Data Extraction

This guide is for reviewers who need to extract quantitative outcome data from studies for meta-analysis.

Prerequisites:

• Data extraction must be enabled on your stage (ask your administrator if unsure)
• Read the Annotating Studies guide first for general annotation instructions

Overview

Data extraction allows you to capture quantitative measurements from studies, including:

• Outcome measurements (means, medians)
• Error values (SD, SEM, IQR)
• Sample sizes (number of animals per group)
• Time points (when measurements were taken)

This data is structured to support meta-analysis calculations.

Understanding the Unit Hierarchy

When data extraction is enabled, you work with a hierarchy of “units” that represent the structure of the study:

Experiment (groups cohorts being compared)
    └── Cohort (group of animals receiving same treatment)
            ├── Disease Model (what condition they have)
            ├── Treatment (what intervention they receive)
            └── Outcome (what is measured)

Why This Hierarchy Matters

• Experiment: Groups related cohorts that are compared against each other (e.g., “24-hour stroke study”)
• Cohort: A specific group of animals that all receive the same procedures (e.g., “Treatment Group A with 10 rats”)
• Disease Model: The condition being studied (e.g., “MCAO” for stroke, or “Sham” for control)
• Treatment: The intervention given (e.g., “Aspirin 100mg” or “Vehicle” for control)
• Outcome: What is measured (e.g., “Infarct Volume” measured in mm³)

Step-by-Step Workflow

The key principle is: Create units first, then link them together.

Step 1: Create Units in Their Respective Tabs

Before you can link anything, you must create the individual units. Work through these tabs:

Disease Model Tab

1. Click “Add Disease Model” (or the + button)
2. Enter a label (e.g., “MCAO” or “Sham surgery”)
3. Set “Control” to Yes if this is a control procedure (e.g., “Sham” = control)
4. Answer any additional questions your administrator has configured
5. Click Save

Treatment Tab

1. Click “Add Treatment” (or the + button)
2. Enter a label (e.g., “Aspirin 100mg” or “Vehicle”)
3. Set “Control” to Yes if this is the control treatment (e.g., “Vehicle” = control)
4. Answer any additional questions your administrator has configured
5. Click Save

Outcome Assessment Tab

1. Click “Add Outcome” (or the + button)
2. Enter a label (e.g., “Infarct Volume”)
3. Set the Average Type - choose Mean or Median based on how the study reports data
4. Set the Error Type - choose SD, SEM, IQR, etc. based on what the study reports
5. Enter the Units (e.g., “mm³”, “score”, “%”)
6. Set “Greater Is Worse” - Yes if higher values indicate worse outcomes
7. Click Save

Step 2: Link Units via Cohort

Once you have created your disease models, treatments, and outcomes, link them together in the Cohort tab:

1. Click “Add Cohort” (or the + button)
2. Enter a label (e.g., “Treatment Group A” or “MCAO + Aspirin”)
3. Use the “Select Disease Models” lookup to choose which disease models apply to this cohort
4. Use the “Select Treatments” lookup to choose which treatments this cohort received
5. Use the “Select Outcomes” lookup to choose which outcomes were measured for this cohort
6. Enter the Number of Animals in this cohort
7. Click Save

Step 3: Link Cohorts to Experiment

Group your cohorts into experiments:

1. Click “Add Experiment” (or the + button)
2. Enter a label (e.g., “Experiment 1” or “24-hour timepoint study”)
3. Use the “Select Cohorts” lookup to choose which cohorts belong to this experiment
4. Click Save

Step 4: Enter Outcome Data

Once the hierarchy is complete (Experiment → Cohort → Outcome), the outcome data entry interface appears:

1. Navigate to the Experiment tab
2. Expand an experiment to see its cohorts
3. Expand a cohort to see its linked outcomes
4. For each outcome, click “Add TimePoint” or the data entry interface
5. Enter:
   • Time: When the measurement was taken (e.g., 24 for 24 hours)
   • Average: The mean or median value from the study
   • Error: The SD, SEM, or IQR value from the study
6. Optionally link to a PDF graph if you’ve uploaded documents
7. Repeat for all timepoints reported in the study

