CAHPS Hospice Survey: the 2025 changes and what moves the scores

The CAHPS Hospice Survey arrives in a mailbox about two months after a death. The person filling it out is a daughter, a husband, a sister — someone who spent the final weeks watching your team work, and who now gets 38 questions about what that felt like. Their answers become your publicly reported quality scores. Which means those scores are, mostly, a record of conversations: whether the nurse explained things plainly, whether anyone asked what mattered to the patient, whether someone called after the death. The survey went through its first major revision in a decade in 2025 — new measures, fewer questions, new mechanics. Here's what changed, where the scores go, and which of them your team can actually move.

The CAHPS Hospice Survey is a standardized CMS survey that asks the family caregiver of a hospice patient, after the patient's death, about the care the hospice provided. The results feed nine publicly reported quality measures, from communication and respect to help with pain and symptoms.

What is the CAHPS Hospice Survey?

CAHPS stands for Consumer Assessment of Healthcare Providers and Systems — a family of standardized experience surveys CMS fields across healthcare settings. The hospice version has one structural difference that changes everything about it: the patient is never the respondent. The survey goes to the primary caregiver listed in the hospice record, usually a family member, starting about two months after the patient's death.

Participation isn't optional. Medicare-certified hospices must contract an approved vendor and field the survey continuously as part of the Hospice Quality Reporting Program; only the smallest providers — fewer than 50 survey-eligible deaths in the prior year — can apply for an exemption. The results go public on Medicare's Care Compare, next to your competitors'.

If you also run hospital or clinic operations, this is the hospice counterpart of HCAHPS: same logic, standardized questions and public scores, but a different respondent in a different moment. An HCAHPS respondent rates their own stay. A CAHPS Hospice respondent rates how your team treated the person they just lost — and how your team treated them. Grief is part of the instrument.

The technical documentation lives at hospicecahpssurvey.org; everything below reflects the current Quality Assurance Guidelines, V12.0.

What changed in 2025

The FY2025 Hospice Final Rule, published in August 2024, put through the survey's first major overhaul since it launched. The revised instrument and the new administration mechanics both took effect for patients who died in April 2025 or later — so every survey your vendor fields now is the new one.

On the questionnaire itself:

• 38 questions, down from 47. The core survey is 31 questions, plus seven about the family member and the respondent. Shorter surveys are a deliberate response-rate play.
• Nine measures instead of eight. The new one is Care Preferences, built on two questions: whether the hospice respected the patient's wishes, and whether it listened to what mattered most to them.
• The family-training measure shrank to one question. What used to be a multi-item composite is now a single item asking whether the team taught the caregiver how to care for the patient.
• The confusing items are gone. Questions about nursing-home settings, moving the patient, and contradictory information were cut after CMS testing.

On how the survey reaches families:

• A prenotification letter now goes out before the first questionnaire, in every mode.
• A web option exists for the first time. The new Web Mail mode emails a survey link, then follows up by post with anyone who doesn't respond.
• The response window stretched from 42 to 49 days.

All three exist for the same reason: response rates in decline. CMS's survey page tracks the official materials and mode specifications if your vendor conversation needs settling.

Where the scores go

Scores publish on Care Compare for every hospice with at least 30 completed surveys, built on eight rolling quarters of data and refreshed quarterly. Star ratings sit on top: a star per measure plus a summary Family Caregiver Survey Rating, which requires 75 completes and updates every other quarter.

Two consequences hide in that arithmetic. First, scores move slowly by design — eight rolling quarters means one bad quarter dilutes across two years of data, and one good one does too. Second, the nine revised measures need eight quarters in the bank before they can publish at all, so the first public scores built entirely on the new survey land in 2027. The ratings families will compare you on then are being generated in visits your team is making this week. There is no cramming for this exam later.

The nine measures

Every measure is built from specific questions with fixed wording — your vendor can't rephrase them and neither can you. Six are composites averaging several questions, one is the single training item, and two are global: a 0-to-10 rating of the hospice, and whether the respondent would recommend it.

One scoring detail worth internalizing before you read any of them: public scores count only the best answer — the share of families who said "always" or "definitely yes." A family that answers "usually" is, for scoring purposes, a family you missed.

The explorer below walks through all nine — what the survey actually asks, and what tends to move each one.

Which measures your team's conversations move

Read the nine as an operator and they sort themselves into two piles: the ones decided by systems, and the ones decided by what your staff said out loud in a living room.

Communication with Family is the biggest composite — five questions, and every one of them describes a conversation. Keeping you informed about your family member's condition. Explaining things in a way that's easy to understand. Listening carefully. Telling you when the team would arrive. There is no clinical intervention hiding in that list; it's five kinds of talking, measured. The listening items in particular reward the therapeutic communication techniques most teams know about and few rehearse.

Care Preferences — the new measure — asks whether the patient's wishes were respected and whether anyone listened to what mattered most. That is a goals-of-care conversation, scored. Not the checkbox version buried in the admission packet: the out-loud version, held early, revisited when things change, and hardest when the family wants everything done. It's the most coachable measure on the survey, and from April 2025 it counts.

Emotional and Spiritual Support includes something most training plans skip entirely: support in the weeks after the death. The bereavement call is the last conversation a family has with your hospice and the freshest memory when the survey arrives — and it's usually delegated to whoever is available, unrehearsed.

Training Family to Care for Patient is now a single question, which concentrates its stakes: one item asking whether your team taught the caregiver what to do. That's a teach-back conversation with a frightened non-clinician, often about medication they're terrified of getting wrong.

Underneath all four sits the conversation the survey never asks about directly: the first one, where a family that isn't ready hears the word hospice. Handle it badly and every later question gets answered through its shadow. Frameworks like the SPIKES protocol exist for exactly this kind of news, but frameworks don't survive contact with a resistant family unless they've been practiced against one.

Two measures mostly don't belong to conversations, and pretending otherwise would waste your training budget.

Getting Timely Help is staffing, triage, and on-call coverage — a family asking for help on a Saturday night is scoring your rota, not your empathy.

Help for Pain and Symptoms is clinical practice: assessment, titration, protocols.

A conversation program won't fix either. What it changes is the margin around them: when the nurse does arrive at 2 a.m., whether the family remembers being listened to; when the pain plan is working, whether anyone explained it plainly enough for the family to trust it.

The practical move is the same one that works for any healthcare roleplay program: pick the three conversations above that scare your team most, write them as scenarios with the resistance real families bring, and give people repetitions before the next real one. If hospice is one of your service lines, we built a dedicated setup for it.

Response rates are a score you control

Before a single answer is recorded, you're already deciding how many surveys come back — and volume is not cosmetic. Thirty completes gets your scores published; seventy-five gets you a summary star; and every response above that makes your scores harder for one angry outlier to drag.

The 2025 mechanics are response-rate machinery, and they only work with good inputs. The web mode needs an email address; the prenotification letter needs a mailing address that's current. Both get captured at admission, months before the survey exists, by whoever fills in the caregiver record. That's the lever: treat caregiver contact information as quality data, not paperwork.

The boundary matters too. You can tell families the survey exists, that it's legitimate, and that it matters to you. You cannot coach answers, ask for top scores, or survey them yourself first — CAHPS rules draw that line firmly, and vendors are required to report attempts to lean on it.