Risk and protective factors associated with intentional selfharm among older community-residing home care clients in Ontario, Canada

Transcription

1 RESEARCH ARTICLE Risk and protective factors associated with intentional selfharm among older community-residing home care clients in Ontario, Canada Eva Neufeld 1, John P. Hirdes 2, Christopher M. Perlman 2 and Terry Rabinowitz 3 1 Centre for Rural and Northern Health Research, Laurentian University, Sudbury, Ontario, Canada 2 School of Public Health and Health Systems, University of Waterloo, Waterloo, Ontario, Canada 3 Psychiatry and Family Medicine, University of Vermont College of Medicine, Burlington, VT, USA Correspondence to: E. Neufeld, PhD, eneufeld@laurentian.ca Objective: We aim to concurrently examine risk and protective factors associated with intentional selfharm among community-residing older adults receiving home care services in Ontario, Canada. Methods: Administrative health data from the home care sector were linked to hospital administrative data to carry out the analyses. Home care data are collected in Ontario using the Resident Assessment Instrument Home Care (RAI-HC), an assessment tool that identifies strengths, preferences and needs of long-stay home care clients. The sample included Ontario home care clients aged 60 years or older assessed with the RAI-HC between 2007 and 2010 (N = 222,149). Multivariable analyses were performed using SAS. Results: Hospital records of intentional self-harm (ISH) were present in 9.3 cases per 1000 home care clients. Risks of ISH included younger age (60 74 years; OR = 3.14, CI: ), psychiatric diagnosis (OR = 2.29, CI: ), alcohol use and dependence (OR = 1.69, CI: ), psychotropic medication (OR = 1.94, CI: ) and depressive symptoms (OR = 1.58, CI: ). Protective effects were found for marital status and positive social relationships, yet these effects were more pronounced for men. Cognitive performance measures showed the odds of ISH 1.86 times higher for older adults with moderate to severe cognitive impairment. Conclusions: This study based on provincial data points to tangible areas for preventative assessment by frontline home care professionals. Of interest were the risk and protective factors that differed by sex. As demand for home care in Canada is expected to increase, these findings may inform home care professionals appraisal and approach to suicide prevention among community-residing older adults. Copyright # 2015 John Wiley & Sons, Ltd. Key words: aging; dementia; RAI-HC; resilience; risk factor; self-harm; suicide History: Received 29 August 2014; Accepted 16 December 2014; Published online in Wiley Online Library (wileyonlinelibrary. com) DOI: /gps.4259 Introduction Older adults have the highest rate of death by suicide across all age groups (WHO, 2005; CCSMH, 2006) but as many as 75% have never made a prior attempt (Conwell et al., 1990). The rate of suicide in later life is particularly higher for older men. Older men between 80 and 84 years have rates of suicide approximately six times greater than older women of the same age (i.e., 23.5 per 100,000 vs. 3.8 per 100,000 (Statistics Canada, 2010)). Suicide prevention research among older adults receiving home care services in Ontario has received little attention in literature. In Canada, approximately one million people are receiving home care services, and the majority is over 65 years old. As the expected demand for home care services increases, frontline home care professionals have a key role in the assessment and prevention of suicide among community-residing older adults.

