Bove G, Nilsson N. Spinal Manipulation in the Treatment of Episodic Tension-Type HeadacheA Randomized Controlled Trial. JAMA. 1998;280(18):1576–1579. doi:10.1001/jama.280.18.1576
From the Center for Biomechanics, Department of Anatomy and Cell Biology, Odense University, Odense, Denmark. Dr Bove is now with the Department of Anesthesia and Critical Care, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Mass.
Context.— Episodic tension-type headache is common and is often treated using
manual therapies. Few data exist for the efficacy of these interventions.
Objective.— To determine the effects of spinal manipulation therapy on adults with
episodic tension-type headache.
Design.— Randomized controlled trial lasting 19 weeks.
Setting.— Outpatient facility of a National Health Service–funded chiropractic
research institution in Denmark.
Participants.— Volunteer sample of 26 men and 49 women aged 20 to 59 years who met
the diagnostic criteria for episodic tension-type headache as defined by the
International Headache Society.
Intervention.— Participants were randomized into 2 groups, 1 receiving soft tissue
therapy and spinal manipulation (the manipulation group), and the other receiving
soft tissue therapy and a placebo laser treatment (the control group). All
participants received 8 treatments over 4 weeks; all treatments were performed
by the same chiropractor.
Main Outcome Measures.— Daily hours of headache, pain intensity per episode, and daily analgesic
use, as recorded in diaries.
Results.— Based on intent-to-treat analysis, no significant differences between
the manipulation and control groups were observed in any of the 3 outcome
measures. However, by week 7, each group experienced significant reductions
in mean daily headache hours (manipulation group, reduction from 2.8 to 1.5
hours; control group, reduction from 3.4 to 1.9 hours) and mean number of
analgesics per day (manipulation group, reduction from 0.66 to 0.38; control
group, reduction from 0.82 to 0.59). These changes were maintained through
the observation period. Headache pain intensity was unchanged for the duration
of the trial.
Conclusion.— As an isolated intervention, spinal manipulation does not seem to have
a positive effect on episodic tension-type headache.
TENSION-TYPE headaches (TTHs) are extremely common, affecting more than
one third of the population1 and accounting
for more than two thirds of all headache episodes.2
The pain of TTHs is usually mild to moderate and typically bilateral. Symptoms
are usually described as pressing or tightening with a nonpulsating quality,
and symptoms characteristic of migraine (nausea, vomiting, photophobia, and
phonophobia) are absent. The International Headache Society (IHS) classifies
persons with 1 to 15 and 16 or more occurrences of TTH per month as having
episodic TTH (ETTH) and chronic TTH, respectively.3
Episodic tension-type headache is by far the more common form, with a 1-year
prevalence of more than 38%.1 Because many
sufferers report interference with their lifestyle, ETTH has a significant
Although ETTH is typically medically treated with nonprescription analgesics,
many people with ETTH seek hands-on therapy, delivered by various groups of
practitioners. Several uncontrolled studies have suggested that a combination
of physical treatment methods can lessen the symptoms of TTH,4- 8
and the single randomized controlled study using IHS criteria suggested a
positive treatment effect of a combination of therapies (including manipulation,
heat, massage, and trigger-point therapy) for chronic TTH,9
compared with a hands-off pharmaceutical treatment protocol. In all these
studies, however, it is unclear whether the effects were due to one, all,
or a combination of the treatment elements. The common element in these studies
was joint manipulation, recently demonstrated to be effective in the treatment
of cervicogenic headache.10 The sole report
using manipulation as an isolated treatment for TTH found a reduction in pain
intensity immediately after the treatment, but the subjects were not further
evaluated.11 The goal of this randomized controlled
trial was to evaluate the long-term effect of joint manipulation as an isolated
intervention for ETTH.
