Australian Health Economics Society

Affiliated with iHEA

Welcome to AHES

Welcome to the website for the Australian Health Economics Society Inc.

The field of health economics in Australia is dynamic and growing. Not only does it attract a great deal of interest and attention from Universities, Governments, and public and private funders and providers of services, it also has a key role in contributing to health policy and evidence based practice in the health system. Health economics researchers and practitioners in Australia have a wide range of disciplinary backgrounds and this diversity is reflected in the Australian Health Economics Society.

Although it is a relatively young academic field, health economics has a very significant history in Australia. AHES was established in the late 1970s through the efforts of economists at the Australian National University, most notably Professor John Deeble. After a one day conference for economists interested in health, the Australian Health Economists Group was formed, which went on to become the Australian Health Economics Society in 1984. This makes AHES one of the longest established health economics associations internationally.  Even before that, health economists had had a significant impact on the landscape of health policy in Australia, with a seminal paper by John Deeble and Dick Scotton published in 1968 in the Australian Economic Review.  This paper was fundamental in the formation of Medibank, the original architecture of Australia’s universal health insurance scheme. The current DRG system and hospital funding arrangements, as well as the key role of health technology assessment in reimbursement decisions through PBAC and MSAC are other areas where the contribution of health economists has been very important.

Health economics in Australia is practiced and taught in a growing number of locations and settings, including government, universities, health service providers, consultancy firms and health-related industry. Please see our Education section for information about courses and our Research section for more information on health economics research centres in Australia.

The field of health economics has also become much broader over the past few decades. As a fundamental component of an efficient and effective health care system, health economics research and analysis now makes important contributions to policy, planning and cost effective care by:

  • Adding valuable information about factors that contribute to and explain health behaviours, health care choices and the health and wellbeing outcomes of the population;
  • Enhancing other health research via concomitant research in the understanding of factors that drive individual decisions in health, on the best ways to fund and provide new services, and on understanding the health system architecture, including funding and delivery arrangements and incentives for efficient and equitable provision of services.

Health economics draws on economic theory, applied micro-econometrics, applied welfare economics, behavioural economics, epidemiology and pharmacoeconomics. Practitioners of health economics include theoretical and applied economists, econometricians as well as researchers and clinicians who undertake health services research. AHES has also grown in membership, with a growing number of Australian and international health economics PhD graduates now choosing to engagein research and practice in Australia.  This is recognised in the Australian Health Economics Research Prize which is awarded bi-annually to the best paper in health economics by Australian based researchers.


The Effect of Personality Traits on Subject Choice and Performance in High School: Evidence from an English Cohort
Authors: Silvia Mendolia, Ian Walker

This paper investigates the relationship between personality traits in adolescence and performance in high school using a large and recent cohort study. In particular, we investigate the impact of locus of control, self-esteem and work ethics at age 15, on test scores at age 16, and on subject choices and subsequent performance at age 17-18. In particular, individuals with external locus of control or with low levels of self-esteem seem less likely to have good performance in test scores at age 16 and to pursue further studies at 17-18, especially in mathematics or sciences.

We use matching methods to control for a rich set of adolescent and family characteristics and we find that personality traits do affect study choices and performance in test scores – particularly in mathematics and science. The results are stronger for adolescents from disadvantaged socio-economic backgrounds. We establish the robustness of our results using the methodology proposed by Altonji et al. (2005) that consists in making hypotheses as to the correlation between the unobservables that determine test scores and subjects’ choices and, the unobservables that influence personality.

Why do multi attribute utility instruments produce different utilities: The relative importance of the descriptive systems, scale and 'micro utility' effects.
Authors: Jeff Richardson, Angelo Iezzi, Munir A Khan

Different multi attribute utility (MAU) instruments produce different utilities for the same health state. This is necessarily a result of differences in the descriptive systems and differences in the utility formula which convert item responses to an index of utility. The utility formulae have two effects. First, they impose an implicit scale upon the data and, secondly, they vary utilities in accordance with item parameters and the interaction of the parameters: an amount which is referred to here as the ‘micro utility effect’. The present paper investigates the relative importance of the three components – the descriptive system, scale and micro utility effects – in each of the pairwise comparisons between five multi attribute utility instruments: viz, EQ-5D, SF-6D, HUI 3, 15D and AQoL-8D.

The five MAU instruments were administered to a sample of 8,022 patients and the public in six countries. Three sets of data were produced for each instrument. The first was obtained by applying the published utility formula to obtain ‘utilities’. The second, an unweighted ‘score’, was calculated from item responses. The third was obtained by applying a linear transformation to the scores to obtain ‘values’ which are on the same scale as the corresponding instrument utilities. For each pairwise comparison of instruments, scale effects were determined by the linear relationship between utilities, the effect of the descriptive system by a comparison of scale adjusted values and the ‘micro utility effects’ by the amount which would convert the difference in values to the difference in utilities

The relationship between health-related quality of life, subjective wellbeing and utility in six countries
Authors: Jeff Richardson, Gang Chen, Munir A Khan, Angelo Iezzi

The upsurge of interest in Subjective Wellbeing (SWB) in the past two decades has led to the advocacy of its use in economic evaluation studies to replace or supplement utility. However the relationship between SWB and utility, as usually measured, remains opaque and it is unclear whether utility and SWB are relatively unrelated constructs or whether the low correlation between them is a result of the use of particular instruments. This paper has four objectives relating to this theme. These are:

i. To document the correlation between utilities predicted by six multi attribute utility (MAU) instruments and four measures of SWB
ii. To explain the ‘content’ of both sets of instruments using the dimensions of the SF-36
iii. To determine the extent to which the reduction in SWB associated with 6 diseases is taken into account (explained) by variation in utility as measured by the EQ-5D, SF-6D, HUI 3, 15D, QWB and AQoL-8D multi attribute utility instruments. .
iv. To determine the extent to which the reduction in utility associated with these diseases is taken into account by variation in subjective wellbeing.

Data were obtained from 8,022 individuals in six countries who completed all of the instruments.

Investigating older people and informal carer’s views and preferences for CDC: A Discrete Choice Experiment
Authors: Billingsley Kaambwa; Emily Lancsar, Nikki McCaffrey, Liz Gill, Ian D Cameron, Maria Crotty, Julie Ratcliffe

Consumer directed care (CDC) is currently being embraced within Australia and internationally as a means to promote autonomy and choice for older people receiving community aged care services (CACSs). CDC involves giving CACS recipients (older people and informal carers of older people) control over how CACSs are administered. However, CDC models have largely developed in the absence of evidence on recipients’ views and preferences. We explored CACS recipients’ preferences for a variety of CDC attributes and identified factors that may influence these preferences, and in turn can be used to inform improved design of future CDC models.

Study participants were recipients of CACSs delivered by five Australian providers. Using a discrete choice experiment (DCE) approach, we investigated which of six CDC attributes (choice of service provider(s), budget management, saving unused/unspent funds, choice of support/care worker(s), support-worker flexibility and level of contact with service coordinator) were important to CACS recipients. The DCE data were analysed using firstly, a mixed logit regression model, which accounts for preference heterogeneity, and secondly using the generalised multinomial logit model, which accounts for preference and scale heterogeneity simultaneously and results compared.

Mean ages for 117 study participants were 80 years (87 older people) and 74 years (30 carers). In order of strength of preference, all participants preferred a CDC approach that allowed them to: save unused funds from a CACS package for future use; have support workers that were flexible in terms of changing activities on the CACS care plan and; choose the support workers that provide their day-to-day CACSs.

DCEs are practical and feasible in populations of older people with varying levels of quality of life status and help to assess preferences for elements of potential future CACSs delivery not yet available in policy.

Real-world observational data in cost-effectiveness analyses: Herceptin as a case study
Authors: Ms Bonny Parkinson, Professor Rosalie Viney, Associate Professor Stephen Goodall, Professor Marion Haas

Introduction: A challenge for decision makers is the potential gap between the estimated cost-effectiveness at the time of funding and real-world cost-effectiveness. There is also increasing pressure to make new drugs available earlier, with decisions based on limited data. Trastuzumab (Herceptin) is a high cost drug with trials reporting increased survival in patients with HER2+ metastatic breast cancer. However, the cost-effectiveness of trastuzumab remains controversial. In Australia the Pharmaceutical Benefits Advisory Committee (PBAC) did not recommend trastuzumab for public funding. Despite this, in 2001 the Government established the Herceptin Program, which resulted in a rich observational dataset.

Aims: To compare the predicted cost-effectiveness of trastuzumab based on trial evidence with real-world cost effectiveness, and to estimate the value of collecting the observational data in the context of a managed entry scheme (MES).

Methods: 1) an economic model was developed using trial evidence and analysed using value of information methods; 2) the observational data was analysed to inform modifications to model parameters; 3) the economic model was adjusted to estimate the real-world cost-effectiveness.

Results: At its current price, the cost-effectiveness of trastuzumab was estimated to be $222,975/QALY, whereas the real-world cost-effectiveness was estimated to be $264,452/QALY. If no price discount had been applied in 2001, the population Expected Value of Perfect Parameter Information (EVPPI) of collecting the observational data was nil. However, if a temporary price discount of 95% had been applied, the population EVPPI was around $70,000 (assuming a threshold of $80,000/QALY).

Conclusion: The real-world cost-effectiveness of trastuzumab is above the range usually considered by the PBAC as being acceptable for public funding. This case study demonstrates the challenges in designing interim funding arrangements such as a MESs.

Changing the game in health economics
Authors: Brita Pekarsky

Non-strategic decision models dominate health technology assessment and the new technology and play a significant role is informing new technology reimbursement decisions in counties such as Australia, Canada, New Zealand and the UK. This role is largely a consequence of non-strategic decision models’ capacity to accommodate, synthesise and analyse the wide ranging evidence that is required to support claims of incremental cost and effect of new technologies. Detailed jurisdictional protocols set out the supporting evidence that must be provided to the decision maker in order for claims to be evaluated. One critical input is not supported by rigorous evidence — the price of the new drug. This omission matters when this price is selected strategically by monopolist firms — the possibility that the firm could sell the drug at a lower price, profitably, cannot be excluded by the largely administrative price data available to decision maker.

This paper outlines the advantages of using games rather than non-strategic decision models to analyse the economics of the drug price. It also presents a technique for health economists to extract the key elements of the economic problem of interest providing a context within which health economists and game theorists can work to develop pertinent and practical games. This technique uses the paradoxes, threats and mistakes that feature in the narratives that surround one of the most contested political economies of health: the price of a new drug.

Timing and efficiency for the collection of health–related quality of life data
Authors: Kim Dalziel, Chris Schilling, Philip Clarke

Health Economists often face the question about how often to collect health-related quality of life data in cohort studies or randomised controlled trials. Health-related quality of life data are collected to provide information on disease burden/wellbeing over time or impact of treatment effect. There are potentially trade-offs in terms of the information collected versus the costs of collection, including the opportunity cost of collecting alternative data. The purpose of this study is to explore this issue using three data sets including chronic disease (diabetes), outcomes following surgery and a longitudinal study (HILDA).

All of these datasets currently collect quality of life data using the EQ5D, SF-36, or SF-12 with 1 or 2 yearly intervals and follow up periods ranging between 5 and 11 years. The data from health-related quality of life questionnaires will be progressively removed in panels to lengthen the period over which data is collected. We will examine how removal of quality of life panels impacts on the mean and variances of other outcome metrics for example the quality of life gain from a successful surgical outcome or the impact on clinical measures. These will be estimated using fixed effects panel regression models.

