J Manipulative Physiol Ther. 2001 (Sep); 24 (7): 433–439 ~ FULL TEXT
Joanne Nyiendo, PhD, Mitchell Haas, DC, Bruce Goldberg, MD, and
Gary Sexton, PhD
Western States Chiropractic College,
Division of Research,
Center for Outcome Studies,
Portland, OR 97230, USA.
BACKGROUND: Few studies exist on the prognostic value of demographic, clinical, or psychosocial factors on long-term outcomes for patients with chronic low back pain.
OBJECTIVE: This study reports on long-term pain and disability outcomes for patients with chronic low back pain, evaluates predictors of long-term outcomes, and assesses the influence of doctor type on clinical outcome.
METHODS: Sixty chiropractic (DC) and 111 general practice (MD) physicians participated in data collection for a prospective, longitudinal, practice-based, observational study of ambulatory low back pain of mechanical origin. The primary outcomes, measured at 6 months and 12 months, were pain (by using the Visual Analog Scale), and functional disability (by using the Revised Oswestry Disability Questionnaire). Satisfaction was a secondary outcome.
RESULTS: Overall, long-term pain and disability outcomes were generally equivalent for patients seeking care from medical or chiropractic physicians. Medical and chiropractic care were comparable for patients without leg pain and for patients with leg pain above the knee. However, an advantage was noted for chronic chiropractic patients with radiating pain below the knee after adjusting for baseline differences in patient and complaint characteristics between MD and DC cohorts (adjusted differences = 8.0 to 15.2; P <.002). A greater proportion of chiropractic patients were satisfied with all aspects of their care (P =.0000). The strongest predictors of primary outcomes included an interaction of radiating pain below the knee with provider type and baseline values of the outcomes. Income, smoking, comorbidity, and chronic depression were also identified as predictors of outcomes in this study.
CONCLUSION: Chiropractic care compared favorably to medical care with respect to long-term pain and disability outcomes. Further study is required to explore the advantage seen for chiropractic care in patients with leg pain below the knee and in the area of patient satisfaction. Identification of patient and treatment characteristics associated with better or worse outcomes may foster changes in physicians' practice activities that better serve these patients' needs.
Key Indexing Terms: Low Back Pain, Predictors, Outcomes, Medical Physicians, Chiropractic
From the FULL TEXT Article:
Eighty percent of adults experience back pain at some time in their lives. [1, 2] Population studies report a 1–month period prevalence of up to 40%. [3, 4] Most persons will cope with their back pain without seeking medical intervention, [5, 6] but 1 in 4 do seek treatment from a health care provider. Once the decision has been made to seek professional care, 70% of patients will go to either a primary care medical physician or a chiropractor. 
In the case of chronic low back pain, the level of careseeking is high.  A North Carolina study reported that, of those seeking care for chronic low back pain, 91% saw a medical doctor and 25% saw a chiropractor.  However, neither chiropractors nor primary care medical physicians have yet been able to successfully predict which patients are most likely to benefit from their care. [8, 9] The enormous personal, financial, social, and societal consequences of chronic low back pain have heightened interest in identifying predictors of outcomes. [1, 10, 11] Studies have variously identified complaint duration, history of low back pain, gradual onset, severity of functional impairment, and radiation of pain into one or both legs as important prognostic indicators. [12–19] More recently, the relationship between psychosocial characteristics and outcomes has also been recognized. [15, 16, 19–22]
By using a practice-based, observational design, [23, 24] we conducted a prospective study of the primary patient outcomes of pain and disability in patients seeking care for ambulatory low back pain from primary care medical physicians (primarily family practice physicians) and chiropractors. We included assessment of a broad range of sociodemographic, psychosocial, health status, and care-seeking variables. Patients were followed for 1 year. This article describes 6–month and 1–year outcomes for pain, disability, and satisfaction in patients with chronic low back pain. It also describes predictors of these outcomes based on initial presentations. Patient-reported complaint characteristics, psychosocial factors, and demographic characteristics are the basic elements of the predictive models. We address the need to evaluate predictors of long-term outcomes for ambulatory patients with low back pain, to evaluate prognostic factors for patients with chronic low back pain, and to assess the influence of doctor type on clinical outcome.