This was a randomized, controlled clinical trial with a blinded observer
designed to evaluate the effect of spinal manipulative therapy in the treatment
of ETTH. The trial was carried out in the outpatient facility of a National
Health Service–funded independent chiropractic research institution
(Nordisk Institut for Kiropraktik, Odense, Denmark). Data were interpreted
by a blinded observer and analyzed at the Center for Biomechanics, Faculty
of Health Science, Odense University, Odense, Denmark. All procedures were
approved by the regional ethics committee (permit No. 97/12), and all subjects
gave verbal and written informed consent prior to participation.
Participants were recruited from the general community by advertisements
in the local newspapers. Initial eligibility screening was done over the telephone,
followed by a personal interview and physical examination. Eligibility criteria
for participation in this study were as follows: (1) fulfillment of IHS criteria
for ETTH,3 with more than 5 but fewer than
15 headache episodes per month; (2) age 20 to 60 years; (3) score for typical
headache intensity between 25 and 85 on a visual analog scale from 0 to 100;
and (4) no relative or absolute contraindications to manipulation. The selection
refinements of criteria 1 and 3 were designed to better allow the possibility
of seeing a change by using more typical rather than very mild or severe cases.
After acceptance, a participant could be excluded for any adverse reaction
to treatment or any event triggering or potentially triggering a change in
headache status (eg, vehicular crash with neck injury).
Participants were informed that the purpose of the trial was to compare
the relative efficacy of 2 physical treatments commonly used for ETTH. Participants
were instructed to continue their normal lifestyle during the trial, including
their usual pattern of medication.
After baseline data collection and randomization (see below), participants
underwent 1 of 2 treatment protocols. All participants received 8 treatments
over 4 weeks, all performed by one chiropractor (Birthe Hove Madsen, DC) and
each lasting approximately 15 minutes. The chiropractor routinely uses both
treatment protocols in daily practice and was confident in performing both
techniques. The manipulation group received joint manipulation of the cervical
spine as determined by the chiropractor based on palpatory examinations and
also deep friction massage (including trigger-point therapy, if indicated)
of the trapezius and muscles deep to it. Specific manipulation maneuvers consisted
of diversified and/or toggle-recoil techniques, depending on the level of
the palpated segmental dysfunction.12 These
techniques use a high-velocity, low-amplitude thrust delivered in a specific
line of drive at the palpated end point of the normal passive range of motion
and are often accompanied by a clicking or cracking noise.
The control group also received deep friction massage. In addition,
this group underwent the application of low-power laser light (Omega Biotherapy,
London, England) to the upper cervical region. No effect, apart from placebo,
can be expected from low-power laser therapy.13
The control group was designed to be an active control group for spinal
manipulation, with the added goal of making the 2 groups more similar. A control
group that receives no treatment is not ideal in procedures that use physical
contact; in this study, every active intervention parameter was identical
in the 2 groups except for the actual manipulation procedure. Our design allowed
evaluation of the contribution of spinal manipulation, as it was the only
difference between the 2 groups.
The trial lasted 19 weeks. In weeks 1 and 2, data were collected to
determine a baseline for the outcome variables. Before starting week 3, participants
were randomized to either the manipulation group or the control group by a
blinded drawing of a ticket by a secretary. In weeks 3 through 6, subjects
were treated 8 times, usually twice a week. Posttreatment data were collected
from headache diaries completed during weeks 7, 11, 15, and 19.
All participants were required to fill out a headache diary14 for each of weeks 1, 2, 7, 11, 15, and 19. Recorded
information included hours per day that headache was present, the intensity
of the headache, and daily analgesic consumption. Additionally, to reveal
any difference between the 2 physical treatment regimens in their ability
to generate a placebo effect, participants scored their expected treatment
outcome on a visual analog scale immediately before randomization and again
immediately after the first treatment session, before a new treatment effect
could have been noted.15
The primary outcome measures were the number of headache hours per day,
mean headache intensity per headache episode, and consumption of analgesics
per day. From a previous study on a similar headache population,10
we expected most patients to stay with their preferred analgesic regimen throughout
the trial period. In case of medication change, 500 mg of acetylsalicylic
acid, 500 mg of paracetamol, and 200 mg of ibuprofen were regarded as equivalent
doses. All data were retrieved from the diaries by assistants unaware of the
treatment group assignment.
