Identifying Psychosis Episodes in Psychiatric Admission Notes via Rule-based Methods, Machine Learning, and Pre-Trained Language Models.

Early and accurate diagnosis is crucial for effective treatment and improved outcomes, yet identifying psychotic episodes presents significant challenges due to its complex nature and the varied presentation of symptoms among individuals. One of the primary difficulties lies in the underreporting and underdiagnosis of psychosis, compounded by the stigma surrounding mental health and the individuals' often diminished insight into their condition. Existing efforts leveraging Electronic Health Records (EHRs) to retrospectively identify psychosis typically rely on structured data, such as medical codes and patient demographics, which frequently lack essential information. Addressing these challenges, our study leverages Natural Language Processing (NLP) algorithms to analyze psychiatric admission notes for the diagnosis of psychosis, providing a detailed evaluation of rule-based algorithms, machine learning models, and pre-trained language models. Additionally, the study investigates the effectiveness of employing keywords to streamline extensive note data before training and evaluating the models. Analyzing 4,617 initial psychiatric admission notes (1,196 cases of psychosis versus 3,433 controls) from 2005 to 2019, we discovered that the XGBoost classifier employing Term Frequency-Inverse Document Frequency (TF-IDF) features derived from notes pre-selected by expert-curated keywords, attained the highest performance with an F1 score of 0.8881 (AUROC [95% CI]: 0.9725 [0.9717, 0.9733]). BlueBERT demonstrated comparable efficacy an F1 score of 0.8841 (AUROC [95% CI]: 0.97 [0.9580,0.9820]) on the same set of notes. Both models markedly outperformed traditional International Classification of Diseases (ICD) code-based detection methods from discharge summaries, which had an F1 score of 0.7608, thus improving the margin by 0.12. Furthermore, our findings indicate that keyword pre-selection markedly enhances the performance of both machine learning and pre-trained language models. This study illustrates the potential of NLP techniques to improve psychosis detection within admission notes and aims to serve as a foundational reference for future research on applying NLP for psychosis identification in EHR notes.

Identifying Psychosis Episodes in Psychiatric Admission Notes via Rule-based Methods, Machine Learning, and Pre-Trained Language Models.

Early and accurate diagnosis is crucial for effective treatment and improved outcomes, yet identifying psychotic episodes presents significant challenges due to its complex nature and the varied presentation of symptoms among individuals. One of the primary difficulties lies in the underreporting and underdiagnosis of psychosis, compounded by the stigma surrounding mental health and the individuals' often diminished insight into their condition. Existing efforts leveraging Electronic Health Records (EHRs) to retrospectively identify psychosis typically rely on structured data, such as medical codes and patient demographics, which frequently lack essential information. Addressing these challenges, our study leverages Natural Language Processing (NLP) algorithms to analyze psychiatric admission notes for the diagnosis of psychosis, providing a detailed evaluation of rule-based algorithms, machine learning models, and pre-trained language models. Additionally, the study investigates the effectiveness of employing keywords to streamline extensive note data before training and evaluating the models. Analyzing 4,617 initial psychiatric admission notes (1,196 cases of psychosis versus 3,433 controls) from 2005 to 2019, we discovered that the XGBoost classifier employing Term Frequency-Inverse Document Frequency (TF-IDF) features derived from notes pre-selected by expert-curated keywords, attained the highest performance with an F1 score of 0.8881 (AUROC [95% CI]: 0.9725 [0.9717, 0.9733]). BlueBERT demonstrated comparable efficacy an F1 score of 0.8841 (AUROC [95% CI]: 0.97 [0.9580, 0.9820]) on the same set of notes. Both models markedly outperformed traditional International Classification of Diseases (ICD) code-based detection methods from discharge summaries, which had an F1 score of 0.7608, thus improving the margin by 0.12. Furthermore, our findings indicate that keyword pre-selection markedly enhances the performance of both machine learning and pre-trained language models. This study illustrates the potential of NLP techniques to improve psychosis detection within admission notes and aims to serve as a foundational reference for future research on applying NLP for psychosis identification in EHR notes.

