Abstract
Objectives. We evaluated factors associated with suicidal behavior and ideation (SBI) during 3 years of follow-up among 89 995 Veterans Health Administration (VHA) patients who underwent major surgery from October 2005 to September 2006.
Methods. We analyzed administrative data using Cox proportional hazards models. SBI was ascertained by International Classification of Disease, 9th Revision codes.
Results. African Americans (18% of sample; 16 252) were at an increased risk for SBI (hazard ratio [HR] = 1.21; 95% confidence interval [CI] = 1.10, 1.32), whereas Hispanics were not (HR = 1.10; 95% CI = 0.95, 1.28). Other risk factors included schizophrenia, bipolar disorder, depression, posttraumatic stress disorder, pain disorders, postoperative new-onset depression, and postoperative complications; female gender and married status were protective against SBI.
Conclusions. The postoperative period might be a time of heightened risk for SBI among minority patients in the VHA. Tailored monitoring and postoperative management by minority status might be required to achieve care equity.
The Veterans Health Administration (VHA) recognizes suicide as a key public health concern and prioritizes outreach to suicidal veterans with implementation of effective prevention measures.1 In civilian and veteran populations, suicide claims more than 36 000 lives each year or 100 lives a day, including approximately 18 veterans; the suicide rate has been increasing since 2000.2–4 Although individuals who commit suicide constitute a small proportion of the population, suicidal thoughts and behaviors are among the strongest predictors of suicide.5 People may become suicidal in response to a life event or psychosocial stressor that overwhelms their ability to cope and control, especially in the presence of a psychiatric disorder.6 Although susceptibility to suicidal behavior and ideation (SBI) may vary across major subgroups in the community, such as racial or ethnic groups, little research on this variability exists, especially under conditions of known stress.
Historically, suicidal behavior in African Americans and Latinos has received little attention because of the limited number of documented suicides among these subgroups.7,8 A full exploration of potential underreporting of SBI by race/ethnicity is beyond the scope of this article, but the role of race/ethnicity in risk of SBI can be explored in the context of receiving care in the VHA. The VHA treats a large patient population, including 25% to 30% non-White veterans9; eliminating known disparities in care is a great concern and high priority. Furthermore, veterans in the general US population may be at greater risk for suicide compared with nonveterans,10 and this trend is not limited to younger veterans. Older veterans continue to have completed suicides rates that exceed those of younger veterans, as noted in both popular and scientific publications.11,12
Veterans with severe mental illness appear to undergo surgery at high rates,13 and it is well-established that those with mental health diagnoses are at higher risk for presenting with SBI.1,14–16 Although we know that many risk factors for suicide have been identified, such as the curvilinear relationship of age with suicide, male gender, and physical and mental disorders, no reports that we know of have explored whether the months following surgery are a time of heightened risk for suicidal behavior, and whether this risk varies by racial/ethnic minority status.
Both pain and depression are commonly experienced by patients who have undergone surgical procedures; symptoms may arise at various times postoperatively, varying from immediately after surgery to months later. Depression in response to an illness or injury that requires surgery can have an added impact on the emotional state of a person and potentiate SBI following surgery. Some studies suggest that pain management is influenced by a patient’s race, with varying pain sensitivity17–19 and treatment20,21 observed among different racial and ethnic subgroups. Studies have shown that African American and Hispanic persons report a lower tolerance for experimentally-induced pain compared with White adults.22,23 There have been, to our knowledge, no assessments of SBI in association with postoperative pain in civilians or in veterans in the VHA system. Because of the high prevalence of mental health comorbidity and chronic pain among VHA surgery patients, they may be an exceptionally at-risk group with multiple risk factors for SBI. These factors are also treatable, so that identifying surgery-related risk factors may provide new intervention points for suicide prevention.
To provide the best integrated medical–mental health care, health care systems need to know whether postoperative SBI risk varies by preoperative factors, suggesting the need to provide interventions to those at high risk for SBI in the context of a health care–related life stressor: major surgery. We investigated SBI as a postsurgical outcome among VHA surgery patients by pre-existing severe mental illness status, race and ethnicity, as well as other demographic characteristics and clinical covariates.
