Search

Older veterans are vastly underrepresented in studies that shape national suicide prevention strategies. This is of great concern because factors that impact younger veterans may not be as robust in later life. Although younger veterans have higher rate of suicide, the highest counts of death by suicide are in older veterans. However, it remains unclear from the extant literature what factors may influence increased or decreased risk of late-life suicide in veterans. The objective of this systematic review was to identify risk and protective factors related to suicide outcomes (i.e., ideation, attempt, death, or suicide-related behavior [SRB]) among older veterans. Furthermore, it offers data regarding future study directions and hypothesis generation for late-life suicide research and for informing potential intervention and prevention efforts in this area. We searched 4 databases from inception up to May 5, 2022. We screened 2,388 abstracts for inclusion and 508 articles required full text review. The final sample included 19 studies published between 2006 and 2022. We found five domains of factors studied (i.e., neuropsychiatric, social determinants of health, aging stereotypes, residential and supportive housing settings, and multifactorial-neuropsychiatric/mental health and physical health) with more risk factors than protective factors reported. Across the three suicide outcomes only neuropsychiatric factors were consistently identified as risk factors. Neuropsychiatric factors also comprised the largest group of risk factors studied. More innovative targets to consider for intervention and more innovative methods to predict suicide in late-life are needed. There is also continued necessity to design suicide prevention interventions for older veterans given lethality trends.

Introduction Given minimum sentencing laws, more arrests in later age, and population aging, the US incarcerated population is aging, with adults 50+ projected to account for 1/3 of this population by 2030. As most incarcerated persons are released, the number of those being released in mid- to late-life is growing. Evidence indicates accelerated aging (e.g., higher prevalence of geriatric conditions and ADL disability) among those 50+ with incarceration histories as compared to their never-incarcerated (NI) peers. Moreover, those reentering the community following incarceration are likely to have many risk factors for Alzheimer's disease and related dementias (AD/ADRD), such as PTSD and low education attainment. However, the burden of AD/ADRD, mild cognitive impairment (MCI), and other related diseases in this population is largely unknown. Our objective was to characterize the prevalence of these conditions in the 50+ reentry population using a national sample. Understanding this burden is a crucial first step to developing evidence-based policies that support their healthcare and reentry needs. Methods We estimated the prevalence of dementia (AD/ADRD and MCI) and other related diseases in 28,267 veterans aged 50+, who were Medicare beneficiaries and reentered the community from incarceration between October 1, 2007 and December 31, 2017 (referred to as “reentry veterans”). In addition, we created a 1:5 matched comparison cohort of 141,335 never-incarcerated veterans (referred to as “NI veterans”) matched on age, year of release/Medicare eligibility year, and sex. To create these cohorts, we linked two national databases: (1) Centers for Medicare & Medicaid Services (CMS), for medical claims/diagnoses and incarceration release date; and (2) National Patient Care Database (NPCD), for VA inpatient and outpatient services. To characterize the prevalence of dementia and related diseases, we used ICD-9 clinical diagnosis codes in inpatient and outpatient medical records from October 1, 2007 to September 30, 2015, and ICD-10 codes through study end. The 10-year prevalence of dementia diagnoses include MCI and AD/ADRD diagnoses: Alzheimer's disease, Frontotemporal dementia, Lewy body dementia, vascular dementia, mixed dementia (2+ dementias), dementia associated with Parkinson's disease, senile dementia, dementia not otherwise specified, and dementia associated with HIV/AIDS. The 10-year prevalence of dementia related diseases included Parkinson's disease, Huntington's disease, and multiple sclerosis. Both 10-year prevalence estimates were determined by reentry and NI status across age groups 50-64, 65-74, 75-84, and 85+. Differences between reentry and NI cohorts age groups were evaluated with chi-square tests. We examined the potential differing trend across age groups by statistically testing the interaction of age groups and reentry status for each disorder using logistic regression. Statistical significance was defined as p Results Both cohorts had an average age of 63 (SD 8) and were 97% male. The reentry cohort was 66% white, 30% black, and 4% other. The never incarcerated cohort was 77% white, 18% black, and 4% other. Figure 1 presents the 10-year prevalence of dementia diagnoses according to age group for both reentry and NI cohorts. The reentry cohort had, on average, a 32% higher 10-year prevalence of dementia diagnosis than the NI cohort (Figure 1), with greatest differences at ages 65-74. At ages 65+, 18% of reentry veterans had an AD/ADRD diagnosis, versus 12% in the NI cohort. The estimates go up to >50% for reentry veterans 85+ versus 43% for NI (p-interaction Conclusions These findings indicate that reentering community after incarceration aged 50+ is associated with higher occurrences of AD/ADRD and MCI, but not necessarily with dementia related diseases. Because we did not have access to the reentry group's incarceration medical records, we were unable to obtain diagnoses during incarceration. This is the first study to characterize the burden of AD/ADRD and related diseases among those with a recent history of incarceration. The higher prevalence of AD/ADRD and MCI in the reentry cohort across age groups may indicate accelerated cognitive aging among those with a history of incarceration. Our findings highlight the importance of identifying those with MCI during incarceration so reentry care transition services can be put into place to slow the progression of cognitive decline. Funding This work was supported by RF1 MH117604 from the National Institute of Mental Health (principal investigators: Lisa C. Barry, Ph.D., M.P.H. and Amy L. Byers, Ph.D., M.P.H.).

