An app for home health care has one job: prove the visit happened.

Six things a visit record has to verify

Federal law defines what an electronic visit verification system has to do, and the definition is a list of six items. The system verifies the type of service performed, the individual receiving the service, the date of the service, the location of service delivery, the individual providing the service, and the time the service begins and ends. That is the entire list. Nothing in it is about quality of care.

The requirement bites where Medicaid pays for a service that needs an in-home visit. States had to require it for personal care services from 2020 and for home health care services from 2023. If Medicaid never pays you, none of it binds you. It is still the cheapest specification of a visit record anyone will hand you.

Two of the six are where homemade tools fail. Location of service delivery means the location, not the caregiver's account of it. Begins and ends means both ends, each captured as it happens. A clock in at the door and a clock out typed up that evening captures five and a half of six. The half is the one in dispute.

States differ on which systems they accept and on whether visit data has to reach a state aggregator. We have not checked any individual state's rules. Ask your state Medicaid agency what it accepts before you build.

42 U.S. Code 1396b(l), electronic visit verification system requirement

A rota, not a work queue

Scheduling is where an in-home care app stops resembling field service software. A field service scheduling app sends the nearest qualified person to the job. A care schedule has to send the same person to the same house. Continuity is most of the service, and a client with dementia should not meet a stranger at seven in the morning.

The second difference is authorization. Hours are approved in advance, per client, per week, by whoever pays. Scheduling past that number is not an overtime problem, it is work nobody will pay for. The screen that matters is the week measured against the authorized hours, with the gap visible before anyone is assigned.

Then the ordinary cruelties of the job. Travel between houses that nobody counts as paid time. A visit window that is not a start time. A caregiver who calls out at six with four visits on her day. An app that cannot move a visit to somebody else in under a minute, on a phone, gets replaced by phone calls by the second week.

The plan of care is the real specification

For a Medicare certified home health agency, the work in the home is set out in an individualized plan of care. The regulation lists what it must contain. Diagnoses. Mental, psychosocial and cognitive status. Services and equipment required. Visit frequency and duration. Medications and treatments. Safety measures. Measurable outcomes and goals. A description of the patient's risk for emergency department visits and hospital re-admission.

The same rule says the plan must be reviewed and revised as often as the patient's condition requires, and no less often than once every 60 days. Verbal orders get documented, signed, dated and timed by whoever takes them, then authenticated and dated by the physician or allowed practitioner. That is a better data model than most product briefs contain.

Private pay companion care sits outside that regulation, and the shape is still worth copying. Whether the visit is skilled nursing or two hours of help with a shower, the caregiver should open the current plan for this client today. Not a note somebody pasted in March. Families see a version of it in a senior care app, and both should read from one plan.

42 CFR 484.60, care planning, coordination of services and quality of care

A house is a hostile place for software

Every hard problem in this category is physical. Basement apartments and steel framed senior housing eat cell signal. The phone is the caregiver's own, three years old, and low on battery by the fourth visit. Hands are gloved, the client is watching, and nobody has two minutes to fight a form.

That argues for very few taps and a record that survives interruption. Clock in, see today's plan, tick what was done, note what changed, capture a signature if the payer wants one, clock out. All of it has to queue and send later, writes included. A design that reads offline but cannot write offline loses the data the app existed to collect.

The other half is the visit that does not happen. Nobody home, client refused, no access to the building: those are outcomes, not blanks, and each needs recording in seconds with a reason. Before building any of it, get clear on what a phone really does with no connection. Most disappointment here starts with an assumption about working without a signal in the home that was never true.