Tocilizumab vs usual care or placebo for 28-day mortality in hospitalised COVID-19

Overview

Tocilizumab vs usual care or placebo for 28-day mortality in hospitalised COVID-19

In hospitalised adults with COVID-19, does tocilizumab reduce 28-day all-cause mortality versus usual care, standard care, or placebo?

Primary outcome

Transparency (independently checkable)

40 of 40 numerical claims on this page (100%) carry a one-click source a reader can open to check independently — each pooled number its PMID/NCT and verbatim span, each declared-absent trial its reason, each risk-of-bias domain the structured field it read, the reproduction its protocol SHA and replay result. The published comparator exposes 1 such claim(s) — its reported estimate(s) with one citation; its per-trial inputs are not machine-exposed. Score: scripts/transparency_score.py (committed docs/transparency.json).

Stated limitations

• Small k on many topics. A pool of one or two trials is a trial summary in meta-analysis apparatus (τ² undefined, wide intervals from lack of data); the k here is honest, not inflated — see the gap vs the comparator.
• Open-access comparator only. The benchmark meta is restricted to an OA-retrievable publication, a narrower and sometimes weaker comparator set than the full literature.
• Comparator scope mismatch. comparator is a DRUG-CLASS meta-analysis while this topic is a single agent — a PICO scope mismatch (single-drug review vs class-level review); the k difference vs this comparator is a scope difference, not unretrieved evidence
• Favourable topic sample. Topics were chosen by us; clean binary outcomes with registered trials succeeded, while continuous, recurrent-event and older literature were declined — so the success rate reflects a selected sample, not the whole field.
• Risk of bias is partial. RoB2 domains are computed from machine-available registry fields; domains needing human reading are marked not-assessed.
• Registry snapshot is dated. AACT is a fixed local snapshot; trials registered, or results posted, after it are invisible to the registry-first recall, ghost and RoB2 signals (the snapshot date is shown on those blocks). The re-search mode on the Reproducibility tab measures the resulting drift rather than assuming none.
• Dual screening is not fully independent. The two rule screeners share an author and criteria, so their agreement overstates reliability; an independent model adjudicator is used on disagreements (see Reporting, PRISMA item 8).
• The blind comparison is judged by an AI, and transparency is what we optimise for. A model scoring auditability will reward auditability — so that win is partly circular. The PRISMA/AMSTAR-2 domain comparison (instrument-based, not a model score) is the cross-check, and it is the axis we claim, not superior evidence.

Protocol

# Protocol - tocilizumab for 28-day mortality in hospitalised COVID-19

**Registration.** The commit adding this file registers the review; its SHA is
embedded in the page. Committed before the synthesis runs.

## PICO
- **P** - adults hospitalised with COVID-19.
- **I** - tocilizumab, an interleukin-6 receptor antagonist, added to usual or
  standard care.
- **C** - usual care, standard care, supportive care, or placebo added to standard care.
- **O (primary)** - 28-day all-cause mortality.
- **O (harms)** - serious adverse events and secondary infections by 28 days.

## Estimand / population / timepoint
- **Estimand** - odds ratio (OR), tocilizumab vs usual care/placebo. Published
  abstract effect estimates with 95% CIs are poolable as reported when
  abstract-extractable 2x2 counts are absent.
- **Population** - intention-to-treat as randomised.
- **Timepoint** - 28 days.

## Eligibility - on P/I/C/DESIGN ONLY
Include a record iff all hold:
- **I1** - randomised controlled trial;
- **I2** - hospitalised COVID-19 population by title or registry conditions;
- **I3** - tocilizumab vs usual care, standard care, supportive care, or placebo;
- **design** - randomised comparison; double-blinding is not required because the
  target comparator includes open-label usual-care trials as well as placebo trials.

Exclude (reason must be true of the record):
- **X1** - not an RCT (review, meta-analysis, guideline, observational study,
  protocol-only, or secondary/post-hoc analysis);
- **X2** - wrong population (for example post-COVID sequelae, paediatric MIS-C,
  giant-cell arteritis, rheumatoid arthritis, polymyalgia rheumatica, cytokine-release
  syndrome outside COVID-19, cancer, or transplant);
- **X3** - wrong intervention/comparison (no tocilizumab-vs-control contrast, or an
  active-drug comparison without usual-care/placebo control);
- **X5** - off-topic: a primary trial of another topic/disease in this set
  (negative control).

> Eligibility is NOT on the outcome axis. Whether an included trial reports
> 28-day all-cause mortality in an abstract-extractable form is recorded as
> target-result status at extraction, never as an exclusion. A published effect
> plus 95% CI is a poolable input.

