J Manipulative Physiol Ther 2005 (Jun); 28 (5): 294–302
Charlotte Leboeuf-Yde, DC, MPH, PhD, Eva N. Pedersen, MS, Peter Bryner, MChiroSc,
David Cosman, DC, Ray Hayek, MChiroSc, PhD, William C. Meeker, DC, MPH,
Junaid Shaik J, MTechChirog, Octavio Terrazas, DC, John Tucker, ME, PhD,
Max Walsh, MSc, MAppSc
Back Research Center,
Backcenter Funen, Denmark.
OBJECTIVE: To replicate a previous study of nonmusculoskeletal responses to chiropractic intervention and to establish whether such responses are influenced by the country of study, chiropractors' attitudes, and information to patients, patients' demographic profiles, and treatment regimens.
METHODS: Information obtained through questionnaires by chiropractors and patients on return visit within 2 weeks of previous treatment from chiropractic practices in Canada, United States, Mexico, Hong-Kong, Japan, Australia, and South Africa. In all, 385 chiropractors collected valid data on 5607 patients. Spinal manipulation with or without additional therapy was the intervention provided by chiropractors. Outcome measures included self-reported improved nonmusculoskeletal reactions (allergy, asthma, breathing, circulation, digestion, hearing, heart function, ringing in the ears, sinus problems, urination, and others).
RESULTS : The results from the previous study were largely reproduced. Positive reactions were reported by 2% to 10% of all patients and by 3% to 27% of those who reported to have such problems. Most common were improved breathing (27%), digestion (26%), and circulation (21%). Some variables were identified that somewhat influenced the outcome: patients informed that such reactions may occur (odds ratio [OR] 1.5), treatment to the upper cervical spine (OR 1.4), treatment to lower thoracic spine (OR 1.3), and female sex (OR 1.3). However, these had a very small "explanatory" value (pseudo R2 3%).
CONCLUSION: A minority of patients with self-reported nonmusculoskeletal symptoms report definite improvement after chiropractic care, and very few report definite worsening. Future studies should use stringent criteria to investigate a possible treatment effect and concentrate on specific diagnostic subgroups such as digestive problems and tinnitus.
From the FULL TEXT Article
The reaction pattern in our international study was found to be similar across countries, and this pattern, in turn, was similar to the one identified in the previous Swedish study. These observations indicate either
(1) that this is indeed a typical pattern of treatment effect that can be observed in chiropractic practice,
(2) that this is how nonmuscular symptoms fluctuate as a function of being under care but not as a direct effect of care,
(3) that this is how they fluctuate because they are under observation, or
(4) that this is simply how nonmuscular symptoms fluctuate in the general population.
Furthermore, a number of variables were identified that may have an effect on the reporting of N-MSRs, although these “explained” an exceedingly small part of the variation (only 3%). These variables were, for example, sex and the spinal area treated. The ORs for reporting improved digestion, allergy, hearing, or urination ranged between 1.4 and 4.1 (Table 4) in patients treated (at least) in occiput to C3 versus those treated in any other area(s). It was also more likely that patients treated in the upper half of the thoracic spine reported improved symptoms in breathing and circulation (both ORs 1.7). Whether the latter reports actually describe N-MSRs as opposed to purely somatic symptoms is, however, uncertain. Problems with breathing, such as pain on inspiration, could be purely mechanical,  and symptoms of diminished circulation are seen to occur with radicular problems of the upper limbs, commonly treated in the thoracocervical regions.  Because most patients were treated in several spinal areas, it was impossible to investigate the outcome pattern for specific spinal areas in relation to specific conditions, such as the upper thoracic spine and hearing, as described by D.D. Palmer. 
Patients who attended a larger number of visits over the past month were also more likely to report at least 1 N-MSR. There are 3 possible reasons for this:
(1) several visits are likely to provide the chiropractor with more opportunities to influence the patient mentally,
(2) several treatments may be needed to produce a physiological effect, and
(3) a larger number of visits occur over a longer duration, giving a greater chance for nonmusculoskeletalsymptoms to fluctuate spontaneously, as previously suggested. 
Interestingly, in the multivariable analysis, the effect of “the number of visits”—variable on the reporting of N-MSRs—was not dependent on the 2 chiropractor-specific variables (“subluxation is important” and “information to patients that N-MSRs are likely to occur”). This indicates that the response observed in relation to the number of visits was not directly related to the amount of information received. Further data interpretation fails to strengthen the second possibility because the number of spinal areas treated is unrelated to the reporting of N-MSRs (contrary to the Swedish study). No conclusions can therefore be drawn on this issue.
