Cheryl Hawk, D.C., Ph.D., Raheleh Khorsan, M.A., Anthony J. Lisi, D.C.,
Randy J. Ferrance, D.C., M.D., and Marion Willard Evans, D.C., Ph.D., C.H.E.S.
Cleveland Chiropractic College,
Kansas City, Missouri 64131, USA.
OBJECTIVES: (1) To evaluate the evidence on the effect of chiropractic care, rather than spinal manipulation only, on patients with nonmusculoskeletal conditions; and (2) to identify shortcomings in the evidence base on this topic, from a Whole Systems Research perspective.
DESIGN: Systematic review.
METHODS: Databases included were PubMed, Ovid, Mantis, Index to Chiropractic Literature, and CINAHL. Search restrictions were human subjects, peer-reviewed journal, English language, and publication before May 2005. All randomized controlled trials (RCTs) were evaluated using the Scottish Intercollegiate Guidelines Network (SIGN) and Jadad checklists; a checklist developed from the CONSORT (Consolidated Standards of Reporting Trials) guidelines; and one developed by the authors to evaluate studies in terms of Whole Systems Research (WSR) considerations.
RESULTS: The search yielded 179 papers addressing 50 different nonmusculoskeletal conditions. There were 122 case reports or case series, 47 experimental designs, including 14 RCTs, 9 systematic reviews, and 1 a large cohort study. The 14 RCTs addressed 10 conditions. Six RCTs were rated “high” on the 3 conventional checklists; one of these 6 was rated “high” in terms of WSR considerations.
(1) Adverse effects should be routinely reported. For the few studies that did report, adverse effects of spinal manipulation for all ages and conditions were rare, transient, and not severe.
(2) Evidence from controlled studies and usual practice supports chiropractic care (the entire clinical encounter) as providing benefit to patients with asthma, cervicogenic vertigo, and infantile colic. Evidence was promising for potential benefit of manual procedures for children with otitis media and elderly patients with pneumonia.
(3) The RCT design is not necessarily incompatible with WSR. RCTs could improve generalizability by basing protocols on usual practice.
(4) Case reports could contribute more to WSR by increasing their emphasis on patient characteristics and patient-based outcomes.
(5) Chiropractic investigators, practitioners, and funding agencies should increase their attention to observational designs.
From the Full-Text Article:
The increasing emphasis on evidence-based health care decision-making requires providers to understand the documented outcomes of their treatments. To better inform this decision-making, the Council on Chiropractic Guidelines and Practice Parameters (CCGPP) developed a process for evaluating the evidence for chiropractic care. Teams of experts on methodology and practice were formed to address various categories of conditions. This paper reports the results of the compilation of evidence related to chiropractic care for patients with nonmusculoskeletal conditions. We defined these, for this review, as conditions in which the primary symptoms are not related directly to the spine or musculature. For operational purposes, our review specifically excluded headaches, for two reasons: First, headaches were included in the CCGPP category of “cervical spine,” and so were addressed by that team; although migraines may not be of musculoskeletal origin, they are often included in headaches studies, along with tension headaches, and it would be difficult to effectively tease out the nonmusculoskeletal and musculoskeletal components.
Second, the topic of manipulative treatment of headache is quite extensive, and would result in an unmanageably large paper if combined with the nonmusculoskeletal literature in general. Previous papers addressing this topic have relied primarily on the results of randomized controlled trials (RCTs), and, because of the paucity of such studies, have concluded that evidence is insufficient. [1, 2] However, recently there has been protest within the scientific community against the near-total reliance on RCTs as a source of evidence.  Particularly for “complementary and alternative medicine” (CAM) practices, observational studies reflecting usual practice are gaining credibility.  This is especially relevant to “body-based” practices, which do not lend themselves readily to blinding. In its 2005 report on CAM, the Institute of Medicine recognized the need to develop scientifically rigorous, yet appropriate, methods to study CAM.  Whole systems research (WSR) is a burgeoning methodological perspective that addresses this need.  It emphasizes the importance of “model validity,” that is, congruence between research methodology and the paradigm of the system being investigated.  Demonstrating the promising nature of WSR, the National Center for Complementary and Alternative Medicine cosponsored a symposium on WSR in 2002.  Application of WSR methods to chiropractic research is as yet only theoretical. 
Therefore, we attempted not only to evaluate papers in accordance with conventional standards, but also to view them through a WSR perspective. The specific aims of this review were to (1) evaluate the published evidence on the effect of chiropractic care, rather than spinal manipulation only, on patients with nonmusculoskeletal conditions; and (2) identify specific shortcomings in the evidence base on this topic, with respect to developing a whole systems approach to research on the effects of chiropractic care.
