Efficacy of manual therapy.

Key Words: Efficacy, Manipulation, Manual therapy, Mobilization.

The therapeutic management of pain, limited motion, and poor posture often involves the use of manual therapy. Many techniques are considered manual therapy procedures, and these techniques often include soft tissue manipulations,[1] massage,[2] manual traction,[3] joint manipulation (short- or long-lever dynamic thrust),[4-6] and joint mobilization.7,8 Clinical interventions using joint mobilization or manipulation have been developed or refined by many authors.[7-15] Joint mobilization is a form of manual therapy that involves low-velocity passive movements within or at the limit of joint range of motion (ROM). In contrast, joint manipulation involves a high-velocity thrust to a joint so that the joint is briefly forced beyond the restricted ROM. In recent years, entry-level physical therapy curricula have expanded the student's exposure to joint mobilization techniques[l6,17] and practicing therapists have enrolled in a greater number of continuing education courses that focus on joint mobilization.[16] The application of manual therapy has been advocated not only for patients with orthopedic disorders but in certain conditions involving central nervous system deficits.[18,19]

In spite of the widespread use of manual therapy in clinical settings, very little is known about the efficacy of these procedures.[20-24] Studies supporting the efficacy of manual therapy have been criticized for not utilizing proper controls or statistical analyses.[25-28] Therefore, the notion of efficacy--whether manual therapy alleviates symptoms of musculoskeletal dysfunction--may not be clearly established in any single investigation because of various flaws in the research design or study procedures (see Di Fabio[23] and discussion below).

Literature reviews on the subject of manual therapy can potentially resolve questions related to efficacy if the review applies criteria that are sensitive to identifying valid clinical trials.[21] Quantitative reviews (meta-analysis) have advantages over traditional literature reviews in that the analytic review process reduces reviewer bias and establishes a measure for the size and statistical significance of the treatment effect.[24,29] The disadvantages of quantitative reviews, however, are that the independent variables are assumed to be similar in order to combine the hypotheses of many studies into one large study group. Given the relatively limited number of controlled clinical trials involving manual therapy, a meta-analysis may combine dissimilar interventions (eg, mobilizations and manipulations) in order to generate meaningful effect sizes through analysis of a critical mass of studies. In addition, the statistical techniques may not be readily available to consumers of manual therapy research.

Several authors[2l,25,27,28,30,31] have proposed various criteria that are intended to assess the validity of a clinical study. in essence, these criteria can be helpful in distinguishing "useful" from "useless" therapy by evaluating claims of efficacy weighed against the integrity of the research design. For example, Deyo[2l] found that many studies of conservative therapies for low back pain failed to randomize subjects, use "blind" raters, measure compliance, or adequately describe cointerventions. Greenland et al[3l] recommended that researchers conducting studies of manual therapy report a list of criteria designed to exclude patients with possible contraindications for manual therapy (ie, osteoporosis), develop accurate descriptions of the treatment interventions, include vocational and psychological profiles of the sample, and study a sufficient number of subjects to enhance the probability of detecting significant differences between control and experimental groups. Gilbert[30] also suggested that the reliability of outcome measures (ie, lift capacity scores, ROM, and pain intensity measures) be reported in order to identify appropriate dependent variables that are also replicable.

The purposes of this article are (1) to establish objective criteria for judging the validity of manual therapy research, (2) to identify and discuss the results of those trials that were determined to be valid demonstrations of treatment efficacy or valid demonstrations of nonuseful therapy, and (3) to determine whether patients who benefit from manual therapy have common characteristics. Efficacy criteria from several previous reviews[2l,24,30,31] were selected in order to evaluate manual therapy research. The literature analysis is presented in two primary sections. Part 1 is a discussion of operational definitions and rationale for selection of specific criteria. Part 2 is an analysis of manual therapy literature based on the systematic application of the efficacy criteria defined in part 1.

Part 1: Defining Selection and Efficacy Criteria

This review will focus on a discussion of manual therapy literature addressing the treatment of somatic pain syndromes (ie, back, neck, and joint pain). To be eligible for selection, a report must have been published in a peer-reviewed journal. Studies published in preliminary form such as abstracts were not included in the literature analysis. In order for the results of a primary investigation to be considered a valid demonstration of the efficacy of manual therapy, the study had to meet the following criteria:

1. Randomization. Concurrent comparisons of randomized groups that received and did not receive manual therapy are an essential element of a valid efficacy study because this procedure reduces confounding effects related to spontaneous recovery, differences in the type of patient, or differences in the prognosis for recovery.

2. Blind Outcome Assessment. Blinded outcome assessment prevents bias based on the investigator's knowledge of patient treatment. Two levels of blinded assessment are possible in a clinical trial of manual therapy[32]: (1) The individual evaluating outcome measures may not know what treatment the patient received, and (2) the patient may not know that he or she received a specific type of therapy (eg, massage versus joint mobilization). Studies that meet both criteria are defined as "double-blind" studies (see, for example, Hoehler et al[32] and Sloop et al33). Reports of studies included in the analysis of literature for this review at a minimum must have included a blinded assessment of relevant outcome measures (eg, pain status, flexibility, time out of work, costs associated with care). Reports of mailed survey questionnaires that measured patient perceptions of treatment benefits, or situations in which subjects independently completed a form describing pain or disability, were judged to have adequately met this criterion, provided that no subjective interpretation by the rater was necessary to score the assessment. For example, Meade et al[34] used the Oswestry back pain questionnaire, which measures pain status and the ability to complete selected functional activities.[35] The weighted value of each item was predetermined. The researchers minimized bias by periodically mailing the questionnaires to all subjects during a follow-up period lasting 2 years.

