Chapter 1: The Quarterly Review

The quarterly outcomes review for Pallinex began at 9:00 on a Tuesday in Conference Room 4B, with the blinds half-open and the wall screen already warm from the previous meeting.

Twelve people were present. Cal Britten from Brand Strategy stood at the front near the screen, one hand resting on the conference table in a way that suggested ease without requiring him to sit. Nora Chen was three seats from the end on the left side, with her laptop closed, a printed deck in front of her, and a pen aligned with the margin of the handout. The coffee in the paper cup by her elbow had gone lukewarm before the meeting started.

Cal moved through the first slides efficiently.

Market share: 62 percent of the prolonged grief disorder therapeutic segment, stable year over year.

Prescription volume: up 4.1 percent from the previous quarter.

Payer retention: unchanged.

Adverse event reports: within expected range.

No one interrupted. This was not the kind of meeting that rewarded interruption. Questions, when they came, were for calibration, not challenge. Slide numbers advanced in the lower right corner. The room's air system made a low, continuous sound above them that Nora had stopped consciously noticing years ago.

Cal clicked to Slide 17.

“Post-market functional recovery outcomes remain strong across all primary cohorts,” he said.

The table on the screen was simple. Three rows, four columns, pale blue shading behind the header. Bereavement. Relational loss. Occupational disruption. Functional recovery rates sat in the rightmost column: 87 percent, 84 percent, 81 percent. Consistent with baseline expectations. Consistent with the previous quarter's summary. Consistent with the original Phase III narrative the company had been telling, in one form or another, since 2028.

Nora looked at the denominators.

She did not know, at first, that she was looking at them. Her eye had already moved there by the time Cal began his sentence about patient adherence. The numbers were not dramatic. That was part of what made them visible. If they had been dramatic, someone else would have noticed them first and the noticing would have been social. This was smaller than that. A shift in the bereavement cohort count. Not large enough to appear unstable. Large enough to alter the shape of the table very slightly if a person knew what the table had looked like three months earlier.

She turned one page back in the printed deck to the prior-quarter comparator. The table had not been included. Only the summary rate. She remembered the denominator anyway.

Not exactly. Close enough.

The bereavement cohort should have widened, not narrowed. The age distribution of current users made that direction more likely than its opposite. Loss, in that cohort, was not a variable that trended down. If fewer people were appearing in the group, then either fewer people were being counted or the threshold for counting them had changed.

Cal said, “We’ve also seen improved onboarding compliance in the occupational disruption segment following the revised intake guidance.”

A woman from Market Access asked about regional payer variation. Someone from Medical Affairs answered. The room moved on around the table while Nora looked at the third line and then at the line above it and then at the total population count at the bottom of the slide. Her pen touched the margin of the handout. She wrote nothing.

Slide 18.

Slide 19.

By the time the meeting ended at 9:46, there had been two questions about reimbursement, one about pediatric off-label chatter in a restricted forum, and none about cohort construction.

People stood. Chairs moved back. Cal began a side conversation near the door with someone from Commercial Analytics. A woman at the end of the table said she had another call in four minutes. Laptops opened. Coffee cups were gathered. The room resumed its ordinary state with the practiced speed of people whose days were divided into discrete forty-five-minute territories.

Nora placed her pen across the handout, slid both into her folder, and left with the others.

The hallway outside 4B was the same hallway it had been at 8:58. Frosted interior glass on one side, framed product timelines on the other, a carpet pattern designed to conceal wear. A digital wellness display near the elevators rotated through messages about hydration, sleep hygiene, and adaptive recovery resources available through Employee Health. A woman Nora knew from oncology smiled at her in passing and said, “Morning.”

Nora said, “Morning,” and kept walking.

Her office was on the sixth floor, one wall glass, one wall shelving, no door. Senior enough for square footage, not senior enough for privacy. She set her folder down, woke her monitor, and opened the post-market surveillance dashboard for Pallinex. The dashboard loaded in layers: company logo, access token prompt, analytics shell, then the quarter's aggregate tables.

She pulled up the bereavement cohort trendline.

The same narrowing was there. Not a collapse. Not an outlier. A drift.

She clicked through the cohort definition history. The field was nested under surveillance protocol parameters, then under confirmatory assessment rules. There had been a revision seven weeks earlier.

Her screen filled with tracked changes.

Version 12.4 to 12.5. Dated seven weeks ago. Confirmatory assessment threshold for bereavement-cohort inclusion revised upward by 0.3 on the composite intake index. The adjustment was small enough to present as calibration. In practice, it would exclude cases with less legible presentations—patients whose bereavement did not score high enough on the intake instrument to count cleanly at the front end, even if they entered treatment later.

A small change to the denominator preserved the rate.

Nora read the revision note twice. Then she opened the metadata.

Approving authority: Dr. Rena Moss, Head of Clinical Analytics.

Rena was her direct supervisor.

Nora sat back in her chair. Not far. An inch, perhaps two. Outside her office bay, someone laughed at something in the corridor. A printer started somewhere behind her, ran for six seconds, then stopped.

