Stabilizing high-dimensional prediction models using feature graphs

We investigate feature stability in the context of clinical prognosis derived from high-dimensional electronic medical records. To reduce variance in the selected features that are predictive, we introduce Laplacian-based regularization into a regression model. The Laplacian is derived on a feature graph that captures both the temporal and hierarchic relations between hospital events, diseases, and interventions. Using a cohort of patients with heart failure, we demonstrate better feature stability and goodness-of-fit through feature graph stabilization.

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.

A discrete choice experiment to examine the preferences of patients with cancer and their willingness to pay for different types of health care appointments

BACKGROUND: This study sought to understand the preferences of patients with cancer and the trade-offs between appointment attributes using discrete choice experiment (DCE). METHODS AND STUDY DESIGN: Patients with cancer at 3 hospitals completed a self-administered DCE. Each scenario described 6 attributes: expertise of health care professionals (HCPs), familiarity of doctors with patients' medical history, waiting time, accompaniment by family/friends, travel time, and out-of-pocket costs. Patient preferences were estimated using logistic regression. Willingness to pay (WTP) estimates were derived from regression coefficients. RESULTS: Of 512 patients contacted, 185 returned the questionnaire. The mean age was 61 years, and 60% of respondents were female. The mean time since cancer diagnosis was 34 months, 90% had received treatment; and 61% had early-stage disease. The most important attributes were expertise and familiarity of doctors with patients' medical history; distance traveled was least likely to influence patient preferences. The WTP analysis estimated that patients were willing to pay $680 (95% CI, 470-891) for an appointment with a specialist, $571 (95% CI, 388-754) for doctors familiar with their history, $422 (95% CI, 262-582) for shorter waiting times, $399 (95% CI, 249-549) to be accompanied by family/friends, and $301 (95% CI, 162-441) for shorter traveling times. Male patients had a stronger preference for accompaniment by family/friends. The expertise of HCP was the most important attribute for patients regardless of geographic remoteness. CONCLUSIONS: Our study can assist the development of patient-centered health care models that improve patient access to experienced HCPs, support the role of primary care providers during the cancer journey, and educate patients about the roles of non-oncology HCPs to cope with increasing demand for cancer care.

The antecedents of health professionals' service recovery performance

 The aim of this thesis is to identify the antecedents to the self-perceived recovery performance of health professionals in Qatar. The quantitative phase of the thesis utilised a survey instrument, which was developed and modified as a result of consultations with experts and piloting. A total of 343 questionnaires were completed and returned by respondents. After determining the adequacy of the congeneric models and the overall measurement model using confirmatory factor analysis (CFA), structural analysis was undertaken to determine the fit of the data to the hypothesised associations between the latent constructs of interest.

The association between self-reported diet quality and health-related quality of life in rural and urban Australian adolescents

OBJECTIVE: This study examines the relationship between diet quality and health-related quality of life (HRQoL) in rural and urban Australian adolescents, and gender differences.

DESIGN: Cross-sectional.

SETTING: Secondary schools.

PARTICIPANTS: 722 rural and 422 urban students from 19 secondary schools.

MAIN OUTCOME MEASURES: Self-report dietary-related behaviours, demographic information, HRQoL (AQoL-6D) were collected. Healthy and unhealthy diet quality scores were calculated; multiple linear regression investigated associations between diet quality and HRQoL.

RESULTS: Compared to urban students, rural students had higher HRQoL, higher healthy diet score, lower unhealthy diet score, consumed less soft drink and less frequently, less takeaway and a higher proportion consumed breakfast (P < 0.05). Overall, males had higher unhealthy diet score, poorer dietary behaviours but a higher HRQoL score compared to females (P < 0.05). In all students, final regression models indicated: a unit increase in healthy diet score was associated with an increase in HRQoL (unstandardised coefficient(B)±standard error(SE); B = 0.02 ± 0.01(SE); P < 0.02); and a unit increase in unhealthy diet scores was associated with a decrease in HRQoL (-0.01 ± 0.00; P < 0.05). In rural students alone, a unit increase in unhealthy diet score was associated with a decrease in HRQoL (B = -0.01 ± 0.00; P = 0.002), and in urban students a unit increase in healthy diet score was associated with an increase in HRQoL (B = 0.02 ± 0.00; P < 0.001).

CONCLUSIONS: Cross-sectional associations between diet quality and HRQoL were observed. Dietary modification may offer a target to improve HRQoL and general well-being; and consequently the prevention and treatment of adolescent health problems. Such interventions should consider gender and locality.

