The Impact of Preexisting Mental Health Disorders on the Diagnosis, Treatment, and Survival among Lung Cancer Patients in the U.S. Military Health System

This article is featured in Highlights of This Issue, p. 1535

There is mounting evidence of a higher all-cause mortality in people with mental illness than general population (1–9). When the causes of mortality were examined, it was found that except for suicide and other unnatural deaths, people with mental health disorders also experienced a higher mortality than general population, including cancer-related mortality (1, 9–14).

Mental health disorders may influence cancer outcomes through an array of socioeconomic, behavioral, and biologic mechanisms. Socioeconomically, people with mental health disorders may be more likely to be unemployed, have lower education level, and live at or below the poverty level; these factors are associated with less access to healthcare, screening services, in-patient care, and out-patient clinical visits, thereby resulting in delayed diagnosis (15–17), inadequate or less aggressive treatments (16, 18, 19), and poorer survival (14–18, 20–25). Behaviorally, mental health disorders are commonly linked to unhealthy lifestyles such as smoking, alcohol and substance abuse, physical inactivity, poor nutrition, and poor physical health, which contribute to cancer progression and mortality (10, 11, 14, 26, 27). It is also possible that biologically, mental illness could affect biochemical pathways, which include increased cell damaging processes and decreased cell protective or restorative processes (28), dysfunctional immunosurveillance, accumulation of somatic mutations, and genome instability (29). These biologic changes may also facilitate cancer progression.

Lung cancer is the leading cause of cancer-related death worldwide (30). Non–small cell lung cancer (NSCLC) comprises 85% to 90% of all lung cancers (30). Previous studies have found increased mortality among patients with lung cancer with mental illness (1, 10, 11, 13, 14, 31, 32). However, it is not clear what factors may contribute to the increased mortality while it can be reasonably postulated that treatment and treatment adherence may differ between patients with and without a mental health disorder. One recent study showed no association between depression and the receipt of surgery, chemotherapy, or radiation among military veterans with NSCLC (31). Another study that was restricted to advanced stage NSCLC veteran patients (stages III and IV) did not find an association between depression and receipt of chemotherapy (33). However, depression was reported to be associated with poor treatment adherence among patients with advanced stage NSCLC (32). To the best of our knowledge, no studies have examined whether mental health disorders overall or specific mental health disorders other than depression could affect stage at diagnosis, cancer treatment, and survival among patients with NSCLC.

In the general population, patients with mental health disorders may differ from those without access to medical care due to unemployment, lower education level, and/or poverty. This differential access to care may affect research results in the general population on cancer diagnosis, treatment, and survival because lack of health insurance, limited access to wellness visits, and underutilization of health care service may be related to delayed diagnosis, inadequate treatment and poor survival among patients (16, 17, 22, 23). Research in a health system that provides universal healthcare could reduce the potential effects of unequal access to care. The U.S. military health system (MHS) provides health care to military personnel, retirees, and their family members and all beneficiaries have equal access to health care regardless of military rank, education, or income. A study within the MHS could demonstrate whether mental health disorders increases the mortality of cancer when there are no or minimal barriers to health care access, thus suggesting potential effects of factors other than those related to health care access. In addition, compared with the general population, military personnel are at higher risk of developing mental health problems (34–36) such as posttraumatic stress disorder related to prior combat. Family members of military personnel are also at increased risk of mental stress due to mobile lifestyle and frequent family separations (37, 38). Therefore, it is important to assess whether mental health disorders are related to cancer outcomes in the MHS. To the best of our knowledge, there have been no epidemiologic studies on preexisting mental health disorders in relation to diagnosis, treatment, and survival of cancer including lung cancer among beneficiaries in the MHS. In this study, we investigated whether preexisting mental health disorders are associated with the stage at diagnosis, cancer treatment received, and survival among patients with NSCLC in the MHS.

