The University of Sheffield
Health Economics and Decision Science

Possible PhD topics

Specific PhD Topics

Measuring health and well-being for public policy

John Brazier

We undertake a wide programme of work in the area of health and well-being in ScHARR and at the Centre for Well-being in Public Policy. We could offer supervision to post graduate research in two areas.

One is concerned with developing tools for measuring and valuing well-being beyond the usual confines of the usual health and health related quality instruments used in health related research. The aim is to develop tools that can be used in economic evaluation for informing public policy. There is interest in supervising post graduate students seeking to develop tools in the area of mental health, public health, and well-being more generally including preference and experience based measures and those concerned with the concept of capabilities. Such work would use and adapt techniques developed in the health arena.

A second area is more concerned with examining the causes of well-being through the analysis of secondary data sets and/or by undertaking micro studies with people experiencing changes in their health and other aspects of their lives, and following through the impact on well-being. Such work may use advanced qualitative and quantitative methods of research.

Equity/ethics/inequalities

Aki Tsuchiya

Health care resources are limited and therefore need to be allocated efficiently, but they also need to be allocated in a fair, equitable manner. One topic where the possible conflict between efficiency and equity occurs is that of severity and disability. Should we, or should we not, give higher priority to improving the health of people who suffer from chronic and severe health problems, even when their treatment is not as cost effective as other treatments? And if yes, for what reason, and by how much? This is an area where the views held by the general public may differ markedly from the views held by those who are actually living with the health problems in question. This leads to two further research questions. First, if people with such health problems adapt to their condition, does this mean they no longer need to be given higher priority, compared to those who (for whatever reason) do not adapt, and why? Second, if general public and patients have different views, then there is an issue of procedural justice: whose views should public policy be based on, and why?

Uncertainty in utilities

Tracey Young

A key problem with preference based measures such as the EQ-5D, SF-5D, HUI, and AQL-5D are that these measures define a large number of unique health states and the consequent need to model health state values from a valuation of a subset of possible states. Health state values present a significant challenge for conventional statistical modelling procedures due to their nature, namely: skewed, truncated, non-continuous and hierarchical. Alternative approaches to modelling health state values using a (nonparametric) Bayesian method and comparing this to conventional regression based methods are currently being explored however, further developmental work into methods for more accurately modelling, for example ordinal data, are needed. A further issue of exploration is developing methods of mapping between non-preference based and preference based measures that reliably generate utilities across the full utility scale and allow for uncertainty in the generation process. It is also important to develop methods that estimate the uncertainty around health state values, combined with other sources of uncertainty, and there use in cost-effectiveness models, the extrapolation of published values to study populations in cost effectiveness models and the analysis of missing or censored outcome data.

Development of descriptive systems for preference based measures

Tracey Young

Existing health related quality of life (HRQL) instruments, for example the SF-6D or the EQ-5D, are not always the most appropriate measure for assessing HRQL in a predefined population, reasons for this include; lack of an appropriate instrument, lack of sensitivity of an instrument, or instruments not being designed to deriving preference based estimates. In these situations new descriptive systems need to be developed from scratch or developed to be used as part of existing HRQL measures. In the development of new descriptive systems qualitative methods will be used to develop descriptive systems, either through the development of dimensions directly or through the generation of pools of items which can then be refined using Rasch and/or factor analysis. The qualitative methods will involve using either focus groups or in depth interviews with patients or the general population to determine the content of a descriptive system and also perhaps further qualitative work to test the descriptive system. Alternatively, descriptive systems can be designed from existing HRQL measures and methods such as Rasch, factor and cluster analysis can be used together alongside other traditional psychometric techniques in the development process. Opportunities exist to refine existing methods and develop new methods for the development of descriptive systems. For example traditional methods of valuation are based on time trade off methods involving the general public, using these methods only a limited number of possible health states can be valued and methods need exploring and developing to minimise the loss of information for the remaining, unvalued, health states, this could involve a regression based mapping exercise from the preference-based scale to the Rasch latent variable scale.

The relationship between discrete choice experiments and QALYs

Donna Rowen

Applications are invited from capable and enthusiastic prospective PhD students to investigate the methodological underpinnings of discrete choice experiments (DCEs) and/or ordinal valuation methods in health care, the methodological underpinnings of quality-adjusted life-years (QALYs), and the extent to which DCEs and/or ordinal valuation methods may be used as a vehicle for estimating utility weights for QALY calculations. Utility weights generated for specific health states relating to several conditions will be generated using a DCE format and then compared to those generated using more standard scaling techniques e.g. standard gamble and time trade-off. The research will discuss the relative advantages and disadvantages of each technique and assess their performance in terms of 1) the validity of the weights generated and 2) their applicability for use in the estimation of QALYs.

Value of Information methods in more complex settings

Alan Brennan

Value of information (VOI) analysis has proven to be sensible and coherent when you have a limited number of strategies for investment and a clear metric for evaluating these, for example cost per QALY of alternative treatments in a specific disease. However, much research investment occurs earlier than the point where that clear decision exists, for example, drug development may be targeted at more than one disease indication. This leads to developing methods to extend VOI to these circumstances, including having multiple objectives and sequential data collection processes.

General Topics

  1. Developing methodology in the valuation of health-related quality of life. Who should be asked? How? Research combining the development and testing of methodologies with case studies concerning nationally or internationally relevant health policy questions is preferred.
  2. Modelling the cost-effectiveness and service impact of alternative interventions and policies using simulation and other approaches.
  3. Bayesian Statistics in Health Economics