The outcome measures were compared between pretreatment and each posttreatment
period within and between the 2 treatment groups. A sample size of 84 was
projected to provide a power of 90% to detect a 1-hour difference between
the 2 groups in mean number of daily headache hours and a difference in headache
intensity of 13/100 per episode, as scored on the visual analog scale. All
tests of hypotheses and reported P values are 2-tailed.
Statistical analysis was conducted on an intention-to-treat basis, using StatView
(Abacus Concepts, Berkeley, Calif).
Participant flow and retention is summarized in Figure 1. Recruitment was conducted over a 9-month period from
February through October 1997. Of the persons who responded to the advertisements
and completed telephone and personal interviews, 82 fulfilled the inclusion
criteria; of these, 7 did not wish to take part in the study, leaving 75
participants to enter the trial. The characteristics of the randomized participants
are shown in Table 1. There were
no statistically significant differences between the 2 groups with respect
to age, sex, mean headache hours per day, mean headache intensity per episode,
or use of analgesics. No side effects were experienced by any participant
in either group.
One participant was excluded in week 5 following a neck injury due to
a car crash. One participant from the placebo group was lost to follow-up
at week 15 and a further 3 (manipulation group, 2; control group, 1) were
lost to follow-up in week 19.
There was no significant difference between the 2 groups for any outcome
variable before or following treatment (Table 1 and Table 2, Figure 2). However, mean headache hours
per day and analgesic use within both groups showed improvement from pretreatment
to week 7. The mean headache hours per day was reduced by approximately 1.5
hours by week 7 (95% confidence interval [CI], −2.4 to −0.6;
Figure 2, B), and this change
did not lessen significantly by the end of the trial. Headache intensity
was unchanged for the duration of the trial (95% CI, −12 to 11; Figure 2, C). Analgesic consumption also
lessened in both groups by week 7 (95% CI, −0.5 to −0.1; Figure 2, B). Of the 74 participants, 48
stayed with the same medication throughout the trial and 25 switched between
2 of the following: 500 mg of acetylsalicylic acid, 500 mg of paracetamol,
or 200 mg of ibuprofen. One participant switched between acetylsalicylic
acid and a morphine preparation; analgesic data for this person were not
included in the results. Analysis of the 48 participants who continued their
same medication yielded similar results (data not shown).
The expected treatment outcome was similar in both the manipulation
group and the control group before (95% CI, 69-83/100 vs 68-80/100) and after
(95% CI, 64-80/100 vs 69-81/100) the initial treatment. The treatment expectation
did not change following the first treatment (95% CI, −2 to 10/100 vs −5
to 7/100), indicating that the 2 treatment methods had no major placebogenic
This study showed that spinal manipulation did not significantly improve
the outcome for ETTH. We point out that our study population was carefully
selected based on accepted criteria for ETTH. In practice, ETTH and cervicogenic
headache can be difficult to differentiate,16
and often occur together.17 Our conclusions
are in stark contrast to those of an earlier and very similar study of cervicogenic
headache, in which the effect of spinal manipulation was quite dramatic.10 These data thus underline the importance of accurate
diagnosis in the selection of headache patients for spinal manipulation.
The only other randomized controlled study including manipulation as
a treatment for ETTH compared a hands-off pharmaceutical treatment (6 weeks
of amitriptyline) with a chiropractic treatment that included personal contact,
heat, massage, ergonomic advice, and spinal manipulation.9
There was a significant decrease in overall symptoms in the chiropractic treatment
group, but this effect could have been due to the higher level of personal
attention given to that treatment group. Our manipulation group was comparable
to that chiropractic group, and we controlled for the treatment elements of
personal attention and hands-on treatment.
As always, when a controlled clinical trial fails to demonstrate a treatment
effect, the question of a possible type II error (ie, overlooking a real difference)
is an issue. In this trial, the number of participants was relatively small,
and it is possible that a larger trial might have identified an effect in
1 or more of the outcome variables. However, the data in Table 2 and Figure 2 suggest
that any such effect, statistically significant or not, would be of little