Data collection poses a significant cost to those conducting research and participants. Value for effort and cost expended need to be considered. The paper will conclude by outlining implications for researchers in regard to protocols for collecting health-related quality of life data in future studies to maximise efficiency.

The mind-body connection: Validating subjective reports of poor health against objective physiological dysregulations
Authors: Claryn Kung; David Johnston; Michael Shields

Recurring socioeconomic surveys often include simple subjective wellbeing measures, on which a large body of economic literature relies when studying the determinants of human wellbeing. Despite their importance, little is known about the information contained in these measures. To better understand this concept, several economists have explored the validity of these subjective reports against observables ranging from self-reported lifestyle habits, to physician-prescribed medications, to objective physiological indicators (‘biomarkers’). This paper aims to contribute to this growing literature by assessing whether reports of poor mental health in response to stressful life events (‘shocks’) are replicable using biomarker levels, in terms of deviations that reflect stress-related physiological dysregulations. Guided by the allostatic load framework (bodily ‘wear and tear’ from chronic stress), we select four biomarkers: cortisol, a primary mediator of stress-related dysregulations; and glycosylated hemoglobin, blood pressure, and c-reactive protein, which are metabolic, cardiovascular, and inflammatory parameters, respectively, that are involved in secondary adaptations to chronic stress (or prolonged cortisol secretion).

We employ the National Child Development Study, a prospective birth cohort study that allows us to not only assess raw correlations but also match respondents on their propensity to experience shocks that are likely to affect their mental health. Propensity scores are predicted from extensive mental health history and labour market history variables, alongside pre-shock demographic characteristics. Conditional on past mental health and economic background, economic shocks predict poorer self-reported health as well as depressive and anxiety symptoms, whereas social shocks only predict the latter. Both types of shocks also predict allostatic biomarker levels in expected directions, but there is no clear correspondence between reportedly poor mental health and associated physiological dysregulations. Although caution is required when interpreting these results, they suggest that subjective reports of health do not necessarily reflect objective physiological dysregulations.

Refugee populations and Emergency Department use in the Adelaide metropolitan area: Examining outcomes and costs using linked, administrative data
Authors: David Banham; Anh-Minh Thi Nguyen; Kay Anastassiadis;

Potentially Preventable Hospitalisations (PPHs) are an indicator of primary health care availability and effectiveness. While PPHs often focus on inpatient stays they are also relevant to emergency department (ED) visits. Specific population groups are vulnerable to higher PPHs. Anecdotally, refugee and asylum seekers (RAS) are one such group. Systematic disparities in hospital contacts have implications for people’s health outcomes and system costs and thus require scrutiny.
This presentation collates empiric evidence of RAS contact with ED services. In particular, the presentation:
• Describes changes in the underlying RAS population and volume of ED contacts, including PPHs;
• Enumerates individuals experiencing these contacts and the frequency of their contact; and
• Estimates the costs associated with any excess presentations.

Countries of birth associated with humanitarian arrivals provide RAS population numbers. Individually linked public hospital records in the Adelaide metropolitan area for 2005-06 to 2010-11 are categorised using Australian Institute of Health and Welfare definitions for PPHs and GP-type presentations. The Pricing Framework for Australian Public Hospital Services 2014-15 provides cost weights. The main outcome measures used are proportions and crude rates of PPHs and costs by geographic region.

Preliminary analyses indicate a doubling of RAS contribution to total ED activity over time. This is not associated with substantive change in ED visit rates per capita. However, results do suggest visiting individuals are less likely to have one presentation and increasingly likely to have three or more ED visits annually.

Compared with the wider community, RAS within Greater Adelaide appear to have poorer utilisation of effective primary care resulting in PPHs to EDs. This indicates an emerging issue of health inequality and cost pressure. The implications of these findings for commissioning services that meet population need in effective, efficient and equitable ways is discussed.

Violent crime, life satisfaction, response heterogeneity and compensation estimates
The dynamics of income-related health inequality across the lifecycle
Authors: Dennis Petrie; Paul Allanson; Paul Calara; Philip Clarke; Ulf Gerdtham

In general, the positive relationship between health and income strengthens at young ages before reaching a peak at middle ages and then weakens at old ages. However, there is limited evidence about what features of the co-evolution of health and income over the life cycle give rise to this age profile in the income gradient in health. In this paper we uncover the changing nature of the inter-dependence between age and income over the life-cycle using recently developed decomposition techniques. We examine the changes in income-related health inequality for rolling age groups by sex for Great Britain (1999-2004). We find that at younger ages health-related income mobility plays the major role with those with poor health falling behind their peers in terms of income growth over time. At older ages income-related morbidity mobility starts to set in with the poor losing health more quickly than the rich but at even older ages income-related mortality plays the major role with the poor more likely to die than the rich though this has a side-effect of weakening the observed positive relationship between income and health for those who remain alive.

Talking about contraception: What doctors decide to discuss with women
Authors: Denzil G. Fiebig; Rosalie Viney; Stephanie Knox; Marion Haas; Deborah Street; Edith Weisberg; Deborah Bateson
What features of ‘consumer-directed’ home-based support services are important to older Australians and their informal carers – development of attributes for a discrete choice experiment
Authors: Dr Nikki McCaffrey; Dr Liz Gill; Dr Billingsley Kaambwa; Prof Ian D Cameron; Emeritus Prof Jan Patterson; Prof Maria Crotty; Prof Julie Ratcliffe

In Australia, newly initiated, publicly subsidised ‘Home Care Packages’ designed to assist older people (≥65 years of age) live in their own home must now be offered on a ‘consumer-directed care’ (CDC) basis by service providers. However, CDC models have largely developed in the absence of evidence on users’ views and preferences. The aim of this study was to determine what features (attributes) of consumer-directed home-based support services are important to older people and their informal carers to inform the design of a discrete choice experiment (DCE). Semi-structured, face-to-face interviews were conducted December 2012-November 2013 with 17 older people receiving home-based support services and 10 informal carers from 5 providers located in South Australia and New South Wales. Salient service characteristics important to participants were determined using thematic and constant comparative analysis and formulated into attributes and attribute levels for presentation within a DCE. Initially eight broad themes were identified: information and knowledge, choice and control, self-managed continuum, effective co-ordination, effective communication, responsiveness and flexibility, continuity, and planning. Attributes were formulated for the DCE by combining overlapping themes such as effective communication and co-ordination, and the self-managed continuum and planning into single attributes. Six salient service features that characterise consumer preferences for the provision of home-based support service models were identified: choice of provider, choice of support worker, flexibility in care activities provided, contact with the service coordinator, managing the budget and saving unspent funds. Best practice indicates that qualitative research with individuals who represent the population of interest should guide attribute selection for a DCE and this is the first study to employ such methods in aged care service provision. Further development of services could incorporate methods of consumer engagement such as DCEs which facilitate the identification and quantification of users’ views and preferences on alternative models of delivery.

The use of multi-attribute utility instruments in paediatric populations
Authors: Gang Chen; Julie Ratcliffe

The measurement and valuation of health forms a major component of economic evaluation in health care and is a major issue in health services research. Health related quality of life (HRQoL) is a multidimensional construct that measures the impact of health or disease on physical and psychosocial functioning. Researchers in health economics and other disciplines are increasingly recognising the importance of the measurement and valuation of HRQoL in both children and adolescents. Presently there are eight multi-attribute utility (MAU) instruments available internationally which have been developed for application in child and/or adolescent samples: the 16D, the 17D, the Health Utilities Index Mark 2 (HUI2), the HUI3, the Assessment of Quality of Life 6-Dimension (AQoL-6D), the Child Health Utility 9D (CHU9D), the European Quality of Life 5 Dimension Youth version (EQ-5D-Y), and the Adolescent Health Utility Measure (AHUM). The paper critically reviewed the development and application of the above eight MAU instruments, discussed the specific challenges of health utility measurement in children/adolescents, compared the pros and cons of the different instruments, and provided some guidance for use of the instruments and areas for further research and instrument development.

Improving the quality of decision making on health interventions: incorporating the long-term economic impacts of informal (unpaid) care
Authors: Deborah Schofield; Rupendra Shrestha; Michelle Cunich; Megan Passey; Emily Callander; Simon Kelly; Robert Tanton; Lennert Veerman; Theo Vos.

Introduction: Currently cost-effectiveness analyses of health interventions significantly underestimate the benefits of interventions, because no methodology exists for accurately incorporating the long-term society-wide productivity costs of informal care.

Aim: To describe a static microsimulation model, called Care&WorkMOD, which will be built to simulate both the current economic costs of informal caregiving in Australia and projecting future costs to 2030.
Methods: Care&WorkMOD will be used to generate snapshots of the prevalence of chronic diseases associated with people leaving the labour force to take on caring responsibilities, demographic information, and economic impacts of informal care based on Australians aged 21-64 years in 2010 every five years from 2010 to 2030. Care&WorkMOD will be built using three databases: i) the base population – i.e. individuals aged 21-64 years with information on their labour force participation, carer status and chronic conditions of self and care recipient from the Australian Bureau of Statistic’s Survey of Disability, Ageing and Caring (SDAC) 2003 and 2009; ii) output datasets from the National Centre for Social and Economic Modelling (NATSEM’s) Australian Population and Policy Simulation Model (APPSIM), which consists of financial information on Australians aged 21-64 years for the years of interest; and iii) data on trends in disease prevalence. These are discussed in the paper.

Conclusions: Microsimulation modelling is a new approach to projecting the costs of informal care for Australia up to 2030; these include changes in labour force participation of carers, personal income, welfare payments, personal savings, personal wealth and superannuation, taxes and national impacts such as reductions in GD related to reduced labour force participation due to caring responsibilities.

Health related quality of life in mothers of children with behaviour problems: a population-based study
Authors: Ha N.D Le; Alisha Gulenc; Lisa Gold; Anna Sakardi; Obioha C Ukoumunne; Jordana Bayer; Melissa Wake; Hariet Hiscock

Background: Child behaviour problems are associated with poor maternal mental health. However, little is known about how maternal health related quality of life (HRQoL) may be affected by child behaviour problems. The aim of the study are: (1) to examine the relationship between mother’s HRQoL and child behavior problems at age 2 years and (2) to investigate whether the relationship between maternal HRQoL and child behaviour problems is independent of maternal mental health.

Design: Cross-sectional survey nested within a population-level, cluster randomised trial (2010–2014) which aims to prevent early child behaviour problems.

Participants: 1160 mothers of eight month old children attending the universal Maternal and Child Health system, in nine Local Government Areas in the state of Victoria, Australia.
Measures: HRQoL was measured using the Assessment of Quality of Life (AQoL6D) and child behaviour was measured using the Child Behavior Checklist (CBCL/1.5-5 years). Maternal mental health was measured using the Depression Anxiety Stress Scale (DASS).

Results: Health related quality of life was lower for mothers with children that had borderline/clinical problems compared to those with children without problems (mean difference -0.14, 95% CI: -0.16 to -0.12 to, p

Cost-effectiveness of intensive chronic condition management for Indigenous people with poorly controlled type 2 diabetes (T2D) by Indigenous Health Workers (IHWs) in remote communities.
Authors: Ha Nguyen, Leonie Segal, Barbara Schmidt, Robyn McDermott

Background: Indigenous Australians experience a disproportionally high burden of diabetes. The “Getting Better at Chronic Care Project” provided intensive chronic condition management for Indigenous adults with poorly controlled T2DM by IHWs in remote Queensland. This paper reports on the implementation costs and cost-effectiveness of the intervention.