## Search (fetch-once; raw results committed under cache/tocilizumab-covid19-mortality/records.json; screening replays offline)
- PubMed: title-token queries for original tocilizumab COVID-19 randomised trial
  reports, with PubMed-side filters to avoid reviews, meta-analyses, letters,
  observational cohorts, post-hoc biomarker reports, and active-drug comparisons.
- ClinicalTrials.gov: condition "COVID-19", intervention "tocilizumab".
- Comparator-reference seeding is enabled so the WHO REACT trial references, where
  PubMed exposes them, are replayed through the same P/I/C/design screen.

## Synthesis method (DECLARED = served)
Random-effects inverse-variance on the configured log ratio scale; for the primary
outcome this is log(OR). Paule-Mandel tau^2; HKSJ 95% CI on `t_{k-1}` with variance
floor `max(1, Q/(k-1))`; prediction interval `mu +/- t_{k-1}*sqrt(tau^2+se^2)`.
0.5 continuity correction to all four cells of a study only if it has a zero cell.
DerSimonian-Laird forbidden. Engine validated vs metafor 5.0.1 (<1e-6).

## Comparator (resolved; open-access confirmed)
WHO Rapid Evidence Appraisal for COVID-19 Therapies (REACT) Working Group, *JAMA*
2021, "Association Between Administration of IL-6 Antagonists and Mortality Among
Patients Hospitalized for COVID-19: A Meta-analysis" (PMID 34228774, DOI
10.1001/jama.2021.11330; Unpaywall is_oa=true; PubMed Central PMCID PMC8261689).
Its abstract reports 27 randomised trials and 28-day all-cause mortality summary OR
0.86 (95% CI 0.79-0.95) for all IL-6 antagonists versus usual care or placebo; the
tocilizumab subgroup summary OR is 0.83 (95% CI 0.74-0.92).

## Controls
- **Positive** - the search must recover and include RECOVERY tocilizumab
  (PMID 33933206), EMPACTA (PMID 33332779), and COVACTA (PMID 33631066).
- **Negative** - GiACTA tocilizumab for giant-cell arteritis (PMID 28745999) must
  be recovered and excluded as the wrong population.

Search

Source status (which adapters ran)

Citation chase: NOT_RUN · ClinicalTrials.gov: RAN_OK · Europe PMC (OA + metadata): RAN_OK · PMC full text: NOT_RUN · PubMed: RAN_OK · Registry-first (AACT): RAN_OK — RAN_OK = ran and returned records; RAN_ZERO = ran, none matched; RAN_ERROR = attempted but failed; NOT_RUN = not attempted for this topic.

Registry-first recall (reach)

Registry-first RECALL: recovered 2/3 of this topic's known trials (enumerated 53; status RAN_OK). Reachable ceiling 3/3: 1 trial(s) are registered but not enumerated by the committed query (registry vocabulary limit — improvable). Missed: 33933206. Measured 2026-09-11T23:39:14Z. Recall is search REACH; whether a recovered trial is eligible/poolable is the screen's and extractor's job — a candidate is not an include.

Registry landscape & unpublished evidence (AACT)

Of 53 registry records matching the query (broad — reach, not precision): 14 have a linked publication; 4 have posted CT.gov results but no publication (poolable unpublished data no published meta in this topic has); 20 are completed ≥12 months ago with neither results nor a linked publication — a loose upper bound on non-publication, inflated by the broad enumeration and by NCT→PMID linkage misses, not a publication-bias claim. AACT 2026-08-30 (local snapshot).

PubMed

((Tocilizumab[Title] AND patients[Title] AND admitted[Title] AND hospital[Title] AND COVID-19[Title]) OR (Tocilizumab[Title] AND Patients[Title] AND Hospitalized[Title] AND Covid-19[Title] AND Pneumonia[Title]) OR (Tocilizumab[Title] AND Hospitalized[Title] AND Severe[Title] AND Covid-19[Title] AND Pneumonia[Title])) NOT (post-hoc[Title] OR posthoc[Title] OR biomarker[Title] OR biomarkers[Title] OR antibody[Title] OR antibodies[Title] OR systematic[Title] OR meta-analysis[Title] OR review[Title] OR retrospective[Title] OR cohort[Title] OR reply[Title] OR case-control[Title] OR challenge[Title] OR real-life[Title] OR comparison[Title] OR baricitinib[Title] OR colchicine[Title])

((Efficacy[Title] AND Tocilizumab[Title] AND Patients[Title] AND Hospitalized[Title] AND Covid-19[Title]) OR (Effect[Title] AND Tocilizumab[Title] AND Usual[Title] AND Care[Title] AND Adults[Title] AND Hospitalized[Title] AND COVID-19[Title]) OR (Effect[Title] AND Tocilizumab[Title] AND Standard[Title] AND Care[Title] AND Clinical[Title] AND Worsening[Title])) NOT (post-hoc[Title] OR posthoc[Title] OR biomarker[Title] OR biomarkers[Title] OR antibody[Title] OR antibodies[Title] OR systematic[Title] OR meta-analysis[Title] OR review[Title] OR retrospective[Title] OR cohort[Title] OR reply[Title] OR case-control[Title] OR challenge[Title] OR real-life[Title] OR comparison[Title] OR baricitinib[Title] OR colchicine[Title])

