https://ivmmeta.com/
•100% of the
32 studies to date report positive effects.
Early treatment is more successful, with an estimated reduction of
85% in the effect measured
using a random effects meta-analysis, RR
0.15
[0.06-0.37].
Prophylactic use also shows high effectiveness.
•100% of the
14 Randomized Controlled Trials (RCTs) report positive effects,
with an estimated reduction of
73%, RR
0.27
[0.15-0.51].
•The probability that an ineffective
treatment generated results as positive as the
32 studies to date is estimated to be 1 in
4 billion (p = 0.00000000023).
| Show forest plot for: |
| All studies |
| Mortality results |
| With exclusions |
| RCTs |
| Early treatment | 85% improvement | RR 0.15 [0.06-0.37] |
| Late treatment | 46% improvement | RR 0.54 [0.39-0.75] |
| Prophylaxis | 90% improvement | RR 0.10 [0.04-0.23] |
| Total | 32 studies | 218 authors | 10,033 patients |
| RCT | 14 studies | 105 authors | 2,225 patients |
Figure 1. A. Random effects
meta-analysis excluding late treatment. Simplified dosages are shown for
comparison, these are the total dose in the first two days for treatment, and
the monthly dose for prophylaxis, for a 70kg person. For full details see the
appendix. B. Scatter plot showing the distribution of effects reported
in early treatment studies and in all studies. C and D. Chronological
history of all reported effects, with the probability that the observed
frequency of positive results occurred due to random chance from an
ineffective treatment.
Introduction
We analyze all significant studies concerning the use of
ivermectin for COVID-19. Search methods, inclusion criteria, effect extraction
criteria (more serious outcomes have priority), all individual study data,
PRISMA answers, and statistical methods are detailed in Appendix 1. We
present random effects meta-analysis results for all studies, for studies
within each treatment stage, for mortality results only, and for Randomized
Controlled Trials (RCTs) only.
We also perform a simple analysis of the distribution of study
effects. If treatment was not effective, the observed effects would be
randomly distributed (or more likely to be negative if treatment is harmful).
We can compute the probability that the observed percentage of positive
results (or higher) could occur due to chance with an ineffective treatment
(the probability of >= k heads in n coin tosses, or the
one-sided sign test / binomial test). Analysis of publication bias is
important and adjustments may be needed if there is a bias toward publishing
positive results.
Figure 2 shows stages of possible treatment for
COVID-19. Prophylaxis refers to regularly taking medication before
becoming sick, in order to prevent or minimize infection. Early
Treatment refers to treatment immediately or soon after symptoms appear,
while Late Treatment refers to more delayed treatment.
Figure 2. Treatment stages.
Results
Figure 3, 4, and 5 show results by treatment stage.
Figure 6 and 7 show forest plots for a random effects meta-analysis of all
studies with pooled effects, and for studies reporting mortality
results only. Table 1 summarizes the results.
| Treatment time | Number of studies reporting positive results | Total number of studies | Percentage of studies reporting positive results | Probability of an equal or greater percentage of positive results from an ineffective treatment | Random effects meta-analysis results |
| Early treatment | 8 | 8 | 100% |
0.0039
0.0039
1 in 256 |
85% improvement RR 0.15 [0.06‑0.37] p < 0.0001 |
| Late treatment | 14 | 14 | 100% |
0.000061
6.1e-05
1 in 16 thousand |
46% improvement RR 0.54 [0.39‑0.75] p = 0.00024 |
| Prophylaxis | 10 | 10 | 100% |
0.00098
0.00098
1 in 1 thousand |
90% improvement RR 0.10 [0.04‑0.23] p < 0.0001 |
| All studies | 32 | 32 | 100% |
0.00000000023
2.3e-10
1 in 4 billion |
77% improvement RR 0.23 [0.16‑0.34] p < 0.0001 |
Table 1. Results by treatment stage.
Figure 3. Results by treatment stage.
Figure 4. Chronological history of early and late
treatment results, with the probability that the observed frequency of
positive results occurred due to random chance from an ineffective
treatment.