Method: The design was a cluster-randomized-controlled-trial with intervention (N=100) and usual care (N=113) groups. Participants were Indigenous adults with poorly controlled T2DM (HbA1c ≥ 8.5%) plus at least one of: chronic obstructive pulmonary disease, coronary heart disease, chronic kidney disease or hypertension. The primary outcome was differential change in HbA1c at 18 months. There were a number secondary outcomes including hospitalisations and use of Medicare services (obtained from administrative data collections) and quality of life measured by the AQoL 4D. Costs of the intervention were derived from project records and covered the costs of the IHWs delivering the service, their training and clinical and operational support from the project team.

Results: The total cost of the project was $2.089 million of which $1.010 million was attributed to delivery of the intervention (the balance being for evaluation related costs). The cost of service delivery was equivalent to about $10,100/per person or an average annual cost of nearly $6,750/person. A preliminary analyses of the outcomes at 18 months, found a non-significantly greater reduction in HbA1c than the control of 0.53% (0.84% vs. 0.31%, p-value=0.12). In terms of hospitalisation rate and quality of life, there were no statistically significantly differential changes in either the intervention and control groups. The intervention thus costs $10,000/person, for a mean 0.53% additional reduction in HbA1c level.

Conclusion: The intervention provided community-based extra support for poorly controlled T2D patients. The impact on HbA1c levels was modest at best and disappointing given the high level of funding. The reasons for the poor result may include the complexity of the patient population; reflected in very high HbA1c levels (mean 10.7% at base line). In addition, the level of implementation was compromised by challenges of recruitment, retention and absenteeism of IHWs. Future interventions may need to look at intervening earlier in disease stage and explore mechanisms to better retain IHWs.

The economics of a GP co-payment – who would pay the price?
Authors: Ian McRae; James Butler
Implications of misclassification errors in empirical studies of adolescent smoking behaviours
Authors: Ijeoma Edoka

Misclassification errors in a dependent variable introduce a downward bias to covariate effects in a binary choice model. Misreporting of smoking behaviours by adolescents has been widely documented; however, the consequences of misclassification errors in empirical studies of adolescent smoking participation have received little attention. This study uses the Health Survey for England (HSE) to investigate the extent and implications of misclassification errors in self-reported smoking behaviours of adolescents aged 11-15 years. Possible solutions to correcting misclassification errors are explored using the modified maximum likelihood estimator (MLE) approach. The HSE contains both a self-reported smoking component and an objective measure of smoking obtained from saliva cotinine assays. Saliva cotinine concentration ≥ 12ng/ml is considered the ‘true’ indicator of adolescent smoking participation against which self-reported smoking participation is compared. The results suggest that smoking is misreported in this age group, resulting in a downwards bias of coefficients and marginal effects in the self-reported smoking participation model. The modified MLE performs moderately well in correcting misclassification when the misclassification probabilities are treated as fixed known parameters, but not as expected when misclassification probabilities are treated as unknown estimable parameters.

Do Financial Incentives trump Clinical Guidance? Hip Replacement in England and Scotland
Authors: Irene Papanicolas; Alistair McGuire

Following devolution in 1999, England and Scotland’s health systems have diverged. One of the major differences that have ensued is the use of payment policies introduced for hospitals, where England adopted an activity based payment mechanism from 2003/4 and Scotland continues to pay by global budgets. This paper investigates the role this financial incentive has played in the selection of treatment across the two countries.

We use Hip Replacement as the treatment of study. Cemented and uncemented prosthesis for Hip Replacement have been around since the 1970s and clinical evidence suggests that both prostheses have comparable rates of success. While the vast majority of Hip Replacements performed in the UK have historically used cemented prostheses, the number of uncemented Hip Replacements performed in both England and Scotland has increased throughout the past decade, albeit at a noticeably different pace. By 2013 in England about half of the total number of Hip Replacements used uncemented prosthesis, compared to about 20% in Scotland. The financial reimbursement for uncemented Hip Replacement in England has been more generous than for its cemented counterpart, despite clinical guidance from the National Institute for Clinical Excellence recommending cemented prostheses. In Scotland, as providers are not reimbursed for cases treated, this financial incentive does not exist.

We use a difference-in-difference estimator to test whether changes in reimbursement across the two countries, as indicated by the change to activity based payment in England had an impact on choice of treatment compared to Scotland. Our results suggest that financial incentives are linked to the faster uptake of the more expensive, uncemented Hip Replacement in England as compared to Scotland.

Child maltreatment, health and socioeconomic disadvantage: pathways and gradients in an intergenerational context
Authors: James C Doidge; Ben Edwards; Daryl Higgins; John W Toumbourou; Suzanne Vassallo; Leonie Segal

Child maltreatment is implicated as both cause and consequence of a wide range of poor health and socioeconomic outcomes. We aimed to explore the role of child maltreatment as a mediator in the intergenerational transmission of disadvantage and poor health, and as a driver of the socioeconomic gradient in health. Using data from a large Australian birth cohort, the Australian Temperament Project (n = 2,443), we first looked at risk factors for child maltreatment and outcomes in early adulthood. After accounting for missing data using a combination of inverse probability weighting and multiple imputation, relationships between these variables were explored using logistic and linear regressions. Risk factors exhibited associations with child maltreatment consistent with established theory and literature. The strongest were indicators of parental mental illness and substance use. Adverse outcomes were demonstrated across health, social and economic domains.

Most research is only able to examine a small part of the complex intergenerational cycle of disadvantage, child maltreatment and poor outcomes. By using a broad range of data, collected over a long period of time from multiple generations of the same families, we are able to examine statistical relationships within the intergenerational context that they occur. Our findings demonstrate some of the broader implications of child maltreatment and identify places for intervening to prevent both child maltreatment and its associated intergenerational consequences. We now intend to build on these preliminary results using path modelling and latent variable analyses to combine different elements of our intergenerational model.

Using eye-tracking to investigate attribute non-attendance in discrete choice experiments
Authors: Jean Spinks; Duncan Mortimer

As discrete choice experiments (DCEs) are increasingly used in health research, so too has scrutiny increased of the underlying assumptions upon which the theory is based. Random utility theory assumes that consumers consider all relevant information when making purchasing decisions and that they maximise their utility gain in a ‘rational’ and ‘consistent’ way. Whilst there is increasing evidence that this assumption does not hold in many situations, especially for complex decisions relating to health and health care, little is known about the extent of any deviation and some of the key drivers. Using eye-tracking technology, which captures the number of times and the duration that a participant looks at any part of a computer screen during the completion of a DCE survey, we can analyse what has become known in the DCE literature as ‘attribute non-attendance’ (ANA). ANA can be directly measured in this analysis when no fixations have been recorded for a particular attribute and thus, the information contained within that attribute cannot have been considered. This approach adds to the previous literature which has relied on deduction and inference to measure ANA in DCEs. Using this approach we analyse the effect of choice set complexity and respondent characteristics on the likelihood of ANA using fixed and random effects models to account for repeated choice set completion. We find that complexity is strongly related to ANA as well as some evidence of heterogeneity in decision making processes. We conclude that this approach provides direct evidence that increased complexity in DCE surveys are related to ANA.

Preferences in perinatal mental health: a discrete choice experiment
Authors: Jemimah Ride; Emily Lancsar

As discrete choice experiments (DCEs) are increasingly used in health research, so too has scrutiny increased of the underlying assumptions upon which the theory is based. Random utility theory assumes that consumers consider all relevant information when making purchasing decisions and that they maximise their utility gain in a ‘rational’ and ‘consistent’ way. Whilst there is increasing evidence that this assumption does not hold in many situations, especially for complex decisions relating to health and health care, little is known about the extent of any deviation and some of the key drivers. Using eye-tracking technology, which captures the number of times and the duration that a participant looks at any part of a computer screen during the completion of a DCE survey, we can analyse what has become known in the DCE literature as ‘attribute non-attendance’ (ANA). ANA can be directly measured in this analysis when no fixations have been recorded for a particular attribute and thus, the information contained within that attribute cannot have been considered. This approach adds to the previous literature which has relied on deduction and inference to measure ANA in DCEs. Using this approach we analyse the effect of choice set complexity and respondent characteristics on the likelihood of ANA using fixed and random effects models to account for repeated choice set completion. We find that complexity is strongly related to ANA as well as some evidence of heterogeneity in decision making processes. We conclude that this approach provides direct evidence that increased complexity in DCE surveys are related to ANA.

Preferences for Scanning Technologies to Diagnose Focal Liver Lesions Elicited via a Discrete Choice Experiment
Authors: Jennifer A Whitty; Alexandra Filby; Adam B Smith; Louise M Carr

Objective: This study aimed to investigate the preferences of an “at risk” population for the characteristics of scanning modalities used in diagnosing focal liver lesions.

Methods: A discrete choice experiment was administered to 504 adults in the UK, recruited via an online survey panel. Age and gender reflected the demographics of patients diagnosed with cirrhosis and liver cancer. Respondents made repeated choices between two hypothetical scans, described according to waiting time for scan and results, procedure type, the chance of minor side effects, and whether further scanning procedures were likely to be required. Choice data were analysed using mixed logit models.

Results: Respondents preferred a shorter waiting time for the scan and results, the procedure to be undertaken with a handheld scanner on a couch instead of within a body scanner, no side effects, and no follow up scans (p≤0.01). The average respondent was willing to wait almost an additional 2 weeks to have the scan if it resulted in avoiding side effects, approximately 1.5 weeks to avoid further procedures or to be likely to be told the results immediately, and almost 1 week to have the scan performed on a couch with a handheld scanner. Substantial preference heterogeneity was observed around desirable imaging characteristics.

Conclusions: On average, the general population sub group most likely to require imaging to characterise focal liver lesions in the UK would prefer contrast enhanced ultrasound over magnetic resonance imaging or computed tomography. Insights into the consumer perspective around the differential characteristics of imaging modalities have the potential to be used to help guide recommendations around the use of these technologies.

Financial support: Financial support for this study was provided to York Health Economics Consortium, UK via a contract with the National Institute for Health and Care Excellence (NICE).

The Effects of Socio-Economic and Risk Factors on Cardiovascular and Circulatory Diseases: With Corrections for Co-Morbidity, Endogeneity and Selection Biases
Authors: Jinjing Li; Yohannes Kinfu

The increasing burden of cardiovascular diseases (CVDs) and the search for ways and means to ameliorate their impact have long generated interest in identifying the factors associated with the disease. Traditionally, two approaches dominated the scene: those based on follow up studies (FuS) in a controlled (often clinical) setting and those that used cross-sectional surveys (CSS). However, both approaches have their own shortcomings. FuS are expensive; they focus on a single health condition and have limited generalizability beyond the study population. On the other hand, CSS, while being less costly and have often larger samples, they lack information on disease progression and timing of explanatory variables. In this paper, we exploit the panel component from the Australian Household, Income and Labour Dynamics survey (HILDA) and re-examine the effects of explanatory variables on CVD. The data set, which has a large sample, and contains information on timing of events that mimics a large-scale clinical experiment and on multiple health conditions allow us to control for co-morbidity, endogeneity and selection biases. The study confirms many existing findings on risk factors such as smoking and physical exercises, but also suggests the importance of SES may have been over-estimated in cross-sectional models.

