Central finding #
Paper (Discussion): “A rare but serious adverse event occurring primarily in women, blood clots in large vessels accompanied by a low platelet count, was rapidly detected by the U.S. vaccine safety monitoring system.”
This MMWR sits in the Janssen pause / TTS moment. Readers are told dual monitoring “rapidly detected” rare clotting via VAERS (17 TTS cases; 88 deaths reviewed), while v-safe mainly shows expected reactogenicity and that 1.4% sought medical care.
Compared to the v-safe dataset for the same Janssen week-after surveys, that understates active surveillance:
Death-coded free text (including thrombosis→heart-failure), ~294 thrombosis-coded
registrants never reconciled with VAERS TTS, and hospitalization checkboxes that usually lack MedDRA
Hospitalisation—even when free text describes admission or elevated D-dimer.
Cohort: Janssen doses 2 Mar–12 Apr 2021 with ≥1 day-0–7 check-in (338,996 registrants; paper 338,765).
Table 3 care-seeking rates match; the gaps are serious free text, coding, and omitted v-safe deaths.
Full write-up: vsafe-analysis/mmwr/mm7018e2/anomalies_report.md.
A. Hospitalizations in v-safe — under-coded #
Paper (Methods): “Enrollees who report seeking medical care are contacted by CDC’s v-safe call center and encouraged to complete a VAERS report, if indicated.”
Day-0–7 hospitalization checkboxes: 170 responses / 128 people. Only
31 (~18%) carry MedDRA Hospitalisation; 43 have no MedDRA at all.
Free text still describes ER admissions, pancreatitis, atrial flutter, appendectomy, and elevated D-dimer workups.
Why this matters: during a TTS crisis, dumping elevated-D-dimer care into
Unevaluable event or omitting Hospitalisation from clear admissions weakens v-safe as a
partner to VAERS clotting surveillance.
Illustrative sample of responses without PT Hospitalisation (139 of 139 under-coded rows with preferred ordering).
B. Emergency visits — under-coded #
1,136 day-0–7 ER checkbox responses; only 170 (~15%) carry an emergency-care– or hospitalisation-like MedDRA term. Table below: sample with free text and no ER-like PT.
C. Thrombosis signal vs VAERS-only TTS #
Paper (Results): “…14 CVST cases had been confirmed… for a total of 17 TTS cases.”
The SAE narrative is VAERS TTS. v-safe Results never mention thrombosis codes. In this cohort’s coding window: 243 Thrombosis, 27 DVT, 12 pulmonary embolism, 1 CVST, plus thrombocytopenia-class terms—about 294 registrants with any thrombosis-related PT (9 also thrombocytopenia-coded). Not confirmed TTS—but unpublished next to the “rapid detection” story.
Sample of up to 250 thrombosis-class registrants (prefer free text; CVST/PE/DVT sorted first).
Unevaluable-only serious free text #
Keyword-filtered sample where free text mentions clot / D-dimer / hospital / stroke / anaphylaxis language but
MedDRA is only weak labels such as Unevaluable event (workplace “I work at the hospital” filtered out).
D. Anaphylaxis in v-safe #
Paper (Results): “Among 79 reports of anaphylaxis after vaccination, four were confirmed…”
v-safe MedDRA includes 36 Anaphylactic reaction and 6 Anaphylactic shock codes in this cohort—unused beside the VAERS anaphylaxis tally.
E. Deaths reported in v-safe #
Paper (Results): “CDC and FDA reviewed 88 reports of death after receipt of Janssen COVID-19 vaccine…”
The paper never mentions v-safe deaths. Under inclusion: 10 MedDRA Death and
1 Cardiac arrest. Free text includes “heart failure caused by a blood clot,”
“I died Wednesday 4/7…,” and “passed away”—omitted while VAERS deaths are reviewed at length.
Especially salient in a TTS paper: 5DU-20396-72475 (Death + Thrombosis + Cardiac failure).
Call-center follow-up #
Among 128 hospitalization reporters, GDIT final outcomes are roughly Successful 54 vs Unsuccessful 74 (see hosp grid). Methods say medical-care reporters “are contacted”; more than half lack a Successful disposition.
Summary #
- Major omission: Death-coded v-safe registrants (incl. clot-related free text) never mentioned
- Reporting gap: ~294 thrombosis-coded v-safe registrants unpublished next to 17 VAERS TTS cases
- Coding failure: ~18% of hosp checkboxes get Hospitalisation PT; ~15% of ER rows get ER-like PT
- OK: Table 3 care-seeking rates match the v-safe dataset
Bottom line: Compared to the v-safe dataset, this paper publishes care-seeking rates but leaves TTS/clotting codes, under-coded hospital admissions, and death-coded free text offstage next to a VAERS-centered “rapid detection” story.