This case study reports the case of a patient who developed a severe headache two weeks after an orthodontic control appointment and whose symptom relief occurred only after the correction of applied occlusal forces. Therefore, this incidence raises the question about the causal relationship between alteration of occlusion and headache. Could this case bring some new information to the old debate: Occlusion and orofacial pain?
From the determinism of Koch’s postulates to the mainly probabilistic model of Bradford Hill’s criteria, causation and causality have been extensively discussed in epidemiology and yet, it is still a challenge over time,1,2 especially when it is applied to the individual, as in a case report. One of the reasons is that causality may have a multidimensional aspect from an etiological point of view, which is something that can make Hill’s third criteria (specificity) inadequate to apply for conditions that depend on multiple factors to be present. That is where John L. Mackie’s INUS model of causation complements Hill’s causal inference.3 INUS is an acronym for an Insufficient but Necessary part of a multifactorial condition, which is itself Unnecessary but Sufficient to the effect when combined. This model considers the possibility that we do not know all of its relevant components. Therefore, individual causes are single factors that by themselves are neither necessary nor sufficient for the effect. There is a sense of necessity, but this occurs only within a certain context or situation. When one of the components is necessary to the whole set of sufficient elements, then we are facing an INUS condition. Orofacial pain falls into this category, since the proponents of the biopsychosocial paradigm describe it in terms of a multifactorial phenomenon.
Hill´s and Mackie´s models could be analytically expressed as follows:
Hill´s: Causal element A → effect B
Causal element A + unknown causal elements X → effect B
Causal element A + (unknown causal elements X and/or anti-causal elements Y) → effect B
Thus, could the accidentally orthodontic bent wire (causal element A) have caused the orofacial pain (effect B)? Could the reversal of this occlusal problem (causal element A) to also remove the symptoms (effect B)?
Even large epidemiologic studies like OPPERA4 could not precisely answer these questions, so could a case report shed some light? Well, as all cases report there are inherent limitations such as: reduced/lack of ability to generalize or to establish a cause-effect relationship, the danger of over-interpretation, publication bias, retrospective design, and a tendency to focus on the unusual. The cause-effect relationship is one of the most critical, even if the observed phenomenon is real, because the causal factors may be outside the observer’s domain (the unknown causal elements of the INUS model). For example, the sudden onset of the headache may be caused by other factors than the accidentally bent wire during orthodontic activation and/or the resolution of the symptoms could be attributed to variables other than the correction of the bent wire (anti-causal elements X of the INUS model), such as a regression to the mean, spontaneous resolution or even a placebo effect. However, case reports also have merits in, among others, detecting novelties, generating hypotheses and allowing an emphasis on an in-depth understanding of certain conditions.
In order to better understand the aforementioned aspects, it is important to separate the observations from the inferences.
The observations are just the pure facts, the phenomenon itself as presented in clinical reality, which are: a) the headache started after the orthodontic intervention and b) it ceased after the correction of the previous intervention.
The inferences, however, are arguments based on the author’s perception of the phenomenon that tries to explain the observations and it is usually biased from individual experience, beliefs and applied methodology. In this paper, the main inference is clearly described in the first paragraph of the discussion, where the author attributes the resolution of the patient’s complaints to the equilibration of occlusal forces and there is also an underlying inference, which is actually the foundation for the main inference, which is that the symptoms were caused by a dental occlusion problem, accidentally induced by the orthodontic treatment. Therefore, the main inference implies extrapolation in the form of endorsement of occlusal therapy and the underlying inference implies causation, once it establishes with the induction a certain type of occlusion problem as an etiologic agent of orofacial pains, in both cases pushing the boundaries of a case report.
These two inferences, however, deserve further examination through two fundamental questions:
- Could the sudden onset of the headache be caused by the accidentally bent wire that changed the occlusal forces?
- Could the correction of these anomalous forces be the reason why the headache and other symptoms were solved?
Notice that there is also a modal logic between the two questions that requires analytic thinking to evaluate the inferences in terms of the possibilities and the probabilities.
- If the response of question (a) is “no,” then (b) is probably also “no.”
If the answer to question (a) is “no,” it is assumed that the alteration of occlusion would have had no causal effect, and the symptom must have come from some other source. Therefore, the correction of occlusion could not have removed the symptom since it would not address the cause. If (b) was a pain relief drug instead of a technologic guided occlusal therapy, then it would be possible to have a “no” for question (a) and a “yes” for (b), once palliative care could, eventually, control the pain without removing the cause. However, this was not the case. It is important to take into account that before the occlusal correction, there was an “alternative (b) intervention,” since the patient tried acetaminophen, ibuprofen, naproxen without any improvement. So, (a) is a necessary condition to (b) and a “no” for both questions implies two coincidences temporally linked but not biologically connected. It is possible, but with what probability?
Even the International Classification of Headache Disorders - 3rd edition (ICHD-3) accepts teeth and TMD as possible causes of headache as it is coded in different parts of the document (codes 11.6, 11.7, 11.9, A11, etc.).5 So, the question (a) is definitely “yes,” which leaves the analytical thinking in terms of probability for question (b) to be either “no” or “yes”, as discussed below:
- If the response of (a) is “yes”, then there is a higher probability (b) could also be “yes.”
