Effect of subcutaneous lidocaine–hydroxypropyl-β-cyclodextrin on quality of life in patients with post-COVID condition

Executive summary

This observational study of 103 patients examined subcutaneous lidocaine–hydroxypropyl-β-cyclodextrin (lidocaine-HP-β-CD) for treating post-COVID symptoms. The intervention is biologically plausible and the study documents improvement in quality-of-life metrics. But the methodology has flaws that prevent any meaningful conclusion about treatment efficacy. The lack of a control group, the high risk of bias from unblinded treatment, the absence of objective outcome measures, and significant commercial conflicts of interest together undermine the study's validity.

Study design

Design type — critical flaw. An interrupted time series with patients as their own controls. There is no control group whatsoever. That makes it impossible to separate a treatment effect from natural recovery, regression to the mean, placebo effect or temporal trends. This is particularly problematic for a condition with a known variable trajectory and spontaneous improvement. The average illness duration of 2.5 years means some patients may have been entering a natural recovery phase.

Blinding — critical flaw. Completely unblinded. Patients knew they were receiving treatment, which creates large potential for placebo and expectation effects, response bias on subjective measures, and performance bias. There was no masking of outcome assessors.

Sample selection

Post-COVID diagnostic criteria — major concern. Duration of at least 12 weeks post-infection is adequate. But COVID confirmation is not specified in the available materials, and hospitalisation status is not clearly specified. An average duration of 2.5 years suggests a moderate-to-severe cohort.

Symptom characterisation — moderate concern. The study mentions post-exertional malaise (PEM) and other ME/CFS-like symptoms, but there is no clear statement about PEM assessment methodology and no use of validated PEM questionnaires such as the DePaul Symptom Questionnaire (DSQ-2). The proportion with ME/CFS overlap is unknown, and there is no characterisation of postural orthostatic tachycardia syndrome or orthostatic intolerance (POTS/OI) or mast cell activation syndrome (MCAS) comorbidities.

Sample characteristics. n=103, adequate for an observational study but insufficient without controls. Mean age 48. Mean duration 2.5 years, so a very chronic cohort. The claim that patients had "no improvement" prior to treatment is impossible to verify.

Outcome measures — critical flaw

Physical Component Score of the 12-item Short Form Health Survey (SF-12). Small effect at 24 weeks, improving from about 30 to about 33 points. This is almost entirely subjective, and so highly susceptible to expectation and placebo effects. Performance bias applies too: patients may try harder at activities because they expect to do better.

SF-12 Mental Component Score. Moderate effect at 24 weeks, improving from about 32 to about 38-39 points. Even more subjective than the physical measure. Mental health improvements can follow from hope, attention and belief in a treatment.

Symptom burden. Improvements across multiple domains — PEM, dyspnea, fatigue, cognitive problems, pain. All patient-reported, all subjective, none validated through objective measures, and all at high risk of reporting bias in an unblinded study.

What is missing. No blood biomarkers, no cardiopulmonary exercise testing, no tilt table testing for POTS/OI, no imaging, no actigraphy, no lean test — no physiological measure that cannot be influenced by effort or perception. This is hard to excuse, for three reasons. These patients were monitored with apps and measurement tools, yet no objective physiological data is reported. The clinic claims lidocaine has anti-inflammatory effects, but there are no inflammation markers. There are claims about autonomic dysfunction, but no autonomic testing.

Bias — extreme

Financial conflicts of interest, severe. The study was funded by Excellent Care Clinics, the private clinic providing the treatment. The lead authors are physicians at that clinic who developed and administer the treatment. The clinic's business model depends on the treatment being seen as effective, and it openly discusses plans for commercialisation and insurance reimbursement. This is a textbook case of research funded by those with a direct financial stake in positive results.

Investigator bias, severe. The researchers developed the treatment protocol themselves, have reputational investment in its success, and are actively promoting it to patients and insurers. No independent oversight is apparent.

Publication bias risk. Were there pilot studies with negative results that were not published? Was the successful cohort selected for publication?

Design choices that amplify bias. No control group, which lets natural history and placebo appear as treatment effect. No blinding, which maximises placebo and expectation effects. Reliance solely on subjective outcomes, which maximises the impact of bias. A 4-week pre-treatment baseline, which creates a regression-to-the-mean opportunity. Patients as their own controls, which is vulnerable to temporal confounding.

Statistical and methodological issues

Regression to the mean — major problem. Patients were selected on severe, longstanding symptoms. Natural fluctuation in chronic conditions means they are likely to improve somewhat regardless. Without controls this cannot be separated from a treatment effect.

Temporal confounding. Treatment started at a specific calendar time, so any concurrent factors — seasonal changes, pandemic phase changes, other treatments, life changes — are confounded with treatment. The 36-week follow-up spans multiple seasons.

Selection bias. Who self-selects to travel to a private Dutch clinic for experimental treatment? Likely a higher-socioeconomic-status, more health-literate, more motivated group, which may have a different trajectory from the general Long Covid population. There is no data on patients who were screened but not included.