Causes of death among Australians with type 1 or type 2 diabetes, 2002-2019.

AIMS: This population-based study sought to explore in detail the conditions driving the diversification in causes of death among people with diabetes. METHODS: We linked Australians with type 1 or type 2 diabetes of all ages on the National Diabetes Services Scheme to the National Death Index for 2002-2019. We investigated the proportional contributions of different causes of death to total deaths over time across eight categories of causes of death, stratified by sex and diabetes type. The underlying causes of death were classified according to the International Classification of Diseases, Tenth Revision codes. RESULTS: Between 2002 and 2019, there was a shift in the causes of death among Australians with diabetes away from cardiovascular disease. The proportion of deaths attributed to cardiovascular disease declined in both sexes (ptrend <0.001), most substantially among women with type 2 diabetes from 48.2% in 2002 to 30.7% in 2019. Among men with type 2 diabetes, cancer replaced cardiovascular disease as the leading cause of death. The proportion of deaths due to dementia increased overall, from 2% in 2002 to over 7% in 2019, and across all age groups, notably from 1% to 4% in those aged 70-79. The proportion of deaths due to falls and Parkinson's disease also increased. CONCLUSIONS: There has been a shift of causes of death among those with diabetes away from cardiovascular disease. The proportion of deaths due to conditions such as dementia and falls is increasing among those with diabetes, which will require consideration when planning future resource allocation.

What can death records tell us about multimorbidity?

BACKGROUND: Multimorbidity has been measured from many data sources which show that prevalence increases with age and is usually greater among women than men and in more recent periods. Analyses of multiple cause of death data have shown different patterns of multimorbidity associated with demographic and other characteristics. METHODS: Deaths in Australia among over 1.7 million decedents aged 55+ were stratified into three types: medically certified deaths, coroner-referred deaths with natural underlying causes and coroner-referred deaths with external underlying causes. Multimorbidity was measured by prevalence of ≥2 causes and analysed over three periods based on administrative changes: 2006-2012, 2013-2016 and 2017-2018. Poisson regression was used to examine the influence of gender, age and period. RESULTS: The prevalence of deaths with multimorbidity was 81.0% for medically certified deaths, 61.1% for coroner-referred deaths with natural underlying causes and 82.4% for coroner-referred deaths with external underlying causes. For medically certified deaths, multimorbidity increased with age: incidence rate ratio (IRR 1.070, 95% CI 1.068, 1.072) was lower for women than men (0.954, 95% CI 0.952, 0.956) and changed little over time. For coroner-referred deaths with natural underlying causes, multimorbidity showed the expected pattern increasing with age (1.066, 95% CI 1.062, 1.070) and being higher for women than men (1.025, 95% CI 1.015, 1.035) and in more recent periods. For coroner-referred deaths with external underlying causes, there were marked increases over time that differed by age group due to changes in coding processes. CONCLUSION: Death records can be used to examine multimorbidity in national populations but, like other data sources, how the data were collected and coded impacts the conclusions.

Changes in anxiety and depression in patients with different income levels through the COVID-19 pandemic.

BACKGROUND: Lower socioeconomic status is known to be associated with high mental health burden, there have been few epidemiological studies showing how socioeconomic status has modified the effect of COVID-19 on anxiety and depression. METHODS: We analyzed data from the National Health Interview Survey in the United States between 2019 and 2021 and used respondents with a documented income-to-poverty ratio as a measure of income level (n = 79,468). We used frequency of medication use and self-reported frequency of anxious and depressive episodes as the main outcome measures. We performed a multivariable logistic regression with a two-way interaction term between income and survey year. RESULTS: We found a statistically significant worsening of depression and anxiety metrics in respondents with higher income levels from 2019 to 2021. We did not observe a significant change in anxiety and depression metrics for low-income respondents over the same period. LIMITATIONS: The data from the NHIS survey is limited primarily by sampling bias (response rate of 50.7 % in 2021), as well as the self-reported nature of the one of the outcome measures. CONCLUSION: These findings suggest that, within the limits of the National Health Interview Survey, mental health outcomes were worse but stable in a socioeconomically disadvantaged demographic between 2019 and 2021. In a higher socioeconomic bracket, mental health outcomes were less severe than the disadvantaged demographic but were worsening at a greater rate.