There are several limitations to this study. First, the number of studies on chiropractic care and/or SMT and other manual therapies for patients with nonmusculoskeletal conditions is relatively small, and the quality of the studies is generally not high. The literature selection was limited to English. It is possible that some studies were missed; however, we used hand searching and input from content experts to ensure a comprehensive search. Another limitation is the possibility of bias in evaluating the studies. We attempted to avoid this by using accepted checklists. A specific limitation to the WSR checklist is that it has not been validated; it must only be viewed as a first attempt to developing a systematic method of representing a WSR perspective.
Implications for chiropractic practice
We have drawn several conclusions, from a pragmatic perspective, regarding our first specific aim, to evaluate the published evidence on the effect of chiropractic care on patients with nonmusculoskeletal conditions.
The adverse effects reported for SMT for all age groups and conditions were rare and, when they did occur, transient and not severe.
Evidence from both controlled studies and usual practice is adequate to support the “total package” of chiropractic care, including SMT, other procedures, and unmeasured qualities such as belief and attention, as providing benefit to patients with asthma, cervicogenic vertigo, and infantile colic.
Evidence was promising for the potential benefit of manual procedures for children with otitis media and for hospitalized elderly patients with pneumonia.
Evidence did not appear to support chiropractic care for the broad population of patients with hypertension, although it did not rule out the possibility that there may be subpopulations of hypertensive patients who might benefit.
Evidence was equivocal regarding chiropractic care for dysmenorrhea and premenstrual syndrome; it is not clear what level of biomechanical force is most appropriate for patients with these related conditions. It does appear that an extended duration of care, over at least 3 menstrual cycles, is more likely to be beneficial.
There is insufficient evidence to make conclusions about chiropractic care for patients with other conditions.
Implications for whole systems research in chiropractic
Regarding our second specific aim, to identify specific shortcomings with respect to developing a whole-systems approach to research on the effects of chiropractic care, we have identified the following issues:
All studies, from case reports to RCTs, should routinely report adverse effects.
Most published RCTs investigating chiropractic care for nonmusculoskeletal conditions have not relied on usual practice in designing their intervention protocols. Some RCTs were designed without benefit of any published observational studies, case series, or case reports. Even in the absence of observational studies, it is possible to demonstrate that the protocol represents usual practice; for example, the Olafsdottir et al.  infantile colic study used a “reference group” of 14 practicing chiropractors to establish the treatment protocol. We recommend that, in the interest of generalizability, investigators carefully review existing observational studies and reports, as well as consult practitioners with experience treating patients with the condition of interest, and design their intervention protocols to reflect these.
Case series and case reports could increase their utility in several ways:
a. Report patient-based outcomes using validated instruments (rather than focusing on clinician-based outcomes);
b. Specifically address occurrence of adverse effects;
c. Describe patient characteristics in greater detail;
d. Routinely include measures of expectation, satisfaction, and other attitudinal assessments.
The RCT design is not necessarily incompatible with WSR. For example, 1 of 6 RCTs scoring high on conventional RCT checklists also scored high with our preliminary list of WSR considerations. Considerations in designing RCTs that are both rigorous by conventional standards yet are consistent with WSR are as follows:
a. In reporting the results of intervention studies, investigators should specify whether care was provided free of charge and/or patients received incentives for participating. Cost is an important consideration, and free care and/or incentives may affect the generalizability of results.
b. As described above, RCT protocols should have greater reliance on procedures and treatment schedules found in usual practice.
c. “Real-life” comparison groups such as no-treatment or standard care are more generalizable; furthermore, using soft-tissue treatment or other procedures that are also used in everyday practice as shams or placebos may confound results.
d. Routinely including patient-based functional outcome measures, satisfaction, and quality of life provides more multifactorial information on treatment effects.
e. Routinely including measures of patient and practitioner preference and expectation provides important information on psychosocial aspects of the clinical encounter that may affect outcomes.
Educate chiropractic investigators, practitioners, and funding agencies as to the value (or in some cases, the existence of) observational designs such as cohort and case–control studies, to avoid use of scarce resources on premature and sometimes poorly conceived RCTs.
Some of the initial work involved in this project is related to the Council on Chiropractic Guidelines and Practice Parameters (CCGPP). We would like to thank John Triano, D.C., Ph.D., CCGPP Research Commission Chair, and Alan Adams, D.C., M.S., M.S.Ed., Research Commission Vice Chair, for their work in developing the groundwork for the CCGPP scientific process. However, this paper represents only its authors’ views, not those of the CCGPP.
We would like to thank Russell Iwami, M.L.S., at National University of Health Sciences library and Diana Salinas, Linda Horat, and Nehmat Saab, M.A., M.L.S., at Southern California University of Health Sciences library for their essential, and generous, contribution to the literature search for this project. Without them this review would not have been possible. We thank Ronald Rupert, M.S., D.C., Parker Research Institute, for contributing his expertise to the literature search. We also thank Maria Dominguez of the Parker Research Institute, Anupama KizhakkeVeettil, BAMS (Ayu), MAOM, of Southern California University of Health Sciences, and Denise Graham of Cleveland Chiropractic College for their assistance in paper retrieval and data management.