Comparing treatments and personal preferences

First specify what the option should improve

Treatment may remove a cause, reduce symptoms, prevent future events, or maintain function. A person’s treatment can have several goals at once, but interventions do not serve them identically. Walking farther after pain relief is meaningful without proving that joint structure has been repaired. Conversely, a preventive treatment may provide benefit without producing an immediate feeling of improvement. Specify the question before ranking the options. AHRQ: The SHARE Approach

Studies often measure blood pressure, laboratory values, or imaging changes. Some measures predict clinical benefit in a validated context; others only demonstrate a biological effect. A surrogate must be evaluated for the particular disease, treatment, and use. The idea that a lower marker is desirable does not prove that every method of lowering it prolongs life. Symptoms, activities, hospitalization, and survival are often more directly relevant. Ask what was measured and which links between the result and the person’s goal remain uncertain. FDA: Surrogate endpoints

Compare benefit with a common denominator

Consider a fictional five-year trial: 100 of 1,000 people on usual care experience the specified event, compared with 70 of 1,000 on a new option. Risk falls from 10% to 7%, an absolute reduction of three percentage points and a relative reduction of 30%. Both describe the same result, but reporting only a thirty-percent reduction conceals the starting probability. The event, period, and hypothetical status should accompany the numbers. Smart Health Choices: Absolute and relative risk

The reciprocal of an absolute reduction of 0.03 is about 33.3, conventionally rounded upward: approximately 34 people treated for five years would prevent one additional event on average under these conditions. This does not identify a guaranteed beneficiary in each group of 34. A lower-risk population could have smaller absolute benefit despite the same relative effect. Number needed to treat must therefore be interpreted alongside its comparator, outcome, and duration rather than be treated as a fixed property of a medicine. Smart Health Choices: Absolute and relative risk

Harms and burdens extend beyond a side-effect list

An option may increase bleeding, infection, or organ injury, while also causing nausea, drowsiness, procedural pain, or recovery time. These outcomes differ in severity, reversibility, and timing. Three prevented disease events and two episodes of discomfort cannot simply be subtracted to produce one unit of net benefit. Describe each outcome and its absolute probability, then explore its practical significance and acceptable trade-offs. FDA: Managing benefits and risks of medicines

Treatment burden also includes dosing, monitoring, travel, cost, caregiving, and lost work. Kidney or liver function, other medicines, allergies, and pregnancy plans can alter risk, so study averages cannot replace clinical history. A plan requiring repeated loss of daily wages may be difficult to sustain. Simplification, alternatives, and resource support deserve consideration rather than immediately labeling difficulty as nonadherence. Review prescription medicines, nonprescription products, and supplements together, with a clear route for reporting and addressing adverse effects. FDA: Managing benefits and risks of medicines

Assess both credibility and applicability

Improvement after treatment does not establish that treatment caused it. Natural fluctuation, simultaneous changes, and measurement error can also affect outcomes. A suitable comparison estimates what might happen without the intervention. Random allocation reduces systematic group differences, and blinding can reduce the influence of expectations on reporting or assessment. Even a strong design needs examination of missing follow-up, outcome selection, sample size, and uncertainty; statistical significance alone does not demonstrate a large clinical benefit. NCI: How clinical trials work

Check whether participants resemble the person considering treatment in stage of illness, age, other conditions, existing therapy, and service setting. An average effect is not identical for everyone, and a striking result in a small subgroup may be chance. Reviews organize research but inherit its limitations. Direct studies and formal evaluations should be connected to practical circumstances. Novelty, price, or endorsement is not a complete benefit-harm comparison. MedlinePlus: Evaluating health information

Preferences develop through informed discussion

Evidence describes possible outcomes; the person can explain which outcomes matter most. With comparable knee pain, one caregiver may prioritize mobility in the coming weeks, while another accepts a longer recovery period for a potential later gain. Preferences are not fixed answers obtained by simply asking which option someone wants. Understanding the alternatives may change the initial view, while clinicians explain which choices are medically reasonable. AHRQ: The SHARE Approach

Where appropriate, include observation or no immediate intervention, with monitoring and consequences of delay. Shared decision-making does not transfer the entire burden of interpreting evidence to the patient. Some people want a clear recommendation and its reasoning; others want family involvement and time to reflect. Language, hearing, anxiety, and information load may require support. Disagreement can lead to clarification or another opinion without treating hesitation about one option as refusal of all medical care. GMC: Dialogue leading to a decision

Turn a choice into a reviewable plan

After choosing, record what will be used, how it starts, when review occurs, which outcomes matter, and what should prompt earlier contact. Functional goals are often useful: reaching a bus stop or sleeping without pain-related waking is clearer than feeling somewhat better. Preventive therapy may produce no short-term symptom change, so review relevant measures, practical difficulties, and adverse effects. Timing should match the intervention’s action and risks. GMC: Dialogue leading to a decision

If an effective option repeatedly causes nausea and missed doses, the response may involve investigating the cause, adjusting treatment, or choosing an alternative rather than merely repeating instructions to persist. Comparing the options again is appropriate when health or priorities change and does not establish that the previous decision failed. Some uncertainty requires more information; other uncertainty can be addressed through an appropriate trial and follow-up. A good decision is a plan informed by evidence and the person’s participation that can be revised in light of actual results. FDA: Managing benefits and risks of medicines

Bilingual terms

临床结局 · clinical outcome

A measure directly reflecting symptoms, function, survival, or another clinical result.

替代终点 · surrogate endpoint

An indicator used in place of a clinical outcome with context-specific supporting evidence.

绝对风险降低 · absolute risk reduction

Risk in the comparator group minus risk in the treatment group.

需要治疗人数 · number needed to treat

Number treated under specified conditions and duration to prevent one additional event on average.

治疗负担 · treatment burden

Time, work, cost, and daily-life changes required to carry out treatment.

Sources and further reading

• AHRQ: The SHARE Approach
• FDA: Surrogate endpoints
• Smart Health Choices: Absolute and relative risk
• FDA: Managing benefits and risks of medicines
• NCI: How clinical trials work
• MedlinePlus: Evaluating health information
• GMC: Dialogue leading to a decision

Original course source-check record: 9 September 2026. Full Chinese and English sentence-by-sentence language review: 14 September 2026. AI editing and language review are not human clinical review. Linked institutions have not participated in or endorsed this course.