A new data driven method for summarising multiple cause of death data.

BACKGROUND: National mortality statistics are based on a single underlying cause of death. This practice does not adequately represent the impact of the range of conditions experienced in an ageing population in which multimorbidity is common. METHODS: We propose a new method for weighting the percentages of deaths attributed to different causes that takes account of the patterns of associations among underlying and contributing causes of death. It is driven by the data and unlike previously proposed methods does not rely on arbitrary choices of weights which can over-emphasise the contribution of some causes of death. The method is illustrated using Australian mortality data for people aged 60 years or more. RESULTS: Compared to the usual method based only on the underlying cause of death the new method attributes higher percentages of deaths to conditions like diabetes and dementia that are frequently mentioned as contributing causes of death, rather than underlying causes, and lower percentages to conditions to which they are closely related such as ischaemic heart disease and cerebrovascular disease. For some causes, notably cancers, which are usually recorded as underlying causes with few if any contributing causes the new method produces similar percentages to the usual method. These different patterns among groups of related conditions are not apparent if arbitrary weights are used. CONCLUSION: The new method could be used by national statistical agencies to produce additional mortality tables to complement the current tables based only on underlying causes of death.

Persistent effects of COVID-19 in patients hospitalized during the first wave of the pandemic: The impact of persistent fatigue on quality of life in a cross-sectional study.

COVID-19 can affect physical and mental health long after acute infection. In this descriptive study, 48 individuals hospitalized for COVID-19 between April and May 2020 were interviewed regarding their experience with COVID-19 after hospitalization. The mean age of participants was 51.1 (±11.91) years (range 25-65 years) and 26 (54.2%) were men. Individuals had a mean of 1.2 (±0.94) comorbidities associated with more severe COVID-19, with hypertension (37.5%) being most common. Nineteen (39.6%) individuals required treatment in the intensive care unit. Participants were interviewed a median time of 553 days (IQR, 405.5-589.0) after discharge from the hospital. Thirty-seven (77.1%) individuals had 5 or more persistent symptoms at time of interview with only 3 (6.3%) experiencing none. The most reported persistent symptoms were fatigue (79.2%), difficulty breathing (68.8%), and muscle weakness (60.4%). Poor quality of life was experienced by 39 (81.3%) participants and 8 (16.7%) had a posttraumatic stress disorder (PTSD) score within the clinical range for diagnosis. For multivariable analyses, persistent fatigue was significantly predicted by number of symptoms during acute COVID-19 (t = 4.4, p < 0.001). Number of symptoms during acute COVID-19 was also significantly associated with persistent dyspnea (t = 3.4, p = 0.002). Higher scores on the Chalder fatigue scale after COVID-19 was significantly associated with poor quality of life (t = 2.6, p = 0.01) and PTSD symptomatology (t = 2.9, p = 0.008). More research is needed to highlight the wide range of resources those suffering from Long COVID require long after discharge.

Analysis of Multiple Causes of Death: A Review of Methods and Practices.

BACKGROUND: Research and reporting of mortality indicators typically focus on a single underlying cause of death selected from multiple causes recorded on a death certificate. The need to incorporate the multiple causes in mortality statistics-reflecting increasing multimorbidity and complex causation patterns-is recognized internationally. This review aims to identify and appraise relevant analytical methods and practices related to multiple causes. METHODS: We searched Medline, PubMed, Scopus, and Web of Science from their incept ion to December 2020 without language restrictions, supplemented by consultation with international experts. Eligible articles analyzed multiple causes of death from death certificates. The process identified 4,080 items of which we reviewed 434 full-text articles. RESULTS: Most articles we reviewed (76%, n = 332) were published since 2001. The majority of articles examined mortality by "any- mention" of the cause of death (87%, n = 377) and assessed pairwise combinations of causes (57%, n = 245). Since 2001, applications of methods emerged to group deaths based on common cause patterns using, for example, cluster analysis (2%, n = 9), and application of multiple-cause weights to re-evaluate mortality burden (1%, n = 5). We describe multiple-cause methods applied to specific research objectives for approaches emerging recently. CONCLUSION: This review confirms rapidly increasing international interest in the analysis of multiple causes of death and provides the most comprehensive overview, to our knowledge, of methods and practices to date. Available multiple-cause methods are diverse but suit a range of research objectives. With greater availability of data and technology, these could be further developed and applied across a range of settings.

