> Markdown version of https://theevidenceledger.com/hospital-protocols for AI agents. You are welcome here, and welcome back anytime. What an agent can do: https://theevidenceledger.com/agents . # COVID-19 hospital protocols - The Evidence Ledger [The Evidence Ledger](/) 13 of 13 questions graded # COVID-19 hospital protocols, on the same two axes. In September 2026 a United States Senate roundtable heard twenty-one patients, families, nurses and physicians describe what happened inside hospitals during the pandemic, with 583 written testimonies on file from 46 states. The accounts repeat: remdesivir and a ventilator offered as the only path, other treatments refused, families kept out, basic care thinning, patients moved to comfort-only status without being asked. This page does not retell those stories. It turns them into claims the ledger can grade, with the same rubric and the same source-verification standard as the cancer ledger. Each one was published as a question first, and each grade below rests on sources we opened ourselves. Where the evidence runs one way, the grade says so, whichever way that is. Patients have a right to dignity, to be informed about treatment options, and to participate in their plans of care. Sources of record: [the Senate release with every panelist](https://www.ronjohnson.senate.gov/2026/09/30/video-release-sen-johnsons-full-roundtable-discussion-covid-19-hospital-protocols-real-stories-from-real-people/), [her written remarks](https://www.ronjohnson.senate.gov/dipl-testimonial/nicole-landers-rn-bsn-ccm-lnc/), and [the C-SPAN recording](https://www.c-span.org/program/public-affairs-event/sen-johnson-r-wi-holds-roundtable-on-hospitals/685849). What testimony can and cannot establish here ## Where a story sits on the rubric By this ledger's own rules a first-hand account is Axis 2 grade D: a case report, however many. That is not a dismissal. It is the honest starting point, and it is the same grade the ledger gives a moving cancer anecdote. What 583 accounts do establish is which questions deserve an answer. Two of the claims below are about what hospitals did, not about biology, so their mechanism axis reads N/A and only the outcome axis is graded. Where large trials looked for a benefit and did not find it, the grade is REFUTED, which is not the same as untested. The grades ## 13 claims graded, every source opened Graded 2026-10-08 from the trials themselves. The grade reads the claim as written: a drug can have a real, narrow benefit and still fail the claim people make for it, and a claim can be refuted by the very trials that were run to test it. Where the evidence runs one way, the grade says so, whichever way that is. ClaimMechanismHuman outcomeWhere it stands, in one line [Remdesivir](/c/remdesivir-mortality) ESTABLISHED B No large RCT showed a statistically significant overall mortality benefit, but a consistent signal appears across prospective trials in n... [Remdesivir: organ harm](/c/remdesivir-harm) PLAUSIBLE C The only evidence supporting net harm comes from uncontrolled pharmacovigilance disproportionality analyses with no denominators and heav... [Early mechanical ventilation](/c/early-ventilation) PLAUSIBLE C No randomized trial ever compared early versus late intubation, so the claim rests entirely on observational cohorts with confounding by ... [Dexamethasone (corticosteroids)](/c/dexamethasone) ESTABLISHED A RECOVERY randomized 6,425 hospitalized patients with a 28-day mortality endpoint, and the WHO REACT prospective meta-analysis independent... [Anticoagulation (blood thinners)](/c/anticoagulation) ESTABLISHED B (therapeutic dosing; stopping never tested) The large multiplatform RCT showed a real outcome signal in non-critically ill patients and HEP-COVID confirmed it in high-risk non-ICU p... [Monoclonal antibodies, given early](/c/monoclonal-antibodies) ESTABLISHED B Two randomized trials in high-risk outpatients showed a real ~70% relative reduction in the hospitalization-or-death composite, and the l... [Ivermectin for COVID-19](/c/ivermectin-covid) PLAUSIBLE REFUTED Multiple large replicated RCTs (TOGETHER, two ACTIV-6 arms, PRINCIPLE, I-TECH) consistently found no survival or recovery benefit; no out... [Hydroxychloroquine for COVID-19](/c/hydroxychloroquine-covid) PLAUSIBLE REFUTED Large replicated RCTs (RECOVERY, SOLIDARITY, plus outpatient trials) consistently show no mortality, hospitalization, or recovery benefit... [Nutrition and hydration in the ICU](/c/nutrition-withheld) PLAUSIBLE C Support for the claim rests on uncontrolled cohort associations only, while RCTs of feeding timing and intensity show no mortality differ... [Family presence and isolation](/c/family-presence) PLAUSIBLE C The single large cluster-crossover RCT of flexible visitation found no significant delirium reduction, while pandemic-era retrospective