Respiratory Symptoms Necessitating Spirometry Among Soldiers With Iraq/Afghanistan War Lung Injury

To the Editor:

We read with interest the recent article by Szema et al¹ on the respiratory symptoms among US Veterans deployed to Iraq (OIF) and/or Afghanistan (OEF).¹ They reported a significant increase in the rates of symptoms and spirometry for veterans deployed to Iraq/Afghanistan compared with troops deployed elsewhere.

We have some concerns regarding the study’s biases and their influence on the observed findings and associated inferences. The study has questionable generalizability. The cohort was based on US Department of Veterans Affairs (VA) users living in Long Island/New York City and likely does not represent all veterans. There are well-established differences in demographics and health status between VA users and nonusers. Veterans Affairs users, on average, have a lower socioeconomic status and overall poorer health status than nonusers.^2–4 OEF/OIF Veterans may be more likely than OEF/OIF-era Veterans (“deployed elsewhere”) to go to a Veterans Affairs Medical Center for reasons related to health care benefits for service-connected illnesses. OEF/OIF combat veterans receive free medical care for any condition related to their service in the Iraq/Afghanistan theatre for 5 years after the date of their discharge or release if they enroll in VA’s health care system.

The study also has several notable biases, which limit the study’s internal validity. Most importantly, the authors do not account for confounding for a variety of established risk factors for respiratory symptoms despite having adequate sample size for multivariable analyses. The authors have data to address potential confounding by smoking, age, and gender but did not employ multivariable analyses to examine the effect of these potential confounders. The authors also fail to discuss several other potential confounders including race/ethnicity, service affiliation (Army, Marine Corps, Navy, and Air Force), and socioeconomic status or rank.

By definition, a confounder (such as smoking) is (1) associated with the exposure (deployment), (2) not influenced by the exposure, and (3) a cause or proxy cause of the outcome (respiratory symptoms requiring spirometry). Smoking is clearly a confounder in the relationship between deployment and respiratory symptoms with spirometry. The authors indicate that smoking is associated with deployment (16% among veterans and 3% among nonveterans) and spirometry evaluation (35% among those with spirometry and 5% among those without spirometry). Using the authors’ data from Tables 1 and 2, smokers have an unadjusted 6-fold increased odds of deployment and 10-fold increased odds of respiratory symptoms. Given that both of these associations are positive, the resulting confounding bias is positive (away from the null hypothesis). Inexplicably, the authors do not address the obvious confounding using analyses stratified by smoking status or multivariable analyses.

Surveillance bias is another potential threat to the validity of this study. There has been a great deal of concern about respiratory conditions among persons deployed to the Persian Gulf region since the 1991 Gulf War.⁵ It is possible that health care providers are more likely to request additional tests, including spirometry, because they were aware of the veteran’s deployment to OEF/OIF. This could lead to respiratory conditions being diagnosed more frequently among deployed veterans, not because they truly have a greater incidence but because they were more likely to be referred for additional testing. This would result in a differential misclassification of disease status. Within the VA, a veteran may also be referred for spirometry on the basis of their current clinical condition or for adjudication of a claim for service-connected disability. The authors neither address this distinction nor identify the reason the veteran was referred for spirometry.

The authors draw conclusions, which are not supported by the study. The authors’ conclusions that Iraq/Afghanistan War Lung Injury (IAW-LI) is “common” and that “rates of symptoms leading to a diagnosis requiring spirometry are high” are not supported by the data. There is no case definition for IAW-LI. By the authors’ definition, IAW-LI can occur only among those veterans deployed to Iraq/Afghanistan. The comparison of those deployed versus nondeployed does not provide any information with respect to IAW-LI because the nondeployed are not at risk for IAW-LI. The authors, therefore, cannot make statements regarding the relative prevalence of IAW-LI. The study concludes that deployed soldiers were more symptomatic and received spirometry more than soldiers stationed “elsewhere.” These findings are likely biased by uncontrolled confounding by smoking and do not provide any indication that respiratory symptoms are widespread among OEF/OIF veterans. Finally, the conclusions regarding lung injury are not supported by data on symptoms and physiologic tests in the absence of pathologic evidence.

The inclusion of appropriate multivariable analyses would address the effect of confounding and more explicitly define the association between deployment to OEF/OIF and development of new-onset respiratory symptoms.

The health effects of deployment to Afghanistan and Iraq are of utmost importance to VA. Well-designed prospective cohort studies comparing deployed and nondeployed veterans are necessary to determine whether deployment to OEF/OIF is associated with respiratory problems.