Quantifying cause-related mortality in Australia, incorporating multiple causes: observed patterns, trends and practical considerations.

BACKGROUND: Mortality statistics using a single underlying cause of death (UC) are key health indicators. Rising multimorbidity and chronic disease mean that deaths increasingly involve multiple conditions. However, additional causes reported on death certificates are rarely integrated into mortality indicators, partly due to complexities in data and methods. This study aimed to assess trends and patterns in cause-related mortality in Australia, integrating multiple causes (MC) of death. METHODS: Deaths (n = 1 773 399) in Australia (2006-17) were mapped to 136 ICD-10-based groups and MC indicators applied. Age-standardized cause-related rates (deaths/100 000) based on the UC (ASRUC) were compared with rates based on any mention of the cause (ASRAM) using rate ratios (RR = ASRAM/ASRUC) and to rates based on weighting multiple contributing causes (ASRW). RESULTS: Deaths involved on average 3.4 causes in 2017; the percentage with >4 causes increased from 20.9 (2006) to 24.4 (2017). Ischaemic heart disease (ASRUC = 73.3, ASRAM = 135.8, ASRW = 63.5), dementia (ASRUC = 51.1, ASRAM = 98.1, ASRW = 52.1) and cerebrovascular diseases (ASRUC = 39.9, ASRAM = 76.7, ASRW = 33.5) ranked as leading causes by all methods. Causes with high RR included hypertension (ASRUC = 2.2, RR = 35.5), atrial fibrillation (ASRUC = 8.0, RR = 6.5) and diabetes (ASRUC = 18.5, RR = 3.5); the corresponding ASRW were 12.5, 12.6 and 24.0, respectively. Renal failure, atrial fibrillation and hypertension ranked among the 10 leading causes by ASRAM and ASRW but not by ASRUC. Practical considerations in working with MC data are discussed. CONCLUSIONS: Despite the similarities in leading causes under the three methods, with integration of MC several preventable diseases emerged as leading causes. MC analyses offer a richer additional perspective for population health monitoring and policy development.

Identifying Diagnoses of Schizophrenia Spectrum Disorder in Large Data Sets.

Objective: The authors used a large clinical data set to determine which index diagnoses of schizophrenia spectrum disorder were new diagnoses. Methods: Using the Massachusetts All-Payer Claims Database (2012­2016), the authors identified patients with a schizophrenia spectrum disorder diagnosis in 2016 (index diagnosis) and then reviewed patients' care histories for the previous 12, 24, 36, and 48 months to identify previous diagnoses. Logistic regression was used to examine patient characteristics associated with the index diagnosis being a new diagnosis. Results: Overall, 7,217 individuals ages 15­35 years had a 2016 diagnosis of schizophrenia spectrum disorder; 67.7% had at least 48 months of historical data. Among those with at least 48 months of care history, 23% had no previous diagnoses. Diagnoses from inpatient psychiatric admissions or among female or younger patients were more likely to represent new diagnoses, compared with diagnoses from most other diagnosis locations or among males or older age groups, and outpatient diagnoses were less likely to represent new diagnoses than were most other diagnosis settings. Reviewing 48 instead of 12 months of data reduced estimated rates of new diagnoses from 112 to 66 per 100,000 persons; historical diagnoses were detected for 61% and 77% of patients with 12 or 48 months of care history, respectively. Conclusions: Examining multiple years of patient history spanning all payers and providers is critical to identifying new schizophrenia spectrum disorder diagnoses in large data sets. Review of 48 months of care history resulted in lower rates of new schizophrenia spectrum disorder diagnoses than previously reported.