This study was based on the MHS, which provides health care to active duty members, retirees, National Guard and Reserve members, and their dependents. The linked database from the Department of Defense (DoD)'s Central Cancer Registry (CCR) and the MHS Data Repository (MDR) was used in this study and described previously (39, 40). The CCR contains information on patients with cancer diagnosed or treated at military treatment facilities (MTF). The CCR data include demographic variables, tumor characteristics, cancer diagnosis, treatment, recurrence, and vital status. Cancer site and histology codes are based on the International Classification of Diseases for Oncology, third edition (ICD-O-3; ref. 41). The CCR registry staff conduct lifetime follow-up on patients. Quality assurance was conducted following the guidelines established by the North America Association of Central Cancer Registries. The MDR contains administrative and medical care information that includes both inpatient and outpatient care provided at MTFs and civilian facilities paid for by the DoD. The MDR database includes information on clinical diagnoses of all medical conditions, which are coded using the International Classification of Disease, 9^th Revision (ICD-9), and diagnostic and treatment procedures, which are coded using ICD-9 or Current Procedural Terminology (CPT) codes. The data linkage project was approved by the Institutional Review Boards of the Walter Reed National Military Medical Center, TRICARE Management Activity, and the NIH Office of Human Subjects Research.

The study subjects were patients diagnosed with histologically confirmed primary NSCLC between 1998 and 2007. NSCLC constitutes about 85% to 90% of all lung cancers (30). The cancer site and histology were classified using the topography (C34.0 to C34.3, C34.8, C34.9) and morphology codes (8050–8078, 8083, 8084, 8250–8260, 8480–8490, 8570–8574, 8140, 8211, 8230, 8231, 8323, 8550, 8551, 8576, 8010–8012, 8014–8031, 8035, 8310, and any other NSCLC codes between 8010 and 8576) of the International Classification of Diseases for Oncology, third edition (ICD-O-3; ref. 41).

Preexisting mental health disorders were defined as diagnoses of mental health disorders during the 2-year period before NSCLC diagnosis (16, 18). Mental health disorder diagnoses were identified using ICD-9 or CPT codes in the MDR data. The following mental health disorder categories were identified using specific ICD-9 codes: any mental health disorder (codes 29, 30, 31); psychotic disorders (codes 295, 297, 298, 293.81, 293.82; refs. 11, 16, 23, 25, 42–44); dementia and other organic psychosis (codes 290, 294; refs. 11, 16); mood disorders (codes 296, 311, 300.4, 309.1, 309.0, 309.4, 301.1, 301.10, 301.12, 301.13; refs. 7, 11, 16, 18, 23, 25, 42–45); substance abuse and dependence disorders (codes 291, 292, 303, 301, 305; refs. 7, 11, 16, 25, 42, 44, 45); anxiety disorders (codes 300 except 300.4, 308.3, 309.81; refs. 25, 42–44), and all other mental health disorders that falls between codes 29 and 31 but not included in any of the above categories.

There were 3 outcomes in this study: cancer stage at diagnosis, receipt of cancer treatments, and all-cause mortality. Cancer stage was defined in accordance of the American Joint Committee on Cancer (AJCC) staging system, the 7^th edition (46). Stage was further grouped into early stage (stages I and II; refs. 47, 48), advanced stage (stages III and IV; ref. 49), and unknown stage (missing information in the data). Cancer treatments (lung cancer specific surgery, chemotherapy, and radiation therapy) were identified and consolidated from both CCR and MDR. Vital status was from the CCR data.

In data analysis, we first presented the distributions of demographic, diagnostic, treatment, and other variables by mental health disorder status. We then used multivariate logistic regression to estimate ORs and 95% confidence intervals (95% CI) of preexisting mental health disorders (any and specific disorders) in relation to tumor stage at NSCLC diagnosis. ORs of mental health disorders were estimated for late stage versus early stage and unknown stage versus early stage, respectively. In regard to cancer treatment (yes or no) related to preexisting mental health disorders, we used multivariate Cox proportional hazards models to estimate HRs and 95% CI. Time to treatment was used in the models and calculated as the interval between NSCLC diagnosis date and first cancer treatment date. Subjects who did not receive any cancer treatment before study end date were censored. For surgery, chemotherapy, and radiation treatments, time to treatment was calculated as the interval between NSCLC diagnosis date and the date of receiving surgery, chemotherapy, and radiation therapy, respectively. Considering recommended treatment guidelines (50, 51), receipt of surgery was assessed among stages I and II patients only and receipt of chemotherapy and radiation therapy was analyzed among stages III and IV patients only. In terms of cancer survival, we first analyzed overall survival by preexisting mental health disorder status using Kaplan–Meir curve and log-rank test. Multivariate Cox proportional hazards models were then used to estimate HRs and 95% CI of mortality associated with preexisting mental health disorders. Survival time was calculated as the interval from the date of diagnosis until death, date of last contact, or the study end date. Subjects who did not die during follow-up were censored. Cox analysis was further stratified by tobacco use (never, former, and current) and by comorbidity index group. Comorbidities were identified from the MDR data. A comorbid condition was considered to be present if at least 1 inpatient record or 3 outpatient records (52, 53) showed the diagnosis prior to the NSCLC diagnosis. The level of comorbidity was categorized according to the Charlson comorbidity index (54). The index score was further grouped into three groups with index score of 0, 1, and 2 or more, respectively. In all logistic and Cox regression analyses, the potential confounders, variables related to both mental health disorders and outcomes, were controlled in the models.