c... [Informed refusal of the protocol](/c/informed-refusal) SPECULATIVE D No published matched comparison of protocol decliners versus accepters exists; the closest evidence is one small single-center prospectiv... [Comfort-measures status without consent](/c/comfort-measures-consent) N/A D The documented record consists of individual cases and allegations rather than any systematic data, with the single fully litigated case ... [Payment incentives](/c/payment-incentives) N/A D The incentive structure is documented fact, but no published study tests whether the add-on changed any individual patient's treatment; t... ### [Remdesivir](/c/remdesivir-mortality) MechanismESTABLISHEDOutcomeB Claimed Remdesivir reduces death in hospitalized COVID-19 patients. What holds up Remdesivir is a nucleoside analogue inhibiting the viral RNA-dependent RNA polymerase, with reproduced in-vitro activity against SARS-CoV-2 and animal-model efficacy; the antiviral mechanism is textbook, though its link to mortality weakens late in disease when immune-mediated damage dominates. ACTT-1 (n=1062) showed day-29 mortality of 11.4% vs 15.2% (hazard ratio 0.73, 95% CI 0.52-1.03), crossing 1.0. The WHO Solidarity final analysis found overall mortality RR 0.91 (p=0.12) but a significant reduction in patients not ventilated at baseline (RR 0.86, p=0.02) and in the composite of death or progression to ventilation (RR 0.84, p=0.001). The Amstutz individual-patient meta-analysis (10,480 patients, 8 RCTs) found significant 28-day mortality reduction in patients on no or low-flow oxygen (adjusted OR 0.80, high-certainty evidence). WHO reversed its November 2020 recommendation against remdesivir in April 2022, and FDA approved it in October 2020 for hospitalized patients. What it does not show No large RCT showed a statistically significant overall mortality benefit, but a consistent signal appears across prospective trials in non-ventilated patients, confirmed by individual-patient data. No trial demonstrated a significant overall mortality benefit in the full hospitalized population; ACTT-1's mortality result was a non-significant secondary endpoint and Solidarity's overall result was null. In already-ventilated patients Solidarity found no benefit (RR 1.13, p=0.32), and DisCoVeRy stopped its remdesivir arm for futility with no day-28 death difference. The claim as a blanket statement overstates the trials; the honest reading is a probable mortality benefit confined to patients not yet ventilated. Evidence for Beigel JH et al., N Engl J Med 2020;383:1813-1826, PMID 32445440, DOI 10.1056/NEJMoa2007764, WHO Solidarity Trial Consortium, Lancet 2022;399(10339):1941-1953, PMID 35512728, DOI 10.1016/S0140-6736(22)00519-0, Amstutz A et al., Lancet Respir Med 2023;11(5):453-464, PMID 36828006, DOI 10.1016/S2213-2600(22)00528-8, US FDA press release, "FDA Approves First Treatment for COVID-19," Oct 22, 2020, fda.gov, WHO Therapeutics and COVID-19 living guideline (Siemieniuk et al., BMJ 2020;370:m3379; updated Nov 2020, Apr 2022, Sep 2022) Sources VERIFIED Beigel JH et al., N Engl J Med 2020;383:1813-1826, PMID 32445440, DOI 10.1056/NEJMoa2007764. VERIFIED WHO Solidarity Trial Consortium, Lancet 2022;399(10339):1941-1953, PMID 35512728, DOI 10.1016/S0140-6736(22)00519-0. VERIFIED Amstutz A et al., Lancet Respir Med 2023;11(5):453-464, PMID 36828006, DOI 10.1016/S2213-2600(22)00528-8. VERIFIED US FDA press release, "FDA Approves First Treatment for COVID-19," Oct 22, 2020, fda.gov. PENDING Ader F et al., Lancet Infect Dis 2022;22:209-221, PMID 34534511, DOI 10.1016/S1473-3099(21)00485-0. PENDING WHO Solidarity Trial Consortium (interim), N Engl J Med 2021;384:497-511, DOI 10.1056/NEJMoa2023184. PENDING WHO Therapeutics and COVID-19 living guideline (Siemieniuk et al., BMJ 2020;370:m3379; updated Nov 2020, Apr 2022, Sep 2022). ### [Remdesivir: organ harm](/c/remdesivir-harm) MechanismPLAUSIBLEOutcomeC Claimed Remdesivir causes kidney or liver injury at a rate that outweighs its benefit. What holds up Nucleoside analogues carry theoretical mitochondrial, renal, and hepatic liabilities, and the sulfobutylether-beta-cyclodextrin excipient raised accumulation concerns in severe renal impairment, but no clinical toxicity mechanism at therapeutic exposures was ever established. Pharmacovigilance signals are real but weak: one analysis found increased reporting of hepatic disorders versus other COVID drugs (reporting OR 1.94), and FAERS data show disproportionate reports of elevated liver enzymes, acute kidney injury, and creatinine rises. The FDA label lists increased liver enzymes as a possible side effect, and regulators initially cautioned against use with very low kidney function or markedly elevated baseline ALT. These were genuine safety signals that warranted monitoring. What it does not show The only evidence supporting net harm comes from