Nonspecific neck pain is a common condition that affects an estimated 70% of persons at some point in their lives, and up to 1 in 5 persons each year. New research suggests that spinal manipulation therapy (chiropractic) or home exercises are equally better than medications for relieving pain. However, there are a number of limitations of this research to consider before leaping to a conclusion that the best medicine for neck pain is no medicine at all.
Researchers at the Northwestern Health Sciences University in Minneapolis, Minnesota, designed a pragmatic, randomized, controlled trial to assess the relative efficacy of spinal manipulation therapy (SMT), medication, and home exercise with advice (HEA) for neck pain in both the short-term (12 weeks) and long-term (52 weeks) [Bronfort et al. 2012]. They enrolled 272 persons aged 18 to 65 years (mean ≈ 48 years) who had nonspecific neck pain for 2 to 12 weeks — most subjects had subacute neck pain lasting more than 4 weeks, beyond the time when acute pain might resolve spontaneously.
Participants were excluded if they had cervical spine instability, fracture, neck pain referred from peripheral joints or viscera, progressive neurologic deficits, existing cardiac disease requiring medical treatment, blood clotting disorders, diffuse idiopathic hyperostosis (excessive bone growth). Additional exclusion criteria were inflammatory or destructive tissue changes of the cervical spine, infectious disease or other severe disabling health problems, substance abuse, pregnancy or breastfeeding, previous cervical spine surgery, and pending or current litigation.
Subjects were randomly assigned to either 12 weeks of SMT (N=91), medication (N=90), or HEA (N=91). Treatment conditions were designed to mimic those that might occur in everyday practice:
- SMT (spinal manipulation therapy) was administered by 6 chiropractors during visits lasting 15 to 20 minutes. The primary focus of treatment was manipulation of areas of the spine with segmental hypomobility by using diversified techniques, including low-amplitude spinal adjustments and mobilization techniques. The specific spinal level to be treated and the number of treatment sessions during the 12-week intervention period was left to the discretion of the provider. Adjunct therapy common to clinical practice included soft-tissue massage, assisted stretching, and hot and cold packs to facilitate the manipulation treatments. Advice to stay active or modify activity was recommended as needed.
- Medication group participants received care by medical physicians, with a focus of treatment on prescription drugs: first line therapy was NSAIDs, acetaminophen, or both. Participants could also receive opioid analgesics and/or muscle relaxants as deemed necessary. Advice to stay active or modify activity was issued as needed. The choice of medications and number of visits was made by the physician, based on the participant’s history and response to treatment.
- HEA (home exercise with advice) was provided in two 1-hour sessions at a university affiliated outpatient clinic. Six therapists provided instruction to participants, with a primary focus on simple self mobilization of the neck and shoulder joints, including neck retraction, extension, flexion, rotation, lateral bending motions, and scapular retraction, with no resistance. The exercise program was individualized to each participant’s abilities, tolerance, and activities of daily living. Participants were instructed to do 5 to 10 repetitions of each exercise up to 6 to 8 times per day [illustrated instructions for the exercises are available here].
The primary outcome measure was participant-rated pain on an 11-point numeric rating scale (NRS, 0 to 10), assessed at 2, 4, 8, 12, 26, and 52 weeks after randomization. Secondary measures included self-reported disability, global improvement, medication use, satisfaction, general health status, and adverse events.
Writing in the Annals of Internal Medicine, the researchers report that, for NRS pain measures, SMT had a statistically significant advantage over medication after 8, 12, 26, and 52 weeks (P≤0.01), and HEA was superior to medication at 26 weeks (P=0.02). No important differences in pain scores were found between SMT and HEA at any time point. Results for most of the secondary outcomes were similar to those of the primary outcome.
In responder analyses, approximately 32% of the SMT group and 30% of HEA participants had a 100% reduction in pain at 12 weeks, and 27% and 36% at 52 weeks, respectively. Meanwhile, only 13% and 17% of medication group participants had 100% pain reduction at 12 and 52 weeks, respectively. It is of interest that, at the 1-year mark, the HEA group actually outperformed SMT in terms of absolute reductions in pain; possibly because those subjects continued the exercise program long after the 12 week intervention period.
The researchers conclude that, for patients with acute and subacute neck pain, spinal manipulative therapy (SMT) is more effective than medication in both the short and long term. However, a couple of instructional sessions on home exercise with advice (HEA) can produce similarly beneficial outcomes.
COMMENTARY: As might be expected, the news media widely and simplistically featured this study as evidence that medications to relieve neck pain are largely ineffective and that chiropractors may offer better solutions. And, the three lead authors of the study — which was funded by the U.S. National Center for Complementary and Alternative Medicine (NCCAM) — are all chiropractors.
In an editorial accompanying the study report, Australian chiropractors Bruce Walker and Simon French  note that a wide variety of therapies have been used for the conservative care of neck pain. Their literature search revealed reviews of different medications, manual therapy (chiropractic), massage, acupuncture, electrotherapy, exercise, traction, patient education, and biopsychosocial rehabilitation. However, a single, most optimally effective therapy for neck pain was not determined.