2 E. Neufeld et al. There are a number of factors reported to increase risk for suicide among older adults including depression and other psychiatric illnesses, previous suiciderelated behavior, pain, poor physical health and functional decline (Waern, 2003; O Connell et al., 2004; Tadros and Salib, 2007; Li and Conwell, 2010; Qin 2011). A growing body of literature also identifies several protective and resiliency factors that act as buffers to suicide risk, such as meaningful relationships, social support and perceived meaning in life (Kposowa, 2000; Awata et al., 2005; Heisel, 2006; Rowe et al., 2006; Heisel and Flett, 2008). A gap exists in the literature that examines how risk and protective factors work in combination to either increase or mitigate suicide risk in later life. A greater understanding of risk and protective factors that confer suicide risk for older adults living in the community can assist in targeting high-risk individuals and provide interventions that can allow them to remain safe and independent in the community. The goals of our current research therefore are to concurrently explore suicide risk and protective factors among community-residing older adults receiving home care services in Ontario, Canada. As the rate of suicide is particularly higher for older men, this study will further explore significant gender differences that could have an impact upon approaches to suicide intervention and prevention. Methodology Sample This research was performed using secondary data that was collected in Ontario s home care sector and linked prospectively to hospital and emergency department records for the purpose of examining intentional self-harm. The sample consisted of Ontario long-stay home care clients, aged 60 years or older, that were assessed with the Resident Assessment Instrument Home Care (RAI-HC) between April 2007 and September 2010 (N = 222,149). Client s initial assessment records in home care were used in the analysis. These were linked to data on corresponding emergency department visits, hospital admission records and admissions to adult mental health beds where the diagnostic codes or problems were coded for intentional self-harm or clients were assessed as a danger/threat to themselves. The RAI-HC is psychometrically strong (Hirdes et al., 2008; Poss et al., 2008) and has several measures of psychiatric symptoms; but, it has no items on suicidal ideation or a history of previous suicide-related behavior. Thus, through linkages between home care data and hospital data, a prospective examination of communityresiding older adults with a record of intentional self-harm is possible. Data sources Ontario home care data were retrieved from the Home Care Reporting System (HCRS) database. The HCRS contains demographic, clinical, functional and resource-utilization information on clients receiving services through publicly funded home care programs in Canada (CIHI, 2012). Information in the HCRS is based on the RAI-HC, which is an assessment tool used to identify the strengths, preferences and needs of long-stay home care clients (Morris et al., 2009). In Ontario, the RAI-HC is the provincially mandated common assessment for clients expected to be on service for greater than 60 days and clients entering into long-term care facilities (CHCA, 2008). Ontario hospital administrative data were retrieved from the National Ambulatory Care Reporting System (NACRS) and the Discharge Abstract Database (DAD). Data on admissions to designated adult mental health beds in Ontario for patients that were assessed as a danger/threat to themselves were retrieved from the Ontario Mental Health Reporting System. These databases are managed by the Canadian Institute for Health Information (CIHI) and were made available through partnerships between CIHI and the University of Waterloo. These four datasets were linked together by CIHI using a common identifier. Personal identifiers and items that might identify an individual were removed to ensure confidentiality and anonymity of the data. Ethics approval was provided for the analyses of the de-identified data in the current study (certificate no ). Outcome measure The dependent variable was intentional self-harm (ISH) operationalized through ICD-10-CA codes in Ontario hospital administration records. Since 2004, Ontario hospitals and emergency departments use ICD-10-CA to classify intentional self-harm (i.e., X60 X84). The ICD-10-CA codes for intentional self-harm were not solely restricted to primary diagnostic types in order to allow for improved detection of self-harm events in a population with higher rates of multi-morbidity. Further, the inclusion of selected accidental poisoning codes in the definition of intentional self-harming