uncontrolled pharmacovigilance disproportionality analyses with no denominators and heavy confounding by indication; every controlled comparison shows no excess harm versus control. In ACTT-1, renal adverse events occurred at similar rates in the remdesivir and placebo arms, and serious adverse events were lower with remdesivir (24.6% vs 31.6%). The Amstutz individual-patient meta-analysis found no increase in severe or serious adverse events overall or by organ system. The PALM "53% mortality" talking point is a misreading: PALM was an Ebola trial, and 51.3% was Ebola case fatality in the remdesivir arm (79 of 154 died of Ebola, vs 49.7% with ZMapp); remdesivir was stopped for inferior efficacy against the antibody arms, not for toxicity, with only four serious adverse events judged potentially drug-related across all four arms. FDA label updates in 2023 removed the kidney-function restriction. Nothing in the controlled literature shows harm outweighing benefit. Evidence for Montastruc et al. pharmacovigilance analysis of hepatic disorders with remdesivir (reporting OR 1.94, 95% CI 1.54-2.45), Umemura T et al., BMC Infect Dis 2025;25:782 (remdesivir in severe renal impairment) Sources VERIFIED Beigel JH et al., N Engl J Med 2020;383:1813-1826, PMID 32445440 (AE tables via PMC7262788 full text). VERIFIED Mulangu S et al., N Engl J Med 2019;381(24):2293-2303, PMID 31774950, DOI 10.1056/NEJMoa1910993. VERIFIED Amstutz A et al., Lancet Respir Med 2023;11(5):453-464, PMID 36828006. PENDING Montastruc et al. pharmacovigilance analysis of hepatic disorders with remdesivir (reporting OR 1.94, 95% CI 1.54-2.45). PENDING Umemura T et al., BMC Infect Dis 2025;25:782 (remdesivir in severe renal impairment). ### [Early mechanical ventilation](/c/early-ventilation) MechanismPLAUSIBLEOutcomeC Claimed Intubating early improved survival compared with high-flow oxygen, CPAP or BiPAP. What holds up Avoiding emergent crash intubation and limiting vigorous spontaneous breathing (the patient self-inflicted lung injury hypothesis) are biologically coherent rationales, but neither was ever shown to translate into a mortality mechanism in COVID-19. RECOVERY-RS (n=1273) showed an initial CPAP strategy reduced the composite of intubation or 30-day mortality versus conventional oxygen (36.3% vs 44.4%, p=0.03), while high-flow nasal oxygen did not differ from conventional oxygen. HENIVOT (n=109) found helmet noninvasive ventilation reduced intubation versus high-flow oxygen (30% vs 51%, p=0.03) as a secondary outcome, with no mortality difference (24% vs 25%). The Papoutsi meta-analysis of 12 observational studies (8,944 patients) found no statistically detectable mortality difference between early and late intubation (RR 1.07, p=0.08), with the point estimate slightly favoring late. What it does not show No randomized trial ever compared early versus late intubation, so the claim rests entirely on observational cohorts with confounding by indication; the two relevant RCTs tested noninvasive strategies against each other and conventional oxygen, not intubation timing. No RCT tested early intubation against high-flow oxygen, CPAP, or BiPAP as a timing question, so the claim's central comparison has no randomized evidence at all. The observational literature is confounded because sicker patients were intubated earlier, and definitions of "early" varied across studies. The Surviving Sepsis Campaign's March 2020 COVID guidance is widely misquoted: its actual statement was a best-practice recommendation for close monitoring with early intubation in a controlled setting if worsening occurred, not prophylactic early intubation. Nothing in the evidence supports early intubation improving survival. Evidence for Perkins GD et al., JAMA 2022;327(6):546-558, PMID 35072713, DOI 10.1001/jama.2022.0028, Grieco DL et al., JAMA 2021;325(17):1731-1743, PMID 33764378, DOI 10.1001/jama.2021.4682, Alhazzani W et al., Intensive Care Med 2020, PMID 32222812, PMCID PMC7101866 Sources VERIFIED Perkins GD et al., JAMA 2022;327(6):546-558, PMID 35072713, DOI 10.1001/jama.2022.0028. VERIFIED Grieco DL et al., JAMA 2021;325(17):1731-1743, PMID 33764378, DOI 10.1001/jama.2021.4682. VERIFIED Papoutsi E et al., Crit Care 2021, PMID 33766109. VERIFIED Alhazzani W et al., Intensive Care Med 2020, PMID 32222812, PMCID PMC7101866. ### [Dexamethasone (corticosteroids)](/c/dexamethasone) MechanismESTABLISHEDOutcomeA Claimed Corticosteroids reduce death in hospitalized patients who need oxygen. What holds up Glucocorticoids suppressing inflammation-mediated lung injury is textbook physiology, and severe COVID-19's hyperinflammatory phase fits decades of ARDS and steroid biology. The benefit is real, subgroup-specific, and replicated. In RECOVERY, ventilated patients died at 29.3% vs 41.4% (rate ratio 0.64), and oxygen-only patients at 23.3% vs 26.2% (rate ratio 