As some may know I am infectious disease doctor. Urinary tract infections (UTI) butter my bread. Figuratively speaking. There is an enormous amount known about the pathophysiology of UTI’s. It is both a common and complex problem. But for all our knowledge, chronic and recurrent UTI’s remain a vexing issue for the patient and the doctor.
One reason people develop recurrent UTI’s is not because of altered chi along meridians altered by needles stuck in the skin distant from the bladder. That would be ridiculous. I like reasoning from basic principles. Given what we know about anatomy, physiology and microbiology, how might acupuncture interfere with the development of a urinary tract infection? Would it prevent colonization with pathogenic E. coli? Prevent retrograde travel of bacteria up the urethra into the bladder? Stop E. coli from binding to uroepithelial cells? Have a bactericidal or bacteriostatic effect?
None of the above seem likely. To my mind, postulating any of the above as a potential mechanism for acupuncture as a preventative for UTI’s would be ludicrous. And spare me your Boosting the Immune System, a concept that exists as a marketing tool, not a useful therapeutic intervention. My boss used to say that many an academic career floundered on attempting to prevent and treat UTI’s using an immune system approach. With some exceptions, and there are always exceptions, recurrent UTI’s in normal humans are usually due to anatomic or microbiological anomalies.
Despite its popularity, it is clear that acupuncture is not based on reality and, like all pseudo-medicine, only has demonstrable efficacy in poorly-designed studies. Acupuncture displays the usual progression of all pseudo-medicines. Increasingly-well-done studies show decreasing effect until a study that removes all bias shows it to be no better than placebo. Which one would expect for an intervention based on fantasy. Prior plausibility (the toy boat of SBM, try saying it three times very fast) would predict that acupuncture is worthless. And that should be acupunctures, all 6 styles are an elaborate ritual with no more likelihood of efficacy than the superstitions in a Budweiser commercial.
Much to my embarrassment, Clinical Infectious Disease (CID), the flagship journal of my specialty, published “Recurrent urinary tract infections among women: comparative effectiveness of 5 prevention and management strategies using a markov chain monte carlo model.” One of the five interventions included in their analysis was acupuncture. Really. They thought a pseudo-medical intervention divorced from reality to be worthy of consideration for the prevention of UTI.
As to be expected, the study generated the usual ‘acupuncture works’ headlines, especially as the analysis suggested acupuncture was second only to antibiotics in preventing UTIs, and better than cranberry pills, estrogens, and symptomatic self-treatment.
Now we get into problems, given my non-statistical brain. What is a markov chain monte carlo model? Fortunately I have a statistician at work who was kind enough to translate statistics into Crislip, a.k.a. dumb it down.
It is a simulation. In this case they simulated 10,000 patients over a year and then summarized what happened on average for each intervention.
Each virtual patient in the model had an assigned probability per day of having a UTI, and there are probabilities for the various interventions being effective. They used an algorithm to determine whether or not a UTI occurs on a given day. That is why “Monte Carlo”: they pick outcomes at random with a certain probability. So if you sample enough virtual patients a UTI will occur roughly the correct percent of the time.
The “Markov Chain” part means that what happened in the past for the virtual patient is irrelevant, all that matters is what state they are in that ‘day’. So it’s a simplifying assumption, which makes evidently makes the simulation much easier.
It is a simulation and, like all simulations, the end results are dependent on how the initial conditions are set.
There is a table of Probability Values for Variables in Model. They have risk reduction probabilities for the interventions evaluated in the simulation:
- Daily antibiotics/nitrofurantoin, 100 mg once daily, risk reduction 0.86
- Acupuncture, risk reduction 0.68
- Estrogen use, risk reduction 0.65
- Cranberry risk reduction 0.50
I appears to me that the initial conditions are set up to show, as the accompanying editorial says, the ‘surprising’ result that acupuncture was second only to antibiotics in UTI prevention. The order of efficacy that came out of the simulation appears to parallel the order of the initial conditions. Go figure. Again, I am no statistician, correct me if I am wrong.
They derived their probability of risk reduction of 0.68 from two clinical trials. I asked my statistician how that number was derived and he said it was a mystery as he could not tell from the paper and that I should email the authors. As I publish this I have not heard back from the author, who said a week ago they would get back to me. When they do, I will post an addendum.
I would have expected the risk reduction probability of acupuncture to be zero, since there is zero prior plausibility acupuncture would work. It made me curious about the articles that demonstrated efficacy. Both were done by the same authors in Sweden. As we learned from Benveniste and homeopathy studies, the same group doing a study does not a reproduction make.
The most recent study, “Acupuncture Treatment in the Prevention of Uncomplicated Recurrent Lower Urinary Tract Infections in Adult Women”, did not have a sham acupuncture group, so the study is worthless. Equally important is the patients were only followed for 6 months, probably not long enough to decrease normal variability of recurrent UTI’s. What is the natural history of uncomplicated urinary tract infections in women (bold added)?
we observed 51 infection-prone women in a standardized fashion for a median of 9 years. During intervals when patients were not receiving antimicrobial prophylaxis, infections occurred at an average rate of 2.6 per patient-year, but the rate varied widely from patient to patient (range 0.3-7.6 episodes per year). Seventy-three percent of the observed episodes were symptomatic, with an 18:1 ratio of cystitis to pyelonephritis episodes. Infectious episodes were strikingly clustered, and rates of infection decreased in the winter months.
