Identifying Predictors of Suicide in Severe Mental Illness: A Feasibility Study of a Clinical Prediction Rule (Oxford Mental Illness and Suicide Tool or OxMIS).

BACKGROUND: Oxford Mental Illness and Suicide tool (OxMIS) is a brief, scalable, freely available, structured risk assessment tool to assess suicide risk in patients with severe mental illness (schizophrenia-spectrum disorders or bipolar disorder). OxMIS requires further external validation, but a lack of large-scale cohorts with relevant variables makes this challenging. Electronic health records provide possible data sources for external validation of risk prediction tools. However, they contain large amounts of information within free-text that is not readily extractable. In this study, we examined the feasibility of identifying suicide predictors needed to validate OxMIS in routinely collected electronic health records. METHODS: In study 1, we manually reviewed electronic health records of 57 patients with severe mental illness to calculate OxMIS risk scores. In study 2, we examined the feasibility of using natural language processing to scale up this process. We used anonymized free-text documents from the Clinical Record Interactive Search database to train a named entity recognition model, a machine learning technique which recognizes concepts in free-text. The model identified eight concepts relevant for suicide risk assessment: medication (antidepressant/antipsychotic treatment), violence, education, self-harm, benefits receipt, drug/alcohol use disorder, suicide, and psychiatric admission. We assessed model performance in terms of precision (similar to positive predictive value), recall (similar to sensitivity) and F1 statistic (an overall performance measure). RESULTS: In study 1, we estimated suicide risk for all patients using the OxMIS calculator, giving a range of 12 month risk estimates from 0.1-3.4%. For 13 out of 17 predictors, there was no missing information in electronic health records. For the remaining 4 predictors missingness ranged from 7-26%; to account for these missing variables, it was possible for OxMIS to estimate suicide risk using a range of scores. In study 2, the named entity recognition model had an overall precision of 0.77, recall of 0.90 and F1 score of 0.83. The concept with the best precision and recall was medication (precision 0.84, recall 0.96) and the weakest were suicide (precision 0.37), and drug/alcohol use disorder (recall 0.61). CONCLUSIONS: It is feasible to estimate suicide risk with the OxMIS tool using predictors identified in routine clinical records. Predictors could be extracted using natural language processing. However, electronic health records differ from other data sources, particularly for family history variables, which creates methodological challenges.

A web-based clinical decision tool to support treatment decision-making in psychiatry: a pilot focus group study with clinicians, patients and carers.

BACKGROUND: Treatment decision tools have been developed in many fields of medicine, including psychiatry, however benefits for patients have not been sustained once the support is withdrawn. We have developed a web-based computerised clinical decision support tool (CDST), which can provide patients and clinicians with continuous, up-to-date, personalised information about the efficacy and tolerability of competing interventions. To test the feasibility and acceptability of the CDST we conducted a focus group study, aimed to explore the views of clinicians, patients and carers. METHODS: The CDST was developed in Oxford. To tailor treatments at an individual level, the CDST combines the best available evidence from the scientific literature with patient preferences and values, and with patient medical profile to generate personalised clinical recommendations. We conducted three focus groups comprising of three different participant types: consultant psychiatrists, participants with a mental health diagnosis and/or experience of caring for someone with a mental health diagnosis, and primary care practitioners and nurses. Each 1-h focus group started with a short visual demonstration of the CDST. To standardise the discussion during the focus groups, we used the same topic guide that covered themes relating to the acceptability and usability of the CDST. Focus groups were recorded and any identifying participant details were anonymised. Data were analysed thematically and managed using the Framework method and the constant comparative method. RESULTS: The focus groups took place in Oxford between October 2016 and January 2017. Overall 31 participants attended (12 consultants, 11 primary care practitioners and 8 patients or carers). The main themes that emerged related to CDST applications in clinical practice, communication, conflicting priorities, record keeping and data management. CDST was considered a useful clinical decision support, with recognised value in promoting clinician-patient collaboration and contributing to the development of personalised medicine. One major benefit of the CDST was perceived to be the open discussion about the possible side-effects of medications. Participants from all the three groups, however, universally commented that the terminology and language presented on the CDST were too medicalised, potentially leading to ethical issues around consent to treatment. CONCLUSIONS: The CDST can improve communication pathways between patients, carers and clinicians, identifying care priorities and providing an up-to-date platform for implementing evidence-based practice, with regard to prescribing practices.

All-cause and external mortality in released prisoners: systematic review and meta-analysis.

OBJECTIVES: We systematically reviewed studies of mortality following release from prison and examined possible demographic and methodological factors associated with variation in mortality rates. METHODS: We searched 5 computer-based literature indexes to conduct a systematic review of studies that reported all-cause, drug-related, suicide, and homicide deaths of released prisoners. We extracted and meta-analyzed crude death rates and standardized mortality ratios by age, gender, and race/ethnicity, where reported. RESULTS: Eighteen cohorts met review criteria reporting 26,163 deaths with substantial heterogeneity in rates. The all-cause crude death rates ranged from 720 to 2054 per 100,000 person-years. Male all-cause standardized mortality ratios ranged from 1.0 to 9.4 and female standardized mortality ratios from 2.6 to 41.3. There were higher standardized mortality ratios in White, female, and younger prisoners. CONCLUSIONS: Released prisoners are at increased risk for death following release from prison, particularly in the early period. Aftercare planning for released prisoners could potentially have a large public health impact, and further work is needed to determine whether certain groups should be targeted as part of strategies to reduce mortality.