3 Self-harm among older adults in home care behavior (i.e., X40 X42 and X46 X47) was performed to improve under-estimation of intentional poisonings as performed in previous studies (i.e., Martens et al., 2004, and Fransoo et al., 2009). Incidents of ISH that were recorded in multiple databases (e.g., emergency department visit followed by hospital admission) used the most recent record date and were counted as one event. Instrumental Activities of Daily Living Scale The Instrumental Activities of Daily Living (IADL) Scale is based on seven items using performance-based ratings. IADL scores range from 0 to 42 and are used to determine IADLabilityanddeclineinhomecareclients(Jónsson et al., 2003; June et al., 2009; Cook et al., 2013). The IADL scale has been validated against the Lawton index with a corresponding coefficient of r=0.81 (Landi et al., 2000). Measures of risk and protective factors All independent variables were derived from the RAI-HC. The RAI-HC includes over 350 data elements measuring sociodemographic information, cognition, mood, behavior, functional status, disease diagnoses, medication, social support and service use. Five summary scales embedded in the RAI-HC were included in the analysis and are described below. These summary scales were collapsed for multivariable analysis using established cutoffs (Hirdes et al., 2011). Depression Rating Scale The Depression Rating Scale (DRS) is based on seven items and is used as a clinical indicator of depression. DRS scores range from 0 to 14. The DRS has been validated against the Hamilton Depression Rating Scale and the Cornell Scale for Depression with corresponding coefficients of r = 0.69 and r = 0.70, respectively (Burrows et al., 2000). Acceptable levels of internal consistency have been reported in samples from complex continuing care hospitals (α = 0.74), nursing home residents (α = 0.85), and among inpatient mental health patients (α = 0.74) (Hirdes et al., 2002; Koehler et al., 2005; Martin et al., 2008). Activities of Daily Living Hierarchy Scale The Activities of Daily Living (ADL) Hierarchy Scale measures four items of ADL performance and categorizes abilities into stages at which they can no longer be performed. Difficulty with performance is scored from 0 (independent) to 4 (total dependence). The ADL Hierarchy reliably assesses change in ADL impairment over time (Morris et al., 1999) and has shown acceptable validation against the Barthel index (r = 0.74) (Landi et al., 2000). Cognitive Performance Scale The Cognitive Performance Scale (CPS) is based on five items and rates the cognitive status of home care clients. CPS scores range from 0 (intact cognition) to 6 (very severe impairment). It has been validated against the Mini Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and the Test for Severe Impairment with coefficients ranging between r = 0.69 and r = 0.81 (Morris et al., 1994; Landi et al., 2000; Jones et al., 2010). Changes in Health, End-stage disease and Signs and Symptoms The Changes in Health, End-stage disease and Signs and Symptoms (CHESS) Scale is a measure of health stability (versus instability) that has been used as a measure of frailty to predict mortality. Scores on the CHESS range from 0 (no instability in health) to 5 (highly unstable health) (Hirdes et al., 2003). CHESS was a strong predictor of survival in long-term care homes. It has been validated for use in home care settings, although correlations between CHESS and the Edmonton Frailty Scale and the Frailty Index were weak (r = 0.39 and r = 0.35) (Armstrong et al., 2010). The Canadian Coalition for Seniors Mental Health identifies several protective factors for suicide in later life (CCSMH, 2006). These were identified in the RAI- HC and included (i) positive social relationships; (ii) low or no indicators of isolation; (iii) having a caregiver with (iv) no indicators of caregiver burden; and (v) having optimism that functional ability will improve. Analysis Frequency tables were generated to examine the distribution of all variables prior to recoding. Demographic variables were recoded into categories where appropriate (e.g., age groupings) or dichotomized for multivariable analysis. Caregiver distress was derived from

4 E. Neufeld et al. endorsing two items in the RAI-HC: (i) caregiver is unable to continue in caring activities and (ii) primary caregiver expresses feelings of distress, anger or depression. Alcohol use and dependence was derived from endorsing two items: (i) the person s alcohol consumption concerns others and (ii) needing a drink first thing in the morning. The receipt of any type of psychotropic medication was combined into a single yes/no psychotropic medication variable. Records of intentional self-harm that were retrieved from the hospital datasets were coded into event (=1) versus no event (=0) to compare between older adults with a hospital record for intentional self-harm to the general home care population. Spearman s rank-order