A total of 5,054 patients with NSCLC were identified from the data. Of them, 1,858 had a history of preexisting mental health disorder. The distributions of patient characteristics by mental health disorder are shown in Table 1. Compared with individuals without any mental health disorder, patients with a diagnosis of preexisting mental health disorder were more likely to be in the age groups of 50–59 and 60–69 years (P < 0.001), female (P = 0.007), and White (P < 0.001). There was a higher percentage of Army beneficiaries with a mental health disorder than other military branches (P < 0.001). Patients with a mental health disorder were more likely to be current smokers (P < 0.001) and have more comorbidities than those without (P < 0.001). Compared with those without mental health disorders, patients with a mental health disorder were more likely to have stage I and less likely to have unknown information on tumor stage (P = 0.002). They were more likely to be diagnosed with moderately differentiated, poorly differentiated, and undifferentiated tumors (P = 0.015) and with large cell histology (P < 0.001; Table 1). Differences between the two groups in cancer treatment, recurrence, and marital status were not significant (Table 1).

Compared with those without a preexisting mental health disorder, patients with any preexisting mental health disorders were 42% less likely to be diagnosed with an unknown stage tumor (OR, 0.58; 95% CI, 0.41–0.80; Table 2). Further analysis showed that compared with patients without any preexisting mental health disorders, the OR of unknown tumor stage was 0.66 (95% CI, 0.40–1.07) for those with one mental health disorder and 0.54 (95% CI, 0.37–0.79) for those with two or more mental health disorders (Table 2). The ORs were not significant in late tumor stage at diagnosis relative to early stage for patients with a mental health disorder compared with those without. In terms of specific mental health disorders, patients with substance abuse disorder were less likely to have an unknown tumor stage (OR, 0.64; 95% CI, 0.43–0.94; Table 2). No significant ORs were observed between other specific mental health disorders and unknown tumor stage or late stage (Table 2).

There were no significant differences in receiving any cancer treatment between the groups by mental health disorder status (Table 3). However, subjects with dementia were less likely to receive treatment than subjects without (HR, 0.74; 95% CI, 0.57–0.97; Table 3). No association was observed between other specific mental health disorders and receipt of any cancer treatment (Table 3). Further analyses on receipt of surgery (stages I and II patients only), chemotherapy (stages III and IV patients only), and radiation therapy (stages III and IV patients only) did not reveal differences between patients with and without mental health disorders (Table 4).

Kaplan–Meier survival curves showed better survival for patients without any mental health disorder than patients with a diagnosis of a mental health disorder (Supplementary Fig. S1; log-rank: P = 0.048). In the multivariate Cox regression model adjusting for confounders, preexisting mental health disorder remained an independent risk factor for mortality (HR, 1.11; 95% CI, 1.03–1.20; Table 5). Furthermore, the HR was 1.08 for those with one mental health disorder (95% CI, 1.00–1.17) and increased to 1.22 (95% CI, 1.08–1.37) for those with two or more mental health disorders. When data were analyzed on a specific mental health disorder, the increased mortality was observed for dementia (HR, 1.43; 95% CI, 1.10–1.86) and anxiety disorder (HR, 1.20; 95% CI, −1.03–1.39; Table 5).

