Vital Signs and Lab Results Extraction Checklist

By Updated
Table of Contents

What this checklist is for

A clinical record often has measurements in many places. Blood pressure can be in a triage note, glucose in a lab table and weight in a nursing note. A reading from home can be in the free text of a letter. This checklist puts every vitals reading and every lab value from one record into one table. Vitals are basic body measurements, such as pulse and blood pressure. For each value, you record the date, the unit, the source and the reference range.

A reference range is the range of results that a lab or a source calls normal. You flag values outside the range, but you do not interpret them. The table helps a reviewer prepare for a visit, a referral or an audit. It is not medical advice. Only a qualified clinician can interpret the values and decide what to do.

What to extract

Extract these vitals, if the record contains them:

  • Blood pressure
  • Heart rate (pulse)
  • Respiratory rate (breaths per minute)
  • Temperature
  • Oxygen saturation (SpO2)
  • Weight and height
  • Body mass index (BMI)
  • Pain score

Also extract every lab value in the record, for example glucose, cholesterol, kidney and liver tests and blood counts. Copy the test name, value, unit and reference range from the report.

The checklist

  1. Find every measurement. Read the vitals flowsheet, which is a table of readings over time. Read the triage notes and the nursing notes. Also read the progress notes, the lab tables and the discharge summary. People often miss values in free text, for example “home BP readings around 150/95”.

  2. Use one row for each reading. If the record has three blood pressure readings, make three rows with their times. Do not average the readings. Do not keep only the last reading.

  3. Record the date and time of each measurement and the section or page of the source. Use the date of the measurement or of the sample collection. Do not use the print date of the report.

  4. Copy the value and the unit exactly. Units differ between labs and countries. For example, a lab can report glucose in mg/dL or in mmol/L. MedlinePlus gives its glucose ranges in both units. Temperature can be in °F or °C, and weight can be in lb or kg. If a value has no unit, put it on the list of open items. Open items are values that you cannot determine.

  5. Keep each blood pressure reading as one pair, for example 128/82 mmHg. Record the systolic value and the diastolic value together. If the record gives the position or the setting, record it. Examples are sitting, standing or a home reading.

  6. Record the measurement method, if the record gives it. Examples are SpO2 on room air or on oxygen, the temperature route, and fasting or random glucose. A pulse oximeter reading is an estimate. The FDA lists factors that can change its accuracy. These factors include poor circulation, skin pigmentation, skin thickness, skin temperature, current tobacco use and fingernail polish.

  7. Use the reference range of the lab. MedlinePlus says that labs often use different reference ranges. It tells you to use the range on your own report. It also says that a result outside the range is not always a sign of a health problem. Copy the range that the report prints next to the result.

  8. Write down the range that you use for each vitals reading. If the record gives no range, use a range that a clinician approved for the age of the patient. For context, MedlinePlus lists these normal values for adults at rest:

    • Temperature: 97.7°F to 99.1°F (36.5°C to 37.3°C)
    • Pulse: 60 to 100 beats per minute
    • Breathing: 12 to 18 breaths per minute
    • Blood pressure: between 90/60 mmHg and 120/80 mmHg

    MedlinePlus adds that normal values change with age, sex, weight, exercise capability and overall health. These values are not a rule for every patient.

  9. If the record lists BMI, make sure that it matches a weight and a height in the same record. The CDC gives these adult BMI categories for adults 20 and older:

    • Underweight: less than 18.5
    • Healthy weight: 18.5 to less than 25
    • Overweight: 25 to less than 30
    • Obesity: 30 or greater

  10. Flag each value as high, low or within range. Use the range in the same row, and name its source. Do not interpret the values. The clinician interprets them.

How to record the result

Use one row for each reading:

Date and time Measure Value Unit Reference range Range source Flag Where found Notes
Blood pressure mmHg sitting, clinic
Glucose, fasting lab report

Under the table, list the open items:

  • Values that you cannot read
  • Values without a unit or a date
  • Readings that differ between two sections of the record

Use this checklist with Claude or ChatGPT

You can paste the checklist and a de-identified record into Claude or ChatGPT. De-identified means that you removed all patient identifiers. Ask the AI tool to build the table. Ask it to give the source of each value. Then you can compare each value with the record.

You are helping me extract measurements from a clinical record into one table. This is a documentation task, not a clinical judgment.

Checklist:
1. Find every vitals reading (blood pressure, heart rate, respiratory rate, temperature, SpO2, weight, height, BMI, pain score) and every lab value anywhere in the document, including free text.
2. Make one row per reading. Do not average or drop repeated readings.
3. For each row, record the date and time measured, the value and unit exactly as written, and the section or page.
4. Keep blood pressure as one systolic/diastolic pair.
5. Note how the value was measured when stated (room air or oxygen, fasting or random, home or clinic).
6. For lab values, copy the reference range printed in the report. Do not supply a range from memory.
7. For vitals, use only the ranges I give below. If none is given, write "no range given".
8. Flag each value as high, low or within range, based only on the range in that row.
9. If BMI is listed, check it against the weight and height in the record.

Output:
- A table with the columns: date and time, measure, value, unit, reference range, range source, flag, where found, notes.
- For each checklist item, one or two lines on how you checked it.
- A list of everything you could not determine (missing units, unreadable values, conflicting readings), and why.

Ranges approved for this review:
[paste ranges, or write "none"]

Record (identifiers removed):
[paste the text or attach the file]

Before you use the result:

  • Remove patient identifiers before you paste or upload anything. Identifiers include the name, date of birth, record number, address and phone number. Also read your workspace’s data settings and the rules of your organization for patient data.
  • Compare every flagged value and a sample of the other values with the record. AI tools can misread scanned tables, take the range from the wrong row or mix up units. They can also report a value that is not in the record.
  • Do not use the table as medical advice. A qualified clinician must review and interpret it.

A docs portal with Ask AI
and support tickets

Connect a repository, or describe your product to start. Each docs site has search, Ask AI, an API console, an MCP server and support tickets.

  • Publishes from GitHub on every push
  • Works with Mintlify repositories
  • API console from your OpenAPI file
  • MCP server and llms.txt for AI agents
  • Ask AI and support tickets
  • Analytics for people and AI agents
Start Free Questions? Ask AI in our docs

No card required.

Share:
Markdown version
↧
Loading PDF…