Early mortality experience in a large military cohort and a comparison of mortality data sources

All-cause mortality among United States military personnel: Findings from the Millennium Cohort Study, 2001-2021

PURPOSE

The goal of this study was to estimate all-cause mortality among Operations Enduring Freedom, Iraqi Freedom, and New Dawn era service members and veterans and to identify protective and risk factors for mortality.

METHODS

Using 20 years of longitudinal data from the Millennium Cohort Study (2001-2021), sequential Cox proportional hazard models were conducted to examine demographic, military, and health-related characteristics associated with all-cause mortality among service members and veterans.

RESULTS

Among 201,619 participants, 3806 (1.9 %) were deceased by the end of the observation period, with an age- and sex-adjusted incidence of 37.6 deaths per 100,000 person-years. Deployed service members had lower all-cause mortality risk than those who did not deploy. Personnel who experienced combat had higher mortality risk compared with those who did not in unadjusted models; this association was nonsignificant after accounting for health-related factors. Enlisted and Army personnel both had a higher mortality risk, while women and Hispanic individuals had a lower risk. Stressful life events, lower physical health related quality of life, problem drinking, and smoking were also associated with greater mortality risk.

CONCLUSION

These profiles may be useful for developing preventive education and intervention efforts in military and veteran populations to reduce premature mortality.

Pregnancy-associated and pregnancy-related deaths in the United States military, 2003-2014

BACKGROUND

The Centers for Disease Control and Prevention has reported a steady increase in the US pregnancy-related mortality ratio since national surveillance began in 1987, although trends are partially induced by concurrent improvements in the identification of pregnancy-related deaths. No previous work has evaluated pregnancy-associated and pregnancy-related deaths among active-duty service members, a population with comprehensive health coverage and stable employment.

OBJECTIVE

This study aimed to assess trends and variations in pregnancy-associated and pregnancy-related deaths in the US military.

STUDY DESIGN

Live births to active-duty service members were captured in Department of Defense Birth and Infant Health Research program data from 2003 to 2014. Pregnancy-associated deaths (deaths temporally related to pregnancy from any cause) were identified through 1 year after pregnancy end date using National Death Index Plus data from the Joint Department of Defense and Department of Veterans Affairs Suicide Data Repository. Pregnancy-associated deaths were classified as pregnancy-related (causally related to pregnancy) based on cause-of-death codes in the National Death Index Plus data, administrative medical encounter data, and medical record review. Pregnancy-related deaths were reported including and excluding deaths from suicide and unintentional overdose. Mortality ratios (deaths per 100,000 live births) were reported overall, triennially, and by selected characteristics; the relative contribution of each cause of death to all pregnancy-associated deaths was reported overall and by age, race and ethnicity, and marital status. Timing of death relative to pregnancy end date was assessed by cause of death.

RESULTS

A total of 179,252 live births occurred to active-duty service members from 2003 to 2014. Pregnancy-associated and pregnancy-related mortality ratios were 41.3 (95% confidence interval, 32.4-51.8) and 18.4 (95% confidence interval, 12.7-25.9), respectively. Excluding deaths from suicide and unintentional overdose, the pregnancy-related mortality ratio was 11.2 (95% confidence interval, 6.8-17.2). Deaths from suicide and unintentional overdose composed a larger proportion of pregnancy-related deaths over time and accounted for 17.6% of all pregnancy-associated deaths. Deaths from other pregnancy-related causes accounted for a greater share of deaths among older vs younger service members (≥30 years: 41.2%; 18-29 years: 22.8%) and non-Hispanic Black vs White service members (33.3% vs 24.1%). Pregnancy-related deaths, excluding suicide and unintentional overdose, were more likely to occur within 42 days of pregnancy end date; in contrast, deaths from suicide, overdose, assault, and undetermined intent were more likely to occur between 42 days and 1 year after pregnancy.

CONCLUSION

Pregnancy-associated and pregnancy-related deaths varied over time and by age and race and ethnicity. Suicide and overdose are major recent causes of pregnancy-related death among active-duty service members.

