Chiropractic treatment including instrument-assisted manipulation for non-specific dizziness and neck pain in community-dwelling older people: a feasibility randomised sham-controlled trial

We conducted a parallel two-arm, randomised, sham-controlled feasibility trial. Participants were allocated to either a chiropractic intervention or sham intervention using a block randomisation procedure.

The human ethics clearance was obtained from RMIT University’s human research ethics committee (HREC) (Approval number 29/13). The trial was registered in the Australian New Zealand Clinical Trials Registry (ANZCTR) (Registration number: ACTRN12613000653763).

All interventions, including the sham intervention, were delivered by either one of two registered practicing chiropractors (depending on availability) who each had at least 20 years of clinical experience. Four intervention sessions were given over four weeks, and the duration of each session was kept to 15 min in both groups.

The recruitment rate was determined by comparing the number of enquiries from each advertising method with the number of participants who were enrolled from each of those methods. Additionally, the inclusion and exclusion criteria were reviewed by examining the frequency and the reasons for exclusion during screening.

Participants’ compliance with the outcome assessment was examined by noting the time taken to complete the baseline questionnaires and assessments, and reviewing if any outcome measures were consistently filled out incorrectly or were incomplete. Compliance with the intervention schedule was examined by measuring drop-out rates and reasons for drop-outs.

We determined if participants dropped out, or had difficulty finding the clinical trial centre, due to its location on the Bundoora campus of RMIT University.

At the conclusion of the follow-up outcome measure assessment, participants indicated which intervention group they believed they were in (sham or chiropractic) to determine the integrity of allocation. All participants were also asked to rate their satisfaction with treatment on a five-point scale (from 1: I feel much worse to 5: I feel much better).

The feasibility of conducting a larger, fully-powered study was assessed on the basis of the costs of: advertising, equipment, and hiring the chiropractors and research assistants to recruit and screen potential participants and administer the interventions.

Harms were defined as adverse consequences of the intervention reported by participants. In accordance with the World Health Organisation’s [45] Conceptual Framework for the International Classification for Patient Safety, the degree of harm was classified as: none (no symptoms detected and no treatment required), mild (minimal or intermediate short term harm caused, and minimal or no intervention required), moderate (permanent or long-term harm caused, or intervention required), severe (major permanent or long-term harm caused, or major surgical/medical or life-saving intervention required) or death (death caused or brought forward). Harms and other reactions to interventions were documented by the treating chiropractor at each intervention session.

As a feasibility study, this trial did not have an a priori calculated sample size. We aimed for a sample size of 40 to generate sufficient information to address the feasibility objectives, particularly the recruitment rate.

Participants were randomly assigned after the baseline appointment using a permuted block randomisation protocol. The randomisation schedule was conducted by an independent statistician before recruitment using a computer generated random list of numbers. This list was used to assign random blocks of four or six participants at a time. Group allocations were concealed by placing them in opaque consecutively numbered sealed envelopes, which were opened, in order, by the treating chiropractor before the first intervention session.

Participants were blinded to group allocation. The researcher performing the outcome assessments was also blinded to group allocation. The chiropractors involved in performing the interventions were blinded to the results of the outcome assessments at pre- and post-intervention.

A total of 24 participants were recruited from 162 telephone enquiries (Fig. 1). The recruitment period ran over nine months, from October 2015 to June 2016; six months over the planned three-month period. Most commonly, screened participants were excluded due to the presence of self-reported diagnosed vestibular and spinal pathologies (n = 27 [20%]), neck pain without symptoms of dizziness (n = 19 [14%]), history of cardiovascular incidents (n = 18 [13%]), recent manual therapy (n = 14 [10%]) and not being able to travel to the research site (n = 16 [12%]). Ten participants (7%), who otherwise could have participated, were excluded due to low performance on the MoCA cognitive function assessment.

Using Facebook as a recruitment method for older Australians was surprisingly successful, with 38 (23%) enquiries; however, it captured many potential participants who were unable to travel to the trial location. The trial was stopped before the target 40 participants was reached due to time and budget constraints.

Clinical outcome measurements took between 60 and 90 min to complete for most participants, and all participants took less than the 2 h that was allocated. Participants were offered a break if they became tired, but this proved to be unnecessary in all cases. Questionnaires were checked by the investigator who pointed out several questions that were often missed and asked the participant to complete them.

