Palliative care focuses on helping relieve pain and prevent the suffering of patients, especially those patients facing advanced cancer and other life-threatening illnesses. More than 80% of large U.S. hospitals have or currently are developing palliative care programs. Although the main goal of palliative care is to improve patient outcomes such as pain and quality of life (QoL), studies have demonstrated that palliative care may also affect survival. Evaluation and improvement of palliative care programs is a primary goal for these research studies. To accomplish this goal, new statistical methods are needed, ranging from improved definition of processes and interventions to standardization of outcomes and improved statistical analysis. In this project, we propose to develop novel methods for the design and analysis of palliative care research studies that have the potential to increase scientific rigor and to fully utilize the valuable data that these trials generate. One of the major statistical issues raised in palliative care studies regards the interplay and tradeoffs between the quality and duration of life, commonly used as primary outcomes. It has been widely noted that patient outcomes worsen at the end of life for patients with advanced cancer. This suggests dependence between survival and longitudinal measurements of patient outcomes. Palliative care intervention studies have reinforced this hypothesis by showing gains in survival and QoL and symptoms including pain. However, this dependence has not been taken into account in analyzing or designing palliative care clinical trials. We have recently developed a method for jointly modeling longitudinal outcomes and survival in palliative care studies using a novel terminal decline model (TDM) where outcomes are modeled in terms of time from death on a retrospective time scale. Randomized "early intervention" or "fast-track" trials (also sometimes called delayed intervention, delayed-start, or wait list designs) have been proposed as an alternative to traditional RCTs comparing palliative care interventions to non-palliative or other supportive care groups. However, many issues remain with respect to implementing such designs. We propose to study this trial design and more general "stepped wedge" designs in the context of TDMs. Specifically, we will address methods for estimating treatment effects and for choosing optimal design features such as overall sample size and the optimal timing of interventions..
Palliative care can lessen suffering from pain and improve outcomes for patients with life threatening conditions. Promising innovations in palliative care are now under study. This grant will create new statistical methods to improve the analysis and design of important ongoing and future palliative care trials.
|Li, Zhigang; McKeague, Ian W; Lumey, Lambert H (2014) Optimal design strategies for sibling studies with binary exposures. Int J Biostat 10:185-96|