Evidence for Schizophrenia-Specific Pathophysiology of Nicotine Dependence.

Tobacco use is the top preventable cause of early mortality in schizophrenia. Over 60% of people with schizophrenia smoke, three times the general prevalence. The biological basis of this increased risk is not understood, and existing interventions do not target schizophrenia-specific pathology. We therefore used a connectome-wide analysis to identify schizophrenia-specific circuits of nicotine addiction. We reanalyzed data from two studies: In Cohort 1, 35 smokers (18 schizophrenia, 17 control) underwent resting-state fMRI and clinical characterization. A multivariate pattern analysis of whole-connectome data was used to identify the strongest links between cigarette use and functional connectivity. In Cohort 2, 12 schizophrenia participants and 12 controls were enrolled in a randomized, controlled crossover study of nicotine patch with resting-state fMRI. We correlated change in network functional connectivity with nicotine dose. In Cohort 1, the strongest (p < 0.001) correlate between connectivity and cigarette use was driven by individual variation in default mode network (DMN) topography. In individuals with greater daily cigarette consumption, we observed a pathological expansion of the DMN territory into the identified parieto-occipital region, while in individuals with lower daily cigarette consumption, this region was external to the DMN. This effect was entirely driven by schizophrenia participants. Given the relationship between DMN topography and nicotine use we observed in Cohort 1, we sought to directly test the impact of nicotine on this network using an independent second cohort. In Cohort 2, nicotine reduced DMN connectivity in a dose-dependent manner (R = -0.50; 95% CI -0.75 to -0.12, p < 0.05). In the placebo condition, schizophrenia subjects had hyperconnectivity compared to controls (p < 0.05). Nicotine administration normalized DMN hyperconnectivity in schizophrenia. We here provide direct evidence that the biological basis of nicotine dependence is different in schizophrenia and in non-schizophrenia populations. Our results suggest the high prevalence of nicotine use in schizophrenia may be an attempt to correct a network deficit known to interfere with cognition.

Geographical variation in hospitalization for psychosis associated with cannabis use and cannabis legalization in the United States: Submit to: Psychiatry Research.

The 2017 National Inpatient Sample database was utilized to investigate the association between cannabis legalization in the United States and hospitalizations for psychosis associated with cannabis use. We compared the odds of hospital discharges for psychosis associated with cannabis use in adults between the Pacific census division (where most states legalized recreational cannabis use) and other divisions using multivariable logistic regression, adjusting for confounders. We calculated a score for each census division representing cannabis legality as the population-weighted sum of state scores: 1=illegal or cannabidiol/low potency cannabis; 2= medical marijuana; and 3=recreational and medical marijuana legalized. Pearson's correlation coefficients (r) quantified the relationship between scores and the proportion of hospitalizations with psychosis associated with cannabis. In 2017, there were an estimated 129,070 hospital discharges for psychosis associated with cannabis use. The Pacific census division had significantly higher odds of discharges than other divisions (adjusted odds ratio 1.55; 95% confidence interval 1.25 - 1.93). There was a significant correlation between the cannabis legality score and proportion of hospital discharges for psychosis associated with cannabis use (r = 0.67, p<0.05). In conclusion, we observed a higher proportion of hospital discharges for psychosis associated with cannabis use in areas with more liberal cannabis legalization laws.

Enhancing Australian Mortality Data to Meet Future Health Information Demands.

The Australian mortality data are a foundational health dataset which supports research, policy and planning. The COVID-19 pandemic necessitated the need for more timely mortality data that could assist in monitoring direct mortality from the virus as well as indirect mortality due to social and economic societal change. This paper discusses the evolution of mortality data in Australia during the pandemic and looks at emerging opportunities associated with electronic infrastructure such as electronic Medical Certificates of Cause of Death (eMCCDs), ICD-11 and automated coding tools that will form the foundations of a more responsive and comprehensive future mortality dataset.