VHA patients who experienced inpatient surgical treatment from fiscal year 2006 (FY2006; October 1, 2005, to September 30, 2006) to FY2009 were eligible for inclusion in the Surgical Treatment Outcomes for Patients with Psychiatric Disorders (STOPP) study, a retrospective cohort study of surgical experiences and outcomes of patients with pre-existing mental illness. For patients with more than 1 surgery date during this period, the first was selected and defined as the index surgery. STOPP patients entered the present study if they underwent major surgery in the VHA in FY2006, were 18 years of age or older at the time of study entry, were US military veterans per VHA priority score (described in the following), and had valid race and ethnicity (n = 103 296; n = 5620 with missing data on race/ethnicity were excluded).24 In addition, patients who underwent surgery in the following areas were retained: organ, bone or joint, cancers, vascular, and amputations. Patients with other surgeries included in the STOPP project were excluded from this report because of small numbers, resulting in a final sample size of 89 995 surgical patients.
Patient demographic characteristics were collected from administrative data extracts and aggregated to create indicators of Hispanic ethnicity, race using patients’ most commonly reported race, and marital status during the baseline year. For race, 3 separate indicators were created, representing White, African American, and other non-White. Patient age, gender, and VHA priority status were based on enrollment and vital status files. VHA priority status consists of 8 categories describing why the veteran is eligible for VHA care. Because the VHA preferentially enrolls the most disadvantaged veterans, and annually cares for approximately 5.5 million of the 23 million living US veterans, VHA priority was established to determine eligibility to receive VHA care and assessment of co-pays. Priority status is associated with both socioeconomic status and severity of illness attributed to military service (i.e., service-connected), making it an important covariate in addition to the case-mix adjusters.25,26 For example, priority 1 status (category 1) means that patients have no co-pays for care or prescription medications, and thus, it measures 1 aspect of access to care by indicating absence of the barrier of cost. For the proposed analyses, an indicator of priority 1 status versus other priority levels was created to serve as a proxy for socioeconomic status.
Other patient attributes entailed the use of patients’ previous year of utilization data to determine severe mental illness status per VHA definition (i.e., schizophrenia, bipolar disorder, major depressive disorder, posttraumatic stress disorder, or no severe mental illness diagnosis) and summary measures of comorbidity using diagnosis codes from inpatient and outpatient records. Two or more outpatient visit dates carrying a diagnosis of schizophrenia (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM] code 295, excluding 295.5) or 1 such inpatient date identified patients with schizophrenia; bipolar disorder (ICD-9-CM codes 296.0–296.1, 296.4–296.8) was similarly defined.27 The criterion of having 2 or more outpatient dates with the same diagnosis applied to identification of posttraumatic stress disorder (PTSD; code 309.81), and major depressive disorder (codes 296.2, 296.3, 311). The Charlson comorbidity score of weighted indicators for 19 conditions associated with postdischarge 1-year mortality and the Selim physical comorbidity score of 30 chronic physical conditions were both created.28–30 These measures captured somewhat different aspects of comorbidity. A co-occurring chronic pain condition (codes 307.81, 337.1 with codes 250.6 or 249.6, 338.2, 338.4, 346, 710–739, 784.0) and postoperative new-onset major depression (exclusive of those with preoperative major depression) were also assessed, as well as indicators of antidepressant use following surgery and postoperative complications resulting in readmission.
Our primary outcome of interest was diagnosed SBI during the 3 years following the index surgery (FY2006–FY2009). SBI was identified by ICD-9-CM diagnosis codes in patient utilization files (V62.84: suicidal ideation, E950-E958: suicide and self-inflicted injury).31 Most SBI diagnoses occurred in inpatient settings. Pre-surgical history of SBI was also determined.
Means and frequencies were calculated for patient baseline demographics, surgical factors, and postoperative characteristics. We then assessed factors associated with the risk of experiencing SBI during a 3-year postoperative period through FY2009 following patients’ index surgery in FY2006. With patients’ race and Hispanic ethnicity as our primary predictors of interest, postsurgical risk of SBI was estimated using a Cox proportional hazards model, adjusting for patient demographics and clinical characteristics. Covariates included age, gender, marital status, and VHA priority 1. For the proposed model, age was divided by 10 to assess a decade association and make the estimated coefficients easier to interpret. Baseline comorbidity and severe mental illnesses, co-occurring pain diagnosis, and SBI before the index surgery were included in the model, as well as postsurgery depression, antidepressant use, and postoperative complications. Results of the Cox proportional hazards model were reported as hazard ratios (HRs) with an accompanying 95% confidence interval (CI). Positive associations between covariates and postsurgical risk for SBI were determined for HRs with values greater than 1, when the 95% CI excludes 1. Similarly, inverse associations were established for estimated HRs between 0 and 1 (95% CI excludes 1).