Hierarchical Bayesian nonparametric models for knowledge discovery from electronic medical records

Electronic Medical Record (EMR) has established itself as a valuable resource for large scale analysis of health data. A hospital EMR dataset typically consists of medical records of hospitalized patients. A medical record contains diagnostic information (diagnosis codes), procedures performed (procedure codes) and admission details. Traditional topic models, such as latent Dirichlet allocation (LDA) and hierarchical Dirichlet process (HDP), can be employed to discover disease topics from EMR data by treating patients as documents and diagnosis codes as words. This topic modeling helps to understand the constitution of patient diseases and offers a tool for better planning of treatment. In this paper, we propose a novel and flexible hierarchical Bayesian nonparametric model, the word distance dependent Chinese restaurant franchise (wddCRF), which incorporates word-to-word distances to discover semantically-coherent disease topics. We are motivated by the fact that diagnosis codes are connected in the form of ICD-10 tree structure which presents semantic relationships between codes. We exploit a decay function to incorporate distances between words at the bottom level of wddCRF. Efficient inference is derived for the wddCRF by using MCMC technique. Furthermore, since procedure codes are often correlated with diagnosis codes, we develop the correspondence wddCRF (Corr-wddCRF) to explore conditional relationships of procedure codes for a given disease pattern. Efficient collapsed Gibbs sampling is derived for the Corr-wddCRF. We evaluate the proposed models on two real-world medical datasets - PolyVascular disease and Acute Myocardial Infarction disease. We demonstrate that the Corr-wddCRF model discovers more coherent topics than the Corr-HDP. We also use disease topic proportions as new features and show that using features from the Corr-wddCRF outperforms the baselines on 14-days readmission prediction. Beside these, the prediction for procedure codes based on the Corr-wddCRF also shows considerable accuracy.

A square peg in a round hole? Approaches to incorporating lifestyle counselling into routine primary health care

Few studies have explored how lifestyle counselling can be integrated into routine practice for primary health care (PHC) clinicians working outside general practice. This paper describes the feasibility of models of lifestyle counselling developed for PHC clinicians working in community health services and the congruence with routine practice. Action research methods were used to develop and implement models of lifestyle counselling in three community health teams. Following a six-month implementation period, semi-structured interviews were conducted with a purposeful sample of participants (n = 30) to explore the appropriateness of implementing risk factor management models in practice. Models were considered appropriate if they fitted the clinician’s philosophy of practice, were relevant to existing work tasks, could easily be integrated into workflow and were perceived as being acceptable to the client. The approach to service delivery and team priorities were also important in influencing which models suited particular teams. Models of lifestyle counselling for PHC clinicians outside general practice should be tailored to the clinicians’ and teams’ way of working and thus may need to be discipline-specific. Engaging PHC clinicians and teams is important in developing models that are acceptable and feasible in everyday practice.

Explaining the variation in the management of lifestyle risk factors in primary health care: a multilevel cross sectional study

Background

Despite evidence for the effectiveness of interventions to modify lifestyle behaviours in the primary health care (PHC) setting, assessment and intervention for these behaviours remains low in routine practice. Little is known about the relative importance of various determinants of practice.

This study aimed to examine the relative importance of provider characteristics and attitudes, patient characteristics and consultation factors in determining the rate of assessment and intervention for lifestyle risk factors in PHC.

Methods

A prospective audit of assessment and intervention for lifestyle risk factors was undertaken by PHC nurses and allied health providers (n = 57) for all patients seen (n = 732) over a two week period. Providers completed a survey to assess key attitudes related to addressing lifestyle issues. Multi-level logistic regression analysis of patient audit records was undertaken. Associations between variables from both data sources were examined, together with the variance explained by patient and consultation (level 1) and provider (level 2) factors.

Results

There was significant variance between providers in the assessment and intervention for lifestyle risk factors. The consultation type and reason for the visit were the most important in explaining the variation in assessment practices, however these factors along with patient and provider variables accounted for less than 20% of the variance. In contrast, multi-level models showed that provider factors were most important in explaining the variance in intervention practices, in particular, the location of the team in which providers worked (urban or rural) and provider perceptions of their effectiveness and accessibility of support services. After controlling for provider variables, patients' socio-economic status, the reason for the visit and providers' perceptions of the 'appropriateness' of addressing risk factors in the consultation were all significantly associated with providing optimal intervention. Together, measured patient consultation and provider variables accounted for most (80%) of the variation in intervention practices between providers.

Conclusion

The findings highlight the importance of provider factors such as beliefs and attitudes, team location and work context in understanding variations in the provision of lifestyle intervention in PHC. Further studies of this type are required to identify variables that improve the proportion of variance explained in assessment practices.

