In short
- House call scheduling works in two steps: a plan that decides which patients each provider should see this week based on visit frequency and capacity, and a daily route that orders those visits by location, time windows and travel time.
- Routing engines such as Google's Route Optimization API solve the second step well, but Google Cloud's HIPAA page does not list Maps Platform or Route Optimization among the services its BAA covers, so send the engine coordinates and opaque IDs and keep names and diagnoses out of the request.
- Each visit record should store the place of service and the clinician's time on qualifying activities apart from drive time, because Medicare's home or residence visit codes and prolonged service rules depend on both.
House call scheduling software decides which patients each provider sees on which day, then orders each day's visits into a route. It has to respect how often each patient needs a visit, how many visits a provider can make in a day, when patients and facilities can receive them, where everyone is, and how long it takes to get between stops. A well-built system keeps planning and routing as separate steps, keeps working when the phone loses signal, syncs with the EHR, and records each visit in the form billing needs.
What does a house call scheduler need to know?
Before any algorithm runs, the data model has to describe the real world. Three objects carry most of it.
- Patients: the visit address and its type (private home, assisted living, nursing home, hospital), geocoded coordinates, access notes such as gate codes or a contact to call on arrival, preferred visit windows, and the visit frequency their care plan sets, such as weekly, every two weeks or specific weekdays.
- Providers: start and end location for the day, working hours, breaks, a daily visit cap, skills or credentials that limit which patients they can see, and planned time off.
- Visits: the visit type, its expected duration on site, whether it is new or follow-up, and its status (planned, confirmed, en route, done, missed, cancelled).
Facilities add their own rules. A nursing home may only accept visits in certain hours, and several patients at one address should be seen in one stop. Model the facility as its own object with its own windows, and attach patients to it.
Why plan first and route second?
Frequency and routing work on different time scales. Frequency is a weekly or monthly question: is this patient due, and which provider should see them? Routing is a daily question: in what order, and when will the provider arrive? Solving both at once makes the plan unstable, because a small change in today's traffic can move a patient to a different week.
- Step one, the plan: for the coming period, list every patient due a visit, assign each to a provider and a day, and respect each provider's daily cap and days off. Keep the same provider for the same patient where you can.
- Step two, the route: for each provider and day, order the assigned visits to cut travel time while meeting time windows, then compute arrival times.
- Step three, the review: a scheduler or the provider checks the result, can move visits by hand, and confirms the day.
One worked example: we built a house-call scheduler for a US clinic where each doctor visits up to 12 patients a day at homes, nursing homes and hospitals. Administrators plan the house calls, and an algorithm builds each doctor's route from the visit frequency (biweekly, weekly or specific days), a working day capped at 12 patients, patient locations and the distances between them, calculated through Google Maps. A team of 5 specialists delivered it in 6 months.
How does route optimization work?
Ordering stops for several vehicles under time windows is a well-known optimization problem, and you rarely need to write your own solver. Google's Route Optimization API is one option. Its documentation describes the inputs and outputs:
- Shipments: tasks with locations, service durations and load sizes. For house calls, each visit is a task, the service duration is the expected time on site, and load does not matter.
- Vehicles: the fleet, with travel modes, start and end locations and capacities. Each provider is a vehicle.
- Time constraints: shipment time windows, driver shift times and required breaks.
- Costs: per-distance and per-hour costs, and penalties for skipped tasks, which tell the solver how much a missed visit matters.
- Outputs: the visit order for each vehicle, which vehicle handles which task, estimated arrival and departure times, and total cost, distance and time per route.
The API offers a blocking method, OptimizeTours, for requests with short solve times, and non-blocking methods for long solves and batches, which return a long-running operation ID. A clinic that plans tomorrow's routes overnight can use the batch path. A scheduler re-routing a day in progress needs the fast one.
What happens when the day changes?
A house call day rarely runs to plan. A patient cancels, a visit runs long, a new urgent visit appears. The software should handle each without rebuilding everyone's day.
- Lock completed and in-progress visits, and re-optimize only what is left.
- Let the provider reorder their own route, since they know the parking, the stairs and the patient who needs a morning visit.
- Show missed visits on a report so they return to the plan for the next available day.
- Notify the next patients when an arrival time moves.
In the clinic's scheduler, doctors see their daily route as a list of patients and addresses or on a map. They change the order, open each patient's records and past house calls, and confirm or cancel visits.
How do you keep patient data out of the mapping service?
A route is health information. A list of addresses visited by a home-based medical practice, with times and names, tells anyone who sees it who is a patient. Check the vendor's BAA coverage before any of it leaves your systems.
Google Cloud's HIPAA page lists the products covered by its BAA. That page does not list Google Maps Platform or the Route Optimization API, and it says the customer is ultimately responsible for evaluating its own HIPAA compliance. If you use a mapping or routing service that a BAA does not cover, design for it:
- Geocode each address once, store the coordinates in your own database, and reuse them.
- Send the routing engine only coordinates, durations, time windows and random IDs that mean nothing outside your system. Never send names, diagnoses or visit reasons.
- Map the results back to patients inside your system.
- Remember that home coordinates can still identify a person. If that risk is not acceptable, run an open-source solver inside cloud services your BAA covers.
The HIPAA minimum necessary standard, 45 CFR 164.514(d), points the same way. It requires covered entities to limit routine disclosures to the amount reasonably necessary for their purpose, and to limit each workforce role's access to the categories of PHI it needs. A scheduling view for office staff who only book visits, for example, needs the address and time, not the chart.
What does each visit record need for billing?
CMS's evaluation and management booklet, MLN006764, sets out the home visit codes. Since January 1, 2023, home services and domiciliary, rest home or custodial care services form one code family, home or residence services, CPT codes 99341 to 99350. They apply to visits in the patient's home or residence, an assisted living facility, a group home not licensed as an intermediate care facility for people with intellectual disabilities, a custodial care facility, and a residential substance abuse treatment facility. Each setting has its own place of service code: 12 (home), 13 (assisted living), 14 (group home), 33 (custodial care) and 55 (residential substance abuse treatment).
- Store the facility type on every address, and derive the place of service code from it, so billing does not depend on someone remembering it per visit.
- Record the provider's time on qualifying activities separately from drive time. In its section on split or shared visits, the booklet lists traveling among activities that cannot be counted toward time.
- Support prolonged services. When the visit level is chosen by time, prolonged home or residence time is reported with HCPCS add-on G0318. The booklet's table sets the threshold at 140 minutes for 99345 and 110 minutes for 99350, counting time from 3 days before the visit to 7 days after.
- Allow the G2211 add-on, which the booklet says may be reported with the home or residence base codes from January 1, 2026.
Not every house call is a physician visit. Nursing and therapy visits under a home health agency follow other rules, so keep the visit type explicit and let billing logic branch on it.
Why does offline mode matter?
Providers work in basements, rural roads and buildings with thick walls. The clinic's scheduler uses a progressive web app on a phone or tablet instead of a native app, and it caches data so work goes on when the signal is poor. If you take this route, cache the day's route and the records for those patients, queue every change made offline, and resolve conflicts on reconnect with a clear rule, such as the provider's on-site status winning over an office change made earlier. Encrypt the cache and clear it at the end of the day.
Which reports does the office need?
The clinic's admin panel has 10 report types: Missed Patients, Completed Visits, Audio Notes, Provider Visit Summary, Referral Census, Referrals, Routes, New Patients, Documents and Weekly Route Overview. Missed visits and route summaries matter most for planning, because they feed the next period's plan. The doctors also add house-call notes in five formats, including text, audio and voice-to-text.
How does it connect to the EHR?
The scheduler needs the patient list, addresses and care plan frequency from the EHR, and the visit outcome and notes going back. In the clinic's system, patients come in from the EHR or are added by hand, and all patient data is encrypted on the server. We have shipped Epic and athenahealth integrations, and we work from the vendor's documentation and sandbox. In the systems we build, each clinician sees only their own patients, and we run a penetration test before launch.
Building or fixing a house call scheduler?
Talk to an engineer at Onix Health about your visit rules, your routing and your EHR. We sign a BAA before anyone on our team sees PHI. Our free five-day technical audit ends with a written report on where your current scheduling flow loses time and where patient data leaves your systems.