The Millennium Cohort Study: The First 20 Years of Research Dedicated to Understanding the Long-Term Health of US Service Members and Veterans

The Millennium Cohort Study, the US Department of Defense's largest and longest running study, was conceived in 1999 to investigate the effects of military service on service member health and well-being by prospectively following active duty, Reserve, and National Guard personnel from all branches during and following military service. In commemoration of the Study's 20th anniversary, this paper provides a summary of its methods, key findings, and future directions. Recruitment and enrollment of the first 5 panels occurred between 2001 and 2021. After completing a baseline survey, participants are requested to complete follow-up surveys every 3 to 5 years. Study research projects are categorized into 3 core portfolio areas (psychological health, physical health, and health-related behaviors) and several cross-cutting areas and have culminated in more than 120 publications to date. For example, some key Study findings include that specific military service-related factors (e.g., experiencing combat, serving in certain occupational subgroups) were associated with adverse health-related outcomes and that unhealthy behaviors and mental health issues may increase following the transition from military service to veteran status. The Study will continue to foster stakeholder relationships such that research findings inform and guide policy initiatives and health promotion efforts.

Variation in model performance by data cleanliness and classification methods in the prediction of 30-day ICU mortality, a US nationwide retrospective cohort and simulation study

OBJECTIVE

There has been a proliferation of approaches to statistical methods and missing data imputation as electronic health records become more plentiful; however, the relative performance on real-world problems is unclear.

MATERIALS AND METHODS

Using 355 823 intensive care unit (ICU) hospitalisations at over 100 hospitals in the nationwide Veterans Health Administration system (2014-2017), we systematically varied three approaches: how we extracted and cleaned physiologic variables; how we handled missing data (using mean value imputation, random forest, extremely randomised trees (extra-trees regression), ridge regression, normal value imputation and case-wise deletion) and how we computed risk (using logistic regression, random forest and neural networks). We applied these approaches in a 70% development sample and tested the results in an independent 30% testing sample. Area under the receiver operating characteristic curve (AUROC) was used to quantify model discrimination.

RESULTS

In 355 823 ICU stays, there were 34 867 deaths (9.8%) within 30 days of admission. The highest AUROCs obtained for each primary classification method were very similar: 0.83 (95% CI 0.83 to 0.83) to 0.85 (95% CI 0.84 to 0.85). Likewise, there was relatively little variation within classification method by the missing value imputation method used-except when casewise deletion was applied for missing data.

CONCLUSION

Variation in discrimination was seen as a function of data cleanliness, with logistic regression suffering the most loss of discrimination in the least clean data. Losses in discrimination were not present in random forest and neural networks even in naively extracted data. Data from a large nationwide health system revealed interactions between missing data imputation techniques, data cleanliness and classification methods for predicting 30-day mortality.

Precancer diagnosis cardiorespiratory fitness, physical activity and cancer mortality in men

BACKGROUND

The preventive role of precancer diagnosis cardiorespiratory fitness (CRF) and physical activity (PA) in cancer mortality is poorly characterized. The aim of this study was to assess the association between CRF, PA at precancer-diagnosis and cancer mortality in men who diagnosed with cancer later in life.

METHODS

A total of 699 men (63±10 years) who were diagnosed with cancer during 7.5±4.9 years from a baseline treadmill exercise test and reported PA were analyzed. Multivariate Cox models for CRF and univariate model for PA were conducted. Population Attributable Risks (PARs%) and exposure impact number (EIN) of low CRF (<5 METs) and inactivity were determined.

RESULTS

During 6.5±5.2 years from cancer diagnosis, 56% died from cancer. CRF was inversely, graded and independently associated with cancer death. A 1-MET increase and categories of moderate and high CRF were associated with 7%, 28% and 51% reductions in risk of cancer death, respectively. Active compared to inactive individuals had a 23% reduced risk of cancer mortality (HR=0.77, 95% CI [0.63-0.94], P=0.01). PARs% of low CRF and inactivity were 4.8% and 9.4%, respectively, while the respective EIN were 3 and 9.

