This is the second installment analysis of a three (and now 4) part series of articles on effects of homeopathy on childhood diarrhea. This second installment elaborates on our findings on data from the second clinical trial in Nicaragua. (1)
I should first explain the title. In order for homeopathy to operate as a base or operating system for medicine “for the 21st century,” the entire system of measurement and of course all physical laws would have to be changed. In analogous political terms, it would be similar to – but more massive a change than – changing a nation from a democracy to a completely different system such as a theocracy with completely different laws and behavior expectations. So…well, it was the best I could think up at the time.
Last time I recounted how the Jacobs ll trial setup was incoherent and unable to produce results that could prove efficacy – unless the differences between treatment and controls were quite large, greater than just barely significant. Most patients were treated differently from others, with multiple preparations (that were in reality the same: pill filler) at differing times during the illness, with each preparation selected according to symptoms that likely varied by the hour, and influenced by memory, well known to be faulty in medical studies.
In fact, given the lack of homogeneity in the trial diagnoses and treatments, outcomes should not have made sense at all. Now I must admit that the thought did not occur to us at the time we undertook the review, nor during the review. If it had, our job would have been easier and the paper shorter.
First, we saw that children with severe dehydration were hospitalized, placed in IV fluids, and not entered in the trial. Milder cases were continued on diet and given standard oral rehydration therapy. They were assessed by homeopathic interviews regarding stool quality, general condition, and emotional states, then randomized to treatment and control groups. The major outcome was stated to have been determined beforehand as the mean number of days to fewer then 3 unformed stools on two consecutive days.
We looked at tables and graphs and saw that the treatment-control differences seemed to be significant and consistent at all days measured.
The primary outcome was: treated, mean of 3.0 days, and for untreated, 3.8 days (P<.05). And, there was a difference at 24, 48, and 72 hours after starting treatment Four other diarrhea indicators showed P=.048, .036, .054, .037.
But the results were actually illustrations of the difference between statistical significance and clinical significance. At 72 hrs, the point of maximum difference (P<0.5,) the difference was between 3 stools per day and 2 per day. The differences on the other days were all less than 1 per day.
We looked further and found recording and computation errors in several tables. None alone seemed large enough to account for the statistical findings, but suggested that enough other errors might lie hidden in the record to account for the possibility.
The most significant finding was the result of culturing the stools. All subjects had stools cultured. Children in the homeopathy treatment group had half again as many positive stool cultures for bacterial pathogens as did the controls. This led us to the possibility that all or many of the children might have been given antibiotics, either through adulterated homeopathic preparations, or at some time outside of the researchers’ observations. A few such patients could have accounted for the difference.
The data were presented as the mean values without data points for all patients. In such presentation, there is a possibility that one or two outliers might have existed in either group or both to have resulted in the differing means. Without the raw data, we don’t know.
Then, as mentioned last time, the data were the results of heterogeneous entry data to which were applied statistics that assume data have some degree of homogeneity. Another problem casting doubt on the meaning of the outcome.
Added to all the irregularities and lack of clinical significance were the claims and language of the authors. The authors referred for authority of homeopathic validity to the discredited Benveniste study in Nature (1998.) That study, if it were valid and authentic, actually disproved the major thesis of homeopathy and showed the selection of dilutions to be arbitrary and effects unpredictable.(2)
The Jacobs paper authors inflated the significance of their findings with phrasing like: “Acute diarrhea is the leading cause of pediatric morbidity and mortality world-wide. In the developing world there are 1.3 billion episodes of diarrhea and 5 million deaths each years from this illness..” But a public health impact is not suggested by this study. Children with severe diarrhea – the ones likely to die – were not included, and sent to a hospital. There was nothing about homeopathy in children with severe diarrhea.
Then they stated, “ Acute childhood diarrhea [is] ideal for …homeopathic trial because …no standard allopathic [sic] treatment would have to be withheld and … the public health importance … is great.” But the trial used children with mild diarrhea. It is hard to imagine a great public health importance for a method that shortens by less than a day a self-limited illness that resolves spontaneously in four days.
We ended up with enough findings to conclude that the study was devised in a way as to render its results nul, to suspect its authors of hyping an idea, and with enough errors to suspect that even more lay hidden in the raw data.
Next time I will explore its sequel, a repeat of the study in… oh, yes, Nepal. Of course. And then on to the pinnacle of the quartet, the meta-analysis.
REFERENCES:
- Jennifer Jacobs, Stephen S. Gloyd, James L. Gale, L. Margarita Jiménez, & Dean Crothers (1994). Treatment of Acute Childhood Diarrhea With Homeopathic Medicine: A Randomized Clinical Trial in Nicaragua Pediatrics, 93 (5), 719-725. PMID: 8165068
- Sampson W. When not to believe the unbelievable. Skeptical Inquirer 1989;14:90-91.
POSTS IN THE SERIES: