Learning vector representation of medical objects via EMR-driven nonnegative restricted Boltzmann machines (eNRBM)

Electronic medical record (EMR) offers promises for novel analytics. However, manual feature engineering from EMR is labor intensive because EMR is complex - it contains temporal, mixed-type and multimodal data packed in irregular episodes. We present a computational framework to harness EMR with minimal human supervision via restricted Boltzmann machine (RBM). The framework derives a new representation of medical objects by embedding them in a low-dimensional vector space. This new representation facilitates algebraic and statistical manipulations such as projection onto 2D plane (thereby offering intuitive visualization), object grouping (hence enabling automated phenotyping), and risk stratification. To enhance model interpretability, we introduced two constraints into model parameters: (a) nonnegative coefficients, and (b) structural smoothness. These result in a novel model called eNRBM (EMR-driven nonnegative RBM). We demonstrate the capability of the eNRBM on a cohort of 7578 mental health patients under suicide risk assessment. The derived representation not only shows clinically meaningful feature grouping but also facilitates short-term risk stratification. The F-scores, 0.21 for moderate-risk and 0.36 for high-risk, are significantly higher than those obtained by clinicians and competitive with the results obtained by support vector machines.

Physical health comorbidities in women with personality disorder: data from the Geelong osteoporosis study

BACKGROUND: Associations between common psychiatric disorders, psychotic disorders and physical health comorbidities are frequently investigated. The complex relationship between personality disorders (PDs) and physical health is less understood, and findings to date are varied. This study aims to investigate associations between PDs with a number of prevalent physical health conditions. METHODS: This study examined data collected from women (n=765;≥25years) participating in a population-based study located in south-eastern Australia. Lifetime history of psychiatric disorders was assessed using the semi-structured clinical interviews (SCID-I/NP and SCID-II). The presence of physical health conditions (lifetime) were identified via a combination of self-report, medical records, medication use and clinical data. Socioeconomic status, and information regarding medication use, lifestyle behaviors, and sociodemographic information was collected via questionnaires. Logistic regression models were used to investigate associations. RESULTS: After adjustment for sociodemographic variables (age, socioeconomic status) and health-related factors (body mass index, physical activity, smoking, psychotropic medication use), PDs were consistently associated with a range of physical health conditions. Novel associations were observed between Cluster A PDs and gastro-oesophageal reflux disease (GORD); Cluster B PDs with syncope and seizures, as well as arthritis; and Cluster C PDs with GORD and recurrent headaches. CONCLUSIONS: PDs were associated with physical comorbidity. The current data contribute to a growing evidence base demonstrating associations between PDs and a number of physical health conditions independent of psychiatric comorbidity, sociodemographic and lifestyle factors. Longitudinal studies are now required to investigate causal pathways, as are studies determining pathological mechanisms.

Invasive infections due to filamentous fungi other than Aspergillus: epidemiology and determinants of mortality

The epidemiology of invasive fungal disease (IFD) due to filamentous fungi other than Aspergillus may be changing. We analysed clinical, microbiological and outcome data in Australian patients to determine the predisposing factors and identify determinants of mortality. Proven and probable non-Aspergillus mould infections (defined according to modified European Organization for Research and Treatment of Cancer/Mycoses Study Group criteria) from 2004 to 2012 were evaluated in a multicentre study. Variables associated with infection and mortality were determined. Of 162 episodes of non-Aspergillus IFD, 145 (89.5%) were proven infections and 17 (10.5%) were probable infections. The pathogens included 29 fungal species/species complexes; mucormycetes (45.7%) and Scedosporium species (33.3%) were most common. The commonest comorbidities were haematological malignancies (HMs) (46.3%) diabetes mellitus (23.5%), and chronic pulmonary disease (16%); antecedent trauma was present in 21% of cases. Twenty-five (15.4%) patients had no immunocompromised status or comorbidity, and were more likely to have acquired infection following major trauma (p <0.01); 61 (37.7%) of cases affected patients without HMs or transplantation. Antifungal therapy was administered to 93.2% of patients (median 68 days, interquartile range 19-275), and adjunctive surgery was performed in 58.6%. The all-cause 90-day mortality was 44.4%; HMs and intensive-care admission were the strongest predictors of death (both p <0.001). Survival varied by fungal group, with the risk of death being significantly lower in patients with dematiaceous mould infections than in patients with other non-Aspergillus mould infections. Non-Aspergillus IFD affected diverse patient groups, including non-immunocompromised hosts and those outside traditional risk groups; therefore, definitions of IFD in these patients are required. Given the high mortality, increased recognition of infections and accurate identification of the causative agent are required.

