The randomized controlled trial design is considered to be the gold standard for clinical research and has been advocated for evaluating the benefits of alternative modes of therapy,
1 including osteopathic manipulative treatment (OMT).
2
When designing a randomized controlled trial, researchers must make several decisions concerning control strategies. They must decide whether to use a sham (false) treatment and, if a sham treatment is chosen, which one is appropriate for the study.
3
The goal of using a sham protocol is to control for the potentially therapeutic effects of touch and belief, which are components of the placebo effect. When participants are blinded to group assignment, the placebo effect can be controlled. However, it is challenging to design a sham protocol that is both ineffective (carries little or no therapeutic effect) and plausible.
In their study of OMT for patients with low back pain, Andersson et al
4 cited the lack of a placebo control group as a design limitation and observed that “it is difficult to develop a placebo for manipulation.”
4 Hondras et al
5 reviewed manual modes of therapy for asthma and made several recommendations for future clinical trials using such modes of therapy. The authors recommended that sham protocols be used and that clinical trials assess the success of the sham protocols to blind subjects to group assignment. Therefore, as part of a randomized clinical trial that examined OMT's effectiveness in boosting immunity in nursing home residents, this study included a poststudy survey to determine the effectiveness of the sham protocols to blind participants and to inquire into participants' assessment of their treatment protocol.
There are few reports of adverse reactions to manipulative modes of therapy in the literature. Vick et al
6 reviewed the safety of manipulative treatments reported in the literature between 1925 and 1993. They found 185 specific reports of adverse events involving serious injuries, including cerebrovascular accidents (66%), disk herniation (12%), and bone fracture and dislocation (8%). The researchers found no reports of minor or transient complications (eg, headaches, sprains, rib fractures) in the literature and speculated that many transient symptoms of injury are unreported. Hondras et al
5 also recommended that future clinical trials of manual modes of therapy report the nature of adverse events. Therefore, researchers in the current study collected data on adverse events that occurred during our study.
Other aspects of the current study are beyond the scope of this article. These highlights have been published as abstracts.
7,8
We obtained informed consent in accordance with the institutional review board standards of the Kirksville College of Osteopathic Medicine of A.T. Still University of Health Sciences in Kirksville, Mo. Consent was obtained directly from all nursing home residents with the capacity to understand the nature of the study and the study procedures. For residents unable to provide consent due to cognitive impairment (but able to cooperate with the protocol treatments), the legal guardians, persons with durable power of attorney, or closest relatives provided the informed consent.
We recruited participants aged 65 years and older from two community nursing homes. Nursing home residents with an acute illness, hypersensitivity to the influenza vaccine, or allergy to egg were excluded from participation, as were those unable to cooperate with study protocol treatments. In addition, residents with acute vertebral or rib fractures, cancer, splenomegaly or history of splenectomy, or at high risk for pathologic fractures were also excluded.
Participants were randomly assigned to groups that either received an OMT protocol or a sham protocol control, stratified by age and mobility. The two age strata fell between 65 and 84 years and 85 years and older. In addition, participants were classified as primarily ambulatory (ambulates daily independently or with an assistive device) or nonambulatory (bed-bound or chair-bound, does not ambulate without maximum assistance) to form the mobility strata. We collected information on participants' demographics, exposure to influenza vaccination the preceding year, and medical histories.
At the start of each protocol session, participants in both groups received a structural examination of the same duration. The treatment group received a standardized OMT protocol designed to boost immune response to the influenza vaccine. The control group received a standardized sham treatment protocol designed to be believable but minimally effective. Influenza vaccinations were administered before the first protocol treatment. Participants received study protocol treatments three times weekly for 2 weeks and then twice weekly for 2 more weeks. Both the OMT and sham treatment protocols were conducted with the participants supine in bed.
The OMT study protocol consisted of two phases. During the first phase, the operator—using the choice of muscle energy, counterstrain, myofascial, direct, or articulatory techniques—treated each patient's specific somatic dysfunction found on structural examination. This phase of the study protocol typically took no more than 5 minutes. The second phase of the OMT study protocol was a standardized application of the following OMT techniques: paraspinal muscle inhibition, rib raising to the paraspinal muscles, thoracic inlet myofascial release, abdominal diaphragm myofascial release, thoracic lymphatic pump, and splenic pump. These techniques have been thoroughly described elsewhere.
