Building a Home Care QA Program When You Don't Have a QA Department

Building a Home Care QA Program When You Don't Have a QA Department

The regulation never asks for a department. 42 CFR 484.65 requires a Medicare-certified home health agency to develop, implement, evaluate, and maintain an ongoing, agency-wide, data-driven quality assessment and performance improvement program. It assigns responsibility for that program to the governing body. It says nothing about headcount, a job title, or a reporting line.

That distinction matters, because most agencies have no dedicated QA function. There were 12,234 home health agencies participating in Medicare in 2024, and most of them are small. In a forty-patient agency, the clinical manager is the QA program, the intake supervisor, and often the weekend on-call nurse.

So the useful question is not how to staff a QA department. It is which artifacts a surveyor will ask to see, and how to produce them with the people already on payroll.

What 484.65 actually requires, standard by standard

Five standards, and they are shorter than their reputation:

• Program scope. The program must be capable of showing measurable improvement in indicators tied to health outcomes, patient safety, and quality of care. You must measure, analyze, and track quality indicators, including adverse patient events.
• Program data. The program uses quality indicator data, including OASIS-derived measures where applicable. The governing body approves how often data is collected and how detailed it is.
• Program activities. Improvement activities focus on high-risk, high-volume, or problem-prone areas. Problems that threaten patient health and safety must be corrected immediately. Adverse events get tracked, causes analyzed, preventive actions implemented — and then you measure whether the fix held.
• Performance improvement projects. Required since July 13, 2018. The number and scope of projects must reflect the scope, complexity, and past performance of the agency's services and operations. Each project must be documented, including the reasons for it and the measurable progress achieved.
• Executive responsibilities. The governing body ensures the program is defined, implemented, and maintained, that improvement efforts address priorities, and that findings of fraud or waste are addressed.

Two things the text does not say are worth noting. It does not set a required number of projects per year. And it does not prescribe a method. A short list of indicators, honest numbers, and documented follow-through satisfies more of the regulation than a polished dashboard with no corrective action behind it.

One scope caveat that trips up multi-line operators: 484.65 binds Medicare-certified home health agencies. Private-pay and Medicaid-only personal care lines are not reached by it. Their quality obligations come from state licensure, which varies substantially from state to state, and from payer contracts. If you run both under one roof, confirm which entity each rule actually reaches before you write one policy for both. Your state survey agency, not a template, is the authority on the licensure side.

Where the errors actually are

If you are choosing what to look at first, let the payment data choose for you. CMS's FY2025 Medicare Fee-for-Service supplemental improper payment data puts the home health improper payment rate at 6.9%, with a 95% confidence interval of 5.3% to 8.5%, and projected improper payments of roughly $1.07 billion. The composition is the part that should redirect your QA plan:

Error type	Share of home health improper payments (FY2025)
Insufficient documentation	49.4%
Medical necessity	30.4%
No documentation	11.8%
Other	4.8%
Incorrect coding	3.6%

Insufficient documentation plus no documentation is more than 61% of the dollars. That is a records problem, not a clinical competence problem. We went deeper on that split in our breakdown of insufficient documentation as the largest home health payment error.

A QA program built only on OASIS outcome measures will never touch the largest category. Outcome measures tell you how patients did. They do not tell you whether the record supports what you billed.

The five artifacts that constitute a program

This is the whole kit. Build these five and you have a defensible program regardless of your org chart.

1. A written QAPI plan, two to four pages, naming your indicators, your data sources, your collection frequency, and who reviews what.
2. A data source inventory — which system each indicator comes from, and who pulls it. If an indicator has no owner and no system, drop it from the plan.
3. An adverse event log with date, description, cause analysis, action taken, and the date you re-measured.
4. One or two project charters, each with a baseline number, a target, an intervention, and a re-measurement date. Two real projects beat six aspirational ones.
5. Governing body minutes showing the body reviewed the data and approved the collection frequency. The regulation puts this on the governing body; minutes are how you show it happened.

Pick indicators you can produce weekly

OASIS-derived measures arrive on a long lag. A program that only runs on them cannot show the immediate correction that the program activities standard calls for. Pair them with near-term operational signals you can count by Friday: referral-to-start-of-care interval, missed visits, complaints that closed without a documented resolution, medication questions escalated to a nurse, patients whose family called more than twice in one week.

Make your review sample defensible

If part of your program involves reviewing patient interactions, the sample size is the whole argument. A two percent review of calls cannot support a claim about the other ninety-eight percent, and a surveyor is entitled to ask what the basis for your conclusion was. We argued that case in detail in 100% call review vs sampling.

Give the program a cadence, not a season

The most common failure mode in a small agency is not a missing plan. It is a plan that gets attention in the two weeks before a survey and none in the ten months between. The program activities standard asks you to correct problems immediately and then measure whether the correction held; neither of those is an annual activity.

A workable cadence for an agency without dedicated QA staff is weekly, monthly, quarterly. Weekly: one person spends thirty minutes pulling the near-term indicators and adding anything new to the adverse event log. Monthly: the clinical manager and administrator review the log, close out completed actions, and re-measure anything that was fixed more than thirty days ago. Quarterly: the governing body sees the data and the project status, and the minutes record it.

Thirty minutes a week is a real budget, and it is roughly what the program costs once the artifacts exist. The expensive part is building them the first time. The same cadence logic applies to agent and clinician coaching, which we broke down in a weekly coaching cadence that works.

Resist one temptation: do not fold your QAPI indicators into a single monthly report that nobody acts on. The regulation asks for measurable improvement and documented corrective action. A report is evidence of measurement. It is not evidence of improvement.

What to do this week

1. Write down the indicators you already produce. Most agencies discover they have six to eight and had never called them a QAPI program.
2. Delete any indicator in your current plan with no data source. Unproducible indicators are a survey finding, not a goal.
3. Open a single adverse event log in whatever tool you already use. Backfill the last ninety days from memory and email.
4. Pick one project tied to documentation sufficiency, since that is where the CERT dollars are. Write a baseline number today, even a rough one.
5. Put QAPI on the next governing body agenda and capture the approval of your collection frequency in the minutes.

Where conversation data fits

A large share of what a home health QA program needs to see happens on the phone and never reaches the record: the referral that stalled, the complaint that was handled verbally, the caregiver who asked the same medication question three times. Call and interaction intelligence platforms, SurfacerIQ among them, exist to turn that conversation volume into reviewable, countable signal rather than audio nobody opens. That is a category of capability worth understanding before you decide your indicators have to come from the EMR alone.

Consider a hypothetical: an agency notices that eleven patients in a month had two or more inbound calls about an unresolved supply issue. None of those calls generated a record entry. That is a project charter with a baseline, found in data the agency already had.

This post is general information, not legal advice; confirm how these requirements apply to your agency with your own counsel and your state survey agency.

If you want to talk through what a QA program looks like when the phones carry most of the signal, talk to us.