Other variables examined in this study did not have an effect on the reported outcome. These variables were level of education, type of work, and type of treatment. The latter finding is interesting in that it shows that classical “spinal adjustments” are no more likely to be associated with N-MSRs than, for example, mobilization. The fact that there is no difference between treatment methods tends to weaken the “treatment effect” explanation and strengthens the alternative explanations as described above.
If we assume that all patients, identified in the study to have a nonmusculoskeletal problem, were able to accurately identify specific N-MSRs and if we also assume that those who reported a change in status did so as a result of the treatment, approximately one quarter of patients with breathing or digestive problems became definitely better, whereas only 1% became “definitely worse.” Approximately one fifth of those with circulation problems, ringing in the ears, or asthma reported definite improvement, and 2% or less became “definitely worse.” The proportions of definitely improved patients were approximately 10% or less for the other conditions studied in this study, with “definite” negative side effects never exceeding 2% (Table 2). If these assumptions are correct, it appears that chiropractic care for this type of conditions is only weakly to moderately successful but, at least, rarely harmful.
When interpreting the results of this study, it is important to keep in mind its weaknesses, including the potential for sampling bias (both of chiropractors and of patients), uncertainty surrounding patients' recall ability, problems with accurate description of variables in the questionnaire, and patients' understanding of these, expectation bias that may have arisen before meeting the present chiropractor, and the possibility that N-MSRs may be missed if they require more than 2 weeks to manifest themselves or if patients are unable to identify these themselves. Another obvious limitation is the absence of a control group to compare the results against, which would be necessary to investigate treatment effects. It is therefore not possible to establish whether patients improved (or worsened) because of the treatment, despite the treatment or regardless of the treatment.
On the other hand, practice-based research is more likely to reflect everyday clinical practice than the procedures used and results obtained in “gold standard”–controlled clinical trials, in which highly selected clinicians and patients participate. Furthermore, the methodology adopted in this study makes it possible to obtain a large sample size at a relatively low cost, which allows for meaningful subanalyses through the use of internal control groups. This made it possible for us to test the influence of various factors that could be suspected to influence treatment outcome. These include information to patients regarding nonmusculoskeletal benefits of chiropractic care and the type and area of treatment.
An important quality issue in this type of study, with volunteer participants, is that of the study's external validity. Our study participants consisted of a large proportion of chiropractors who believed firmly in the subluxation and in N-MSRs. It is not known if they are representative of the general chiropractic population. In fact, it is possible that chiropractors who elected to participate in this study were proponents of the concept under scrutiny. In a previous study of selected clinicians, the percentage of patients consulting their chiropractor for a nonmusculoskeletal complaint (10%) was approximately the same as in the present study (8%).  A previous Australian practice-based study noted a change in digestive symptoms similar to our results.  Furthermore, the fact that the outcome pattern in our study was so similar to that of the previous Swedish study, which was carried out among chiropractors generally without strong convictions concerning nonmusculoskeletal effects, is yet another argument in favor of the possibility that the patient profile in relation to N-MSRs is largely unaffected by the chiropractors' beliefs and attitudes.
An additional method to investigate the external validity in studies in which particular characteristics of some clinicians can result in a nonrandom aggregation of patients with particular features is to control statistically for the “clustering effect.” In our study, such clustering effect could arise as a result of the research officers' choice of participating chiropractors and also because of the individual chiropractors' selection of patients. According to our results, the variable “treating chiropractor” did not remove any of the previously noted associations with the outcome variable, but the variable “research officer” did have the effect of removing the link between “subluxation” and the outcome variable. This means that at least some research officers invited chiropractic participants who were similar in their beliefs on the subluxation issue.
The findings in the present study were largely similar to those of the previous Swedish study. A minority of patients with self-reported nonmusculoskeletal symptoms report definite improvement after chiropractic care, and very few report definite worsening. Some factors relating to the chiropractor, the treatment, and the patient were found to be weakly associated with the outcome but these factors “explained” only a small fraction, approximately 3%, of the variance.
It is recommended that further research in this area would concentrate on specific disorders that are most likely to produce positive results, such as specifically identified subgroups of digestive problems or tinnitus, and that such research, whether purely experimental or clinical, use stringent research criteria such as random allocation, objective measurements, sham treatment, and observer blindness.