Purpose/Objectives: Two analytical approaches are described for a randomized trial testing interventions for symptom management.
Design: To compare an intention-to-treat with a perprotocol approach.
Setting: Patients were accrued from six cancer centers.
Sample: 94 men and 140 women with solid tumors were accrued.
Methods: An intention-to-treat approach (as randomized) and per-protocol analyses (at least one symptom reaching threshold and one follow-up intervention) were compared. The analysis determines how each approach affects results. A two-arm, six-contact, eight-week trial was implemented. In one arm, nurses followed a cognitive behavioral protocol. In the second arm, a non-nurse coach referred patients to a symptom management guide.
Main Research Variables: Trial arm; summed severity scores; interference-based severity categories at intake, 10 weeks, and 16 weeks; site; and stage of cancer.
Findings: Each arm produced a reduction in severity at 10 and 16 weeks with no differences between arms. In the per-protocol analyses, symptoms reported at the first contact required more time to resolve. Older patients exposed to the nurse arm resolved in fewer contacts.
Conclusions: The intention-to-treat analyses indicated that both arms were successful but offered few insights into how symptoms or patients influenced severity. Per-protocol analyses (intervention and dose), when, and which strategies affected symptoms.
Implications for Nursing: Each analytical strategy serves a purpose. Intention-to-treat defines the success of a trial. Per-protocol analyses allow nurses to pose clinical questions about response and dose of the intervention. Nurses should participate in analyses of interventions to understand the conditions where interventions are successful.