Understanding Lookup Questions

Lookup questions automatically show options from other categories:

Lookup Question	Shows Options From
“Select Disease Models”	Disease Model tab
“Select Treatments”	Treatment tab
“Select Outcomes”	Outcome Assessment tab
“Select Cohorts”	Cohort tab

Important: You must create units in their respective tabs before they appear in lookup questions. If a lookup shows no options, go back and create the units first.

Understanding the “Control” Checkbox

The “Control” checkbox asks: “Is this specific unit a control procedure?”

This is asking about the unit itself, not about the group:

Unit Type	Control = Yes (examples)	Control = No (examples)
Disease Model	Sham surgery, No induction, Naive	MCAO, Stroke induction, Disease model
Treatment	Vehicle, Saline, No treatment	Aspirin, Active drug, Intervention

Common confusion: This is NOT asking “Is this the control group?” The control group is determined by the combination of units in a Cohort. For example, a cohort with “Sham surgery” (control) + “Vehicle” (control) would be the control group.

Example: Stroke Study

Here’s how you might extract data from a typical stroke study:

Study reports: Two groups - Treatment (MCAO + Aspirin) vs Control (MCAO + Vehicle), measuring infarct volume at 24h and 48h.

1. Create Disease Models:

   • “MCAO” (Control = No)

2. Create Treatments:

   • “Aspirin 100mg” (Control = No)
   • “Vehicle” (Control = Yes)

3. Create Outcomes:

   • “Infarct Volume” (Average Type = Mean, Error Type = SD, Units = mm³, Greater Is Worse = Yes)

4. Create Cohorts:

   • “Treatment Group”: Link to MCAO, Aspirin 100mg, Infarct Volume. Enter n=10.
   • “Control Group”: Link to MCAO, Vehicle, Infarct Volume. Enter n=10.

5. Create Experiment:

   • “Main Experiment”: Link to both cohorts

6. Enter Outcome Data:

   • For Treatment Group → Infarct Volume: Add timepoints at 24h and 48h with means and SDs
   • For Control Group → Infarct Volume: Add timepoints at 24h and 48h with means and SDs

Troubleshooting

Lookup Shows No Options

• Cause: Units haven’t been created yet
• Solution: Go to the respective tab (Disease Model, Treatment, Outcome Assessment) and create the units first

Can’t See Data Extraction Categories

• Cause: Data extraction may not be enabled on your stage
• Solution: Ask your project administrator to enable data extraction in stage settings

Outcome Data Entry Not Appearing

• Cause: The hierarchy is incomplete
• Solution: Ensure you have:
  1. Created at least one Experiment
  2. Created at least one Cohort linked to the Experiment
  3. Created at least one Outcome linked to the Cohort
  4. The full chain must be complete: Experiment → Cohort → Outcome

Not Sure What Average/Error Type to Use

• Check how the study reports the data
• “Mean ± SD” = Average Type: Mean, Error Type: SD
• “Mean ± SEM” = Average Type: Mean, Error Type: SEM
• “Median (IQR)” = Average Type: Median, Error Type: IQR
• If unclear, consult your project administrator or protocol

Quick Reference Checklist

Use this checklist for each study:

• Disease Models created - With labels and control flags
• Treatments created - With labels and control flags
• Outcomes created - With label, average type, error type, units, greater is worse
• Cohorts created - Linked to disease models, treatments, outcomes; with number of animals
• Experiments created - Linked to cohorts
• Outcome data entered - TimePoints for each Experiment → Cohort → Outcome combination

Related Guides

• Annotating Studies - General annotation guide for reviewers
• Stage Settings - How administrators enable data extraction
• Glossary - Definitions of key terms