3. Criteria for Selecting Subjects. There is considerable variability in the course of syndromes associated with spinal pathology.[22] The lack of consensual research findings[24] May be due to the difficulty diagnosing the precise source of the symptoms and a high incidence of both spontaneous recovery and reoccurrence of symptoms.[26,36] In addition, objective tests of spinal function may not reflect organic changes in the neuromusculoskeletal system because of the confounding influences of excessive illness behavior[37] or emotional stress.[38] Therefore, a description of the criteria for selecting subjects to participate in the study of manual therapy (or for eliminating subjects from participation) was required.

4. Description of Intervention. A sufficient description of the intervention was required to properly identify and categorize studies using mobilization (low-velocity joint movements within the available ROM), manipulation (high-velocity thrust momentarily exceeding the available ROM), or some combination of these two manual therapies. Nwuga,[39] for example, used a "lumbar oscillatory rotation ... in a push-relax sequence" that fits the operational definition for mobilization stated above, In contrast, Hoehler et al[32] described a rotational manipulation to the lumbosacral spine using a "short, high-velocity thrust" that met the operational criteria for manipulation. The use of the term "manipulation," however, did not always correspond with the operational definition. Coxhead et al,[40] for example, used the term "manipulation" but described graded articulations that were consistent with the definition of mobilization. Therefore, the description of the procedure took precedent over the label applied to the type of manual therapy. In some cases, the authors cited references that were a compilation of many manual therapy procedures rather than specifically outlining the intervention within the text of the report.[34,41,42] Unless stated otherwise, an author's reference to the treatment techniques of Maitland[7,13] or Kaltenborn[12] was considered primarily mobilization. In contrast, manipulative therapy was attributed to a reference of "chiropractic" treatment. Reports of studies addressing the efficacy of traction (manual or mechanical) without including some other form of mobilization or manipulation were excluded from review.

5. Statistical Analysis. The study had to compare mobilization, manipulation, or a combination of these two interventions in a patient sample with some control or comparison group. The comparisons had to be described in terms of statistical significance (P<.05), using any appropriate statistical test.

6. Statistical Power. A conclusion that manual therapy had no effect on the outcome measures may be incorrect if the sample size was not sufficient. Sample size estimates are dependent on the precision of the scales used to evaluate outcome measures and the power of the statistical test. Greenland et al,[31] for example, estimated that 100 subjects per group were needed to achieve 70% power using a five-point scale to assess changes in pain intensity following manual therapy. For studies demonstrating no difference between experimental and control groups, an estimate of the level of Type 11 error (probability of falsely accepting the null hypothesis) should be reported. in order for a study to be included in the analysis of literature, the author(s) reporting negative findings had to state that a power analysis was completed and that the sample size was adequate to detect significant differences between groups. It is important to emphasize that reports of negative findings might be valid, but this determination cannot be definitive without a power analysis.

Part 2: Evaluation of the Literature

Methods

Potentially relevant studies were identified through a computer-assisted bibliographic search of the Medline database. In addition, a manual examination of the references contained in the retrieved reports provided the location of additional reports. The search procedures yielded a total of 146 nonoverlapping titles with key words that were potentially relevant to the efficacy of manual therapy. Examples of key words were "efficacy," "mobilization,l" "manipulation," "chiropractic," "evaluation," "comparative study," and "single blind." The abstracts or full reports were then evaluated based on the operational definitions of the criteria outlined in part 1. One hundred five of the 146 studies did not address some form of manual therapy as a primary intervention, or the reports were letters to the editor, announcements, special communications, or review articles. Of the remaining 41 studies, 18 studies did not include a control or comparison group, a blinded assessment, or a statistical analysis of outcome measures (Tab. 1). Nine controlled studies yielded negative results, but the authors did not justify the sample size (Tab. 2). Fourteen studies met the inclusion criteria and were considered acceptable for the analysis of literature (Tab. 3). [TABULAR DATA 1, 2, 3, OMITTED]

The literature reviewed in this article was organized by three factors: the anatomical region of intervention, the goal of the research, and the technique of manual therapy (manipulation, mobilization, or a combination of each procedure). The goal of the study was defined as either explanatory or pragmatic.[31] An explanatory goal seeks to separate and then to measure physiological or psychological responses to manual therapy (eg, the amount of joint surface distraction). An explanatory approach might identify some mechanism that helps to explain how manual therapy works. In contrast, a pragmatic goal seeks to determine whether manual therapy "offers relief comparable or superior to that produced by conventional medical intervention."[31(p673)] The "primary" intervention was defined as the manual therapy technique that was compared with some control or comparison group.

No studies that met the criteria for this review could be found in the following categories: (1) studies involving manual therapy of the peripheral joints; (2) explanatory trials using lumbar or cervical manipulation, mobilization, or a combination of both manual techniques for each region; and (3) pragmatic trials involving sacroiliac joint dysfunction (SIJD).

Results

Valid pragmatic studies involving lumbar spine 

1. Manipulation Compared with Mobilization. Meade et al[34] conducted a multicenter trial to compare chiropractic care (n=378) with conventional hospital outpatient management of 339 patients with low back pain. A preliminary analysis was reported by Meade et al.[43] Subjects were excluded from the study if they showed evidence of nerve root compression or irritation, structural deformities of the spine, or osteopenia or if litigation was pending. Subjects were allowed to participate if they were between 18 and 65 years of age and if they had not been treated within the past month. Over half of the subjects in each group reported a duration of symptoms in the current episode of pain greater than 1 month. Hospital-based interventions involved a wide range of procedures including mobilization and manipulation of the lumbar spine as well as exercises, back school, and lumbar traction.[43] In contrast, chiropractic treatment involved manipulation in 99% of the patients treated. Approximately 10 treatment sessions were provided within 3 months of the initiation of the study. Pain status and the ability to complete selected functional activities were measured using the Oswestry back pain questionnaire.[35] Chiropractic manipulation was more effective than hospital outpatient management with respect to decreasing pain and improving mobility.[34] In addition, the benefits of chiropractic care continued to be superior to hospital-based treatments for 6 months to 2 years following intervention. Straight leg raising (SLR) and lumbar flexion showed a greater degree of improvement following chiropractic treatment, but these particular interventions were not assessed blindly.