Differentiating sub-groups of online depression-related communities using textual cues

Depression is a highly prevalent mental illness and is a comorbidity of other mental and behavioural disorders. The Internet allows individuals who are depressed or caring for those who are depressed, to connect with others via online communities; however, the characteristics of these online conversations and the language styles of those interested in depression have not yet been fully explored. This work aims to explore the textual cues of online communities interested in depression. A random sample of 5,000 blog posts was crawled. Five groupings were identified: depression, bipolar, self-harm, grief, and suicide. Independent variables included psycholinguistic processes and content topics extracted from the posts. Machine learning techniques were used to discriminate messages posted in the depression sub-group from the others.Good predictive validity in depression classification using topics and psycholinguistic clues as features was found. Clear discrimination between writing styles and content, with good predictive power is an important step in understanding social media and its use in mental health.

Evaluation of the Counterweight Programme for obesity management in primary care: a starting point for continuous improvement

Background Evaluation for obesity management in primary care is limited, and successful outcomes are from intensive clinical trials in hospital settings.

Aim To determine to what extent measures of success seen in intensive clinical trials can be achieved in routine primary care. Primary outcome measures were weight change and percentage of patients achieving ≤5% loss at 12 and 24 months.

Design of study Prospective evaluation of a new continuous improvement model for weight management in primary care.

Setting Primary care, UK.

Method Primary care practice nurses from 65 UK general practices delivered interventions to 1906 patients with body mass index (BMI) ≥30 kg/m2 or ≥28 kg/m2 with obesity-related comorbidities.

Results Mean baseline weight was 101.2 kg (BMI 37.1 kg/m2); 25% of patients had BMI ≥40 kg/m2 and 74% had ≥1 major obesity-related comorbidity. At final data capture 1419 patients were in the programme for ≥12 months, and 825 for ≥24 months. Mean weight change in those who attended and had data at 12 months (n = 642) was −3.0 kg (95% CI = −3.5 to −2.4 kg) and at 24 months (n = 357) was −2.3 kg (95% CI = −3.2 to −1.4 kg). Among attenders at specific time-points, 30.7% had maintained weight loss of ≥5% at 12 months, and 31.9% at 24 months. A total of 761 (54%) of all 1419 patients who had been enrolled in the programme for >12 months provided data at or beyond 12 months.

Conclusion This intervention achieves and maintains clinically valuable weight loss within routine primary care.

Morbidity patterns among the underweight, overweight and obese between 2 and 18 years: population-based cross-sectional analyses