9,10
The sham treatment protocol also consisted of two phases. During the first phase, the operator took several minutes to purposefully and systematically auscultate participant's heart and lungs. During the second phase, the operator—sitting next to the patient at bedside—placed both hands under the participant and lightly touched the paraspinal muscles along the thoracic rib cage, taking care not to move fascia or articulate the ribs. Next, the operator—at the head of the bed—purposefully palpated the cervical spine, observing areas of restriction and tissue texture changes. The operator then tested for fascia direction preference in the thoracic inlet and laid hands lightly on the thoracic inlet—without moving—for approximately 1 minute. The operator continued by placing one hand on the thoracolumbar junction and lightly placing the other hand on the epigastric region for 1 minute—without moving either the hands or the fascia. Next, the operator lightly placed both hands on the anterior chest wall and diagnosed the ribs for pump handle and bucket bail restriction with regular respirations. Finally, the operator placed both hands on the cervical region and lightly massaged the cervical soft tissue.
Approximately 15 minutes were required to deliver either the OMT protocol or the sham protocol, so contact time with the operator in both groups was the same. The first phase of each protocol treatment lasted between 2 and 5 minutes, while the second phase lasted between 8 and 10 minutes.
All participants received conventional medical care as directed by their primary care physicians. Those physicians were blinded to group assignment, as were the participants, participants' familes, nursing home staff, and those collecting outcome data. To evaluate the success of the sham protocol to blind participants, a posttreatment questionnaire was given 1 week after the final treatment protocol by an individual who was also blinded to group assignment.
We used two methods to evaluate the frequency and severity of adverse effects from the treatment protocols. For the first method, we requested—before the beginning of the second through the tenth treatment sessions—that each participant report adverse effects potentially associated with the prior treatment session. For the second method, in the posttreatment questionnaire, we requested that participants report adverse side effects potentially associated with treatment sessions.
We then calculated summary statistics on demographic and medical history data, as well as on data derived from the posttreatment questionnaire. Both method groups were compared on age, using a Mann-Whitney test, and on categorical demographic and medical history data, using the Fisher exact test. Data collected from the posttreatment questionnaire were compared, using Fisher exact tests, to evaluate the effectiveness of the sham treatment protocol.
Twenty-two nursing home residents were recruited. Eight participants were excluded from the posttreatment survey: 5 participants were excluded (3 in the treatment group and 2 in the control group) because they did not remember receiving study protocol treatments due to dementia, 2 were excluded because they were unable to communicate due to prior stroke-related expressive aphasia, and 1 was unavailable at the time of the survey due to hospitalization. Fourteen participants (seven in each group) remembered the treatment sessions well and were able to respond to the survey. Finally, 2 participants with histories of dementia (one in each group) had mild cognitive impairment, but remembered treatments well enough to respond to the survey.
The demographic characteristics of the fourteen participants are summarized in
Table 1. The groups were not significantly different in regard to age, gender, and ambulatory status, and there were no significant differences for history of comorbid medical problems of dementia, chronic obstructive pulmonary disease, diabetes mellitus, hypertension, stroke, coronary artery disease, and congestive heart failure. None of the participants of either group were active tobacco users. Finally, in addition to receiving the influenza vaccine at study baseline, all participants had received the influenza vaccine 1 year before the study began.