A series of commentaries highlighted several perceived flaws in the study design and conclusions presented by Meade et al.[34] it was noted that patients seen by chiropractors received more treatments (44%) compared with those seen in the hospital.[44,45] In addition, the early intervention and specialized care provided by chiropractors working in a private clinic were contrasted with longer waiting times for initial appointments and more generalized care provided in a socialized hospital system.[45,46]

In a more focused study comparing the effects of manipulation versus mobilization of the lumbar spine, Hadler et al[47] recruited volunteers by placing advertisements in the newspaper. In order to be eligible for the study, subjects were required to meet six criteria: (1) age between 18 and 40 years, (2) symptoms lasting no more than 1 month, (3) no other episode of back pain within 6 months prior to the study, (4) no workers' compensation and the incident causing the injury could not be work related, (5) no previous treatments involving spinal manipulation, and (6) willing to travel to the test site.

Fifty-four subjects were randomly assigned to one of two groups. One group received manipulation (n=26), and the control group (n=28) was given mobilization of the lumbar spine without a manipulative thrust. A total of five treatment sessions were provided within 2 weeks from the beginning of the project. Subjects who experienced symptoms for only 2 to 4 weeks prior to receiving manipulation reported greater and more rapid improvement of mobility and pain compared with subjects receiving mobilization. These results were based on the subjects' responses to a questionnaire.

Sclerosing injections,* analgesics, and exercise have been applied in conjunction with manipulative therapy in order to develop a broad-based approach for treating patients with chronic low back pain. Ongley et al[48] studied 81 patients with chronic low back pain (>1 year duration) who had not responded to previous nonsurgical management. Subjects were recruited through a mailing to previously registered patients attending a multispecialty clinic. The patients included in the mailing list were randomly selected from a larger population of patients attending the clinic, and the mailings were made without regard to previous complaints of back pain. Patients were rejected from the study if they were not between the ages of 21 and 70 years or if they were involved in litigation or receiving workers' compensation, pregnant or contemplating pregnancy, overweight, or diabetic. Additional reasons for excluding subjects included osteoporosis, alcohol abuse, upper (rather than lower) back pain, cardiac disease, and overt psychopathology. The experimental group (n=40) received sclerosing injections together with a single forceful rotational manipulation of the spine. This procedure was followed by weekly proliferant injections and lumbar flexion exercises for a total intervention dispersed across 6 weeks. The control group (n=41) received a "nonforceful manipulation" (mobilization) of the lumbar spine with injection of a placebo (saline) proliferant and a reduction of analgesics. The experimental group reported greater improvement in disability and pain scores compared with the control group 6 months following treatment. These data were based on the patients' response to a disability questionnaire. The authors speculated that manipulation ruptured microadhesions that formed in response to connective tissue immobilization. It was suggested that the "proliferant" injections then served to reverse the incompetence of lumbar and sacroiliac ligaments by inducing fibroplastic hyperplasia and stimulating growth of collagen.[48]

2. Manipulation Compared with a Control Group. Manipulation appears to have a definite immediate effect on the symptoms associated with low back pain. Sanders et al[49] selected 18 subjects, aged 22 to 56 years, with acute back pain of less than 2 weeks' duration. Subjects were "naive" to chiropractic adjustive manipulation and agreed to discontinue any medications 48 hours prior to the study. The perception of pain indicated by each subject on a five-point visual analog scale was measured before and immediately after lumbar manipulation. The results were compared in an experimental group (n=6), a control group that received no treatment (n=6), and a "sham" group (n=6) that received light touch without manipulation. A slight, but significant, reduction in the pain scores of the experimental group was found immediately following a single high-velocity, low-amplitude adjustive manipulation focused at the L4-5 to S1 segments.

Hoehler et al[32] studied 95 patients with restricted or painful spinal ROM who had no previous experience with manipulative therapy. These subjects were eligible for the study because they had no previous experience with spinal manipulation, were not receiving disability income or involved in litigation, had no previous back surgery, were not obese, and did not report drug abuse. All subjects demonstrated hyperalgesia or restricted/ painful vertebral ROM. Manipulation versus a placebo treatment consisting of soft tissue massage) provided immediate alleviation of low back pain (see also Fig. 1 in Tobis and Hoehler[50]). However, at the termination of treatment, which entailed a "variable" number of sessions, and at 3 weeks following treatment, no difference could be found between groups.

The immediate reduction in pain symptoms following lumbar manipulation has been noted previously (see Fig. 2 in Glover et al[51]) and is consistent with the findings that the resistance to passive stretch of the hamstring muscles was reduced immediately on the painful side following rational manipulation of the lumbar spine compared with the measurements taken from a healthy control group.[52] Glover et al[51] provided only one session of lumbar rotational manipulation and studied only patients n=43), ages 21 to 64 years, with unilateral symptoms consisting of pain between the inferior angle of the scapula and the lower end of the sacrum. Patients were assigned to groups based on having a pain duration of greater than or less than 7 days. Patients with pain lasting less than 7 days were found to report more immediate relief of symptoms. Fisk[52] selected 10 patients, 25 to 55 years of age, who also had unilateral symptoms, but the primary requirement was that the patient have a reduction in passive hamstring muscle stretch on the painful side. The mean duration of symptoms was 15 days, with a range of 2 to 35 days.

