Note from the National Guideline Clearinghouse (NGC): The National Institute for Health and Clinical Excellence (NICE) commissioned an independent academic centre to perform a systematic literature review on the technology considered in this appraisal and prepare an assessment report. The assessment report for this technology appraisal was prepared by the Centre for Reviews and Dissemination, University of York (See the "Companion Documents" field.)
Data Extraction Strategy
Data extraction was conducted by one reviewer using predefined data extraction forms in a Microsoft Access database and checked by a second reviewer. Any disagreement was resolved by consensus and, if this was not reached, a third reviewer was consulted.
The type of data that was extracted and summarised included specific details about the interventions, the population investigated, and the outcome measures used. Trials that had been reported in multiple publications were collated and reported only once.
Where sufficient data were presented, an estimation of the treatment effect along with the 95% confidence interval (CI) was calculated for each individual trial. Where possible this was done on an intention to treat basis. For dichotomous outcome measures the relative risk (RR) or hazard ratio (HR) was calculated and for continuous outcomes the median or mean difference (MD) was used. For survival data or other time-to-event data the hazard ratio was reported where presented in the included trial. If Kaplan Meier curves were presented, the p value of the log rank test was presented, where performed. Median survival times were also reported, where given in the trial.
In order to assess the economic data in terms of the clinical effectiveness of the intervention (i.e., the direction of the cost-effectiveness data and the magnitude of effectiveness data), each trial was given a summary grading (A-I) according to the level and direction of dominance (i.e. whether the intervention of interest should be preferred over the comparator). Extended dominance indicates that both the effectiveness data and the economic data support the use of either the intervention or the comparator and the decision on resource allocation is clear. When either the economic or the effectiveness data supports the intervention/comparator, the dominance is said to be partial or weak and a decision can still be made. However, if there is no dominance indicated then further incremental cost analysis may be required in order to estimate the incremental cost-effectiveness ratio. This is important in helping the decision- making process. Figure 1 in the Assessment Report (see the "Availability of Companion Documents" field) illustrates all of the possible permutations, and was used to assign each trial a summary grading.
Quality Assessment Strategy
The methodological quality of each included trial was assessed using predefined checklists. Two reviewers conducted this process independently. Any disagreements were resolved by consensus and a third reviewer was consulted if required. Quality criteria included method of randomisation, allocation concealment, baseline comparability of identified prognostic characteristics (which were identified as being treatment free interval, disease bulk, number of previous regimens, age, histology and performance status), presentation of eligibility criteria, reporting of co-interventions, loss to follow-up <20%, handling of withdrawals and use of intention to treat analysis. Blinding was also assessed, although it is acknowledged that blinding is often impossible in trials of cancer treatment.
Methods of Analysis/Synthesis
Results of data extraction and quality assessment are presented in structured tables and also as a narrative summary. Where new trials were found which impact on the results of the original review, the results of the original review are also presented.