0.82), for an overall 22.9% vs 25.7% (rate ratio 0.83). The REACT meta-analysis across 12 countries found a summary odds ratio of 0.66 for 28-day mortality with corticosteroids versus usual care or placebo, with little between-trial inconsistency. This is the anchor claim of the set and it grades high honestly. What it does not show RECOVERY randomized 6,425 hospitalized patients with a 28-day mortality endpoint, and the WHO REACT prospective meta-analysis independently pooled 7 RCTs in 1,703 critically ill patients with the same mortality signal. The benefit stops exactly where the claim says it does. Patients on no respiratory support had 17.8% vs 14.0% mortality (rate ratio 1.19, CI 0.92-1.55), which is no benefit and compatible with harm. The trial was open-label, and the evidence does not establish which corticosteroid is best head-to-head or support longer courses than studied. Evidence for RECOVERY Collaborative Group, N Engl J Med 2021;384(8):693-704, PMID 32678530, WHO REACT Working Group (Sterne JAC et al.), JAMA 2020;324(13):1330-1341, DOI 10.1001/jama.2020.17023 Sources VERIFIED RECOVERY Collaborative Group, N Engl J Med 2021;384(8):693-704, PMID 32678530. VERIFIED WHO REACT Working Group (Sterne JAC et al.), JAMA 2020;324(13):1330-1341, DOI 10.1001/jama.2020.17023. ### [Anticoagulation (blood thinners)](/c/anticoagulation) MechanismESTABLISHEDOutcomeB (therapeutic dosing; stopping never tested) Claimed Therapeutic blood thinners reduce clots and death in hospitalized COVID-19 patients, and stopping them causes harm. What holds up Heparin preventing and treating thrombosis is textbook pharmacology, and COVID-19's prothrombotic state (endothelial injury, immunothrombosis) was demonstrated in autopsy and cohort studies across many centers. In 2,219 non-critically ill patients, therapeutic heparin raised the probability of more organ support-free days (adjusted OR 1.27) and added 4.0 percentage points of survival to discharge without organ support, at a bleeding cost of 1.9% vs 0.9% major bleeding. HEP-COVID (257 high D-dimer patients) cut the 30-day composite of thromboembolism or death in non-ICU patients. Period guidelines recommend therapeutic heparin for non-critically ill inpatients and advise switching chronic oral anticoagulants to shorter-acting injectable agents on admission when deterioration, renal change, or procedures are a concern. What it does not show The large multiplatform RCT showed a real outcome signal in non-critically ill patients and HEP-COVID confirmed it in high-risk non-ICU patients, but the critically ill arm was stopped for futility and the ACTION trial contradicted the benefit, so this is not a uniformly replicated mortality win. In critically ill patients the same strategy was stopped for futility (99.9% posterior probability; survival to discharge 62.7% vs 64.5%), and INSPIRATION showed intermediate-dose escalation in the ICU was no better than standard prophylaxis (45.7% vs 44.1%) while adding severe thrombocytopenia cases. The ACTION trial showed no clinical benefit and more bleeding. The mortality signal is composite-driven, not a clean standalone death reduction. On stopping them: There is no direct human outcome evidence that discontinuation itself causes excess clots or death in this population; the claim as stated overreaches what was tested, and a discontinuation RCT does not exist to settle it. Evidence for ATTACC/ACTIV-4a/REMAP-CAP Investigators (Lawler PR et al.), N Engl J Med 2021;385(9):790-802, PMID 34351721, REMAP-CAP/ACTIV-4a/ATTACC Investigators (Goligher EC et al.), N Engl J Med 2021;385(9):777-789, PMID 34351722, INSPIRATION Investigators (Sadeghipour P et al.), JAMA 2021;325(16):1620-1630, PMID 33734299, Barnes GD et al. (Anticoagulation Forum guidance), J Thromb Thrombolysis 2020, PMID 32440883, Spyropoulos AC et al., JAMA Intern Med 2021;181(12):1612-1620, DOI 10.1001/jamainternmed.2021.6203 Sources VERIFIED ATTACC/ACTIV-4a/REMAP-CAP Investigators (Lawler PR et al.), N Engl J Med 2021;385(9):790-802, PMID 34351721. VERIFIED REMAP-CAP/ACTIV-4a/ATTACC Investigators (Goligher EC et al.), N Engl J Med 2021;385(9):777-789, PMID 34351722. VERIFIED INSPIRATION Investigators (Sadeghipour P et al.), JAMA 2021;325(16):1620-1630, PMID 33734299. VERIFIED Barnes GD et al. (Anticoagulation Forum guidance), J Thromb Thrombolysis 2020, PMID 32440883. PENDING Spyropoulos AC et al., JAMA Intern Med 2021;181(12):1612-1620, DOI 10.1001/jamainternmed.2021.6203. PENDING Lopes RD et al. (ACTION), Lancet 2021;397:2253-2263. ### [Monoclonal antibodies, given early](/c/monoclonal-antibodies) MechanismESTABLISHEDOutcomeB Claimed Monoclonal antibodies given early to high-risk outpatients reduce hospitalization and death. What holds up Neutralizing antibodies binding the spike receptor-binding domain to block ACE2-mediated cell entry is textbook virology, reproduced