Comparing Variants of Lead and Lag Time TTO

As reported in an earlier post,  OHE was awarded a Policy Research Programme grant by the Department of Health that focuses on three aspects of health status indexes.  The second of these is a study on Time Trade Off (TTO) methodology, focusing on Lead and Lag Time TTO (LT-TTO). TTO is crucial because it helps translate patients’ EQ-5D data into values that can underpin health care allocation and access decisions. The National Institute of Health and Clinical Excellence (NICE), for example, uses TTO values in assessing new technologies.

The objective of this research was to develop a better understanding of the characteristics of the data generated by LT-TTO, improve data collection tools, and provide information for selecting the particular variant of LT-TTO to be used in subsequent research. Now available as OHE Research Paper 10/02, the results of this research already are having an important impact. Research to pilot the use of the methods reported in our paper now is underway in four countries.  An additional four countries will use our results as the basis for further methodological research.  These studies, coordinated by the EuroQol Group, will lay the foundations for future EQ-5D-5L value sets studies.

Background

The estimate of Quality Adjusted Life Years (QALYs) a patient gains from treatment requires that the length of life gained be adjusted by its quality – i.e., the perceived value of life to the patient experiencing a particular health state.  These values (weights) are anchored on a scale where 1 is full health and 0 is dead. Health states perceived to be worse than dead have values of less than 0.

The ‘Time Trade Off’ method is widely used to obtain these values, but it presents some important problems. In particular, the method cannot adequately handle very poor health states that people may consider to be so bad they are ‘worse than being dead’. In such cases, the TTO must switch to a different questioning process to capture value, creating problems for the comparability and interpretation of values less than zero. In previous research,  LT-TTO proved capable of producing weights for states both better and worse than dead.

Aims

The aims of this research were (1) to investigate the values generated from LT-TTO using different combinations of the length of time the individual experiences full health and a particular health state; the order of these states also was varied (Lead v. Lag Time TTO), (2) to gauge whether values generated from these methods concur with participants’ views as to whether the states are better or worse than dead, and (3) to explore a range of methods for handling the preferences of those whose distaste for very poor health states is so great that they are willing ‘use up’ all their lead time to avoid them.

Methods

A sample of 200 members of the general public valued five health states, using two of four variants of the LT-TTO: a lead time of 10 years followed by a health state duration of 20 years; a lead time of 5 years followed by a health state duration of 1 year; a lead time of 5 years followed by a duration of 10 years; and a duration of 5 years of a health state followed by with a lag time of 10 years. Participants also responded to a range of supplementary tasks and other questions.

Results

Values are influenced by the length of the lead time relative to the health state duration. Longer lead times enable somewhat more preferences to be captured, but appear to exert a framing effect on values. Lag time TTO results in both a greater willingness to trade off for mild states of poor health and trading off less time for severe states. Of those who valued the worst health state as less than 0, 70% also expressed the view that this state was worse than dead.

Conclusions

LT-TTO confers an important advantage over TTO by providing a single method capable of generating values greater and less than 0 that seem broadly in keeping with participants’ stated views about those states being better or worse than dead.  However, values are sensitive both to the length of time in full health relative to the duration of the state to be valued and to the order in which these occur (lead vs. lag time). For those who use up their lead time, we show that additional ways of eliciting these preferences (via additional questioning) are feasible, as is modelling those values (via survival analysis). A small (<5%) group of participants remain, however, whose preferences are so ‘extreme’ that they cannot be captured by any approach.

Download Devlin, N et al. (2010) A comparison of alternative variants of the Lead and Lag Time TTO. Research Paper 10/02. London: Office of Health Economics

Posted in EQ-5D and PROMs, Health Care Systems, Health Technology Assessment | Tagged Grants