Randomised controlled trials (RCTs) are often costly, slow, and logistically challenging - ChatGPT (2025)

In this article, we asked ChatGPT to provide a better solution for randomised controlled trials (RCTs). Here's what ChatGPT said:

You're absolutely right—randomised controlled trials (RCTs), while considered the gold standard for evidence, are often costly, slow, and logistically challenging. Here are some pragmatic, more feasible alternatives or complementary approaches that can provide valuable evidence—especially in real-world settings or for repurposed/low-cost interventions:

---

✅ 1. Pragmatic Clinical Trials (PCTs)

• What: Conducted in real-world settings with broad patient inclusion criteria.

• Advantage: Less strict protocol than traditional RCTs, more generalisable, lower cost.

• Use Case: E.g., testing off-label or repurposed drugs like ivermectin or fenbendazole in actual outpatient settings.

---

✅ 2. N-of-1 Trials

• What: A single patient undergoes multiple treatment periods with different therapies in random order.

• Advantage: Useful for rare diseases or personalised treatment; faster and cheaper than large trials.

• Use Case: Individualised testing for patients unresponsive to standard cancer therapies.

---

✅ 3. Synthetic Control Arms

• What: Use historical data or real-world evidence (RWE) to replace or supplement the control group.

• Advantage: Reduces need to randomise patients to placebo; saves time, money, and ethical concerns.

• Use Case: Oncology studies where standard-of-care is already well documented.

---

✅ 4. Real-World Evidence (RWE) from Observational Studies

• What: Retrospective or prospective cohorts using EMRs, patient registries, claims data, etc.

• Advantage: Large, inexpensive data sets reflecting everyday practice; valuable especially when RCTs are not feasible.

• Use Case: Evaluating long-term effects or safety profiles of interventions already in use.

---

✅ 5. Case Series / Retrospective Case Reviews

• What: Compile and analyse existing anecdotal or clinical case data.

• Advantage: Fast, useful for hypothesis generation or rare outcomes.

• Use Case: Early-stage evaluation of repurposed or natural treatments in cancer (e.g. fenbendazole and ivermectin projects).

---

✅ 6. Bayesian Adaptive Trials

• What: Uses Bayesian statistics to continuously update probability of treatment effect as data accumulates.

• Advantage: More efficient, allows for early stopping or arm-dropping; flexible.

• Use Case: Small-scale oncology studies where multiple treatment arms are tested.

---

✅ 7. Digital Trials / Remote Monitoring

• What: Use apps, wearables, and telemedicine to gather patient-reported outcomes and data.

• Advantage: Reduces infrastructure cost, improves scalability, improves compliance.

• Use Case: Lifestyle or dietary interventions; long-term safety/effectiveness follow-ups.

---

🔁 Hybrid Strategy Recommendation:

For controversial or off-label interventions (e.g. repurposed drugs, supplements), a retrospective case series → matched cohort from real-world data → pragmatic or adaptive trial pathway offers progressive evidence-building without jumping straight into a multi-million dollar RCT.

A meta-analysis of meta-analyses

According to Massimaux (Substack):

The faithful followers of mainstream medicine, and thus indirectly of the big pharmaceutical industry, who understand something about statistics, have been parroting throughout the COVID pandemic that the vast majority of studies on cheap treatments such as ivermectin, hydroxychloroquine, vitamin D, melatonin, zinc and others are observational, i.e. non-randomized controlled studies, and therefore have a built-in defect and can only detect correlation (association), but not causation. In other words, they have no value in public health.

For example, if ivermectin was given to one group of patients, and another group of patients refused or were unable to receive it, and if the ivermectin group had a significantly lower mortality rate, even if the results were adjusted for many baseline parameters, it could not be concluded that ivermectin was the one that reduced mortality because the patients in the groups were not randomly selected in advance (randomized). While this is potentially a valid argument, rejecting observational studies a priori is, to put it mildly, wrong. What is the contribution of c19early.org to this issue? They decided to test the hypothesis: do the pooled effects of randomized studies differ statistically significanly from the pooled effects of non-randomized studies? If there is a statistically significant difference between them, then, in general, it is most likely due to the unequal baseline characteristics of the treated and untreated groups in non-randomized studies.

Can you guess what their result was? I couldn’t believe it either when I looked at the so-called forest plot of the mega meta-analysis of 102 meta-analyses. The result in their own words:

“For the 102 treatments we cover, there is no difference in results between RCTs & observational studies, RR 1.00 [0.92-1.08].

For the subgroup of high-profit treatments, there is a non-significant trend towards greater efficacy in RCTs. This may be related to financial conflicts of interest.”

In simple terms, this result is interpreted as follows: observational, non-randomized controlled trials of COVID treatments were as valid as randomized trials in determining causal relationships between treatments and clinical outcomes.

By the way, this is not the first meta-analysis of meta-analyses to compare randomized and non-randomized trials. In 2014, Engelmayer et al. published a Cochrane review/meta-analysis of 15 reviews that included 1583 meta-analyses for 228 different medical conditions with an average of 178 studies per paper. They concluded that:

“on average, there is little evidence for significant effect estimate differences between observational studies and RCTs”

End of story.