[CORRECTION] The quality of “ HIV” studies may be even worse than ever
More PrEP nonsense, now for “cisgender women”!
My initial email contained a typo. The total number of participants was 6296. I apologize for the oversight.
They even admit it. Here is yet another poorly designed study on “PrEP efficacy.” This was an eight year analysis (2012 to 2020) of 11 post-approval studies conducted in Kenya, South Africa, Botswana, India, and the U.S., so these results are pooled. It’s interesting to note that the total number of “incident HIV infections” (32 out of 6296 participants) that occurred in the studies, plural, is so low as to render their statistical conclusions highly suspect. Here is the paper:
I’m going to break with my usual approach here and highlight something the authors stated right near the end, in the “Limitations” section, because it’s important. It highlights the tattered state of modern academic research. A study with so many methodological flaws perhaps should not have been published at all; scientific papers are accepted largely on the basis that they provide something new, and these PrEP studies read more like the MMWR, which is not an academic journal. I’d like to focus on this, although I will report the findings of the study and their significance, if any, because this is the really important part. All emphasis is mine.
Here’s the relevant section:
This analysis has several limitations. First, data were pooled from heterogeneous postapproval studies that differed in important ways, including (1) approaches to collecting the objective and subjective measures of adherence, (2) varied geographies with different HIV incidence, (3) distinct inclusion and exclusion criteria reflecting diverse behaviors associated with HIV acquisition among the participants, (4) approaches to mode of PrEP delivery (eg, home PrEP delivery, use of mobile clinic), and (5) differences in sociodemographics related to country setting. Second, adherence data, especially from objective measures, were only available for a subset of participants, limiting generalizability. The use of plasma tenofovir concentrations to categorize participants into adherence categories may be susceptible to misclassification because plasma tenofovir has a short window of detection. However, no changes were observed in sensitivity analyses excluding the 46 participants for whom plasma tenofovir measures were used in group-based trajectory modeling (data not shown). Importantly, objective adherence data were available only for a minority of participants and were largely representative of the total population. As previously reported, it was also observed that participants’ self-reported adherence was substantially overestimated compared with objective measures. This reinforces the notion that considerable caution should be applied when interpreting self-reported adherence to PrEP. Third, the accuracy of the group-based trajectory methodology used could be improved with a greater sample size and longer duration of follow-up.
Let me comment first on the fact that this report has so many limitations and that they were so bad that they had to admit it within the paper. It is not at all unusual for researchers to report on limitations of their studies, but they are usually much more minor than these. Among the five limitations mentioned, #1 and potentially #3 are especially relevant, although the really relevant part is the group classification.
Limitation #1 is that “data were pooled from heterogeneous postapproval studies that differed in important ways, including (1) approaches to collecting the objective and subjective measures of adherence.” This basically means that the data collected using these “differing approaches” are at best highly suspect, due in part to the inconsistent methods of data-gathering that were employed. This is not totally unusual in pooled studies or meta-analyses, but those studies tend to be much larger in size.
Limitation #2 is nothing new in “HIV” research; see Henry Bauer’s analysis of the remarkable consistencies in patterns of “HIV positivity” in differing geographic locations, and among racial groups in a manner totally inconsistent with an infectious agent. Limitation #3 is confusingly worded; “distinct inclusion and exclusion criteria reflecting diverse behaviors associated with HIV acquisition” could mean anything ranging from excluding participants with certain behaviors and/or conditions to intentionally manipulating the patient pool to encourage a particular outcome, though the latter seems unlikely in a pooled study. Without further information, I can’t really say for sure what that means, but it seems significant. Limitations #4 and #5 are differing “modes of PrEP delivery” and “differences in sociodemographics related to country setting.” #4 is pretty standard, but #5 seems to be an iteration of #2. Regardless, this study basically pooled a bunch of subset studies that were inconsistently designed and that produced so few “seroconversions” as to cast doubt on the level of accuracy they claim in their conclusions.
The second bolded statement is really important. We’ve discussed before that the estimates for “PrEP efficacy” don’t come from actual real data, but rather from inferred effectiveness assuming perfect adherence, which was mostly derived from measuring tenofovir concentrations in dried blood spots or plasma, and correlating those with “HIV incidence.” These are already some serious mental gymnastics, but it gets worse. Let me repeat what they say about “tenofovir concentrations:”
The use of plasma tenofovir concentrations to categorize participants into adherence categories may be susceptible to misclassification because plasma tenofovir has a short window of detection.
Translation: using drug concentrations as a measure of adherence is almost useless. So why do they continue to do this? I think we know why. There are already so many problems with this study.
Let’s move on to the results. I’ve already let you know it’s underwhelming, so I won’t go into it too deeply. From the “key points:”
Findings In this pooled analysis of 6296 participants from 11 postapproval studies of PrEP in cisgender women, the overall HIV incidence was 0.72 per 100 person-years. The HIV incidence rates per 100 person-years for different PrEP adherence trajectories were 0 for consistently daily (7 doses/week), 0.13 for consistently high (4-6 doses/week), 0.49 for high but declining (from a mean of 4-6 doses/week and then declining), and 1.27 for consistently low (less than 2 doses/week) adherence.
Meaning In a pooled analysis of postapproval studies, cisgender women with daily or consistently high adherence to emtricitabine and tenofovir disoproxil fumarate for PrEP experienced very low HIV incidence.
