Central finding #
Paper (Results): “Hospitalization was reported by 55 (0.03%) registrants. Among 45 registrants with information… 29 indicated that the hospitalization was unrelated to vaccination, 13 completed a VAERS report, and three did not wish to complete a VAERS report.”
Paper (Table 2): Hospitalization row printed as 0 / 0 / 0 / 0 / 0 for every age group and the total.
Paper (Abstract): “Health care providers and patients can be reassured that adverse events reported after a bivalent booster dose are consistent with those reported after monovalent doses.”
The MMWR’s pitch is familiar reactogenicity, <1% medical care, hospitalizations framed as vanishingly rare and mostly “unrelated,” and reassurance that nothing new appeared versus monovalent doses. Table 2 goes further: it prints zero hospitalizations in every column—even though Results admit 55.
Compared to the v-safe dataset (~240,217 reconstructed; paper 211,959): about
62 week-1 Hospitalization checkbox registrants; roughly 50 lack a
Hospitalisation preferred term on those responses (including MI, ICU anaphylaxis, chest pain + LOC).
Separately: Chest pain 637, Dyspnoea 980, Palpitations 428,
Myocarditis 5, Pericarditis 7, Death 5—unpublished on the v-safe Results page.
Full write-up: vsafe-analysis/mmwr/mm7144a3/anomalies_report.md.
A. Table 2 prints zero hospitalizations while the text admits 55 #
Paper (Table 2 vs Results): Hospitalization = 0 everywhere in the table; narrative reports 55 (0.03%).
Under reconstruction: 62 distinct registrants with a day-0–7 Hospitalization checkbox.
Of 87 such responses, about 61 lack Hospitalisation on that response.
Coded examples without the hosp PT still include myocardial infarction, anaphylaxis + intensive care,
chest pain with loss of consciousness, and atrial fibrillation with stent placement.
Why this matters: An all-zero table plus soft “unrelated” language teaches clinicians that hospitalization after bivalent boosters is a non-issue. The dataset shows dozens of hospital checkboxes— including cardiac and ICU-level coding—while failing to attach the standard hospitalization code on most rows.
B. Soft disposition vs failed call-center follow-up #
Paper (Results): Of 45 hospitalizations with follow-up information, 29 “unrelated,” 13 completed VAERS, 3 declined.
Among the 62 week-1 hospital registrants with a GDIT outcome recorded: about Successful 34 / Unsuccessful 28. Nearly half never got a successful closeout— so “unrelated” and “declined” sit beside a large silent remainder folded into “45 with information.”
C. Chest pain / myocarditis / death unpublished next to “reassured” #
Paper (Abstract / VAERS): Providers and patients can be “reassured” events are consistent with monovalent doses. VAERS: 5 myocarditis, 4 pericarditis, 36 deaths among serious reports.
v-safe Results never publish chest-pain, dyspnea, palpitation, myocarditis, pericarditis, or death counts for this bivalent cohort. In the coding window: Dyspnoea 980, Chest pain 637, Palpitations 428, Pericarditis 7, Myocarditis 5, Death 5, plus PE/DVT/CVA/MI preferred terms.
Free text in the death/myocarditis set includes “Myocarditis relapse,” hospital admission for Afib & pericarditis, “We think shot killed her,” “Stroke; respiratory failure; death,” and “massive hemorrhage of the brain.”
Why this matters: Reassurance is memorable. Leaving hundreds of cardiac-symptom codes—and non-zero myocarditis/pericarditis/death preferred terms—off the v-safe Results page understates what active surveillance recorded while VAERS alone carries the serious-cardiac and death narrative.
Death / myocarditis / pericarditis MedDRA (all).
Chest pain MedDRA sample (100 of 637; free-text prioritized).
Clot / stroke / MI MedDRA.
Serious free-text narratives (sample).
D. Hospitalisation preferred terms exist elsewhere #
About 175 distinct registrants carry a Hospitalisation preferred term in the coding window
(table shows 80 illustrative unique cases). The preferred term was available—yet most week-1 hospital checkboxes lack it.
E. Emergency visits in week 1 #
ER checkbox responses (day 0–7): about 277 registrants—consistent with Table 2’s ~0.1% emergency line. Context for how much acute care sits under the “<1% medical care” headline; detailed ER grid omitted to keep the page focused on hospitalizations and cardiac coding.
Summary #
- Table contradiction: Hospitalization printed as 0 everywhere vs narrative 55 (reconstructed 62)
- Under-coding: most week-1 hospital checkboxes lack Hospitalisation PT despite MI / ICU anaphylaxis / cardiac coding
- Follow-up: ~28/62 Unsuccessful GDIT among hospital reporters beside soft “unrelated / declined” prose
- Reporting gap: Chest pain 637 / Dyspnoea 980 / Palpitations 428 / Myocarditis 5 / Pericarditis 7 / Death 5 unpublished on v-safe Results
Bottom line: Compared to the v-safe dataset, this brief gets everyday reactogenicity roughly right— but visually zeros hospitalizations in Table 2, softens dozens of hospital cases, and leaves cardiac-symptom and death coding off the v-safe Results page beside a “providers and patients can be reassured” message.