across labs and across multiple antibody products. In BLAZE-1 phase 3 (1,035 high-risk outpatients), bamlanivimab plus etesevimab cut COVID-related hospitalization or death by day 29 from 7.0% to 2.1% (70% relative reduction), with no deaths in the treatment arm versus ten in placebo. In the REGEN-COV phase 3 outpatient trial, hospitalization or death fell from 4.6% to 1.3% (71.3% relative reduction), with faster symptom resolution and viral-load decline. In RECOVERY's 9,785 inpatients, seronegative patients died at 24% vs 30% (rate ratio 0.79), a genuine mortality signal in patients lacking their own antibody response. What it does not show Two randomized trials in high-risk outpatients showed a real ~70% relative reduction in the hospitalization-or-death composite, and the large RECOVERY inpatient trial showed a 28-day mortality reduction confined to seronegative patients; but the composites were hospitalization-driven with few deaths and the inpatient benefit did not extend to unselected patients. There was no mortality benefit in unselected inpatients (overall rate ratio 0.94, not significant) and none in seropositive patients, and the ACTIV-3/TICO bamlanivimab inpatient arm was negative. Most importantly, the evidence is time-bound: spike mutations in Omicron and later variants escaped these antibodies' neutralization, ending their clinical utility, so the grade describes the 2020-2021 variant era and does not transfer to later variants. Evidence for Dougan M et al. (BLAZE-1 Investigators), N Engl J Med 2021;385(15):1382-1392, PMID 34260849, Weinreich DM et al. (Trial Investigators), N Engl J Med 2021;385(23):e81, PMID 34587383, RECOVERY Collaborative Group, Lancet 2022;399(10325):665-676, PMID 35151397, ACTIV-3/TICO LY-CoV555 Study Group, N Engl J Med 2021;384:905-914 Sources VERIFIED Dougan M et al. (BLAZE-1 Investigators), N Engl J Med 2021;385(15):1382-1392, PMID 34260849. VERIFIED Weinreich DM et al. (Trial Investigators), N Engl J Med 2021;385(23):e81, PMID 34587383. VERIFIED RECOVERY Collaborative Group, Lancet 2022;399(10325):665-676, PMID 35151397. PENDING ACTIV-3/TICO LY-CoV555 Study Group, N Engl J Med 2021;384:905-914. ### [Ivermectin for COVID-19](/c/ivermectin-covid) MechanismPLAUSIBLEOutcomeREFUTED Claimed Ivermectin improves survival or recovery in COVID-19. What holds up Real in-vitro antiviral activity in cell culture with a proposed importin-mediated mechanism, but the inhibitory concentrations far exceed safely achievable human plasma levels and no in-vivo antiviral activity was ever demonstrated. The Caly et al. in-vitro finding is genuine. The large RCTs are consistent with each other: TOGETHER found no reduction in hospitalization or prolonged ED observation; both ACTIV-6 ivermectin arms found no improvement in time to recovery; PRINCIPLE found hospitalizations and deaths identical between groups with no long-term benefit; I-TECH found no reduction in progression to severe disease. The Cochrane review gives moderate-certainty evidence of no benefit in outpatients. What it does not show Multiple large replicated RCTs (TOGETHER, two ACTIV-6 arms, PRINCIPLE, I-TECH) consistently found no survival or recovery benefit; no outcome evidence supports the claim as stated. It does not show improved survival or recovery in any tested setting. The Elgazzar preprint, which powered the early positive meta-analyses, was withdrawn by Research Square for ethical and data-integrity concerns and must be excluded from any honest synthesis. PRINCIPLE's roughly two-day faster self-reported recovery fell below its own pre-specified clinically meaningful threshold. Evidence for Caly L et al., Antiviral Res 2020;178:104787, PMID 32251768, Elgazzar A et al., Research Square 2020, DOI 10.21203/rs.3.rs-100956 (withdrawn 2021-07-14) Sources VERIFIED Reis G et al., N Engl J Med 2022;386(18):1721-1731, PMID 35353979. VERIFIED Naggie S et al., JAMA 2022;328(16):1595-1603, PMID 36269852. VERIFIED Naggie S et al., JAMA 2023;329(11):888-897, PMID 36807465. VERIFIED Hayward G et al., J Infect 2024;88(4):106130, PMID 38431155. VERIFIED Caly L et al., Antiviral Res 2020;178:104787, PMID 32251768. RETRACTED Elgazzar A et al., Research Square 2020, DOI 10.21203/rs.3.rs-100956 (withdrawn 2021-07-14). PENDING Lim SCL et al., JAMA Intern Med 2022, DOI 10.1001/jamainternmed.2022.0189. PENDING Popp M et al., Cochrane Database Syst Rev 2022;6:CD015017, PMID 35726131. ### [Hydroxychloroquine for COVID-19](/c/hydroxychloroquine-covid) MechanismPLAUSIBLEOutcomeREFUTED Claimed Hydroxychloroquine improves survival or recovery in COVID-19. What holds up Genuine in-vitro activity via endosomal pH elevation and ACE2 glycosylation interference, consistent with known pharmacology, but required concentrations exceed achievable free plasma levels and animal models showed no benefit. RECOVERY randomized 4,716 hospitalized