About

8.1 What Is Data Integrity?

Theory: What Is Data Integrity?

In Week 7 the question was “Who should be allowed to see the data?” (confidentiality). This week the question is “Can we trust the data?” That is integrity, the “I” in the CIA triad. Availability, the “A”, comes in Week 9.

A leak is loud: if 1,000 patient records appear on the internet, everyone knows within a day. A change can be silent: if one patient’s penicillin allergy is quietly deleted from the EHR, nobody notices until the patient is given penicillin. That is why integrity failures can be more dangerous than confidentiality failures, even though they make fewer headlines.

Definition

Data integrity means data is accurate and trustworthy, and it is changed only by authorized people, in authorized ways.

The definition has two separate parts, and both must hold:

PartMeaningWhere it comes from
1. Authorized peopleOnly people with the right permission can change the data.Access control (Week 6)
2. Authorized waysEven the right person must change data through the correct process. A doctor should not edit a lab result directly; the result must come from the lab process.New this week (integrity models, Part 3)

Common mistake: thinking integrity is just access control. A doctor has permission to change medical records, but if she types over a lab value directly instead of asking the lab to re-run the test, the person is authorized while the way is not. Integrity is broken.

The Four Dimensions of Integrity

DimensionMeaningHospital example
AccurateThe value is correct.The doctor ordered 4 units, and the record says 4 units.
CompleteNothing is missing.The allergy list has every known allergy.
ConsistentSystems agree.The lab system and the EHR show the same result.
CurrentThe data is up to date.The medication list changes when the doctor stops a drug.

A record can fail any one dimension while passing the others. A deleted allergy is accurate for the allergies that remain, but it is no longer complete.

Data Integrity vs. System Integrity

Data integritySystem integrity
QuestionIs the data itself correct?Does the system work as designed, untampered?
ExamplesA lab value, a drug dose, a bank balanceThe EHR software, the server, a software update
Typical failureA dose typed as 40 instead of 4An update that silently removes a validation check

The two are linked: a system whose integrity is broken (bad update, tampered code) will soon produce data whose integrity is broken.

Why Integrity Matters

Every clinical and business decision is made on data. When data is wrong, it still looks normal. People act on it with confidence, and no one notices until harm happens. A missing record is obvious; a wrong number in the right field is not.

Exam tip: classifying C, I or A. Ask what went wrong with the data:

  • The wrong person saw it → Confidentiality.
  • The data is wrong, missing or changed → Integrity.
  • The right person cannot get to it → Availability.

Worked Examples

Exercise 1

Is each problem about C, I, or A?

  1. A nurse opens the lab results of a patient she is not treating.
  2. After a system update, a lab result shows 5.0 instead of 0.5.
  3. The EHR is down for three hours during a storm.
Show solution ↓
Solution

Apply the three questions from the tip above.

  1. Nothing was changed and the system works; the problem is that someone without a need to know saw the data. → Confidentiality.
  2. The value itself is wrong (ten times too high) and the cause is a change to the system. → Integrity (a data-integrity failure caused by a system-integrity problem: the update).
  3. The data is still correct and private, but authorized staff cannot use it for three hours. → Availability.

Answer: 1 = C, 2 = I, 3 = A.

Exercise 2

Which is more dangerous for MediCare Plus, and why?

  • A. The data leaks: 1,000 patient records appear on the internet. Everyone knows within one day.
  • B. The data changes: one patient’s penicillin allergy is silently deleted from the EHR. No one notices.
Show solution ↓
Solution

Compare the two on detection and harm:

  • Leak (A): a serious confidentiality breach, but it is detected within a day. The hospital can notify patients and regulators and contain it. Nobody’s treatment is directly affected.
  • Silent change (B): the record still looks normal, so staff trust it. Nobody is looking for the problem. The first sign may be an allergic reaction when penicillin is prescribed, so the harm is physical and possibly fatal.

B is more dangerous for patient safety, because an undetected integrity failure leads people to act confidently on wrong data, and harm happens before anyone notices. (A is still a serious incident, just a different kind of damage.)

Exercise 3

A doctor has full rights to edit patient records. She notices a lab potassium value that looks wrong, so she opens the lab result and types in the value she expects. Has integrity been violated? Use the definition of integrity in your answer.

Show solution ↓
Solution

Check both parts of the definition:

  1. Authorized people? Yes. The doctor has permission to change records.
  2. Authorized ways? No. A lab result must come from the lab process (re-test or the lab’s own correction procedure). Typing over it bypasses that process, and her expected value may itself be wrong.

Because the second part fails, yes, integrity is violated: the person is authorized but the way is not. The correct action is to ask the lab to re-check the sample and correct the result through the lab’s process.

Exercise 4

For each situation, name the dimension of integrity (accurate, complete, consistent, current) that fails.

  1. A patient’s medication list still shows a drug the doctor stopped last week.
  2. The pharmacy system shows the patient’s weight as 60 kg, but the EHR shows 75 kg.
  3. A discharge summary is saved without the list of new prescriptions.
  4. A blood pressure of 120/80 is recorded as 210/80.
Show solution ↓
Solution

Match each to the definition of the dimension:

  1. The list was right once but was not updated → Current.
  2. Two systems disagree about the same fact → Consistent.
  3. Part of the record is missing → Complete.
  4. The value itself is wrong → Accurate.

Answer: 1 current, 2 consistent, 3 complete, 4 accurate.