Care Home Medication Audit: A Monthly Template and AI Action Log

Run a monthly medication audit by sampling records and counts against a fixed checklist, scoring each item yes, no or not applicable, and logging every "no" as an action with an owner and a date. AI can summarise the anonymous totals and draft action wording, but it never sees resident identifiers and never judges clinical significance.

Medicines are among the highest-risk records in a care home, so the boundary has to be firm. Treat the audit as a records-and-counts exercise: is the chart complete, does the stock match, was the paperwork done. Anything that needs a clinical view, such as whether a missed dose mattered, goes to a person with the right competence and your normal incident route.

Follow me on Instagram@sagnikteaches

What this audit covers, and what it leaves alone

The audit answers one question: can we show, from our records, that medicines were handled the way our policy says? It does not review whether a prescription is appropriate, whether a resident should be on a medicine, or whether an error caused harm. Those belong to prescribers, pharmacists and your clinical lead.

Connect on LinkedInSagnik Bhattacharya

Who does what in the month

Split the work so one person does not check their own shifts. A workable pattern for a 30-bed home: a senior from a different unit carries out the checks on days 1 to 5 of the month; the deputy compiles the totals on day 6; the registered manager reviews and signs the action log on day 8; actions are due within 14 days unless the manager sets a different date. Put those days in your policy so the audit is a routine rather than something done "when there is time".

Subscribe on YouTube@sagnikteaches

The checklist: 20 items in four groups

Use this as a starting point and adapt it to your medicines policy and the systems you use, whether paper or electronic. For each sampled chart or item record yes, no or not applicable. A "no" always needs a short note, written without resident names (use the resident's audit code, which is a number your home assigns and keeps in a separate key).

Group A: administration records

1. Every scheduled dose in the audit period has a signature or an approved code.
2. Any non-administration code has a reason recorded.
3. Where a medicine was refused or unavailable, the follow-up is recorded, for example who was told.
4. "When required" medicines record the reason given, the time, and the outcome.
5. Handwritten entries were made only where your policy allows them, and a second trained member of staff checked and signed each one before it was first used.
6. Allergy and sensitivity information is present and matches the care record.
7. Any change instructed by a prescriber is dated, signed and traceable to its source.

Group B: stock and counts

8. The balance of each sampled boxed medicine matches the record.
9. Controlled medicines register entries are complete and balances agree on the day.
10. Date opened is written on liquids, creams and eye drops that need it.
11. Nothing past its expiry or "use by" date remains in the trolley or cupboard.
12. Medicines for return or disposal are logged and stored separately.

Group C: storage and environment

13. Fridge and room temperatures were logged on every day in the period, within the limits in your policy.
14. Out-of-range temperatures have a recorded action.
15. The trolley and cupboards are locked when unattended, and key handling follows policy.
16. Labels are legible and match the chart.

Group D: governance

17. Staff who administered medicines in the period have a current competency record.
18. Last month's actions are closed or have a revised date and reason.
19. Every incident, near miss or concern in the period has a matching incident report.
20. The audit was done by someone who did not administer on those shifts.

Where your policy sets numeric limits, such as a temperature range, put the figure in the checklist wording itself so the checker does not rely on memory. No figures are given here, because they vary with the product and your pharmacist's advice.

A worked example: one month, 12 charts, 20 items

Illustrative figures for a 30-bed home. The checker samples 12 charts and runs the 20-item checklist. Not every item applies to every chart, and groups C and D are checked once for the home rather than per chart, so the numbers below are the answers recorded.

That makes 11 "no" findings. A percentage can be calculated, but resist ranking units by percentage alone, because a score of 92% can hide a single missed controlled-medicine entry that matters more than ten small gaps. The action log is the real output.

Turning findings into an action log

Each "no" becomes a row: audit code, group and item number, what was found, action, owner (a role, not an individual's name, where you will share the summary), due date, status and date closed. Illustrative rows:

Where AI helps: anonymous totals in, a tidy summary and drafts out

AI earns its place after the checking, not during it. Once the findings are totalled, a tool can turn a column of numbers and short notes into a half-page summary for the provider meeting and propose action wording. It can also compare this month with the last three to spot a pattern that a human reading each month in isolation would miss.

What goes in: totals per group, findings written with item numbers and a short factual description, and last month's totals. What stays out: resident names or codes if your key could be rejoined by the recipient of the output, medicine names tied to a person, room numbers, photographs, staff names, and any free text that was copied from a chart. If a finding cannot be described without a resident detail, describe it generically ("one non-administration code with no reason") and keep the detail in your own record.

Check with your data-protection lead before using any tool, and apply the rules in the staff policy for AI in care homes. If your medicines system has built-in AI features, ask the supplier how they handle data; the care home software checklist lists the right questions.

The prompt

You are helping a care home manager summarise a monthly medication record audit. The data below contains counts and short, anonymous findings only. Write:
1. A five-line summary: sample size, items checked, number of "no" findings by group, and the main theme.
2. A comparison with last month's totals: say what rose or fell. Do not guess causes.
3. For each finding, one suggested action in plain words, with a suggested owner role and a due-in-days figure that the manager can change.
4. A line "Items needing senior clinical or manager judgement" listing any finding that involves stock differences, controlled medicines, or a possible missed dose. Do not assess seriousness.
Do not give clinical advice. Do not infer identities. Do not invent findings. If a number does not add up, say so.

