68-3

66/130

262only 10% of the variance5), and a study by Powell et al. (2000) similarly found that predictors correctly identified a mere 2% of suicide patients6). Chronic risk factors such as past suicide attempts and mental illness have been considered ill-suited by practitioners from the perspective of clinical effec-tiveness, as such risk factors cannot assess suicide risks that fluctuate over time. Recently, however, studies that have highlighted the value of proximal, near-term risk factors have been gaining attention. For example, Rudd et al. (2006) proposed a set of warning signs of suicide risk in the near term─in time frames of minutes, hours, and days7). Yaseen et al. (2019) proposed diagnostic criteria for predicting imminent suicide risk, namely the suicide crisis syndrome (SCS)8). However, one issue is that the warning signs do not necessarily arise only immediately before death by suicide9). One issue with the SCS is that it predicts post-dis-charge suicidal behaviour only in a limited number of cases8).Other authors have suggested that as interven-tions based on clinical risk assessments are insuffi-cient in preventing suicides among psychiatric in-patients, it is necessary to focus on a mix of other strategies, such as providing a safe environ-ment and observing patients effectively10-12). Menon (2013) argued that suicide risk should be assessed in three areas, namely risk factors, warning signs (current mental state), and protective factors4). However, he did not clarify the warning signs that are particular to psychiatric in-patients. Clinical inquiries, structured interviews, and self-assess-ment questionnaires have limited value in assessing suicide risk, given that some patients have attempted suicide despite denying having suicidal ideation13, 14) and in cases where the patient has trouble communicating or shows a decline in comprehension, or has limited trust in healthcare professionals. It is possible to obtain information through alternative means, such as by having ward staff monitor the in-patients’ behaviour or chat with them. Thus, it should be possible to improve the precision of risk assessments by looking out for specific objective signs that indicate imminent suicide risk in the in-patients’ behaviour in the ward, instead of relying on diagnostic records alone. We conducted a retrospective study of the objec-tive signs of imminent suicide risk among psychi-atric in-patients as drawn from a combination of diagnostic and nursing records (nurses’ recorded observations of patient behaviour in the ward).SampleWe analysed the diagnostic and nursing records of 18 in-patients (6 male, 12 female) maintained by a psychiatric hospital. These people were found to have died by suicide during their hospital stay, between March 2008 and July 2019 (11 years and 4 months). In each case, the surviving family members did not object to their data being used in this study. The sample included cases in which the patient died by suicide during temporary leave and in which the patient was rushed to hospital after a suicide attempt and subsequently died in the hospital. It may be unclear in some cases as to whether the death really was by suicide or acci-dent (the person is no longer around to confirm this). Accordingly, in this study, we defined suicide as an event in which the person committed an act that resulted in their death, irrespective of whether or not they anticipated this fatal outcome15). We considered utterances, behaviours, attitudes, and mental states observed over a period of 14 days before the patient’s fatal suicide attempt as signs of imminent suicide risks.Preparing the Data to be AnalysedIn Resource A, we compiled the patients’ basic information, which comprised diagnostic records made at time of admission, their diagnosis, period of hospitalisation, and history of suicide attempts, psychiatric symptoms, suicide in their families, and diseases. In Resource B, we arranged the diagnostic and nursing records into a time series for a period of 14 days before the fatal suicide attempt. In both Resources A and B, all data were anonymised.Analysis 1From Resources A and B, we aggregated the cases in which the patient had attempted suicide in the past; the patient’s problem list at the time of admission mentioned suicidal behaviour or ideation; and the patient showed signs of suicide in the 14 days before the fatal suicide attempt, including suicidal utterances, self-injury, and unsuccessful suicide attempts.Materials and Methods