Education-related inequalities in cause-specific mortality: first estimates for Australia using individual-level linked census and mortality data.

BACKGROUND: Socioeconomic inequalities in mortality are evident in all high-income countries, and ongoing monitoring is recommended using linked census-mortality data. Using such data, we provide the first estimates of education-related inequalities in cause-specific mortality in Australia, suitable for international comparisons. METHODS: We used Australian Census (2016) linked to 13 months of Death Registrations (2016-17). We estimated relative rates (RR) and rate differences (RD, per 100 000 person-years), comparing rates in low (no qualifications) and intermediate (secondary school) with high (tertiary) education for individual causes of death (among those aged 25-84 years) and grouped according to preventability (25-74 years), separately by sex and age group, adjusting for age, using negative binomial regression. RESULTS: Among 13.9 M people contributing 14 452 732 person-years, 84 743 deaths occurred. All-cause mortality rates among men and women aged 25-84 years with low education were 2.76 [95% confidence interval (CI): 2.61-2.91] and 2.13 (2.01-2.26) times the rates of those with high education, respectively. We observed inequalities in most causes of death in each age-sex group. Among men aged 25-44 years, relative and absolute inequalities were largest for injuries, e.g. transport accidents [RR = 10.1 (5.4-18.7), RD = 21.2 (14.5-27.9)]). Among those aged 45-64 years, inequalities were greatest for chronic diseases, e.g. lung cancer [men RR = 6.6 (4.9-8.9), RD = 57.7 (49.7-65.8)] and ischaemic heart disease [women RR = 5.8 (3.7-9.1), RD = 20.2 (15.8-24.6)], with similar patterns for people aged 65-84 years. When grouped according to preventability, inequalities were large for causes amenable to behaviour change and medical intervention for all ages and causes amenable to injury prevention among young men. CONCLUSIONS: Australian education-related inequalities in mortality are substantial, generally higher than international estimates, and related to preventability. Findings highlight opportunities to reduce them and the potential to improve the health of the population.

Unintentional and intentional drug poisoning deaths, Australia, 2012-2016: Drug pattern profile and demographic characteristics.

BACKGROUND: Identifying differences in unintentional versus intentional drug poisoning deaths can inform targeted prevention. This study aimed to: compare unintentional versus intentional drug poisoning deaths by drug involvement, age and sex; describe patterns of drug involvement by intent; and describe common drug patterns by age and sex. METHODS: Cases comprised deaths among Australians aged ≥15 where drug poisoning was the underlying cause (Cause of Death Unit Record File 2012-2016). Sex, age, and drug involvement were analysed by intent using logistic regression. RESULTS: Of 7994 deaths, 71% were unintentional and 24% intentional. Compared with unintentional deaths, intentional deaths were more likely among females (OR 1.31 [95% CI 1.16-1.48]) and those aged 55+ (1.50 [1.25-1.81] for 55-64 years; 3.79 [3.07-4.66] for 65+ years, compared to 35-44 years), and were more likely to involve hypnosedatives (2.11 [1.87-2.39]), other psychotropic medicines (1.58 [1.39-1.78]), non-opioid analgesics and anaesthetics (1.48 [1.25-1.73]). Common unintentional profiles comprised: opioids (excluding heroin); heroin; alcohol; opioids with hypnosedatives; opioids with hypnosedatives and other psychotropic medicines; stimulants; other psychotropic medicines; and opioids with other psychotropic medicines. Unintentional deaths involving heroin or stimulants only had a greater proportion of males (79.0% and 83.4%, respectively) and younger individuals aged 15-34 (30.3% and 39.5%, respectively). Common intentional profiles comprised: hypnosedatives; other psychotropic medicines; opioids (excluding heroin); hypnosedatives with other psychotropic medicines; opioids with hypnosedatives; and opioids with hypnosedatives and other psychotropic medicines. CONCLUSION: The demographic and drug involvement profile of intentional and unintentional deaths were distinct, suggesting different approaches to prevention are necessary.