The language makes it look pretty good; after all, the “HIV incidence” is so much lower in the consistent adherence group! They even have regression models to back it up! However, there are a few problems with this. First of all, as we just discussed, the methodology used to classify patients into these groups is suspect to begin with. Also, have you noticed that they use all this fancy language about “infections per 100 person-years” without mentioning (initially, you have to dig for it) how many “seroconversions” even occurred? That seems like rather an important piece of information. You have to read a bit to get to that number, and the number of women that developed “HIV positivity” during this study was 32 of 6,296. This number is further split into groups as follows:
None of the 498 women with consistently daily adherence acquired HIV. Only 1 of the 658 women with consistently high adherence acquired HIV (incidence rate, 0.13/100 person-years [95% CI, 0.02-0.92]). The incidence rate was 0.49 per 100 person-years (95% CI, 0.22-1.08) in the high but declining adherence group (n = 1166) and 1.27 per 100 person-years (95% CI, 0.53-3.04) in the consistently low adherence group (n = 632).
The 1 woman in 1156 represents an incidence of 0.1%; that leaves the 31 remaining “incident HIV infections,” which are very tricky to analyze because the researchers admit that “measures of adherence” were only available for 2,955 women, leaving 3,342, or more than half of the study participants, completely unclassified. Among those that were classified:
A total of 12 incident HIV infections were observed among the 2955 participants with adherence data, compared with 20 diagnoses among 3341 participants without adherence data.
At least we have some numbers to work with here. Let’s consider the group for whom adherence data were available, leaving aside the categorization issues we’ve already discussed. We have 1 “infection” among 1156, as mentioned, or 0.1%, leaving 11 “infections” among the remaining 1,799 women, a 0.6% incidence. Again, we’re talking differences in very small numbers over the course of twelve years!
Furthermore, “The number of participants with objective adherence data declined from 147 at week 16 to 22 at week 96.” I infer from this that the remaining women in the group of 2,955—the vast majority—provided only subjective adherence data. These are already differences in small numbers, but these are minuscule. It’s no wonder their confidence intervals are so wide. We also lack information about what other characteristics these women might have had that affected their propensity to test “HIV positive,” although we do know that half of them were nulliparous, so for at least those women, pregnancy would not have been a confounding factor. Who knows what else was at play.
I don’t think I need to explain any further why this tiny number of “incident infections” (0.5% of total participants over twelve years) can not possibly give the level of statistical confidence that the authors would like to imply. We’re almost done.
The study concludes by stating that “Among the nearly 3000 individuals with adherence data, the effectiveness of F/TDF in preventing HIV infection was very high in women who demonstrated consistently high (4-6 doses/week) or consistently daily (7 doses/week) adherence.” I think we’ve addressed the problems with drawing this conclusion.
Where this gets interesting is that the article concludes with a small missive “cautioning” women that “PrEP forgiveness” (that’s not creepy language AT ALL) is much lower among “cisgender women” than it is among MSM, so better take those pills every day to treat a condition that you wouldn’t have even if it existed, ladies!
Current guidance for PrEP use among cisgender women highlights the more rigid recommendations relative to those for MSM. In its 2021 PrEP Clinical Guideline Update, the US Centers for Disease Control and Prevention advised clinicians to counsel patients on the importance of daily dosing, noting that there is less “forgiveness” for missed PrEP doses for cisgender women than for MSM. These recommendations are largely based on pharmacokinetic modeling that suggests that cervicovaginal tissue concentrations of tenofovir diphosphate are lower than those observed in rectal tissue with similar dosing, although a correlation between tissue tenofovir diphosphate concentration and PrEP efficacy has not been definitively demonstrated.
I’m sorry, but this is just total nonsense. It’s nonsense. No matter how they dress it up with all these studies and deceptive language and misclassifications and differences in small numbers, it doesn’t hide the fact that there is a massive, accelerating push to profit off of selling very pricey, dangerous medications to people that are designed to treat a condition they don’t even have and calling that the apex of medical science. This insanity has to stop.
To support my work on Substack, please purchase my book for yourself or for a friend, and leave a review on Amazon. You can learn about efforts to ban my book here. You can also buy my new book Almost Cancelled. Or consider becoming a free or paid subscriber!
If you’re a new reader and would like some background as to my views on HIV AIDS, including the “existence” question, please refer to this post and the links contained therein. My interview with Sam Bailey is also a great introduction.
Have you seen this??
"Removing HIV from cells with CRISPR?!? What could go wrong?
https://www.bbc.com/news/health-68609297?utm_source=newsletters.tip.news&utm_medium=newsletter&utm_campaign=political-age-limits-north-dakota-may-try
I was going to comment on the fairly wide confidence intervals also. A few years back I listened to a presentation about statins, and the guy (who has a Ph.D. in biology of some specific discipline that I don't remember now) said that you have to treat 60 people to see one 'benefit' as classified by a reduction in a cardiac event that statins are claimed to reduce. But he said that all those 60 'treated' individuals are subject to the side effects of the statins.
I can't help but look at this research in the same light. We see them stratifying people according to adherence as if that's all that matter when taking these drugs. How they affect your health in 'non-target' ways is just as important. How many of these women over 12 years had decreasing bone density that will cause them long term problems as an example?
This is the problem with 'HIV' research, nothing else matters but reducing 'incidence'.