Time Trade-Off (TTO): Eliciting Health State Utility Values

The time trade-off (TTO) method is one of two elicitation techniques — the other is the standard gamble — used to derive the health state utility values that quality-adjusted life year (QALY) calculations depend on. A utility value anchors a health state on a scale where 1.0 represents full health and 0 represents death, with some elicitation protocols permitting negative scores for states a respondent rates as worse than death. TTO is also the method behind most of the standardized national value sets used in health technology assessment — including the EQ-5D tariffs that underpin NICE’s technology appraisal process — which is why the mechanics of the task matter even to researchers who never run an elicitation study themselves and instead apply an existing published value set inside a CHEERS-reported cost-utility analysis.

What the TTO Elicitation Task Asks a Respondent to Do

In its classic form, developed by George Torrance in the early 1970s and refined through the 1980s, the TTO task presents a respondent with two alternatives that both end in death: living a fixed duration T (conventionally ten years) in a specified impaired health state, or living a shorter duration x in full health. An interviewer varies x — typically through a titration (“ping-pong”) procedure that narrows in from above and below — until the respondent is indifferent between the two options. At that indifference point, the ratio x/T is taken as the utility value for the health state: a respondent who would give up two of ten years to live in full health instead of a described impaired state is implying a utility of 0.8 for that state.

Because both options are certain outcomes differing only in how long they last, TTO asks the respondent to reason about duration and trade-off, not about probability. That is generally considered an easier and more intuitive task than the standard gamble, described below, and is a large part of why TTO became the standard method for large-scale general-population valuation surveys, where the same protocol has to be delivered consistently by non-specialist interviewers to thousands of respondents with no background in expected-utility reasoning.

Handling Health States Rated Worse Than Death

A basic TTO trade can only produce values between 0 and 1, because giving up years from a positive baseline duration cannot express “worse than being dead.” Valuation studies — including the EuroQol Group’s original UK TTO study behind the EQ-5D-3L value set — handle this with a modified procedure: once a respondent indicates the impaired state is worse than death, the choice is re-posed as a trade between a period in the impaired state followed by death, and a shorter period of immediate death. A corresponding formula converts this second trade onto the same scale, extending it down to -1. That is why published EQ-5D value sets routinely list health states with negative utility scores, and it is worth checking a value set’s own documentation before assuming 0 is its practical floor.

TTO Versus the Standard Gamble

The standard gamble (SG) asks a structurally different question. Instead of trading duration, the respondent chooses between living the remainder of their life in the impaired health state for certain, or accepting a gamble with probability p of full health and (1-p) of immediate death — and p is varied until the respondent is indifferent. The two methods differ in ways that matter for how their results should be used:

• Respondent burden. TTO’s certain-outcome, duration-based framing is widely reported as easier for respondents to understand and complete than SG’s probabilistic lottery, which asks people with no statistical training to reason consistently about risk — a demand that shows up as more erratic or inconsistent responses in general-population interviewing.
• Theoretical grounding. This is where SG has the stronger direct claim. The standard gamble literally is a choice under uncertainty, so its indifference point maps directly onto a von Neumann–Morgenstern expected-utility index — the theory the task was built to instantiate. TTO’s trade is a choice under certainty, so treating its output as an expected-utility-theory utility requires an added assumption (that the QALY model is linear in duration, i.e. “utility independence” between health state and time), which the task itself does not test and need not hold.
• Empirical divergence. The two methods do not reliably return the same number for the same described health state. A substantial comparative literature finds standard gamble valuations tend to run higher than time trade-off valuations for equivalent states, a pattern generally attributed to probability weighting and risk attitude entangled in the SG task versus time-preference distortions embedded in a duration trade. Neither method is a bias-free measurement of one underlying construct, which is part of why HTA bodies specify which single method and protocol a reference-case value set must use rather than treating TTO and SG results as interchangeable.