Is change in global self-rated health associated with change in affiliation with a primary care provider? Findings from a longitudinal study from New Zealand

To investigate the association of self-rated health and affiliation with a primary care provider (PCP) in New Zealand.
Methods

We used data from a New Zealand panel study of 22,000 adults. The main exposure was self-rated health, and the main outcome measure was affiliation with a PCP. Fixed effects conditional logistic models were used to control for observed time-varying and unobserved time-invariant confounding.
Results

In any given wave, the odds of being affiliated with a PCP were higher for those in good and fair/poor health relative to those in excellent health. While affiliation for Europeans increased as reported health declined, the odds of being affiliated were lower for Māori respondents reporting very good or good health relative to those in excellent health. No significant differences in the association by age or gender were observed.
Conclusions

Our data support the hypothesis that those in poorer health are more likely to be affiliated with a PCP. Variations in affiliation for Māori could arise for several reasons, including differences in care-seeking behaviour and perceived need of care. It may also mean that the message about the benefits of primary health care is not getting through equally to all population groups.

Factor structure and measurement invariance of the Health Education Impact Questionnaire: does the subjectivity of the response perspective threaten the contextual validity of inferences?

Objective:
On-going evidence is required to support the validity of inferences about change and group differences in the
evaluation of health programs, particularly when self-report scales requiring substantial subjectivity in response generation are used as outcome measures. Following this reasoning, the aim of this study was to replicate the factor structure and investigate the measurement invariance of the latest version of the Health Education Impact Questionnaire, a widely used health program evaluation measure.
Methods:
An archived dataset of responses to the most recent version of the English-language Health Education Impact
Questionnaire that uses four rather than six response options (N=3221) was analysed using exploratory structural equation
modelling and confirmatory factor analysis appropriate for ordered categorical data. Metric and scalar invariance were
studied following recent recommendations in the literature to apply fully invariant unconditional models with minimum
constraints necessary for model identification.
Results:
The original eight-factor structure was replicated and all but one of the scales (Self Monitoring and Insight) was
found to consist of unifactorial items with reliability of ⩾0.8 and satisfactory discriminant validity. Configural, metric and scalar
invariance were established across pre-test to post-test and population sub-groups (sex, age, education, ethnic background).
Conclusion:
The results support the high level of interest in the Health Education Impact Questionnaire, particularly for use as a pre-test/post-test measure in experimental studies, other pre–post evaluation designs and system-level monitoring and evaluation.

Economic models of preventive dentistry for Australian children and adolescents

Purpose:
To identify economic evaluation models and parameters that could be replicated or adapted to construct a generic model to assess cost-effectiveness of and prioritise a wide range of community-based oral disease prevention programmes in an Australian context.
Methods:
The literature search was conducted using MEDLINE, ERIC, PsycINFO, CINHAL (EBSCOhost), EMBASE (Ovid), CRD, DARE, NHSEED, HTA, all databases in the Cochrane library, Scopus and ScienceDirect databases from their inception to November 2012.
Results:
Thirty-three articles met the criteria for inclusion in this review (7 were Australian studies, 26 articles were international). Existing models focused primarily on dental caries. Periodontal disease, another common oral health problem, was lacking. Among caries prevention studies, there was an absence of clear evidence showing continuous benefits from primary through to permanent dentition and the long-term effects of oral health promotion.
Conclusion:
No generic model was identified from previous studies that could be immediately adopted or adapted for our purposes of simulating and prioritising a diverse range of oral health interventions for Australian children and adolescents. Nevertheless, data sources specified in the existing Australian-based models will be useful for developing a generic model for such purposes.

Cycling as a part of daily life: A review of health perspectives

Health aspects of day-to-day cycling have gained attention from the health sector aiming to increase levels of physical activity, and from the transport and planning sector, to justify investments in cycling. We review and discuss the main pathways between cycling and health under two perspectives — generalizable epidemiological evidence for health effects and specific impact modeling to quantify health impacts in concrete settings. Substantial benefits from physical activity dominate the public health impacts of cycling. Epidemiological evidence is strong and impact modeling is well advanced. Injuries amount to a smaller impact on the population level, but affect crash victims disproportionately and perceived risks deter potential cyclists. Basic data on crash risks are available, but evidence on determinants of risks is limited and impact models are highly dependent on local factors. Risks from air pollution can be assumed to be small, with limited evidence for cycling-specific mechanisms. Based on a large body of evidence, planners, health professionals, and decision-makers can rest assured that benefits from cycling-related physical activity are worth pursuing. Safety improvements should be part of the efforts to promote cycling, both to minimize negative impacts and to lower barriers to cycling for potential riders.