Table 1
Demographic Characteristics of Subjects (N = 14) *
Variable | Treatment Group (n = 7) | Control Group (n = 7) |
---|
Age, y | 82.3 ± 4.7 | 84.3 ± 5.6 |
Gender, No. (%) | | |
Female | 6 (86) | 6 (86) |
Male | 1 (14) | 1 (14) |
Race/ethnicity, No. (%) | | |
White | 7 (100) | 7 (100) |
Received flu shot in past year, No. (%) | | |
Yes | 6 (86) | 7 (100) |
Not known | 1 (14) | 0 (0) |
History of dementia, No. (%) | | |
Yes | 1 (14) | 1 (17) |
No | 6 (86) | 5 (83) |
History of chronic obstructive pulmonary disease, No. (%) | | |
Yes | 1 (14) | 1 (17) |
No | 6 (86) | 5 (83) |
History of diabetes mellitus, No. (%) | | |
Yes | 2 (29) | 3 (50) |
No | 5 (71) | 3 (50) |
History of hypertension, No. (%) | | |
Yes | 5 (71) | 3 (50) |
No | 2 (29) | 3 (50) |
History of stroke, No. (%) | | |
Yes | 3 (43) | 0 (0) |
No | 4 (57) | 6 (100) |
History of coronary artery disease, No. (%) | | |
Yes | 0 (0) | 3 (50) |
No | 7 (100) | 3 (50) |
History of congestive heart failure, No. (%) | | |
Yes | 3 (43) | 3 (50) |
No | 4 (57) | 3 (50) |
Table 1
Demographic Characteristics of Subjects (N = 14) *
Variable | Treatment Group (n = 7) | Control Group (n = 7) |
---|
Age, y | 82.3 ± 4.7 | 84.3 ± 5.6 |
Gender, No. (%) | | |
Female | 6 (86) | 6 (86) |
Male | 1 (14) | 1 (14) |
Race/ethnicity, No. (%) | | |
White | 7 (100) | 7 (100) |
Received flu shot in past year, No. (%) | | |
Yes | 6 (86) | 7 (100) |
Not known | 1 (14) | 0 (0) |
History of dementia, No. (%) | | |
Yes | 1 (14) | 1 (17) |
No | 6 (86) | 5 (83) |
History of chronic obstructive pulmonary disease, No. (%) | | |
Yes | 1 (14) | 1 (17) |
No | 6 (86) | 5 (83) |
History of diabetes mellitus, No. (%) | | |
Yes | 2 (29) | 3 (50) |
No | 5 (71) | 3 (50) |
History of hypertension, No. (%) | | |
Yes | 5 (71) | 3 (50) |
No | 2 (29) | 3 (50) |
History of stroke, No. (%) | | |
Yes | 3 (43) | 0 (0) |
No | 4 (57) | 6 (100) |
History of coronary artery disease, No. (%) | | |
Yes | 0 (0) | 3 (50) |
No | 7 (100) | 3 (50) |
History of congestive heart failure, No. (%) | | |
Yes | 3 (43) | 3 (50) |
No | 4 (57) | 3 (50) |
×
The results of the poststudy survey of the nursing home study are presented in
Table 2. There was no difference between groups for correctly guessing which treatment protocol they believed they had received. Three participants in the treatment group correctly believed they received OMT, 3 in the control group incorrectly thought they received OMT, and 4 in each group could not identify the treatment protocol they received. None of the participants believed they received the sham treatment protocol. The groups were also not significantly different in reporting the health benefits and adverse effects associated with treatments nor in their reporting of the enjoyment of treatments and their willingness to recommend these treatments to others.
