The Work Beneath the Conversation: Articulation Work

"You're so good with families." It's sincere, it's usually accurate, and it names the surface part of the job that's easy to see.

I argued last week that our own trials keep locating the effect of specialist palliative care somewhere other than the family meeting. If the work happens in the reconciling and the phone calls and the months of contact, why does everyone describe it as a discreet conversation — including us?

The answer turns out to be sitting next door in a discipline few of us trained in. Medicine draws almost nothing from sociology, anthropology, or psychology, which is a strange gap for a profession whose entire object is people under duress. So we hit phenomena those fields mapped forty years ago and build our own vocabulary from scratch, having no idea the map exists. Palliative care gets this worse than most specialties, because our substrate is social organization and meaning-making, and nothing in our training equips us to name either one.

A Word for It

Anselm Strauss spent years inside hospitals watching what actually got patients through treatment, and named a category of labor nobody had accounted for. Any project — an illness, an admission, a discharge — is a mass of tasks distributed across people, and those tasks do not line up automatically. Somebody has to make them line up. Strauss called that articulation work.

Somebody has to notice that two services changed the same medication. Somebody has to get the consultant and the primary team to say the same number out loud. Somebody has to find out that the plan depends on a home health visit that was never going to happen.

That is most of what a palliative care team does, and it has forty years of scholarship behind it explaining why nobody sees it.

Nothing in the System Is Built to Show It

Susan Leigh Star and Strauss later worked out the mechanism. No work is inherently visible or invisible. We see work through a selection of indicators — a strained muscle, a finished object, a completed form — and whoever controls the indicators controls what gets to count as work at all.

Articulation work fails every indicator a hospital collects. Star's formulation is that it's invisible to rationalized models of work, which is a claim about what those models are able to represent. They represent encounters, orders, and documented findings. Articulation work produces an absence of trouble.

Two things follow. Routine buries it further: work expected of a discipline recedes into the background and goes functionally unseen even when it happens in plain sight. And when productivity gets quantified through indirect indicators, the legitimacy of the work passes to whoever handles the indicators — people who have never watched the work happen.

So the well-run family meeting looks easy from outside, and easy is the only thing the record holds. The compliment describes the only surviving evidence.

What Gets Priced

Palliative medicine has no discreet procedure. We bill: evaluation and management, time, prolonged services, a handful of advance care planning codes. Every one of them "pays" for an encounter.

Articulation work happens between encounters. The call to the outpatient cardiologist is not a visit. Catching that two services adjusted the same diuretic is not a visit. Learning that a patient's coverage lapsed in March is not a visit. Identifying anticipatory grief is not a visit. The work that makes the meeting possible produces no claim (although it might produce enough to justify the occasional complex care management code in the ambulatory setting), so the only consistently reimbursable artifact of a week's labor is the forty minutes at the end.

This is a measurement inheritance rather than a plot. The apparatus was built to price what's observable inside a room, and it does that reasonably well. Articulation work is defined by happening between rooms. A system that can only see encounters will price a specialty whose product lives between them at close to zero, and it will do so without anyone deciding to.

One Tuesday

A patient with advanced heart failure, third admission in ten weeks. The consult question is goals of care.

Over four days: the nurse practitioner finds that two services adjusted the diuretic independently and neither reconciled it, which explains the readmission better than anything in the discharge summary. The physician spends forty minutes across three phone calls getting the outpatient cardiologist and the inpatient team to agree on a prognostic range, because they'd been giving the family different ones for a month. The social worker learns the patient's son has been sleeping in his car in the visitor lot and hasn't told anyone, which is why he's exhausted and irritable. The chaplain identifies the patient doesn't know who they are meant to be in the world anymore.

Then the family meeting happens. It takes forty minutes. It goes well.

In the chart, in the billing, and in the program's quarterly report, that week is one consult and one family meeting. Everything that made the meeting possible left no trace. And the compliment afterward — you're so good with families — describes the only thing that survived.