Objective To compare occurrence of clinically diagnosed psychiatric disorders and suicidal behavior (mental health disorders) across dementia subtypes in the largest healthcare system in the United States. Methods We aggregated two national databases (Department of Veterans Affairs [VA] National Patient Care Database, National Suicide Prevention Applications Network [SPAN]) and estimated 2-year prevalence of mental health disorders across five dementia subtypes during fiscal years 2012–2013. Using VA healthcare systems throughout the United States, the sample included 56,296 older patients (≥50 years) with Alzheimer's disease (AD; n = 30,578), vascular dementia (VD; n = 17,924), frontotemporal dementia (FTD; n = 1,181), Lewy body dementia (LBD; n = 3,194), and mixed dementia (MD; n = 3,419). Mental health disorders were determined by International Classification of Diseases, Ninth Revision, Clinical Modification codes and the National SPAN. Results Roughly 25% of patients had at least one mental health disorder, with 2-year prevalence reaching 30%–45% in FTD, VD, LBD, and MD. Compared with other subtypes, patients with FTD had the highest prevalence of mood (19%), anxiety (20%), and substance use (19%) disorders, as well as suicidal behavior (4%), with nearly 0.5% with a suicidal plan/attempt. Those with VD also showed a high prevalence of these disorders (14%–17%). Although patients with LBD and MD had a slightly lower prevalence of mood and anxiety disorders (12%–15%), they had a much lower prevalence of substance use disorders (9%) and suicidal behavior (2%). Patients with AD had the lowest 2-year prevalence of all mental health disorders ( Conclusion Occurrence of mental health disorders is high and differs across dementia subtypes, highlighting the importance of reducing the burden of mental health disorders in dementia subtypes.

