth, the pulmonary function tests abnormalities of the FAs in this study do not meet the definition of COPD by GOLD criteria as the FAs had FEV1
within the normal predicted range and as they did not have any respiratory complaint at baseline. Traditionally, both COPD diagnosis and severity evaluation have been based on spirometry 
, and change in FEV1
over time is still the most widely accepted measure of disease progression. However, FEV1
has limitations as it measures only one aspect of the disease and is not predictive of disease progression, especially in early disease 
. In addition, COPD patients with similar FEV1
may show very different underlying pathologies, for example predominantly airspace disease (i.e. emphysema) or disease of the airways, as manifested by increased airway wall thickness 
. Thus, we believe that despite the “normal” GOLD classification of the FAs in our cohort, the physiologic abnormalities that we have observed in them are best explained and most consistent with presence of COPD and emphysema 
. Finally, we estimated the SHS exposure through the proxy of years of pre-ban employment using our occupational/employment questionnaire 
. This method of SHS assessment may not be adequately sensitive for true amount of SHS exposure and may also be prone to recall bias. However, the cabin SHS exposure, which is relatively readily quantified using employment history, was much higher than levels experienced outside aircraft cabin in most circumstances 
, and dwarfs the non-cabin SHS exposure for these pre-ban FAs.
In conclusion, in this cohort of never-smoking pre-ban FAs with remote but significant history of cabin SHS exposure, we found physiologic abnormalities consistent with presence of emphysema and/or COPD. The FAs with lower diffusing capacity had exercise limitation and lower increase in their diffusing capacity with increasing workload and increasing pulmonary blood flow suggesting that those with lower resting diffusing capacity had reduced pulmonary capillary bed recruitment. Exposure to SHS in the aircraft cabin seemed to be a predictor for lower diffusing capacity during exercise in those with higher history of SHS exposure.