Five-week Outcomes From a Dosing Trial of Therapeutic Massage for Chronic Neck Pain

Karen J. Sherman, PhD, MPH; Andrea J. Cook, PhD; Robert D. Wellman, MS; Rene J. Hawkes, BS; Janet R. Kahn, PhD; Richard A. Deyo, MD, MPH; Daniel C. Cherkin, PhD


Ann Fam Med. 2014;12(2):112-120. 

In This Article


Recruitment and Follow-up

Among the 1,027 people successfully assessed for eligibility between July 2010 and August 2011, we found 728 to be ineligible and 91 declined (Figure 1). Of the remaining 228 who were randomized, 37 to 39 persons were allocated to each of the 5 dosing groups or to the single wait list control group. The large majority of participants (86%) were recruited from Group Health. Overall, follow-up was 97%, with group-specific rates ranging from 93% to 100%.

Figure 1.

Trial flow.
a Most common reasons for ineligibility: 237 (32.6%) insufficient neck pain; 176 (24.6%) neck pain too complex; 74 (10.2%) prior massage; 140 (19.2%) could not attend treatment clinics.

Baseline Characteristics

Baseline characteristics were well balanced across groups, except for the percent of participants of white, non-Hispanic race/ethnicity and the percent having more than 7 days of usual activity restricted because of neck pain ( Table 1 ). Study participants typically had moderately severe neck pain, but relatively few reported substantial activity limitations due to their pain.

Treatment Adherence

Treatment adherence, defined as attending at least 75% of the assigned dose, was at least 95% in 4 massage dosing groups and 84% in the fifth group (30 minutes 3 times weekly). Because self-care recommendations were proscribed, it was not surprising that only 6 of 184 participants (3.3%) in the massage groups reported doing self-care activities they claimed were recommended by the study massage therapists, with those recommendations spread evenly among the groups.

Nonstudy Treatments

The use of medication as a nonstudy treatment varied across groups. Among those randomized to 60 minutes 3 times weekly treatments, medication use in the prior week dropped from 71.8% at baseline to 34.2%, and in the 30 minutes 3 times weekly group, medication use increased from 48.7% to 67.7%. In all other treatment groups, the absolute percentage change in medication use varied between −13.2% and 2.6%. Medication use in the control group increased slightly from 56.8% to 62.9%. Approximately one-third of participants reported doing neck exercises at least 3 times per week at both baseline and 5 weeks. Overall, 11% of participants made visits to health care professionals during the 5 weeks of treatment, mostly primary care physicians and chiropractors, with the highest percentage seen in the wait list control group (17%).

Neck Dysfunction and Neck Pain Intensity

A higher proportion of participants randomized to any dose of massage reported clinically important improvements in both the NDI and neck pain intensity relative to those randomized to the wait list control condition ( Table 2 ). These differences were statistically significant only for the more frequent 60-minute treatment groups, however: the adjusted likelihood of improvement in NDI score for the 2 times weekly and 3 times weekly groups relative to the control group was 3.41 (95% CI, 1.05–11.08; P = .04) and 4.98 (95% CI, 1.64–1 5.17; P = .0 05), respectively. The adjusted likelihood of improvement in neck pain intensity was 2.30 (95% CI, 1.26–4.18; P = .007) and 2.73 (95% CI, 1.52–4.91; P = .001), respectively.

Although the adjusted mean NDI scores worsened in the wait list control group after 5 weeks, they improved in all the massage groups ( Table 3 ). Mean reductions in NDI from baseline were significantly greater in the massage groups than in the control group except for the 30-minute 3 times weekly dose. Improvements in adjusted mean neck pain intensity were significantly greater than that in the control group only in the 60-minute treatment 2 and 3 times weekly and in the 30-minute treatment 2 times weekly groups.

We performed linear tests for trend in the outcomes based on the number of 60-minute treatments per week (0 to 3 massages per week). These tests showed a significant dose-dependent benefit for both the adjusted NDI and neck pain intensity. For each additional weekly massage, there was an estimated −1.81-point improvement in NDI (95% CI, −2.52 to −1.10; P <.001) and an estimated −0.75-point improvement in neck pain intensity (95% CI, −1.01 to −0.47; P <.001).

Other Outcomes

There were no statistically significant differences between groups for any secondary outcomes except for the proportion of participants who reported their neck pain was much better or completely gone (omnibus P <.001) ( Table 4 ). Three massage groups were significantly more likely to report this level of improvement than the control group: benefit was evident the 30 minutes 3 times weekly group (20.4%; 95% CI, 10.2%–40.6%); the 60 minutes 2 times weekly group (18.9%; 95% CI, 10.0%–35.8%), and the 60 minutes 3 times weekly group (40.6%; 95% CI, 27.8%–59.5%). Moreover, the last group fared significantly better than all other treatment groups.

Adverse Events

During the 4-week treatment period, 10 participants (5.2% of the total) reported 14 adverse events (11 mild and 3 moderately severe) at least possibly related to massage. All these events were related to pain, primarily spine pain. Adverse event frequencies were similar in participants attending 30-minute and 60-minute treatments (4% vs 6%, respectively) and in those attending 1, 2, or 3 times per week (7.9% vs 2.6% vs 6.7%, respectively).