Somewhat transient effects following manipulation were also reported by Evans et al.[53] These authors utilized two groups of patients in a simple "crossover" design for two continuous 3-week intervals. The crossover design exposed each group to control and treatment interventions. Those beginning with manipulative treatment during the first 3-week period then received the control regimen, which consisted of no treatment and discontinuation of analgesics, during the second 3-week period. The group initially receiving the control regimen then "crossed over" to receive spinal manipulation for the last 3 weeks of the study. Spinal motion was increased significantly in patients with low back pain during the 3-week period of manipulation, followed by a significant decrease in motion during the 3 weeks without manipulation.[53]

3. Mobilization Compared with a Control Group. Only one acceptable study was located that described the effect of lumbar mobilization with respect to a control group. Nwuga[39] limited his study to patients, ages 20 to 40 years, with disk protrusion confirmed by myelographic and electrodiagnostic tests and nerve root compression inferred by lower-extremity reflex changes. In addition, the patients could have only unilateral symptoms of 2 weeks' duration (maximum) and could not have received prior treatment for the "condition." Lumbar rotatory oscillations combined with a "push-relax" technique were utilized in one group (n=26) to reach each subject's painful ROM. A control group (n=25) received shortwave diathermy and lumbar flexion exercises. All subjects were given instructions for proper lifting and posture. Nwuga[39] reported that the mobilization group showed significant increases in lumbar motion and SLR compared with controls. The functional significance of improvement in the SLR was not described.

4. Combination Manipulation/Mobilization Compared with a Control Group. Farrell and Twomey[54] studied two groups of patients (n=24/group) with a duration of low back pain symptoms of less than 3 weeks. Subjects were accepted to participate in the study if they were 20 to 65 years of age and reported painful lumbar movement or SLR with pain localized in the paravertebral region between the 12th thoracic level and the gluteal folds. Potential subjects were also required to be pain-free for 6 months prior to the onset of the current episode of low back pain. Exclusion criteria included previous treatment for the current episode of pain, pregnancy, previous lumbar surgery, systemic disease, and altered sensation, reflexes, or lower-extremity muscle weakness. The experimental group received a combination of passive mobilization and manipulation, whereas the control group was given microwave diathermy, abdominal exercises, and ergonomic instruction. Patients were considered symptom-free when functional activities could be performed essentially without pain and when measurements of lumbar movement and SLR could be made without a report of pain. The group receiving manual therapy achieved symptom-free status approximately 1 week sooner than the control group.[54] However, 91% of all the subjects recovered from their symptoms within 4 weeks. Others[36,55] have also observed spontaneous recovery in patients with low back pain. Gibson et al[56] have argued that spontaneous recovery may not be attributed to the natural history of low back dysfunction because the outcome of treatment in their trial was unrelated to initial severity, duration of symptoms, or observed trends toward improvement or deterioration at the time of entry into the study. Unfortunately, a power analysis was not reported for this clinical trial (Tab. 2).

Stratification of samples based on key variables may provide a unique perspective regarding the type of patient most likely to benefit from manual therapy. Mathews et al[57] divided a sample of patients, 18 to 60 years of age, with low back pain into two groups: One group of subjects (n=132) had limited passive SLR, and the other group of subjects (n=33) had no SLR limitation. Prospective subjects with abnormal spinal radiograph, blood count, erythrocyte sedimentation rate, and alkaline phosphatase and urine test results were excluded from the study. Mobilization and manipulations were given daily "if indicated" for 2 weeks. Control groups (one group with and the other without limited SLR) were given infrared treatments for 15 minutes, three times weekly. All groups were presented with instructions for proper lifting and "offered a spinal corset." Subjects with SLR limitations showed significantly higher recovery rates based on pain scores compared with subjects in the control group without SLR limitations. In contrast, there was no difference in recovery between manual therapy and control groups for subjects without initial SLR limitations. The relatively small sample size for the experimental group without SLR limitations and the loss of statistical power were discussed as factors contributing to the negative finding in this component of the study.

Valid pragmatic studies involving cervical spine 

1. Manipulation Compared with Mobilization. Parker et al[41] recruited volunteers through the "media" who were less than 55 years of age and had a diagnosis of migraine headache. The duration, chronicity, and frequency of migraine headache were not specified. The authors compared chiropractic manipulation of the spine (n=30) and manipulation performed by medical practitioners (n=27) with spinal mobilization for treatment of migraine headaches (n=28). Improvement was demonstrated in all three groups, and no difference in outcome measures involving pain intensity and disability were found when systematic pair-wise comparisons were completed. A subsequent report[58] suggested that a power analysis was done for the original data. in addition, a reanalysis of the data using different statistical methods apparently did not alter the interpretation of the original results.

2. Manipulation Compared with a Control Group. Howe et al[59] conducted a randomized control trial with a blind assessment of pain perception and cervical motion in 52 patients. The subjects were selected from a private practice and were between 15 and 65 years of age. Subjects were included if they had pain related to the cervical spine and limited cervical segmental motion. Patients with spinal cord involvement, rheumatoid arthritis, cancer, and psychiatric illness were excluded from the study. The selected sample was treated with azapropazone, an anti-inflammatory and analgesic drug. Half of the subjects (n=26) also received cervical manipulation. The group receiving manipulative treatment was found to have a significant improvement in cervical rotation (maintained at a 3-week follow-up assessment), which was on the order of 5 degrees. There was an improvement in lateral cervical flexion and a significant reduction in pain and stiffness involving the neck immediately following treatment, but these effects were not observed at a 1-week follow-up. Howe et al noted that the spontaneous improvement in control subjects who had no simultaneous improvement in cervical motion negated the significance of lasting improvement in those receiving manipulation.