Nicotine acutely alters temporal properties of resting brain states.

BACKGROUND: Nicotine-dependent individuals have altered activity in neurocognitive networks such as the default mode (DMN), salience (SN) and central executive networks (CEN). One theory suggests that, among chronic tobacco smokers, nicotine abstinence drives more DMN-related internal processing while nicotine replacement suppresses DMN and enhances SN and CEN. Whether acute nicotine impacts network dynamics in non-smokers is, however, unknown. METHODS: In a randomized double-blind crossover study, 17 healthy non-smokers (8 females) were administered placebo and nicotine (2-mg lozenge) on two different days prior to collecting resting-state functional magnetic resonance imaging (fMRI). Previously defined brain states in 462 individuals that spatially overlap with well-characterized resting-state networks including the DMN, SN, and CEN were applied to compute state-specific dynamics at rest: total time spent in state, persistence in each state after entry, and frequency of state transitions. We examined whether nicotine acutely alters these resting-state dynamics. RESULTS: A significant drug-by-state interaction emerged; post-hoc analyses clarified that, relative to placebo, nicotine suppressed time spent in a frontoinsular-DMN state (posterior cingulate cortex, medial prefrontal cortex, anterior insula, striatum and orbitofrontal cortex) and enhanced time spent in a SN state (anterior cingulate cortex and insula). No significant findings were observed for persistence and frequency. CONCLUSIONS: In non-smokers, nicotine biases resting-state brain function away from the frontoinsular-DMN and toward the SN, which may reduce internally focused cognition and enhance salience processing. While past work suggests nicotine impacts DMN activity, the current work shows nicotinic influences on a specific DMN-like network that has been linked with rumination and depression.

Reporting of alcohol as a contributor to death in Australian national suicide statistics and its relationship to post-mortem alcohol concentrations.

AIM: To describe the assignment of International Classification of Disease (ICD)-10 alcohol codes as underlying or contributory causes of death by the Australian Bureau of Statistics during mortality coding for suicides according to the blood alcohol concentration (BAC) detected at autopsy. DESIGN: Population-based case-series descriptive analysis. SETTING AND PARTICIPANTS: Data for all alcohol-related (Alc+) suicide deaths (aged 15+) in Australia from 2010-2015 (n = 3132) from the National Coronial Information System. MEASUREMENTS: Alc+ suicides were categorised as those with a post-mortem BAC ≥0.05 g/100 mL. The outcome variable was whether the case was assigned an ICD-10 alcohol code (F10.0-F10.9, R78.0, T51, X45 and/or X65). We estimated OR for the assignment of codes in Alc+ suicides using BAC as the key predictor. We also examined several covariates that have been implicated in the risk of Alc+ suicides. FINDINGS: An ICD-10 alcohol code was assigned during the mortality coding process in 47.6% (n = 1491) of Alc+ suicides. Higher BAC was associated with higher odds of having a code assigned; cases with a BAC over 0.20 g/100 mL over were twice as likely to have an alcohol code assigned (adjusted OR [AOR] = 2.06, 95% CI = 1.59, 2.67) compared with cases with a BAC of 0.050-0.075 g/100 mL. Compared with New South Wales, higher likelihood of code assignment was found in Northern Territory (AOR = 3.85, 95% CI = 2.32, 6.63) and Western Australia (AOR = 2.89, 95% CI = 2.27, 3.68). Compared with 15-24 year olds, 25-44 (AOR = 0.79, 95% CI = 0.63, 0.99) and 65-84 year olds (AOR = 0.63, 95% CI = 0.43, 0.93) were less likely to have a code assigned. CONCLUSIONS: An ICD-10 alcohol code was not assigned as an underlying or contributory cause of death in over half of suicides in Australia (2010-2015) with a BAC ≥0.05 g/100 mL. The higher the BAC detected at autopsy, the more likely cases were to be assigned an alcohol code during the mortality coding process.