Past CHE Economic Evaluation Seminars 2009

10 December 2009

Title: Mapping specific disease QOL profiles to Utilities: the example of the EORTC QLQ-C30
Speaker: Ralph Crott, Centre for Reviews and Dissemination, University of York

Abstract:
Background: Although cancer-specific Health-Related Quality of Life (HRQoL) are commonly included in randomized clinical trials or other prospective non-randomized clinical studies, it is rare that preference-based instruments is used that allow the calculation of a Utility weight suitable for in estimating Quality-adjusted Life-Years (QALYs) gained.

Objective: To develop a mapping algorithm to transform the EORTC QLQ-C30 questionnaire responses into EQ-5D derived Utilities.

Study Design: Retrospective data analysis of a multicentre, multicountry prospective clinical trial in breast cancer patients.

Methods: Regression analysis of individual pairs of EQ-5D and QLQ-C30 scores.

Results: A model that explained 80% of the variance was developed to estimate EQ-5D Utilities from QLQ-C30 scores at individual level. From these reliable group level means and deviations can be derived.

Conclusions: Mapping from QLQ-C30 scores to EQ-5D derived utilities when only QLQ-C30 data are available has been shown to be possible with good accuracy. Validation of the proposed algorithm in other external clinical datasets should be encouraged.

19 November 2009

Title: Cost effectiveness and cost equity analysis of uncomplicated malaria management in Africa: A pilot study
Speakers: Manuel Espinoza and Susan Griffin, CHE

Abstract: Health inequalities have become a major concern among decision makers. However, few studies have incorporated equity concerns into the economic evaluations of health care technologies. The current study addresses a cost-effectiveness analysis of different strategies to manage a patient with uncomplicated fever (suspected malaria) including diagnostic and therapeutic options. In addition, a cost-equity analysis has been performed considering differences in access to health clinics and adherence to treatment by socio-economic status. Inequalities were measured using a Concentration Index and Equity Ratios. The opportunity cost was estimated comparing the most cost-effective alternative against the most cost-equitable strategy in terms of benefits forgone to achieve a certain number of equality units. A hypothetical therapeutic alternative associated with 100% adherence was incorporated into the model in order to have a better representation of the contrast between equity and efficiency. Twenty strategies were included in the cost-effectiveness analysis. Ten strategies were selected for the equity analysis. The opportunity cost between efficient and equitable strategies was estimated for a set of threshold values. Implications of these results in terms of the equity-effectiveness loop framework will be discussed.

13 October 2009

Title: Fragile - handle with care! Issues in the long-term extrapolation of individual patient level data for cost-effectiveness modelling
Speaker: Andrea Manca, Centre for Health Economics, University of York
Authors: A Manca, P Saramago, MJ Sculpher, P Royston, M Parmar, A Copas

Abstract:
Background
Much of the evidence informing cost effectiveness models comes from RCTs which are short in durations. However, resource allocation decisions often require long term estimates of costs and quality adjusted survival. Thus, researchers are expected to extrapolate RCT results beyond the observed period, with or without the benefit of additional external information. Survival regression models (SRMs) can naturally be used to implement such an extrapolation. Two issues must be considered though when carrying this out. First, how well SRMs fit/predict the RCT data, and second, the extent to which their results can be used to inform decision models aimed at producing (beyond trial) long-term cost-effectiveness estimates.

Aims
This paper identifies a set of desirable features SRMs should have for use in cost effectiveness modelling. Distinguishing between within and beyond-trial prediction, the manuscript discusses the extent to which competing SRMs satisfy the proposed 'wish list'. The candidate SRMs are tested using a real life dataset. Methods SRMs for IPD and probabilistic decision models.

Data
RITA3 trial.

Results
SRMs for cost effectiveness modelling must be able to capture the time-varying nature and shape of the baseline risk, as well as estimate the effect of covariates on the hazard rate. Standard parametric SRMs may be too rigid. The Cox model is limited by how it handles baseline hazard. The Royston Parmar flexible parametric model provides a middle ground solution and has a series of desirable features for use in CEA. Extrapolation beyond trial follow up for cost effectiveness purposes, though, ideally requires mature survival data and external information.