CONTEXT: No study has documented how symptomatic morbidity varies across the body mass index (BMI) spectrum (underweight, normal weight, overweight and obese) or across the entire child and adolescent age range. OBJECTIVE: To (1) quantify physical and psychosocial morbidities experienced by 2-18-year-olds according to BMI status and (2) explore morbidity patterns by age. DESIGN, SETTING AND PARTICIPANTS: Cross-sectional data from two Australian population studies (the Longitudinal Study of Australian Children and the Health of Young Victorians Study) were collected during 2000-2006. Participants were grouped into five age bands: 2-3 (n=4606), 4-5 (n=4983), 6-7 (n=4464), 8-12 (n=1541) and 13-18 (n=928) years. MAIN MEASURES: Outcomes-Parent- and self-reported global health; physical, psychosocial and mental health; special health-care needs; wheeze; asthma and sleep problems. Exposure-measured BMI (kg m(-2)) categorised using standard international cutpoints. ANALYSES: The variation in comorbidities across BMI categories within and between age bands was examined using linear and logistic regression models. RESULTS: Comorbidities varied with BMI category for all except sleep problems, generally showing the highest levels for the obese category. However, patterns differed markedly between age groups. In particular, poorer global health and special health-care needs were associated with underweight in young children, but obesity in older children. Prevalence of poorer physical health varied little by BMI in 2-5-year-olds, but from 6 to 7 years was increasingly associated with obesity. Normal-weight children tended to experience the best psychosocial and mental health, with little evidence that the U-shaped associations of these variables with BMI status varied by age. Wheeze and asthma increased slightly with BMI at all ages. CONCLUSIONS: Deviation from normal weight is associated with health differences in children and adolescents that vary by morbidity and age. As well as lowering risks for later disease, promoting normal body weight appears central to improving the health and well-being of the young.

Anxiety in children with attention-deficit/hyperactivity disorder

OBJECTIVES: Although anxiety is common in children with attention-deficit/hyperactivity disorder (ADHD), it is unclear how anxiety influences the lives of these children. This study examined the association between anxiety comorbidities and functioning by comparing children with ADHD and no, 1, or ≥2 anxiety comorbidities. Differential associations were examined by current ADHD presentation (subtype). METHODS: Children with diagnostically confirmed ADHD (N = 392; 5-13 years) were recruited via 21 pediatrician practices across Victoria, Australia. Anxiety was assessed by using the Anxiety Disorders Interview Schedule for Children-IV. Functional measures included parent-reported: quality of life (QoL; Pediatric Quality of Life Inventory 4.0), behavior and peer problems (Strengths and Difficulties Questionnaire), daily functioning (Daily Parent Rating of Evening and Morning Behavior), and school attendance. Teacher-reported behavior and peer problems (Strengths and Difficulties Questionnaire) were also examined. Linear and logistic regression controlled for ADHD severity, medication use, comorbidities, and demographic factors. RESULTS: Children with ≥2 anxiety comorbidities (n = 143; 39%) had poorer QoL (effect size: -0.8) and more difficulties with behavior (effect size: 0.4) and daily functioning (effect size: 0.3) than children without anxiety (n = 132; 36%). Poorer functioning was not observed for children with 1 anxiety comorbidity (n = 95; 26%). Two or more anxiety comorbidities were associated with poorer functioning for children with both ADHD-Inattentive and ADHD-Combined presentation. CONCLUSIONS: Children with ADHD demonstrate poorer QoL, daily functioning and behavior when ≥2 anxiety comorbidities are present. Future research should examine whether treating anxiety in children with ADHD improves functional outcomes.

A systematic review of the impact of weight loss on cancer incidence and mortality

Obesity is well recognized as a significant risk factor for certain cancers; however, a corresponding risk reduction with weight loss is not yet clearly defined. This review aims to examine the literature investigating the effect of all types of weight loss on cancer incidence and mortality, and to more clearly describe the relationship between these two factors. A literature search identified 34 publications reporting weight loss data in relation to cancer incidence or mortality. All except one were observational studies and the majority used self-reported weights and did not define intentionality of weight loss. 16/34 studies found a significant inverse association between weight loss and cancer incidence or mortality. The remainder returned null findings. The observed association was more consistently seen in studies that investigated the effect of intentional weight loss (5/6 studies) and the risk reduction was greatest for obesity-related cancers and in women. In conclusion, intentional weight loss does result in a decreased incidence of cancer, particularly female obesity-related cancers. However, there is a need for further evaluation of sustained intentional weight loss in the obese with less reliance on self-reported weight data and more focus on male populations.