Table 2
Results From Follow-up Questionnaire in the Influenza Vaccination Study
Variable | Yes, No. (%) | No, No. (%) |
---|
Did your health benefit from receiving OMT? | | |
Treatment | 5 (71) | 2 (29) |
Control | 4 (57) | 3 (43) |
Did you have any adverse effects from OMT? | | |
Treatment | 1 (14)* | 6 (86) |
Control | 1 (14)† | 6 (86) |
Do you believe you received OMT? ‡ | | |
Treatment | 3 (43) | 0 (0) |
Control | 3 (43) | 0 (0) |
Did you enjoy receiving OMT? | | |
Treatment | 6 (86) | 1 (14) |
Control | 6 (86) | 1 (14) |
Would you recommend the treatment you received to others? | | |
Treatment | 5 (71) | 2 (29) |
Control | 6 (86) | 1 (14) |
Table 2
Results From Follow-up Questionnaire in the Influenza Vaccination Study
Variable | Yes, No. (%) | No, No. (%) |
---|
Did your health benefit from receiving OMT? | | |
Treatment | 5 (71) | 2 (29) |
Control | 4 (57) | 3 (43) |
Did you have any adverse effects from OMT? | | |
Treatment | 1 (14)* | 6 (86) |
Control | 1 (14)† | 6 (86) |
Do you believe you received OMT? ‡ | | |
Treatment | 3 (43) | 0 (0) |
Control | 3 (43) | 0 (0) |
Did you enjoy receiving OMT? | | |
Treatment | 6 (86) | 1 (14) |
Control | 6 (86) | 1 (14) |
Would you recommend the treatment you received to others? | | |
Treatment | 5 (71) | 2 (29) |
Control | 6 (86) | 1 (14) |
×
Four of the survey participants had OMT before the study. Two of these individuals in the treatment group correctly believed they had received OMT, while one of these individuals in the control group incorrectly believed he had received OMT. Another of these individuals in the control group was unsure whether he had received OMT or the sham treatment protocol.
Table 3 lists by case summary responses to the poststudy questionnaire and each participant's Mini-Mental Status Examination (MMSE) score. A score of 23 or lower is generally considered to be “impaired.” At least 3 of 7 responders in the treatment group and 1 of 7 responders in the control group had MMSE scores considered to be impaired. One individual in the treatment group was recovering from acute illness on the day the MMSE scores were collected and therefore was not available for the survey, and 2 individuals in the control group refused to participate in the MMSE.
Table 3
Case Summaries Listed by Mini-Mental Status Exam Scores for OMT and Sham Groups Who Responded to the Poststudy Questionnaire
| Factor | | | | | |
---|
Group | MMSE Score * | Which Group? | Health Benefit? | Enjoy Treatment? | Would Recommend? | Adverse Effects? |
---|
OMT | | | | | | |
1 | No data† | Not sure | No | Yes | Yes | No |
2 | 13 | Not sure | No | No | No | No |
3 | 16 | OMT | Yes | Yes | Yes | No |
4 | 20 | Not sure | Yes | Yes | No | No |
5 | 25 | OMT | Yes | Yes | Yes | No |
6 | 25 | Not sure | Yes | Yes | Yes | Yes |
7 | 30 | OMT | Yes | Yes | Yes | No |
Sham | | | | | | |
1 | No data‡ | OMT | Yes | Yes | Yes | No |
2 | No data‡ | OMT | Yes | Yes | Yes | Yes |
3 | 13 | Not sure | No | Yes | Yes | No |
4 | 28 | Not sure | Yes | Yes | Yes | No |
5 | 29 | OMT | Yes | Yes | Yes | No |
6 | 29 | Not sure | No | Yes | Yes | No |
7 | 30 | Not sure | No | No | No | No |
Table 3
Case Summaries Listed by Mini-Mental Status Exam Scores for OMT and Sham Groups Who Responded to the Poststudy Questionnaire
| Factor | | | | | |
---|
Group | MMSE Score * | Which Group? | Health Benefit? | Enjoy Treatment? | Would Recommend? | Adverse Effects? |
---|
OMT | | | | | | |
1 | No data† | Not sure | No | Yes | Yes | No |
2 | 13 | Not sure | No | No | No | No |
3 | 16 | OMT | Yes | Yes | Yes | No |
4 | 20 | Not sure | Yes | Yes | No | No |
5 | 25 | OMT | Yes | Yes | Yes | No |
6 | 25 | Not sure | Yes | Yes | Yes | Yes |
7 | 30 | OMT | Yes | Yes | Yes | No |
Sham | | | | | | |
1 | No data‡ | OMT | Yes | Yes | Yes | No |
2 | No data‡ | OMT | Yes | Yes | Yes | Yes |
3 | 13 | Not sure | No | Yes | Yes | No |
4 | 28 | Not sure | Yes | Yes | Yes | No |
5 | 29 | OMT | Yes | Yes | Yes | No |
6 | 29 | Not sure | No | Yes | Yes | No |
7 | 30 | Not sure | No | No | No | No |
×
The responses of those participants with MMSE scores in the “normal” range indicate uncertainty about group assignment. Of the 3 participants in the treatment group with MMSE scores greater than 23, 1 incorrectly believed he received a sham treatment, and 2 correctly believed they received OMT. Of the 4 participants in the control group with MMSE scores greater than 23, 1 incorrectly believed he received OMT, and 3 were unsure of group assignment.