correlations were used to examine variable relationships and check for multicollinearity using a threshold of 0.7 (Tabachnick and Fidell, 2001). Widowhood and married marital status were highly correlated (r = 0.78, p < ); therefore, predictive models were examined that tested these variables separately (i.e., model 1 tested married marital status and model 2 tested widowed marital status). As an initial test, the forward selection method of model building was used to observe the unique contribution of independent variables that have either a risk or protective relationship to intentional self-harm as supported in research literature. All independent variables were then manually entered into the regression model to test combinations of variables the forward selection method may have missed because of order-of-entry effects and to test interaction effects between gender and risk and protective variables. Receiver Operating Characteristic (ROC) curves and Hosmer and Lemeshow s goodness-of-fit tests were used to determine goodness of fit. All analyses were performed using SAS software. Results Among the 222,149 older adults in home care, there were 2077 hospital recordings of ISH. The prevalence of ISH was 9.3 cases per 1000 older adults. The average time between home care assessments and hospital recordings of ISH was approximately 2 years (+483.6). Rates of ISH among men was 1.1 (CI = ) and 1.0 (CI = ) among women. The ISH sample was younger (mean age 75.3 years, SD 8.7, vs years, SD 8.5, respectively) with more cases of divorced/separated marital status than the general home care population. Additional descriptive information is reported in Table 1. Table 2 shows the results for the two final logistic regression models predicting ISH while adjusting for relevant covariates. The variables in model 1 with the greatest odds of ISH were younger age (60 74 years; OR = 3.14, 95% CI: ), any psychiatric diagnosis (OR = 2.29, 95% CI: ), use of psychotropic medication (OR = 1.94, 95% CI: ), alcohol use and dependence (OR = 1.69, 95% CI: ) and higher scores on depression measures (OR = 1.58, 95% CI: ). Positive social relationships were protective as the odds of ISH among older adults that reported positive relationships were approximately 25% lower than older adults reporting no positive social relationships (95% CI: ). Unexpectedly, impairments in activities of daily living also lowered the odds of ISH (OR = 0.51, 95% CI: ). A diagnosis of dementia was not an independent risk factor until entered in an interaction. The main effects in model 2 are nearly identical to those presented in model 1. Widowed marital status was protective in model 2 as the odds of ISH were approximately 30% lower than older adults that were not widowed (95% CI: ). A significant quadratic effect for scores on the CPS was present in both models. This effect showed that the odds of ISH increased up to scores of 4 and 5 on the CPS and then decreased. At its highest point, the odds of ISH were 1.86 times higher for older adults with CPS scores of 5, indicating moderate to severe cognitive impairment (Figure 1). Significant interaction effects were found in both models between sex, married marital status, widowed marital status, dementia diagnosis and positive social relationships. Interestingly, the effects were mainly significant for older men than older women (Figure 2). For example, the odds of ISH among older men that were not married were 1.27 times higher (95% CI: ) than older men that were married. There was no effect for women between married marital status and ISH. Among older men that were widowed, the odds of ISH were 1.30 times higher (95% CI: ) than older men that were not widowed. In this case, the odds of ISH were 21% lower (95% CI: ) among older widowed women. When examining dementia, the odds of ISH among older men were approximately two times higher (95% CI: ) than older men without a dementia diagnosis. There was no effect for women between dementia and ISH. Lastly, there was 23% lower odds of ISH among older men reporting positive social relationships (95% CI: ); however, the effect was not significant for older women.