After enrolment in the trial, participant compliance with the four interventions were deemed acceptable, with only two drop-outs. One drop-out from the sham group had a spontaneous aggravation of a lower back complaint unrelated to intervention and another participant in the chiropractic intervention group did not start the intervention due to inability in making the travel commitment.

Travel to the outer-suburban university location was a barrier for 16 (12%) potential participants. Additionally, one drop-out was due to difficulties with travel. Participants often became lost on their initial visit to the campus. While temporary parking permits were provided, the participants had to pick these up from the security station. This was sometimes confusing for them.

Advertising costs totalled AUD$43,679. Minor equipment costs were AUD$395. Two registered chiropractors were employed part-time as research assistants to set up the procedures, screen participants and quantify outcome measures; the cost for this was AUD$3033. In addition to one of the investigators (MFA), another experienced registered chiropractor was employed to be available to provide the weekly interventions at a cost of AUD$10,866. Therefore, total costs amounted to AUD$57,973. This translated to AUD$2635 per participant. The costs (per participant) of recruitment and intervention were AUD$2141 and AUD$494 respectively. These costs excluded the salary of the senior author (MFA) and the PhD scholarship of the first author (JCK).

Six (27%) participants reported harms. All harms were mild, including increased neck pain (chiropractic n = 2, sham n = 1), headache (chiropractic n = 1, sham n = 1) and mid-back pain (chiropractic n = 1).

Many clinical outcome measures were used in this study (Table 1). Participants had moderate intensity of dizziness at baseline [mean (SD)] in both the chiropractic group and the sham group (Table 2). DHI scores were also similar at baseline and improved in both groups post-intervention [chiropractic 28.33 (14.37) to 40.77 (12.48); sham 44.00 (16.97) to 36.40 (20.11)]. Similarly, NDI scores were reduced post-intervention [chiropractic 24.94 (12.87) to 19.07 (12.50); sham 24.18 (8.22) to 22.8 (6.2)]. Fifty eight percent of the chiropractic group showed a clinically-significant improvement (of at least 19%) in NDI scores compared to 30% of the sham group (Table 3). The DHI scores improved by the clinically significant amount (of at least 18%) in 67% of the chiropractic group compared to 50% of the sham group. Mood was generally low, with participants commonly reporting symptoms of depression, anxiety and stress on the DASS. Concerns of falling were high in both groups at baseline [chiropractic 26.00 (5.61); sham 29.00 (5.71)], and reduced slightly in both groups [chiropractic 24.42 (5.21); sham 26.7 (6.29)]. All participants were able to complete the physical functional tasks.

The sample size for a fully-powered trial (derived from data in this feasibility trial with an effect size of d = 0.38), using the DHI as the primary outcome measure, would require a group size of 150 (i.e. 75 per group). Alternatively, using NDI as the primary outcome measure (with an effect size of d = 0.46), would require a group size of 222 (i.e. 111 per group). These calculations exclude provision for XX%? drop-outs.

We aimed for a sample size of 40 participants. However, recruitment did not reach this pre-defined arbitrary number, even with a six-month extension to the recruitment period. If conducted in the same setting, to reach the estimated sample size of 150 participants (using DHI as the primary outcome measure) would take more than four years, assuming a similar recruitment rate. This study found that using online recruitment methods could be useful in targeting older Australians over a wider geographic area. On the other hand, newspaper advertisements, while more expensive, captured a local population. Considering the relatively high proportion (12%) of participants who could not make the travel commitment to our single location, for future studies we recommended having multiple sites with sufficient geographical spread to increase recruitment and retention or performing the study in a facility with a high concentration of elderly people with neck pain and dizziness such as a specialty clinic.

