Family Caregivers Call Before Patients Churn

Family Caregivers Call Before Patients Churn

The daughter calls on a Thursday. She is not complaining. She asks whether the aide is coming at the usual time, mentions that her mother was "a bit off" after the last visit, and says she has been looking at a different agency "just to understand the options." The call lasts four minutes. It is logged as a scheduling inquiry.

Eleven days later the patient transfers. Nothing in your system predicted it, because the thing that predicted it was that phone call, and the phone call was not data.

This is the most under-instrumented signal in home health operations, and the reason is structural: your two main feedback instruments — the patient experience survey and the clinical record — are both built in ways that systematically miss family caregivers.

What HHCAHPS Tells You, and When

Start with what the national data actually says. Per the current CMS HHCAHPS national file:

Measure	National result
Team gave care in a professional way	89%
Team communicated well with patients	86%
Rated the agency 9 or 10	85%
Team discussed medicines, pain, and home safety	82%
Would definitely recommend the agency	79%
Completed surveys	1,007,318
Response rate	22%

Two of these deserve an operator's attention.

79 percent would definitely recommend. That is the lowest measure on the list, and it is the one closest to the commercial reality of referral relationships and family sentiment. Roughly one patient in five is not a definite advocate. Nationally.

82 percent said the team discussed medicines, pain, and home safety. So in close to one in five cases, the patient does not recall the conversation that is supposed to be the core of the visit happening at all. Whether it did not happen or did not land is, for retention purposes, almost beside the point.

The trouble is timing and sampling. A 22 percent response rate means roughly four of five sampled patients said nothing. And nonresponse here is not random.

Who HHCAHPS Is Designed Not to Ask

This is the part most operators have never read, and it changes how much weight the survey deserves as a retention instrument.

The HHCAHPS Protocols and Guidelines Manual sets eligibility criteria. To be surveyed, a patient must be at least 18, have had care paid for by Medicare or Medicaid, have had at least one skilled visit (nursing, PT, OT, or speech) in the sample month, and have had at least two such visits during the lookback period. Patients who are deceased are not eligible. Neither are patients currently receiving hospice care. Agencies must also exclude patients who "requested that the HHA not release their name and contact information to anyone other than agency personnel."

Line those exclusions up against the population you would most want to hear from:

• The patient who left after one visit. Excluded — the two-visit lookback threshold removes exactly the early-churn case.
• The patient who died. Excluded.
• The patient who transitioned to hospice. Excluded.
• The patient who asked you not to share their contact information. Excluded — and that request is itself often a signal.
• Private-pay and commercially-insured patients. Outside the Medicare/Medicaid criterion entirely.

None of that is a flaw in HHCAHPS. The survey is built to measure the experience of patients who received a course of care, and it does that competently. It is simply not an early-warning system, and it was never meant to be one. Treating your star rating as a churn metric means reading a lagging, filtered, retrospective measure of the patients who stayed.

The patients who left are not in the denominator. Their families, however, called you first.

Caregivers Are Already in the Conditions of Participation

There is a regulatory point here that gets missed. The CoPs do not treat the family caregiver as a bystander.

Under 42 CFR 484.60(d), the agency must "coordinate care delivery to meet the patient's needs, and involve the patient, representative (if any), and caregiver(s), as appropriate, in the coordination of care activities," and must "ensure that each patient, and his or her caregiver(s) where applicable, receive ongoing education and training provided by the HHA, as appropriate, regarding the care and services identified in the plan of care." Paragraph (c) requires the agency to "promptly alert" the practitioner to changes in the patient's condition or needs — and a caregiver on the phone is frequently how the agency learns of one.

Separately, 42 CFR 484.50(e) requires the agency to investigate complaints made by the patient, the representative, "and the patient's caregivers and family," and to "document both the existence of the complaint and the resolution of the complaint." Caregivers are named in the text.

So the obligation to listen to the daughter already exists. What usually does not exist is any record that you did.

This post is general information about regulatory requirements, not legal advice. State licensure rules and payer contracts may add requirements; confirm your own.

What the Pre-Churn Call Sounds Like

Across agencies the pattern is consistent enough to describe. These are illustrative composites, not findings from any customer:

• Logistics questions that are really reliability questions. "Is she coming Tuesday?" asked three times in two weeks is not a scheduling issue. It is a trust issue, and schedule volatility is the cause.
• The named-aide question. "Will it be Marcus again?" Caregiver continuity is the relationship. Once the family starts asking who is coming, the relationship has moved to the staffing roster.
• Comparison language. "My neighbor uses—" or "I was reading about—". Unprompted mention of an alternative is close to an explicit exit signal, and it almost never gets escalated.
• Clinical worry routed to a non-clinical person. A caregiver describing a symptom to a scheduler is both a safety event and a retention event. Whether it reaches a clinician is the whole question.
• Repeat calls about the same unresolved thing. Third call about the same supply order is not three inquiries. It is one failure with a growing cost.
• The call that goes quiet. A family that called weekly and then stopped has often already decided.

Every one of these is detectable in the call itself and invisible in the disposition code. "Scheduling inquiry" is true and useless.

Do This Week

1. Pull your last 20 transfers and discharges-against-advice. For each, list every inbound call from that household in the 30 days before. You are looking for one number: how many had three or more caregiver calls. Most agencies find the majority did.
2. Count caregiver calls as a category. If your system cannot distinguish "patient called" from "family member called," that is the first fix, and it is usually a one-field change at intake.
3. Flag two phrases. Comparison language and the named-aide question. You do not need a model to start — a human reviewer with two criteria and a week of calls will produce a list worth acting on.
4. Route clinical content out of scheduling. Write down what a scheduler is supposed to do when a caregiver describes a symptom, then ask two schedulers what they currently do. The delta is your escalation gap.
5. Call back the repeat callers. Pull anyone who called three or more times in the last fortnight and have a supervisor call them proactively. This is the highest-yield hour on this list and it costs nothing.
6. Stop using your star rating as a retention metric. Keep improving it; it matters for referrals. But put caregiver call volume and repeat-call rate on the same dashboard, because those move first.

The economics of why this is worth the effort are in the hidden cost of patient churn in home health. The survey-side analysis — why the communication composite sits above the recommend score, and what that implies — is in this piece on the HHCAHPS communication gap. And the single highest-leverage window for all of it is covered in the 72 hours after discharge.

The Point

You are measuring retention with an instrument that excludes the patients who left, surveys a fifth of the rest, and arrives months late. Meanwhile the families of the patients about to leave are calling you, for free, in advance, and telling you — in ordinary language, buried in a question about Tuesday's schedule — that they are shopping.

SurfacerIQ works in the category of conversation intelligence for healthcare contact centers, which here means making caregiver calls countable and searchable so patterns across a household surface before the transfer request does. The underlying move is simply to treat the family's calls as a dataset rather than as interruptions.

If you want to see what your pre-churn call pattern looks like, talk to us.