返回提示词目录第 1299 页

研究证据与本地情境适用性评估

帮你梳理研究与实际场景在人群、实施和结局上的差异,形成适用性判断与需专业确认的清单。

完整提示词

<context>
A study can be well done and still not apply to you. Risk-of-bias tools judge whether the result is true for the people studied; applicability (external validity, transferability) asks whether it will hold for your people, in your setting, delivered your way. The differences that matter are specific: who was excluded (often the older, sicker, poorer or more complex people a service actually sees), how the intervention was delivered (expert staff, extra funding, high fidelity in a trial), what the comparator was (usual care in another country may be much better or worse than yours), how long outcomes were followed, and baseline risk, which changes the absolute benefit even when the relative effect holds. Explanatory trials run under ideal conditions transfer less well than pragmatic ones. Not every difference matters: the question is whether there is a plausible reason the mechanism or the size of the effect would change.
</context>

<task>
Judge whether this study applies to my setting.
<study>
{{study}}
</study>
<my_context>
{{context}}
</my_context>
{{#decision}}The decision: {{decision}}{{/decision}}

1. Summarise the study's population, setting, intervention or exposure as delivered, comparator, outcomes and follow-up from the text, noting exclusion criteria and how many screened people were enrolled if stated.
2. Compare each dimension with the user's context: population (age, severity, comorbidity, socioeconomic and cultural factors), setting (country, system, level of care or schooling, era), delivery (who delivered it, training, intensity, fidelity, resources), comparator (what "usual practice" meant), outcomes (whether they are the ones the user cares about, and over what time), and baseline risk.
3. For each difference, say whether there is a plausible reason it would change the direction or size of the effect, and which way.
4. Say how pragmatic or explanatory the study looks and what that means for transfer.
5. List what to check locally before relying on the result: data on the local population's baseline rates, capacity to deliver as in the study, acceptability, costs, and any local pilot or audit that would test it.
6. Give a verdict: likely applicable, applicable with adaptations (name them), or unclear or unlikely (and why), tied to the decision if given.
</task>

<constraints>
- Use only the study text and the context given. Where the study does not report something needed (exclusions, delivery details), say "not reported" and treat it as uncertainty, not as a match.
- Do not judge the study's internal validity here beyond a one-line note if a serious flaw is obvious; point to a risk-of-bias appraisal instead.
- Do not decide for the user. The verdict concerns transfer of evidence; costs, values, local guidance and professional judgement also shape the decision.
- Avoid generic caveats such as "results may not generalise"; every difference listed must be specific to this study and this context.
- If the context is too vague to compare (for example "my clinic"), ask for the three or four facts that matter most.
</constraints>

<output_format>
## Fit in brief
Two or three sentences.
## Differences that matter
Table: dimension | study | your setting | likely effect on the result.
## Differences that probably do not
Bullets with one-line reasons.
## What to check locally
Checklist.
## Verdict
One line verdict, then two or three sentences.
</output_format>

可替换内容

  • {{study}}
  • {{context}}
  • {{#decision}}
  • {{decision}}
  • {{/decision}}

分类与标签

学习教育、文献评估、外部效度、证据应用、情境比较

前往 AI 图片创作