Skip to main content
Paste your PICO extraction schema, attach a clinical trial paper, and instruct Anara to extract only from the attached document. It returns a completed PICO table with GRADE quality scoring, every field traced to the source text, and blanks for anything not reported. For HEOR analysts, medical affairs teams, and regulatory writers who need extraction that stands up to audit.

​
1. Describe the task

The dossier requires a PICO extraction table for every included study. Each field must be traceable to the paper. If a field is not reported, the cell stays blank. You cannot fill gaps from memory, from similar trials, or from what the endpoint usually is in this indication. The constraint is the methodology. Anara applies your extraction schema to the attached paper, reads only that document, and returns the completed table with a GRADE quality assessment. Fields the paper does not report are marked not reported. The constraint stated clearly in the prompt is what makes the output auditable. Here a health economics and outcomes researcher at a pharmaceutical company is extracting PICO data and GRADE quality scores from a randomized controlled trial on a PCSK9 inhibitor for use in an indirect treatment comparison for a payer submission. 
I am extracting clinical trial data for a HEOR dossier. Attached is a single RCT paper. Apply my extraction schema below using ONLY the information in this paper. Do not use external knowledge, prior studies, or general clinical assumptions to fill any field. If a field is not reported in this paper, write "Not reported."

PICO schema:
- Population (inclusion/exclusion criteria, sample size, demographics)
- Intervention (drug, dose, regimen, duration)
- Comparator (active or placebo; dose and regimen)
- Primary outcomes (measure, result, 95% CI, p-value as reported)
- Secondary outcomes (list each with results as reported)
- Follow-up duration
- Study design and blinding
- Industry sponsorship (yes/no, which sponsor)

After the PICO table, provide a GRADE quality assessment: rate the evidence quality as high, moderate, low, or very low, and state the reason for any downgrade (risk of bias, indirectness, imprecision, inconsistency, publication bias).

At the bottom, quote the three passages from the paper that most directly support the primary outcome data.

​
2. Give Anara context

Required context
  • The clinical trial paper as a PDF upload or from your library.
  • Your PICO extraction schema. Paste it directly into the prompt.
Optional context
  • One completed row from your existing extraction table. Anara matches the format and level of detail of your prior work.
  • The indication or drug class. Helps Anara identify the primary outcome measure when the paper uses indication-specific terminology.

​
3. What Anara creates

A completed PICO extraction table drawn exclusively from the attached paper, with a GRADE quality assessment and three quoted source passages at the bottom for audit verification. Every filled field has a traceable basis in the text. Every unfilled field is marked not reported rather than inferred. The output is the audit-ready extraction unit you paste into your evidence table and verify against the quoted passages before the row enters the dossier.

​
4. Follow-up prompts

​
Run GRADE quality ranking across a folder of papers

When you have completed single-paper extractions and want to synthesize quality across the evidence base. 
I have extracted PICO data from 15 RCTs in my library folder. Search across all 15 papers and rank them by GRADE evidence quality from high to very low. For the papers rated low or very low, summarize the main reasons for downgrade and note which limitation is most common across the set.

​
Check a specific field against the source text

When a filled cell looks too precise and you want to confirm it came from the paper. 
For the "primary outcome result" cell you filled: quote the exact sentence from the paper that reports this value. If you derived the figure from a calculation or table rather than a verbatim statement, identify the specific table and row number.

​
Compare outcomes across two included studies

When you need to flag heterogeneity before completing a meta-analysis input. 
Compare the primary outcome measures and effect sizes in this paper to the paper I uploaded last session. Do they use the same outcome measure? Are the populations sufficiently similar for a direct comparison? Flag differences that would require adjustment in an indirect treatment comparison.

​
5. Tricks, tips, and troubleshooting

​
How you word your prompt shapes what you get

The constraint statement is the most important part of the prompt. Without “only use information from this paper,” Anara may fill fields from general clinical knowledge when the paper is silent, producing a table that looks complete but is not traceable. State the fallback explicitly: “if not reported, write Not reported.” The three source passages at the bottom make every row self-auditing.

​
Check the output against your own understanding

Anara reads the paper as text. If the primary outcome data is in a figure rather than a table, note this in the prompt: “the primary outcome data is in Figure 2, not in the results text.” For GRADE assessments, Anara applies the framework from your uploaded guidance documents. It does not assess inconsistency across studies without access to the other papers; provide that context in the prompt if it affects the downgrade decision.

​
What to do with the output next

Paste the row into your evidence table and verify the three quoted passages against the PDF before the row enters the dossier. Use the cross-paper GRADE ranking follow-up to identify which studies carry the most evidentiary weight.