Introduction Late life is often associated with declining health, with an estimated 11-32% of community-dwelling older adults living with frailty. Frailty, a durable phenotype associated with depression, is often measured as an accumulation of physical and mental deficits. This underlying distress may be associated with suicidal thoughts. Frailty has never been assessed as a risk factor for suicide on a national level but may be useful as a prognostic marker. Our primary objective was to determine the relationship of frailty to subsequent risk of suicide attempt and death by suicide across age groups in a large national cohort of veterans aged 50+. With >70% of the veteran population aged 50+ and ∼70% of veteran suicides occurring in older veterans, this study has potential to be highly impactful. Methods We performed a longitudinal cohort study on a national sample of mid to late-life veterans: All veterans aged 50+ who used VA healthcare services during study baseline of fiscal years 2012-2013 (N=5,059,526) with ongoing follow-up through the end of calendar year 2017. We linked four national databases: (1) National Patient Care Database (NPCD), for VA inpatient and outpatient services; (2) Centers for Medicare & Medicaid Services (CMS), for non-VA inpatient and outpatient services; (3) Suicide Prevention Applications Network (SPAN), for information on non-fatal suicide attempts; and (4) Mortality Data Repository (MDR), for cause-specific death information and dates. Demographic information for age, gender, and race/ethnicity was obtained from NPCD and/or CMS records. To study the role of frailty and its connection to late-life suicide-related behaviors, we used a cumulative deficit method and frailty index (FI) developed by Orkaby et al. (2019). This index integrates 31 variables (deficits) from ICD-9/10 diagnoses and procedure codes and HCPCS (Healthcare Common Procedure Coding System) codes from NPCD and CMS data. Variables relate to morbidity (e.g., diabetes), functional status (e.g., use of durable medical equipment), cognition and mood (e.g., depression), sensory loss (e.g., blindness), and other geriatric conditions (e.g., weight loss in the past year). FI scores were calculated at baseline index date of study entry and within two years prior, summing the number of accrued deficits then dividing by 31 (total number of possible deficits), producing scores from 0 to 1. We defined Non-frail as FI≤0.1, Pre-frail as 0.1 0.4. Suicide attempt and death outcomes were derived from SPAN and MDR databases through 2017. Non-fatal suicide attempts were derived from SPAN, based on standardized clinician reports uploaded to the national database. Fatal suicide attempts were derived from MDR with ICD-10 codes X60-X84 and Y87.0. Cumulative incidence of any suicide attempt was plotted by patient age in years across frailty categories. We used Cox proportional hazards regression to examine time to suicide attempt at follow up, with age as the time scale and censoring at end of study. We examined multivariable models adjusted in steps: Model 1, unadjusted; Model 2, adjusted for demographics (race, gender, region); Model 3, Model 2 + adjusted for substance use disorder and PTSD; Model 4, Model 3 + adjusted for Charlson Comorbidity Index (CCI). We did not adjust for depression as it was included in the FI. We adjusted for CCI to examine the FI's robustness as a marker for suicide risk in the context of another global index. Results Our sample's average age was 68 (SD 11), 94% male, 77% white, and 14% black. Fifty-one percent of veterans aged 50+ were non-frail, while 26% were pre-frail, 12% mildly frail, 6% moderately frail, and 4% severely frail. The average FI score was 0.10. There were 29,679 suicide attempts at follow up, with 9,659 fatal attempts. Cumulative incidence rates of suicide attempt were significantly higher for more frail than less frail individuals (Figure 1). Our fully adjusted model (Table 1, Model 4) showed a strong dose-response relationship between frailty severity and risk of suicide attempt (fatal and non-fatal), with hazard ratios (HR) increasing from 1.55 for Pre-frail to 2.50 for Severely frail individuals. An attenuated HR effect was seen for fatal attempts: ranging from 1.29 for Pre-frail to 1.56 for Severely frail individuals. Furthermore, we found a significant interaction of FI by age groups 50-64 and 65+ (p-interaction Conclusions Our findings suggest a cumulative deficit FI may be a strong prognostic marker associated with risk of suicide attempt in young-old to oldest-old beyond other factors such as substance use disorder, PTSD, and CCI. This study provides clinicians with a better understanding of potential frailty-related suicide risk in patients aged 50+. Funding I01 CX001119-05 (PI: Byers) Veterans Administration Clinical Science Research & Development