Their microbiologic definition of UTI is probably outdated and would fail to diagnose many UTI’s:
bacteriuria (10^5 or more colony-forming units per milliliter of uropathogen or any amount of Staphylococcus saprophyticus).
As a recent NEJM study, “Voided Midstream Urine Culture and Acute Cystitis in Premenopausal Women”, suggests 10^2 CFU of E. coli is predictive of UTI.
And with a p value of 0.08, it is far from the .005 that would be suggestive.
An earlier study, “Acupuncture in the prophylaxis of recurrent lower urinary tract infection in adult women”, has the same fatal flaws they were to repeat later: inadequate duration of follow up, an inadequate microbiologic definition of UTI (although of course they could not have known), poor statistical validity and very small numbers in each arm of the study (acupuncture (27), sham acupuncture (26), and control (14)).
There was no microbiologic information in the latest study, but the early study reported 5 patients with enterococcus as the pathogen, with the NEJM article suggesting that enterococcus is not a pathogen in this population.
In the latest study no mention was made as to what mechanisms was used to get follow up, in the earlier study it was patient initiated. We have no way of knowing, given the self-limited nature of cystitis, if patients became symptomatic and did not bother to come in or sought care elsewhere.
And, importantly in the first study, no mention was made as to whether blinding to sham or real (as if there is a difference) acupuncture was successful, which the authors recognize was a flaw with plenty of opportunity for the clever Hans effect (unconscious cuing):
The study otherwise mainly applied a single-blind design, since patients were not informed about the type of treatment they were receiving. It is however difficult to ensure that there was no non-verbal communication about treatment type between the acupuncturist and the patient, and patients receiving real acupuncture may have felt an increased anticipation of an effect.
Two profoundly flawed studies from which no valid conclusions can be made about the efficacy of acupuncture, much less an estimated acupuncture risk reduction of 0.68. There was a recent review of UTI prevention in the Journal of Urology with the same conclusion:
Cranberries decreased urinary tract infection recurrence (2 trials, sample size 250, Jadad score 4, RR 0.53, 95% CI 0.33-0.83) as did acupuncture (2 open label trials, sample size 165, Jadad score 2, RR 0.48, 95% CI 0.29-0.79)
This is a classic example of the difference between evidence and science-based medicine. There is evidence for acupuncture that stinks on ice if anyone would take a moment to read the articles. A science-based consideration of reality, anatomy, physiology, microbiology would suggest that any real risk reduction by acupuncture should be zero. Prior plausibility would suggest any positive outcome in an acupuncture study would be a false positive due to methodologic errors in the study. As we have discussed at great length over the years, there is zero reason to suspect any true efficacy of acupuncture beyond study bias and poor methodologies, which were rife in these studies.
That being said, the authors say:
Somewhat surprisingly, we found that acupuncture was the next most effective prevention method. Acupuncture’s high efficacy may be a function of publication bias, as there were fewer studies on acupuncture compared to other management
No. Acupuncture’s high efficacy was a function of profoundly horrible studies and an estimation of risk reduction that is too large by 0.68. No one involved gives the appearance of having read the original papers to see if the estimated risk reduction is reasonable. The editors and reviewing peers evidently did not as well.
It is an oddity of medicine. I would wager that astronomy journals do not publish editorials touting astrology as a solution for difficult problems. Similarly, psychology journals do not look to psi and chemistry journals do not advocate the methods of alchemy. In medicine, the editors have no problem with suggesting nonsense on the basis of GIGO in their journal. The accompanying CID editorial says:
For example, in an individual case, a clinician and patient working together may opt to use acupuncture in combination with cranberry juice and self- directed therapy at the first sign of symptoms.
Did the editorial writer read the paper? The simulation evaluated cranberry pills, not juice. Goodness gracious great balls of fire, an editorial in CID recommending two simultaneous worthless therapies, acupuncture and cranberry juice. I would so love to have the editors and authors come up with a plausible explanation for how acupuncture may work for UTI prevention and why this simulation had any validity as to the efficacy of acupuncture after reading the original papers. I did not go back and read the papers touting cranberry pills, but my father taught me to judge a risk reduction by the company it keeps.
I used to have a slightly smug attitude about CID since it had been the only journal I regularly read that had not published an article erroneously promoting pseudo-medicine based on bad studies. No longer. And to add salt to the wounds, it comes out of the institution where I did my fellowship. When ID falls for pseudo-medicine (because we are the best and brightest in medicine) we are indeed doomed.
I am so bummed. I am going to have to look into treatment for reactive depression. I wonder if acupuncture would work…
*A much expanded version of an essay from the Society for Science-Based Medicine blog.