5 Self-harm among older adults in home care Table 1 Descriptive characteristics comparing older home care clients with a record of intentional self-harm (ISH) to the general home care population, Ontario, (N = 222,149) ISH sample (N = 2077) Non-ISH sample (N = 202,720) %(n) % (n) Sociodemographic Age*** years 45.4 (943) 24.2 (48,995) years 38.5 (800) 42.0 (85,156) 85+ years 16.1 (344) 33.8 (68,569) Sex*** Female 59.8 (1,242) 64.2 (130,091) Male 40.2 (835) 35.8 (72,629) Marital status*** Married 42.7 (874) 41.1 (83,236) Never married 6.5 (135) 4.6 (9,413) Divorced 15.6 (323) 6.8 (13,647) Widowed 34.5 (716) 46.6 (94,463) Risk factors Any psychiatric diagnosis*** 34.4 (715) 11.1 (22,492) Alcohol use and dependence*** 4.0 (83) 1.3 (2,687) Daily pain*** 56.7 (1,178) 51.6 (104,623) Psychotropic medications*** 66.6 (1,383) 39.9 (80,890) Alzheimer s disease/dementia*** 21.5 (429) 16.1 (31,253) DRS score *** (932) 61.3 (130,432) (590) 22.3 (45,214) (555) 13.4 (27,074) ADL Hierarchy Scale score *** (1,509) 69.1 (140,023) (432) 21.9 (44,369) (136) 9.0 (18,328) IADL Capacity Scale score ** (92) 5.5 (22,205) (575) 24.4 (49,389) (1,410) 70.2 (142,226) CPS score *** (1,191) 65.8 (133,339) (795) 30.7 (62,324) (91) 3.5 (7,057) CHESS Scale score *** (633) 29.6 (59,917) (1,220) 58.1 (117,784) (224) 12.3 (25,019) Protective factors Positive social relationships*** 77.2 (1,603) 86.5 (175,283) Not isolated* 50.1 (1,040) 52.8 (106,956) Caregiver available*** 50.0 (1,039) 52.0 (105,390) Caregiver distress*** 20.9 (435) 16.2 (32,854) Optimism to improve 21.8 (452) 22.5 (45,596) DRS, Depression Rating Scale; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; CPS, Cognitive Performance Scale; CHESS, Changes in Health, End-stage disease and Signs and Symptoms. *p < **p < ***p < Divorced includes separated marital status. 0 = no indicators of depression; 1 2 = some indicators of depression; 3+ = indicators of probable depression. 0 = no impairment; 1 2 = some functional impairment; 3+ = severe functional impairment. 0 = no difficulty; 1 2 = some difficulty; 3+ = great difficulty. 0 1 = cognitively intact to borderline cognitive impairment; 2 3 = mild to moderate cognitive impairment; 4+ = cognitively impaired. 0 = no health instability; 1 2 = some health instability; 3+ = moderate to high health instability. Discussion This study utilized provincial data from the home care sector in Ontario to identify evidence for risk and protective factors associated with intentional self-harm among older adults. The findings point to tangible areas for preventative assessment by frontline home care professionals. The prevalence of older home care clients with a hospital record of ISH was low (9.3 cases per 1000 home care clients), although this is consistent with previous Canadian data (CIHI, 2004). In the current study, older men were more likely to have a record of ISH than older women, which is also consistent with previous literature (CCSMH, 2006; CDC, 2007). Multivariable analyses demonstrated that the odds of ISH increased for older adults with a

6 E. Neufeld et al. Table 2 Multiple logistic regression models predicting intentional self-harm among older home care clients, Ontario, (N = 222,149) Model 1 Model 2 Independent variables Adjusted odds ratio (95% CI) Adjusted odds ratio (95% CI) Sociodemographic Aged years 3.14 ( )*** 2.95 ( )*** Aged years 1.67 ( )*** 1.61 ( )*** Sex (male) 1.27 ( )** 1.13 ( ) Married 1.00 ( ) Widowed 0.79 ( )** Risk factors Any psychiatric diagnosis 2.29 ( )*** 2.31 ( )*** Alcohol use and dependence 1.69 ( )*** 1.76 ( )*** Psychotropic medication 1.94 ( )*** 1.94 ( )*** DRS score ( )*** 1.30 ( )*** DRS score ( )*** 1.57 ( )*** ADL Hierarchy Scale score ( )*** 0.71 ( )*** ADL Hierarchy Scale score ( )*** 0.50 ( )*** Daily pain 1.14 ( )* 1.14 ( )* Cognitive Performance Scale 1.32 ( )*** 1.32 ( )*** Cognitive Performance Scale ( )* 0.97 ( )* Dementia diagnosis 0.91 ( ) 0.92 ( ) Protective factors Positive social relationships 0.77 ( )*** 0.85 ( )* Interaction effects Sex * married 0.75 ( )* Sex * dementia diagnosis 1.88 ( )*** 1.75 ( )*** Sex * widowed 1.46 ( )** Sex * positive social relationships 0.79 ( )* Individual variables were tested using Wald chi-square tests. AUC, area under the curve; CI, confidence interval; Df, degrees of freedom; DRS, Depression Rating Scale; ADL, Activities of Daily Living. DRS: 0 = no indicators of depression; 1 2 = some indicators of depression; 3+ = indicators of probable depression. ADL: 0 = no impairment; 1 2 = some functional impairment; 3+ = severe functional impairment. *p < **p < ***p < Model 1 tests married marital status; AUC = 0.74; goodness of fit χ 2 = 14.56, df =8,p = Model 2 tests widowed marital status; AUC = 0.74; goodness of fit χ 2 = 9.11, df =8,p = Versus age = 85+. Versus not married/divorced/widowed. Versus married/not married/divorced. Versus DRS = 0. Versus ADL = 0. psychiatric diagnosis, indicators of alcohol use and dependence, psychotropic medication and indicators of probable depression. Risk was considered to be highest for the young old in the years age group and among males. Positive social relationships emerged as protective against ISH after adjusting for risk factors. The findings from the predictive models are consistent with previous research on psychiatric illness and depression associated with intentional self-harm (Hansen et al., 2003; Oquendo et al., 2004; Qin, 2011). Although the majority of people with a psychiatric diagnosis neither attempt nor die by suicide, the heightened potential risk warrants that adequate assessment and continued monitoring