Attrition. The extension to 36 weeks covered only the first group, and it is not explained why. How many dropped out? Were dropouts due to adverse events or lack of efficacy?

Biological plausibility — moderate to good

Lidocaine has documented anti-inflammatory properties, immune dysregulation is implicated in post-COVID, targeting lymph nodes for immune modulation is conceptually reasonable, and there is some pre-clinical evidence for lidocaine effects on neutrophil extracellular traps (NETs) and cytokines.

But: is subcutaneous administration actually achieving therapeutic levels systemically? No pharmacokinetic data is provided. There is no measurement of inflammation markers to show target engagement. Previous studies used IV lidocaine, and a different route may have different effects. The HP-β-CD formulation is novel, with no data on whether it actually improves delivery.

On dose and safety, home self-administration raises adherence and monitoring questions, and "the majority reported non-serious skin problems at the injection site" is not quantified.

Interpretation and spin — moderate concern

What they claim: "78% reported clear and clinically relevant improvement"; "significantly improved quality of life and symptom burden"; "offers a scalable and potentially disease-modifying approach".

What the data shows: subjective improvements in unblinded patients without controls; small-to-moderate effect sizes on subjective scales; no evidence of disease modification, since there are no objective measures of pathology; no evidence of mechanism, since there are no biomarkers.

"Disease-modifying" is an extraordinary claim with no supporting evidence here. Media coverage emphasised "80% improvement" without context about the study's limitations, and the clinic is already marketing the treatment while calling for further research.

What this study actually demonstrates

It shows that patients receiving this treatment reported feeling better, that the treatment appears tolerable (though adverse event data is incomplete), and that there is enough of a signal to warrant properly controlled investigation.

It does not demonstrate treatment efficacy, does not rule out placebo effect, does not demonstrate a mechanism of action, and does not establish that the treatment beats natural history.

Critical missing elements

1. A control group, placebo or usual care
2. Blinding, at minimum of outcome assessors
3. Objective outcome measures
4. Biomarker data on inflammation and immune function
5. Detailed safety data with full adverse event characterisation
6. Pharmacokinetic data on lidocaine levels actually achieved
7. Target engagement data, as proof of mechanism
8. Independent funding and investigators
9. A pre-registered analysis plan
10. Clear diagnostic criteria, especially for the PEM/ME/CFS phenotype

Comparison to field standards

This study would not meet quality standards for regulatory approval, practice guideline inclusion, Cochrane systematic review inclusion (too high a risk of bias), or insurance coverage decisions in most healthcare systems.

For reference, the PACE trial — rightly criticised for methodological flaws — at least had control groups, blinding of outcome assessors though not participants, and some objective measures in the form of actigraphy. This study has none of those, and adds severe financial conflicts of interest.

Recommendations

For patients. These results cannot be used to judge whether the treatment works. The improvements may reflect placebo, natural recovery or other factors. Paying privately for an unproven treatment should be considered carefully. Wait for properly controlled trials.

For researchers. Run a randomised placebo-controlled trial with adequate blinding (saline injections as placebo), pre-registered primary outcomes, a mix of subjective and objective measures, independent funding, and longer follow-up to assess durability. Add mechanistic studies: measure inflammatory markers before, during and after treatment; do pharmacokinetic and pharmacodynamic (PK/PD) analysis; identify responder characteristics. And phenotype better: separate PEM-predominant from other phenotypes, assess POTS/OI and MCAS systematically, and check whether mechanism differs by subgroup.

For clinicians. This is preliminary, hypothesis-generating data only. The treatment cannot be recommended on this evidence. Explain the limitations to patients who are considering it.

Final assessment

Grade: 3.5 out of 10.

Why it does not pass: the absence of a control group is fatal for causal inference; there is no blinding on highly subjective outcomes; the financial conflicts of interest are severe; there are no objective outcome measures; and the risk of bias is too high to draw meaningful conclusions.

What prevents a lower score: the sample size is adequate for this design, the biological mechanism is plausible, effect sizes appear consistent across outcomes, follow-up duration is reasonable, and symptom tracking is detailed.

Why not higher: it would need a control group, blinding, objective measures and independent funding to score above 5. The current design cannot distinguish efficacy from bias, placebo or natural history.

Conclusion

This study is the kind of preliminary, commercially motivated research that has dogged the ME/CFS and Long Covid fields. The intervention deserves proper investigation, but this paper cannot support conclusions about efficacy. No control group, no blinding, only subjective outcomes and severe conflicts of interest together produce an evidence base that is close to worthless for clinical decision-making.

The biological rationale is reasonable enough to justify a proper randomised controlled trial (RCT), but patients and insurers should not be making treatment decisions on this data. The enthusiastic media coverage and clinical promotion ahead of adequate evidence is concerning, and potentially harmful to the field's credibility.

The most charitable reading is that this is very preliminary data suggesting a signal worth investigating properly. The concerning reading is that it is commercially motivated research designed to generate marketing material for a private clinic's profit-making treatment, with methodology inadequate to determine efficacy. The truth is probably in between, but the evidence quality is far too low to make any claim about treatment efficacy.