A HIPAA compliant software development checklist is mostly a list of engineering tasks: who can log in, what gets logged, how data is encrypted in transit and at rest, how sessions end, how you prove a record has not been altered. Those come from the technical safeguards in 45 CFR 164.312, and they are your developer's job. The rest of HIPAA, the policies, risk analysis, training and contracts, stays with the practice or health business that owns the data. If you are serving patients in the UK or EU, the GDPR adds a second list that most HIPAA checklists skip.

This post gives you both lists, shows where they overlap, and draws the line between what to ask of your developer and what you have to do yourselves. It does not make you compliant. No checklist does, and we will explain why near the end.

What a HIPAA compliant software development checklist can and cannot do

HIPAA's Security Rule has three groups of safeguards: administrative, physical and technical. Software touches all three, but only the technical safeguards can be built into code. The administrative ones (a documented risk analysis, a security officer, workforce training, sanctions policy, contingency planning) are organisational. The physical ones (facility access, workstation security, device disposal) are mostly handled by your hosting provider and your own office practices.

So when a developer says "we build HIPAA compliant software", read it as "we build the technical safeguards the Security Rule describes". That is a real and valuable thing. It is not the same as your organisation being compliant, and a buyer who confuses the two ends up with a secure app and an unfinished compliance programme.

Three terms are worth fixing before the checklist:

  • Protected health information (PHI) is health information that can identify a person, held by a covered entity or its business associate. A name next to an appointment type is PHI. Anonymised aggregate counts are not.
  • Covered entity is the clinic, hospital, insurer or clearinghouse. Business associate is any vendor that handles PHI for them, which includes your developer if they can see live data, and your hosting provider.
  • Required versus addressable: each technical safeguard has implementation specifications. Required ones must be done. Addressable ones must be done or replaced by a documented equivalent. Addressable does not mean optional.

The five technical safeguards, translated into code

The technical safeguards section lists five standards. Here is what each one means when a senior engineer is building a patient portal, a telehealth app or an appointment system.

Access control

Every user has a unique identifier, so actions can be traced to a person and nobody shares logins. There is an emergency access procedure, so a clinician can reach a record when the usual route fails and the access is logged. Sessions end automatically after inactivity. Data at rest is encrypted, with keys managed separately from the data. In a web app this translates to role-based permissions enforced on every request (not just hidden in the interface), a break-glass role with its own audit trail, server-side session timeouts, and encrypted database volumes and backups.

Audit controls

The system records who did what to which record and when. For PHI this means read access as well as writes: the fact that a receptionist opened a patient's file matters even if nothing changed. Audit logs need to be tamper-evident, retained for a defined period, and queryable, because the point of a log is to answer "who looked at this record in March" quickly during an investigation.

Integrity

You need a way to show PHI has not been altered or destroyed improperly. In practice: database constraints, checksums or hashes on stored documents, versioned records rather than in-place edits for clinical data, and backups that are tested by restoring them, not just by confirming they ran.

Person or entity authentication

The system verifies that someone is who they claim before granting access. Password rules alone are weak; multi-factor authentication for staff accounts is the norm now, and anything that integrates with another system (an EMR, a lab) should authenticate the system too, through signed tokens or mutual TLS rather than a shared password in a config file.

Transmission security

PHI in transit is protected against interception and alteration. That means TLS on every connection, including internal ones between services, no PHI in URLs or query strings where it lands in server logs, and care with email and SMS, which are not secure channels unless the content is limited to "you have a new message, log in to read it".

If you are deciding whether to build a portal at all, our earlier post on patient portal development cost for a small clinic puts these safeguards in budget terms.

The GDPR and UK GDPR items a health app adds

HIPAA checklists for developers tend to stop at the safeguards above. If your app serves anyone in the UK or EU, the GDPR and UK GDPR add requirements of a different kind. They are less about how data is protected and more about why you hold it and what people can ask you to do with it.

  • Lawful basis and the special category condition. Health data is special category data. You need a lawful basis for processing it and a separate condition for special category data (explicit consent, or provision of health care, among others). The app has to record which applies, and consent screens have to be real choices, not pre-ticked boxes.
  • Data minimisation and retention. Collect only what the purpose needs, and delete or anonymise when the purpose ends. Retention rules belong in code (scheduled deletion, anonymisation jobs), not in a policy nobody runs.
  • Rights requests. People can ask for a copy of their data, correction, erasure where applicable, and restriction. The portal should be able to export one person's complete record and delete or anonymise it on request without an engineer running database queries by hand.
  • Data protection impact assessment. Large-scale processing of health data almost always needs one before you launch. Your developer contributes the technical description; your organisation owns the assessment.
  • Breach notification. The ICO's guidance on responding to a breach sets a 72-hour window for notifying the regulator where a breach is likely to risk people's rights. That is only achievable if the system can tell you what was accessed, by whom, and which patients were affected. The audit log you built for HIPAA is the same log that makes this possible.
  • International transfers. If data leaves the UK or EU (a US-hosted database, a US-based support engineer), you need a transfer mechanism and a record of it. Hosting region is a design decision, not an afterthought.
  • UK specifics. The Data Protection Act 2018 sits alongside the UK GDPR, and anything used in NHS care will be asked for the NHS Data Security and Protection Toolkit and, for digital tools, the Digital Technology Assessment Criteria. Plan for them if the NHS is a customer.

