Absolute Benefits of Medical Therapies for Breast and CRC
Absolute Benefits of Medical Therapies for Breast and CRC
Background: Phase III randomized clinical trials (RCTs) have become larger and are powered to detect small absolute benefits. Temporal changes in absolute benefits of experimental medical therapies reported in RCTs are unknown.
Methods: We identified all RCTs with sample size ≥200 evaluating experimental medical therapies for breast and colorectal cancer published from 1975 to 2007. We assessed changes over three decades in absolute differences in time-to-event end points between experimental and control arms by (i) the usual method (i.e. at one point) and (ii) as the area between time-to-event curves up to a predefined time.
Results: We identified 236 eligible RCTs of which 57% (N = 135) evaluated adjuvant treatments. Experimental treatments became more often compared with active treatments (48% versus 59% versus 81%; P < 0.0001). Median absolute benefits of experimental adjuvant treatments decreased but outcomes in control arms improved with time. For RCTs evaluating metastatic disease, there were no changes in absolute benefit over time but incremental monthly costs of new approved treatments increased with time by 100-fold (P < 0.0001).
Conclusion: In RCTs of breast and colorectal cancer, new effective adjuvant treatments show decreasing absolute benefit, while new treatments of metastatic disease show unchanging levels of benefit at rapidly escalating costs.
New treatments in oncology are accepted following demonstration of increased efficacy or decreased toxicity in phase III randomized clinical trials (RCTs). The method of reporting RCTs has been shown to influence health professionals' adoption of new treatments with higher endorsement when numerically higher values of relative benefit are presented as compared with numerically smaller values of absolute benefit. Despite Consolidated Standards of Reporting Trial's recommendations, only 5% of authors expressed their findings in terms of absolute benefits (i.e. absolute risk reduction and/or number needed to treat) in RCTs with statistically significant treatment effects published in major medical journals.
Statistical power determines the chance of detecting a result at a certain level of significance and does not indicate whether the difference between standard and experimental arms has clinical importance. The perception of a 'positive' trial is complex and there is no consistent definition of clinical benefit. With a very large sample size, almost any difference, no matter how meaningless from a clinical standpoint, may be statistically significant. The size of RCTs for breast, colorectal and lung cancer has increased over the last 30 years and therefore RCTs have the potential to detect smaller absolute differences between experimental and standard treatment arms.
Absolute benefits of experimental therapies in time-to-event end points are usually measured in phase III clinical trials at one point [e.g. 5-year overall survival (OS) or median survival] but this method ignores most of the time-to-event data. Assessment of absolute benefit as the area between time-to-event curves up to a given time is an alternative, which uses all of the data from time-to-event curves, does not require assumptions about their shape or of proportional hazards and might be more informative than the usual method.
Recently, we demonstrated that relative benefits of experimental medical therapies in RCTs of breast, colorectal and lung cancer did not change significantly over time. Here, we build on previous work and report changes over time in absolute benefits of experimental medical therapies from RCTs of breast and colorectal cancer. Absolute benefits in time-to-event end points are assessed: (i) at one point, as usually reported, and (ii) as the area between the time-to-event curves up to a given time. We also evaluate changes in costs of new medical therapies for metastatic breast and colorectal cancer and levels of endorsement of new medical therapies by health professionals.
Abstract and Introduction
Abstract
Background: Phase III randomized clinical trials (RCTs) have become larger and are powered to detect small absolute benefits. Temporal changes in absolute benefits of experimental medical therapies reported in RCTs are unknown.
Methods: We identified all RCTs with sample size ≥200 evaluating experimental medical therapies for breast and colorectal cancer published from 1975 to 2007. We assessed changes over three decades in absolute differences in time-to-event end points between experimental and control arms by (i) the usual method (i.e. at one point) and (ii) as the area between time-to-event curves up to a predefined time.
Results: We identified 236 eligible RCTs of which 57% (N = 135) evaluated adjuvant treatments. Experimental treatments became more often compared with active treatments (48% versus 59% versus 81%; P < 0.0001). Median absolute benefits of experimental adjuvant treatments decreased but outcomes in control arms improved with time. For RCTs evaluating metastatic disease, there were no changes in absolute benefit over time but incremental monthly costs of new approved treatments increased with time by 100-fold (P < 0.0001).
Conclusion: In RCTs of breast and colorectal cancer, new effective adjuvant treatments show decreasing absolute benefit, while new treatments of metastatic disease show unchanging levels of benefit at rapidly escalating costs.
Introduction
New treatments in oncology are accepted following demonstration of increased efficacy or decreased toxicity in phase III randomized clinical trials (RCTs). The method of reporting RCTs has been shown to influence health professionals' adoption of new treatments with higher endorsement when numerically higher values of relative benefit are presented as compared with numerically smaller values of absolute benefit. Despite Consolidated Standards of Reporting Trial's recommendations, only 5% of authors expressed their findings in terms of absolute benefits (i.e. absolute risk reduction and/or number needed to treat) in RCTs with statistically significant treatment effects published in major medical journals.
Statistical power determines the chance of detecting a result at a certain level of significance and does not indicate whether the difference between standard and experimental arms has clinical importance. The perception of a 'positive' trial is complex and there is no consistent definition of clinical benefit. With a very large sample size, almost any difference, no matter how meaningless from a clinical standpoint, may be statistically significant. The size of RCTs for breast, colorectal and lung cancer has increased over the last 30 years and therefore RCTs have the potential to detect smaller absolute differences between experimental and standard treatment arms.
Absolute benefits of experimental therapies in time-to-event end points are usually measured in phase III clinical trials at one point [e.g. 5-year overall survival (OS) or median survival] but this method ignores most of the time-to-event data. Assessment of absolute benefit as the area between time-to-event curves up to a given time is an alternative, which uses all of the data from time-to-event curves, does not require assumptions about their shape or of proportional hazards and might be more informative than the usual method.
Recently, we demonstrated that relative benefits of experimental medical therapies in RCTs of breast, colorectal and lung cancer did not change significantly over time. Here, we build on previous work and report changes over time in absolute benefits of experimental medical therapies from RCTs of breast and colorectal cancer. Absolute benefits in time-to-event end points are assessed: (i) at one point, as usually reported, and (ii) as the area between the time-to-event curves up to a given time. We also evaluate changes in costs of new medical therapies for metastatic breast and colorectal cancer and levels of endorsement of new medical therapies by health professionals.