Valid explanatory studies involving sacral spine. Cibulka et al[60] used sacral manipulation in a group of 10 patients, 15 to 47 years of age, with SIJD. The criteria for identifying SIJD were a minimum of three positive test results from a pool of four clinical measures: standing flexion test, prone knee flexion test, supine long-sitting test, and comparison of posterior superior iliac spine heights in the sitting position. Subjects were excluded from participation if they were pregnant; diagnosed with ankylosing spondylitis; had an alteration of lower-extremity reflexes, sensation, or muscle strength; or showed an SLR of less than 45 degrees. In addition, the authors excluded subjects who exhibited nonorganic physical signs associated with their low back pain.

The study by Cibulka et al[60] was classified as "explanatory" research because the authors intended to study alterations in pelvic spatial orientation rather than changes in symptoms induced by manipulation of the innominate bone. This study differs from the citation for Cibulka in Table 1. The degree of innominate bone tilt before and after a single manipulative treatment) was compared with similar measurements in a group of patients with SIJD (n=10) who did not receive manipulation. Subjects in the experimental group showed a significant change in the angle of pelvic inclination when pretest and posttest measurements were compared, When the factors of "side" (manipulated versus nonmanipulated innominate bone) and test (pretest versus posttest) were averaged, however, Cibulka et al[60] were unable to demonstrate a significant main effect for "group" (manipulative therapy versus no manipulative therapy). This finding was partly due to the variability of pelvic inclination measurements observed within the experimental group subjects. Coefficients of variation calculated from their data (see Tab. 1 in Cibulka et al[60]) ranged from 147% to 660%.

Discussion

The results of this review provide clear evidence that manual therapy can be an effective modality when used to treat patients who have somatic pain syndromes. There may be a difference in efficacy, however, between manipulation and mobilization therapy. The vast majority of valid efficacy studies (11/14) involved some form of manipulation as the primary intervention (Tab. 3). In contrast, only 22% (2/9) of the studies with negative results (showing no difference between control and manual therapy groups) utilized manipulation as the primary intervention (Tab. 2). By comparison, there was one acceptable efficacy study in favor of mobilization as a primary intervention[39] versus three studies with negative results (Tab. 2). Therefore, additional work is needed to establish mobilization as an effective therapy.

Further comparisons between studies that demonstrated significanty greater improvement in patients receiving manual therapy compared with controls (Tab. 3) and studies that demonstrated no significant differences between manual therapy and control groups (Tab. 2) may provide some insight concerning (1) the specific manual procedures used to successfully treat patients who have somatic pain syndromes and (2) the type of patient likely to benefit from this form of therapeutic intervention.

Manual therapy techniques vary widely, and several reports have been criticized because descriptions of the manipulative or mobilizing techniques would not allow for replication of the study (see Deyo,[21] Gilbert,[30] Greenland et al,[31] and Hadler et al[47]). The nature of clinical intervention often requires a change in the treatment protocol based on the patient's ongoing response to manual therapy.[13,31] This "ongoing" reevaluation process further increases treatment variability. Nearly all of the reports of acceptable studies in the current literature analysis, however, described mobilization or manipulative procedures with adequate detail.[dagger] The majority of these procedures involved rotational manipulations of the lumbosacral spine. Three valid reports lacked complete treatment descriptions.[34,41,59] Meade et al[34] reported that the treatments were based on the judgment of the chiropractor or hospital team. Although 99% of the group treated by chiropractors received manipulation, the time over which 10 treatments could be given ranged from 3 months to 1 year, at the discretion of the manipulator. The practitioners providing manual therapy in the study by Parker and colleagues[41] were "free to perform" mobilizations or manipulations on any area of the spine. Howe et al[59] maintained that a detailed description of the manipulation was not appropriate to include in their report. Therefore, it was difficult to know precisely what treatment was provided.

Reports of negative findings (Tab. 2) were mixed regarding the specificity of treatment; some reports provided detailed accounts of the manual therapy intervention,[6,61,62] whereas others reported little detail[33,56] or indicated that the technique was at the discretion of the researcher.[63] The absence of standardized treatment protocols may be one factor contributing to negative findings, because inconsistencies between practitioners will introduce an additional source of measurement variability. it should also be recognized, however, that clinical studies involving multiple centers[34,63] may face practical limitations in standardizing procedures across large numbers of practitioners.

In an attempt to identify the type of patient who would most likely experience functional improvement or a reduction in symptoms, common features of the patient samples (reported for two or more investigations) for valid efficacy studies of the lumbar spine were analyzed. Based on a limited number of studies (Tab. 3), a patient profile was developed and is outlined in Table 4. The duration of symptomatic relief varied from immediate improvement following treatment with no difference between manual therapy and control groups over longer terms[32,47,49,51,54] to measurable improvement 2 years following the termination of intervention.[34] Patients were commonly excluded from a study if there was pending litigation or issues involving workers' compensation, radiographic evidence of structural deformity, frank neurological signs (ie, depressed reflexes), or evidence of systemic infection. A positive SLR test alone was not routinely considered to be a reason for excluding subjects from clinical trials, presumably because SLR-evoked symptoms may be caused by mechanisms other than neurological impingement.[25] It should be noted that the characteristics forcing subject exclusion from manual therapy studies need further elaboration. Some of these exclusion criteria are listed in the preliminary profile of subjects most likely to benefit from manual therapy (eg, no frank neurological signs or workers' compensation). It would be premature to infer that patients with neurological signs, for example, could not benefit from manual therapy. The subject exclusion criteria in many studies did not allow for direct testing to make this kind of inference. Therefore, the tentative profile (Tab. 4) will require modification as future investigations systematically include subjects who had previously been excluded from manual therapy research. [TABULAR DATA OMITTED]

