Another recent correlational study on the relationship between talcum powder (TP) and female reproductive tract cancers was published. This research originated in the 1970s when gynecologists at one hospital noted that an unusually high number of women were experiencing these cancers. They reviewed medical records and, in looking at possible causative agents, noted that many of the affected women used TP. They published their findings, and additional case review studies at other institutions followed; they, too, found a correlation between TP and these cancers.
After five decades, one would think that this would be, if not settled research, a reasonably established relationship, but it is not; several meta-analyses determined that the majority of studies were of mediocre quality, questioning their findings, and it was concluded that if there was an association, it was weak at best. And aside from costing Johnson & Johnson a grotesque sum of money in questionable lawsuits, nothing was accomplished to forward women’s health.
Of note, the cancer site in these studies has migrated, over time, from the uterus to the ovaries, while the putative causative agent remains unidentified. Initially, it was thought to be the minuscule amount of asbestos in the original formulation, but this was removed ±20 years ago. It would appear that either another component or talcum itself might be the culprit, assuming TP plays a role in these cancers.
Perhaps more significantly, the mechanism by which TP, an inert substance, migrates from the external genitalia through the vaginal barrel to an internal location remains unelucidated. In reproductive-aged women, there is a strong outward fluid flow in the barrel. This is why spermatozoa swimming in groups are advantaged over singletons, in that they can largely nullify the current’s effects, increasing potential fertilization. How does immobile powder navigate this current? Additionally, there is a gap between the ovarian end of a fallopian tube and an ovary. Mature ova are unable, at times, to negotiate this gap. By what contrivance does TP cross the divide?
It is a statistical axiom that correlation does not equal causation. Nonetheless, the relationships between cigarette smoking and cancer and asbestos and lung disease are correlational. Both have been extensively supported by animal and in-vitro (laboratory) studies. However, I have seen no published animal or lab studies supporting the putative relationship between TP and cancer. Absent these, the repeatedly found correlation, while interesting, is medically meaningless.
Research involves three types of variables: independent (IV), dependent (DV), and extraneous (EV) or confounding. IVs represent potential cancer cause(s), while the DV is cancer. An EV is anything that may influence the relationship between the independent and dependent variables. These can be multitudinous, and their control is essential to achieve an accurate picture of the relationship between cause and outcome.
Correlational studies, especially those involving records reviews and/or span decades, as did the talcum studies, have significant weaknesses pertaining to IV determination and confounder control. Medical records (MRs) have long been considered an excellent source of patient data for retrospective studies, but they are unstandardized as to content, which makes accurate compilation of potential IVs and confounders difficult. The authors of a recent review of MR research concluded that while they have “potential” as information sources, they had not yet achieved their full value, given their lack of standardization. Further independent variable problems are related to the sampling method and sample size, as both can limit generalizability only to women who participated in the study.
Also, a routine clinical query about TP use is uncommon, and it would likely be necessary to seek out and specifically question women about this use. By the time the research is conducted, some would have expired, and their data lost; yielding skewed results and false conclusions. It would also be necessary not only to know if women used talcum but to explore how they used it. To further complicate information collection, it is well documented that personal recollections are often inaccurate and may be biased by the phrasing of inquiries.
In some of the studies, the determined correlation factor was not significant but “approached significance,” and the researchers valued it as indicating a relationship. Yet, the value of such results has been the subject of many years of debate, as claiming that a non-significant value still signifies significance eliminates the importance of establishing a level of significance, a cornerstone of research.
Of necessity, researchers would have to use a statistical analysis that controlled the effect of each confounder to assess any relationship between TP and cancer. The early studies, those used to sue Johnson & Johnson, made no attempt to account for these, negating their value. Some later studies were more sophisticated but still demonstrated only weak correlations. Given these repeated poor statistical findings, no supporting animal and/or in-vitro research, and no clear causative mechanism, is continued interest in talcum powder realistic? If research in this area is to continue, rather than further correlational repetitions, researchers need to perform animal or laboratory analyses if they wish to establish a link between talcum and cancer. To do otherwise does nothing to validate the assumption of the cause or to aid women’s health.
M. Bennet Broner is a medical ethicist.