This month:
[paste totals and findings]
Last month:
[paste totals]

What a good output looks like, and what to fix

An illustrative reply: "Summary: 12 charts sampled, 152 items checked, 11 'no' findings (administration 6, stock 3, storage 1, governance 1). Main theme: incomplete reasons and signatures on handwritten or non-administration entries. Compared with last month: administration findings rose from 4 to 6; stock findings unchanged. Items needing senior judgement: stock finding B9-1 (a controlled medicines balance difference)." Check three things before it goes anywhere. First, do the numbers match your tally? 6 + 3 + 1 + 1 is 11, so yes. Second, has the tool softened anything? If a stock difference appears as "minor discrepancy", reword it to the factual description. Third, has it stated a cause ("due to staff shortages") that nobody recorded? Delete it.

A second example: a small home with a different problem

Take a 12-bed home with one senior per shift and no deputy. The same 20 items would be too heavy, so the manager runs a "short form" every month (items 1, 2, 4, 5, 8, 9, 11, 13, 18 and 19) on every chart, because there are only twelve, and the full list every third month. Illustrative result for one month: 87 item checks (seven chart items on twelve charts, plus the three home-level items), 4 "no" findings, all in group A, and three of them on "when required" records where the reason was ticked but the outcome was blank. The pattern is clear without any tool: staff record why but not what happened. The action is a one-line change to the chart prompt and a five-minute huddle at handover, with a re-check in two weeks. This is the kind of finding that a percentage hides and a short, anonymous written summary shows.

"When required" records: the item with the thinnest trail

Medicines given as needed leave a thinner trail than scheduled ones. Three things should be present: why it was offered, what was given at what time, and what the effect was when staff checked. Say the checker finds an entry that reads "given, 14:20" and nothing else. In the log it becomes: "A4-1: reason and outcome not recorded on one when-required entry." The suggested action from the tool might read "Remind staff to record reason and outcome for every when-required dose". That is fine as a draft. What the manager adds is the human decision: whether to raise it at the next team meeting, whether the chart layout makes the outcome easy to skip, and whether the prescriber's instructions need clarifying with the pharmacist. A tool cannot make those calls, and it should not try.

Briefing the person who does the checking

The quality of the audit depends on the checker, so give them a one-page brief: the sample list, the checklist, the rule about not using resident names in notes, and what to do on the spot if they find something urgent. "On the spot" matters. If the checker finds a stock difference in a controlled medicine, a cupboard left open or an entry that suggests a dose was missed, they stop the audit and tell the senior on duty at once, following your policy. The audit form is not the place for urgent concerns to sit for a week. Add a line at the top of the form: "If you find anything that needs action today, act first, then record it here."

Also say what the checker should not do: correct entries themselves, add missing signatures from memory, or discuss findings with residents. Corrections follow your normal route for amending a medicines record. Widely used care home medicines guidance describes that route as a single line through the error, with the correction, a signature, the date and the time written beside it, and never correction fluid. Add an item for it if your charts show crossings-out, because an over-written dose is as hard to audit as a missing one.

Reading the trend: a six-month table

A single month tells you little. Keep a trend table and ask the tool to comment on it only after you have looked yourself. Illustrative figures, one row per month:

The picture is of findings falling from 11 to 4, but the repeat column shows that three findings from month 1 came back in month 2. That is the number to put in front of the registered manager. When you ask the tool for commentary, tell it to describe movement only and to avoid causes. If it writes "improvement is due to better staff engagement", delete it, because nothing in the table says that.

Choosing the sample so it is fair

A sample is fair when no one can predict it and it covers the whole home over time. Number the residents by an audit code from 1 to the number of beds. For a sample of twelve in a 30-bed home, use a random-number method, or "start at the day of the month, then every third code", and record which method you used. Add one chart from each unit, one from each shift pattern and, where relevant, one from every resident with a recent change to their medicines, because a change is a common point for transcription errors. Rotate so every resident is checked at least once a quarter. Record the codes used each month in the audit file, not in the AI prompt.

Stock counts: the part AI should not touch

Counts and balances are arithmetic on real boxes in a real cupboard. If a controlled medicines balance does not agree with the register, a person must follow your policy at once: recount with a second person, check the previous entries, and escalate as required. A tool cannot do any of this, and the audit's role is to record that it happened. Never paste balances for individual medicines into a chat tool. Keep AI to the summary after the count has been resolved or escalated.

Where this audit goes wrong

Checking the tidy charts. Letting the unit choose which charts to show you gives a flattering result. Draw the sample yourself from a full list, for example every third resident on a numbered list.

A score replaces the actions. A page of percentages looks impressive and changes nothing. Review the action log first at every governance meeting.

The same finding returns each month. A repeating finding means the action did not address the cause. Add a column "repeat?" to the log, and when a finding repeats twice, move it from a reminder to a change in routine: a new item on the shift checklist, a different tool, a short retraining.