The first aim of this study was to identify long-term pain and disability outcomes for patients with chronic low back pain seeking care from community-based MDs and DCs. Because patients with radiating leg pain often have a more prolonged recovery and tend to be more disabled than patients with uncomplicated low back pain, [36, 37] stratification by level of leg pain was used to adjust for case-mix differences in the MD and DC groups. This strategy was suggested by Selim in his 1998 report of findings on patients with chronic low back pain in the Veterans Health Study.  Interestingly, in our study, the patients who appear to have received the greatest comparative benefit from chiropractic care were those who had radiating leg pain below the knee. Much better pain and disability outcomes were realized for chiropractic patients at 6 months and at 1 year. Patients with leg pain above the knee and those without leg pain generally had equivalent outcomes in the 2 provider groups. Large group differences might be found in the short term, but are much less likely to continue in the long term, where natural history clearly plays an important role. It is interesting to note that, in this study, clinical benefit appears to stabilize at 6 months, with little improvement thereafter.
Our finding of an advantage for chiropractic care in the most disabled group — the subgroup of patients with radiating pain below the knee — is intriguing and requires further study, including randomized clinical trials, because we can never be certain that we took into account all clinically important baseline differences between medical and chiropractic patients. There may be factors involved in patient progress, especially with the most severely disabled patients, that have a negative impact on outcomes. An influential predictor, if present and unaccounted for in data collection or statistical adjustment, could have resulted in poorer outcomes for the MD patients. Asking questions of these data leads to still more questions. We wonder whether the subgroup of chronic medical patients with pain below the knee are so much more impacted by pain and disability than chiropractic patients with pain below the knee that, despite treatment rendered, they are unable to get better. Alternatively, the differences in outcomes may be the result of some effect of chiropractic care. Finally, it remains to be investigated whether psychosocial or experiential factors may have been operative in this patient cohort (pain below the knee). It is not known whether different clinical experiences, such as number of visits or nature of the doctor-patient interaction, could account for the differences in outcomes.
Patients with radiating pain below the knee have the poorest prognosis, and the subset of this group with chronic low back pain are often the most difficult to treat. [36, 39] We plan to investigate this further by analyzing care provided (data were collected on practice activities) for this patient subgroup.
The second aim of this study was to identify independent predictors of long-term outcomes for each provider group based on initial presentations. Duration of pain at the initial visit has been identified as a prognostic indicator in most earlier studies, [12–18, 20, 21] and chronicity is associated with poorer outcomes for low back pain patients. Thus, we chose to restrict these analyses to patients with chronic presentation. The strongest predictors of patient outcomes included an interaction of radiating pain below the knee with provider type, and the baseline values of the outcomes.
The literature is inconclusive on the prognostic value of leg pain, [12–15, 19, 22, 40, 41] but is fairly constant in reporting that the most consistent predictors of outcomes are pain intensity and disability at the initial visit. [12, 15, 17, 18, 20, 21] Our findings concur with these previous studies; baseline pain and disability are strong predictors for chronic patients. This study also extends the application of previous reports to long-term outcomes.
Health status and outcomes were correlated, but the regression model retains comorbidity (related to general health) and depression (related to mental health) instead of health status. Two previous studies identified depression as a prognostic indicator for poor outcomes in acute patients. [16, 19] Reports are mixed on the prognostic value of a variety of other psychosocial indicators. [15, 20–22] Although smoking and income were identified as predictors of outcomes for patients with chronic low back pain in this study, previous reports have been inconsistent on their prognostic value. [15, 16, 42]
We acknowledge certain limitations in this study. The participating doctors comprised a convenience sample. It is not known whether the doctors or their patients are entirely representative of community-based practice. As a practice-based study, subjects are, by design, those patients who seek care from the participating physicians and clinics. Self-selection to a treatment group is a known bias in observational studies. One can never be sure that all of the vital factors affecting patient outcomes are known and are reflected in the data. The small numbers of patients with pain below the knee limits generalization of findings. Finally, as an observational study, we cannot presume cause and effect.
With the exception of leg pain and its interaction with provider type, the predictors of outcomes in our study generally correspond to those reported in previous studies of primary care patients. It should be noted that, in these models, history was not generally a strong predictor independent of the other variables investigated. Studies of prognostic indicators often differ in design, recruitment site, patient eligibility, patient characteristics, operational definitions, outcome measures, follow-up periods, loss to follow-up, sample size, and statistical modeling techniques used. These variations may confound interpretation when trying to apply findings to the clinical setting. Still, these studies are important because they can assist physicians in predicting and discussing prognosis with the patient.