Recommendations for Future Manual Therapy Research

Comparisons among studies on low back dysfunction would be facilitated if some uniform screening procedure were used to select subjects for study. Future studies might consider guidelines such as those proposed by Frymoyer et al[64] to evaluate sample characteristics. An overview of this screening format is shown in the Appendix. The procedures used to screen eligible patients should also be coupled with appropriate follow-up intervals for reassessment and a mechanism to evaluate patient attrition.

The clinical application of manual therapy techniques will often be coupled with other therapeutic modalities (or "co-interventions") in an attempt to achieve the best possible outcome. Co-interventions include exercise and ergonomic instruction,[39] heat,[51] and proliferant injection.[48] Multiple interventions, however, may prevent the attribution of positive results to a single isolated procedure. Ongley et al[48] utilized proliferant injection, analgesics, and manipulation in their experimental group and provided controls with a placebo injection, a reduction of pain medication, and a "nonforceful" mobilization of the spine. Both groups received exercise; therefore, the only equivalent cointervention was exercise instruction. Deyo[21] and Gilbert[30] suggested that equivalent co-interventions and a measure of compliance (particularly home exercise programs) are needed to establish a valid clinical trial, and researchers seeking to extend the work of Ongley et al[48] should consider this proposal.

Manual therapy may be considered a "passive" treatment from the perspective that it does not require active involvement from the patient. There is evidence to suggest that patients who take an active role in the rehabilitation process may enhance their chances of recovery.[65] Saal and Saal[65] reported that 92% (n=59) of patients with lumbar radiculopathy returned to work following a treatment program that included a back school and "stabilization" exercise training. When patients receiving workers' compensation were partitioned out of the sample, 86% of those receiving insurance compensation (n=11) returned to Work.[65]

Outcome measures such as return-to-work rate or use of sick leave were used by Saal and Saal[65] and others.[62] These types of outcomes should receive more attention in future studies of manual therapy because they allow for a more meaningful description of the efficacy of manual therapy and collateral interventions. in addition, the reliability of outcome measures should be clearly demonstrated in future studies concerning manual therapy.

Leboeuf[66] found that failure to use blinded assessors and naive study subjects was a common flaw in the study design of contemporary chiropractic research. The possibility that rater bias may influence outcome measures cannot be overlooked when a clinical trial is "open" (not blind). The majority of studies that did not include blinded assessment or statistical comparisons also demonstrated results favoring manual therapy (Tab. 1). This observation is consistent with summaries of other noncontrolled studies cited in previous reviews.[23,25]

Randomization is considered an essential feature of a valid clinical trial.[21,30,31] Previous reviews on the efficacy of spinal manual therapy[23,24] showed that studies implemented without random assignment of subjects to control and manual therapy groups were more likely to produce results supporting the use of mobilization or manipulation. These earlier conclusions must now be modified in view of the current literature analysis. Significant improvements in pain, flexibility, and disability status have since been reported using randomized clinical trials (Tab. 3).

Nearly all of the valid efficacy studies reviewed were pragmatic in nature. There is a clear need for additional explanatory research. Only Cibulka et al[60] completed an explanatory trial, which demonstrated a change in pelvic spatial orientation following manipulation. Alterations in the spatial orientation of the skeletal structure following manipulation have also been reported by Scheibel and Debusschere,[67] but healthy volunteers were studied. Therefore, the study did not meet the criteria for comparing a patient sample with some control or comparison group.[double dagger]

Conclusions

This analysis of literature identified 14 studies that were judged to be valid demonstrations of the efficacy of manual therapy in the treatment of patients who have somatic pain syndromes. Overall, there was clear evidence to justify the use of manual therapy, particularly manipulation, in the treatment of patients who have back pain (Tab. 3). Only 1 of the 14 studies did not support the use of manipulation. There was a paucity of valid explanatory research in all areas and a particular absence of controlled trials involving manual therapy applied to the peripheral joints. As the state of the art of manual therapy continues to develop, future researchers should focus on providing replicable descriptions of treatment interventions and measures of compliance and on designing studies with equivalent co-interventions. Alternative approaches should also be considered such as the use of single-subject research designs that incorporate the subjects as their own controls.[68]

Acknowledgment

I want to express my sincere appreciation to Cindy Kelley, a physical therapy student at the University of Minnesota, for her invaluable assistance with the preparation of this manuscript.

Appendix. Essential Content for Describing Subjects in Clinical Studies of Manual Therapy[a] 1. Basic demographics and patient descriptors (ie, age, sex, height, weight) 2. Taxonomic questions applicable to well-understood pathologies (eg, questions relative to sciatica) 3. Resultant morbidity and mortality (ie, hospitalizations, operations, disabilities) 4. Exposure to drugs, alcohol, cigarettes 5. Other lifestyle habits (eg, frequency and intensity of exercise) 6. Ergonomic information (A) Self-report of "injury" mechanisms with some attempt to quantify work factors involved (B) Job descriptors 7. Pain descriptors (A) Visual analogue scales (B) Pain drawings 8. Psychological covariates (A) Self-report of activities of daily living and disability (B) Measure of depression [a]Modified from Frymoyer et al, Spine. 1991;16:681-682.