Older adults experiencing neck pain and dizziness often have co-morbidities. However, chiropractic intervention is unlikely to impact dizziness due to known vestibular or neurological origin, dizziness as a result of postural hypotension, or poly-pharmacy. Therefore, we recommend that the same exclusion of people with vestibular or neurological causes of dizziness be used for the larger study. Excluding these participants based on self-reported previous diagnoses, however, may not capture these individuals accurately. We recommend basing a future, larger trial in a rehabilitation setting such as a dizziness/falls clinic which provides access to clinical expertise and equipment to rule out vestibular, neurological, hypotensive and pharmacological causes of dizziness. Alternatively, the larger trial can be based on an effective referral system from a network of general medical practitioners. Exclusion of people based on cognitive-function testing has been shown to reduce the generalizability of findings [46], particularly in older people with pain [47]. However, the validity of self-reported measures of pain and function depends on intact memory and executive function. Participants in this trial who were excluded based on cognitive function were significantly disappointed to the extent that one of them lodged a complaint to the ethics committee. Future studies should consider how participants with potential impairments in cognitive function can be included, using outcome measures that are still able to capture self-reported pain and function. Alternatively, if a threshold of cognitive function is used as an exclusion criteria in future studies, procedures need be in place to direct excluded participants to providers of therapeutics for neck pain and dizziness to avoid disappointment.

The assessment regime was not too onerous for the participants and was completed in a timely manner. However, several participants missed individual questions on the questionnaires, and had to be prompted by the investigator to fill these in. It was necessary to have an investigator review completed questionnaires to check that all questions had been completed at the end of the assessment session.

The drop-out rate for participants was acceptable, with less than 15% for each group. However, this was for a relatively short intervention schedule of 4 visits over 4 weeks. It cannot be determined from this study if a longer, more intensive or less intensive schedule would have good compliance.

Conducting the trial at a university campus meant that some participants became lost trying to find the building location. Future studies should be conducted in an easy to find location with convenient car parking facilities, and ideally with a choice of several sites to capture participants who cannot travel long distances.

The protocol of using the Activator II™ instrument (set on zero) as a sham-chiropractic intervention appeared to achieve sufficient blinding in participants. This tool appears to be a useful blinding tool for future similar studies, particularly ones in which the experimental intervention consists of Activator II™-instrument delivered manipulation.

The costs and time to recruit sufficient numbers may be a challenge for a larger fully-powered RCT. Use of a network of chiropractic intervention sites may increase feasibility of recruitment. The cost of AUD$2415 per participant may prove prohibitive if only small grant funding is available. To reduce this expenditure, the larger study could be based in a dizziness/falls clinic of a general or rehabilitation hospital. The use of a specialised or hospital recruitment setting would necessitate modification of this protocol, and our results may not be reflective of the participants recruited in such settings.

Fifteen out of 23 participants did not report any harms. Mild harms such as transient increases in neck pain or headache are common following chiropractic intervention [28]. However, participants in the sham group also reported these harms, so these may be related to natural and non-specific effects [44].

Trials of non-pharmacological interventions for pain and dizziness in older people are scarce. This trial provides useful information in the Australian context on recruiting older people, and blinding for spinal manipulation, both of which are challenging. This is important information for future research. Furthermore, this was a feasibility study for determining effectiveness rather than efficacy. This necessitated that the intervention given reflected a ‘real-world’ combination of intervention strategies that Australian chiropractors would provide. Effectiveness studies by nature are not mechanistic and cannot identify the ‘active ingredient’ in the intervention package. But they do have higher external validity in their relevance and applicability to actual practice. In this sense, this was a trial comparing usual chiropractic care with sham chiropractic care. The intervention combination used here reflects the practice approach of a majority of Australian chiropractors (unpublished data), and follows contemporary practice guidelines for the treatment of the elderly [38]. However, it does not reflect every chiropractor’s practice style, particularly in its exclusion of manual manipulation of the neck. This limits the relevance of this study to trials of manual neck manipulation, as the biomechanics of manual manipulative thrusts are likely to be different from those delivered by an Activator instrument.

This trial was limited by the short-term follow-up, and no conclusions can be drawn about compliance with longer follow-up times. While the results of this trial advocate for conducting a fully-powered RCT at multiple locations, it did not test the feasibility of a protocol to ensure consistent recruitment and data collection across several sites. These issues should be investigated before such large-scale multi-centre studies are attempted. Another limitation of this study is that the participants were excluded based on self-reported previous diagnoses of dizziness, and were not uniformly screened by specialist medical staff to exclude other causes of dizziness. This may have made the cohort of participants somewhat heterogeneous. However, this heterogeneity reflects private practice that takes place within the primary care setting. Furthermore, this study is limited by including participants with very low intensities of dizziness and neck pain. There was no threshold for severity or intensity of dizziness or neck pain for inclusion. Setting of minimum DHI and NDI scores as inclusion criteria for future studies is recommended, although this would lead to a lower proportion of interested participants being eligible.