Measuring Social Preferences for Treating the Permanently Disabled and Chronically Ill: Eliminating a Framing Effect
Authors: John McKie; Jeff Richardson; Jan Abel Olsen

Evidence suggests that the value members of the public place on saving the lives of patients with a disability or chronic illness differs depending on whether the patients were previously disabled or ill (the pre-existing scenario) or will be left so after treatment (the onset scenario). This is so even when the future QALYs (quality-adjusted life years) are identical. Further, the value people place on saving the lives of patients in both cases can differ significantly depending on the order in which subjects are presented with the scenarios. For example, the value subjects place on saving the life of a healthy patient who will left paraplegic after treatment may differ ten-fold depending on whether they have previously been asked about life-saving treatments for pre-existing paraplegics. If social preferences are to be taken into account in economic evaluations this presents a problem. Which set of preferences should count? Using data obtained from a series of in-depth, small-group discussions with members of the public we explain how we eliminated the order effect, report the distributional preferences of participants in its absence, and present the reasons for participant’s choices.

Preference-based outcomes: representative or age-matched general population values?
Authors: J Karnon; JE Brazier; RM Ara; BA van Hout

Assessing hospital performance using hospital administrative data
Chair: Tony Scott, University of Melbourne
Session Organiser: Jongsay Yong
Authors: Ying Chen; Terence Cheng; Choon Cheng; Tony Scott; Vijaya Sundararajan; Jongsay Yong

Health care costs and quality of care are two major concerns of all governments. In Australia, hospital care accounts for almost 40% of total health care expenditure, around 3% of GDP, and expenditure on public hospitals grows at 5-6% per year in real terms. Although expenditure is growing, hospital performance does not appear to keep pace. There are continuing concerns about the performance of public hospitals across Australia. These concerns range from quality and safety, waiting times, financial deficits, cost-shifting, workforce shortages, the role of technology, and the interaction with the private hospital and aged care sectors. These issues can have significant impacts on the quality of care and health outcomes of patients. This session consists of four presentations on using hospital administrative data to shed light on hospital performance such as quality of care and hospital efficiency, and the key drivers that are correlated with hospital performance.

Together these presentations address a number of important issues pertaining to hospital performance; they also highlight how hospital administrative data could be used to address these and other related issues.

Presentation 1: Linkage of the Victorian hospital administrative datasets and its value for research and policy development

Presenter Ying Chen, Victorian Department of Health

This presentation provides an overview of linked hospital administrative datasets available for researchers from the Victorian Department of Health. It highlights the enhanced value that data linkage adds to research using hospital administrative data, leading to greater possibilities in answering a broader range of research questions. The advantages and limitations of the Victorian linked hospital administrative data are also discussed and the current barriers and challenges around data linkages are explored.

The Victorian Data Linkages Unit (VDL) was established in 2009 with combined funding from the Australian and Victorian Governments as the Victorian node in the national Population Health Research Network (PHRN), to build a data linkage infrastructure capable of securely managing health information for health and medical research.

The Victorian Data Linkage Map contains enduring links between key hospital datasets and mortality datasets. In addition, VDL has responded to the needs of the researchers as well as government programs in undertaking the project specific linkages linking key hospital datasets to other health datasets. VDL has developed sophisticated data linkage methods and strategies by using privacy preserving methods to link together records pertaining to the same individual within or across different administrative datasets. De-identification and encryption rules are applied to remove identifying information about patients, service providers and service dates. The hospital datasets are enriched with socio-economic-geospatial indicators at small area level.

The benefits of hospital administrative data linkage are recognised as a resource for public good. However, there are also constraints that limit linkage of hospital administrative data to other datasets for research purposes. The barriers include current legislative framework governing collection and disclosure of information, insufficient common identifiers across data collections, quality of data collections, and different perspectives of data custodians versus researchers.

Presentation 2: Tracking hospital quality using Victorian hospital administrative data

Presenter Choon Cheng, Victorian Department of Health

This paper measures risk-adjusted mortality and readmission rates of Victorian hospitals using hospital admission data for an 11-year period from 2001/02 to 2011/12. For the purpose of risk adjustment, a standard logistic model is developed to take into account differences in patient characteristics including age, gender, principal diagnosis, admission type, comorbidity and care type. In total more than 23 million admission episodes occurring in 310 hospitals are used in the analysis, which provides an overview of the quality of hospital care in Victoria. Preliminary findings suggest that quality of care has been improving over time, but small public hospitals consistently lag behind other hospital types. Further analyses suggest that there exist considerable degrees of heterogeneity among small public hospitals. Considerable degrees of heterogeneity also exist among private hospitals, although private hospitals generally perform better than public hospitals in risk adjusted mortality and readmissions.

Presentation 3: Understanding hospital productivity using Victorian hospital administrative data

Presenter Jongsay Yong, University of Melbourne

Abstract This paper investigates the productivity of Victorian public hospitals using hospital admission data for the five-year period 2007/08 to 2011/12. We construct measures of total factor productivity using aggregate output and input indices. Aggregate output index is constructed by enumerating all in-patient activities of a hospital and are aggregated using diagnostic related group costs as weights. Aggregate input is constructed by adding total dollar costs of labour, capital, and materials. Total factor productivity (TFP) is computed as the ratio of aggregate output index to aggregate input index. Preliminary results suggest that, without taking into account changes in quality, hospital TFP did not register any material changes during the five-year period. Closer inspection further reveals that tertiary hospitals generally perform best in productivity among all hospitals, and small hospitals tend to exhibit greater degrees of variations in TFP growth than other hospitals.

Presentation 4: An investigation of cream-skimming using hospital administrative data

Presenter Terence Cheng, University of Melbourne

In this talk Terence Cheng will present the latest developments from a research project where we investigate the presence of cream-skimming behaviour using the novel approach of analysing hospital transfers. Using Victorian hospital administrative data we examine if patients transferred between public and private hospitals differ systematically in the severity and complexity of their medical conditions. We find that patients with higher disease severity are more likely to be transferred from private to public hospitals whereas the opposite is true for patients transferred to private hospitals. We also find that patients transferred from private to public hospitals stayed longer and cost more than private-to-private transfer patients, after controlling for patients’ observed health conditions and personal characteristics. Our findings are consistent with cream skimming behaviour by private hospitals.

Retrospective economic evaluation of childhood pneumococcal vaccination in Australia
Authors: Anthony T. Newall; Josephine F. Reyes; Peter McIntyre; Robert Menzies; Philippe Beutels; James G. Wood

Cost-effectiveness analyses are primarily used to provide evidence for policy and funding decisions before health interventions are implemented. Retrospective economic evaluations have received less attention in the literature. By making use of data collected post-implementation, retrospective cost-effectiveness analyses of vaccine programs may substantially reduce uncertainty and provide a better assessment of the value for money achieved. As a case study, we conducted a retrospective economic evaluation to assess the cost-effectiveness of the nationally funded 7-valent pneumococcal conjugate vaccine (PCV7) program in Australia. This is the first study specifically designed to evaluate the cost-effectiveness of an existing vaccine program in Australia. We designed a static, deterministic model that describes transitions between health states for a population over time. We investigated health-related outcomes on which PCV7 may have had an impact, primarily invasive and non-invasive pneumococcal disease. We applied a healthcare perspective and included costs for the vaccination program as well as healthcare utilisation. Quality-of-life utility weight estimates were attached to the different health states in the model. The primary outcome calculated was an incremental cost per Quality Adjusted Life Year (QALY) gained. We conducted one-way and probabilistic sensitivity analysis, and examined a number of different scenarios. In the scenario where observed declines in all health outcomes were attributed to PCV7, the ICER was below A$50,000 per QALY gained. However, in scenarios where only declines with more robust evidence for attribution to the program were included, the cost-effectiveness was less clear-cut. We discuss how our results compare to previous pre-implementation cost-effectiveness studies on PCV7 in Australia, and how methodological choices impact on cost-effectiveness results. Retrospective cost-effectiveness analysis can provide important insight in our understanding of the value for money achieved by vaccination efforts.

Cost of hospital and health service utilisation of very pre-term infants from the ProPrems Neuro study
Authors: Kah-Ling Sia; Lisa Gold; Sue Jacobs; Jeanie Cheong; Gillian Opie; Suzanne M. Garland; Leah Hickey; Rosemarie Boland; Chelsea Webster for the ProPrems Study Group

Patient contrbutions in Solomon Islands and Papua New Guinea: Whether and how to eradicate them
Authors: Katherine Gilbert; Duncan Mortimer; Nicole Au; Rohan Sweeney; Jean Spinks; Ajay Mahal; Wayne Irava; Paula Lorgelly

The Solomon Islands Ministry of Health and Medical Services (MHMS) delivers health care to over 500,000 people in predominantly rural areas across 10 provinces, including numerous remote islands. It has a budget of SBD 552 million (AUD 81 million), which is sourced from annual appropriations from national revenue and from development partners as part of the Health Sector Support Program (HSSP).

The MHMS is committed to providing free universal access to health services; and the current National Health Act restricts the collection of fees at health facilities to hospitals. A recent Patient Exit Survey undertaken in Solomon Islands found that despite this explicit restriction patient contributions are common across all facility levels for consultations and medical record books, excluding the National Referral Hospital in Honiara.

The accompanying Health Facilities Costing Survey found that revenue collected from patient contributions are the only sources of revenue managed by facilities. Health workers reported that these contributions are used to cover gaps in the salaries and supplies provided by the national and provincial health administrations.

A detailed analysis of who makes patient contributions suggests that the poor and women pay such fees more often, but they do not appear to be a barrier to access and are waived for those who cannot afford to pay. Evidence from a recent study in PNG (Wiltshire and Mako 2014) shows a similar context. While eradicating such fees is consistent with the MHMS policy and promotes the WHOs universal health care policy, it presents administrative challenges for the MHMS and has additional cost implications in getting cash to health facilities, and supporting them to manage it appropriately.

This paper will explore the options available to the MHMS and other low and middle income countries who wish to abolish user fees.

Primary care in Australia: towards better quality and value for money
Chair: Denzil Fiebig; presenting authors: Ian McRae; Rachael Moorin; Anthony Scott; Chunzhou Mu

Primary care in Australia: towards better quality and value for money

Chair: Professor Denzil Fiebig

Primary care has been the subject of substantial policy changes in recent times. Proposed government reforms have focused on changes to the funding of GP services as well as on organsisational aspects, including the establishment of regionally-based primary health networks (PHN). This organised session will examine some of the key policy challenges, covering issues relating to funding, efficiency, quality and equity in the field of primary care. Four papers from leading researchers in the fields of health economics and policy will be presented in this session:

The economics of a GP co-payment

, Dr Ian McRae*, Australian National University
The Australian Government’s proposal for a $7 GP co-payment, if implemented, will impact on patients’ out-of-pocket costs, their attendance at surgeries and on GP earnings. We explore both estimated supply and demand responses for the overall GP market and possible responses in different market segments.

Capturing regularity of MBS funded primary care services in WA: Evaluation of its relationship to frequency and cost

, Associate Professor Rachael Moorin*, Curtin University
The rapidly ageing population, high prevalence of chronic illnesses, and ever increasing expectations in service delivery are placing the Australian health care system under increasing pressure. Approaches to reduce this unsustainable demand include policies to improve the regularity and quality of healthcare provided by GPs. This presentation will showcase methodology developed to evaluate service regularity as distinct from frequency of GP visits using longitudinal whole-of-population linked individual level Medicare and hospitalisation data.

What explain the use of GPs?

Ms Chunzhou Mu*, Professor Jane Hall CHERE, University of Technology, Sydney
Geographic variation in GP use is a persistent finding across Australia. Some of this variation may be warranted, reflecting differences in, for example, population health status. However, variation may also be indicative of health care inefficiency and inequity. This paper examines the amount of geographic variation in GP use, and the factors that influence that variation. Results suggest that health care needs, age, obesity, lower socio-economic status are significant drivers of health care use. The availability of more GPs is also associated with higher GP use but the supply of more specialists reduces the use of GPs. A better understanding of the sources of the geographic variation can inform policy interventions to prioritise resource allocation and improve health care delivery efficiency.

Do financial incentives influence GPs’ decisions to provide after hours care? A discrete choice labour supply model.

Barbara Broadway, Guyonne Kalb, Anthony Scott*, Jinhu Li, Melbourne Institute of Applied Economic and Social Research, The University of Melbourne
The aim of this paper is to examine the effects of changes in financial incentives on whether GPs provide after hours care, using data from the first wave of the MABEL survey. A discrete choice labour supply model is used where decisions on total hours worked are combined with decisions on whether or not to provide after hours care. Sixty percent of men and 41% of women GPs do some after hours work. Preliminary results show that if earnings from working after hours increase by 1%, then the probability of undertaking after hours work increases by around 0.1%. Women with children are less likely to increase their after hours workload in response to a wage increase, compared to women with no children and men. Overall, the results suggest that GPs are not very responsive to increases in earnings for after hours work. Family factors and whether the GP is self-employed seem to play a larger role.

Following the presentations, there will be an opportunity for discussion among participants and the authors.

* presenting author

Health, austerity and economic crisis: assessing the short-term impact in OECD countries
Authors: Kees van Gool; Mark Pearson

The economic crisis that started in 2008 has had a profound impact on the lives of citizens. Millions of people lost their job, saw their life-savings disappear and experienced prolonged financial hardship. The economic crisis has also led a number of OECD governments to introduce austerity measures to reduce public deficits. The combined effects of economic crisis, austerity and reforms have led many OECD health systems into unchartered territory.

This paper looks at the impact of economic crisis on health and health care. It summarises findings from the published literature on the effects of economic crisis that took place over recent decades. This paper also analyses the empirical relationship between unemployment and health care use, quality and health outcomes, using data from OECD Health Statistics. In doing so, it investigates whether the effects of unemployment on health outcomes have been extenuated by austerity measures.

Results show that economic downturns are associated with adverse outcomes for some, but certainly not all, health indicators. During times of economic crises, mental health deteriorates and the prevalence of communicable diseases appears to rise, but at the same time there are fewer deaths from transport accidents. There is less consistent evidence on the relationship between economic conditions and overall health outcomes such as mortality and health care quality but there is evidence that higher rates of unemployment are strongly linked to lower health care use. Austerity measures appear to have led to a decline in hospital admissions but not to reductions in pharmaceutical consumption. The short-term effects examined in this paper suggest that austerity measures have had mixed success in protecting patients from reduced health care access, but there remains an important need to actively monitor the wider long-term health impact of the economic crisis.

Pensions health and retirement

Kompal Sinha, Centre for Health Economics ,Monash University

Anthony Harris ,Centre for Health Economics, Monash University

The aim of paper is to assess the level of income needed in retirement to maintain a satisfactory life and slow the rate of health decline. This is an alternative to the usual assessment of the adequacy of income in retirement based on pre-retirement income replacement. Using the HILDA data from Australia, this paper estimates the short and long term health effects of retirement accounting for the simultaneity of health and retirement in an instrumental variable panel data econometric model. We initially assume that human, and financial capital is exogenous at retirement. The results suggest that retirement has a positive effect on health and wellbeing initially but in the longer term has a negative effect on health mitigated by pension income and wealth. Health in the bottom income quintile wears out a good deal faster after retirement than does health in the top quintile, and in the bottom quintile men’s health deteriorates more rapidly than women’s health.

Assessment of Community Based Health Insurance In Sunsari Disrtict
Red Herring in Vistula River - New Evidence on Time to Death
Authors: Maciej Lis

The aim of article provide evidence on the role of age and proximity to death as drivers of health care expenditure (HCE). For our analysis a unique dataset on 14-types of HCE by age, sex and distinction for decedents and survivors has been delivered by Polish National Health Fund. We decompose the differences in health expenditure of 1-year cohort groups to quantify the contribution of survivors, decedents, mortality-rate and population changes. Then we compare the effects of assuming age-dependence and proximity-to-death-dependence for total health care expenditure. We find out that HCE depend on the proximity to death, but age-dependence cannot be left out. We also quantify the effects of both models for ageing to show that their implications hugely contradict. Using exponential model with parameter transition of cost increse before death we show that the health expenditure start to rise at about 10 years before death and that the implications of time-to-death and age-dependent model are considerable for the effects of ageing on health care expenditure and public finance.

Multi-method Patient Flow Simulation Models – An Opportunity to Engage to Influence Efficiency and Policy
Authors: Mark Mackay; Shaowen Qin; Dale Ward

Despite the introduction of the national emergency access target key performance indicator, many Australian public emergency hospitals continue to struggle with delays in the discharge of patients from the emergency department or admission into a hospital bed within 4 hours. High levels of bed occupancy and increasing demand for services contribute to this situation. Given expected health care budget reductions and increases in demand this situation is unlikely to change significantly, if at all.

The health care system is a complex system. Patient wait and crowding issues are not simple issues and arise as the outcomes of many contributing factors. Simulation tools exist that enable health care operational managers and health care policy makers to explore the complex issue of patient flow in hospital environments. While there are many examples of simulation being applied to the redesign or design of emergency departments in the literature, the use of simulation as a tool that is regularly employed by decision-makers in the health sector has yet to occur. The reluctance may be attributed to a variety of factors including a lack of skills, the perception of cost, the lack of data and confusion regarding methods.

Previously simulation models have typically been developed using either a system dynamics or discrete event approach. AnyLogic, an off-the-shelf simulation software package, provides for agent-based and multi-method simulation.

We describe a multi-method modelling project designed to create a simulation model of patient flow that describes the progression of patients through the emergency department of a large teaching hospital into general medicine wards and then discharge. We also describe how this model has been used to engage with those interested in seeking improved efficiencies and understanding future scenarios, and how the model has been used to engage in policy debate.

A profile of disability in Cambodia
Authors: Michael Palmer, The University of Melbourne; Jenny Williams, The University of Melbourne

Globally, people with disabilities are over-represented in the poorest nations. Accordingly, the United Nations has flagged disability as a key development issue in meeting the United Nations Millennium Development Goals, which aim to eradicate world poverty by improving health, education and other outcomes. In an effort to meet the Millennium Development Goals, governments of several developing countries have introduced legislation designed to equalise the rights and opportunities of people with disabilities. To date, there has been little robust analysis of the economic profile of persons with disabilities in development countries. This represents a considerable knowledge gap that has practical implications for the development of efficacious strategies to improve the lives of people with disabilities. This research proposes to take the first steps in filling this gap by analysing a range of life outcomes for persons with disabilities as articled in Cambodia’s 2009 Law on the Protection and the Promotion of the Rights of Persons with Disabilities. Pooling six independent cross-sections of the Cambodian Socio-Economic Survey, the study will evaluate the impact of disability on poverty, employment, education, health and health care outcomes. The study draws the first representative profile on the lives of persons with disabilities in the country.

Cost Effectiveness of Aggressive Risk Factor Management in Atrial Fibrillation
Authors: Rajeev Pathak; Michelle Evans; Melissa Middeldorp; Abhinav Mehta; Rajiv Mahajan; Darragh Twomey; Anand N. Ganesan; Walter Abhayaratna; Dennis H. Lau; Prashanthan Sanders

Investing in primary health care reduces the risk of diabetes-related hospitalisation of Aboriginal and Torres Strait Islander people with type 2 diabetes in north Queensland, Australia
Authors: Odette Gibson; Robyn McDermott; Adrian Esterman; Ming Li; Mark Wenitong; Leonie Segal

Objective: To estimate the impact of per capita primary health care (PHC) resourcing on the risk of diabetes-related hospitalisation over a 5 year period (2000-2005) in a cohort of Aboriginal and Torres Strait Islander Australian adults with type 2 diabetes (T2DM).

Design: Retrospective cohort study using linked community survey data, hospitalisation and death records.

Setting: Twenty one remote communities in far north Queensland, Australia.

Participants: Adults with existing or diagnosed with T2DM during a health check offered between1999-2000 (n=366).

Main outcome measure: Diabetes-related hospitalisations

Results: There was an inverse relationship between per capita PHC investment and the risk of diabetes-related hospitalisation over a 5 year period. Each AUD $1,000 per capita PHC spending was associated with a 24% reduction in hospitalisation risk (adjusted RR = 0.76, 0.60 to 0.95). This effect was strongest for extra spending on Aboriginal and Torres Strait Islander health workers (adjusted RR=0.21, 0.06 to 0.80) and medical officers (adjusted RR = 0.22, 0.05 to 0.95), but for extra employment (i.e. FTE) of medical officers (adjusted RR = 0.10, 0.01 to 1.01).

Conclusion: Investing in PHC in an under-served remote high risk population with T2DM prevents diabetes-related hospitalisations. Aboriginal and Torres Strait Islander health workers contribute significantly, as members of a multi-disciplinary PHC team, to reducing the risk of diabetes-related hospitalisations of people living in remote areas with T2DM.

Mental Health and Disengaged Youth
Authors: Paul Amores; Robert Breunig; Tue Gorgens

Low socio-economic status youth are found to have worse mental health and be more economically inacte. Poor mental health is correlated with lower levels of economic activity, particularly employment.

While current activity tends to be state dependent, at least half of those who were economically inactive at age 18 were fully engaged at 20. Similarly, mental health is found to be neither permanent nor totally transitory.

It is concerning that the persistence of inactivity, and relationship between mental health and economic outcomes, is found to be stronger among disadvantaged youth.

Data is from the Youth in Focus project, which covers almost all Australians born between 1 October 1987 and 31 March 1988, including disadvantaged youth and comparison groups of those from middle and upper-middle income families. Administrative data provides us with information about the incidence, timing and intensity of welfare receipt while the young person was growing up.

The challenges of undertaking a cost benefit analysis of a large scale, multifaceted community-based nutrition project in Indonesia
Authors: Lisa Cameron; Paula Lorgelly; Kompal Sinha

One in three children under five in Indonesia is stunted. Stunting negatively affects future productivity, as stunted children are at a higher risk of experiencing chronic disease, impaired cognitive development and reductions in academic achievement. The a priori economic argument for developing programs to address stunting would thus appear strong. Increasingly though, to justify investment in such health initiatives, governments and funding agencies require a calculated estimate of a program’s expected monetary return. Indonesia’s Community-based Nutrition Program (CBNP) is a large-scale, complex, nutrition project with a budget in excess of USD150m. Its activities include the provision of conditional cash transfers to villages (with the promise of larger future grants if health indicators improve); training health providers; conducting sanitation activities; providing micronutrients to pregnant women and young children; results-based financial incentives to health providers; and a communications outreach campaign. We were tasked with calculating a single economic rate of return (ERR) for the entire project, ERRs for specific project components, and the returns flowing to the poor and women. Budget costs were compared to estimated benefits – comprising projected future increases in individuals’ incomes associated with the improved productivity resulting from reductions in low birth weight, stunting and malnutrition, and savings to the health sector from diarrhoea and chronic disease averted. Benefits flowing from the economic stimulus associated with the cash grants were also estimated. We highlight the challenges of undertaking a modelled economic evaluation in a lower-middle income country – including identifying the (often scarce) relevant evidence of the effects of the various interventions and particularly the monetary benefits; defining beneficiary groups; the difficulty of modelling benefits for separate project arms when there are likely to be interactions between the program’s different components; and the attempts to model uncertainty and sensitivity in the results.

