How to Schedule Caregivers and Track Visits for Home Health in 2026

Home health scheduling is the hardest scheduling problem in healthcare. The visits are not in your building. They are spread across a metro area, sometimes across county lines, in apartments without elevators, on streets with no parking, behind dogs the intake form did not warn the nurse about. The caregivers are independent enough to need autonomy and constrained enough that a fifteen-minute drive overrun cascades into a missed visit two hours later. Every miss has three failure modes stacked on top of each other: the patient does not get care, the agency may not get paid (Medicaid won't reimburse a visit with no Electronic Visit Verification record), and the surveyor has another data point if a Conditions of Participation review hits next quarter. Layer in caregiver retention — turnover in home health aide and CNA roles has stayed above thirty percent in many markets — and bad routing isn't just an operational headache, it's a workforce problem.

This guide is for agency owners, schedulers, and operations leads who want a procedural, no-fluff playbook for getting visit scheduling and tracking right. We'll walk through six steps: mapping patient visit patterns, clustering caregivers to geography, configuring EVV (federal mandate plus state Medicaid layers), automating daily visit routing, building real-time missed-visit alerting, and tracking the KPIs that actually move the needle. We'll close with the most common mistakes agencies make and how Deelo's Practice app handles each of these workflows in one place.

Before the steps, picture the goal state — because most agencies aren't aiming at the same target. A well-run home health schedule, on a good Tuesday, looks like this:

The caregiver opens the mobile app at the start of shift. Today's six visits are listed in geographic order. Drive time between each is pre-calculated against current traffic. The first patient is a Medicare PDGM episode, second visit of the certification period, RN visit, sixty minutes scheduled. The caregiver drives, parks, walks to the door, and clocks in via GPS — the location matches the patient's home address within a fifty-foot radius, EVV captures the timestamp, and the visit becomes billable. If GPS is unavailable (basement, dead zone, rural concrete), the caregiver clocks in via telephony from the patient's home phone instead, and the system records that fallback. The visit note is filed at the bedside on a tablet — vitals, wound photos, OASIS items if it's a recert, narrative — and signed before the caregiver gets back in the car.

Meanwhile, in the office: the dispatcher's screen shows every caregiver as a colored dot on a map. Green means on-site and clocked in. Yellow means en route. Red means a visit window is closing and the caregiver hasn't clocked in yet — which triggers a phone call before it becomes a missed visit. At end of day, every visit either has a complete EVV record (clock-in, clock-out, GPS or telephony) or is flagged for manual reconciliation with a documented reason. Payroll runs against actual clocked time plus mileage. Claims go out the next morning with EVV data attached. That's the target. The next six steps are how you get there.

You cannot schedule what you have not characterized. Before you tune the system, build a clean picture of what each patient on census actually needs.

For every active patient, capture six fields:

Visit frequency — once weekly, twice weekly, three times weekly, daily, or PRN (as needed). Medicare PDGM patients tend to cluster at one to three visits per week per discipline; Medicaid waiver patients may run daily aide visits. Frequency drives capacity planning more than any other variable.

Visit duration — RN assessments and OASIS visits often run 60-90 minutes. Routine skilled nursing visits run 45-60. Home health aide personal care visits run 60-120 depending on the care plan. PT and OT therapy visits run 45-60. Don't average — use the actual scheduled duration per discipline per patient, because a 30-minute miscalibration across 200 visits a week is 100 hours.

Discipline required — RN, LPN/LVN, PT, OT, SLP, MSW, HHA. Skill mix matters because not every caregiver can cover every visit. An aide cannot do a recertification OASIS. A PT cannot deliver wound care.

Preferred visit window — 8-10 AM, 10 AM-12 PM, 12-2 PM, 2-4 PM, 4-6 PM. Many patients have hard constraints (dialysis at 1 PM, family member only home before noon, sundowners patient who is calmer in the morning). Capture these and respect them.

Continuity preference — does this patient want the same caregiver every visit? For Alzheimer's, hospice, and complex wound cases, continuity is clinically meaningful. For routine medication management, it's less load-bearing.

Geographic identifier — full street address plus a coarser bucket (ZIP code, neighborhood, or service zone). The bucket is what you'll cluster on in Step 2; the address is what EVV will validate against in Step 3.

Most agencies discover, when they actually pull this data into a spreadsheet, that 15-25% of their patient records are stale (wrong frequency on file, outdated preferred window, address changed after a hospital discharge). Clean the data before you tune the schedule.

Random caregiver-to-patient assignment is the most expensive scheduling mistake in home health. A caregiver who drives 14 miles between visits twice a day is doing 28 miles of unbillable mileage at IRS-rate reimbursement, plus 60-80 minutes of unpaid windshield time, plus the cumulative fatigue that drives caregiver attrition. The fix is territory clustering.

Divide your service area into territories. The right number depends on density — for an urban agency with patients packed into a ten-mile radius, three to five territories may be enough. For a suburban or mixed agency, eight to twelve. Each territory should be defined by ZIP code, neighborhood, or driving-distance polygon, with the goal that any two patients within the same territory are no more than 15-20 minutes apart in normal traffic.

For each territory, assign a primary caregiver pool — the two to four caregivers who will see the majority of visits in that zone. Skill-mix the pool: at least one RN, one LPN, one HHA, plus PT/OT coverage. The pool is who the scheduler defaults to when assigning visits in that territory. Out-of-pool assignments still happen (someone calls out, a new admission lands in a sparse territory), but the default is clustering.

Three practical effects: drive time per visit drops 25-40% for most agencies that move from random assignment to clustered, mileage reimbursement becomes more predictable in your monthly P&L, and continuity of care improves naturally because the same small pool is rotating through the same patients. Caregivers also start learning their territory — the parking, the apartment building call boxes, the patients' family dynamics. That tacit knowledge is part of why clustered caregivers retain longer.

EVV is not optional. The 21st Century Cures Act federally mandates EVV for Medicaid personal care services and home health services, and CMS has enforced this through state Medicaid agencies. Every state now has an EVV requirement, though aggregator models differ (open model, closed model, MCO-specific). If your agency bills any Medicaid line, you need EVV running on every applicable visit, with six data points captured on each: type of service, individual receiving the service, date, location, individual providing the service, and time the service begins and ends.

Choose your capture mode. GPS-based check-in via the caregiver's mobile app is the default for most agencies in 2026. The caregiver opens the app at the patient's home, the app captures GPS coordinates, the system validates that the coordinates fall within an acceptable radius of the patient's address on file, and the timestamp is logged. Clock-out repeats the same validation. This is fast (under five seconds per check-in) and tamper-evident.

Configure telephony as fallback. Some patients live in basements, rural dead zones, or concrete buildings where GPS is unreliable. Some don't have a phone the caregiver can use, but most do. Telephony EVV uses the patient's landline or cell — the caregiver dials a toll-free number, enters their credentials, and the system captures the calling phone number as the location proof. Configure your platform so that if GPS fails three consecutive attempts at the geofence, the app prompts the caregiver to use telephony instead. Train caregivers on the fallback before they need it.

Plan for dead-zone exceptions. Some visits will fail both modes. Your platform should let the caregiver record the visit offline, sync when they return to coverage, and flag the record for supervisor reconciliation. The reconciliation should require a written reason ('GPS unavailable, no phone in home, patient confirmed visit') and a supervisor signature. Auditors look at exception rates — under 5% of total visits is healthy; over 10% will draw scrutiny.

Reconcile manual edits. Sometimes a caregiver forgets to clock in until they're in the car. Sometimes they clock out at the wrong moment. Edits to EVV records should be possible, but every edit should require a reason code, a supervisor approval, and a permanent audit trail. Edits without reason codes are the single most common compliance finding in EVV audits.

Manual scheduling — a person dragging visits around a calendar — works at fifteen patients. It breaks at fifty. By the time you're at a hundred and fifty active patients, manual scheduling burns one full-time scheduler per fifty caregivers and still produces routes that experienced caregivers know are sub-optimal.

Automated routing takes the patient pattern data from Step 1, the territory clustering from Step 2, and the caregiver availability for the day, and produces a draft schedule the scheduler reviews and approves. The good rule sets layer four constraints in priority order:

Constraint 1 — Skill mix. The visit needs an RN; only RNs are eligible. The visit needs a Spanish-speaking aide; only Spanish-speaking aides are eligible. Don't compromise on this; bad skill matches cause real clinical risk.

Constraint 2 — Continuity of care. If this patient has seen Caregiver A for the last six visits, Caregiver A is the strong default for visit seven. Override only when A is unavailable.

Constraint 3 — Drive-time minimization. Among eligible caregivers, prefer the one whose other scheduled visits create the shortest total drive time. This is where the territory pool from Step 2 pays off — the scheduler isn't picking from forty caregivers; they're picking from three or four in the right territory.

Constraint 4 — Caregiver preferences. Caregiver B doesn't drive after 5 PM; Caregiver C doesn't take stairs; Caregiver D requests no more than five visits a day. Encode these as soft constraints — the system tries to honor them but can override with scheduler approval if coverage demands it.

Run the auto-assign at the start of each scheduling cycle (most agencies run weekly or bi-weekly, with daily adjustments). The scheduler reviews the draft, hand-corrects exceptions, and publishes. Caregivers see the new schedule in the mobile app the moment it's published. Last-minute changes (call-outs, new admissions, hospital discharges) trigger a re-route of the affected day, not the whole week.

Missed visits in home health are not a minor inconvenience. Clinically, the patient missed a wound dressing change or a medication reconciliation. Operationally, the visit may not be billable. Compliance-wise, repeated missed visits without documented reasons can land in a CoP review. Financially, Medicare's PDGM episode pays based on the assumption that the plan of care is delivered — chronic missed visits put episodes at risk and can trigger Low Utilization Payment Adjustments (LUPAs) that crater margin.

The alert tree should be three tiers:

Tier 1 — Approaching window. Fifteen minutes before the scheduled visit window closes, if the caregiver hasn't clocked in, the system pings the caregiver via push notification. Most caregivers are already on-site or en route and just forgot to clock in; the ping resolves it.

Tier 2 — Window closing. Five minutes before the window closes with still no clock-in, the alert escalates to the scheduler or on-call dispatcher. They call the caregiver. Most often, the caregiver is stuck at the previous visit and needs the rest of the day re-routed.

Tier 3 — Visit missed. The window closed without a clock-in. The visit is now flagged as missed. The system routes a notification to the patient's clinical lead (the case manager or DON), who calls the patient, attempts a same-day rescue assignment if anyone is available, and documents the missed-visit reason in the chart.

For every missed visit, log a root cause: caregiver call-out, traffic delay, prior visit overrun, patient cancellation, patient not home, agency-side scheduling error, EVV system failure. After thirty days, run the missed-visit report and look at the root-cause distribution. If 'prior visit overrun' is your top reason, your visit durations in Step 1 are wrong. If 'caregiver call-out' is your top reason, you have a workforce problem to address upstream of scheduling.

Most home health dashboards measure too much and improve nothing. Pick five KPIs, watch them weekly, and act on them.

Missed visit rate. Total missed visits divided by total scheduled visits. Best-in-class agencies hold this under 2%. Anything above 5% is a problem worth a project plan.

On-time arrival rate. Visits that started within the agreed window divided by total visits. Target: 90%+ for windowed visits. This correlates strongly with patient satisfaction scores in HHCAHPS surveys.

Caregiver utilization. Billable visit hours divided by paid hours (including travel and admin). Healthy range is 65-78% — too low means caregivers are paid for non-productive time, too high means you're underpaying for travel and risking burnout. Bench against your own trend, not industry numbers.

Drive time per visit. Average windshield minutes between consecutive visits. The territory clustering in Step 2 should pull this down month over month. If it isn't, your territories are wrong or the auto-router is being overridden too often.

EVV exception rate. Visits with manual edits, missing GPS/telephony data, or supervisor reconciliations divided by total visits. Target: under 5%. Anything above 10% is an audit risk and worth a workflow review.

Run these five every Monday morning. Show them to your DON and your operations lead. The agencies that retain caregivers, hold margin, and pass audits are the ones that close the loop on these numbers, not the ones with the prettiest dashboard.

Deelo's Practice app is built around the workflow above, with the operations and clinical sides on the same data layer instead of stitched together across multiple vendors.

Caregiver dispatch with drive-time awareness. Patients are tagged with a service territory and a preferred caregiver pool. The auto-router builds the daily schedule against skill mix, continuity, drive time, and caregiver soft constraints, then surfaces the draft for the scheduler to review and publish.

Electronic Visit Verification. GPS-based clock-in and clock-out via the mobile app, with telephony fallback for dead-zone visits. Six required data points captured on every applicable visit. Manual edits require a reason code and a supervisor approval, and every change is in a permanent audit trail.

Point-of-care mobile documentation. Caregivers see their visit list, drive directions, patient context (allergies, recent vitals, plan of care, last visit note), and document the current visit on the device. Visit notes file at the bedside, not at the office desk three hours later. Offline mode buffers writes when coverage drops and syncs when it returns.

Real-time missed-visit alerting. The three-tier alert tree (approaching window, window closing, visit missed) runs against scheduled visits and EVV clock-in status. Notifications route to caregiver, dispatcher, or case manager based on tier.

Operational and clinical KPIs in one dashboard. Missed visit rate, on-time arrival, caregiver utilization, drive time per visit, EVV exception rate — plus the clinical and revenue KPIs from the same patient and visit data, so the DON and the operations lead are looking at the same source of truth.

Pricing. Plans run $19, $39, and $69 per seat per month, with the Practice app and the broader Deelo platform (CRM, scheduling, billing, helpdesk, automation) included. No per-visit fees, no per-chart fees.