CONCLUSIONS

Higher CRF and being active at precancer-diagnosis were associated with lower cancer mortality and longer survival time in men who developed cancer later in life. Screening and intervening for low CRF and inactivity as risk factors during middle-age and maintaining at least moderate CRF and activity levels may be effective strategies for prevention of cancer mortality.

Cardiorespiratory Fitness, Adiposity, and Cancer Mortality in Men

OBJECTIVE

This study sought to evaluate the association between cardiorespiratory fitness (CRF) and cancer mortality in men with overweight and obesity.

METHODS

Maximal exercise testing was performed in 3,610 men (58.8 ± 17.5 years) (n = 2,100 with overweight and n = 1,510 with obesity) free from malignancy at baseline who were followed for 12.3 ± 7.4 years. Body mass index of 25.0 to 29.9 kg/m² for overweight and ≥ 30.0 for obesity categories was used. Hazard ratios and population-attributable risks (PAR) were determined.

RESULTS

During the follow-up period, 11.1% and 9.1% died from cancer among those who had overweight and obesity, respectively. CRF had an inverse and graded association with cancer mortality. Compared with low CRF (< 5 metabolic equivalents), moderate and high CRF levels were associated with 48% and 79% reduced risks for cancer mortality in men who had overweight (P < 0.001) and 55% and 83% lower risks in those who had obesity (P < 0.001), respectively. Low CRF had PARs of 9.3% and 10.5% for cancer mortality in subjects who had overweight and obesity, respectively.

CONCLUSIONS

Among men with overweight and obesity, higher CRF is associated with lower cancer mortality. Eliminating low CRF as a risk factor would potentially prevent a considerable number of cancer deaths and reduce the associated societal and economic burden in these high-risk populations.

Impact of Combat Deployment and Posttraumatic Stress Disorder on Newly Reported Coronary Heart Disease Among US Active Duty and Reserve Forces

Background—

The recent conflicts in Iraq and Afghanistan have exposed thousands of service members to intense stress, and as a result, many have developed posttraumatic stress disorder (PTSD). The role of military deployment experiences and PTSD in coronary heart disease (CHD) is not well defined, especially in young US service members with recent combat exposure.

Methods and Results—

We conducted a prospective cohort study to investigate the relationships between wartime experiences, PTSD, and CHD. Current and former US military personnel from all service branches participating in the Millennium Cohort Study during 2001 to 2008 (n=60 025) were evaluated for newly self-reported CHD. Electronic medical record review for International Classification of Diseases, Ninth Revision, Clinical Modification codes for CHD was conducted among a subpopulation of active duty members (n=23 794). Logistic regression models examined the associations between combat experiences and PTSD with CHD with adjustment for established CHD risk factors. A total of 627 participants (1.0%) newly reported CHD over an average of 5.6 years of follow-up. Deployers with combat experiences had an increased odds of newly reporting CHD (odds ratio, 1.63; 95% confidence interval, 1.11–2.40) and having a diagnosis code for new-onset CHD (odds ratio, 1.93; 95% confidence interval, 1.31–2.84) compared with noncombat deployers. Screening positive for PTSD symptoms was associated with self-reported CHD before but not after adjustment for depression and anxiety and was not associated with a new diagnosis code for CHD.

Conclusions—

Combat deployments are associated with new-onset CHD among young US service members and veterans. Experiences of intense stress may increase the risk for CHD over a relatively short period among young adults.

Mood disorders and risk of lung cancer in the EAGLE case-control study and in the U.S. Veterans Affairs inpatient cohort

Background

Mood disorders may affect lung cancer risk. We evaluated this hypothesis in two large studies.