Introduction Although posttraumatic stress disorder (PTSD) may increase risk of suicide in mid- to late-life, findings have been controversial, in large part, because studies have not accounted for comorbid disorders. Moreover, little is known about association between mid- to late-life PTSD and apparent accidental death by overdose, which could be due to suicidal intent. Thus, we conducted a comprehensive study to examine PTSD and risk of death by suicide, suicide attempt, and unintended death by overdose among veterans in mid- to late-life. Methods This nationwide cohort study included all veterans ≥50 years (N=1,038,980) seen in the Veterans Health Administration starting 2012-2013 (baseline) and followed through December 31, 2017. Veterans with a PTSD diagnosis were propensity-matched 1:1 with patients without PTSD. They were matched on age, gender, race/ethnicity, income, education, homelessness, service connection, Charlson comorbidity index, and neuropsychiatric disorders. Dates of suicide attempt and cause-specific mortality at follow up were defined using the National Suicide Prevention Applications Network and the Mortality Data Repository, respectively. PTSD was identified by two or more ICD-9 codes in electronic medical records at baseline. PTSD and risk of suicide-related outcomes were assessed using multivariable models. Fine-Gray proportional hazards regression was used to examine time to outcome event at follow up. Models were next adjusted in steps, with the final adjustment accounting for sociodemographics, Charlson comorbidity index, and neuropsychiatric diagnoses. Other outcomes examined using similar methods included death by drug overdose (i.e., antiepileptics/sedative-hypnotics, narcotics, and unspecified/other cause-specific drug categories) for suicide, unintentional, and undetermined death. Results Of 1,038,980 patients in our study, 519,490 had PTSD. The mean participant age was 63.56, 59,542 (5.73%) were women, and 765,572 (73.68%) were non-Hispanic White. Although firearms were by far the leading method of suicide death for those with and without PTSD, there was no significant between-group difference related to firearms. However, veterans with PTSD were significantly more likely to die by suicide by drug overdose compared with those without a PTSD diagnosis. As shown in Table 1, after adjusting for sociodemographics, Charlson comorbidities, and neuropsychiatric diagnoses, the hazard ratio for risk of suicide in veterans with PTSD was 1.57 (95% CI, 1.51, 1.63) for any suicide attempt; 1.09 (95% CI, 1.01, 1.19) for death by suicide; 1.43 (95% CI, 1.32, 1.56) for drug overdose death overall; 1.53 (95% CI, 1.21, 1.94) for suicide death by drug overdose. Further analyses explored type of drug involved in overdose and demonstrated that veterans with PTSD had hazard ratios of 1.44 (95% CI, 1.26, 1.65) for narcotics and 1.48 (1.31-1.68) for other /unspecified drugs (Figure 1). We also found that having PTSD increased unintentional death, suicide death, and undetermined intent for narcotics by 1.4 to 2.4 times, and for other/unspecified drugs by 1.4 to 1.8 times. Conclusions The findings suggest that veterans with PTSD aged ≥50 years in VA healthcare, have persistently elevated risk for suicide attempts and suicide death, particularly by drug overdose. This current work may inform clinical practice targeting late-life suicide prevention efforts by promoting improved prescribing and drug-monitoring. Future medication recommendations to treat PTSD may need to consider less risky drugs than narcotics, as well as emphasize inquiring about potential illicit drug use and access, for older adults. More research is needed to elucidate the relationship of PTSD to increased risk of mid- to late-life death by suicide and unintended drug overdose, exploring what other and unspecific drugs do and do not exacerbate this risk. Funding This work was supported by Award Number I01 CX001119 from the Clinical Science Research & Development Service of the U.S. Department of Veterans Affairs (VA) Office of Research and Development (PI: Dr. Byers). Support for the Mortality Data Repository (MDR; previously known as the Suicide Data Repository or SDR) and Suicide Prevention Applications Network (SPAN) is provided by the VA Center of Excellence for Suicide Prevention. Support for VA/CMS data is provided by the Department of Veterans Affairs, VA Health Services Research and Development Service, VA Information Resource Center (Project Numbers SDR 02-237 and 98-004).

Objective Little is known about key factors associated with use of mental health services across the life course. This study determined key socioeconomic, social support, psychiatric, and medical predictors of services use in younger, middle, and older age. Design, Setting, Participants, Measurements The sample included 3,708 adults with DSM-IV–based mood, anxiety, and substance use disorders in the Collaborative Psychiatric Epidemiology Surveys. Key predictors of mental health services use for each age group were systematically determined by multivariable models, and exploratory analyses examining potential effect modification by race–ethnicity and sex were assessed by interaction terms. Statistical analyses included complex design-corrected and weighted logistic regression analyses that provide results generalizable to the United States. Results Psychiatric and medical issues such as prior suicidal behavior, comorbid psychiatric disorders, and perceived cognitive impairment increased odds of mental health services use in younger, middle, and older age. Chronic medical conditions also influenced services use in younger and older age, with their impact on use across age potentially modified by racial–ethnic disparities (p interaction = 0.01). Moreover, socioeconomic factors like marital status influenced use in middle and older age, where being divorced, separated, widowed, or never married encouraged use. The effect of marital status on use across age was also potentially modified by racial–ethnic disparities (p interaction = 0.02). Conclusions Key socioeconomic, social support, psychiatric, and medical predictors uniquely influence use of mental health services across the life course. These findings will help inform efforts to encourage greater services use by adults across the life course in need of care.