References

[1] Cottingham JT, Porges SW, Richmond K. Shifts in pelvic inclination angle and parasympathetic tone produced by Rolfing soft tissue manipulation. Phys Ther. 1988;68:1364-1370. [2] Licht S. Massage, Manipulation, and Traction. New Haven, Conn: Elizabeth Licht, Publisher; 1960. [3] Zylbergold RS, Piper MC. Cervical spine disorders: a comparison of three types of traction. Spine. 1985;10:867-871. [4] Stoddard A. Manual of Osteopathic Technique. 2nd ed. London, England: Hutchinson Books Ltd; 1962. [5] Cyriax JH. Textbook of Orthopedic Medicine, Volume II: Treatment of Manipulation, Massage, and Injection. Baltimore, Md: Williams & Wilkins; 1971. [6] MacDonald RS, Bell CM. An open controlled assessment of osteopathic manipulation in nonspecific low-back pain. Spine. 1990;15:364-370. [7] Maitland GD. Peripheral Manipulation. 2nd ed. Boston, Mass: Butterworth; 1977. [8] Kessler RM, Hertling D. Management of Common Musculoskeletal Disorders. Philadelphia, Pa: Harper & Row, Publishers Inc; 1983. [9] Di Fabio RP. Thoracolumbar spine examination. In: Wadsworth CT, ed. Manual Examination and Treatment of the Spine and Extremities. Baltimore, Md: Williams & Wilkins; 1988: 58-76. [10] Di Fabio RP. Sacroiliac examination. In: Wadsworth CT, ed. Manual Examination and Treatment of the Spine and Extremities. Baltimore, Md: Williams & Wilkins; 1988:76-89. [11] Grieve GP. Mobilization of the Spine. Edinburgh, Scotland: Churchilll Livingstone; 1979. [12] Kaltenborn FM. Mobilization of the Spinal Column. Wellington, New Zealand: New Zealand University Press; 1970. [13] Maitland GD. Vertebral Manipulation. 5th ed. London, England: Butterworth & Co (Publisher) Ltd; 1986. [14] Cyriax JH. Textbook of Orthopaedic Medicine, Volume I: Diagnosis of Soft Tissue Lesions. 8th ed. London, England: Bailliere Tindall; 1982. [15] McKenzie RA. The Lumbar Spine: Mechanical Diagnosis and Therapy. Waikanae, New Zealand: Spinal Publications; 1981. [16] Ben-Sorek S, Davis CM. Joint mobilization education and clinical use in the United States. Phys Ther. 1988;68:1000-1004. [17] Lee M, Moseley A, Refshauge K. Effect of feedback on learning a vertebral joint mobilization skill. Phys Ther. 1990;70:97-104. [18] Cochrane CG. Joint mobilization principles: considerations for use in the child with central nervous system dysfunction. Phys Ther. 1987;67:1105-1109. [19] Brooks SC. Coma. In: Payton OD, Di Fabio RP, Paris SV, et al, eds. Manual of Physical Therapy. New York, NY: Churchill Livingstone Inc; 1989:215-238. [20] Progress in back pain? Lancet. 1981;1:977-979. [21] Deyo RA. Conservative therapy for low back pain: distinguishing useful from useless therapy. Jama 1983;250:1057-1062. [22] Waddell G. A new clinical model for the treatment of low back pain. Spine. 1987;12: 632-644. [23] Di Fabio RP. Clinical assessment of manipulation and mobilization of the lumbar spine: a critical review of the literature. Phys Ther. 1986;66:51-54. [24] Ottenbacker K, Di Fabio RP. Efficacy of spinal manipulation/mobilization therapy: a meta-analysis. Spine. 1985;10:833-837. [25] Haldeman S. The Clinical Basis for Discussion of Mechanisms of Manipulative Therapy. New York, NY: Plenum Publishing Corp; 1978: 53-75. [26] Nachemson AL. A critical look at the treatment of low back pain. In: Goldstein M, ed. The Research Status of Spinal Manipulative Therapy. Bethesda, Md, NINCDS Monograph No. 15, US Dept of Health Education, and Welfare publication NIH 76-988, 1975:287-293. [27] Hoehler FK, Tobis JS. Appropriate statistical methods for clinical trials of spinal manipulation. Spine. 1987;12:409-411. [28] Brunarski DJ. Clinical trials of spinal manipulation: a critical appraisal and review of literature. J Manipulative Physiol Ther. 1984;7: 243-249. [29] Cooper HM. Scientific guidelines for conducting integrative research reviews. Rev Educ Res. 1982;52:291-302. [30] Gilbert JR. Management of low-back pain in family practice: a critical review. Can Fam Physician. 1986;32:1855-1861. [31] Greenland S, Reisbord LS, Haldeman S, Buerger AA. Controlled clinical trials of manipulation: a review and a proposal. J Occup Med. 1980;22:670-676. [32] Hoehler FK, Tobis JS, Buerger AA. Spinal manipulation for low back pain. JAMA 1981; 245:1835-1838. [33] Sloop PR, Smith DS, Goldenberg E, Dore C. Manipulation for chronic neck pain: a double-blind controlled study. Spine. 1982;7: 532-535. [34] Meade TW, Dyer S, Browne W, et al. Low back pain of mechanical origin: randomized comparison of chiropractic and hospital outpatient treatment. BMJ. 1990;300:1431-1437. [35] Fairbank JCT, Davies JB, Coupar J, O'Brian JP. The Oswestry low back pain disability questionnaire. Physiotherapy. 1980;66:271-273. [36] Jayson MI, Sims-Williams H, Young S, et al. Mobilization and manipulation for low-back pain. Spine. 