The analysis stratified by tobacco use or comorbidity index is shown in Table 6. Significantly increased mortality associated with mental health disorders was observed among current tobacco users (HR, 1.18; 95% CI, 1.04–1.33) but not among previous users (HR, 1.01; 95% CI, 0.91–1.13) or never users (HR, 1.29; 95% CI, 0.89–1.88). The higher mortality associated with mental health disorders was also observed among patients who did not have comorbidity (HR, 1.15; 95% CI, 1.01–1.31) and those with low comorbidity index (HR, 1.18; 95% CI, 1.01–1.39) but not among those with high comorbidity index (HR, 1.07; 95% CI, 0.95–1.20; Table 6). However, in these stratified analyses, the 95% CIs were overlapped.

Our results showed that in MHS, a system providing universal care to all beneficiaries, compared with patients with NSCLC without any mental health disorder, patients with a diagnosis of mental health disorder did not present with more advanced disease at diagnosis and were less likely to have unknown information on staging. There were no differences in receipt of cancer treatment between those with and without mental health disorders. However, patients with pre-cancer mental health disorders experienced higher mortality than those without.

Higher mortality of cancer among people with mental health disorders has been reported. Kisely and colleagues (10) studied all-cause mortality of 135,442 patients with cancer, including patients with lung cancer, from 1988 to 2007 using Western Australian Cancer Registry and national morbidity data systems and reported increased mortality rates for major cancers in psychiatric patients as compared with sex- and age-matched general population. The rate ratio for lung cancer was 1.24 (95% CI, 1.11–1.40). Two other studies in the same area observed similar excess lung cancer mortality in people with psychiatric illnesses (1, 11). In another study based on the linked data from death certificates and the mental health system in Ohio, Musuuza and colleagues (14) reported higher standardized mortality rates (SMR) of lung cancer among people with mental illness than the general population, with SMRs ranging from 3 to 5 times higher stratified by race and gender, respectively (14). With respect to specific mental health disorders, a French study observed a more than 2-fold increase in the mortality of lung cancer among men with schizophrenia compared with the general population (13). Two other cohort studies identified psychologic stress (12) and schizophrenia (21) as significant may exist for the observed associations in these studies. A recent study in veterans reported increased mortality among patients with lung cancer with preexisting depression (31) and depression was associated with poor survival in a small Mexican study (32). It is not clear what underlying factors might be for the observed associations in these studies.

One of the possible factors related to the poor survival of patients with cancer with mental illness may be socioeconomic disadvantages, which may limit their access to health care and services, resulting in delayed diagnosis or later cancer stage at diagnosis (15–17, 22), insufficient treatment (16, 18, 23), and therefore poor survival (15, 16, 18, 23). In lung cancer, however, there are few studies examining these possible factors. Sullivan and colleagues (31) found no association between depression and receipt of surgery, chemotherapy, or radiation in a veteran population. Another study restricted to advanced stage veteran patients did not find an association between depression and receipt of chemotherapy (33). The authors concluded that other factors not included in their analysis may have contributed to the poor survival among patients with depression. Nevertheless, in other cancers, multiple large-scale studies have demonstrated that patients with mental health disorders tended to present with later stage tumors and were less likely to receive cancer treatments (16, 18, 23). In a study in Taiwan, Chang and colleagues (23) found that patients with oral cancer with mental illness were less likely to undergo surgery with or without adjuvant therapy than those without mental illness. In the study by Kisely and colleagues (10), psychiatric patients had a higher proportion of cancer with metastases than the general population. They were also less likely to receive surgery of colorectal, breast, and cervical cancers; radiotherapy for breast, colorectal, and uterine cancers; and chemotherapy sessions (10). A SEER-Medicare–based study (16) found that old patients with colon cancer with preexisting mental health disorders were more likely to be diagnosed at an unknown stage and less likely to have received any treatment (surgery, chemotherapy, or radiation therapy) and less likely to receive chemotherapy (stage III patients), suggesting that individuals with mental health disorders may not be able to complete the diagnostic procedures for cancer staging and treatments, likely due to psychosocial and economic barriers to medical care (16).