Why Standardized Value Sets Use TTO in General-Population Samples

The EuroQol Group’s original Measurement and Valuation of Health (MVH) study elicited TTO utility values from a representative sample of the general adult population — not from patients living with the health states being valued — to build the first EQ-5D-3L value set. The same underlying design, TTO-based valuation of a representative population sample, has since been repeated country by country, because a value set calibrated on one country’s preferences is not assumed to transfer cleanly to another’s. For the newer, larger EQ-5D-5L descriptive system, the EuroQol Valuation Technology (EQ-VT) protocol pairs composite TTO with a discrete choice experiment component, a design related to the discrete choice experiments used elsewhere in preference research, to model values across the much larger number of health states a 5-level instrument generates.

Sampling the general population rather than affected patients is a deliberate methodological choice, not a shortcut, for two standard reasons: a QALY used to inform a shared healthcare budget decision is meant to reflect societal preferences about health states, not only the preferences of people currently adapted to living in one of them (patients frequently report higher utility for their own condition than an outside observer would predict, a well-documented adaptation or “response shift” effect); and using one standing, published, general-population value set — rather than re-eliciting utilities from whatever sample happens to be in a given trial — is what makes QALYs comparable across unrelated disease areas and different studies, which is the entire premise of cost-utility analysis as a resource-allocation tool. NICE’s reference case for technology appraisal, and ISPOR’s good-practice guidance on health-state valuation, both specify essentially this workflow: collect a validated descriptive-system instrument such as EQ-5D-5L or the SF-36-derived SF-6D from participants, then apply the appropriate published general-population value set to convert their descriptive profiles into utility scores.

Practical Guidance for Designing or Reviewing a Utility-Based Study

If a study needs health state utilities, the default path is not to run a fresh TTO or SG elicitation. It is to collect a validated patient-reported outcome measure with an existing value set — EQ-5D-5L is the default choice for most HTA submissions — and score responses against the value set for the jurisdiction the analysis is meant to inform. Running an original elicitation study is reserved for cases where no adequate value set exists for the relevant health states or population, since primary valuation work is its own substantial undertaking with its own sampling, training and consistency requirements.

When reviewing someone else’s cost-utility analysis, check which value set was applied and to which country’s population it was elicited from, and confirm that matches the jurisdiction the analysis is meant to inform. Applying, say, a UK TTO-derived tariff to a trial intended to support a decision in a different health system is a real methodological choice with consequences for the resulting incremental cost-effectiveness ratio, not a formality to skip past in a methods section.

Frequently Asked Questions

What is the time trade-off method used for?

TTO elicits a numeric health state utility value — a weight between roughly -1 and 1 — by asking a respondent how many years of full health they would consider equivalent to a longer period spent in a specified impaired health state. Those utility values are the weighting factor QALY calculations multiply against time spent in a health state.

Is the time trade-off method the same as the standard gamble?

No. Both elicit health state utility values, but TTO poses a choice between two certain outcomes of different durations, while the standard gamble poses a choice between a certain outcome and a probabilistic gamble between full health and death. TTO is generally easier for respondents to complete; the standard gamble has the more direct theoretical grounding in expected-utility theory, because it is itself a choice under uncertainty.

Can a time trade-off utility value be negative?

Yes. Standard TTO protocols cannot express “worse than dead” using the basic duration trade, so valuation studies use a modified lead-time procedure for respondents who consider a state worse than death, extending the scale down to -1. Published value sets, including EQ-5D tariffs, list some health states with negative utility scores.

Why don’t researchers just elicit new utility values for every study?

Standardized, general-population value sets exist specifically so QALYs are comparable across different diseases, interventions and studies — the core purpose of cost-utility analysis. Re-eliciting utilities from each trial’s own (often patient) sample would reintroduce the adaptation and small-sample variability that standardized value sets are designed to avoid, and HTA reference cases generally require an established value set unless none is adequate for the health states in question.

What kind of sample are EQ-5D value sets built from?

A representative sample of the general adult population of a given country, not patients with the specific conditions being valued. This reflects the reasoning that a value set used to inform societal healthcare resource allocation should represent broad societal preferences about health states rather than only the preferences of people currently living in them.