To evaluate the cost-effectiveness of deep tissue massage (‘massage’), strengthening and stretching exercises (‘exercises’) or a combination of both (‘combined therapy’) in comparison with advice to stay active (‘advice’) for subacute and persistent neck pain, from a societal perspective.
We conducted a cost-effectiveness analysis alongside a four-arm randomized controlled trial of 619 participants followed-up for one year. Health-related quality of life was measured using EQ-5D-3L and costs were calculated from baseline to one year. The interventions were ranked according to quality adjusted life years (QALYs) in a cost-consequence analysis. Thereafter, an incremental cost per QALY was calculated.
In the cost-consequence analysis, in comparison with advice, exercises resulted in higher QALY gains, and massage and the combined therapy were more costly and less beneficial. Exercises may be a cost-effective treatment compared with advice to stay active if society is willing to pay 17 640 EUR per QALY. However, differences in QALY gains were minimal; on average, participants in the massage group, spent a year in a state of health valued at 0.88, exercises: 0.89, combined therapy: 0.88 and, advice: 0.88.
Exercises are cost-effective compared to advice given that the societal willingness to pay is above 17 640 EUR per year in full health gained. Massage and a combined therapy are not cost-effective. While exercise appeared to have the best cost/benefit profile, even this treatment had only a modest benefit and treatment innovation is needed. Advice to stay active remains as a good therapeutic alternative from an economical perspective.
Exercises are cost-effective compared to advice for persistent neck pain.
Massage is not cost-effective compared to advice for persistent neck pain.
A combined therapy is not cost-effective compared to advice for persistent neck pain.
Differences between massage and/or exercises and advice in QALY gains are minimal.
Advice is a good alternative for persistent neck pain from an economic perspective.
Neck pain is a leading cause of disability worldwide and it is especially high among people between 25 and 64 years old ( ). Since most people with neck pain are in working age, this translates into high costs for the society due to productivity loss ( ). In Sweden in 2017, neck pain accounted for 4% (95% CI 3.3–4.8%) of the total years lived with disability and its burden is increasing ( ). Although most costs due to neck pain come from productivity loss ( ), the associated healthcare utilization are also an important source of expenditures. Health care costs are higher for care related to subacute and chronic then for acute neck pain ( ).
The evidence suggest that non-pharmacological therapies are effective for the clinical management of neck and associated disorders ( ). Recommended interventions for non-acute neck pain include massage ( ) and exercises ( ; ). Results from the STONE trial suggest that deep tissue massage and strengthening and stretching exercises are more effective in improving self-perceived recovery than advice to stay active in patients with disabling subacute and chronic non-specific neck pain ( ).
Few studies have evaluated the cost-effectiveness of non-pharmacological therapies and none have specifically evaluated massage for neck pain from an economical perspective. The existing evidence regarding its cost-effectiveness is inconclusive ( ). Nonetheless, massage for low back pain has been shown not to be cost-effective when provided alone, but it is cost-effective when combined with exercises ( ). On the other hand, exercises provided alone ( ) or in combination with advice ( ) are cost-effective for neck pain. In addition, one session of education is cost-effective for whiplash associated disorders ( ). An economic evaluation of these types of therapies for disabling non-specific neck pain is warranted.
This study evaluates the cost-effectiveness of deep tissue massage, strengthening and stretching exercises, or a combination of both compared with advice to stay active for subacute and persistent non-specific neck pain from a societal perspective.
Material and methods
We conducted the Stockholm Neck (STONE) trial, a randomized controlled trial to determine the effectiveness of deep tissue massage therapy (hereafter massage), strengthening and stretching exercises (hereafter exercises), a combination of deep tissue massage and strengthening and stretching exercises (hereafter combined therapy), compared with advice to stay active ( ). The study included subjects aged 18–70 years with subacute (30–90 days duration) or persistent (≥90 days duration) disabling neck pain with or without headache and/or radiating symptoms. Those with pain intensity <2/10 and pain-related disability <1/10 on the Numerical Rating Scales (NRS) were not included ( ; ). A study coordinator prepared blocks of sequentially numbered sealed envelopes (40 for each intervention). Participants were randomly allocated to one intervention when they contacted the study coordinator. The intervention was revealed after baseline assessment at the research clinic. Blinding of participants or therapists was not possible due to the nature of the interventions. The recruitment and data collection occurred between Nov 2015 and Nov 2017. The sample size was calculated to detect a relative risk of 1.2–1.3 at 12 months follow-up for a minimal clinically important improvement in pain intensity and pain-related disability. A more detailed description of the trial’s methodology is provided elsewhere ( , ).
All interventions were provided by naprapaths (medical professionals who focus on the management of musculoskeletal pain with a combination of manual therapy, exercise and/or advice) either under training (but with experience in the provision of massage and physical training) or licensed. The therapists underwent training sessions (3-h sessions twice) and regular supervision to ensure a standardized provision of all the interventions in the STONE trial. The number of sessions followed the usual practice in primary care in Sweden and were considered to be sufficient to achieve the purpose of the interventions; up to six sessions within six weeks were offered to the participants in the massage, exercises and combined therapy groups and up to three sessions in the advice to stay active group. All the sessions were provided individually.