At first glance, this present study by Bronfort et al.  seems adequately designed and well powered, with sufficient enrollment at the start to detect significant differences between groups (if any existed). As is typical of pain research, however, a closer inspection of the study design, execution, and outcomes reveals some important areas for consideration when assessing the validity of results (some of these also were noted by Walker and French, 2011):
- Effect sizes for NRS-score differences between groups were small, even when statistically significant [Cohen’s d scores, not calculated by the researchers; discussed in prior UPDATE here], and there was overlap in the 95% Confidence Intervals between groups for NRS-score reductions from baseline. Therefore, differences in pain-relief effects between the 3 treatments were not of large proportions.
- The subjects for study were very carefully selected patients, with uncomplicated, nonspecific neck-pain conditions that had not reached chronic status at the time they began therapy. Furthermore, baseline pain levels (mean±SD NRS score = 5.0 ± 1.6) were of moderate intensity. Since there was no placebo or “wait list” group, the extent to which this sort of neck pain might resolve on its own in this patient population, at least to some extent, is undetermined. Also, the extent to which a hands-on therapy, like SMT, might have exerted placebo effects is unknown but could have been of some significance.
- There were a disproportionate percentage of females in the medication group (72%) compared with the SMT (58%) and HEA (66%) groups; although, the impact of this on outcomes is unknown. By the end of the year of observation there also had been many more study discontinuations in the medication group (55) compared with 26 in the SMT group and 38 in the HEA group. At critical followup points during the course of the study, the numbers of missing subjects may have weakened analytical power to detect valid statistically significant differences between groups, which could have particularly affected medication group outcomes.
The researchers conducted an intent-to-treat analysis, which included all subjects initially enrolled; however, for unexplained reasons, they considered that only 12 participants were lost due to nonrandom reasons, and this might have biased results. They did not conduct a per-protocol analysis to determine if subjects actually completing all phases of the study might have had different outcomes than were reported.
- Effects of medications are clouded by several factors. First, it appears that subjects in the SMT and HEA groups were taking over-the-counter analgesics during the course of the study; although, the number of days per week of such use declined as a result of treatment. Still, this might have been a confounding factor [see Appendix Table 1 in the study report].
Secondly, 90% of subjects in the medication group were prescribed a combination of NSAIDs, opioid analgesics, and muscle relaxants. Whether this was too much or too little medication could be important and account for the excessive proportion of study discontinuations (61%) in this group; too much medication might have promoted intolerable side effects, while too little could have provided insufficient pain relief to make a difference — either of these may have encouraged drop outs.
- It also is unfortunate that the researchers did not include vitamin D status as a potential factor in nonspecific neck pain. As we have consistently noted in many UPDATES articles [series here], vitamin D deficiencies can sometimes account for musculoskeletal pains of unknown etiology, which might be resolved at least to some extent by adequate supplementation.
- The researchers do not report treatment adherence with either medications or home exercises, and this could have made a difference. They note that 60% of medication-group subjects experienced adverse effects during the study, primarily dry mouth, gastrointestinal complaints, cognitive symptoms, and drowsiness. Meanwhile, 46% of HEA group and 40% of SMT group participants experienced aggravation of pain due to their therapies. Only SMT therapy was provider-administered, during which compliance was beyond control of the patients; in the other two groups, adherence could be self-modified by patients. In the case of medications, nonadherence also might have ultimately led to study discontinuations.
- According to some critics [eg, Science-Based Medicine blog here], there is considerable controversy surrounding the validity of chiropractic care with its emphasis on adjusting spinal subluxations, or vertebral misalignments, to treat a variety of conditions. However, Bronfort and colleagues  make no mention of subluxations and their protocol seemed focused on mobilization of the cervical skeleton and musculature, including stretching and massage.
Still, in this present study, success of SMT might have been to an extent dependent on the expertise and skills of the individual chiropractor. As indicated above, 4 out of 10 SMT-group subjects complained of increased pain; it also should be noted that, in the wrong hands, rigorous manipulative chiropractic therapies focusing on the neck can incur a rare but potentially catastrophic risk of vertebral artery stroke [Walker and French 2011].
Furthermore, SMT in this study appeared to incorporate elements of massage therapy. In a previous UPDATE [here] we noted that massage was superior to medications or exercises for easing discomfort and improving function in lower back pain, with benefits lasting up to 6 months.
- Finally, this study did not examine relative costs of the 3 therapies or cost-effectiveness. If self-treatment by home exercises, at minimal cost, is effective for this type of neck pain, as the study suggests, it could make a big difference in future resource allocation.
In sum, by attempting to construct a pragmatic trial — that is, approximating how care is actually delivered in everyday practice — the researchers seem to have encountered a number of uncontrolled clinical factors that could have biased outcomes in this study. Perhaps, the most important message from this research by Bronfort et al. appears to be that medications as a sole therapy for nonspecific acute or subacute neck pain are probably an incomplete solution for many patients.
Patients and their healthcare providers need to be more proactive in combining appropriate medications with dynamic and safe therapies that help to improve fitness and health of the involved anatomical structures. Whether chiropractic, clinical massage, or simple and inexpensive at-home exercises are best for individual patients seems largely unresolved at present and may depend in part on patient preferences and their economic resources.
> Bronfort G, Evans R, Anderson AV, et al. Spinal Manipulation, Medication, or Home Exercise With Advice for Acute and Subacute Neck Pain. Annals Int Med. 2012(Jan);156(1 part 1):1-10 [article PDF here].
> Walker BF, French SD. Pain in the Neck: Many (Marginally Different) Treatment Choices [editorial]. Annals Int. Med. 2012(Jan);156(1 part 1):52-53 [editorial PDF here].