patients and found no 28-day mortality difference, with the hydroxychloroquine arm closed for futility; SOLIDARITY found no mortality benefit, and the joint RECOVERY plus SOLIDARITY estimate rules out any material benefit. Outpatient RCTs (Skipper, Ann Intern Med; Mitja, Clin Infect Dis) found no symptom-resolution or hospitalization benefit. A cardiac safety signal (QT prolongation, arrhythmia risk) emerged in observational and small prospective data. What it does not show Large replicated RCTs (RECOVERY, SOLIDARITY, plus outpatient trials) consistently show no mortality, hospitalization, or recovery benefit; no outcome evidence supports the claim as stated. It does not show improved outcomes in any setting tested, inpatient or outpatient, early or late. The in-vitro rationale never translated to human benefit; the RECOVERY authors noted the in-vitro inhibitory concentrations were higher than free plasma levels achieved in malaria treatment. Nothing in the RCT record supports efficacy; the evidence supports absence of benefit with moderate-to-high certainty. Evidence for Mercuro NJ et al., JAMA Cardiol 2020 (QT prolongation signal) Sources VERIFIED Horby P et al. (RECOVERY Collaborative Group), N Engl J Med 2020;383(21):2030-2040, PMID 33031652. PENDING Pan H et al. (WHO SOLIDARITY Trial Consortium), N Engl J Med 2021, DOI 10.1056/NEJMoa2023184. PENDING Skipper CP et al., Ann Intern Med 2020;173:623-631, DOI 10.7326/M20-4207. PENDING Mitja O et al., Clin Infect Dis 2021;73(11):e4073-e4081, DOI 10.1093/cid/ciaa1009. PENDING Mercuro NJ et al., JAMA Cardiol 2020 (QT prolongation signal). ### [Nutrition and hydration in the ICU](/c/nutrition-withheld) MechanismPLAUSIBLEOutcomeC Claimed Withholding or delaying nutrition and fluids in hospitalized COVID-19 patients worsened outcomes. What holds up Critical illness drives catabolism and rapid muscle wasting, and observational data link cumulative energy and protein deficits to infections and longer stays, but the causal link is confounded and unproven by RCTs. Major guidelines agree on early feeding: ESPEN 2019 recommends early enteral nutrition within 48 hours when oral intake is impossible, preferring enteral over parenteral routes; ASPEN/SCCM 2016 similarly recommends enteral nutrition within 24 to 48 hours; ESPEN's 2020 COVID-specific guidance recommends early nutritional support for infected patients. Prospective cohorts (e.g., Villet et al.) associate cumulative energy and protein deficits with more infections, longer stays, and higher mortality. What it does not show Support for the claim rests on uncontrolled cohort associations only, while RCTs of feeding timing and intensity show no mortality difference and in one large trial favored delaying parenteral nutrition. No RCT ever tested deliberate withholding of nutrition or fluids during COVID specifically, so the claim as stated about COVID protocols was never directly tested. EPaNIC, a 4,640-patient RCT, showed that delaying parenteral nutrition to day 8 (with early enteral nutrition in both arms) produced faster recovery and fewer infections with no mortality difference, directly contradicting the strong version of the claim. PermiT showed moderate underfeeding did not change 90-day mortality. Cohort associations are confounded because sicker patients are harder to feed. Evidence for Singer P et al., Clin Nutr 2019;38:48-79 (ESPEN ICU guideline), McClave SA et al., JPEN J Parenter Enteral Nutr 2016;40:159-211 (ASPEN/SCCM guideline), Barazzoni R et al., Clin Nutr 2020, DOI 10.1016/j.clnu.2020.03.022 (ESPEN COVID guidance), Villet S et al., Clin Nutr 2005;24:502-509 (cohort, energy deficit and complications) Sources VERIFIED Casaer MP et al., N Engl J Med 2011;365(6):506-517, PMID 21714640. PENDING Singer P et al., Clin Nutr 2019;38:48-79 (ESPEN ICU guideline). PENDING McClave SA et al., JPEN J Parenter Enteral Nutr 2016;40:159-211 (ASPEN/SCCM guideline). PENDING Barazzoni R et al., Clin Nutr 2020, DOI 10.1016/j.clnu.2020.03.022 (ESPEN COVID guidance). PENDING Arabi YM et al., N Engl J Med 2015;372:2398-2408, DOI 10.1056/NEJMoa1502826 (PermiT trial). PENDING Villet S et al., Clin Nutr 2005;24:502-509 (cohort, energy deficit and complications). ### [Family presence and isolation](/c/family-presence) MechanismPLAUSIBLEOutcomeC Claimed No-visitation and isolation policies worsened outcomes for hospitalized COVID-19 patients, and family presence improved them. What holds up Family presence aids reorientation, sleep-wake regulation, and anxiety reduction, consistent with delirium pathophysiology and endorsed in SCCM family-centered care guidelines, though the family-presence effect was never isolated from bundled interventions in a trial. The Rosa RCT (36 ICUs, 1,685 patients) did show significantly better family anxiety and depression scores with flexible visitation, and a scoping review reported