Exercises (strengthening and stretching exercises)
Exercises targeted the following muscle groups: deep flexors of the neck, chest musculature, scapula musculature, jaw musculature and extensors of the neck. The sessions lasted 35 min on average. In addition to the supervised sessions, participants were advised to repeat the provided exercises at home. To facilitate that the execution of the exercises followed a proper technique, they were filmed with their own smartphone during the supervised sessions.
Massage (deep tissue massage)
The components of the deep tissue massage therapy were: massage of muscles of the neck, upper back, jaw and chest, as well as management of active and latent points producing concordant signs and stretching of chest. The techniques used were effleurage, firm motion involving compression and pressure release and deep muscle/fascia massage to areas that produced concordant symptoms. The sessions lasted 35 min on average.
Combined therapy (combination of strengthening and stretching exercises and deep tissue massage)
The combined therapy started with the same techniques as in the exercises groups and finished with the techniques in the massage group. Similar to the exercises group, in addition to the supervised sessions, participants were advised to repeat the provided exercises at home. Sessions lasted 50 min divided equally to the two modalities.
Advice (advice to stay active)
The intervention in the advice group consisted of a motivating discussion using evidence-based information on spinal pain in addition to an educational booklet. The advice group was the control group. Sessions lasted 20–25 min.
Participants filled out questionnaires at baseline, seven weeks, three months, six months and one year. We measured health-related quality of life using the EQ-5D-3L questionnaire ( ) (not measured at seven weeks), and based on this, utility was calculated using Swedish experience-based utility values (using the time trade-off method) ( ).
Pain intensity and pain-related disability were measured with numerical rating scales (NRS) from the Chronic Pain Questionnaire ( ); information on perceived recovery ( ) as well as sickness absence were also collected. These outcomes were, however, not included in the economic evaluation. A more detailed description of procedures and findings regarding the effectiveness of the interventions (in terms of pain intensity, pain-related disability, perceived recovery and sickness absence) has been reported elsewhere ( , ).
We collected information on direct medical costs (interventions, appointments with different health providers and prescribed drugs), direct non-medical costs (over the counter medication, naturopathic medication and time allocated to attend appointments) and indirect costs (sickness absence). Costs used in the economic evaluation were calculated as a product of items of resources used. The unit cost of each type of item associated with the therapy as described in the section below. All the costs were collected in Swedish crowns (SEK) and converted to Euros (1 EUR = 9.94 SEK ( )) for the year 2017/2018 (which correspond to the period in which the randomized controlled trial was conducted) and no discounting was applied since the interventions and the follow-ups occurred within one year.
Direct medical and non-medical costs were calculated by multiplying the number of each resource use at the individual level with its corresponding unit cost. No hospitalizations costs were recorded. Visits to physiotherapists and medical doctors are subsidized by the Swedish healthcare system and patients usually pay a copayment only. We assigned the fee paid by the patients plus the compensation given by the city council ( , ). The cost of the interventions in the trial were obtained from market prices of visits to naprapaths, depending on the duration of the intervention, rather than from a direct calculation based on salaries, overhead costs, supplies and others, since that information was not available. The cost of imaging was calculated from the average of the market prices in Stockholm. For prescribed and over the counter (OTC) medication, we took the selling price available on (TLV from its name in Swedish). For naturopathic medications, price information was obtained from the local and online retailers and assigned to each participant individually. The Swedish healthcare system subsidizes prescribed medications, imaging and appointments with medical doctors, physiotherapists and, to some extent, psychotherapists (high-cost threshold) in order to protect patients from high expenditures ( ). This reduction in costs was taken into account during the cost calculations.
The unit costs assigned were as follows: 1019 SEK/appointment with a medical doctor (103 EUR), 564 SEK/appointment with a physiotherapist (57 EUR), 661 SEK/appointment with a naprapath (67 EUR), 607 SEK/appointment with a chiropractor (61 EUR), 657 SEK/appointment with an osteopath (66 EUR), 506 SEK/appointment with a masseur (51 EUR), 863 SEK/appointment with a psychotherapist (87 EUR), 159 SEK/session of Yoga or similar interventions (16 EUR), 546 SEK/appointment with other providers not listed above (mostly alternative medicine or fitness-related) (55 EUR), 609 SEK for x-rays (61 EUR), 1795 SEK for computerized tomographies (181 EUR), 2428 SEK for magnetic resonances (244), 10.6 SEK/day for prescribed medications on average (1.1 EUR), 5.7 SEK/day for over the counter medications on average (0.6 EUR), and 7.8 SEK/day for naturopathic medications on average (0.8 EUR).
Production loss and opportunity costs
Indirect costs were included in terms of losses to paid production (i.e., sickness absence). Participants reported the number of days they had been off work due to neck pain. We followed the human capital approach ( ) (the value of people’s contribution is based on what they are paid) to calculate the cost of production loss. We multiplied the number of days off work by the mean salary based on the self-reported occupation ( ). Students and pensioners were assigned zero as value and volunteer time was deemed minimal.
