Does case misclassification threaten the validity of studies investigating the relationship between neck manipulation...

The purported relationship between cervical manipulative therapy (CMT) and stroke related to vertebral artery dissection (VAD) has been debated for several decades. A large number of publications, from case reports to case-control studies, have investigated this relationship. A recent article suggested that case misclassification in the case-control studies on this topic resulted in biased odds ratios in those studies.

Given its rarity, the best epidemiologic research design for investigating the relationship between CMT and VAD is the case-control study. The addition of a case-crossover aspect further strengthens the scientific rigor of such studies by reducing bias. The most recent studies investigating the relationship between CMT and VAD indicate that the relationship is not causal. In fact, a comparable relationship between vertebral artery-related stroke and visits to a primary care physician has been observed. The statistical association between visits to chiropractors and VAD can best be explained as resulting from a patient with early manifestation of VAD (neck pain with or without headache) seeking the services of a chiropractor for relief of this pain. Sometime after the visit the patient experiences VAD-related stroke that would have occurred regardless of the care received. This explanation has been challenged by a recent article putting forth the argument that case misclassification is likely to have biased the odds ratios of the case-control studies that have investigated the association between CMT and vertebral artery related stroke. The challenge particularly focused on one of the case-control studies, which had concluded that the association between CMT and vertebral artery related stroke was not causal. It was suggested by the authors of the recent article that misclassification led to an underestimation of risk. We argue that the information presented in that article does not support the authors' claim for a variety of reasons, including the fact that the assumptions upon which their analysis is based lack substantiation and the fact that any possible misclassification would not have changed the conclusion of the study in question.

Current evidence does not support the notion that misclassification threatens the validity of recent case-control studies investigating the relationship between CMT and VAD. Hence, the recent re-analysis cannot refute the conclusion from previous studies that CMT is not a cause of VAD.