Audit fatigue. Twenty items on twelve charts every month is a lot for a small team. Split the checklist so groups A and B run monthly and C and D alternate, if your policy allows, or reduce the sample when there have been three clean months. Do not drop it altogether.

Findings kept in the AI tool. The tool is not the record. Save the signed audit and action log in your own system, and decide with your data-protection lead whether prompts and outputs should be deleted from the tool's history.

Keeping the audit useful after the first few months

Audits decay when they become a ritual. Every quarter, ask the checkers what they dislike about the form, which item never produces a finding, and which finding the form does not capture. Drop or reword items that always score yes, and add one for any new risk, such as a new medicines system or a change of pharmacy supplier. Keep a version number and date on the checklist so that old audits can be read against the form used at the time. Small, regular edits keep the audit honest without making it longer.

Signing off the audit

Add a final block to the audit form: the date, the person who checked, the person who reviewed, a statement that the AI summary was read against the raw totals, and a tick that any incident or concern was reported through the normal route. Two signatures make a record that stands up when you are asked about it later.

Month to month, build up a small trend sheet: sample size, "no" findings by group, repeat findings, actions closed on time. After six months you will see whether the work is changing outcomes. The wider monthly audit round and the mock inspection tutorial show how to combine this with other checks, and the training tracker covers competency dates. If you would like someone to look at your forms with you, that is what my AI implementation consultation covers.

Medication audit questions care managers ask

Can AI tell us whether a medication error was serious?

No. Judging the clinical significance of an incident belongs to a trained senior, your medicines policy, the prescriber or pharmacist and your incident process. The AI step here only summarises counts of missing signatures, gaps and stock differences, and drafts action wording for a person to approve.

What should never be pasted into an AI tool?

Resident names, dates of birth, room numbers, government identifier numbers, medicine names linked to a person, photos of charts and anything that could identify someone. Use audit codes and totals only. If in doubt, leave it out and do that step by hand.

How many records should we sample each month?

Use a sample you can complete properly every month, and sample more where you have found problems. One common approach is every chart on a small unit and a fixed share on a large one. Agree the method in your medicines policy so month-to-month figures are comparable.

Who should sign off the audit?

A senior who is competent in medicines management, usually the registered manager or a deputy, and ideally not the person who did the checking. Add a second signature for any finding that involves a stock discrepancy or a possible missed dose.

Further reads

• A Simple AI Risk Register for Small Businesses (With Template) — A way to record the risks of using AI inside your own audits.
• AI Tools for Facility Management in a Care Home: Maintenance Logs — Another dated-checks log: maintenance and safety tests.
• Care Home Daily Notes: Better Wording With AI, Checked by Staff — A sampling routine for daily notes, built the same way.
• Care Home Menu Planner: Four-Week Cycle With AI and Dietary Checks — Another resident-safety record: dietary needs and allergens.
• How to Stop AI Tools Training on Your Business Data — Settings that stop a tool keeping what you paste in.
• How to Roll Out an AI Policy So Staff Actually Follow It — How to get a policy followed on the floor.
• Domiciliary Care Software UK: What the AI Features Actually Do — Birdie, CareLineLive and Nourish compared on the AI each really has, what it costs per hour or per carer, and the inspection evidence it creates.
• AI Policy for Aged Care: What to Cover and What Staff Must Not Paste — A checklist for an Australian aged care provider's AI policy: approved tools, the never-paste list, privacy impact assessment, consent and an incident log.
• Care Home Activities Calendar: Plan a Month With AI and Resident Input — Turn coded resident interest sheets into a monthly activities calendar with group and one-to-one slots, a printable version and a staff check.
• Care Home Policies and Procedures UK: Keeping Them Current With AI — Use AI to compare each care home policy with current standards, keep tracked changes and a review log, and get staff sign-off before anything is published.
• Care Inspectorate Self-Evaluation: Draft the Evidence Summary With AI — A Scotland-specific worked self-evaluation: quality framework headings, an evidence summary drafted with AI prompts, and a provider who owns every judgement.
• Care Staff Supervision Template: Notes and Action Logs With AI — A care-specific supervision agenda, record format and action log, with AI drafting the actions from your notes while staff data stays inside approved tools.
• Infection Control Audit Action Plan: Turn Findings Into Fixes With AI — Paste anonymised audit findings into AI, get grouped fixes with owners and dates, then a re-audit sheet that proves each one closed.
• AI Use in Long-Term Care Facilities: Where Resident Data Must Not Go — A long-term care operator's guide to AI: which jobs are safe, which resident details never leave your system, and the staff rules to print.
• AI and Senior Living: Where a Small Care Home Should Start — A small care home's starting order for AI: family updates, activities, rotas and enquiry replies first, safeguarding and resident records kept human.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: published care home medicines guidance on handwritten administration records (second trained member checks and signs), corrections (single line, signature, date and time, no correction fluid) and 'when required' medicines, checked 4 November 2026; your own medicines policy. The checklist is a record-completeness template, not clinical advice.