Using the results of Phase III studies in clinical practice depends on how representative study participants are of the clinical population to whom the results will be applied. The closer the characteristics between the subgroup who participate in a clinical trial and the whole population, the easier it is for clinicians to apply the results directly to the patient that he/she is treating. Trial participation is generally more happenstance than a systematic sampling of a population and is limited by eligibility criteria that do not reflect the entire clinical population.
Phase III study populations tend to be younger with fewer comorbidities and not represent the gender and ethnicity of the target population, limiting generalizability of results. When moving from Phase III studies to Phase IV postmarketing studies, there will be differences in the populations prescribed the medications and the outcomes achieved, highlighting the gap between gold-standard evidence from Phase III randomized studies and the application of that evidence in real-world practice. Palliative care services are referral based, yet there are no standard national or international referral criteria. As such, the populations served by specialist palliative care services are ill defined, complicating the problem of defining a "representative" population even further. Key characteristics have been suggested to aid generalizability of research findings in palliative care, although these are poorly reported.
In palliative care, large-scale Phase III, symptom control effectiveness studies are being conducted successfully. This includes the work of the Australian National Cancer Symptom Trials (CST) group. Can the findings from these studies be generalized to the broader palliative cancer care population in the same health system?
The aim of this analysis was to compare key demographic factors between participants with cancer in these Phase III studies and people referred to nontrial specialist palliative care services using standard prospective data collection. The null hypothesis was that there was no difference between participants' characteristics.