Among the 89 995 patients studied, 2836 had a postoperative SBI diagnosis over the next 3 years (3.2%; Table 1). The sample averaged 64 years of age (SD = 12) and included 4% women, 6% Hispanic, and 18% African American veterans; 884 individuals identified themselves as both Hispanic and African American (1%). Most patients qualified for VHA care via low income (41% priority 5), whereas 23% were categorized with high levels of service-connected disability (priority 1). Among the types of surgical operations studied, 25 656 patients had organ-related operations, 21 964 bone or joint-related, 8331 cancer-related, 41 532 vascular operations, and 2889 had amputations (Table 2).
TABLE 1— Characteristics of a Sample of Veterans Undergoing Major Surgery in the Veterans Health Administration by Race: Surgical Treatment Outcomes for Patients with Psychiatric Disorders Study, United States, 2006–2009
| Characteristics | African American (n = 16 252; 18.1%), No. (%) or Mean ±SD | Other Race (n = 2164; 2.4%), No. (%) or Mean ±SD | White (n = 71 579; 79.5%), No. (%) or Mean ±SD | Total (n = 89 995), No. (%) or Mean ±SD |
| Female | 899 (5.5) | 123 (5.7) | 2472 (3.5) | 3494 (3.9) |
| Hispanic | 884 (5.4) | 308 (14.2) | 4059 (5.7) | 5251 (5.8) |
| Married | 6314 (38.9) | 1121 (51.8) | 37 277 (52.1) | 44 712 (49.7) |
| Substance abuse diagnosis | 5504 (33.9) | 627 (29.0) | 21 117 (29.5) | 27 248 (30.3) |
| Served in Operation Enduring Freedom/Iraqi Freedom | 112 (0.7) | 23 (1.1) | 399 (0.6) | 534 (0.6) |
| Pain disorders | 5175 (31.8) | 789 (36.5) | 23 116 (32.3) | 29 080 (32.3) |
| Use of antidepressants, preoperative | 4807 (29.6) | 755 (34.9) | 25 128 (35.1) | 30 690 (34.1) |
| Severe mental illness, preoperative | ||||
| Schizophrenia | 557 (3.4) | 45 (2.1) | 1177 (1.6) | 1779 (2.0) |
| Bipolar disorder | 233 (1.4) | 47 (2.2) | 1389 (1.9) | 1669 (1.9) |
| PTSD | 1137 (7.0) | 228 (10.5) | 4597 (6.4) | 5962 (6.6) |
| Depression | 354 (2.2) | 64 (3.0) | 1791 (2.5) | 2209 (2.5) |
| No severe mental illness | 13 971 (86.0) | 1780 (82.3) | 62 625 (87.5) | 78 376 (87.1) |
| VHA priority status | ||||
| Priority 1 (50%–100% disabled by military service; no co-pays) | 3908 (24.0) | 663 (30.6) | 15 748 (22.0) | 20 319 (22.6) |
| Priority 2 (30%–40% disabled) | 1034 (6.4) | 150 (6.9) | 4055 (5.7) | 5239 (5.8) |
| Priority 3 (10%–20% disabled or special cohorts; e.g., POW) | 1455 (9.0) | 196 (9.1) | 6464 (9.0) | 8115 (9.0) |
| Priority 4 (catastrophically disabled/homebound) | 1916 (11.8) | 168 (7.8) | 5696 (8.0) | 7780 (8.6) |
| Priority 5 (very low income) | 6509 (40.1) | 777 (35.9) | 29 569 (41.3) | 36 855 (41.0) |
| Priority 6 (era-related cohorts such as post-Gulf War) | 154 (0.9) | 23 (1.1) | 889 (1.2) | 1066 (1.2) |
| Priority 7 (agreed to co-pays for all care) | 289 (1.8) | 49 (2.3) | 1984 (2.8) | 2322 (2.6) |
| Priority 8 (agreed to co-pays for all care) | 987 (6.1) | 138 (6.4) | 7174 (10.0) | 8299 (9.2) |
| Age, y | 61.5 ±12.2 | 63.1 ±12.1 | 64.9 ±11.6 | 64.2 ±11.8 |
| Selim physical comorbidity score | 4.4 ±2.2 | 4.5 ±2.2 | 4.5 ±2.3 | 4.5 ±2.3 |
| Charlson comorbidity score | 2.8 ±2.6 | 2.6 ±2.5 | 2.6 ±2.5 | 2.6 ±2.5 |
Note. POW = prisoner of war; PTSD = posttraumatic stress disorder; VHA = Veterans Health Administration. The sample size was n = 89 995.