Conclusions
Beyond trial extrapolation of time-to-event data without external information must be handled with care and the uncertainty associated with extrapolation should be captured as fully as possible in models.

17 September 2009

Title: Appropriate perspectives for health care decisions
Speaker: Simon Walker, Centre for Health Economics, University of York

Abstract: Decisions based on cost-effectiveness analysis compare the health benefits expected to be gained by using a technology with the health that is likely to be forgone as a result of the displacement of other activities that improve health due to the additional costs falling on the health care budget. This approach to making decisions in health care will be suitable if the social objective is to improve health; that the measure of health gained and forgone captures enough socially valuable aspects of health to be useful; and that the budget for health care ought to be regarded as fixed.
However, a broader societal perspective would also consider effects outside the health care sector including: the direct costs of care borne by patients and carers that do not fall on the health care budget, and the indirect external effects on the rest of the economy.

The problem for policy is that, in the face of budgets set by a socially legitimate higher authority (government), it is not clear how or whether a broader social perspective, which would include all these effects on all sectors, should be implemented; particularly if transfers between sectors are not possible. It poses the question of how the trade offs between health, consumption and other social arguments, as well as the valuation of
market and non market activities, ought to be done. This is particularly
acute when there is no consensus on how to prescribe social choice; each alternate view generating potential conflicts with other agreed social objectives.

The seminar will briefly cover the current policy environment in the UK and the rest of the world. A framework for examining the introduction of health care technologies be introduced and various decision rules will be analyzed with regards to both marginal and non marginal changes. Issues relating to questions of value judgements and facts will then be discussed. Four case studies will be presented with results from both a health care and a societal perspective.

21 July 2009

Title: Towards a more universal approach in health valuation
Speaker: Benjamin M. Craig, Moffitt Cancer Center and University of South Florida

Abstract: By polling individual responses to hypothetical scenarios, valuation studies estimate population preferences toward health on a quality-adjusted life year (QALY) scale. The scenarios typically involve tradeoffs in time (time trade-off, TTO), risk (standard gamble, SG), or number of persons affected (person trade-off, PTO). This paper provides a parsimonious theoretical framework that unites TTO, SG and PTO techniques under a common estimator that avoid the arbitrary replacement of worse-than-dead (WTD) responses.

As an example, we estimate QALY values for 242 EQ-5D health states using time trade-off responses from the seminal UK Measurement and Valuation of Health (MVH) study. The revised estimates helped identify an average bias of
0.21 in the original UK estimates. The rule of thumb is that if the original UK QALY value is less than 0.5, add 0.25 (e.g., 0.25 QALY becomes 0.5 QALYs).

The procedures of data manipulation and resulting bias (known to some as the
N3 coefficient) have been well-recognized and critiqued in the field, yet they remain the conventional standard in the analysis of TTO, SG and PTO responses. Once forgone, trade-off QALY estimates demonstrates greater convergent validity with VAS and ordinal QALY estimates.

Link to paper: http://www.pophealthmetrics.com/content/7/1/3

18 June 2009

Title: Impact of structural assumptions within breast cancer natural history models on estimates of the benefits of screening
Speaker: Jason Madan, University of Sheffield

Abstract: A range of health economic models have been used to evaluate breast cancer screening programmes. These models vary in the way they represent breast cancer progression, and all have short-comings in terms of how well they represent current knowledge of cancer biology. To illustrate the impact of model structure on predictions of the survival benefit from screening, a range of natural history / survival models of increasing sophistication were fitted to a common dataset. These were combined with treatment/survival models matched to each structure, to assess the impact of model design on survival and resource use. A wide range of predictions of both intermediate parameters (sojourn time, test sensitivity) and survival were obtained. This demonstrates the dependence of modelling results on choices made in the model development process, and the importance of reviewing the clinical plausibility of mathematical statements in healthcare modelling.

13 May 2009

Title: The value of added years of life as a function of age, prognosis and quality of life
Speaker: Professor Ben Van Hout, Pharmerit Ltd, UK.

Abstract: OBJECTIVE: Do people weigh gains in life years differently when patients differ in age (but not in life expectancy), life expectancy (but not in age) or QOL (but not age or life expectancy)?