Only one participant in each group reported adverse effects from OMT (
Table 2). In the treatment group, one participant reported stiffness and was unable to get out of bed the morning after the first treatment. In the control group, one participant reported “a little muscle soreness in the abdomen.” Of interest, this was the individual in the control group with a history of dementia who refused to participate in the MMSE.
Questioning participants before treatment sessions 2 through 10 about adverse effects from treatment sessions generated 8 reports of adverse effects—7 in the treatment group and 1 in the control group.
In the treatment group, all 7 reports were generated from 2 individuals. One of those individuals reported mild musculoskeletal soreness following the second, third, fourth, and sixth treatment sessions. The other individual reported moderate musculoskeletal soreness after the sixth, seventh, and eighth treatment sessions. The discomfort was transient, and both participants completed the study.
Specific musculoskeletal complaints reported by participants in the treatment group were “low back pain,” “sore and tired all over,” “sore on left side,” “some tenderness” to the splenic pump, and “sore in back and shoulders.”
The only adverse musculoskeletal effect in the control group was reported by an individual who complained of “aching bones” for the entire first week after the influenza vaccination. The operator characterized the reaction as moderate and unlikely to have been the result of the sham protocol treatment. The individual reported no adverse effects from the sham protocol treatment in the poststudy survey.
This study adds to a growing body of literature on the use of sham treatments in manual medicine research. To our knowledge, this is the first clinical trial to document transient adverse effects associated with non–high-velocity, low-amplitude (HVLA) manipulation techniques in a nursing home population. Considering the small cohort size, the findings should be interpreted with caution. These findings primarily apply to a narrow population (frail nursing home residents), though we believe that the general principles learned from this project have broader application. As nursing home residents have increased infection rates, poor mobility, and multiple comorbid conditions, they are likely to benefit from manual modes of therapy designed to boost immunity, improve circulating bodily fluids, increase mobility, and relieve pain.
A variety of control strategies are documented in the literature. Kwoh
2 advocates using the osteopathic structural diagnostic examination as a sham treatment for patients with acute neck pain. Buerger
11 used “soft tissue massage” as the sham treatment and rotational manipulation as the active treatment in a study of patients with low back pain. Buerger used a posttreatment questionnaire to show that participants in the study were successfully blinded to group assignment.
In chiropractic treatment of low back pain, Hawk et al
12 used a handheld activator adjusting instrument with a pressure setting of zero for sham treatments. In this study, 8 of 14 interviewed subjects believed the placebo had a treatment effect. Balon et al
13 used a series of maneuvers, including “soft tissue massage” and “low-velocity, low-amplitude nondirectional maneuvers,” to create a believable sham treatment in patients with childhood asthma. A poststudy questionnaire indicated that 63% of the participants were unsure as to which group they had been assigned, and there was no difference between groups on the percent correctly guessing their group assignment.
Noll et al
14,15 used a light-touch standardized protocol as a sham treatment in two randomized controlled trials of elderly patients hospitalized with pneumonia. “Light touch” was applied to the same areas of the body for the same approximate duration and in the same sequence as in the active manipulation protocol. However, neither study used a poststudy survey to assess the effectiveness of the sham protocol for blinding participants.
Wells et al
16 used a sham protocol in a randomized controlled clinical trial on the efficacy of OMT to improve gait performance in patients with Parkinson's disease. To create a plausible sham protocol, researchers conducted structural examinations and took the same joints manipulated in the treatment group through passive range of motion without reaching the joints' range-of-motion limits. The duration of the sham protocol session was approximately equal to the active manipulation session. Wells et al
16 considered it unlikely that participants could distinguish between group assignment because all subjects were generally naive to the nature of OMT. However, no survey data were reported to confirm this assumption.