In this case, if (b) is “no,” then it is assumed that the occlusal problem caused the symptoms, but it was resolved because of factors other than the correction of its own cause (the unknown factors of the INUS model). Is this possible? Yes, it seems illogical, but it is definitely possible. Let’s consider that the occlusion correction was not done, the symptoms could improve anyway overtime due to any other biological variable, such as tissue adaptation, reduction of inflammation, etc. This is called regression to the mean. Symptoms, especially a subjective one such as pain, are not linear. After a peak it tends to decrease to the mean pain level, as it slowly progresses to chronification. Biologically, it is a process that needs time, but the patient improved just minutes after the intervention! This increases the probability of (b) to be “yes,” but there is still a possibility to be a spontaneous regression or a placebo effect.
A spontaneous regression just minutes after the intervention is very unlikely since the symptoms were in an escalating process and not responsive to palliative care. Even without knowing the pathophysiology behind the symptoms, it would be very unscientific to assume they would just disappear in a miraculous way. In this scenario, could a placebo effect be more likely to occur?
Well, spontaneous regression and placebo effect share an important common feature; they are both subjective, especially dealing with pain, which is a non-measurable symptom. That’s why Santorio Santorii, an Italian physiologist, invented the thermometer in 1612 and is still celebrated as the Father of Experimental Medicine. His work also represents the beginning of the explanation of the physiological process of the human body in terms of the laws of physics.6
He realized that asking a patient if he still has fever after administration of a drug (or a placebo in an experiment) was not as accurate as measuring him with a thermometer, to properly determine if the drug was effective in reducing a fever. This was the beginning of biometric instrumentation already in the 17th century.
When a placebo drug is administered, it is more difficult to ponder. In both cases, a physician would rely basically on timing (considering the standard time known for the drug to produce an effect) and the patient’s perception of the symptom. In this example, monitoring with a thermometer gives the professional more precision to determine if there is any decrease of the fever’s temperature within the time-period that he is expecting the drug to act. This improves the odds of him correctly detecting a causal-effect relationship between the intervention (drug) and the symptom (fever), having more accuracy to rely on the selected intervention or change it.
Pain, itself, is completely subjective and non-measurable, but many other pathophysiological processes are completely measurable. Peripheral sympathetic constriction, that is directly implicated in some form of pain can be measured by digital thermal imaging, muscle function can be studied by applying computed surface electromyography, mandibular movement patterns can be registered with a jaw tracking device and occlusal forces can be recorded with a T-scan unit.
Once there is objective data available, the placebo effect is very unlikely to occur or, if it does, there is very little effect. In a meta-analyses published in 20017 and an update study in 2004,8 Which encompassed 156 trials and 11,737 patients, it was found that in studies with a binary outcome, such as improved or not improved, the placebo group had no statistically significant improvement over the no-treatment group. Similarly, there was no significant placebo effect in studies in which objective outcomes (such as blood pressure) were measured by an independent observer. Placebo effects could only be documented in studies in which the outcomes were subjectively reported by the subjects themselves.
At the individual level, the use of T-scan recordings allowed the authors to go beyond the subjective nature of pain and objectively document the pathological effect of the accidental bent wire upon the dental structures and also the physiological improvements in occlusal dynamic forces after the removal of the bent wire (in a biological gradient and direct chronological relation to the subjective symptom relief). This biomechanical, physiological improvement can be pointed to as a clear plausible biological explanation for symptom improvement, since it reduces stress upon the teeth, muscles and joints, which are highly infiltrated with sensory structures. Therefore, the debate over placebo vs. actual efficacy of the intervention, in this particular case report, seems to tend heavily to the later, in other words, the inferences of the authors are highly likely to be correct.
It is important to note that the temporality, the biological plausibility and the biological gradient are key points of Bradford Hill’s criteria for causation.6 Also, the pathophysiological data measured with T-scan reveals part of the puzzle that constitutes the unknown elements of John Mackie’s INUS mode.3
In this sense, this simple case report:
Reinforces how the understanding, the etiology and the pathophysiology are all heavily technology-dependent;
Challenges the taxonomic classification;
And reflects the way clinicians “verify” and “falsify” scientific arguments by observing how the arguments explain (or not) the clinical reality.
The item (1) compromises those large trials that are limited to using psychometric questionnaires and subjective physical examination for diagnosis and then treat purely subjective orofacial pain. Without proper tools to access pathophysiology and biomechanics, the “unknown elements” of Mackie´s INUS model are not controlled.
The item (2) compromises taxonomic classification such as the ICHD-3 and RDC-TMD (now DC-TMD) that are completely dependent upon subjective symptoms. These classifications give no clues about etiology and/or pathophysiology, generating aberrant categorization. Without biometric data, this patient could easily be classified in the group of tension type headache, a primary condition typically treated with pharmacology, while with the biometric data this same patient was moved to the category of headache secondary to a dental condition, requiring a completely different treatment.
And finally, the item (3). When a hypothesis such as “there is no relationship between occlusion and orofacial pain,” as proposed by the biopsychosocial paradigm, does not survive Popper’s falsification principles and does not explain the clinical reality once more phenomena (measured pathophysiology) are considered, a paradigm shift occurs. It has long been a tradition to limit the validity of a treatment option, such as pharmacological and palliative care for orofacial problems, by showing individual cases where this kind treatment is either contraindicated or has been ineffective.9 Thus, case-studies such as this one can and do add knowledge to medical and dental diagnosis and treatment planning.
Conflict of Interest Statement
The author reported no conflict of interest.