Threshold adjustments happened. Post-market surveillance protocols evolved. Confirmatory criteria were not sacred. If a cohort was too noisy, if intake consistency was weak across sites, if one variable was generating too much heterogeneity for meaningful quarter-over-quarter comparison, then somebody tightened a rule. There were memos for this. Governance pathways. Statistical justifications. Most of them were reasonable.

The problem was not that the threshold had changed.

The problem was the direction of the change.

Nora opened the underlying rationale attachment.

“Revision implemented to improve cohort specificity and maintain interpretive reliability across post-market functional recovery reporting.”

Maintain interpretive reliability.

She read the sentence again. It was well written. That was part of its function. It described a technical objective without describing its cost. Higher specificity meant fewer ambiguous cases. Fewer ambiguous cases meant cleaner outcome reporting. Cleaner outcome reporting meant stable recovery rates. Nothing in the sentence was false.

Her inbox flashed at the lower corner of the monitor. A calendar reminder for an 11:30 methodology sync. She dismissed it without opening the invite.

Instead she searched the archived protocol language for the phrase “confirmatory assessment threshold.” There were six instances. Then she searched “bereavement cohort inclusion.” Twelve instances. Then “neuroimaging.” No results in the surveillance protocol.

That absence did not mean anything yet. Not by itself. The surveillance protocol was not the Phase III protocol. Post-market datasets were built to track function and safety at scale, not mechanistic depth. You did not run neuroimaging endpoints through a commercial monitoring instrument any more than you ran fMRI in an employee wellness check. The scale was wrong. The cost was wrong. The system was not built for that.

Still, she left the search result open for a moment longer than was necessary.

At 10:18, she stood, took her badge, and walked to Rena Moss’s office.

Rena’s assistant looked up from her desk. “She’s in with Regulatory.”

Nora nodded. “Do you know for how long?”

The assistant glanced at the internal calendar and gave the small apologetic smile of someone who had no power over time. “Hard to say. Maybe twenty?”

Nora said thank you and returned to her office.

She did not email. Email made shape too early. She preferred to ask simple questions in rooms where answers had to occupy air before they became language on a screen.

She reopened Slide 17 from the meeting deck and set it beside the revision document. Rate on the left. Threshold on the right. Result and mechanism. She looked from one to the other until the relation between them was stable in her mind.

At 11:30 she attended the methodology sync for oncology and spoke exactly once, to point out that a missingness assumption in one subgroup analysis had been treated as random when the dropout pattern suggested otherwise. The correction was noted. Someone thanked her. The meeting ended. She returned to her desk and opened the original Phase III trial archive for Pallinex.

The file was older than the current interface and loaded more slowly.

Outside, through the glass, the day had brightened without becoming sunny. Vantara’s campus buildings stood in their arranged intervals of steel, warm stone, and professionally maintained landscaping. Employees crossed the central courtyard carrying lunch containers and phones, moving at the speed of people whose biometric badges would register their return to the building within acceptable ranges. Near the fountain, a freestanding display advertised Adaptive Recovery resources in soft blue lettering.

At 12:14 the Phase III protocol opened.

Nora went first to endpoints.

Primary endpoint: functional recovery, defined as a composite measure of productivity return, actigraph-assessed sleep normalization, reduction in self-reported distress, and restoration of social engagement frequency to baseline range within the adaptive window.

Secondary endpoints: safety, metabolic panels, dropout rates.

She scrolled more slowly.

There were many ways for a trial to answer a question incorrectly. Some were crude. Some were elegant. The elegant ones were the hardest to see because they resembled practicality. Cost constraints. Logistical complexity. Endpoint selection aligned with regulatory utility. Study design was an architecture of exclusions as much as inclusions. A trial did not merely measure what it measured. It created a world in which unmeasured things had no standing.

Nora stopped at a footnote near the bottom of page 38.

Neuroimaging endpoints were considered and excluded due to cost constraints and logistical complexity in a multi-site trial design.

She kept her eyes on the sentence.

Excluded due to cost constraints and logistical complexity.

A reasonable footnote. More than reasonable. A defensible one. Multi-site neuroimaging was expensive, difficult to standardize, vulnerable to site-level variance. Any review board in 2026 would have accepted the decision without difficulty if the primary goal was scalable approval. Functional recovery metrics were cleaner. Employers understood them. Regulators understood them. Insurers understood them. People went back to work. People slept. People reported improvement.

The thing those metrics did not ask was whether grief itself had resolved, or only become inaccessible.

Nora looked away from the screen and toward the glass wall of her office, but she was not seeing the courtyard now. She was seeing her own kitchen from the previous evening. The framed photograph of her mother on the counter where Ming had placed it months ago. Liling Chen in a blue sweater, seated at a restaurant table with one hand lifted halfway toward the camera because she disliked being photographed but tolerated it for family. Nora had looked at the frame while waiting for water to boil and had felt the factual recognition of a known face.

Nothing else.

Not pain. Not warmth. Not the pressure behind the sternum that belonged there. Only identification.

She turned back to the protocol and read the footnote again.

At 12:31, Rena’s assistant sent a message saying Dr. Moss was free at 1:00 if Nora still wanted to stop by.

Nora typed, Yes. Thank you.

Then she closed the message, left the Phase III protocol open on one monitor and the cohort revision on the other, and sat very still in the interval before lunch, with the architecture of two separate decisions beginning to take the shape of the same question.