Older veterans have the highest number of lives lost to suicide, make up majority of the veteran population, and are highly likely to experience conditions (e.g., chronic pain, sleep disorders, musculoskeletal) associated with commonly prescribed medications that are potential markers for suicide risk (hereafter referred to as “high-risk” drug categories), including benzodiazepines, sedative-hypnotics, opioids, antidepressants, antipsychotics, and antiepileptics. The research presented in this session will highlight important patterns in high-risk drug prescribing and use and related outcomes in late life. The presentations will highlight various groups of older veterans that may be important to consider and, yet, neglected in the polysubstance and polypharmacy and suicide prevention conversation, including those who recently attempted suicide, veterans with late-life posttraumatic stress disorder (PTSD), and older veterans transitioning from prison to community. Drs. Maust and Morin will present findings on polypharmacy prescribing and use of high-risk medications in late life. First, drawing from the literature and his own research studies, Dr. Maust will provide a big picture overview of trends in polypharmacy among older adults in the U.S., including older veterans. Next, Dr. Morin will describe high-risk profiles of use in the 90 days prior to a first documented suicide attempt among veterans >=65 years who attempted suicide between 2012 and 2014, using Veterans Health Administration (VHA) medical record data. In addition, she will examine associations of these profiles with diagnostic, lethality, and utilization factors. Drs. Byers and Barry will speak about more specific subgroups of older veterans, with recommendations for possible intervention to reduce suicide, unintended death by overdose, and substance use disorder (SUD)-related emergency department (ED) visits and hospitalizations, emphasizing importance of care transition models for prevention. Dr. Byers will present a cohort study of veterans >=50 years seen in VHA. Findings provide evidence that late-life PTSD increases risk of suicide attempts and suicide by overdose (and accidental and undetermined death), independent of comorbid disorders. Furthermore, findings will be presented about the relationship of late-life PTSD to cause-specific high-risk drugs. Finally, Dr. Barry will provide an example of increased risk for SUD-related ED visits and hospitalizations (found in the literature to be strongly associated with suicide attempts) among older veterans transitioning from prison to community. Findings show that those re-entering society following prison sentence are at significantly higher risk for SUD-related ED visit or hospitalization compared to matched controls. These results highlight urgent need for assistance during this vulnerable period of transition.

Objectives To determine if late-life posttraumatic stress disorder (PTSD) is associated with cardiovascular disease in a sample of older veterans, and whether the association is independent of medical and psychiatric comorbities. Design Retrospective cohort study conducted using the Department of Veterans Affairs (VA) National Patient Care Database (2000–2011). Setting VA medical centers in the United States. Participants A total of 138,341 veterans 55 years and older without cardiovascular disease at study baseline (2000–2003). Measurements PTSD and cardiovascular disease (as defined by diagnoses of: cerebrovascular disease [CVD], congestive heart failure [CHF], myocardial infarction [MI], and peripheral vascular disease [PVD]) were identified by ICD-9 codes during study baseline (2000–2003) and follow-up (2004–2011), respectively. Results 3% of veterans (N = 4,041) had a baseline diagnosis of PTSD. Unadjusted increased risk of incidence of CVD was 80%, CHF was 56%, MI was 82%, and PVD was 60% in veterans with PTSD compared with those without PTSD. After adjustment for demographics, medical comorbidities, substance use, and psychiatric comorbidities, veterans with late-life PTSD were at a 45% increased risk for incident CVD, 26% increased risk for incident CHF, 49% increased risk for incident MI, and 35% increased risk for PVD compared with veterans without late-life PTSD. Conclusions Findings highlight the longitudinal impact of PTSD on increasing the incidence of cardiovascular disease in older adults. This study implies the need for greater monitoring and treatment of PTSD in older persons, particularly older veterans, to assist in preventing adverse outcomes, such as cardiovascular disease, over the long term.

