In order to address suicide prevention and mental health care in older people living in long-term care facilities, it is essential to deepen the understanding of suicidal ideation and its manifestation in this context. However, most studies have addressed the phenomenon from a predictive variable’s perspective, but very few with a person-oriented approach like Latent Profile Analysis (LPA). As Lee (2023) points out, since the causal relationship between suicidal ideation and suicide has not yet been fully established, detailed analyses are needed to identify target groups and thus design more effective prevention and early intervention strategies. Only one study has used latent profile analysis to identify health profiles in older people living in nursing homes (Yuan et al., 2022) but they only take into account self-reported suicidal ideation at admission, not during the stay.

In this sense, the present study applies a LPA to characterise different clinical profiles of suicidal ideation in this population. LPA is a person-centered approach that identifies unobserved subgroups based on individuals’ response patterns across multiple observed variables, thus focusing on configurations of characteristics rather than on isolated variables. As Ferguson et al. (2020) point out, LPA has three characteristics: individual differences are important and must be taken into account in the explanation of a phenomenon; these differences occur within a logic that allows grouping into patterns or profiles; and finally, these profiles are significant and occur in different individuals. Therefore, the aim of this article is to explore the existence of mental health profiles in nursing homes so that appropriate detection and prevention programmes can be designed.

A total of 231 older adults aged 60–97 years (M = 78.99, SD = 8.83) participated in the study, including 95 men (41.1%) and 136 women (58.9%). Participants resided in facilities that provide long-term care and assistance to older adults, generally in a situation of physical or cognitive dependence or in a context of social vulnerability. From a pull of 30 residential centres, 9 were randomly selected, and from these, those who met the study’s inclusion criteria were invited to apply. Approximately 25–30 residents took part from each participating centre.

The inclusion criteria for participation in the study were: Being 60 years of age or older and having preserved cognitive abilities or presenting only mild cognitive impairment (MCI), that is, obtaining scores above 19 on the Mini-Mental State Examination (MMSE), in its adapted Spanish version (Mini-Mental Cognitive, MEC) (Lobo et al., 1999). This criterion aimed to ensure that all participants had sufficient cognitive capacity to provide informed consent and to reliably complete the self-report instruments. The mean MMSE score was 29.17 (SD = 3.99), with participants with MCI representing 16.4% of the total sample. Participants with MCI were included because they were able to respond to all questionnaires during individualized interviews conducted by trained staff. Including this group ensures that the sample reflects the institutionalized population and allows for the capture of emotional and psychological aspects related to these participants.

Exclusion criteria: Having suffered an acute illness or hospitalisation or unstable chronic illness in the last month or having a serious illness; or having a diagnosis of mental illness in an acute phase or with significant psychiatric decompensation at the time of assessment, which could hinder the ability to provide informed consent or respond reliably. This evaluation was carried out by the healthcare team at each care home, including the centre’s physician and psychologist.

Scores on the Mini-Mental State Examination (MMSE) scales, and information for inclusion and exclusion criteria were extracted through the residential management software used in Spanish residential centres. This database systematically collects and stores the results of the assessments carried out by the professionals of the centres every six months.

After approval of the research project by the management of the residential centres, older people living in the selected nursing homes were informed of the research objectives. Those who met the inclusion criteria and agreed to take part in the study signed an informed consent form before the questionnaires were completed. Individualised, semi-structured interviews were conducted in a room in the residential centre, ensuring a quiet and distraction-free environment. The duration was approximately 60–80 min per person. The assessors were three psychologists previously trained by the study authors. The data were then entered anonymously into an SPSS database for analysis.

Since Likert-type response scales can present difficulties for older people, especially those with mild cognitive impairment or information processing problems, a colour-coded support system was implemented. Each response option within the scales was visually represented by a progressive colour palette, assigning a different colour to each response category: from cool tones for lower intensity responses to warm tones for higher intensity responses. This was intended to facilitate the understanding of the progression of responses, optimise the accessibility of the scale for people with difficulties in verbally differentiating the options and reduce ambiguity in the interpretation of responses.

