What Normal Looks Like: A Family-Authored Record for Post-Acute and Home Care

The relevant exclusion is statutory rather than accidental. The HITECH Act of 2009 funded electronic record adoption through incentive payments to eligible professionals and eligible hospitals, and the Office of the Assistant Secretary for Planning and Evaluation records that it “did not include financial incentives for LTPAC providers” ^[1]. Skilled nursing facilities, home health agencies, rehabilitation and long-term acute care hospitals, and assisted living communities were therefore left to fund the transition themselves.

The consequence is measurable. Among hospitals ineligible for the incentive program, 6% of long-term acute care hospitals, 4% of rehabilitation hospitals, and 2% of psychiatric hospitals had at least a basic electronic record, against 12% of the short-term acute care hospitals that were eligible ^[2]. Those figures are from 2012 and adoption has since risen everywhere. What has not changed is the second-order effect, which is the subject of this paper: the settings that were excluded never acquired a patient-facing portal, so the family of a person receiving care in them has no read access to the record of that care.

The affected population is large and is growing quickly. The National Alliance for Caregiving and AARP report 63 million family caregivers in the United States, approximately one in four adults, up from roughly 43 million a decade earlier, an increase of about 45% in ten years ^[3].

It would be convenient if the missing artifact were a record that nobody keeps. It is not. The aide keeps a visit note, the agency keeps a plan of care, and the facility keeps an assessment. Each is a record of the same person, none is assembled with the others, and each one ends when that party’s involvement does. Only the family has a continuous view of the person, but that view only lives in a group text or email chain.

The design of this work rests on primary research rather than on inference from the literature. Between September 4th and September 16th, 2026, we conducted twenty interviews: sixteen with family members coordinating the care of a parent or grandparent, and four on the paid side, including a home care aide and the executive directors of two assisted living communities. Fifteen of the family interviews concerned care that is currently being delivered or was delivered within the past four years.

This is a convenience sample and should be read as one. Respondents were reached through personal and referred networks, several were known to the author beforehand, and the count is small. The figures below are reported as counts of what was said rather than as rates in any population, and they are stated here because the two patterns they describe are strong enough to have changed what we are building.

The first pattern concerns what people reject. Every respondent who was asked about a system that would prompt a family member to do something rejected it, four out of four at the point the finding was recorded. The reasons differed and were specific in each case. One family had already run a shared calendar of assigned tasks, and it had ended through attrition rather than disagreement. One respondent observed that the aides in question might not be able to use an application at all. Another said there was nothing in his day for which he required a reminder, because the mail arrived at his house and the appointments were already on his work calendar.

The second pattern concerns what people ask for without being asked. Five respondents independently described a system that would observe and report, before any such system had been described to them, and eight of eight said they wanted one when it was described. What they described was passive: a short account of how the person was that day, assembled out of what the people already in the house were already doing, and sent to whoever in the family runs things.

These two patterns are one finding rather than two. The prompting half of the idea and the observation half have opposite evidence, and they differ in who is asked to act. Prompting asks an unpaid family member to do something in response to a message. Observation asks nobody to do anything and reports what already happened. Demand for the second appears to track the number of paid parties in the household rather than the distance between the family and the person, which is the opposite of what we assumed at the outset.

One further observation bears on delivery rather than on demand. A nurse who worked nights in a skilled nursing facility reported that the resident’s baseline is recorded in the care plan at intake and is not read on the night shift, because nobody is directed to read it; the route by which it actually travels is the oral report at the change of shift. An assisted living director described the same constraint from the other side: a written family summary is useful, but it reaches the aide covering a floor only if it has been compiled into that aide’s task list.

The artifact is a record of the baseline, authored by the family. Its content is not novel, and two existing forms have carried most of it for over a decade. The Alzheimer’s Society’s “This is me,” endorsed by the Royal College of Nursing since 2010, is completed by the people who know the person best, states explicitly that it is not a medical document, and is intended to be read by care staff in an unfamiliar setting ^[4]. Section F of the federal nursing home resident assessment asks every resident about bedtime, bathing, snacks, and family involvement in discussions of care, and directs staff to interview the family where the resident cannot answer ^[5]. The field list should therefore begin from these rather than from nothing.

What those forms lack is everything around the fields. Each carries one author for the whole document rather than a source per entry; neither has a change history; neither provides a way for several family members to keep it current; and neither has any route to the next setting except being carried there. Those are the properties we are building, and they are the reason this is a record rather than a form.

Three commitments follow.

First, every entry carries its source and its time. A household’s knowledge is contested more often than it is missing: one family member reports that the person is eating and another reports that she is not, and the value of the record depends on its being able to hold both, attributed, rather than on its resolving them. We currently model this on the provenance and relationship shapes already defined in HL7 FHIR, which makes the later expansion into clinical data a mapping rather than a rewrite.

Secondly, the record is built for delivery rather than for storage. The evidence in Section 2 indicates that a document which must be read in order to work will not be read, because the worker who most needs it is the one covering an unfamiliar floor on an unfamiliar night. The design requirement that follows is that the record compile into the form the receiving worker already acts on, which is a task list in a facility and a shift note in home care, rather than into a page that someone must open.

Thirdly, the record is built to leave. The federal element nearest to this content is Care Experience Preference, added in version 4 of the United States Core Data for Interoperability ^[6]. It is one field, defined to hold a person’s goals and priorities for the experience of their care, so a record of the kind described here would have to be flattened into it rather than carried by it. The envelope in which a record of this kind would travel between post-acute settings was published as a standard for trial use in May 2026 ^[7]. Neither carries a family-authored record today. We currently treat them as the export path rather than as the model.

This is not a medical device. It does not diagnose, it does not detect, grade, or characterize pathology, and it does not recommend or modify treatment.

It does not prompt or assign tasks to members of a family. That is a deliberate exclusion rather than an omission, and Section 2 states the evidence for it.

It does not monitor the person. The record is assembled from what the paid parties in the household already produce and from what the family chooses to write down, and it adds no sensor, camera, or wearable to the home.

It does not replace the clinical record, the plan of care, or the facility’s assessment, and it is not a substitute for the judgment of the clinicians and caregivers who use it.

One limitation deserves to be stated rather than footnoted. A written record of a person’s daily care is discoverable, and a family that keeps one may find it read back to them in a dispute they did not anticipate. We regard this as a design constraint on what the record retains and on who can see it, and we currently do not consider it solved.

There is no product. Nothing described here is available, and nothing is for sale.

The work is at the stage of customer discovery, and it is being run as a hypothesis with a decision date. Six criteria were written down before the first interview, and the evidence will be scored against them on October 2nd, 2026 rather than against how the conversations felt. Four households, three of them close to the author, have offered to begin paying a subscription in the middle of October. None has been asked for money, no money has been taken, and we do not currently treat any of them as revenue.

Teleos Labs is participating in MIT’s AI Studio (MAS.665) this autumn, which concludes with a demonstration day on December 3rd, 2026.

We are currently looking for two kinds of conversation: with families coordinating care for an aging parent, and with the operators and staff of assisted living communities, home care agencies, and skilled nursing facilities. The form below reaches the author directly.

If you are coordinating a parent’s care, or you run or work in a community, agency, or facility that delivers it, we would like twenty minutes of your time. Interviews are confidential, and nothing you say is shared with anyone else in your family.