Measurement invariance of English and French Health Education Impact Questionnaire (heiQ) empowerment scales validated for cancer

PURPOSE: If measurement invariance (MI) is demonstrated for a scale completed by respondents from two different language groups, it means that the scale measures the same construct in the same way in both groups. We assessed MI of the French- and English-language versions of the five Health Education Impact Questionnaire (heiQ) empowerment scales validated for the cancer setting. METHODS: Data came from two cross-sectional studies of Canadian cancer survivors (704 English, 520 French). Single-group confirmatory factor analysis (CFA) was used to test whether the hypothesized factor structure of the French-language heiQ empowerment scales fit the data. Multi-group CFAs were conducted to assess different levels of MI conditions (configural, metric, scalar, strict, as well as MI of factor variances, covariances, and latent means) of the French- and English-language heiQ empowerment scales. RESULTS: The correlated five-factor model showed good fit in both language groups (goodness-of-fit indices: CFI ≥ .97; RMSEA ≤ .07). Goodness-of-fit indices and tests of differences in fit between models supported MI of the five-factor model across the two language groups (∆CFI ≤ -.010 combined with ∆RMSEA ≤ .015). CONCLUSIONS: The French- and English-language heiQ empowerment scales measure the same five dimensions of empowerment in the same way across both language groups. Thus, any observed similarities or differences between French- and English-speaking respondents completing these scales are valid and reflect similarities or differences in empowerment across language groups, not measurement artifact. Consequently, heiQ empowerment data from English- and French-speaking respondents can be directly pooled or contrasted in data analyses.

Housing tenure and affordability and mental health following disability acquisition in adulthood

Acquiring a disability in adulthood is associated with a reduction in mental health and access to secure and affordable housing is associated with better mental health. We hypothesised that the association between acquisition of disability and mental health is modified by housing tenure and affordability. We used twelve annual waves of data (2001-2012) (1913 participants, 13,037 observations) from the Household, Income and Labour Dynamics in Australia survey. Eligible participants reported at least two consecutive waves of disability preceded by two consecutive waves without disability. Effect measure modification, on the additive scale, was tested in three fixed-effects linear regression models (which remove time-invariant confounding) which included a cross-product term between disability and prior housing circumstances: housing tenure by disability; housing affordability by disability and, in a sub-sample (896 participants 5913 observations) with housing costs, tenure/affordability by disability. The outcome was the continuous mental component summary (MCS) of SF-36. Models adjusted for time-varying confounders. There was statistical evidence that prior housing modified the effect of disability acquisition on mental health. Our findings suggested that those in affordable housing had a -1.7 point deterioration in MCS (95% CI -2.1, -1.3) following disability acquisition and those in unaffordable housing had a -4.2 point reduction (95% CI -5.2, -1.4). Among people with housing costs, the largest declines in MCS were for people with unaffordable mortgages (-5.3, 95% CI -8.8, -1.9) and private renters in unaffordable housing (-4.0, 95% CI -6.3, -1.6), compared to a -1.4 reduction (95% CI -2.1, -0.7) for mortgagors in affordable housing. In sum, we used causally-robust fixed-effects regression and showed that deterioration in mental health following disability acquisition is modified by prior housing circumstance with the largest negative associations found for those in unaffordable housing. Future research should test whether providing secure, affordable housing when people acquire a disability prevents deterioration in mental health.

Health-related quality of life and willingness to pay per quality-adjusted life-Year threshold—A study in patients with epilepsy in China

Objectives
To assess the health-related quality of life (HRQOL) and willingness to pay (WTP) per quality-adjusted life-year (QALY) amount of patients with epilepsy in China.
Methods
Adults with epilepsy and a healthy control were recruited in two tertiary hospitals in China. Participants completed two indirect utility elicitation instruments (Quality of Well-being Scale-self administered version and EuroQol five-dimensional questionnaire) and a WTP questionnaire. Correlations between sociodemographic or epilepsy-specific variables (age of epilepsy onset, duration of epilepsy, seizure types, types of antiepileptic drug treatment, etc.) and HRQOL or WTP/QALY were assessed to identify the candidate predictor. Multiple linear regression models were adopted to investigate the predictive performances of identified candidate predictors. Data analyses were performed on SPSS 20.0 (SPSS, Inc., Chicago, IL).
Results
For utilities of both the Quality of Well-being Scale-self administered version and the EuroQol five-dimensional questionnaire, patients with epilepsy had statistically lower values than did the control group (P < 0.0001). In terms of the WTP/month, the percentage of WTP accounting for the monthly income and the WTP/QALY values from the epilepsy group were substantially higher than those from the control group (P < 0.0001).
View the MathML sourceWTP/QALY=12×WTPMonth1−Utility(CurrentHealth)
Turn MathJax on
The multiple linear regression model identified working status (P = 0.05), seizure types (P = 0.022), income (P = 0.006), and self-rating health state (P < 0.05) as predictors of HRQOL while income (P = 0.000) and self-rating health state (P < 0.05) statistically contributed to the variations in WTP/QALY value for the epilepsy group.
Conclusions
Patients with epilepsy had substantially lower HRQOL than did the healthy population.