Does HIPAA Require Patient-Level Audit Logs?
HIPAA audit log requirements don't list fields. See why breach scoping and risk-based enforcement push PHI data shares toward patient-level, six-year logs.

Not in so many words. HIPAA's audit log requirements are short. The Security Rule tells you to record and examine activity in systems that contain electronic PHI. It doesn't list the fields a log must capture, and HHS has said as much. Still, if you hold PHI for many patients, the rest of the rules push hard toward patient-level access logs that are tamper-resistant, and our conservative reading of the documentation rule is to keep them for six years. The breach notification rules ask you to identify the individuals affected. Enforcement asks whether your controls match your risk.
That gap between what the text says and what an investigation will ask is where a lot of health-tech teams get surprised. This post walks through it with a scenario from our vCISO work and gives you a table to test what your current logs can actually prove. It's general guidance, not legal advice.
What does the HIPAA text actually require?
The audit controls standard, 45 CFR 164.312(b), asks covered entities and business associates to implement mechanisms that "record and examine activity in information systems that contain or use electronic protected health information." A separate required specification in 164.308 asks you to regularly review records of system activity, "such as audit logs, access reports, and security incident tracking reports."
That's the whole text. OCR's January 2017 newsletter on audit controls says directly that the Security Rule "does not identify what information should be collected from an audit log or trail." Instead, you decide based on your risk analysis and your organization's capabilities. The same newsletter does give examples. Application audit trails can log "the creating, reading, editing, and deleting of application records associated with ePHI." It also says logs should be protected from tampering and that access to them should be restricted.
So the honest starting point is that log scope is a documented, risk-based decision. Anyone who tells you HIPAA mandates a specific field list is overstating it.
A composite scenario: query logs aren't patient logs
Here's a pattern we've seen in vCISO work, with details changed. A health-tech startup has a read-only warehouse share of a hospital customer's EHR data, PHI included. Its logging looks solid on paper. For every query, it records who ran it, when, the full query text and the tables or views it touched.
During a security review, the question came up: if an employee looked at records they shouldn't have, could the company say which patients were affected? With query-level logs, the honest answer was "maybe, with effort." The logs showed that a query ran against a view. They didn't show which patient rows came back.
Our advice was to add patient-level access logs that record which workforce member accessed which patient's records, store them so they can't be quietly edited and keep them for at least six years. That's a reasonable recommendation. It's also new logging infrastructure, and it surfaced in the middle of a customer deal.
What can your current logs prove?
Run your logging through the questions an investigator, a customer's security team or your own incident responders will ask. The middle column is typical of query-level logging. The right column is what patient-level logs add.
| Question you'll be asked | What query-level logs can show | What patient-level logs add |
|---|---|---|
| Which patients did this user access during the incident window? | The queries and the tables or views touched, but not the rows returned | A list of patient identifiers you can give the covered entity for notification |
| Was PHI actually viewed or acquired? | That a query ran, and sometimes how many rows it returned | Which records were returned to which person |
| Who looked at this specific patient's record last year? | Only by re-running past queries against the data as it existed then, if that's still possible | A direct lookup by patient and date range |
| Can you show the log wasn't altered? | Depends on where it's stored and who can administer it | Little on its own. Proof comes from append-only storage and separate administration, which you need for any log you rely on. |
| Can you answer for an event several years back? | Depends on how long your platform keeps query history, which may be far shorter than six years | Whatever retention you set and enforce |
If most of your answers land in the middle column, you can probably reconstruct some of the picture. Reconstruction is weaker evidence, though. It depends on past data states and on logs that may themselves be editable. An immutable record of who accessed which patient is the strongest position to argue from.
Why does breach scoping push you toward patient-level logs?
Under 45 CFR 164.402, an impermissible use or disclosure of PHI is presumed to be a breach. That presumption holds unless a risk assessment shows a low probability that the PHI was compromised. One of the factors in that assessment is "whether the protected health information was actually acquired or viewed."