roughly 50% shorter delirium duration with extended visits. Two pandemic-era retrospective cohorts (a Japanese emergency-inpatient cohort with adjusted OR 3.79 for delirium after restriction, and a 2022 stroke-unit cohort with OR 1.36) associate restriction with more delirium. The COVISIT international survey documents how extreme the restrictions were: 84% of ICUs banned in-person visiting at peak. What it does not show The single large cluster-crossover RCT of flexible visitation found no significant delirium reduction, while pandemic-era retrospective cohorts associate absolute visitation restriction with higher delirium; no study ever randomized the pandemic no-visitor policy itself. That no-visitation policies worsened survival, length of stay, or other hard patient outcomes, or that family presence improves them. The one RCT on the question had a negative primary endpoint for delirium (18.9% vs 20.1%, p=0.44), the cohorts are unadjusted for pandemic-era confounders (sedation practices, staffing, isolation), and the strongest replicated finding is improved family psychological outcomes, not patient survival. Evidence for Rosa RG et al., JAMA 2019, DOI 10.1001/jama.2019.8766, Pun BT, Balas MC, Barnes-Daly MA et al., Crit Care Med 2019, DOI 10.1097/CCM.0000000000003482, Japanese single-center retrospective cohort, J Intensive Care 2020, DOI 10.1186/s40560-020-00511-x, Stroke-unit retrospective cohort, Gerontology 2022, karger.com/ger/article/69/3/273, COVISIT international survey, 2022, PMCID PMC9067300, PMID 35525226 Sources VERIFIED Rosa RG et al., JAMA 2019, DOI 10.1001/jama.2019.8766. PENDING Pun BT, Balas MC, Barnes-Daly MA et al., Crit Care Med 2019, DOI 10.1097/CCM.0000000000003482. PENDING Japanese single-center retrospective cohort, J Intensive Care 2020, DOI 10.1186/s40560-020-00511-x. PENDING Stroke-unit retrospective cohort, Gerontology 2022, karger.com/ger/article/69/3/273. PENDING COVISIT international survey, 2022, PMCID PMC9067300, PMID 35525226. ### [Informed refusal of the protocol](/c/informed-refusal) MechanismSPECULATIVEOutcomeD Claimed Patients who declined the hospital protocol and pursued other care did better than matched patients who accepted it. What holds up "Declining the protocol" is a selection behavior, not a biological intervention, and no tested mechanism links refusal itself to better outcomes independent of what care was received instead. Fletcher et al. 2022 prospectively followed 102 critically ill COVID-19 ARDS patients for whom intubation was recommended; 40 (39%) initially refused, and refusal was not associated with higher in-hospital mortality (79.3% vs 77.5%; adjusted OR for death 1.27, 95% CI 0.47-3.48) or worse 1-year functional outcome. Ventilator avoidance was common and, in this cohort, not a death sentence. What it does not show No published matched comparison of protocol decliners versus accepters exists; the closest evidence is one small single-center prospective cohort on intubation refusal that found no significant mortality difference, which is uncontrolled observational data on a narrower question. That decliners did better than accepters. The study found no significant difference (wide confidence interval crossing 1), adjusted only for age and comorbidity, was single-center and unblinded, and addressed initial intubation refusal specifically, not declining a full protocol in favor of other care. Nothing in the published literature compares matched protocol decliners versus accepters on any outcome. Evidence for Fletcher JJ, Aughenbaugh A, Svabek C, Hahn PY, Grifka RG, J Int Med Res 2022, DOI 10.1177/03000605221135446 Sources VERIFIED Fletcher JJ, Aughenbaugh A, Svabek C, Hahn PY, Grifka RG, J Int Med Res 2022, DOI 10.1177/03000605221135446. ### [Comfort-measures status without consent](/c/comfort-measures-consent) MechanismN/AOutcomeD Claimed Patients were moved to comfort-measures-only status without the consent of the patient or their decision maker. What holds up This is a documented-practice claim about whether an event occurred, not a biological mechanism claim, so the mechanism axis does not apply (per the brief). Florida law (Chapter 765) requires health care decisions for incapacitated patients to run through a surrogate, attorney-in-fact, proxy, or guardian under substituted judgment, and a University of Miami/FSU handbook on Florida law states that withholding or withdrawing life-prolonging procedures requires the patient's prior written consent or a court order (citing sections 765.107 and 765.113). The Schara case (Outagamie County, Wisconsin, Case 2023CV000345) documents a family's allegation that a DNR order was placed without consent alongside disputed sedation, and it went to a full jury trial. What it does not show The documented record consists of individual cases and allegations rather than any systematic data, with