TABLE 2— Operations and Postoperative Characteristics, Suicidality Among Veterans by Race Following Surgery: Surgical Treatment Outcomes for Patients with Psychiatric Disorders Study, Veterans Health Administration, United States, 2006–2009
| Characteristic/Index Surgery | African-American, No. (%) | Other Race, No. (%) | White, No. (%) | Total, No. (%) |
| Organ | ||||
| Digestive, excluding liver/spleen | 1995 (12.3) | 252 (11.6) | 7800 (10.9) | 10 047 (11.2) |
| Urinary system | 1588 (9.8) | 158 (7.3) | 5845 (8.2) | 7591 (8.4) |
| Hernia | 574 (3.5) | 72 (3.3) | 2946 (4.1) | 3592 (4.0) |
| Cholecystectomy | 474 (2.9) | 82 (3.8) | 2498 (3.5) | 3054 (3.4) |
| Appendectomy | 207 (1.3) | 39 (1.8) | 1126 (1.6) | 1372 (1.5) |
| Bone or joint | ||||
| Hip/knee | 1491 (9.2) | 240 (11.1) | 7325 (10.2) | 9056 (10.1) |
| Musculoskeletal | 1072 (6.6) | 163 (7.5) | 4670 (6.5) | 5905 (6.6) |
| Back or spine | 987 (6.1) | 125 (5.8) | 3798 (5.3) | 4910 (5.5) |
| Fractures | 277 (1.7) | 44 (2.0) | 1772 (2.5) | 2093 (2.3) |
| Cancers | ||||
| Head/neck cancer | 111 (0.7) | 12 (0.6) | 775 (1.1) | 898 (1.0) |
| Lung cancer | 202 (1.2) | 25 (1.2) | 1061 (1.5) | 1288 (1.4) |
| Prostate cancer | 557 (3.4) | 40 (1.8) | 1469 (2.1) | 2066 (2.3) |
| Colorectal cancer | 710 (4.4) | 101 (4.7) | 3268 (4.6) | 4079 (4.5) |
| Vascular | ||||
| Other vascular surgery | 6242 (38.4) | 860 (39.7) | 29 602 (41.4) | 36 704 (40.8) |
| Coronary artery bypass graft | 551 (3.4) | 126 (5.8) | 4151 (5.8) | 4828 (5.4) |
| Amputations | ||||
| Below knee amputation | 169 (1.0) | 0 (0) | 417 (0.6) | 596 (0.7) |
| Above knee amputation | 144 (0.9) | 0 (0) | 311 (0.4) | 463 (0.5) |
| Other amputation | 459 (2.8) | 41 (1.9) | 1330 (1.9) | 1830 (2.0) |
| Postoperative characteristics | ||||
| Suicidal behavior and ideation | 642 (4.0) | 64 (3.0) | 2130 (3.0) | 2836 (3.2) |
| Died within 30 d | 1259 (7.7) | 183 (8.5) | 5867 (8.2) | 7309 (8.1) |
| Died within 3 y | 4198 (25.8) | 508 (23.5) | 18 639 (26.0) | 23 345 (25.9) |
| Postoperative major depression | 2096 (12.9) | 308 (14.2) | 10 203 (14.3) | 12 607 (14.0) |
| Postoperative use of antidepressants | 5054 (31.1) | 771 (35.6) | 26 162 (36.5) | 31 987 (35.5) |
| Readmission for complications | ||||
| Myocardial infarction | 723 (4.4) | 109 (5.0) | 4052 (5.7) | 4884 (5.4) |
| Deep vein thrombosis | 294 (1.8) | 31 (1.4) | 993 (1.4) | 1318 (1.5) |
| Pneumonia | 566 (3.5) | 89 (4.1) | 2773 (3.9) | 3428 (3.8) |
| Respiratory failure | 184 (1.1) | 18 (0.8) | 715 (1.0) | 917 (1.0) |
| Sepsis | 252 (1.6) | 22 (1.0) | 838 (1.2) | 1112 (1.2) |
| Wound infection | 75 (0.5) | 14 (0.6) | 363 (0.5) | 452 (0.5) |
| Other postoperative readmission | 398 (2.4) | 66 (3.0) | 2245 (3.1) | 2709 (3.0) |