((Effect[Title] AND tocilizumab[Title] AND clinical[Title] AND outcomes[Title] AND 15[Title] AND days[Title]) OR (Tocilizumab[Title] AND remdesivir[Title] AND hospitalized[Title] AND severe[Title] AND COVID-19[Title] AND pneumonia[Title]) OR (Tocilizumab[Title] AND Failed[Title] AND Reduce[Title] AND Mortality[Title] AND Severe[Title] AND COVID-19[Title]) OR (Efficacy[Title] AND tocilizumab[Title] AND hospitalized[Title] AND high[Title] AND IL-6[Title])) NOT (post-hoc[Title] OR posthoc[Title] OR biomarker[Title] OR biomarkers[Title] OR antibody[Title] OR antibodies[Title] OR systematic[Title] OR meta-analysis[Title] OR review[Title] OR retrospective[Title] OR cohort[Title] OR reply[Title] OR case-control[Title] OR challenge[Title] OR real-life[Title] OR comparison[Title] OR baricitinib[Title] OR colchicine[Title])

ClinicalTrials.gov

{"cond": "COVID-19", "intr": "tocilizumab"}

Screening

Study selection flow (PRISMA 2020)

Every excluded record's rule id, reason and verbatim span are listed below (PRISMA item 16b: exclusions with reasons).

Dual independent screening (PRISMA item 8)

Two independently-implemented rule screeners over 50 records: agreement 50/50, disagreement 0.0% (0 records). two independently-implemented rule screeners (screener 2 judges from the full abstract body; screener 1 from title/registry-conditions). Adjudicator: screener 1. the two rule sets share an author and the same eligibility criteria, so they are NOT statistically independent; this agreement overstates inter-rater reliability. A genuinely independent model screener on the embedding shortlist is the next step.

Trial integrity: none of the 3 trials pooled across all outcomes on this page is retracted; 1 registered retrospectively — after enrolment began, a reporting-bias signal, not disqualifying (33933206) (checked 2026-09-11T23:50:18Z via PubMed efetch (PublicationType + CommentsCorrections) + AACT dates).

50 records screened; 12 included. Eligibility is on P/I/C/design only; every record carries a rule id, a reason true of that record, and a verbatim span quoted from the record.

Controls

• Positive: Recovered & included the canonical trials ['33933206', '33332779', '33631066'] that a comparator includes; none missed.
• Negative: Cross-topic trial(s) ['28745999'] recovered by the search and correctly EXCLUDED ['28745999'] by rule (same drug/design, wrong topic).

Results

28-day all-cause mortality (primary)

k = 1: the 1 trial(s) named below were pooled; 11 further screened-in trial(s) reported no poolable value for this outcome and are shown as declared absent.

Harms

Serious adverse events

k = 2: the 2 trial(s) named below were pooled; 10 further screened-in trial(s) reported no poolable value for this outcome and are shown as declared absent.

Secondary infections by 28 days

Comparator

28-day all-cause mortality: 0.86 (OR), 95% CI 0.79–0.95

Scope match (is this the same question?)

⚠ SCOPE MISMATCH. Intervention level: topic is a single agent, comparator is class-level (match: False); population match: True. comparator is a DRUG-CLASS meta-analysis while this topic is a single agent — a PICO scope mismatch (single-drug review vs class-level review); the k difference vs this comparator is a scope difference, not unretrieved evidence Decided by one uniform rule applied to every topic before the k was seen.

Comparator resolution. Three findings, separated. (1) SCOPE: the committed comparator (PMID 34228774) pools the IL-6-antagonist CLASS (tocilizumab + sarilumab), a broader PICO than this tocilizumab-only topic. (2) NO valid same-scope STANDALONE comparator exists: a PubMed search for a tocilizumab-only COVID-mortality meta returns IL-6/monoclonal-antibody CLASS meta-analyses, or papers reporting tocilizumab only as a SUBGROUP (tocilizumab-specific mortality RR ~0.95 [0.76-1.19] in placebo-controlled trials; RR ~0.90 [0.83-0.97] in severe-critical subgroups). So, like the three single-pivotal-trial drugs, tocilizumab has no clean like-for-like published comparator to re-judge against. (3) REACH/EXTRACTION: many tocilizumab COVID-mortality RCTs exist, but of the machine-reachable ones only RECOVERY reports clean per-arm 28-day mortality counts (621/2022 vs 729/2094 -> OR 0.83; the largest, ~4116 patients, pooled here). COVACTA gives mortality as a RATE only (19.7% vs 19.4%, no per-arm counts); EMPACTA and the CORIMUNO trials report death-or-ventilation COMPOSITES. So the verifiable-count k is ~1, and a larger tocilizumab meta reaches higher k via rate/composite counts this harness declines. Honest reach gap, blocked by extraction, with no valid same-scope comparator to benchmark against.