1981;6:409-416. [37] Hirsch G, Beach G, Cooke C, et al. Relationship between performance on lumbar dynamometry and Waddell score in a population with low-back pain. Spine. 1991;16:1039-1043. [38] Bigos SJ, Battie MC, Fisher LD. Methodology for evaluating predicative factors for the report of back injury. Spine. 1991;16:669-670. [39] Nwuga VCB. Relative therapeutic efficacy of vertebral manipulation and conventional treatment in back pain management. Am J Phys Med. 1982;61:273-278. [40] Coxhead CE, Inskip H, Meade TW, et al. Multicentre trial of physiotherapy in the management of sciatic symptoms. Lancet. 1981; 1(8229):1065-1068. [41] Parker GB, Tupling H, Pryor DS. A controlled trial of cervical manipulation for migraine. Aust NZJ Med. 1978;8:589-593. [42] Mealy K, Brennan H, Fenelon GCC. Early mobilization of acute whiplash injuries. BMJ. 1986;292:656-657. [43] Meade TW, Browne W, Mellows S, et al. Comparison of chiropractic and hospital outpatient management of low back pain: a feasibility study. J Epidemiol Community Health. 1986;40:12-17. [44] Smidt GL. Research study analysis. Journal of Orthopaedic and Sports Physical Therapy. 1991;13:288-291. [45] Saunders HD. Commentary four. Journal of Orthopaedic and Sports Physical Therapy. 1991;13:297-299. [46] Paris SV. Commentary two. Journal of Orthopaedic and Sports Physical Therapy 1991; 13:294-295. [47] Hadler NM, Curtis P, Gillings DB, Stinnett S. A benefit of spinal manipulation as adjunctive therapy for acute low-back pain: a stratified controlled trial. Spine. 1987;12:703-706. [48] Ongley MJ, Klein RG, Dorman TA, et al. A new approach to the treatment of chronic low back pain. Lancet. 1987;18:143-146. [49] Sanders GE, Reinert O, Tepe R, Maloney P. Chiropractic adjustive manipulation on subjects with acute low back pain: visual analog pain scores and plasma [Beta]-endorphin levels. J Manipulative Physiol Ther. 1990;13:391-395. [50] Tobis JS, Hoehler FK. Musculoskeletal manipulation in the treatment of low back pain. Bull NY Acad Med. 1983;59:660-668. [51] Glover JR, Morris JG, Khosla T. Back pain: a randomized clinical trial of rotational manipulation of the trunk. Br J Ind Med. 1974;31:59-64. [52] Fisk JW. A controlled trial of manipulation in a selected group of patients with low back pain favoring one side. NZ Med J. 1979;90:288-291. [53] Evans DP, Burke MS, Lloyd KN, et al. Lumbar spinal manipulation on trial, part I: clinical assessment. Rheumatol Rehabil. 1978; 17:46-53. [54] Farrell JP, Twomey LT. Acute low back pain: comparison of two conservative treatment approaches. Med J Aust. 1982;1:160-164. [55] Dixon ASJ. Progress and problems in back pain research. Rheumatol Rehabil. 1973;12: 165-175. [56] Gibson T, Grahame R, Harkness J, et al. Controlled comparison of short-wave diathermy treatment with osteopathic treatment in non-specific low back pain. Lancet. 1985; 1(8440):1258-1261. [57] Mathews JA, Mills SB, Jenkins VM, et al. Back pain and sciatica: controlled trials of manipulation, traction, sclerosant and epidural injections. Br J Rheumatol. 1987;26:416-423. [58] Parker G, Pryor D, Tupling H. New Zealand inquiry into chiropractic. Med J Aust. 1980;1:103-105. [59] Howe DH, Newcombe RG, Wade MT. Manipulation of the cervical spine: a pilot study. JR Coll Gen Pract [Occas Pap]. 1983;33:574-579. [60] Cibulka MT, Delitto A, Koldehoff RM. Changes in innominate tilt after manipulation of the sacroiliac joint in patients with low back pain: an experimental study. Phys Ther. 1988; 68:1359-1363. [61] Godfrey CM, Morgan PP, Schatzker J. A randomized trial of manipulation for low-back pain in a medical setting. Spine. 1984;9:301-304. [62] Bergquist-Ullmann M, Larsson U. Acute low back pain in industry: a controlled prospective study with special reference to therapy and confounding factors. Acta Orthop Scand Suppl. 1977;170:1-110. [63] Doran DML, Newell DJ. Manipulation in treatment of low back pain: a multicentre study. BMJ. 1975;2:161-164. [64] Frymoyer JW, Nelson RM, Spangfort E, Waddell G. Clinical tests applicable to the study of chronic low-back disability. Spine. 1991;16:681-682. [65] Saal JA, Saal JS. Nonoperative treatment of herniated lumbar intervertebral disc with radiculopathy: an outcome study. Spine. 1989;14: 431-437. [66] Leboeuf C. A review of data reports published in the Journal of Manipulative and Physiological Therapeutics from 1986-1988. J Manipulative Phyiol Ther. 1990;13:89-95. [67] Scheibel A, Debusschere M. Les modifications du stabilogramme qu'entrainent deux ajustements osteopathiques sont coherentes avec l'orientation sagittale ou frontale qu'ils impliquent. Agressologie. 1991;32:134-136. [68] Center DB, Leach Ra. The multiple baseline across subjects design: proposed use in research. J Manipulative Physiol Ther. 1984;7: 231-236.