of suicide ideation is an approach to prevention among this group. Interventions aimed at early treatment for depression, alcohol use and pain are especially critical. An unexpected finding was the lowered risk of ISH with increased impairment in ADL ability. This might suggest that functional impairment in ADL is protective against suicide-related behavior. Previous literature suggests the opposite: that impairments in ADL and anticipatory fear of losing ADL ability increase risk for suicide-related behavior (Leenaars, 1992; Bauer et al., 1997; Lester et al., 2004). Therefore, it appears that the relationship between ISH and physical ability or disability is more complex. As the current study modeled suicide behavior, as opposed to suicide ideation, it is

7 Self-harm among older adults in home care Adjusted Odds Ratios (95% CI) Cognitive Performance Scale Figure 1 Adjusted odds ratios and 95% confidence intervals (CI) of the Cognitive Performance Scale scores in the prediction of intentional selfharm among older home care clients, Ontario, (N = 222,149). Figure 2 Interaction effects between sex and risk and protective factors in the prediction of intentional self-harm among older home care clients, Ontario, (N = 222,149). *p < 0.05, **p < 0.001, and ***p < CI, confidence interval. possible that impairments in ADL work differently in terms of risk between thinking about suicide versus acting on suicidal thoughts. Of the five protective factors assessed, positive social relationships emerged as significant both independently and in the interaction with sex. This suggests that positive social relationships mitigate suiciderelated behavior among older adults and is protective in the face of risk factors that increase odds of ISH (Rowe et al., 2006). Increasing opportunities for positive social relationships among home care clients is warranted and may be an under-recognized suicide prevention strategy. Significant interactions in the predictive models demonstrated a gendered effect for certain risk and protective factors. The odds of ISH were higher for older men that were not married, although this effect is not present for older women. A similar pattern of increased odds of ISH was observed for older men that were widowed. Older women that were widowed, however, had lower odds of self-harm. These results support interesting patterns related to theories of marriage and widowhood in later life and associated risk and protective effects of each (Van Grootheest et al., 1999; Kposowa, 2000; Martin-Matthews, 2011). Positive social relationships were an independent protective factor to ISH in the current study, which has been supported in previous research (Clarke et al., 2004; Heisel, 2006). However, the interaction between positive social relationships and sex showed that this protective effect was only statistically significant for older men. A further examination of these gendered effects and protective factors is suggested. A curvilinear pattern of the scores on the CPS among older adults with ISH was an indication to test a quadratic effect in the multiple regression model. The results of the test show increasing odds of ISH as cognitive performance becomes more impaired. However, the increased odds reach a peak on the CPS, at which point risk declines when cognitive performance is most impaired. Although a controversial topic, these findings suggest that cognitive impairment, to a certain degree, is a significant risk factor for suicide-related behavior among older adults even when controlling for other risk factors. Margo and Finkel (1990) hypothesized that the loss of functioning that accompanies cognitive decline is often ignored as a danger signal for suicide. Of note, the dementia diagnosis was not an independent risk factor until entered as an interaction with sex. Results of the current study demonstrated a higher risk of suicide-related behavior for older men with a dementia diagnosis than older women in the same group. While it may be difficult to discern the intent of the self-harming behavior, these results indicate that the sub-population of home care clients with cognitive impairments should not be assumed incapable of suicide-related behavior, particularly older men. The findings from this research should be interpreted with the following limitations in mind. The dataset was drawn using a cross-sectional sample of older home care clients in Ontario between 2007 and 2010 with a subsequent hospital administration record for intentional self-harm. The results are applicable to older adults receiving home care in Ontario