This session will highlight the particular challenges of addressing suicide among older veterans in VHA care using epidemiological and novel statistical approaches, and offer a promising way forward for early identification and prevention of late-life suicide. The research presented here will highlight various unique factors in older veterans that may be missing from the suicide prevention conversation, including the saliency of physical health, pain, high-risk subpopulations, and psychosocial interventions. Drs. Byers and Morin will present findings on late life suicide in VHA using medical record data for novel investigations of administrative data. First, using her database of over 5 million veterans receiving VHA care, Dr. Byers will provide a big picture overview of nationally based estimates of the occurrence of suicide-related outcomes, including non-fatal attempts and death by suicide, in veterans over age 50. Additionally, she will discuss factors that influence risk of a first attempt in later life, as well as factors that further influence, among survivors, who goes on to repeat an attempt and who does not. Next, Dr. Morin will describe physical and mental health comorbidity profiles of 2,131 veterans over 65 who attempted suicide between 2012 and 2014, and last saw a primary care provider prior to their attempt. The relatively minimal depression prevalence as well as high incidence of chronic pain in more than half of these attempters, as well as higher rates of fatality using firearms among these individuals compared to those with multiple mental health comorbidities will be discussed. These findings raise awareness about possible points of intervention in a group that may otherwise not be considered to be as high-risk as a population with significant comorbid mental health concerns. Drs. Barry and Van Orden will speak about more specific subgroups of older veterans, with recommendations for possible intervention to reduce unintended death and suicide, emphasizing the importance of transition of care models for prevention of late-life suicide. Dr. Barry will provide an example of increased risk for suicide and unintended death among older veterans transitioning from prison to the community. Findings show that those re-entering society following a prison sentence are at significantly higher risk for suicide attempt by multiple means, with higher rates of fatality compared to matched controls. These results highlight the urgent need for assistance during this vulnerable period of transition. Finally, Dr. Van Orden will discuss transition of care models more specifically and their importance in combating loneliness, a major risk factor for late-life suicide. She will share compelling findings related to a novel intervention currently being studied on the benefits of volunteer-provided peer companionship for older veterans transitioning from community living center (CLC) stays back into the community. This intervention is targeted to support upstream suicide prevention.

Objective The purpose of this study was to determine the impact of psychotherapy on cognitive functioning in older adults with late-life depression (LLD) and executive dysfunction. Methods Two hundred twenty-one adults aged 60 years and older participated in a randomized clinical trial comparing the efficacy of Problem Solving Therapy (PST) and Supportive Therapy (ST) for LLD. Cognitive performance on seven tests of executive functioning, verbal learning, and memory was evaluated at baseline, after 12 weeks of treatment, and at 24 weeks after the completion of treatment. Results Performance on a measure of executive functioning with a significant information processing speed component (Stroop Color and Word Test) improved after treatment, F (1, 312) = 8.50, p = 0.002, and improved performance was associated with a reduction in depressive symptoms but not treatment type. Performance on other measures of executive functioning, verbal learning, and memory did not change significantly after 12 weeks of psychotherapy treatment. Conclusion Our results suggest that improvements in cognitive functioning after psychotherapy treatment for depression in older adults with executive dysfunction are likely focal and not distributed across all cognitive domains. Although previous analyses reported that PST was superior to ST in the treatment of depression, this analysis indicated no difference between the two treatments with regard to improvements in cognitive functioning.

Objective To determine whether less severe depression spectrum diagnoses such as dysthymia, as well as depression, are associated with risk of developing dementia and mortality in a "real-world" setting. Design Retrospective cohort study conducted using the Department of Veterans Affairs (VA) National Patient Care Database (1997–2007). Setting VA medical centers in the United States. Participants A total of 281,540 veterans aged 55 years and older without dementia at study baseline (1997–2000). Measurements Depression status and incident dementia were ascertained from International Classification of Diseases, Ninth Revision codes during study baseline (1997–2000) and follow-up (2001–2007), respectively. Mortality was ascertained by time of death dates in the VA Vital Status File. Results Ten percent of veterans had baseline diagnosis of depression and nearly 1% had dysthymia. The unadjusted incidence of dementia was 11.2% in veterans with depression, 10.2% with dysthymia and 6.4% with neither. After adjusting for demographics and comorbidities, patients diagnosed with dysthymia or depression were twice as likely to develop incident dementia compared with those with no dysthymia/depression (adjusted dysthymia hazard ratio [HR]: 1.96, 95% confidence interval [CI]: 1.71–2.25; and depression HR: 2.18, 95% CI: 2.08–2.28). Dysthymia and depression also were associated with increased risk of death (31.6% dysthymia and 32.9% depression versus 28.5% neither; adjusted dysthymia HR: 1.41, 95% CI: 1.31–1.53; and depression HR: 1.47, 95% CI: 1.43–1.51). Conclusions Findings suggest that older adults with dysthymia or depression need to be monitored closely for adverse outcomes. Future studies should determine whether treatment of depression spectrum disorders may reduce risk of these outcomes.