Health Shocks and Household Welfare in Zambia: An assessment of changing risk
Authors: Peter Hangoma; Arild Aakvik; Bjarne Robberstad

Health shocks consist of illness, injury and death which may reduce household consumption through increased medical spending and reduced labor supply or productivity. Given that the government of Zambia undertook extensive health and economic reforms between 1994 and early 2000, our main objective was to assess the relationship between health shocks and household welfare before the reforms were fully implemented (pre_2000) and after (post_2000).

We assessed the risk of health shocks (injury and death) on household welfare using five waves of nationally representative household survey data collected between 1996 and 2006. We carefully modelled the relationship between injury and death on one hand and outcomes (Food consumption, Nonfood expenditure, Medical Expenditure and categories of income) on the other hand in a Seemingly Unrelated Regression framework (SUR) taking survey design into account.

The negative impact of health shocks on household welfare increased post_2000. Pre_2000, injury did not have any effect on food and non-food consumption. The picture however changed post_2000 when injury in a household reduced food consumption by 8%. In terms of sources of risk, post_2000, medical expenditure became a major source of risk with injury increasing medical expenditure by 29%; medical expenditure did not pose a risk pre_2000. Similarly, the reduction of non-farm business income from an injury increased from 17% pre_2000 to 27% post_2000. In terms of coping, injury increased the likelihood of selling assets by 5% pre_2000 and 6% post 2000. On the other hand, death reduced wage income by 3.7% pre_2000 and this worsened to 13% post_2000.

The post_2000 period saw an increase in the risk of increased medical spending and reduction in income as a result of health shocks. Selling assets was the key coping strategy but this was not sufficient to prevent the fall in household consumption.

Impact of price on initiation and persistence of statin use in the United States
Authors: Dennis Petrie; Philip Clarke; Joshua Salomon

Background: Statins are one of the most commonly prescribed medications in the United States and have been shown to be extremely effective at reducing the risk of major cardiovascular events and mortality. In recent years many types of statins have come off patent and this has produced dramatic declines in price.

Objective: To examine whether the fall in price has resulted in an increase in the initiation and/or persistence of statin use.

Design: Repeatedly conducted 2-year overlapping panel, nationally representative survey of non-institutionalized adults from 1996-2011.

Setting: United States.

Participants: Adults aged 40-75 years included in the Medical Expenditure Panel Survey between 1996-2011.

Measurements: Out-of-pocket price paid for statins and the self-reported use of statin medications over a two year period.

Results: There were large decreases in the real out-of-pocket prices paid for statins from 2006 which resulted in the average price more than halving; from $1.87 per tablet in 2006 to $0.46 in 2011. While there is some evidence of a significant effect of changes in the price on the initiation rate of statin use, there was a larger impact on persistence with a $1 drop in the per tablet price of statins resulting in a 27% drop in the OR of discontinuation. These price reductions are estimated to translate into nearly 1 million more Americans between 40-75 years in 2011 continuing to take their statin medications and approximately 750,000 more starting statin use each year due to the reduction in price.

Conclusion: While out-of-pocket prices seem to have a smaller impact on individuals initiating statin use it plays an even more important role in encouraging individuals to continue taking their medication.

Cost-effectiveness implications of displacement of oncology therapies
Authors: Philip Haywood, Rosalie Viney, Marion Haas

The development of new therapies in oncology has the potential to displace older therapies from their current positions. This may impact on the cost-effectiveness. This paper evaluates the impact of displacement on cost-effectiveness and models the outcomes.

RCTs have been synthesized via a meta-analysis and conclusions drawn about the impacts of displacement on characteristics of oncology protocols (a collection of therapies). Specifically the impact of a protocol being displaced from one of therapy to the next line of therapy has been assessed on overall survival, progression free survival and toxicity.
The impact of a protocol being displaced a line of therapy is to decrease a protocols effectiveness and increase the rate of toxicity per unit time. The line of therapy a protocol is used within impacts on the characteristics of the protocol. However the size of the impact is uncertain because of the data reported in the trials and the necessary restrictions on the methods of synthesis.

The potential impacts of the characteristics of a protocol altering as it is displaced are explored using modelling. The impact of displacing a protocol is to increase the cost-effectiveness of the protocol. Increasing the toxicity and reducing the effectiveness of a therapy in displacing it from one line of therapy to the next increases the cost-effectiveness of the therapy. The net economic loss associated with the increase in cost-effectiveness is highly variable. Key determinants include the mechanism of reimbursement.

The cost-effectiveness of a protocol is dependent on which line of therapy it is used in. Without pricing changes, on average, the cost-effectiveness worsens as a protocol is displaced from one line of therapy to the next. However the size of the opportunity cost of the health forgone from the worsening cost-effectiveness is highly variable.

Anchoring vignettes and the EQ-5D: Adjusting for systematic differences in reporting behaviour
Authors: Rachel Knott; Paula Lorgelly; Nicole Au; Bruce Hollingsworth

Preference-based measures of health are often used in economic evaluations to calculate quality adjusted life years (QALYs). Increasingly they have been used as a measure of population health status; for example, comparisons of EQ-5D health profiles and utility values have been made by socio-economic status and behavioural risk factors. However the EQ-5D as a subjective measure is likely to suffer from mis-reporting and response category differential item functioning (DIF), like other self-rated measures of health, and failure to correct for it may lead to conclusions that are misleading.

This paper explores the use of anchoring vignettes to test and correct for systematic differences in responses and improve inter-group comparisons of EQ-5D profiles and indices. We apply hierarchical ordered probit (HOPIT) models to individual domain responses and assess the impacts of DIF in EQ-5D reporting according to gender, age, income and country of birth. We find considerable variation in the use of response categories across all EQ-5D domains, which alters relative differences in EQ-5D summary indices across respondent sub-groups.

Quality of life in older adults following a hip fracture: an empirical comparison of the ICECAP-O and the EQ-5D instruments
Authors: Rachel Milte; Maria Crotty; Michelle Miller, Craig Whitehead, Julie Ratcliffe

Aim: This study compares the performance of the ICECAP-O and the EQ-5D instruments for the measurement and valuation of quality of life in a population of older adults following a hip fracture.

Methods: 87 consenting older adults completed the ICECAP-O and EQ-5D instruments following surgery to repair a hip fracture, with family members recruited as proxy respondents for patients with significant cognitive impairment. Values were calculated for both instruments. For the EQ-5D, values are generated on the zero=death and 1=full health quality adjusted life years (QALY) scale whereas for the ICECAP-O values are generated on the zero=no capability and 1=full capability scale.The results were compared to value generated from a general population sample of older adults.

Results: Mean utility scores for ICECAP-O (0.639 SD 0.206) and EQ-5D (0.545 SD 0.251) were found to be lower than the general population norms for both instruments (ICECAP-O: 0.798 SD 0.165, p=

Retirement and consequences for health in Australia
Authors: Kostas Mavromaras, Sue Richardson, Rong Zhu

This paper estimates the causal effect of retirement incidence and duration on health outcomes among mature age Australians. We utilize pension eligibility age in the context of Fixed Effects Instrumental Variables estimation. We find that becoming retired has positive and significant effects on health, both physical and mental, and more so for women. We find that longer time spent in retirement confers clear additional health benefits for women, but not clearly so for men. We also find robust evidence that health improvements due to retirement can only be traced for people with below median health status.

Projection of economic impacts of chronic conditions leading to early retirement of older working age Australians: Results from a microsimulation model Health&WealthMOD2030.
Authors: Deborah Schofield; Rupendra Shrestha; Michelle Cunich; Simon Kelly; Lennert Veerman; Robert Tanton; Megan Passey

Premature retirement due to illness can have significant economic impacts. Individuals are affected through reduced income and savings; governments through decreased taxation revenue and increased expenditure on income support payments. These economic impacts are likely to increase in future with an ageing workforce and increasing trends in disease prevalence.

We have a developed a microsimulation model, Health&WealthMOD2030 to estimate the projected economic impacts of illness leading to early retirement to 2030, sourcing the base population from the Australian Bureau of Statistics 2003 and 2009 Surveys of Disability, Ageing and Carers datasets. Projected estimates of income, taxation, income support payments, savings and superannuation from a dynamic microsimulation model APPSIM and the disease trends from ACE-prevention models were synthetically matched with the base population to project forward the economic impacts of early retirement due to ill health to 2030 for the Australian population aged 45 to 64 years old.

This paper provides projections of the economic costs associated with illness-related early retirement for both individuals, in terms of income lost, and government, in terms of taxation revenue lost and increase in government support payments, for every five years from 2010 to 2030. On average, annual earnings lost of those not in the labour force due to illness, when compared to those in full time employment and have no chronic condition, is estimated to be about $43000 in 2010, which is estimated to increase to about $60000 by 2030. The estimated annual national loss of earnings due to illness-related early retirement is about $11 billion in 2010, estimated to increase to about $20 billion by 2030.

These results highlight the need for effective prevention or treatment of illness to keep older working age Australian in the labour force to reduce the immediate and the long-term negative economic consequences.

Social Network and Mental Health: Evidence from the Rural-to-Urban Migration in China
Authors: Xin Meng; Sen Xue

This paper attempts to identify the causal effect of social network on mental health of the Chinese rural-to-urban migrants. The Chinese migrants are a vulnerable group to the mental diseases. The empirical analysis is conducted based on a unique migrant survey from the Rural-to-Urban Migration in China Project, which includes the most up-to-date information about Chinese rural migrants. Using OLS and FE models, we find that larger network is significantly correlated with better mental health. To mitigate the endogeneity bias, we employ a novel instrumental variable approach. In particular, we use the rainfall at migrants’ hometown and mobile phone ownership as the instrumental variables. Both IV and FEIV estimates indicate that social network is beneficial in reducing mental health problem, and the effect is larger and significant for the migrants who are less mobile, have smaller network or are with more serious mental problem.

Social Network and Mental Health: Evidence from the Rural-to-Urban Migration in China
Authors: Xin Meng; Sen Xue

This paper attempts to identify the causal effect of social network on mental health of the Chinese rural-to-urban migrants. The Chinese migrants are a vulnerable group to the mental diseases. The empirical analysis is conducted based on a unique migrant survey from the Rural-to-Urban Migration in China Project, which includes the most up-to-date information about Chinese rural migrants. Using OLS and FE models, we find that larger network is significantly correlated with better mental health. To mitigate the endogeneity bias, we employ a novel instrumental variable approach. In particular, we use the rainfall at migrants’ hometown and mobile phone ownership as the instrumental variables. Both IV and FEIV estimates indicate that social network is beneficial in reducing mental health problem, and the effect is larger and significant for the migrants who are less mobile, have smaller network or are with more serious mental problem.

Economic Implications of Cardiovascular Disease Management Programs: Moving Beyond One-off Experiments
Authors: Shoko Maru; Joshua Byrnes; Melinda J. Carrington; Simon Stewart; Paul A Scuffham

Despite the expectation that cardiovascular disease management programs (CVD-MP) may reduce costs while improving health outcomes, the results of cost-effectiveness analyses remain equivocal. Substantial variation in economic analyses of CVD-MPs hinders not only the proper assessment of cost-effectiveness but also the identification of potential predictors of cost-effectiveness. The objective of this review is to explore how much of the variability of cost-effectiveness can be explained by the characteristics of patients or intervention evaluated rather than the differences in study methodology and reporting practice.

Based on two systematic reviews we recently conducted, we further explored the impact of methodological and reporting variation on the cost-effectiveness of multi-component CVD-MPs by introducing scenarios that could lead to different policy, funding or clinical decisions. We then focused on ambiguity or equivocation, an issue most commonly observed in the published reports on CVD-MPs, which hinders the assessment of net intervention effects and the generalizability of findings.

