Letter to the Editor:
This brief correspondence is to address concerns regarding our manuscript, “Study of Active Duty Military Personnel for Environmental Deployment Exposures: Before and After Deployment Spirometry (STAMPEDE II)” by Morris et al published in the May 2019 issue of Respiratory Care.1 Questions have been raised in several forums regarding the statistical analysis used in the original article. The original data involved the longitudinal collection of spirometry and impulse oscillometry values from soldiers prior to and after deployment to the theater of operations in southwest Asia. Although the overall mean values for spirometry demonstrated a slight increase for the group, there was concern that the data were not analyzed correctly for individual variations associated with each subject’s change in spirometry prior to and after deployment. The original data in the article were in fact analyzed using a paired t test approach, which takes advantage of the correlation between before and after deployment measures.1 This was incorrectly reported as a Student t test in the article.
The presentation of the data as grouped means gave the impression that individual variation was not addressed. However, the original paper did include results of a very specific analysis of subjects with obstruction (as defined by NHANES III reference values) and noted 116 (13.7%) subjects in the cohort with this finding. It was further noted that only 29 individuals with normal spirometry before deployment developed airway obstruction after deployment with a significant decrease in FEV1/FVC. Con-versely, an equal number of individuals with baseline obstruction had a similar increase in FEV1/FVC. Additionally, further analysis of spirometry values based on risk factors (eg, smoking, asthma, and elevated body mass index) did not demonstrate any statistical differences before and after deployment as shown in Table 4 of the original article.
To further emphasize the individual variation, additional data are presented with this letter. Figure 1 presents the distributions for changes in FEV1 and FVC after deployment. The median change for FEV1 was 0.9% (interquartile range 6.3–4.6%) and for FVC was 0.6% (interquartile range 5.7–5.1%). As expected, there is an equal distribution of values centered on the mean. We further updated both mean spirometry and impulse oscillometry values (Table 1). The addition of correlations for the pulmonary function data comparing the before and after deployment values shows a tight clustering of data. Note that r2 represents the proportion of reduction in the within-subject variability (from before to after deployment) from the overall variability. Reported interindividual variability for spirometry is ± 5%, whereas the population variability may be ± 20%.2 Updated data analyzing comorbidities show similar findings with the addition of Δ change and correlation values (Table 2).
Distribution box plots on the changes in spirometry values from before deployment to after deployment. Values 2.5 times the standard deviation outside of the median are represented by dots.
Mean Pulmonary Function Testing Values
Pulmonary Function Test Values and Comorbidities
We likewise examined the percentage of individuals who saw at least a 10% decrease in their percent of predicted values for spirometry. There were 64 subjects with at least a 10% decrease in FVC; only 2 would have been classified as meeting criteria for restrictive indices. Of the 54 subjects with a 5–10% decrease, only 3 met restrictive criteria. Furthermore, of the 68 subjects with a 10% decrease in FEV1 after deployment, only 2 would have been reclassified as obstructive indices. With a 5–10% decrease, 8 subjects would have been reclassified as obstructive. Overall, only 15 (1.8%) subjects had a significant change to reclassify their spirometric indices in consideration of further evaluation for underlying lung disease. This small percentage does not support the routine use of spirometry in the deployed population in the absence of respiratory findings. Given the relatively brief length of deployments and low level of potential exposures, detecting spirometric changes would be unlikely compared to intense exposures seen in the World Trade Center collapse.3 However, additional longitudinal studies of lung function in the deployed population are warranted.
Footnotes
- Correspondence: Michael J Morris MD. E-mail: michael.j.morris34.civ{at}mail.mil.
- Copyright © 2020 by Daedalus Enterprises