Methodology/Principal Findings

We examined 1,939 lung cancer cases and 2,102 controls from the Environment And Genetics in Lung cancer Etiology (EAGLE) case-control study conducted in Italy (2002–2005), and 82,945 inpatients with a lung cancer diagnosis and 3,586,299 person-years without a lung cancer diagnosis in the U.S. Veterans Affairs Inpatient Cohort (VA study), composed of veterans with a VA hospital admission (1969–1996). In EAGLE, we calculated odds ratios (ORs) and 95% confidence intervals (CI), with extensive adjustment for tobacco smoking and multiple lifestyle factors. In the VA study, we estimated lung cancer relative risks (RRs) and 95% CIs with time-dependent Poisson regression, adjusting for attained age, calendar year, hospital visits, time within the study, and related previous medical diagnoses. In EAGLE, we found decreased lung cancer risk in subjects with a personal history of mood disorders (OR: 0.59, 95% CI: 0.44–0.79, based on 121 lung cancer incident cases and 192 controls) and family history of mood disorders (OR: 0.62, 95% CI: 0.50–0.77, based on 223 lung cancer cases and 345 controls). The VA study analyses yielded similar results (RR: 0.74, 95% CI: 0.71–0.77, based on 2,304 incident lung cancer cases and 177,267 non-cancer person-years) in men with discharge diagnoses for mood disorders. History of mood disorders was associated with nicotine dependence, alcohol and substance use and psychometric scales of depressive and anxiety symptoms in controls for these studies.

Conclusions/Significance

The consistent finding of a relationship between mood disorders and lung cancer risk across two large studies calls for further research into the complex interplay of risk factors associated with these two widespread and debilitating diseases. Although we adjusted for smoking effects in EAGLE, residual confounding of the results by smoking cannot be ruled out.

Comparison of demographic characteristics, surgical resection patterns, and survival outcomes for veterans and nonveterans with non-small cell lung cancer in the Pacific Northwest

INTRODUCTION

Lung cancer is a leading cause of death in the United States and among veterans. This study compares patterns of diagnosis, treatment, and survival for veterans diagnosed with non-small cell lung cancer (NSCLC) using a recently established cancer registry for the Veterans Affairs Pacific Northwest Network with the Puget Sound Surveillance, Epidemiology, and End Results cancer registry.

METHODS

A cohort of 1715 veterans with NSCLC were diagnosed between 2000 and 2006, and 7864 men were diagnosed in Washington State during the same period. Demographics, tumor characteristics, initial surgical patterns, and survival across the two registries were evaluated.

RESULTS

Veterans were more likely to be diagnosed with stage I or II disease (32.8%) compared with the surrounding community (21.5%, p = 0.001). Surgical resection rates were similar for veterans (70.2%) and nonveterans (71.2%) older than 65 years with early-stage disease (p = 0.298). However, veterans younger than 65 years with early-stage disease were less likely to undergo surgical resection (83.3% versus 91.5%, p = 0.003). Because there were fewer late-stage patients among veterans, overall survival was better, although within each stage group veterans experienced worse survival compared with community patients. The largest differences were among early-stage patients with 44.6% 5-year survival for veterans compared with 57.4% for nonveterans (p = 0.004).

CONCLUSIONS

The use of surgical resection among younger veterans with NSCLC may be lower compared with the surrounding community and may be contributing to poorer survival. Cancer quality of care studies have primarily focused on patients older than 65 years using Medicare claims; however, efforts to examine care for younger patients within and outside the Department of Veterans Affairs are needed.

Assessing nonresponse bias at follow-up in a large prospective cohort of relatively young and mobile military service members

BACKGROUND

Nonresponse bias in a longitudinal study could affect the magnitude and direction of measures of association. We identified sociodemographic, behavioral, military, and health-related predictors of response to the first follow-up questionnaire in a large military cohort and assessed the extent to which nonresponse biased measures of association.

METHODS

Data are from the baseline and first follow-up survey of the Millennium Cohort Study. Seventy-six thousand, seven hundred and seventy-five eligible individuals completed the baseline survey and were presumed alive at the time of follow-up; of these, 54,960 (71.6%) completed the first follow-up survey. Logistic regression models were used to calculate inverse probability weights using propensity scores.

RESULTS

Characteristics associated with a greater probability of response included female gender, older age, higher education level, officer rank, active-duty status, and a self-reported history of military exposures. Ever smokers, those with a history of chronic alcohol consumption or a major depressive disorder, and those separated from the military at follow-up had a lower probability of response. Nonresponse to the follow-up questionnaire did not result in appreciable bias; bias was greatest in subgroups with small numbers.

CONCLUSIONS

These findings suggest that prospective analyses from this cohort are not substantially biased by non-response at the first follow-up assessment.