We identified key limitations around which the research of CVD-MPs tend to pivot, such as poor third-party reproducibility, variations in handling of costs arising from improved survival, competing risks surrounding hospitalization events, or the assumed durability of post-intervention effects. To date, these limitations have hindered the ability to move beyond one-off experiments (a single study-based economic evaluation) and time-/place-specific attributes, leading to the substantial variation in the economic implications of CVD-MPs.

The current evidence supporting CVD-MP is affected by both methodological variation and inadequate or insufficient reporting of cost components, thereby true heterogeneity of interest is obscured. Further standardization using practical tools used in our recent systematic reviews such as a checklist for the reporting standards of health economic evaluation reporting will increase transparency and comparability across studies. These efforts should help translate research findings into unambiguous recommendations for decision makers.

Explaining gender differentials in child mortality in India: Trends and determinants
Authors: Shrikant Kuntla

Background: In spite of the growing number of literature on child mortality in India, the reasons behind the changes and factors affecting gender differentials in child mortality is poorly understood. Studies, which have assessed the trends have not understood change in terms of within and between changes in gender differentials in child survival. Further, the studies focused on the determinants were estimated factors contributing to child survival but not the factors contributing to gender differentials in child survival.

Objective: To investigate the progress in gender differentials in child mortality, in India, in terms of within and between group changes. To identify the factors explaining the gender differentials in child mortality.
Data & methods: Utilising National Family Health Survey (NFHS) data, 1992 to 2006, Patty and Oaxaca decomposition analyses have been done to study the progress and critical predictors of gender differentials in child mortality in India.

Results: The results of patty’s Gini decomposition change in gender differentials, in child mortality, reveal that change (Gini index from 0.36 to 0.24) is greatly driven by change in within inequality (Gini index from 0.18 to 0.14) which is also a major contributor to it. Among the factors affecting to gender differentials in child mortality, breastfeeding (87%), birth order (4.25%), antenatal care (3.93%) and the mother’s age (2%) emerged as critical contributors for the excess female child mortality in India.
Conclusion: This study suggests that any efforts to remove gender differences in child survival should more focus on within female children discrimination across different population sub-groups alongside male-female discrimination. Also need to cut-down the discrimination in proximate factors such as breast feeding, nutritional and health care requirements, neglect of higher order female births and need to increasing mean age of women at child birth.

How do drug prices respond to a change from external to internal reference pricing? Evidence from a Danish regulatory reform
Authors: Ulrich Kaiser; Susan J. Mendez

Reference pricing is a widely used cost containment tool where the maximum reimbursement a patient can obtain is determined by prices of substitute drugs. Reference prices are usually set using either domestic prices (“internal” reference pricing) or foreign prices (“external” reference pricing). We study the effect of a change from external to internal reference pricing using a Danish regulatory reform and explore how the reform had different effects on different therapeutic markets

For the analysis we consider three markets that differ in their treatment duration, these are: anti-cholesterol drugs for a chronic condition, antiulcerants for a semi-chronic condition, and antibiotics for the treatment of an acute condition. We find that, in general, the switch from external to internal reference pricing led to a substantial reduction in pharmacy purchase prices. The effects are stronger for chronic conditions than for acute conditions. Furthermore, the reform affects generics and parallel imports more than original products. Finally, we find that the reform reinforces the effects of competitive pressure.

Do you have to win it to fix it? A longitudinal study of lottery winners and their health care behaviours.
Authors: Terence C Cheng (UMelb); Joan Costa-i-Font (LSE); Nattavudh Powdthavee (LSE)

In this talk I will present on-going work on a research project where we exploit lottery wins to investigate the effects of exogenous changes in individuals’ income on health care seeking behaviours in the United Kingdom. We study this using longitudinal data of roughly 15000 lottery winners from the British Household Panel Survey to analyse the decision to use health care and the choice of private (non-NHS) versus NHS care for a range of health care services. Our results indicate that winners with larger wins are more likely to choose private care over NHS care. For individuals without private medical insurance, the results suggest that lottery winnings are used towards affording the out-of-pocket payments associated with the use of private hospital care. For individuals with private medical insurance, winners with larger wins are more likely to choose private care for dental services and preventive health services (e.g. eye, cervical examinations). I will discuss our findings in relation to the literature on the relationship between income and health expenditures, and broader issues surrounding public-private mix in health care.

Expenditure on non-communicable diseases in China: Results from a National Health Accounts Framework
Authors: Tiemin Zhai, Yuhui Zhang, Peipei Chai, Qiang Wei, John Goss

Rationale:Rapid socio-economic development, dramatic changes in people’s way of life and population aging have led the case prevalence of non-communicable diseases (NCDs) in China to increase rapidly, from 15.8% in 1998 to about 20.0% in 2008. These trends beg a series of questions. Most notably, how many resources have been spent in preventing, managing and treating NCDs so far; what diseases consumed most of health resources, and how much was paid by individuals so that the country is able to formulate health financing policies that balance the needs of the population and the country’s capacity to pay for services.

Objective: To analyse expenditure on prevention, management and treatment of NCDs by financing source, health provider and health function in China.
Methodology:Analysis was based on the latest version of the OECD and WHO System of Health Accounts methodology. Relevant data for the study were drawn from 3 major sources, namely the hospital revenue and expenditure data from the Ministry of Health, individual service utilization data obtained from a field survey conducted in 4 provinces and China’s 4th National Household Health Survey.

Result: Total expenditure on NCDs accounted for nearly 70% of total health expenditure(THE), but of these less than 2 percent were spent on prevention. Of those NCDs, CVDs consumed about 34% of the expenditure. Results also showed that reimbursement through health insurance was relatively lower for NCDs.

Conclusion: NCDs accounted for a large proportion of THE, but we also know that a large promotion of NCDs can be prevented by appropriate primary health care interventions. Therefore, the reimbursement policies should be direct towards encouraging these types of interventions. On the other hand, if these types of interventions discouraged by the reimbursement policies this will raise health financing sustainability issues.

The implications of different methods for modelling time from event til death for health economic analyses
Authors: Tom Lung, Dennis Petrie, Alison Hayes, Bjorn Eliasson & Philip Clarke

Aims: Estimating time after event parametric models from survival analysis are often used to extrapolate into the future for predicted costs and outcomes. These predictions are used in cost-effectiveness analysis to form the basis of medical decision making. This paper aims to investigate the implications of using different methods for estimating the risk in the present and in the future.

Methods: This study is based on data from the Swedish National Diabetes Register linked to patient level hospital records, prescription data and death records. We selected patients with Type 1 diabetes who visited a clinic 2002-2010 and experienced a major cardiovascular complication after their clinic visit. We estimated a parametric model for mortality following a cardiovascular complication in two ways: Using age as the underlying risk; Using time from CVD event as the underlying risk. A simulation model estimating cardiovascular incidence, mortality and a competing risk model was used for illustrative purposes in a cost-effectiveness setting.
Results: Immediately following a CVD event, there is an increased hazard, which subsides in the medium term but increases again in the long-term, mimicking the shape of a “bathtub”. The results suggest using age as it adjusts for this effect post-event, producing better predictions in the future than using time from event. When extrapolating into the future, the two models predict opposite sloping directions of hazard which has different implications for cost-effectiveness analysis of interventions following a CVD event.

Conclusions: When estimating parametric models following a CVD event in Type 1 diabetes patients, our paper suggests using age as the underlying hazard provides better predictions when using these models to predict long-term outcomes and costs.

What is the best model to use to evaluate an outpatient orthopaedic service - DES or Markov cohort?
Authors: Lachlan Stanfield; Tracy Comans; Paul Scuffham

Social Determinants of HIV: A Study of Women in Kerala
Authors: Vaibhav Khandelwal

Linking hospital accreditation scores and hospital acquired infection rates: the quest for a safety and quality indicator
Authors: Virginia MUMFORD; David GREENFIELD; Kevin FORDE; and Jeffrey BRAITHWAITE.

Background: While health service accreditation programs, particularly in acute care, are well established and have seen rapid spread, little is known about the costs of the accreditation process. Current research in the acute sector is generally based on self-reported single studies with little evidence of costs being assessed over a range of acute care facilities. We describe a survey to evaluate the incremental costs of accreditation in five acute hospitals in three different states in Australia.
Methods: We reviewed the stakeholders of accreditation costs with a focus group. We then constructed a purpose-built audit tool using previous survey methodology and policy evaluation tools. We piloted the survey with one institution and completed the survey with four other acute care facilities. We convened an expert panel to assess the validity of both methodology and scalability of the results, and conducted a sensitivity analysis.
Results: Accreditation costs varied from 0.03% to 0.60% of operating costs per year, averaged across the four-year accreditation cycle. Higher costs were associated with the surveys years and with smaller facilities. Scaling up our costs nationally results in total accreditation costs of $40.8 million, equivalent to 0.11% of acute public hospital recurrent expenditure in 2012 fiscal year.
Conclusions: This is the first time accreditation costs have been independently evaluated across a wide range of hospitals. This methodology can be readily adapted to assess accreditation costs across the acute, primary and aged care health services in any health system. A better understanding of the costs allows policy makers to assess alternative quality improvement strategies, and to review accreditation and standards-setting methods, especially for smaller hospitals.

The Effects of Air Pollution on Health: Evidence from China
Authors: Xiaobo He; Rong Zhu; Hongjia Zhu

Using three waves from the China Health and Nutrition Survey (CHNS, 2004, 2006 and 2009), this paper examines the causal effects of air pollution on health in China. The measures of health rely on results of biomarker testing. In addition, this study sheds light on people’s response to local air pollution. To mitigate potential endogeneity bias in ordinary least squares (OLS) estimates, this study utilizes wind speed as an instrumental variable (IV) for the endogenous variable of interest, i.e. air pollution measured by Air Pollution Index (API). The IV estimates show that more severe air pollution causes lower density of hemoglobin whereas higher density of platelet, white blood cell, insulin and uric acid in blood. This study also provides evidence that heavier air pollution leads to changes in human behavior: i) more likelihood of having supplementary medical insurance, ii) less likelihood of drinking alcohol everyday, and iii) less consumption of beer and wine per week.

Preferences across attributes for use in health care priority setting – a systematic review of empirical evidence
Authors: Yuanyuan Gu, Emily Lancsar, Cam Donaldson, Peter Ghijben, Jim Butler

Achieving efficiency in non-market decision-making in health care is often interpreted as requiring QALY maximisation. This approach implies that QALYs are homogeneous across a population, i.e. that “a QALY is a QALY is a QALY” to whomsoever it accrues. However, QALY maximisation may not result in the most equitable distribution of QALYs across society. QALY gains across a population can differ over at least three dimensions: the characteristics of beneficiaries, the characteristics of health gains, and other important contextual factors.

In this review we search for studies using empirical data to elicit preferences across attributes for use in health care priority setting. The objectives of the review are to identify a list of attributes that can be classified into the three dimensions, to extract distributional weights, and to review preference elicitation methods and approaches that have been used for estimating distributional weights. General patterns from the data are inferred, pros and cons of each method are discussed, and future research directions are suggested.

The empirical evidence in these papers has concentrated mainly on age and severity. Overall, size of health gain, severity, age, lifestyle, and having dependants have been found to be important factors for prioritisation; gender and socioeconomic status have been found to be less important; and a premium for end-of-life treatment has not always been supported. It is implausible to synthesise results and draw firm conclusions because of heterogeneity in the use of approaches and questionnaires. Only a small number of studies estimated distributional weights, mostly for age or/and severity. Neither an agreed set of weights nor an agreed method has emerged from the literature. Early studies typically examined the importance of factors in isolation while the recent trend has been towards more realistic scenarios where multiple attributes are allowed to vary.