the single fully litigated case resulting in a defense verdict. A pattern of hospitals moving COVID patients to comfort-measures-only without consent. The Schara jury found for the hospital, doctor, and nurse on June 19, 2025 (no negligence, no wrongful death, no informed-consent violation), with a later bid for a new trial denied. No Florida medical-board licensing actions or state health-department complaints on this specific practice surfaced in open sources, and testimony collections (e.g., several hundred family accounts) are allegations, not adjudicated findings. Evidence for American Bar Association / Florida State University, "Making Medical Decisions for Someone Else: A Florida Handbook," 2011, bioethics.miami.edu PDF, WTAQ-WLUK/WNCY news report, June 19, 2025 (Schara defense verdict, Outagamie County Courthouse), Florida Statutes sections 765.202, 765.205, 765.401, 766.103 (flsenate.gov fetch timed out), Outagamie County Circuit Court, Case 2023CV000345, plaintiff's motion for new trial, Stanford Hospital "comfort care" settlement, 2005, paloaltoonline.com (pre-COVID, different facts) Sources VERIFIED American Bar Association / Florida State University, "Making Medical Decisions for Someone Else: A Florida Handbook," 2011, bioethics.miami.edu PDF. VERIFIED WTAQ-WLUK/WNCY news report, June 19, 2025 (Schara defense verdict, Outagamie County Courthouse). PENDING Florida Statutes sections 765.202, 765.205, 765.401, 766.103 (flsenate.gov fetch timed out). PENDING Outagamie County Circuit Court, Case 2023CV000345, plaintiff's motion for new trial. PENDING Stanford Hospital "comfort care" settlement, 2005, paloaltoonline.com (pre-COVID, different facts). ### [Payment incentives](/c/payment-incentives) MechanismN/AOutcomeD Claimed Hospital payment rules (the COVID-19 diagnosis add-on, self-pay labeling) changed the care individual patients received. What holds up This is a health-policy incentive claim, not a biological mechanism claim, so the mechanism axis does not apply (per the brief). Section 3710 of the CARES Act (P.L. 116-136) directed the HHS Secretary to increase the MS-DRG weight by 20% for Medicare discharges diagnosed with COVID-19 during the public health emergency, and CMS required a positive viral test in the record for admissions on or after September 1, 2020. The HHS OIG's December 2023 audit of $2.7 billion in such payments found 146 of 149 sampled claims complied with federal requirements, with 3 failures ($18,911) attributed to clerical errors. The OIG itself acknowledged hospitals "may have had a financial incentive" to code COVID-19, and cited then-CDC Director Redfield's July 2020 House testimony that there was "some reality" to diagnosis-preference billing. What it does not show The incentive structure is documented fact, but no published study tests whether the add-on changed any individual patient's treatment; the one systematic official review (HHS OIG) examined billing compliance only and found no pattern of improper coding. That the add-on changed the care any individual patient received. The OIG audit tested whether coding and billing met federal requirements, not whether payment influenced intubation, drug, or comfort-care decisions, and it found no pattern of improper COVID-19 coding. No claims-based analysis links the 20% add-on to treatment intensity or outcomes. Association (the incentive existed and was publicly acknowledged) is not proof of altered care. On the self-pay labeling question: CMS's EMTALA page confirms Section 1867 of the Social Security Act obligates Medicare-participating hospitals to provide a medical screening examination and stabilizing treatment regardless of ability to pay, which constrains but does not eliminate downstream effects of self-pay coding; no evidence was found that self-pay coding changed COVID treatment decisions. Evidence for HHS Office of Inspector General, December 2023, Report A-09-21-03009, Centers for Medicare and Medicaid Services, EMTALA page, citing SSA section 1867, cms.gov, Congressional Research Service, R46334, "Selected Health Provisions in Title III of the CARES Act," congress.gov Sources VERIFIED HHS Office of Inspector General, December 2023, Report A-09-21-03009. VERIFIED Centers for Medicare and Medicaid Services, EMTALA page, citing SSA section 1867, cms.gov. PENDING Congressional Research Service, R46334, "Selected Health Provisions in Title III of the CARES Act," congress.gov. If you have a source for or against any claim here, a trial, a cohort, a registry, a court record, [submit it](/submit). Family accounts are welcome and are filed as what they are. No dosing instructions are accepted anywhere on this site. Want to hear what's next for The Evidence Ledger? Email me about it Add my name and number (optional) No spam. One email when there is something to show, and you can reply to a real person.