In our study, no participant believed he or she received a placebo treatment and either reported being unsure of his treatment protocol or believed he received OMT. Also, most participants in both groups “believed their health benefited from the treatment,” “enjoyed the treatment sessions,” and “would recommend the treatments to others.” This suggests that the sham treatment protocol was successful for blinding most of the participants to group assignment.
The purpose of a sham treatment is to introduce a measure of uncertainty about group assignment, control for the expectation that an intervention will heal, and control for the therapeutic effect of human touch and attention. From this perspective, the sham treatment protocol was successful.
Several factors may have made this particular group easier to blind than other populations. First, this cohort was relatively inexperienced with OMT. Only 4 of 14 study participants reported that they had received OMT before. Hondras et al
5 recommended that clinical trials of manipulation use participants who are naive to manipulation, as it is generally assumed that those with experience will be difficult to blind using a sham treatment protocol. To our knowledge, this assumption has never been tested. In our limited sample, those who had received OMT correctly guessed group assignment only half of the time.
Another reason for the success achieved in blinding in this study is that the treatment protocol uses a number of gentle techniques—myofascial, soft tissue, and lymphatic pumping techniques—that more closely resemble the “light touch” used in the sham treatment protocol. These techniques are different than the direct HVLA techniques with which the four subjects with prior OMT were familiar. Experience with one type of manipulation may not translate into familiarity with other types of manipulation.
Mein et al
17 argued that general ignorance of manual medicine diversity has led to shortcomings in current research methodology for manual medicine. The authors note the rich variety in the types of manual medicine techniques in use today.
In the search for a believable sham protocol, a pitfall is developing one that is essentially another form of manipulation, rather than a truly ineffective or minimally effective sham. Furthermore, any form of touch may have an accompanying beneficial physiologic effect. All sham interventions can be conceptualized as comparing one form of manipulation we consider effective to another form we consider ineffective. Therefore, sham-controlled trials may underestimate the actual benefit of manual therapy.
To address the dilemma of not knowing the relative contribution of the sham treatment, Hondras et al
5 recommended that investigators use deferred or no treatment control group in their study design in addition to a sham control. Perhaps, the ideal randomized clinical trial in manual medicine would include three arms: an active treatment group, a sham group, and a no-treatment group. However, such a study design is more costly and requires a larger number of participants for adequate statistical power.
Cognitive impairment and memory loss may also have been a factor in the success in blinding in this population, though all participants in the poststudy questionnaire asserted that they remembered the treatments well and could answer the questions. Four of the participants in the treatment group and three participants in the control group either had no recorded MMSE or scored in the “impaired” range on the examination. The importance of participants having an accurate memory in this study may be debatable because the placebo effect is highly belief-dependent, rather than dependent on factual recollections. Central to the placebo effect is the belief that the intervention is beneficial.
As long as a participant remembers receiving treatments and believes the treatments were beneficial, a placebo effect is likely. It could be argued that the greater gullibility of a population with mild memory impairment is an advantage in manual medicine research, as this might make blinding easier. In our study, however, participants with MMSE scores in the “normal” range did not do much better at guessing group assignment than those with “moderately impaired” MMSE scores.
Osteopathic manipulative treatment in elderly people has been advocated as safe and beneficial having few adverse effects, especially when administered in the proper “dosage.”
18,19 Yet, several participants in our previous clinical trial of OMT for elderly patients hospitalized with pneumonia had transient muscle soreness after manipulation, causing them to drop out of the study.
20
Many of the same non-HVLA techniques used in the hospital-based study were also used in the current nursing home study. Musculoskeletal soreness after the treatment protocol again appeared to be a relatively common adverse effect. Fortunately, the effects were relatively mild and transient and not severe enough to compel any of the participants to withdraw from the study. Considering the frailty of the population, the treatment protocols were remarkably well-tolerated. However, future studies of OMT in elderly people should list—during the informed consent process—muscle and joint soreness after treatment as a potential adverse effect.