Assessing changes in stroke survival for the Northern Territory public hospitals, 1992-2013: A marginal structural analysis
Authors: Yuejen Zhao

Objective: To investigate changes in stroke long-term survival for the public hospitals in the Northern Territory (NT) using routinely collected hospital data.

Design and setting: a longitudinal study of stroke patients consecutively admitted to the public hospitals between 1 July 1992 and 30 June 2013.
Main outcome measures: Stroke survival hazard ratios (HR) by time trend, key demographics, stroke subtypes and comorbidities. Kaplan-Meier method and proportional hazards Cox regression were used for survival analysis. A marginal structural Cox model was applied to adjust for time-dependent age confounder and loss to follow-up.

Results: There were 4754 stroke patients admitted over the study period. Of the total, there were 3540 (74.5%) new cases and 837 (17.6%) recorded deaths due to stroke. The distribution of stroke type was ischaemic stroke 46.9%, haemorrhagic stroke 23.8%, and type undetermined 29.3%. Age of stroke onset among Indigenous patients was 51.7, 12.3 years younger than non-Indigenous patients (64.0). The marginal structural survival analysis indicated that Indigenous stroke patients were more likely to die than non-Indigenous patients (HR=1.65, P

Childhood Injury and Family Socio-Economic Determinants: A Longitudinal Analysis in Australia
Authors: Anthony Niu; David Johnston; Nicole Au; Michael Shields

Injury is the leading cause of mortality and morbidity among children and adolescents around the world. A substantial body of empirical literature documents a strong inverse relationship between family socio-economic status (SES) and child injury occurrence. There is, however, a significant lack of longitudinal studies. This paper endeavours to fill this research gap, and examines the potential parental SES gradients in childhood injury in Australia. We employ a four-wave panel data set from the Longitudinal Study of Australian Children (LSAC). Based on different measures of injury, we utilise a number of fixed-effects panel-data models, and include a comprehensive set of socio-demographic controls. This empirical strategy allows us to account for unobserved child- and family-specific characteristics in the estimation. Our results overall indicate that there is no strong family SES gradient in child injury in Australia. We do, however, observe some unanticipated positive income effects on the likelihood of child injury. A child raised by a high income family is estimated to have a 40% higher odds of suffering an injury than a child from a very-low income family. Further, this estimated marginal odds ratio more than doubles to 94% if we limit the estimation to fractures and dislocations. Our findings are inconsistent with many previous international studies, which may be explained by differences in empirical methods employed and study contexts. Methodologically, our study firstly controls time-invariant individual heterogeneity and thus produces more robust and consistent estimates. On the contextual side, Australia has already developed the best child injury prevention systems in the world, which may explain the absence of socio-economic differentials in child injury risk propensity in Australia.

A New Cost-effectiveness Model of Screening for and Treatment of Osteoporosis is Constructed and Validated
Authors: Lei Si; Tania M. Winzenberg; Qicheng Jiang; Andrew J. Palmer

Aim The objective of this study was to document and validate a new cost-effectiveness model of screening for and treatment of osteoporosis.
Methods A state-transition microsimulation model with a lifetime horizon was constructed that captured 5 competing events (hip fracture, vertebral fracture, wrist fracture, other fracture, death). Tracker variables were used to record the evolving simulated patient’s history, such as fracture events, duration of treatment and time since last screening. To test the validity of the model, an osteoporosis screening and treatment strategy for Chinese post-menopausal women was adopted. Goodness of fit analysis was performed for internal validation by comparing the fracture incidences from model prediction with model input data. Additionally, sensitivity analysis was undertaken for internal validation to evaluate whether change in cost-effectiveness was coherent within the plausible range of assumed parameters. External validity was tested by comparing predicted life expectancy and lifetime hip, vertebral fracture risk with the corresponding reported values.

Results Age-specific hip, vertebral and wrist fracture incidences were accurately reproduced compared to data used in model input (the regression line slope was 0.99, R2=1.00). Cost per quality-adjusted life-year gained by the osteoporosis screening and treatment strategy was estimated at $28,168 in base-case analysis, moreover, it changed consistently in deterministic and probabilistic sensitivity analysis. Predicted life expectancy and 10-year fracture risk for Chinese women at age 65 was 19.49 years and 9.3% and predicted lifetime hip fracture and vertebral fracture risk at age 60-year were at 4.3% and 30.7% respectively. These were all consistent with reported values, demonstrating the good external validity of the model.
Conclusions Our model was reliable with good internal and external validity, model reliability ensures it’s validity in economic evaluations in osteoporosis.

Instruments used to measure quality of life in older people – a systematic review
Authors: Norma B. Bulamu; Billingsley Kaambwa; Julie Ratcliffe

Aim: This paper describes the methods and results of a systematic review to identify instruments that have been used in measuring quality of life outcomes in older people, with a primary focus on preference based instruments suitable for application within economic evaluation in the aged care sector.

Methods: To identify instruments used to measure quality of life in older people, online databases searched were PubMed, Medline, Scopus, and web of science, PsycInfo, CINAHL and informit. Studies that met the following criteria were considered: 1) study population exclusively above 65 years of age 2) measured quality of life outcomes as indicators of health status or HRQoL through use of an instrument, and 3) published in journals in the English language after 1990.

Findings: The most commonly used generic preference based instruments were the EQ-5D, HUI3, and AQOL. Of the older people specific instruments, the ICECAP-O was the most commonly used instrument followed by the WHOQoL-Old, the OPQOL the CASP-19 and the ASCOT.
Conclusion: In the absence of a single ideal instrument for cost utility analysis undertaken in the aged care sector, this review recommends the use of a generic preference based instrument, the EQ-5D to obtain QALYs, in combination with the OPQOL or the ICECAP-O or the ASCOT to facilitate measurement of broader quality of life as defined by older people.

Exploring heterogeneity in count data: A study of PBS utilisation in children with language impairment.
Authors: Paula Cronin


The use of count data to explore health utilisation is well documented and poisson regression models provide a standard framework for this type of analysis. In this paper I review and compare methods that deal with two forms of heterogeneity or over-dispersion in panel count models.I explore the Longitudinal Study of Australian Children K dataset to analyse one count variable- PBS utilisation. A central focus of this investigation was whether children who are screened as having language impairment (LI) have higher PBS utilisation and over what time periods.

This study seeks to explore the determinants of health care in different parts of the utilisation distribution. It exploits the panel nature of the data over five waves and provides insights into the timing of health care utilisation. It is the first Australian study that we are aware of that addresses this issue in an econometric framework.

A sample from the Longitudinal Study of Australian Children (LSAC) (n=3490, t=5 ) was estimated using negative binomial (random effects) models. I compare these models with the pooled poisson (robust SEs), pooled negative binomial and zero inflated negative binomial model to study the effects of language impairment (measured by PPVT (Peabody Picture Vocabulary Test ) on PBS utilisation (no of scripts).

The negative binomial (random effects) with Mundlak adjustment outperforms all models and the results show that having a positive language impairment screen at 4/5yrs is a significant predictor of between 0.76 and 1.33 higher PBS scripts for children in the highest income households. This is despite the lower income children reporting the highest rates of LI (36% screened). The results point to the existence of inequity in health care utilisation, which is found to be distributed according to wealth rather than according to need.

The Impact of War on Human Capital of Poor Household Children in Ethiopia: Evidence from Natural Experiment Using Ethiopian-Eritrean War
Authors: Samuel Weldeegzie

Many developing countries are exposed to various shocks, both natural and human initiated. War is man-made shock that can pose serious risk to the wellbeing of individuals, and may have long term intergenerational implications, as it may damage early childhood human capital, with possibly lasting impacts by inducing intergenerational poverty trap. Yet, despite these concerns, there has been relatively little attention in the empirical literature.

In May 1998 a border dispute between Ethiopia and Eritrea resulted in a war lasting until June 2000, which, besides the loss of life, has come with significant economic cost and affected individual livelihoods in both countries. However, the specific effects on human capital of a child born in Ethiopia still remain unclear, partly because of limited availability of suitable data.

This paper uses the Young Lives data set for Ethiopia containing information on an array of child, maternal and household characteristics that track the development of young (born after war) and old (born before war) cohorts of children over time. And by combining event data with location, a difference-in-difference approach can be used to assess whether the war has had a significant impact on child human capital and whether this impact was different for boys and girls.

The results show that children who resided in the war region (and survived) have, on average, 0.31 lower standard deviations height-for-age z-scores (2.31cm), as compared to children that were not directly exposed to the war, slightly higher for boys than for girls. This is a disconcerting finding, especially for children from poor backgrounds, as early life outcomes can have lasting impacts on education and earnings during adulthood.
The findings suggest that interventions that try to address the lives of people who reside in conflict or war affected areas should also focus on health and nutrition of children.

Price Offers of Pharmaceutical Procurement in China: Evidence from Guangdong Province

Using a sample from the pharmaceutical procurement in Guangdong over the period from 2007-2009 and data of all the pharmaceutical manufacturers, this paper analyzes the determinants of bidding patterns in the presence of sample selection. Bidding behaviors between different groups are also examined, and price differentials are further decomposed to explain factors that account for the gap. The study finds the evidence that high level of competition and the capability of producing varieties of drugs drive manufacturers to bid lower. Winning experience enables bidders to behave aggressively. A longer distance from the procurement location relates to higher bids. The decomposition results indicate that the characteristics of bidders are not the main reason for the large price differentials between groups.

Are fruit and vegetables good for our mental and physical health?
Authors: Redzo Mujcic

Education and Health Knowledge: Evidence from UK Compulsory Schooling Reforms
Authors: Agne Suziedelyte

The economic costs and consequences of providing antenatal lifestyle advice for women who are overweight or obese: the LIMIT randomised trial
Authors: Prof. Jodie Dodd; Sharmina Ahmed; A/Prof. Wendy Umberger; Prof. Jon Karno; Andrea Deussen

Maternal obesity is associated with adverse pregnancy and birth outcomes. The objective of this study is to evaluate the overall cost effectiveness of antenatal dietary and lifestyle interventions for pregnant women who are overweight and obese pregnant. The study uses data from a multicentre randomized trial conducted in three hospitals in South Australia. Women having a singleton pregnancy, between 10-20 weeks’ gestation and BMI≥25 were randomised to receive either standard care (n = 1104) or a comprehensive dietary and lifestyle intervention (n = 1108). Cost effectiveness analysis was completed considering two significant clinical outcomes of the trial: infants born to women receiving lifestyle advice were significantly less likely to have birth weights above 4000g and were.less likely to suffer moderate to severe respiratory distress syndrome (RDS) compared to infants born to women receiving standard care. Direct out-patient, in-patient and intervention costs for up to 4 weeks after birth of the baby were calculated. The SF-36 QoL was completed by participants at trial entry, 28- and 36 weeks of gestation and 4 months post partum. No significant mean cost differences were found between the two groups. However, the cost effectiveness acceptability curve (CEAC) illustrates that if $20,000 is the threshold value (i.e. the maximum policy makers would be willing to pay), then the probability of lifestyle advice being a cost effective intervention for reducing the risk of infants with birth weight above 4000g is 0.85 compared to 0.15 for standard care. Similarly, using the same threshold value, the probability of lifestyle advice being a cost effective intervention for avoiding an infant suffering from moderate to severe RDS would be 0.64 compared to 0.37 for standard care group. In terms of changes in QoL, the combined treatment and time effects were not significantly different between two groups.