Who Ends Up Doing It

Paula England, Michelle Budig and Nancy Folbre measured the care penalty: net of education, experience, cognitive and physical skill demands, authority, union status, industry, and the share of women in the job, working in an occupation that develops other people's capabilities costs about five to six percent in wages. The same person earns less while in care work and more after leaving it.

The exception in their data is interesting: registered nurses and other licensed non-physician clinicians show a premium rather than a penalty, and the authors offer two explanations: a large rise in demand for nursing, and licensure — an occupational closure that makes the work legally somebody's, which raises what it pays. Chaplaincy has no such closure. Neither does the coordination itself.

A later analysis by Asaf Levanon with England and Paul Allison settled the direction across fifty years of census data. When an occupation's share of women rises, its relative pay falls afterward. The reverse pathway — women moving into jobs whose pay had already dropped — turns up in the 1950s and then mostly stops appearing. The devaluation follows the women in.

Look at who does the articulation work in the health system. Nursing, social work and chaplaincy are overwhelmingly women's professions, and so is hospice and palliative medicine, more than medicine as a whole. The work most exposed to this pattern is also the work least protected by any formal claim on it. This is a point I've touched on before.

I want to be precise about the inference. These findings concern occupations in a labor market, not subspecialties inside one. Whether the mechanism reaches a medical subspecialty the way it reaches an occupation is a reasonable extrapolation and not a demonstrated result. What's demonstrated is that the pattern is real, that it runs in the direction people usually assume backward, and that our team composition sits squarely in its path.

Where the Fragmentation Is Worst

Articulation work scales with fragmentation.

A patient with continuous coverage, one health system, an established primary care relationship, adaptive coping, and a family fluent in institutional English generates modest articulation demand. A patient whose Medicaid lapsed, whose specialists sit in three unconnected systems, who needs an interpreter for every encounter, and whose county has one home health agency generates enormous demand. The articulation work of making the second patient's plan real can run to days.

None of it produces a billable event. So the programs doing the most articulation work serve the patients whose care generates the least, and when those programs get evaluated on encounters per clinician they look like the least productive services in the system. A safety-net practice and a well-resourced academic one can do identical clinical work and appear, in the only numbers anyone reports, to be doing radically different amounts of it.

Star and Strauss state the equity version outright: when a system ignores how visible and invisible work relate to each other, the cost gets pushed onto whoever has the least standing to refuse it. The devaluation here is regressive, and it arrives through a metric nobody designed to do that.

Division of Labor, Division of Rights

Strauss drew one more distinction in that same paper, and it explains why competence on the unit rarely converts into standing anywhere else. The division of labor describes who does which tasks. The division of rights describes what people can claim, impose, negotiate over and enforce — and rights hold, he noted, only where the resources exist to enforce them: legal, financial, organizational.

Palliative care has an excellent division of labor and almost no division of rights. On most units everyone knows what we do and many defer to it. In the apparatus that decides what counts as a distinct kind of expertise, articulation work isn't a recognized object at all.

The gap produces three predictable failures.

Trivialization. The work becomes a personality trait. "So good with families" converts expertise into temperament, and temperament can't be taught, hired for, or funded. A department can budget for a skill. Nobody budgets for a disposition.

Instrumentalization. The work gets valued for what it saves — length of stay, ICU days, cost avoidance. A service justified by savings is secure exactly as long as the savings hold up, and it gets renegotiated by whoever runs the analysis next.

Displacement. The function gets unbundled and reassigned. Someone decides the conversation is a discrete task, writes it into a protocol, and gives it to whoever is available.

We already know what displacement produces, because it's been tested. Carson's trial gave specialists the meeting without the continuity, without symptom management, and without most of the team. It produced nothing. That's what an unbundled function looks like when it runs — the conversation without the articulation work underneath it, delivered by the right people, to no effect.