METHODS: Trade off questions were developed searching for indifference between giving healthy life years to patients with different ages, prognoses and quality of life. Data come from 46 heart failure patients, 60 healthy controls and 180 students. For age, as well as prognosis and QOL, 6 comparative sets were developed. Each respondent answered 2 questions of each set and 2 combination-questions. Ordered logistic regression was used in combination with conditional linear regression for 'extreme' answers. Answers are 'extreme' when, for example, 1 extra life year in a young patient is preferred to 10 in an old patient or when respondents can't choose.

RESULTS: More than 40% of the answers are 'extreme'. Respondent prefer to give to the young and to those who are worse off, either now (quality of life), or in the future (prognosis). Elderly individuals, more often, prefer not to prioritize. It is estimated that an additional life year in a 20 year old is worth 12.8 times the value of an additional year in an 80 year old. An additional life year given to someone with a life expectancy of 5 years is worth 2.12 times that of one given to someone with a life expectancy of 10 years. An additional life year in someone with a utility that is 0.25 lower than someone else is worth 2.45 times more.

CONCLUSIONS: All results indicate that people do not think that a QALY is a QALY and that the value of life years depends on the age of the respondent, the prognoses of the patients and the patients' current quality of life.

16 April 2009

Title: Assessing the acceptability and usefulness of formal methods of value of information analysis for informing resource allocation decisions for healthcare and healthcare research
Speakers: Claire McKenna and Susan Griffin

Abstract: At least three conceptually distinct but simultaneous decisions can be identified for those responsible for allocating resources within the healthcare system:

What intervention/technology should be adopted in practice given the existing evidence base and the uncertainty surrounding outcomes and resource use?
Is additional evidence required to support the adoption of a particular intervention/technology in practice?
Are implementation strategies required to get adoption decisions implemented into clinical practice?

Formal analytic methods have been developed to address these questions by establishing the cost-effectiveness of alternative healthcare programmes and the maximum returns to investment in further research and/or implementation efforts.

This study aims to assess these formal analytic approaches (value of information analysis and value of implementation analysis) by applying current methods to a case study. The case study relates to the CRASH trial. The CRASH trial was a high profile large randomised placebo controlled trial of the effect of early administration of a 48 hour infusion of corticosteroid (methylprednisolone) on the risk of death and disability after traumatic brain injury. Before CRASH, existing trials were too small to demonstrate or refute the possibility of a clinically important benefit of steroids on death and disability following brain injury. Steroid use was variable across the NHS for the estimated 10,000 persons experiencing severe brain injury per year. CRASH was needed to reliably establish whether there was a benefit or whether steroids were in fact killing patients.

A retrospective value of information analysis and cost-effectiveness analysis has been conducted to evaluate the use of steroids in brain injury. Using this we assess whether formal approaches could have aided decisions about steroid use by quantifying:

Whether there should have been widespread adoption or rejection of steroids in practice given the evidence that existed prior to the CRASH trial.
What value would have been placed on obtaining further evidence to inform that decision prior to the CRASH trial.
What value would have been placed on implementation strategies to reduce variation in practice prior to the CRASH trial.

The presentation is targeted to a non-technical audience. Before these formal methods can be used to inform resource allocation decisions in practice, they must be understood and viewed as useful by those with the remit to make those resource allocation decisions, many of whom may not specialise in health economics.

12 March 2009

Title: Comparing multiple treatment effects in observational studies with marginal structural models
Speaker: David Suarez, PhD, Epidemiology and Assessment Unit, Parc Taulí Foundation, Autonomous University of Barcelona, Sabadell, Spain.

Abstract: Observational studies represent the real clinical situation better than randomized clinical trials (RCTs). However, control of confounding in observational studies remains challenging. Here, we review marginal structural models (MSMs) and show how they are useful when comparing the effects of multiple treatments on outcomes in observational studies. To illustrate the application of MSMs when patients may receive several treatments, we have reanalyzed the effects of antipsychotic medication on achieving remission in schizophrenia using data from the SOHO study, a 3-year observational study of health outcomes associated with the treatment of schizophrenia. The MSM showed qualitative differences in some comparisons in which the conventional analysis obtained results that were not consistent with previous RCTs. MSMs might provide a better control for confounding than conventional methods by improving the adjustment for treatment group differences in observational studies, which may approximate their results to those of RCTs.