Ethical Breach in Sleep Bruxism Study also Destroyed its Scientific Credibility

Yesterday, I discussed a new study published in the July issue of Tobacco Control which reports the results of a randomized controlled trial designed to study the relationship between secondhand smoke exposure and sleep bruxism in children. I argued that the study violated ethical principles of research conduct because parents in the control group were instructed to continue smoking in the presence of their children (while parents in the intervention group were instructed to stop smoking in the presence of their children).

Today, I comment further on why the study was unethical and I explain why this serious ethical breach also destroyed the study's scientific credibility.

The Ethical Breach

As I noted yesterday, the Helsinki declaration emphasizes that an effective intervention (or treatment) cannot be denied to human research subjects unless it is associated with only a small increase over minimal risk. Clearly, the withholding of any treatment for the control group in this study represented a large increase over minimal risk. Even the study authors acknowledge this critical point. Thus, the study violated the Helsinki declaration.

But it gets even worse. What was the intervention delivered to research subjects that according to the study, resulted in the removal of secondhand smoke exposure for children in that group? It was simply instructing the parents not to smoke around their children. If reducing secondhand smoke exposure is as simple as instructing parents not to smoke around their children, then it becomes even more of an ethical breach to withhold this very simple intervention from the control group.

The Rest of the Story

The breach was more than just an ethical one. It also resulted in the destruction of the scientific integrity of the study. Let me explain why.

Recall that the authors of the study defended the protocol by noting that after randomization, 27 families refused to participate in the study. Many families randomized to the intervention group withdrew from the study because they did not feel they would be able to refrain from smoking in the presence of their children. And many families randomized to the control group withdrew from the study because they were concerned about the risks of secondhand smoke and did not want to continue to expose their children to these risks.

Therefore, what remained in the control group were parents who were not particularly concerned about the risks of secondhand smoke exposure and who had no interest in reducing or eliminating their children's exposure. The investigators themselves acknowledge that every parent remaining in the control group met this "criterion": "All the parents of group 2 remaining in the trial were those who reported not being able to reduce children's exposure to SHS." 

Given that there were only 107 families in the study, the withdrawal of 27 families who recognized that the study was unethical represents a huge proportion of the eligible sample.

What this means is that the randomization was destroyed. This was not a randomized study, as the paper claims. Instead, it was essentially a self-selecting study where two very different groups of families were obtained. In fact, they couldn't have been more different. One group was very concerned about the health of their children and wanted to reduce their children's exposure to secondhand smoke. The other group was apparently not that concerned about the health of their children had no desire to reduce their children's exposure.

In other words, in purporting to examine the effect of secondhand smoke on the elimination of sleep bruxism in children, what the study actually investigated was the role that parental concern over child health plays in eliminating sleep bruxism.

It may well be that the higher rate of reduction of sleep bruxism in the intervention group was not due to reduction of secondhand smoke exposure, but to the fact that these parents were more concerned about their children's health and may have intervened in other ways to try to help their children overcome the problem of sleep bruxism. Parents in the control group likely did much less to work with their children to eliminate bruxism.

Given this plausible alternative explanation for the observed study findings, I believe that the study conclusions are invalid.

The rest of the story is that this study should not have been published, not only because the research violates basic ethical principles of research conduct but because the study has no scientific validity.