When there is a breach, a business associate must notify the covered entity. Under 164.410, that notice has to identify, "to the extent possible," each individual whose PHI has been, or is reasonably believed to have been, accessed, acquired, used or disclosed. The covered entity then notifies those individuals under 164.404. Under 164.414, the burden of showing that notifications were made, or that no breach occurred, sits with the covered entity or business associate.
Put those together. If you can't show which records an employee accessed, you may not be able to argue that most patients' data wasn't viewed. The safer reading becomes that everyone in the dataset may be affected. For a vendor holding a hospital's data, that can mean reporting everyone in the dataset to the covered entity, which then has to notify them, not just the handful of patients whose records were actually viewed.
How long should you keep HIPAA audit logs?
The Security Rule's documentation standard, 45 CFR 164.316(b), requires a written record of any action, activity or assessment that the rule requires you to document. That record must be kept for six years from creation or from when it was last in effect, whichever is later.
The rule doesn't say "audit logs" in that sentence. Our conservative interpretation is to treat audit logs as records of activity and keep them for six years. NIST's log management guide, SP 800-92, lists this among HIPAA's log management needs, noting that documentation of actions and activities needs to be retained for at least six years. That's an interpretation, not settled law. HHS's Security Rule guidance page lists NIST publications such as SP 800-66 Rev. 2 as informational resources, not legally binding guidance. Some organizations document a shorter retention period based on their risk analysis. If you do, write down why.
What's the counterargument, and where does it break down?
The counterargument is fair. The text isn't prescriptive. OCR's own newsletter says scope comes from your risk analysis. NIST SP 800-66 Rev. 2 frames audit scope as something you determine from your risk assessment, and it poses "what should the audit record include?" as a question for you to answer. A company with a small, tightly controlled PHI footprint can reasonably decide that system-level and query-level logging is proportionate.
Where it breaks down is scale. Under the Security Rule's flexibility standard, what counts as reasonable and appropriate depends partly on the probability and criticality of risks to ePHI. A startup with read access to records for a very large patient population carries about as much risk as a data holder can. A decision to skip patient-level logging there needs a strong written rationale, and it will be tested at the worst possible moment.
Healthcare buyers also bring their own frame of reference. Health IT certified to ONC's audit-logging criteria must be able to record actions on electronic health information following parts of ASTM E2147-18, a standard built around logging access to patient-identifiable information. In our experience, hospital security teams that work with those EHR audit trails expect vendors to answer at a similar level.
HHS also proposed a stronger Security Rule update in January 2025, including revised audit trail and logging requirements. The latest regulatory agenda lists it as a long-term action with final action estimated for July 2027. Until a final rule exists, it doesn't change what applies today.
What does this mean for a health-tech vendor?
In our experience, HIPAA's ambiguity is one reason large health systems often ask vendors for HITRUST certification, frequently the risk-based r2. A shared, assessed control framework replaces a debate about what "reasonable and appropriate" means. Our comparison of HITRUST and SOC 2 covers how that differs from a SOC 2 report. If you're heading that way, our guide to choosing a HITRUST consultant helps you scope the engagement.
The practical lesson is about timing. A single logging question can turn into unplanned engineering work in the middle of a deal. Decide your logging scope during your risk analysis, before a customer's security team decides it for you. Record the decision, the data volumes it assumed and the retention period. Our breakdown of SOC 2 vs. HIPAA explains why an existing SOC 2 report won't settle this question either.
Workstreet's vCISO team works through decisions like this with health-tech companies as part of their HIPAA programs.
Get a vCISO review of your PHI logging scope before your next healthcare deal reaches security review.
HIPAA audit log questions
Does HIPAA say how long to keep audit logs?
Not explicitly. The Security Rule requires six-year retention for documentation of required actions and activities. Our conservative position is to treat audit logs as that kind of record and keep them for six years.
Is a query log enough for HIPAA?
It can be, if your documented risk analysis supports it. For large PHI datasets, query logs often can't show which patients were accessed. That gap matters when you have to scope a breach.
Do de-identified datasets need patient-level logs?
Data that meets HIPAA's de-identification standard isn't PHI, so the Security Rule doesn't apply to it. Confirm that the dataset actually meets that standard before you scale back logging.



.avif)