## Integration of survival and QL

The concept of combining QL and survival into a single statistic is very appealing, as it overcomes the subjective balancing of quality and quantity of life. An argument against simple survival analysis is that there are only two 'states,' alive and dead [29]. Amongst those alive a patient who is bedridden with severe symptoms is considered equivalent to one who is active and asymptomatic. Quality Adjusted Life Years (QALYs) are calculated on the basis that a value between zero and one can be assigned to various health states, and that the time spent in each state can be multiplied by this value to express survival in terms of QALYs.

The QALYs require firstly the definition of a number of health states, and secondly weights to be assigned to each. How to achieve this is not obvious, although some have attempted to do this [30]. Clinicians and trialists have in the main not embraced QALYs. There is a feeling perhaps that this is oversimplification, and there is still a desire to consider the various domains and items of QL separately and make individual treatment decisions, although this puts a particular onus on clarity of presentation of the results.

An alternative method of combining QL and survival is TWIST (time without symptoms or toxicity) which is a summation of survival time during which patients had no symptoms or toxicity [31,32]. It is calculated by subtracting from the overall survival time, periods of time when symptoms, or other clinical events, were present. This requires defining what events (or what severity of events) are relevant and what the time penalty for each should be. It is important that all patients are beyond the period of toxicity before calculations are performed, as different patterns of toxicity (i.e. early or late) maybe different in the different groups. A variant of TWIST is Q-TWIST (quality adjusted time without symptoms or toxicity) [33]. Here again health states that are clinically relevant to the disease and/or treatment are chosen (such as toxicity, asymptomatic, relapse), and scores between zero and one are assigned to each. The particular score assigned is called a utility, in effect the weight of importance placed on each health state, and these can be generated from patients, clinicians or previous work.

0 0