This study is part of a research project that was approved by the Ethics Committee of the University of Valencia (2024-PSILOG-3281474). All procedures performed were in accordance with the ethical principles set out in the Declaration of Helsinki. Written informed consent was obtained from all participants, who were previously informed about the objectives of the study, the confidentiality of the data and their right to withdraw at any time.

To carry out the LPA, the steps proposed by Ferguson et al. (2020) were followed. First, data cleaning and checking for missing data were performed. After this, 5 iterative models were tested, following the recommendations of Tein et al. (2013) using Mixture type and Robust Maximum Likelihood Estimator (MLR). Then, each model is compared against the previous model or models to make a decision regarding the number of latent profiles in the data using the common fit indices in LPA: Akaike’s Information Criterion (AIC), Bayesian Information Criterion (BIC), Sample-Adjusted BIC (SABIC), Lo–Mendell Ruben (LMR), and Bootstrap Likelihood Ratio Test (BLRT). Regarding BIC, SABIC, and AIC, lower values indicate a better fit; however, the lowest value is relative (Masyn, 2013). Therefore, attention should be paid to the magnitude of the difference. The LMR test and the Bootstrap Likelihood Ratio Test (BLRT) compare the current model to a model with k-1 profiles. Entropy was also considered, where values of 0.80 or higher provide evidence that the classification of individual profiles in the model occurs with minimal uncertainty (Celeux and Soromenho, 1996; Tein et al., 2013), and the percentage of the sample included in the smallest profile, which is recommended to be no lower than 5% (Ferguson et al., 2020).

After selecting the model that best fits the data, it was interpreted through the mean scores on each variable (depression, hopelessness, burden, self-efficacy, purpose of life, and resilience) for each profile. Finally, to examine whether the resulting profiles differed in suicidal ideation after adjusting for potential confounding variables, an Analysis of Covariance (ANCOVA) was conducted. Suicidal ideation was entered as the dependent variable, profile membership as the fixed factor, and age, sex and perceived health as covariates. Adjusted marginal means were calculated and pairwise comparisons were conducted using Bonferroni correction. A p-value < 0.05 indicated statistical significance. All analyses were conducted using Mplus 8.11 (Muthén and Muthén, 2017) and IBM SPSS 28.0.

To examine whether suicidal ideation differed across the four latent profiles after adjusting for potential confounding variables, an Analysis of Covariance (ANCOVA) was conducted with profile membership as the fixed factor and age, sex, and perceived health as covariates. The overall model was significant, F(9,221) = 19.64, p < 0.001, explaining a substantial proportion of the variance in suicidal ideation (partial η² = 0.444). Among the covariates, perceived health was a significant predictor, indicating that poorer perceived health was associated with higher suicidal ideation, F(1,221) = 5.23, p = 0.023. In contrast, age was not a significant predictor, F(1,221) = 0.07, p = 0.795. Sex showed a significant main effect, with women reporting higher suicidal ideation than men, F(1,221) = 20.33, p < 0.001. Importantly, although women and individuals with poorer perceived health showed higher overall suicidal ideation, the pattern of differences between profiles remained consistent across these variables, with the High Risk group exhibiting the highest levels and the Optimal Mental Health group the lowest.

Crucially, the effect of latent profiles remained highly significant after covariate adjustment, F(3,221) = 31.02, p < 0.001, partial η² = 0.296. Adjusted marginal means indicated that the High Risk profile showed the highest level of suicidal ideation (M = 7.09), followed by the Burdensomeness (M = 3.16) and Weakened Personal Strengths profiles (M = 3.02). The Optimal Mental Health profile showed the lowest levels (M = 0.88). Post-hoc pairwise comparisons with Bonferroni correction demonstrated that the High Risk profile differed significantly from all other profiles (p < 0.001), and that the Optimal Mental Health profile presented significantly lower suicidal ideation compared to Burdensomeness (p = 0.001) and Weakened Personal Strengths (p < 0.001). No significant differences were found between the Burdensomeness and Weakened Personal Strengths profiles.

