Marco's Morning Report

Saturday, August 15, 2026 · Cape Coral, FL

Morning, Marco. The federal page you were told to stop trusting finally updated — and it moved by more than three thousand cases in a single jump, which tells you something about how the counting works. Also: a 21,000-woman study that will land in a family meeting before it lands in a guideline, and a fight over who gets your ER's patient names. Coffee, then sleep.

A · Top Stories

Three items. All moved in the last 48 hours.

Update: the cyclospora page caught up — and it jumped from 6,358 to 9,481 cases in one revision

Yesterday the outbreak page had been frozen since August 5 while the outbreak wasn't. It has now moved: the multistate iceberg-lettuce outbreak stands at 9,481 illnesses across 17 states, with at least 398 hospitalizations and two deaths, both in Michigan. That's an increase of more than 3,100 cases in a single update. Maine and Massachusetts are the newly added states, and FDA has begun onsite inspection and sampling at Taylor Farms de México. The most useful number in the release is the one nobody headlines: last illness onset August 3.

Why it matters to you. The last-onset date is the clinical line. Product with any plausible shelf life is long gone, so a new watery diarrhea presenting this weekend is very unlikely to belong to this cluster — but the wider surveillance number is the one that should keep the test on your list. CDC has logged roughly 13,900 lab-confirmed domestic cyclosporiasis cases since May 1 against about 1,180 in the same window last year. That is a season, not an outbreak, and it doesn't end when the lettuce does.
Suggested action: you can put the federal page back in rotation as a denominator — but note how it updates. It caught up in a lump, not a stream. If you quote a number at a group meeting, say the date it was published, not "currently."

CDC investigation update · FDA outbreak page · Food Safety News, Aug 13

Estrogen-only menopausal hormone therapy linked to 39% lower odds of dementia — with autopsy data behind it

Stanford investigators published in Neurology on August 12 an analysis of 21,462 women drawn from national datasets, and the coverage caught up over the last 48 hours. Women who had used estrogen-only menopausal hormone therapy had 39% lower odds of a clinical dementia diagnosis and 35% lower odds of Alzheimer's pathology at autopsy, with biomarker profiles consistent with less amyloid accumulation. They also performed better on memory testing and functional independence measures. The autopsy arm is what makes this more than another observational hormone-therapy headline — the association survives when you look at brains rather than billing codes.

Why it matters to you. Not because it changes anything you prescribe — it doesn't, and the authors say so plainly. It matters because this is exactly the study that reaches a family before it reaches a guideline. You will get the question from the daughter of an 82-year-old with delirium on top of baseline cognitive impairment: "she was taken off estrogen in 2003, did that cause this?" The honest answer has three parts: observational design, historical treatment patterns that don't map onto today's prescribing, and estrogen-only regimens (i.e., post-hysterectomy women), which is a different population from combined therapy.
Suggested action: bank one sentence for the family conversation — "this was an association in women who were treated decades ago, and the researchers themselves say it can't show cause." Saying it before the family Googles it is worth ten minutes of your night.

Stanford Medicine · Medscape · Scientific American

KFF: about $60B in federal Medicaid hospital payments across 37 states sits above the new OBBBA caps

A KFF analysis circulating this week puts numbers on the state directed payment (SDP) limits in the 2025 reconciliation law. Previously states could direct managed-care plans to pay hospitals up to average commercial rates; the new limits cap those payments at or near Medicare levels. KFF estimates roughly $60 billion in federal Medicaid spending across 37 states and D.C. would exceed the new ceilings once fully implemented. Eight states account for half of the exposure: California ($7.4B), Illinois ($4.0B), Kentucky ($3.9B), Texas ($3.5B), North Carolina ($3.4B), Louisiana ($3.3B), Arizona ($3.0B), Michigan ($2.6B).

Why it matters to you. Third squeeze in three weeks pointing at the same place. You've already got the statutory site-neutral cut in the CY2027 OPPS rule (comments close August 31) and the FY2027 IPPS rule effective October 1. SDP caps are the Medicaid-side version. None of these change a single order you write — they change what your hospital can afford to keep open, which shows up in your world as bed availability, case-management staffing, and which post-discharge clinics still take Medicaid in ninety days.
Suggested action: when the site-neutral comment letter goes around before August 31, this is the second paragraph. Not "payment is going down" — but that three separate mechanisms hit the same transitions-of-care infrastructure on overlapping timelines, and nobody has modeled them together.