The Bind We Make

We describe the conversation (our "procedure"), which is the visible tenth of the work. Then we describe the savings (that sweet, sweet ROI), which is a byproduct we don't control. Both things are true. Both get programs funded. Neither says what happens on a Tuesday.

That's a reasonable thing to do with the vocabulary available to us. It's also a trap. A field that spends twenty-five years describing itself as expert at difficult conversations will find that medicine has learned the description, believed it, and priced accordingly. We were selling the parts that were legible, because those were the parts we had words for even if they are a fraction of what we do each day.

Anticipating Your Objections

Every specialty does coordination work. True, and by far the strongest point here. Cardiology reconciles medications, calls consultants, and fixes broken discharges. The difference is what else is in the bag. Cardiology also bills an ablation, and the ablation makes the specialty visible whether or not anyone notices the coordination. For us, articulation work is the whole product. A pricing system blind to it is blind to everything.

You're tacking sociology on where it doesn't belong. Fair, and the test is whether the framing generates anything. It generates one concrete practice — log the between-encounter work as its own category and report it next to consult volume, so that three calls to a cardiologist exist somewhere in the institution's account of itself. It generates a prediction about which programs will look worst on productivity metrics. And it generates a testable claim about where articulation demand concentrates. If none of that survives contact with data, the framing was decoration.

Nursing ran the first part of that experiment already. Through the 1990s, nurses built a classification of nursing interventions specifically to pull their work out of the background and into the record, in an era when nursing notes were routinely discarded after discharge. One nurse's summary of the problem was I am not a bed. Star and Strauss watched that campaign and named its cost: every gain in granularity is also a gain in surveillance, and specification invites Taylorism. There's a difference between recording that a clinician did the counseling and scripting what they should have said. Build the category at the level of the task and leave the discretion alone.

Naming the work won't change the price. Probably right in the short term. The places where an enforceable claim actually gets made — payment definitions, quality specifications, accreditation language — move over years, which is cold comfort to a program facing a budget review in March. Nothing gets contested that can't first be named, though naming only opens the argument.

Where I Might Be Wrong

Strauss developed this watching a health system that no longer exists — before managed care, before the electronic record, before hospital medicine. Importing the concept wholesale may bring assumptions that stopped holding decades ago.

The care-penalty literature contains a real exception. Licensed clinical occupations came out ahead in England's data, not behind, which cuts against the simplest version of this argument. It's why I've kept the claim narrow — about enforceable rights rather than about women's work being discounted across the board.

The devaluation argument extrapolates from occupational labor markets to a medical subspecialty, and I haven't shown the mechanism transfers.

I've described how the pricing works without arguing about how prices get set. That's a real limit here, and a reader who wants to press on it should.

Making invisible work visible has costs that the invisible-work literature is more honest about than I've been. Documentation invites surveillance, specification erodes discretion, and both land hardest on the disciplines with the least protection — the same disciplines doing most of the work.

And the objection I find hardest sets up where this goes next. Maybe the work really is worth less. Maybe a system that prices articulation work at nothing has correctly identified that most of it compensates for failures that shouldn't exist, and paying for it would entrench the fragmentation it exists to survive. If that's right, every hour we spend making a broken arrangement work is an hour of pressure removed from fixing it. I think that objection is partly correct, and it's the subject of the next post.

Final Thoughts

"Articulation work" explains the compliment, the billing, the productivity metric, and why the function keeps getting handed to whoever's free.

Sociology has been studying how institutions distribute invisible labor. Anthropology has been studying how people make meaning when the world stops making sense. Psychology has been studying how anyone tolerates knowing what's coming. That's the substrate of palliative care, and we're trained in little of it — which is why we keep building vocabulary for things that already have names.

I am a palliative care physician, educator, and professional strategery expert. Known for turning rounds into rants and rants into teaching points. Rounds & Rants represents my views — not those of any institution or professional membership organization where I hold a role. I don't write on their behalf and they don't vet what I publish.