High Risk profile is characterised by high levels of depression and hopelessness, as well as high perceived burden. Depression is one of the risk factors in older adults that has received the most scientific evidence (Beghi et al., 2021; Hernandez et al., 2021; Ki et al., 2024); Depression often coexists with other emotional and cognitive difficulties, such as feelings of uselessness or a lack of motivation. It has also been linked to an increased perception of burden, particularly among older adults in nursing homes who have functional limitations (Zhang et al., 2021). In residential settings, depression can be overlooked or mistaken for normal ageing, despite its strong predictive value for suicidal thoughts (Redondo et al., 2025). Several studies have confirmed the direct association between hopelessness and suicidal ideation in older adults, even when controlling for depression and functional impairment (Cukrowicz et al., 2011; Ribeiro et al., 2018). In this sense, hopelessness intensifies the subjective experience of suffering and, when combined with the perception of burden, forms a high-risk psychological profile.

In addition, individuals in this group show low scores on self-efficacy, resilience and purpose in life. This combination of factors indicates a marked vulnerability (Beach et al., 2021; Olatunji et al., 2020; Liu et al., 2024), which requires immediate and specialised interventions. Strategies targeting this profile should include immediate psychological care, combined with suicide prevention programmes tailored to older people, interventions focused on strengthening purpose in life and resilience, enhancing self-efficacy, and actions to improve perceived social support and reduce emotional burden (Okolie et al., 2017; Heisel and Flett, 2022; Treichler et al., 2020; Ki et al., 2024).

Within the moderate risk profiles, two subgroups are distinguished. Burdensomeness profile, which is characterised by the presence of low depressive and hopelessness symptoms together with a notable perception of burden, although these individuals retain certain levels of resilience, self-efficacy and purpose in life. This group would benefit from preventive interventions that enhance social support, encourage participation in group activities and strengthen the perception of belonging and usefulness since, as indicated by Van Orden et al. (2012), reducing their perceived burden could be an effective way to improve their well-being and quality of life. This perceived burden, which can include physical, social and emotional components (McPherson et al., 2007), is more prevalent in those living in residential care, where the need for help with daily activities is often greater (Conwell et al., 2011; Zhang et al., 2021).

On the other hand, the Weakened Personal Strengths profile shows low levels of depression, hopelessness and perceived burden, but also low purpose, resilience and self-efficacy was observed. Low purpose increases vulnerability to suicidal ideation and low resilience and self-efficacy makes them less able to cope with complexities. For this group it is essential to implement interventions that work on purpose in life, as they could be innovative ways to prevent suicidal behaviour in older adults (Heisel et al., 2020; Ostafin and Proulx, 2020).

Purpose in life is a factor that, not only prevents suicidal ideation (Beach et al., 2021; De Brigard, 2021), but also reduces stress, depression and anxiety. It is a method of coping with human problems that cause suffering and plays a mediating role in health according to longitudinal studies (Alimujiang et al., 2019; Soucase et al., 2023). In this line, the development of psychological interventions that can enhance modifiable intrapersonal factors such as resilience (Treichler et al., 2020) or coping strategies (Gysin-Maillart et al., 2020) could be innovative ways to prevent suicidal behaviour in older adults (Ki et al., 2024).