HIPAA vs GDPR for software: the overlap and the gaps

Area HIPAA (US) GDPR and UK GDPR
Who it applies to Covered entities and business associates Any organisation processing personal data of people in the UK or EU
What data Protected health information All personal data; health data is a special category
Access control, audit logs, encryption Explicit technical safeguards Expected as "appropriate technical measures", not itemised
Why you hold data Permitted uses and disclosures Lawful basis plus a special category condition
Individual rights Access and amendment Access, correction, erasure, restriction, portability, objection
Contracts with vendors Business associate agreement Processor agreement with mandatory clauses
Breach reporting To HHS and affected individuals, timing by breach size To the regulator within 72 hours, to individuals if high risk
Hosting location No explicit rule Transfer rules apply when data leaves the UK or EU

The practical reading: build the HIPAA technical safeguards and you have most of the "appropriate technical measures" the GDPR expects. Then add the GDPR items that HIPAA does not have: lawful basis records, retention automation, rights-request tooling, a processor agreement with every vendor, and a hosting region decision.

The checklist to hand your developer

This is the developer-side list. Ask for each item to be shown working before launch, not described in a proposal.

  1. Unique user accounts with role-based permissions enforced on the server for every request.
  2. Multi-factor authentication for all staff and admin accounts.
  3. Automatic session expiry, with the timeout configurable per role.
  4. An emergency access role that is time-limited and logged separately.
  5. Encryption in transit on every connection, internal services included.
  6. Encryption at rest for the database, file storage and backups, with keys held outside the data store.
  7. Audit logging of reads and writes to patient records, tamper-evident, retained for an agreed period, and searchable by patient and by user.
  8. Versioned clinical records (no silent overwrites) and integrity checks on stored documents.
  9. No PHI in URLs, logs, error messages, push notification bodies, emails or SMS.
  10. Backups that are encrypted, stored in the agreed region, and restored successfully in a test at least before launch.
  11. One-person data export and deletion or anonymisation, runnable by an administrator.
  12. Scheduled retention jobs that delete or anonymise data when the agreed period ends.
  13. A test environment that uses synthetic data only, so developers and support staff never need live PHI.
  14. Dependency and vulnerability scanning in the build pipeline, with a process for patching.
  15. A written description of the architecture and data flows, suitable for your risk analysis and impact assessment.

The last item is the one most often missed and most often needed. Your compliance adviser cannot assess a system nobody has described.

Who owns what: clinic and developer responsibilities

The clearest way to avoid gaps is to agree the split before the build starts. Here is the division we use on healthcare projects.

Responsibility Developer Clinic or health business
Technical safeguards in the checklist above Builds and demonstrates Reviews and accepts
Risk analysis and risk management plan Supplies architecture and data flow documentation Owns and maintains
Policies, training, sanctions None Owns
Business associate and processor agreements Signs as a vendor where PHI is handled Issues and tracks for every vendor
Hosting provider selection and agreement Recommends; configures Signs the agreement
Impact assessment (UK and EU) Contributes the technical sections Owns and approves
Breach response Provides logs and technical investigation Decides, notifies, records
Ongoing patching and monitoring Under a support retainer Funds and reviews reports

Notice how often the developer "contributes" and the organisation "owns". That is the correct shape. A vendor who promises to own your compliance is promising something they cannot deliver.

What a checklist cannot do for you

Three limits are worth stating plainly.

First, HIPAA has no official certification. Any vendor advertising a HIPAA certificate for software is describing a private audit or a self-assessment, which may be useful evidence but is not a regulatory status. The same goes for the GDPR: there are approved certification schemes in some areas, but ticking a list is not one of them.

Second, safeguards decay. A system that passed every item at launch can fail six months later because a library was not patched, a new feature logged a field it should not have, or a contractor was given an admin account and never removed. Compliance is maintained, not achieved, which is why healthcare software needs a support arrangement with security patching in scope rather than an ad hoc fix when something breaks.

Third, the checklist cannot tell you whether your product should collect the data in the first place. Data minimisation is a design question. A booking system that stores a reason-for-visit free text field has created a special category data store that may not have been needed. The cheapest safeguard is the data you never collect.

Build the safeguards in from the first sprint

The decision for a buyer is simple to state: insist that the technical safeguards for healthcare software are in scope from the first sprint, in writing, with a demonstration of each before launch, and keep the administrative side with your own organisation or a compliance adviser. Retrofitting access control and audit logging onto a finished app costs more than building them in, and the finished app usually has to be partly rebuilt to accommodate them.

We build patient portals, telehealth and appointment systems with these safeguards as standard; the detail is on our healthcare software development page. If you already have a system and are not sure which items it meets, a code audit is the fastest way to find out; our earlier post explains what a code audit covers.

If you are planning a healthcare build or trying to bring an existing one up to standard, book a free 30-minute discovery call. A senior engineer talks through your data, your users and the regulations that apply, and we send a written approach, timeline and estimate within two business days.