The prevalence of comorbid personality disorders in treatment-seeking problem gamblers: a systematic review and meta-analysis

The aim of this study was to systematically review and meta-analyze the prevalence of comorbid personality disorders among treatment-seeking problem gamblers. Almost one half (47.9%) of problem gamblers displayed comorbid personality disorders. They were most likely to display Cluster B disorders (17.6%), with smaller proportions reporting Cluster C disorders (12.6%) and Cluster A disorders (6.1%). The most prevalent personality disorders were narcissistic (16.6%), antisocial (14.0%), avoidant (13.4%), obsessive-compulsive (13.4%), and borderline (13.1%) personality disorders. Sensitivity analyses suggested that these prevalence estimates were robust to the inclusion of clinical trials and self-selected samples. Although there was significant variability in reported rates, subgroup analyses revealed no significant differences in estimates of antisocial personality disorder according to problem gambling severity, measure of comorbidity employed, and study jurisdiction. The findings highlight the need for gambling treatment services to conduct routine screening and assessment of co-occurring personality disorders and to provide treatment approaches that adequately address these comorbid conditions.

Areas of controversy in neuroprogression in bipolar disorder.

OBJECTIVE: We aimed to review clinical features and biological underpinnings related to neuroprogression in bipolar disorder (BD). Also, we discussed areas of controversy and future research in the field. METHOD: We systematically reviewed the extant literature pertaining to neuroprogression and BD by searching PubMed and EMBASE for articles published up to March 2016. RESULTS: A total of 114 studies were included. Neuroimaging and clinical evidence from cross-sectional and longitudinal studies show that a subset of patients with BD presents a neuroprogressive course with brain changes and unfavorable outcomes. Risk factors associated with these unfavorable outcomes are number of mood episodes, early trauma, and psychiatric and clinical comorbidity. CONCLUSION: Illness trajectories are largely variable, and illness progression is not a general rule in BD. The number of manic episodes seems to be the clinical marker more robustly associated with neuroprogression in BD. However, the majority of the evidence came from cross-sectional studies that are prone to bias. Longitudinal studies may help to identify signatures of neuroprogression and integrate findings from the field of neuroimaging, neurocognition, and biomarkers.

Impact of comorbidities and gender on the use of coronary interventions in patients with high-risk non-ST-segment elevation acute coronary syndrome

OBJECTIVES: To determine the impact of gender and comorbidity on use of coronary interventions in patients diagnosed with high-risk non-ST-segment acute coronary syndrome (NSTEACS). BACKGROUND: Guidelines recommend the use of coronary angiography for all patients diagnosed with NSTEACS with high-risk features, except in the presence of severe comorbidities. However, little is understood about the relationship between gender, comorbidity, and the use of coronary interventions. METHODS: Retrospective analyses of the Victorian Admitted Episodes Data Set (VAED) including all patients diagnosed with NSTEACS with high-risk features on their first admission for ACS between June 2007 and July 2009. Hierarchical logistic regression models and correspondence analyses were used to understand the relationship between gender, comorbidities, and the use of coronary interventions. RESULTS: Out of 16,771 NSTEACS patients with high-risk features, 6,338 (38%) were female. Females were older than males (aged ≥75: 62% vs 39%, p < 0.001) and more likely to have multiple comorbidities (≥2: 66% vs 59%, p < 0.001). After adjusting for potential confounders, females were more likely to receive no coronary intervention than males with a similar number of comorbid conditions (no comorbidities: OR 1.62, 95% CI 1.28-2.05; 1 comorbidity: OR 1.67, 95% CI 1.44-1.93; 2 comorbidities: OR 1.93, 95% CI 1.66-2.23; ≥3 comorbidities: OR 1.42, 95% CI 1.27-1.60). CONCLUSIONS: Lower rates of coronary intervention in females persisted after adjusting for number of comorbidities which suggests that gender may bias decisions regarding referral for coronary intervention in high-risk NSTEACS independent of other factors.

