|| Checking for direct PDF access through Ovid
Since most advanced cancers are still incurable, oncologic clinical research pays considerable attention to palliation, increasingly valuing subjective measures of outcome such as quality of life (QoL). We reviewed randomised clinical trials (RCT) of cytotoxic or hormonal treatments in advanced breast cancer (ABC), published before December 2003, to evaluate the methodological quality of QoL assessment and assess its added value (over classical clinical endpoints (CCE), i.e. survival, response, time to progression, toxicity) in the choice of the best treatment option.RCTs were classified according to treatment characteristics and the CCEs. A descriptive analysis was based on the methodological aspects of QoL assessment and the clinical value of QoL findings was judged by counting the frequency of reporting in the study abstracts and the assessment of QoL combined with CCEs.We retrieved 33 eligible RCTs (10,791 patients); only 20 reported the number of patients considered in QoL principal analysis and only 69% of randomized patients were included in such analyses. A total of 17 different QoL questionnaires were used, 11 only once. QoL assessment lasted from less than 12 weeks to progression, and timing of questionnaires from 2 to 12 weeks. Compliance rates were 85.7% for baseline forms and 67% for overall assessment, but this information was available for only 18 and 20 trials, respectively. Wide variability emerged in analysis strategies and statistical approaches. QoL findings were reported in 12 study abstracts (37% of patients). Eight studies reported a significant difference in QoL scores but since QoL data often failed to parallel the clinical findings (e.g. better QoL scores were reported in two of 17 trials with better CCEs and in six of 20 with significant differences in toxicity profiles), the QoL added value was difficult to ascertain and, on the whole, only moderate.In ABC trials, QoL assessment added relatively little value to CCEs in helping select the best treatment option, apparently largely because of sub-optimal methodological standards.