AI (Acupunctures Inevitable) Slop

Years ago, when I was a mere lad using Word Perfect on MS DOS, it was 4 a.m. and I just finished the final draft of a grant proposal that was due to the next day. I spun around in my chair and managed to kick out the power cord. I was a Mac user then, but had to use a PC for work. Mac had autosave. Word Perfect? Nope. Its one imperfection, I suppose. And I had not saved for hours, assuming autosave. All that work lost. I felt like crying. But in the process I became hyper compulsive about back-ups. One copy on the computer, one on an SD card, one on the web for any active project and backup every day. And then the whole computer backed up once a week.

Somehow my first draft of this essay vanished. No backup anywhere. Poof. Must have been around the time my prior machine died. So have I have recreate all that work from scratch and there are few things as boring as recreating prior work.

Expletive deleted. But here goes.

These are odd times. Fringe ‘medicine’ has become mainstream and mainstream medicine has become fringe. So, is writing about SCAMs now doing what the occasional critic has suggested: that we debunk mainstream medicine? No quite yet. But we are getting there.

This showed up in my feeds: Study supports acupuncture as a safe and effective treatment for chronic low back pain in older adults.

Really? News to me.

Most of the coverage on the internet reported pretty much the same thing:

According to a recent study, older Americans with chronic low back pain who received acupuncture had greater improvement in physical function and reduced pain than those who received usual medical care only, generally prescribed medications or physical therapy.

Cut and paste forever. Not that anyone took the time to read the original report with a critical eye.

Does repetition make it so? Maybe for big lies. But howz about the middlin’ lie that acupuncture has efficacy for back pain? Data to date suggests nope.

The meta-analysis suggests the usual beer goggle effect from acupuncture: subjective but not objective improvements;

Immediately after the acupuncture treatment, we found statistically significant differences in pain reduction between acupuncture and sham or placebo therapy (standardized mean difference (SMD) -0.40, 95% CI -0.54 to -0.25; I2 7%; 753 participants; 9 studies), but there were no differences in function

But as is often the case, different meta-analyses can give different conclusions, so take your pick. It is the old support not illumination metaphor. The metas often note these are the results are based on poor quality studies. I have never been a fan for metas when the subject is fictional. I still think if you put a large number of cow pies in a pile, you don’t get gold, just a large pile of BS.

But maybe they have done the clinical trial to end all clinical trials for acupuncture and back pain. I mean, after decades of time and money spent on profoundly flawed trials, in 2025 did someone finally do it right?

What do you think? So I went to the source: Acupuncture for Chronic Low Back Pain in Older Adults A Randomized Clinical Trial.

This is a pragmatic study. A pragmatic study is defined as one that ignores all the subtleties and biases that might render a study negative. Pragmatic SCAM studies are likely to be positive, although I cannot find where that has been rigorously evaluated.

They compared acupuncture, extended acupuncture, and usual care for non-specific low back pain. No sham control. No real blinding (just between those who received usual or extended acupuncture). No objective endpoints, all patients reported outcomes.

You know in advance that acupuncture will be superior to usual care. We have known for decades the powerful placebo effect of invasive procedures staring with mammary artery ligation for angina. No sham, no way to know if the effects one sees are the real deal or not. No sham, no conclusion. Know sham, know conclusion.

And so it was a ‘positive’ study. I mean, why even bother to waste the time and money on such a worthless study. It tells us nothing and does not add to our understanding. To quote the bard,

It is a study

Told by PhD,

full of sound and fury, Signifying nothing.

And they find that acupuncture works. My usual quibble. Acupuncture. Singular.

What is acupuncture? This study used acupuncture based on traditional Chinese acupuncture, as if this were a single entity. Why not Korean or Japanese or ear or head or any of the other literally too numerous to count versions of acupuncture? There are probably as many forms/styles of acupuncture as there are practitioners, since the intervention is completely divorced from reality and each practitioner makes it up as they go along, i.e., individualize the care. Pft.

To bring the point home, the patients were referred to one of 50 acupuncturists, so there were 50 different interventions.

You would think that after 3000 years they would know how to needle a common problem like low back pain.ling a

Nope. They had to use a Delphi method (no, not inhaling hallucinogenic gas to see into the future) to determine where to needle and decided

Through the Delphi process, our Acupuncture Advisory Panel recommended including a total of 113 named acupoints (214 acupoints total if counting bilaterally): low back acupoints (29; 58 if bilateral; 4 central for total of 62), acupoints on the mid and upper back (33; 66 if bilateral; 2 central for total of 68), front of the body, including distal leg acupoints (33; 66 if bilateral; central points for total of 72) and ear acupoints (6; 12 if bilateral). Ashi points (tender upon palpation) are also permitted.