Gender differences in presentation, coronary intervention, and outcomes of 28,985 acute coronary syndrome patients in Victoria, Australia

BACKGROUND: Differences in demographics, presenting characteristics, and treatment of heart disease in women may contribute to adverse outcomes. The purpose of this paper was to describe gender differences in the epidemiology, treatment, and outcomes of all admissions for acute coronary syndrome (ACS) in Victoria that occurred between June 2007 and July 2009. METHODS: We undertook a retrospective cohort study of all patients admitted to Victorian hospitals with a first time diagnosis of ACS. Use of angiograms, percutaneous coronary intervention (PCI), coronary artery bypass graft (CABG), and adverse outcomes (death and/or unplanned readmission) were compared by gender and hierarchical logistic regression models were used to account for confounding variables. RESULTS: Of a total of 28,985 ACS patients, 10,455 (36%) were women. Compared with men, women were older (aged ≥75 years: 54% vs 31%; p < .001), more likely to present with multiple comorbidities (>1 comorbidity: 53% vs 46%; p < .001), and more likely to be diagnosed with non-ST-segment elevation ACS (86% vs 80%; p < .001). Women were less likely to receive coronary interventions (angiogram: adjusted odds ratio [aOR], 0.71; 95% CI, 0.66-0.75; PCI: aOR, 0.73; 95% CI, 0.66-0.80; CABG: aOR, 0.58; 95% CI, 0.53-0.64). Adverse outcomes were similar in women and men after accounting for confounding variables. CONCLUSIONS: Our results show that women in Victoria were less likely to receive coronary interventions after an admission for ACS. Clinicians should be wary of inherent gender bias in decisions to refer patients for angiography.

Gaps in understanding health and engagement with healthcare providers across common long-term conditions: a population survey of health literacy in 29 473 Danish citizens

OBJECTIVES: To (1) quantify levels of subjective health literacy in people with long-term health conditions (diabetes, cardiovascular disease, chronic obstructive pulmonary disease, musculoskeletal disorders, cancer and mental disorders) and compare these to levels in the general population and (2) examine the association between health literacy, socioeconomic characteristics and comorbidity in each long-term condition group.

DESIGN: Population-based survey in the Central Denmark Region (n=29,473). MAIN

OUTCOME MEASURES: Health literacy was measured using two scales from the Health Literacy Questionnaire (HLQ): (1) Ability to understand health information and (2) Ability to actively engage with healthcare providers.

RESULTS: People with long-term conditions reported more difficulties than the general population in understanding health information and actively engaging with healthcare providers. Wide variation was found between disease groups, with people with cancer having fewer difficulties and people with mental health disorders having more difficulties in actively engaging with healthcare providers than other long-term condition groups. Having more than one long-term condition was associated with more difficulty in engaging with healthcare providers and understanding health information. People with low levels of education had lower health literacy than people with high levels of education.

CONCLUSIONS: Compared with the general population, people with long-term conditions report more difficulties in understanding health information and engaging with healthcare providers. These two dimensions are critical to the provision of patient-centred healthcare and for optimising health outcomes. More effort should be made to respond to the health literacy needs among individuals with long-term conditions, multiple comorbidities and low education levels, to improve health outcomes and to reduce social inequality in health.

Timing of androgen-deprivation therapy in patients with prostate cancer with a rising PSA (TROG 03.06 and VCOG PR 01-03 [TOAD]): a randomised, multicentre, non-blinded, phase 3 trial