In contrast to the above previous studies that found patients with mental health disorders were more likely to be diagnosed at an advanced stage or unknown stage and less likely to receive recommended treatment, our study showed that there was no difference in the stage at diagnosis between patients with mental health disorders and those without. Similarly, patients with mental health disorders were less likely to have unknown information on tumor stage. Regarding cancer treatment, with the exception of patients with dementia who received less treatment which was largely driven by less surgery, there was no overall difference between those with and without a mental health disorder. The lower likelihood of having unknown stage and the absence of differences in receiving cancer treatments may be attributed to a high accessibility to health care for beneficiaries in MHS, in which there are no or minimum cost barriers to laboratory tests, radiology workup, and cancer prevention and surveillance services (55). Moreover, in early 2000, the MHS launched the initiative of integration behavioral health services into primary care services (56), which have facilitated the care and follow up of beneficiaries with mental health disorders in the system. The integration of mental healthcare into primary care services in MHS may have reduced the treatment disparity generally found in the general population. Consistent with our results, two studies in the Veteran Affairs (VA) system, a system providing care to all beneficiaries, did not observe an association between depression and treatment. A recent study in the VA system also did not find differences in the stage at diagnosis or delay in care for colorectal, urothelial, and head and neck cancer between patients with and without mental health problems (57). The authors concluded that mental health is not a barrier to cancer staging or treatment in the VA system that integrated mental health care and routine health care (57).

While cancer diagnosis and treatment related to health care access and socioeconomic barrier might not account for the survival disadvantage among those with mental health disorder in MHS, unhealthy lifestyles and poor physical health commonly found in patients with mental health disorder may affect survival indirectly (10, 11, 14, 26, 27). In our study population, there were higher percentages of tobacco users and higher percentages of people with comorbidities in the mental health disorder group. However, the decrease in survival among patients with mental health disorders was still observed after adjusting for tobacco use and comorbidity in survival analysis in our study. In addition, we stratified the survival analysis by tobacco use and comorbidity, and the results remained similar between the subgroups, suggesting the effects of factors other than smoking and comorbidity. The second factor is that patients with mental illness may be more likely to commit suicide, which contributes to overall mortality. However, the rate of suicide might not be high enough to account for the difference in survival. Another factor may be the time interval between treatments and treatment frequency. This difference between patients with and without mental health disorders warrants future study. This study only evaluated receipt of treatment, but adherence to treatment recommendations may differ between patients with versus without mental health disorders, which may contribute to the overall survival difference. Finally, mental health disorders may affect interactions and communications between patients and their providers, and patient–provider communications have been shown to influence clinical outcomes in patients with lung cancer (21, 58, 59)

Our study was based on the existing cancer registry and medical claims data and thus there are limitations due to the use of such data for research, such as lack of detailed information and inaccurate records found in secondary data sources. Also, the numbers in specific mental health disorder groups were small and thus further research with a larger number of patients is warranted.

In conclusion, our study suggests that health care access and socioeconomic barriers commonly found to be responsible for the poor survival in the studies conducted within the general population may not explain the worse survival among patients with mental health disorders. There may be other factors at play and additional research is warranted to further examine the increased mortality of lung cancer among patients with mental health disorders.

No potential conflicts of interest were disclosed.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of the Departments of the Navy and Army, the Uniformed Services University of the Health Sciences, the Department of Defense, National Cancer Institute, or the U.S. Government. Nothing in the presentation implies any Federal/DoD endorsement.

Conception and design: J. Lin, C.A. Carter, C.D. Shriver, K. Zhu

Development of methodology: J. Lin, K. Zhu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.A. Carter, K. Zhu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Lin, C.A. Carter, J.A. Nations, W.F. Anderson, K. Zhu

Writing, review, and/or revision of the manuscript: J. Lin, K.A. McGlynn, C.A. Carter, J.A. Nations, W.F. Anderson, C.D. Shriver, K. Zhu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.D. Shriver

Study supervision: J. Lin, C.D. Shriver, K. Zhu

The authors thank the following institutes for their contributions and support for the original data linkage project: ICF Macro, Kennel and Associates, Inc., TRICARE Management Activity, Armed Forces Institute of Pathology, National Cancer Institute, U.S. Military Cancer Institute. We thank Dr. Shelia H. Zahm for her comments on the manuscript.

This project was supported by John P. Murtha Cancer Center, Walter Reed National Military Medical Center via the Uniformed Services University of the Health Sciences under the auspices of the Henry M. Jackson Foundation for the Advancement of Military Medicine and by the intramural research program of the National Cancer Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.