In the multivariable model, in veterans who received major surgery in FY2006, the risk of SBI was greatly increased for those with a history of SBI (HR = 5.4; 95% CI = 4.7, 6.2), schizophrenia (HR = 3.3; 95% CI = 2.9, 3.8), bipolar disorder (HR = 3.0; 95% CI = 2.6, 3.4), or new-onset postoperative depression (HR = 2.8; 95% CI = 2.6, 3.1). African American veterans had an approximately 21% increased hazard rate of SBI (HR = 1.2; 95% CI = 1.1, 1.3). Other risk factors for SBI included having postoperative complications, preoperative chronic pain, and a higher burden of chronic illness per the Selim Index, whereas being female, older, or married was protective. When holding all other variables constant, an increase in the Selim comorbidity score of 1 corresponded to an increase of 5% in the rate of SBI. When holding all other variables constant, an increase in age of 1 decade corresponded to a decrease of 28% in the hazard rate of SBI. Female, married, and priority 1 veterans similarly displayed a decreased risk of SBI estimated at 18%, 24%, and 12%, respectively. However, an association between Hispanic ethnicity and increased risk for SBI following surgery was not observed (HR = 1.10; 95% CI = 0.95, 1.28; Table 3). In an exploratory model, type of surgery was also not related beyond cancer-related surgeries, which were negatively associated with SBI.
TABLE 3— Factors Associated With Suicidal Behavior and Ideation During the Three Years Following Major Surgery: Surgical Treatment Outcomes for Patients with Psychiatric Disorders Study, Veterans Health Administration, United States, 2006–2009
| Variables | HR (95% CI) |
| Age in decadesa | 0.72 (0.69, 0.75) |
| Femalea | 0.82 (0.71, 0.95) |
| Marrieda | 0.76 (0.70, 0.82) |
| Hispanic | 1.10 (0.95, 1.28) |
| African Americana | 1.21 (1.10, 1.32) |
| Other non-White race | 0.80 (0.62, 1.02) |
| Priority 1 (no co-pay; highly disabled)a | 0.88 (0.81, 0.96) |
| Selim physical comorbidity score (range = 0–34)a | 1.05 (1.02, 1.07) |
| Charlson comorbidity score (range = 0–19) | 1.00 (0.98, 1.02) |
| Schizophreniaa | 3.31 (2.85, 3.84) |
| Bipolar disordera | 2.99 (2.60, 3.44) |
| Major depressive disordera | 1.27 (1.09, 1.48) |
| Post-traumatic stress disordera | 1.48 (1.32, 1.67) |
| Postoperative major depressive disordera | 2.81 (2.58, 3.06) |
| Antidepressant usea | 1.90 (1.73, 2.09) |
| Postoperative complicationsa | 1.35 (1.22, 1.49) |
| History of suicidal behavior and ideationa | 5.37 (4.68, 6.15) |
| Pain disordersa | 1.13 (1.05, 1.22) |
Note. CI = confidence interval; HR = hazard ratio. The sample size was n = 89 995.
a 95% confidence level excludes 1.0.