Since we did not have specific information on means of transportation to and from the clinic or whether they used that time from work (or leisure time), no costs for that concept were assigned. However, overtime cost was used for the time invested in attending appointments, both the ones provided as part of the trial and the ones incurred by the participants after that. An hour was assumed for each attended appointment and the average salary for Sweden was used (33 700 SEK = 3391 EUR). ( ).
A typical month was assumed to consist of 21 working days or 168 h.
A recommended broader societal perspective on the impact of resource use was followed when conducting the economic evaluation. This implies that all relevant costs and effects of the intervention and non-specific neck pain are considered, regardless of who pays or benefits. An intention to treat principle was followed ( ). Costs and benefits were analyzed within one year horizon (the RCT followed participants for one year) ( ). No discounting was applied ( ). Quality-Adjusted Life Years (QALYs) were calculated over one year using the standard calculation method (area under the curve) ( ). Quality of life is measured between 0 and 1. A QALY is equivalent to one year in perfect health ( ). Thereafter, a cost-consequence analysis ( ) was performed in order to trim off strongly dominated alternatives, ranked by QALYs. If one of the interventions resulted in larger effects and higher cost than the next one in the QALYs-based rank, the relative cost-effectiveness of the intervention was calculated in terms of incremental cost-effectiveness ratio (ICER) by dividing the incremental costs by the incremental QALYs.
In order to account for the missing values in costs and QALYs, we did multiple imputation using QALYs at baseline (baseline utility) and provided intervention as predictors. We used predictive mean matching to impute QALYs and a regression function to impute data on costs. The economic value of providing an intervention was represented by the ICER. The differences in costs and QALYs between therapies was visually represented in a cost-effectiveness plane by using bootstrapping with 5000 replications. Seemingly unrelated regression was used to calculate the bootstrapped total costs and QALYs. To summarize the impact of uncertainty on the ICER, a cost-effectiveness acceptability curve (CEAC) was used to show the probability that the intervention will be cost-effective at different willingness to pay thresholds. All analyses were performed using Microsoft Excel and Stata 14.1 ( ).
An a posteriori sensitivity analysis was performed by excluding the costs from sickness absence. We observed that this item showed the highest difference between therapies. In addition, we observed differences at baseline in terms of the proportion of persons reporting sickness absence the past six months (Advice 23%, 95% CI: 16–32; Massage 29% 95% CI: 21–38; Exercises 31% 95% CI: 23–39 and, Combined therapy 31% 95% CI: 23–39). This supported the decision for conducting the sensitivity analysis. An ICER and a CEAC were obtained using the same procedures as in the main analysis.
Flow of participants in the STONE trial
After assessment of eligibility, 619 subjects were included in the study ( Fig. 1 ). After randomization, 145 subjects were allocated to the massage group, 159 to the exercises group, 168 to the combined therapy group and 147 to the advice group. The drop-out rate was higher in the advice group (21%) ( Fig. 1 ). There were no differences in baseline demographic characteristics in between dropouts and those remaining in the study (P-values > 0.05).
Participants gave their consent to participate in the STONE trial and for their data to be analyzed. The trial was approved by the Regional Ethics Committee in Stockholm (Registration number: 2014/755-31/3). The trial was registered 3 July 2014 with registration number ISRCTN01453590.
Most participants were females, the mean age was 46 years, most participants had completed more than 12 years of education, the most common occupations were managerial or those requiring a high university degree ( Table 1 ). Most participants had had neck pain for at least one year and reported a gradual onset of pain. The advice group had lower proportion of persons reporting at least one day of sickness absence related to neck pain in the past six months ( Table 1 ).
|n = 147||n = 145||n = 159||n = 168|
|Age, mean (SD)||46 (13)||48 (14)||47 (14)||45 (14)|
|12 years or less||54||(37)||43||(30)||55||(35)||65||(39)|
|More than 12 years||93||(63)||102||(70)||104||(65)||103||(61)|
|Occupation category ( )|
|Managerial or high university degree||71||(48)||56||(39)||61||(39)||75||(45)|
|University degree, administration and client-oriented||34||(23)||41||(28)||53||(33)||33||(20)|
|Service, care, sales, construction, transportation or short education||27||(19)||23||(16)||26||(16)||33||(20)|
|Duration of neck pain|
|12 + months||88||(60)||92||(64)||88||(55)||96||(57)|
|Characteristics of pain onset|
|Pain intensity at baseline, mean (SD)||5.8 (1.5)||5.9 (1.3)||6.1 (1.3)||6.1 (1.5)|
|Pain-related disability at baseline, mean (SD)||4.3 (1.7)||4.4 (2.0)||4.2 (1.8)||4.3 (1.7)|
|Previous episodes of NP (at least one)||75||(51)||67||(46)||74||(47)||90||(54)|
|Sickness absence in the past six months (at least one day)||34||(23)||42||(29)||49||(31)||52||(31)|