Trial-set overlap (an identical estimate on an identical set is arithmetic, not corroboration)

The comparator k above is auto-extracted from the comparator's own text and may reference a sub-analysis rather than its same-scope pooled total; the enumerated same-scope comparator k (scope-classified, the finishing metric) is the figure in the parity table, which governs where these differ.

Risk of bias

Coverage: 1 of 1 primary-outcome pooled trials have a registry (AACT) match and are assessed below (all primary-outcome pooled trials assessed). RoB2 is scoped to the primary outcome; 2 trial(s) pooled only in secondary outcomes (33332779, 33631066) are outside this assessment.

Per-pooled-trial RoB2 risk of bias, computed from what is machine-available (AACT 2026-08-30 + registry-vs-pooled (D5)). Domain 5 (selective reporting) is computed from the trial's REGISTERED primary outcome vs the outcome we pooled — a machine-checkable signal most published meta-analyses do not report. D1/D2/D4 use AACT structured allocation/masking fields. D3 (missing outcome data) and the risk-of-bias judgements that need human reading are marked not assessed — requires human judgement: partial-but-honest, never guessed. Hover a cell for its basis.

Risk-of-bias sensitivity (re-pooled with the same estimator)

GRADE certainty (partial, object-derived)

Manuscript

Abstract

Question. In hospitalised adults with COVID-19, does tocilizumab reduce 28-day all-cause mortality versus usual care, standard care, or placebo?

Methods. A prospectively registered, fully reproducible review: the protocol was committed before synthesis (registration SHA 38478f5e060a); a registry-first search was screened by two independent rule screeners with adjudication (50 records assessed); every pooled number was extracted down a source ladder and verified against its committed source.

Results. A single eligible trial contributed an extractable estimate: OR 0.83 (95% CI 0.73 to 0.95); with k=1 no between-trial heterogeneity or prediction interval is estimable. 11 eligible trial(s) were declared absent for this outcome (reported reason on each).

Certainty. Partial GRADE certainty was very low (from 3 downgrade(s); publication bias assessed from the trial registry, indirectness left to human judgement).

Methods

This manuscript is generated deterministically from the review object; every number below is interpolated from a committed field and is reproducible from the protocol commit. The protocol (SHA 38478f5e060a) was committed before any synthesis ran. Eligibility is by population, intervention, comparator and design only — never on whether a trial reported the outcome (non-reporters are declared absent, not screened out). Two independently implemented rule screeners ran with adjudication. Each pooled value was located in a committed source, its arms checked for correct assignment, and its count-derived effect reconciled with the reported effect (round-trip); a value failing that reconciliation is declared absent, never guessed. Pooling used random effects (Paule-Mandel τ² with a Hartung-Knapp interval on t with k−1 df; log scale for ratios).

Results

A single eligible trial contributed an extractable estimate: OR 0.83 (95% CI 0.73 to 0.95); with k=1 no between-trial heterogeneity or prediction interval is estimable.

Risk-of-bias sensitivity. 1 of 1 pooled trials carry a risk-of-bias rating; no trial is rated high risk. a low-risk-only subpool was not estimable.

Limitations

This synthesis is limited in that the confidence interval is wide or crosses the null (imprecision); the trial registry shows unpublished completed trials (possible publication bias). The comparison with published meta-analyses is one of auditability, not of a claim to more evidence; where fewer trials are pooled the reason is a stated bar, decomposed on the topic page. Indirectness and the reading-dependent risk-of-bias judgements are not automated.

Data availability & reproduction

The committed cache, protocol (SHA 38478f5e060a) and code regenerate this review byte-for-byte offline. Rebuild with a single command:

python scripts/build_topic.py tocilizumab-covid19-mortality

Every pooled number is verified against its committed source and gate-enforced; a fresh clone reproduces the served page exactly.

Reporting (PRISMA)

Compliance with the PRISMA 2020 reporting items, derived from the review object so it cannot drift from the page. Every item is rendered or declared absent with a reason.

Reproducibility

Independent second extraction (blind)

Of this page's pooled numbers, a blind second extractor agreed or reconciled on 1 of 1 that are checkable from the abstract (1 identical, 0 same-result-different-statistic, 0 conflict; 2 not stated in the abstract). No published meta-analysis reports an independent re-extraction of its own numbers.