8 E. Neufeld et al. but may not represent all older adults in the province with the experience of self-harm. Records of intentional self-harm in the DAD and NACRS data sets were drawn from the diagnostic code fields, of which there are up to 25 code fields in the DAD. This was purposefully done to draw a larger sample size of older adults with the experience of intentional self-harm. This approach yielded a large sample of 2077 older adults with a record of intentional self-harm within a representative sample of 222,149 older adults. This is a strength of the current study as it is very difficult to recruit a sample this large for the purpose of studying a sensitive and stigma-laden topic such as suicide. However, the use of administrative health datasets such as the NACRS has been criticized as ICD- 10-CA codes are likely not 100% accurate. Nonetheless, examining data from home care and hospital administrative records ensured little selection bias in the older adult sample, as the data collection tools are provincially mandated in Ontario. Conclusions The results of this study describe a sub-population of older adults in the home care sector who may be at risk of suicide-related behavior. By examining predictors associated with ISH versus predictors of suicide ideation, it is possible to conclude more accurately which characteristics increase risk and which characteristics are protective against acts of intentional self-harm. The multi-dimensional profile that was created of older adults at higher risk in the home care population is an important approach to developing targeted suicide prevention initiatives in the home care sector. This study of older adults at risk of ISH in the province of Ontario adds to the existing evidence of suicide risk in later life and the importance of concurrently examining protective factors that might mitigate suicide risk. The results showed several areas that should be assessed by home care professionals to reduce risk in the older home care client. Of notable interest were the risk and protective factors that differed by sex. As demand for home care in Canada is expected to increase, these findings may inform home care professionals appraisal and approach to suicide prevention among community-residing older adults. Conflict of interest None declared. Key points Suicide-related behavior among older home care clients is an understudied area of research. Characteristics that increase risk or protect/ mitigate suicide-related behavior among older home care clients differ by gender. Study findings support evidence of an increased risk for suicide-related behavior among older home care clients with cognitive impairment. Acknowledgements This research was supported by funding from the Toronto Central Community Care Access Center. This study is part of the National Population Health Study of Neurological Conditions. The membership of Neurological Health Charities Canada and the Public Health Agency of Canada are acknowledged for their contribution to the success of this initiative. Funding for the study was provided by the Public Health Agency of Canada, project no / The opinions expressed in this document are those of the primary author and do not necessarily reflect the official views of the Public Health Agency of Canada. JPH s participation is supported through the Ontario Home Care Research and Knowledge Exchange Chair funded by the Ontario Ministry of Health and Long Term Care. References Armstrong JJ, Stolee P, Hirdes JP, Poss J Examining three frailty conceptualizations in their ability to predict negative outcomes for home-care clients. Age Ageing 39: Awata S, Seki T, Koizumi Y, et al Factors associated with suicidal ideation in an elderly urban Japanese population: a community-based cross-sectional study. Psych Clin Neurosci 59: Bauer MN, Leenaars A, Berman AL, et al Late adulthood suicide: a life-span analysis of suicide notes. Arch Suicide Res 3: Burrows AB, Morris JN, Simon SE, et al Development of a minimum data setbased depression rating scale for use in nursing homes. Age Ageing 29: Canadian Coalition for Seniors Mental Health (CCSMH) The assessment and treatment of depression. National Guidelines for Seniors Mental Health. Baycrest: Toronto, ON. Canadian Home Care Association Portraits of Home Care in Canada. Canadian Home Care Association: Ottawa, ON. Canadian Institute for Health Information (CIHI) National trauma registry analytic bulletin: Hospitalizations due to suicide attempts and self-inflicted injury in Canada, CIHI: Toronto (ON). Canadian Institute for Health Information Home Care Reporting System (HRCS). Available at types+of+care/ community+care/home+care/ hcrs_metadata. Centers for Disease Control and Prevention [CDC] Nonfatal self-inflicted injuries among adults aged 65 years United States, MMWR CDC Surveill Summ 56: Clarke DE, Colantonio A, Heslegrave R, et al Holocaust experience and suicidal ideation in high-risk older adults. Am J Geriatr Psychiatry 12:

9 Self-harm among older adults in home care Conwell Y, Rotenberg M, Caine E Completed suicide at age 50 and over. JAm Geriatr Soc 38: Cook RJ, Berg K, Lee KA, et al Rehabilitation in home care is associated with functional improvement and preferred discharge. Arch Phys Med Rehab 2. doi: /j.apmr Fransoo R, Martens P, Burland E, et al Manitoba RHA Indicators Atlas Manitoba Centre for Health Policy: Winnipeg, MB. Hansen PEB, Wang AG, Stage KB, Kragh-Sorensen P Comorbid personality disorder predicts suicide after major depression: a 10-year follow up. Acta Psychiatr Scand 107: Heisel MJ Suicide and its prevention among older adults. Can J Psychiatr 51: Heisel MJ, Flett GL Psychological resilience to suicide ideation among older adults. Clin Gerontol 31: Hirdes JP, Smith TF, Rabinowitz T, et al The Resident Assessment Instrument- Mental Health (RAI-MH): inter-rater reliability and convergent validity. J Behav Health Serv Res 29: Hirdes JP, Frijters DH, Teare GF The MDS-CHESS Scale: a new measure to predict mortality in institutionalized older people. J Am Geriatr Soc 51: Hirdes JP, Ljunggren G, Morris JN, et al Reliability of the interrai suite of assessment instruments: a 12-country study of an integrated health information system. BMC Health Serv Res 8: 277. Hirdes JP, Mitchell L, Maxwell C, White N Beyond the iron lungs of gerontology : using evidence to shape the future of nursing homes in Canada. Can J Aging 30: Jones K, Perlman CM, Hirdes JP, Scott T Screening cognitive performance with the Resident Assessment Instrument for Mental Health Cognitive Performance Scale. Can J Psychiatr 55: Jónsson PV, Guðmundsdóttir H, Friðbjoernsdóttir F, et al Health care needs and quality of life of elderly in home care in Reykjavik, Laeknabladid 89: June KJ, Lee JY, Yoon JL Effects of case management using Resident Assessment Instrument-Home Care (RAI-HC) in home health services for older people. J Korean Acad Nurs 39: Koehler M, Rabinowitz T, Hirdes JP, et al Measuring depression in nursing home residents with the MDS and GDS: an observational psychometric study. BMC Geriatr 5. doi: / Kposowa AJ Marital status and suicide in the National Longitudinal Mortality Study. J Epidemiol Community Health 54: Landi F, Tua E, Onder G, et al Minimum data set for home care: a valid instrument to assess frail older people living in the community. Med Care 38: Leenaars A Suicide notes of the older adult. Suicide Life-Threat Behav 22: Lester, D, Wood, P, Williams, C, Haines, J Motives for suicide- a study of Australian suicide notes. Crisis 25: Li LW, Conwell Y Pain and self-injury ideation in elderly men and women receiving home care. J Am Geriatr Soc 58: Margo GM, Finkel JA Early dementia as a risk factor for suicide. Hosp Comm Psychiatr 6: Martens PJ, Fransoo R, McKeen N, et al Patterns of Regional Mental Illness Disorder Diagnoses and Service Use in Manitoba: A Population-Based Study. Manitoba Centre for Health Policy: Winnipeg, MB. Martin L, Poss JW, Hirdes JP, et al Predictors of a new depression diagnosis among older adults admitted to complex continuing care: implications for the depression rating scale (DRS). Age Ageing 37: Martin-Matthews A Revisiting widowhood in later life: changes in patterns and profiles, advances in research and understanding. Can J Aging 30: Morris JN, Fries BE, Mehr DR, et al MDS Cognitive Performance Scale. J Gerontol 49: M174 M182. Morris JN, Fries BE, Morris SA Scaling ADLs within the MDS. J Gerontol A Biol Sci Med Sci 54: M546 M553. Morris JN, Fries BE, Bernabei R, et al interrai Home Care (HC) Assessment Form and User s Manual. Version 9.1. interrai: Washington, DC. O Connell H, Chin A, Cunningham C, Lawlor BA Recent developments: suicide in older people. BMJ 329: Oquendo MA, Galfalvy H, Russo R, et al Prospective study of clinical predictors of suicidal acts after a major depressive episode in patients with major depressive disorder or bipolar disorder. Am J Psychiatr 161: Poss JW, Jutan NM, Hirdes JP, et al A review of evidence on the reliability and validity of minimum data set data. Healthc Manage Forum 21: Qin P The impact of psychiatric illness on suicide: differences by diagnosis of disorders and by sex and age of subjects. J Psychiatr Res 45: Rowe JL, Conwell Y, Schulberg HG, Bruce M Social support and suicidal ideation in older adults using home healthcare services. Am J Geriatr Psychiatry 14: Statistics Canada Suicides and Suicide Rate, by Sex and by Age Group. CANSIM, Table and Catalogue no. 84-F0209-X. Statistics Canada: Ottawa, ON. Tabachnick B, Fidell L Using Multivariate Statistics, 4 th Edition. Allyn & Brown: Toronto, ON. Tadros G, Salib E Elderly suicide in primary care. Int J Geriatr Psychiatry 22: Van Grootheest DS, Beekman ATF, Broese van Groenou MI, Deeg DJH Sex differences in depression after widowhood. Do men suffer more? Soc Psychiatry Psychiatr Epidemiol 34: Waern M Physical illness and suicide in the elderly. Geriatrics Aging 6: World Health Organization [WHO] Distribution of suicide rates (per 100,000), by gender and age, Available from: