Hospice Visits in the Last Days of Life: 48.3%

Hospice Visits in the Last Days of Life: 48.3%

Two numbers from the CMS Hospice — National Data file, both covering January 2023 through December 2024, should be read next to each other.

• Visits near death: 89.3% of decedents received at least one visit from a skilled nurse or social worker in the last three days of life.
• Hospice Visits in the Last Days of Life (HVLDL): 48.3% received such visits on at least two of the final three days.

Somebody showed up nine times out of ten. Somebody stayed present fewer than five times out of ten.

That gap is the entire subject of this post, and it is not primarily a clinical staffing question. It is a question of whether your hospice knew, in time, that a patient was in their final days.

What HVLDL actually measures

The CMS HVLDL measure specifications define the numerator as patient stays “in which the patient and/or caregiver received in person visits from registered nurses or medical social workers on at least two of the final three days of the patient’s life.” The denominator is “all Medicare hospice decedents discharged to death within the time period of the data.”

Precision matters here, because HVLDL is narrower than people assume:

• Only RN and medical social worker visits count. Aide visits, chaplain visits and LPN visits do not.
• Visits must be in person.
• Patients receiving continuous home care, respite or general inpatient care during the final three days are excluded, as are patients enrolled fewer than three days.
• Visits occurring after the time of death do not count.

That last exclusion is worth sitting with. A team that arrives two hours late has, for measurement purposes, not arrived.

The Hospice Care Index says the same thing from a different angle

The Hospice Care Index bundles ten claims-based indicators. The national values in the same CMS file tell a coherent story about continuity.

HCI indicator	National value	What it counts
Gaps in nursing visits	52.4% of elections	At least one gap between nursing visits exceeding seven days, among stays of 30 days or more
Skilled nursing minutes on weekends	9.6% of minutes	Share of total skilled nursing minutes on routine home care days that fall on Saturday or Sunday
Late live discharges	37.4% of live discharges	Live discharges occurring on or after 180 days of election
Early live discharges	7.3% of live discharges	Live discharges early in the election period
Visits near death	89.3% of decedents	At least one RN or social worker visit in the final three days
HCI overall score	8.8 of 10	Composite across all ten indicators

Weekends are 28.6% of the calendar. Weekend skilled nursing accounts for 9.6% of skilled nursing minutes on routine home care days. That arithmetic is not a scandal — routine home care staffing is legitimately weighted toward weekdays — but it does mean that a patient who enters their final three days on a Friday is entering them against the thinnest part of the week.

Meanwhile, more than half of elections lasting 30 days or more contain at least one stretch of more than seven days without a nursing visit. During a gap like that, your information about the patient comes from exactly one place: the phone.

The regulation requires presence, not recording

42 CFR 418.100(c) requires that “nursing services, physician services, and drugs and biologicals… must be made routinely available on a 24-hour basis 7 days a week,” with other covered services available on a 24-hour basis when reasonable and necessary.

42 CFR 418.56 requires the interdisciplinary group, in collaboration with the attending physician, to “review, revise and document the individualized plan as frequently as the patient’s condition requires, but no less frequently than every 15 calendar days,” incorporating the updated comprehensive assessment and noting progress toward goals.

Neither provision requires you to record a phone call. What 418.56 does require is that changes in condition make their way into the plan of care. A daughter telling your on-call nurse at 11 p.m. that her mother stopped eating two days ago is a change in condition. Whether it reaches the plan of care depends entirely on what happens to that call.

This post is general information about federal hospice requirements, not legal advice. State licensure rules and Medicaid hospice programs add requirements that vary considerably; review your own obligations with counsel.

Where the signal arrives, and where it dies

Prognostication in the final days is imprecise, and no measure can fix that. But in most hospices the earliest indication that a patient has turned is not a clinical assessment. It is a family member describing something on the phone, often after hours, often to whoever is covering.

Consider a hypothetical: a patient at day 140 of an election, stable, on a routine visit schedule. Saturday evening the son calls the answering line. He says his father has not taken fluids since Thursday and his breathing has changed. The on-call nurse gives sound advice about positioning and comfort, and logs the call as a symptom-management call. No triage. Monday the case manager reads the log, calls, and schedules a visit for Tuesday. The patient dies Monday night.

There was no clinical failure in that sequence. Every person did their job. The call contained the information that would have changed the schedule, and the system had no mechanism for the call to change the schedule.

That is what a 48.3% national rate looks like from the inside. Not neglect. A signal that arrived and did not route.

Do this week

Pick a bounded review. You can do this without new software.

• Pull the last 15 decedents who did not meet HVLDL — those with fewer than two RN or MSW visits in the final three days — and list every inbound call on those cases in the preceding seven days.
• Read the call logs for condition-change language. You are looking for the family vocabulary that precedes active dying: stopped eating, stopped drinking, sleeping all day, breathing changed, not waking up, confused, will not take pills. Mark which calls contained one of these and did not produce a visit within 24 hours.
• Check the day of the week. Sort your HVLDL misses by the day the final three days began. If Friday and Saturday are overrepresented, your problem is coverage design, not clinical judgment.
• Audit one specific handoff: after-hours call to next-business-day case manager. Time-stamp when the call came in and when the case manager acted. That interval is your real exposure.
• Write down the escalation trigger explicitly. Most hospices have never stated, in writing, which caller-reported symptoms require a same-day visit rather than telephone advice. If your on-call staff are improvising that judgment at 11 p.m., they will improvise inconsistently.

Do not treat this as a chart audit. The charts will look fine. The information you need is in the calls.

The measurement caveat, stated honestly

HVLDL is claims-based and its denominator is restricted. Hospices with a high share of very short stays, or heavy use of general inpatient and continuous home care in the final days, will see a smaller denominator and a score that reflects fewer patients than their census suggests. A low score deserves investigation, not an assumption. Comparing your rate to 48.3% is a starting question, not a verdict.

The broader trend is worth tracking too. The CMS Hospice Monitoring Report shows the overall live discharge rate rising from 16.9% in FY2021 to 19.1% in FY2025. Combined with a 37.4% late live discharge indicator, that suggests continuity over long elections is under real pressure across the industry — not only at the end.

What to change

The structural fix is to stop treating after-hours calls as a message-taking function and start treating them as clinical intake. That means a defined symptom vocabulary, a written escalation trigger, and a review loop that runs before the next visit is scheduled rather than after the patient dies. We laid out the general version of this shift in from reactive to proactive compliance monitoring, and the taxonomy work it depends on in building a custom call taxonomy. The same pattern shows up in home care as what a missed visit sounds like before it appears in EVV data.

SurfacerIQ belongs to the category of tools that read every call rather than a sample, and surface condition-change language and unclosed commitments as they occur. The point is not more data. The point is that the Saturday-evening call reaches the person who builds Monday’s schedule.

If your HVLDL rate is below 48.3% and your charts do not explain why, the explanation is on your phone lines. Talk to us.