J Manipulative Physiol Ther 2005 (Oct); 28 (8): 555–563 ~ FULL TEXT
Mitchell Haas, DC, Rajiv Sharma, PhD, Miron Stano, PhD
Center for Outcomes Studies,
Western States Chiropractic College,
Portland, OR 97230, USA.
OBJECTIVES: To identify relative provider costs, clinical outcomes, and patient satisfaction for the treatment of low back pain (LBP).
METHODS: This was a practice-based, nonrandomized, comparative study of patients self-referring to 60 doctors of chiropractic and 111 medical doctors in 51 chiropractic and 14 general practice community clinics over a 2-year period. Patients were included if they were at least 18 years old, ambulatory, and had low back pain of mechanical origin (n = 2780). Outcomes were (standardized) office costs, office costs plus referral costs for office-based care and advanced imaging, pain, functional disability, patient satisfaction, physical health, and mental health evaluated at 3 and 12 months after the start of care. Multiple regression analysis was used to correct for baseline differences between provider types.
RESULTS: Chiropractic office costs were higher for both acute and chronic patients (P < .01). When referrals were included, there were no significant differences in either group between provider types (P > .20). Acute and chronic chiropractic patients experienced better outcomes in pain, functional disability, and patient satisfaction (P < .01); clinically important differences in pain and disability improvement were found for chronic patients only.
CONCLUSIONS: Chiropractic care appeared relatively cost-effective for the treatment of chronic LBP. Chiropractic and medical care performed comparably for acute patients. Practice-based clinical outcomes were consistent with systematic reviews of spinal manipulation efficacy: manipulation-based therapy is at least as good as and, in some cases, better than other therapeusis. This evidence can guide physicians, payers, and policy makers in evaluating chiropractic as a treatment option for low back pain.
From the Full-Text Article:
Back pain is experienced by 80% of adults during their lives [2, 3] and accounts for 2.5% of US health care expenditures.  Arguably, the relative cost-effectiveness of medical and chiropractic care is an urgent economic and health policy issue, one for which evidence is especially limited. Much of the recent work on cost-effectiveness has been conducted abroad. [25, 26, 27] With cost structures in the United States that are very different from other countries, [48, 49] our work fills important information gaps that can help with policy and health plan decisions. We include a broad set of outcomes indicators as well as comprehensive cost data for large samples of patients. Furthermore, we have been able to adjust both costs and outcomes for a variety of confounding factors to provide clear relative cost indicators.
Our study had several important findings. First, office costs alone are not appropriate outcomes for a comparison of medical and chiropractic care. Medical office costs do not include physical therapy, whereas physical modalities are usually performed in chiropractic offices.  These and other referral costs (advanced imaging and other provider care) appear to be the great equalizers for medical and chiropractic care. The appropriateness of advanced imaging and referral were not investigated in this study. Clearly, over- and underuse could have a dramatic effect on relative cost-effectiveness.
Chiropractic appears relatively cost-effective compared with medical care for the treatment of chronic LBP in pain and functional disability improvement. This was evidenced by a relative clinical benefit, particularly in the short term, concomitant with no difference in total costs. The picture for acute patients is somewhat less clear. There was only a small advantage for chiropractic care in outcomes with additional but statistically insignificant costs.
Two recent randomized trials addressed cost-effectiveness of manipulation/chiropractic care. Using a formal analysis, a trial in the United Kingdom found that manipulation is cost-effective for back pain.  Kominski et al  found, at an 18-month follow-up, that chiropractic care was more expensive than medical care, but chiropractic care with physical modalities was less expensive than medical care with physical therapy. Outcomes were comparable across the 4 groups. This study supports our contention that ancillary care such as physical modalities need to be considered in cost-effectiveness studies. The absence of group differences in outcomes at 18 months is consistent with our study findings reported previously; chiropractic and medical care differences vanished between 12 and 24 months. 
Although most cost comparisons have been favorable to chiropractic, several studies for the United States have reported that chiropractic care costs more than treatment provided by primary care physicians. [19, 20] For example, general practitioners had the lowest charges over episodes of care, with DCs and orthopedists the highest, in a study using 1974 to 1982 data from the RAND Health Insurance Experiment. 
In particular, our findings were in contrast to the seminal, nonrandomized comparative study by Carey et al,  who found equivalent outcomes but the highest costs for urban DCs and orthopedists and the lowest for primary care and health maintenance organizations. However, their cost data reflected charges rather than payments, which are often much lower than charges. Their costs were also evaluated for a single episode, rather than a fixed period. Many investigators believe that the episode is the appropriate unit of analysis.  However, costs over a fixed period capture recurrences and, thus, may be the more practical approach from the perspective of payers and policy makers.