From FY2006 to FY2009, African American race was associated with an increased likelihood of experiencing SBI in the 3 years following major surgery in the VHA. Although the experience of major surgery might not typically be thought of as associated with mental illness, in an integrated care delivery system such as the VHA, the entirety of a patient’s health care, which addresses mental and physical disorders, can be captured and examined for insight into an individual’s overall experience, as well as providing an overview of unexpected areas of disparity or nonequivalent risk. In terms of mechanism, SBI represented untreated mental anguish, primarily depression, which might lead to decreased healing and increased new-onset disease after surgery; this is an understudied area.32,33 Consistent with compassionate care and the VHA’s mission to reduce veteran suicide, providers must address not only physical pain or other postoperative symptoms, but also new or exacerbated mental distress. Postoperative recovery is a process that can be problematic, marked by pain and disability, and for some, suicidal ideation occurs in the process of seeking a solution to the problem.34 The VHA has a system-wide suicide prevention program that includes outreach to patients identified by a provider as at-risk for suicide; a diagnosis of SBI would flag a patient as at-risk.
To help eliminate disparities in mental health outcomes after surgery and to help manage the care of those at heightened risk for suicide after surgery, patient activation interventions might need to be tailored for cultural subgroups to enable minority patients to better cope with the perioperative period. According to some studies, minority race/ethnicity patients reported more severe pain and pain-related interference with physical function than their White counterparts.22,35 They were more likely to be prescribed medication that was inadequate for their pain intensity after a major surgery.35 Was this an issue of inadequate pain management? Prescriber characteristics and habits, or patient communication difficulties and intrapersonal barriers to requesting and using pain medications must be explored in future research to resolve this question. Alternatively, the association observed might be the result of a confounder associated with both race and postoperative SBI, such as cultural or socioeconomic factors, that led African American veterans with complex mental and physical health needs to use the VHA for inpatient and outpatient care.
Mental health professionals observed that individuals often reported suicide as an option that represented relief from long-term suffering or the burden they felt they placed on others.36 According to Contrada et al., there were 4 phases of the surgery experience37; each phase presented its own unique challenges and coping issues. Phase I included immediate postoperative complications. Phase II included the pain, discomfort, fatigue, and reduced capacity for physical activity after surgery. Phase III included the inability to enact social roles. Phase IV included long-term management of a possibly chronic medical condition. The challenge for clinicians is to recognize and identify high risk SBI in those moving through these phases after surgery. Effective clinical care for mental, physical, and substance disorders is critical to protect against SBI, especially at times of stress that might include major health care events.38,39 Differences in postoperative coping by race/ethnicity should be assessed in future research to spur development of tailored postoperative care plans that take into account variations in ability to cope with pain, disability, and recovery.
We relied on retrospective archival data, and therefore, had no measures of subjective experience or severity of illness and pain. The presence of non-White patients in the sample depended on the military enlistment demographic characteristics of up to 50 years ago; more recent military cohorts had higher proportions of African American and Hispanic veterans than did those who served in the Vietnam War, Korean War, or earlier wars. Most patients were men; results might not generalize to women. Results may not also be applicable to nonveterans; the VHA caters to the most disadvantaged veterans, who are not representative of US residents. SBI occurring in other hospital systems was not captured. Rates reported likely underestimated actual rates. Patients who underwent surgery in the VHA might be systematically different from nonsurgery VHA patients. In particular, younger veterans were underrepresented; results might not generalize well to new veterans from Iraq or Afghanistan (n = 534; < 1% of our sample). Mitigating these limitations were the size of the cohort, which allowed a broad overview of the issues, and the importance of understanding suicide prevention for citizens cared for in the publicly funded federal health care system.
Our findings demonstrated that the months following surgery might be a heightened time for SBI among minority populations treated in the VHA, with African American patients experiencing a 21% increased rate of this outcome. To eliminate health disparities in perioperative health care, the VHA and other health care institutions must be continually vigilant in monitoring and managing the care of those experiencing health care–related pain and anxiety because health care encounters designed to increase their life expectancy might put them at heightened risk for suicide.
Acknowledgments
This work was supported by a research award from Veterans Health Administration, Health Services Research & Development #IIR-09-335 to Central Texas Veterans Health Care System to partially support the salaries of L. A. Copeland, J. E. Zeber, M. J. Pugh, and D. J. MacCarthy. Additional support was provided by the Center for Applied Health Research, a center jointly sponsored by Central Texas Veterans Health Care System and Scott & White Healthcare, in Temple, Texas, including the Scott & White Minority Health Research Predoctoral Fellowship awarded to R. T. McIntyre.
Note. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.
Human Participant Protection
Our study was approved by the institutional review boards of the Central Texas Veterans Health Care System and the South Texas Veterans Health Care System before initiation.