Figure 5. Chronological history of prophylaxis results.
Figure 6. Random effects meta-analysis for all studies.
Figure 7. Random effects meta-analysis for
mortality results only.
Randomized Controlled Trials (RCTs)
Results restricted to Randomized Controlled Trials (RCTs) are
shown in Figure 8, 9, and 10, and
Table 2. RCT results are similar to non-RCT results.
Evidence shows that non-RCT trials can also provide reliable results.
[Concato] find that well-designed observational studies do not
systematically overestimate the magnitude of the effects of treatment compared
to RCTs. [Anglemyer] summarized reviews comparing RCTs to
observational studies and found little evidence for significant differences in
effect estimates. [Lee] shows that only 14% of the guidelines of
the Infectious Diseases Society of America were based on RCTs. Evaluation of
studies relies on an understanding of the study and potential biases.
Limitations in an RCT can outweigh the benefits, for example excessive
dosages, excessive treatment delays, or Internet survey bias could have a
greater effect on results. Ethical issues may also prevent running RCTs for
known effective treatments. For more on issues with RCTs see
[Deaton, Nichol].
Figure 8. Randomized Controlled Trials. The
distribution of results for RCTs is similar to the distribution for all other
studies.
Figure 9. Random effects meta-analysis for
Randomized Controlled Trials only.
| Treatment time | Number of studies reporting positive results | Total number of studies | Percentage of studies reporting positive results | Probability of an equal or greater percentage of positive results from an ineffective treatment | Random effects meta-analysis results |
| Randomized Controlled Trials | 14 | 14 | 100% |
0.000061
6.1e-05
1 in 16 thousand |
73% improvement RR 0.27 [0.15‑0.51] p < 0.0001 |
| Randomized Controlled Trials (excluding late treatment) | 7 | 7 | 100% |
0.0078
0.0078
1 in 128 |
83% improvement RR 0.17 [0.08‑0.36] p < 0.0001 |
Table 2. Summary of RCT results.
Exclusions
To avoid bias in the selection of studies, we include all
studies in the main analysis. Here we show the results after excluding studies
with critical issues likely to alter results, non-standard studies, and
studies where very minimal detail is currently available.
[Soto-Becerra] is a database analysis covering anyone
with ICD-10 COVID-19 codes, which includes asymptomatic PCR+ patients.
Therefore many patients in the control group are likely asymptomatic with
regards to SARS-CoV-2, but in the hospital for another reason. For those that
had symptomatic COVID-19, there is also likely significant confounding by
indication. Several factors are consistent with this - all treatments are
worse than the control group at 30 days, KM curves show that more than the
total excess mortality at 30 days occurred on day 1, and at the latest
followup all treatments show lower mortality than control. The machine
learning system used also appears over-parameterized and likely to result in
significant overfitting and inaccurate results. Note that this study provides
both 30 day mortality and weighted KM curves up to day 43 for ivermectin, we
use the day 43 results as per our protocol. [Vallejos] reports
prophylaxis results, however only very minimal details are currently available
in a news report. We include these results for additional confirmation of the
efficacy observed in other trials, however this study is excluded here.
[Hellwig] provide an analysis of African countries and COVID-19 cases
as a function of whether widespread prophylactic use of ivermectin is used for
parasitic infections. Since this is a different kind of study to the typical
trial, it is excluded here. [Krolewiecki] show a concentration
dependent antiviral activity of ivermectin whereby the viral decay rate for
patients with ivermectin >160ng/mL was 0.64 log10
copies/reaction/day versus 0.13 for control. However, they do not provide the
results for the entire treatment group vs. control.
Summaring, the studies excluded are as follows, and the
resulting forest plot is shown in Figure 11.
[Carvallo], control group formed from cases in the same hospital not in the study.
[Hellwig], not a typical trial, analysis of African countries that used or did not use ivermectin prophylaxis for parasitic infections.
[Soto-Becerra], substantial unadjusted confounding by indication, includes PCR+ patients that may be asymptomatic for COVID-19 but in hospital for other reasons.
[Vallejos], detail too minimal.