KFF analysis · HFMA

B · The Debate

One topic, both sides, no verdict.

Should a federal safety agency be able to pull identifiable emergency-department records on every patient?

Why it's live right now: the Consumer Product Safety Commission is rebuilding the National Electronic Injury Surveillance System — "NEISS-R" — expanding from roughly 70 hospitals in 36 states to 100 hospitals in all 50, and from about 400,000 records reviewed a year to roughly 2 million, routed through a contractor, Konza Health, on a $15.9M award. Reporting in late July established that CPSC asked hospitals for names, addresses and diagnoses for all ED patients, not just product-related injuries. On August 12 the American Hospital Association filed a comment letter asking the agency to pause and scale the request back — which is what moved this from a reporting story to a live policy fight.

The case for

Injury surveillance built on hand-abstracted, de-identified summaries from seventy hospitals is a 1970s instrument being used to police a 2020s consumer market. Acting CPSC Chairman Peter Feldman has framed the rebuild as the agency "investing in AI-enabled workflows that improve the quality and quantity of injury surveillance data," with the digital infrastructure to handle a much larger volume of electronic health records. The argument underneath: you cannot detect a rare defect — a battery that ignites, a crib that entraps, a recalled device that keeps injuring people after the recall — from a 400,000-record sample abstracted by hand, and every year of undercounting is measured in children. Automated extraction also reduces the manual handling of sensitive records that the current system depends on, which the agency argues is a privacy improvement, not a cost.

CPSC on NEISS · NEISS 2026 fact sheet

The case against

A product-safety agency asking for every ED patient's name and address is collecting far past its own mandate — the requested categories include injuries CPSC does not regulate. The American Hospital Association, in its August 12 letter, asked CPSC to collect only what is "minimally necessary" for injury surveillance, to explain how it will comply with data privacy and security law, and to keep hospital participation voluntary; AHA reports its members are confused about whether participation is mandatory at all, since CPSC and the contractor have both used that word. The Consumer Federation of America — not an industry group — raised the same objection from the consumer side, questioning whether identifiable personal health data is needed to do the job. Several systems, including Mass General Brigham, Harborview and Henry Ford Health, have already declined or paused.

AHA, Aug 12 · Consumer Federation of America · KFF Health News

Where it might land: the unresolved question is whether participation is mandatory or voluntary — CPSC and Konza have said mandatory, AHA is asking the agency to say voluntary in writing. That single word decides whether this is a privacy debate or a compliance deadline.

C · On My Radar

Noteworthy, not urgent.

D · Trends to Watch

Signals, not stories.

E · Ideas & Opportunities

Small, actionable, optional.

H · Background / Already Covered
  • Cyclospora background. Taylor Farms de México pulled central-Mexico iceberg lettuce from the US market on July 17; exposure window ran roughly June 22 – early August. Both deaths were in Michigan, both in comorbid patients. Covered daily since July 27.
  • Childhood vaccine executive order (Aug 10) and the state split. The West Coast Health Alliance and the Florida AAP chapter are continuing to endorse the AAP schedule. No new development in the last 24 hours. Covered Aug 13–14.
  • Cuba's grid. No seventh nationwide collapse since the August 3 restoration. Fidel Castro's centenary was marked August 13 with exhibitions and a colloquium in Havana. Covered Aug 4–13.
  • Puerto Rico drought baseline. July was San Juan's driest month in 120+ years and its fourth hottest; about 25% of the island in severe drought, 36% moderate. Covered Aug 2–13.
  • Ambient AI scribes. Previously covered as a Section B debate (May 26, June 14, July 20) and in the SHM 2026 Workforce Experience Report — 64% of hospitalists using AI, 73% reporting benefit, only about a third adequately trained. Today's item is the hospital-level adoption map, not the clinician-level data.
  • Reacher S4 premiered August 12 on Prime Video, three episodes then weekly; Tomatometer settled around 87%.
📚 Archive · last 7 morning reports

A page that caught up all at once, a study that will reach a family before it reaches a guideline, and a Green Lantern show critics like. Sleep well, Marco.