When the Record Doesn't Follow the Patient: Documentation Lessons from the Clancy Trial

There is no comfortable way to write about this case, and we won't pretend otherwise.

In January 2023, Lindsay Clancy's three children — Cora, Dawson, and Callan — died in their home in Duxbury, Massachusetts. Clancy is now standing trial for their killings. Her defense is that she was not criminally responsible, arguing she was in the grip of postpartum psychosis and bipolar illness. Prosecutors argue she acted deliberately. As of this writing, the prosecution has rested after roughly 70 witnesses, the defense has begun its case, and no verdict has been returned.

We are not going to litigate that question here, and neither should any clinician reading this. What a jury decides about criminal responsibility is a legal determination made on a full evidentiary record that none of us have seen.

But something else surfaced in weeks of testimony that belongs in a clinical conversation: a detailed, sworn account of what a fragmented mental health record actually looks like from the inside. For those of us who write notes, request releases, and coordinate care for a living, that account is worth sitting with.

What the testimony described

Three threads from the treatment record, as described in court, are worth naming.

Clinicians who were treating the same patient at the same time did not have each other's information. A psychiatric nurse practitioner at a perinatal behavioral health program testified that she never spoke with the psychiatrist who was also treating Clancy, never accessed that psychiatrist's records, and was not aware Clancy had seen her roughly 14 times over four months. Asked on cross-examination whether it would have mattered to know the other clinician's diagnosis and treatment plan, she acknowledged that it would have — and said she never found out.

Records from higher levels of care did not reach outpatient providers. According to testimony, the nurse practitioner had no records from a Rhode Island hospital postpartum program, and no records from a five-day inpatient admission at McLean Hospital. The treating psychiatrist testified that she did not obtain or request records from Clancy's prior outpatient psychiatric treatment before beginning care, and learned of the January 2023 inpatient stay only after Clancy had already been discharged.

The medication history lived in pieces. Reporting on the trial record describes 13 different psychiatric medications prescribed across multiple providers in the months before the deaths, including antidepressants, benzodiazepines, and an antipsychotic that was reportedly discontinued near the end. No single clinician appears to have been looking at that list in its entirety.

The psychiatrist's own framing was that she trusted her patient to disclose what was clinically relevant. That is not an unusual position. It is close to the default in outpatient behavioral health.

This is not a story about two clinicians

It would be easy, and wrong, to read the above as an indictment of the individual providers involved. They were working inside a system that does not make coordination the path of least resistance: separate EHRs that don't talk to each other, releases that expire or never get signed, discharge summaries that route to a primary care provider and stop there, and caseloads that leave no billable time for the phone call that would have closed the loop.

It's also worth being clear about what was being missed. Postpartum psychosis occurs in roughly one to two births per thousand and typically emerges within days to weeks of delivery. It is a psychiatric emergency, with real risk of suicide and of harm to the infant. And it does not appear as a distinct diagnosis in the DSM — it is captured through a peripartum-onset specifier or folded into other categories. Perinatal psychiatry researchers interviewed during the trial have made the point plainly: clinicians are largely not trained to recognize it, and it is missed often. A history of bipolar disorder is the single largest risk factor, and a fluctuating presentation means a patient can look reassuring in a 30-minute appointment and be acutely unwell that night.

Which is precisely why the record matters. When a condition waxes and wanes, and when patients minimize, the longitudinal record and collateral information are not administrative overhead. They are the diagnostic instrument.

What this means for your documentation

None of the following is novel. All of it is the kind of thing that quietly slides when the schedule is full.

Write down what you don't know

A note that says "prior treating psychiatrist identified; records requested 3/14, not yet received; treatment plan and diagnosis currently unknown to this clinician" is a substantively different document than one that is silent on the question. It flags an open loop for you, for your coverage, and for anyone reading the chart later.

Treat releases as a clinical task, not a front-desk task

Signed at intake, scoped to the actual providers involved, and revisited when a new provider enters the picture — including after any hospitalization.

Reconcile medications from the source, not from recall

Ask for the prescriber and the pharmacy. Check the prescription monitoring database where you have access. Document the full list, the prescriber for each, and any recent changes or discontinuations, with dates.

Document risk in specifics

"Denies SI" carries far less information than what was asked, what was answered, what the patient's baseline looks like, who else is in the home, and what the plan is if things change overnight. The same applies to intrusive thoughts: note whether they are ego-dystonic obsessions or a loss of contact with reality, because that distinction changes the entire disposition.

Record the coordination attempts that failed

The call you placed and never got returned belongs in the chart. It is both a clinical fact and a fair account of what you did.

Capture perinatal specifics

Delivery date, sleep, feeding, and any personal or family history of bipolar disorder or prior postpartum episode. And a caution worth stating: the EPDS screens for depression. A reassuring score does not rule out psychosis.

The point underneath

Good documentation is usually framed defensively — what will protect you if this is ever reviewed. That framing is understandable, and it is too small.

The chart's real job is to make the next clinician's thinking possible: the covering colleague, the ED physician at 2 a.m., the prescriber who inherits your patient in six months, and often your own future self, who will not remember what you noticed on a Tuesday in March. In this case, according to testimony, that job went undone across several settings at once — and no individual clinician had a complete picture of a patient who was, by every account, actively seeking help.

That is the part we can act on. Not the verdict, and not the question of blame, but the ordinary work of making sure the record follows the patient.