Our results were consistent with Carey et al  and a trial by Cherkin et al  in finding greater satisfaction with chiropractic care than with other interventions. We do not know how to value satisfaction against costs at this time but feel that satisfaction is an outcome that merits consideration in cost-effectiveness studies.
The RAND  study provides an example of cost-minimization analysis, a method that is,  “appropriate if the alternatives have identical consequences” including “side effects and adverse events.” Despite these caveats, cost minimization has been the dominant methodology used in US cost analyses. In a subsequent example, patients with back and neck pain treated by chiropractors in one health maintenance organization had lower costs than those treated by other providers.  The authors recognized that they did not control for differences in comorbidities, chronic illnesses, or severity but only inferred from other data that there were no substantial differences in underlying illnesses.
A more widely cited study applied an incremental spending methodology to a large database of fee-for-service patients with LBP. [17, 18] Chiropractic users had far lower outpatient and total costs for their episodes of care than nonusers. Although the analysis included controls for differences in patients' insurance and sociodemographic characteristics, controls for the severity of the condition and health status of the patient were limited. The study also did not include any patient outcomes measures. In the large managed care network study in California, where members with chiropractic coverage showed lower annual health care expenditures and lower use rates per episode of back pain than those without chiropractic coverage,  there were no patient outcomes measures that could lead to stronger evidence of chiropractic's relative cost-effectiveness. Our contribution examined both costs and outcomes to report results through easily understood incremental cost-effectiveness ratios.
Nevertheless, several limitations may have affected the study outcomes and generalization of findings. It is well-known that observational studies are more susceptible to bias than randomized controlled trials from unknown factors associated with patients and providers. Control for relevant confounding variables would have the greatest validity in inferring that the costs and outcomes are not attributable to other extraneous factors in observational studies. [32, 33] Our study statistically controls for a broad set of potentially confounding variables to evaluate cost and effectiveness in actual practice when patients can select the providers of their choice. A well-designed observational study can thus overcome a major weakness of randomized trials, their artificial design and limited generalization to clinical practice.  Only large, pragmatic, randomized trials that do not control patient management can yield more accurate estimates of adjusted cost and outcomes differences between medical and chiropractic care.
Hospitalization/surgical costs were not available for our analysis. Because there was a greater referral rate for surgical evaluation from MDs and the hospitalization rate is known to be higher for medical patients,  it is likely that inclusion of hospitalization/surgery would have increased medical costs disproportionately.
Over-the-counter (OTC) drug costs were also excluded from the analysis. We found OTC drug costs difficult to estimate, because the data collected did not account for the large variation in drug type and pill dosage. Drug costs appeared to be relatively small compared to provider costs, so bias was probably small. It is unknown whether there was differential consumption of OTCs between chiropractic and medical patients.
Caution must be taken in generalizing study findings from a regional study to national practice. Chiropractic scope of practice varies from state to state,53 permitting different modalities for the treatment of LBP. For example, Oregon's scope of practice included physical modalities, whereas neighboring Washington's did not. Caution must also be used in light of the continual evolution in health care financing and reimbursement mechanisms. The study controlled for some differences in patients' insurance characteristics, and these results will be reported elsewhere. However, the study design, conceived in the early 1990s, did not anticipate the extent of the shift toward managed care or of other developments such as consumer-driven health plans.
This study supports the generalizability of systematic reviews of the efficacy of spinal manipulation for pain and functional disability to the effectiveness of chiropractic care in clinical practice. Our findings are consistent with the review findings that spinal manipulation–centered therapy is as least as good as, and in some cases, better than other treatments of LBP. [8, 9, 10] Although randomized trials found an advantage for chiropractic care in costs, our study leaned toward comparability.
Chiropractic patients with chronic LBP showed an advantage over medical patients in pain, disability, and satisfaction outcomes without additional costs. Chronic pain and disability outcomes were clinically important in the short term and of lesser magnitude in the long term. Satisfaction with chiropractic care was considerably greater for both acute and chronic patients at both time points. Although the advantages in pain and disability were small for acute patients with LBP, it is important to consider that these gains can be obtained with, at most, small increased costs. With their mission to increase value and respond to patient preferences, health care organizations and policy makers need to reevaluate the appropriateness of chiropractic as a treatment option for LBP.
Chiropractic care is relatively cost-effective compared with primary medical care for the treatment of chronic LBP, particularly in the short term.
Chiropractic and medical care are comparable in cost and effectiveness for acute LBP.
Healthcare organizations and policy makers should consider the appropriateness of chiropractic as a treatment option for LBP.