Supportive care, survivorship and clinical evidence
Align treatment with the actual goal
Palliative care addresses symptoms, function, communication and support alongside disease-directed therapy when appropriate. It is not restricted to the final days or defined by a single code-status decision. Pain assessment includes mechanism, intensity, function, previous treatment, adverse effects and the person's priorities. Opioid rotation, rescue treatment and adjuvant approaches require careful review rather than an automatic equianalgesic substitution. Myoclonus and hyperalgesia can suggest neurotoxicity, but dehydration, organ dysfunction and other causes also need consideration. A fixed high-volume hydration instruction is not appropriate for every person near the end of life.
Support the whole treatment course
Antiemetic prevention reflects emetogenic risk and individual circumstances. Nutrition, fertility preservation, sexual health, rehabilitation, fatigue and psychological support can matter before, during and after treatment. A survivorship plan summarizes relevant therapy, possible late effects, surveillance and coordination; it does not guarantee that every symptom is cancer-related. Secondary-cancer surveillance depends on the actual exposure, age, sex and applicable framework. Family participation supports care but does not automatically override the current informed preferences of a person who has decision-making capacity.
Define the endpoint before interpreting a trial
Overall survival, progression-free survival, response, quality of life and adverse events measure different outcomes. A progression advantage is not proof of an overall-survival or quality-of-life advantage. Censoring requires assumptions about what is observed and why follow-up ends. A hazard ratio describes a time-to-event comparison under the relevant model, not a fixed absolute probability difference for every patient. For example, a hazard ratio of 0.65 under proportional hazards indicates an estimated 35 percent lower hazard, not an automatic 35-percentage-point reduction in cumulative event risk.
Evaluate size, uncertainty and applicability
Confidence intervals, prespecified analyses, multiplicity, missing data, crossover and subsequent treatments affect interpretation. A noninferiority conclusion needs the correct endpoint direction and a justified prespecified margin; absence of statistical significance for superiority is insufficient. Predictive values depend on prevalence, even when sensitivity and specificity are held fixed in a supplied exercise. A biomarker associated with outcome is not necessarily predictive of differential treatment benefit. Finally, determine whether the studied population, disease state and prior treatment match the case. Honest discussion includes what is known, the magnitude of potential benefit and harm, uncertainty and the person's values.