Background Androgen-deprivation therapy is offered to men with prostate cancer who have a rising prostate-specific antigen after curative therapy (PSA relapse) or who are considered not suitable for curative treatment; however, the optimal timing for its introduction is uncertain. We aimed to assess whether immediate androgen-deprivation therapy improves overall survival compared with delayed therapy. Methods In this randomised, multicentre, phase 3, non-blinded trial, we recruited men through 29 oncology centres in Australia, New Zealand, and Canada. Men with prostate cancer were eligible if they had a PSA relapse after previous attempted curative therapy (radiotherapy or surgery, with or without postoperative radiotherapy) or if they were not considered suitable for curative treatment (because of age, comorbidity, or locally advanced disease). We used a database-embedded, dynamically balanced, randomisation algorithm, coordinated by the Cancer Council Victoria, to randomly assign participants (1:1) to immediate androgen-deprivation therapy (immediate therapy arm) or to delayed androgen-deprivation therapy (delayed therapy arm) with a recommended interval of at least 2 years unless clinically contraindicated. Randomisation for participants with PSA relapse was stratified by type of previous therapy, relapse-free interval, and PSA doubling time; randomisation for those with non-curative disease was stratified by metastatic status; and randomisation in both groups was stratified by planned treatment schedule (continuous or intermittent) and treatment centre. Clinicians could prescribe any form and schedule of androgen-deprivation therapy and group assignment was not masked. The primary outcome was overall survival in the intention-to-treat population. The trial closed to accrual in 2012 after review by the independent data monitoring committee, but data collection continued for 18 months until Feb 26, 2014. It is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12606000301561) and ClinicalTrials.gov (NCT00110162). Findings Between Sept 3, 2004, and July 13, 2012, we recruited 293 men (261 with PSA relapse and 32 with non-curable disease). We randomly assigned 142 men to the immediate therapy arm and 151 to the delayed therapy arm. Median follow-up was 5 years (IQR 3·3–6·2) from the date of randomisation. 16 (11%) men died in the immediate therapy arm and 30 (20%) died in the delayed therapy arm. 5-year overall survival was 86·4% (95% CI 78·5–91·5) in the delayed therapy arm versus 91·2% (84·2–95·2) in the immediate therapy arm (log-rank p=0·047). After Cox regression, the unadjusted HR for overall survival for immediate versus delayed arm assignment was 0·55 (95% CI 0·30–1·00; p=0·050). 23 patients had grade 3 treatment-related adverse events. 105 (36%) men had adverse events requiring hospital admission; none of these events were attributable to treatment or differed between treatment-timing groups. The most common serious adverse events were cardiovascular, which occurred in nine (6%) patients in the delayed therapy arm and 13 (9%) in the immediate therapy arm. Interpretation Immediate receipt of androgen-deprivation therapy significantly improved overall survival compared with delayed intervention in men with PSA-relapsed or non-curable prostate cancer. The results provide benchmark evidence of survival rates and morbidity to discuss with men when considering their treatment options. Funding Australian National Health and Medical Research Council and Cancer Councils, The Royal Australian and New Zealand College of Radiologists, Mayne Pharma Australia.

Analysis of the impact of limitation of medical treatment orders during unplanned transfers from sub-acute care to emergency departments

BACKGROUND: The impact of limitation of medical treatment orders (LOMT) on patient outcomes following transfer from sub-acute care to the Emergency Department remains unclear.

METHODS: Retrospective medical record review of 431 adult in-patients who required ambulance transfer following clinical deterioration during a sub-acute care admission during 2010.

RESULTS: Common reasons for transfer were respiratory (18.9%) or neurological (19.0%) conditions; 35.7% (154/431) were transferred within one week of sub-acute care admission. LOMT orders were in place for 37.8% (n=163) patients who were older (p<0.001), with more comorbidities (p<0.005), specifically cardiac, renal and pulmonary disease than patients without LOMT. Patients with LOMT orders had more physiological abnormalities before transfer; tachypnoea (43.7% vs 28.6%), hypoxaemia (63.5% vs 48.4%) and severe hypoxaemia (27.6% vs 14.5%). There were no differences in rates of admission, cardiac arrest, Medical Emergency Team activation or ICU admission. For admitted patients, those with LOMT orders had significantly (p≤0.005) higher mortality: in-hospital (21.9% vs 11.3%); 30 days (23.9% vs 12.3%) and 60 days (28.2% vs 13.4%).

CONCLUSIONS: Patients with LOMT had higher levels of comorbidity and were more acutely ill during their sub-acute care admission. Once transferred those with a LOMT had similar rates of cardiac arrest, MET activation and unplanned ICU admission, but higher mortality.