Optimal Mental Health profile, represents well-functioning individuals. Their situation highlights the importance of promoting social participation, the establishment of support networks and the consolidation of effective coping mechanisms in the older population. Health, understood as a fundamental human right, must be approached from a biopsychosocial model that recognises the dynamic interaction between biological, psychological and social factors in people’s well-being (WHO, 2024). This conception is aligned with the notion of active and healthy ageing promoted by the WHO, which seeks not only the absence of disease, but full participation in social, emotional and community life (WHO, 2020). In this sense, this profile reflects not only an absence of risk factors, but also the presence of protective factors, such as self-efficacy, resilience and purpose in life, that allow older people to face the challenges of ageing from a position of autonomy and dignity. This result challenges ageist conceptions that associate institutionalisation with situations of dependency and deterioration (Xu et al., 2022), recalling that not all older people living in residential care homes feel a burden, nor do they live their old age from emotional discomfort. This profile therefore posits a form of holistic wellbeing that makes sense of the idea that ageing with health is also living with purpose (Seligman, 2011; Alimujiang et al., 2019).

This work not only allows us to identify clinically differentiated profiles that can guide more tailored preventive strategies, but also provides evidence on the discriminative capacity of these profiles in key variables such as suicidal ideation. The High Risk profile showed significantly higher scores than the other profiles in terms of risk variables, thus validating its usefulness as a clinically prioritised group requiring immediate action (tertiary prevention). Conversely, the Optimal Mental Health profile exhibited the lowest levels of suicidal ideation, necessitating universal prevention measures and well-being promotion (primary prevention). This profile significantly diverged from the intermediate Burdensomeness and Weakened Personal Strengths profiles, for which secondary and selective prevention measures (individual or group) would be necessary.

This approach does not replace clinical assessments or crisis protocols; rather, it complements them by offering a more understandable and applicable view in contexts where specialised training is limited, such as in many nursing homes (Redondo et al., 2025). Unlike traditional screening, which focuses on current ideation, profiles enable the identification of psychological patterns that predispose individuals to varying degrees of vulnerability. For instance, an individual may not express active suicidal thoughts, yet exhibit a severely impaired psychological profile characterised by low resilience, low purpose and high hopelessness, representing a ‘silent’ risk that conventional questionnaires might overlook.

Although the psychological profiles retain their explanatory power beyond sociodemographic variables, the ANCOVA analyses showed that both sex and perceived health influence overall levels of suicidal ideation. Consistent with recent reviews, evidence indicates that men exhibit higher rates of suicide mortality and lower emotional expression, whereas women tend to report greater suicidal ideation and affective symptomatology (Shelef, 2021; Connors et al., 2025). Additionally, although research specifically addressing self-rated health is limited, multiple studies concur that physical and cognitive decline increases vulnerability to suicidal thoughts and behaviours (Chauliac et al., 2020; Beghi et al., 2021).

From a health psychology perspective, it is crucial to understand the psychological profiles of older people living in nursing homes in order to develop effective, person-centred interventions that promote mental health and prevent suicidal thoughts. By identifying distinct patterns of risk and protective factors, this study provides a useful framework for designing preventive strategies that transcend clinical diagnosis and integrate psychosocial dimensions such as resilience, self-efficacy and purpose in life (Ki et al., 2024). These findings underline the need to address mental health in a holistic manner in older age, especially in residential settings where psychological vulnerability may go unnoticed.

A limitation of this study is that the LPA requires the selection of a limited number of variables, which implies that some relevant aspects of mental health have been left out of the model, such as loneliness, anxiety, coping or spirituality. Moreover, although suicidal ideation has been considered as one of the key variables, passive suicide, a particularly frequent manifestation in residential settings (Lee, 2023), has not been explicitly included, and which could provide valuable information for a better characterisation of risk profiles.

Additionally, it should be acknowledged that the profiles obtained largely refer to participants without cognitive impairment and with full ability to respond to all questionnaires during the interview, as only a small proportion of the sample presented MCI. Therefore, future replications should be conducted exclusively with participants without cognitive impairment.

Finally, the cross-sectional design prevents us from establishing causal relationships between the variables analysed and the profiles identified. Future research should consider longitudinal designs to examine the evolution of the profiles over time and the effectiveness of personalised interventions based on these profiles.