So it is likely each patient received a unique needling. The same response to a wide variety of interventions suggests a placebo effect. Which we know is the case, that acupuncture is only a fancy schamncy theatrical placebo. But the study design was such that if you give the article a casual reading, you would not notice this fatal flaw, one amoung many.

But I wondered. Why do they keep doing these awful, useless clinical trials that waste time and money?

I do not think the authors are stupid. Willfully ignorant of the issues of most SCAM clinical trials? True believers in nonsense? Is there a special resource for SCAM providers where they have lousy protocols produced on demand that will only confirm the usefulness of their nonsensical intervention? Seems that way.

For hoots and giggles, I asked ChatGPT for a “protocol for evaluating the efficacy of acupuncture for low back pain.”

Did a pretty good job and certainly much more rigorous than this study.

• Type: Prospective, randomized, controlled, patient-blinded, assessor-blinded clinical trial
• Arms:

1. True Acupuncture
2. Sham Acupuncture (placebo control)
3. Usual Care (optional third arm)

• Randomization ratio: 1:1 (or 1:1:1 if 3 arms)
• Blinding:
• Patients blinded between true vs. sham
• Outcome assessors blinded
• Practitioners not blinded (inherent to acupuncture)

5. Participant Selection

Inclusion Criteria

• Adults aged 18–75
• Chronic low back pain ≥ 12 weeks
• Pain score ≥ 4/10 on average
• Ability to give informed consent

Exclusion Criteria

• Previous acupuncture for LBP in last 12 months
• Radicular pain or progressive neurological deficit
• Severe spinal pathology (fracture, cancer, infection)
• Coagulopathy or anticoagulation contraindicating needling
• Pregnancy
• Current participation in another interventional trial

6. Interventions

A. True Acupuncture (STRICTA-compliant)

• Points selected: e.g., BL23, BL25, BL40, GV3, GV4, plus additional individualized distal points
• Needle type: sterile, single-use stainless steel needles
• Insertion depth: per standard anatomical guidelines
• De qi sensation: elicited when appropriate
• Retention time: 20–30 minutes
• Number of sessions: 10–12 treatments over 6 weeks

B. Sham Acupuncture

Options:

• Non-penetrating Streitberger needles
• Superficial needling at non-acupuncture points
• Same schedule and duration as true acupuncture
• No elicitation of de qi

C. Usual Care (optional arm)

• Standard primary care management (analgesics, exercise advice)
• No acupuncture allowed during study period

7. Outcome Measures

Primary Outcome

• Pain intensity measured by Numeric Rating Scale (NRS 0–10) at 6 weeks.

Secondary Outcomes

• Functional disability: Oswestry Disability Index (ODI)
• Quality of life: SF-12 or EQ-5D
• Patient global impression of change (PGIC)
• Use of analgesic medications
• Durability: pain and ODI at 3 and 6 months post-treatment
• Safety: adverse events, needling complications

8. Sample Size Calculation

• Based on detecting a 1-point difference on the NRS (clinically meaningful)
• Assumptions: SD ≈ 2.0, α = 0.05, power = 0.8
• Required sample ≈ 64 participants per group
• After dropout (15–20%), target enrollment ≈ 75–80 per group

(Adjust once pilot data or institutional parameters are known.)

9. Randomization and Allocation

• Computer-generated randomization list
• Block randomization to balance groups
• Allocation concealed using sealed opaque envelopes or REDCap randomization module
• Only the acupuncturist has access to group assignment

10. Blinding Procedures

• Participants informed that they may receive one of two types of acupuncture with no disclosure about which is “real”
• Sham needles specifically designed to maintain blinding
• Outcome assessors and statisticians fully blinded

11. Data Collection Timeline

It even asked

Would you like this turned into a full IRB-ready protocol

Grok came up with a similar protocol, although the Acupuncture Arms included a Nazi salute. Joke.

The old joke was, to err is human, to really screw up takes a computer.

There is a lot of fretting about AI slop drowning out real music, art, etc. At least when it comes to acupuncture study protocols, if you want state-of-the-art slop, you can still rely on humans to give the purist form. Not so sure we should be proud of that.

Image by Tim McReynolds from Pixabay