Introduction

Stroke is a major health problem worldwide and represents one of the most common causes of disability with regard to its impact on functional limitations [1]. In addition, because of the aging population, the absolute number of strokes is expected to increase in the coming years. According to the Global Burden of Disease Study (GBD) [2], the socioeconomic burden of stroke has increased over time, although there has been a decrease in its prevalence [3]. Stroke imposes a high burden in terms of direct and indirect costs. On the one hand, indirect costs occur because of lost productivity due to long-term disability and restricted social functioning leading to a detriment to the patient’s quality of life (QoL) as well as premature death [4, 5]. On the other hand, direct costs of care occur resulting from the engagement of health professionals, hospital services [6], medications [7], etc.

Upper motor neuron lesions may result in positive symptoms like spasticity and negative symptoms like weakness or loss of dexterity [8]. Both result in some degree of functional limitation affecting the individual’s QoL, as well as somatosensory impairments, also related to activity limitations [9]. Physical therapy treatments can be combined with other pharmacological interventions and/or other medical treatments, such as antispastic drugs or botulinum toxin type A (BTX-A) infiltration. Recently, non-pharmacological approaches have been used, such as dry needling (DN) of myofascial trigger points (MTrPs), which is increasingly used to treat neurological conditions such as stroke [10], Parkinson’s disease [11], and multiple sclerosis [12]. Although the reasons for the increase in non-pharmacological treatments such as DN are not clear, the following factors could be relevant: (1) from the patient’s perspective, there is a need to shift to more patient-centered treatments, where patients are more involved in decision making about different treatment alternatives; and (2) from the professional and health system perspective, it is important to consider the high costs of pharmacological treatments such as BTX-A infiltration.

BTX-A is the most potent neurotoxin known, and its paralytic effect is due to the blockade of neuromuscular transmission [13]. On the other hand, DN acts by mechanically impairing sensory or motor nerves and dysfunctional motor endplates that contribute to the abnormal functioning of muscular contractile elements [14]. Therefore, the main difference between the two is the mechanism of action, as BTX-A works via chemical denervation while DN induces mechanical damage in the MTrP region [14]. MTrP region (also called Trigger Point Zone, Trigger Spot, or Trigger Area) is understood as “a focus of hyperirritability in a tissue that, when, compressed, is locally tender and, if sufficiently hypersensitive, gives rise to referred pain and tenderness, and sometimes to referred autonomic phenomena and distortion of proprioception” [15]. DN has been also related to pain modulation and has been shown to achieve a washout of sensitizing substances in the MTrP region [16]. DN is considered to be an effective and safe treatment to improve function and spasticity in stroke patients [10, 17, 18] when applied by an experienced physiotherapist. Moreover, although DN may have some adverse effects such as bruising, bleeding, and pain, it does not have the other adverse effects that BTX-A can have in the short and long term. In the short term, the diffusion of toxin from the injected muscle into neighboring muscles may cause undesirable weakness [19] and, depending on the location, this spread can be dangerous and lead to adverse events such as dysphagia, dysarthria, dysphonia, or respiratory compromise [20] amongst others. The long-term adverse effects include chemodenervation leading to muscle atrophy [19]. However, when compared with BTX-A, DN has fewer long-lasting positive effects [14], which requires a greater number of treatment sessions.

Different studies have carried out economic analyses of the aforementioned treatments for post-stroke spasticity. The most studied to date are different variants of botulinum toxin: abobotulinumtoxinA (aboBoNT-A), onabotulinumtoxinA (onaBoNT-A), and incobotulinumtoxinA (incoBoNT-A) [21]. Although recently there have also been secondary analyses of DN clinical trials in patients with subacute [22] and chronic stroke [23], there are no comparative studies or reviews that have included both of these treatment approaches.

It is important to analyze the clinical effectiveness of both pharmacological and non-pharmacological treatments with respect to their economic impact to inform clinical decision-making. However, only a few studies have been done from this perspective and there exists a great heterogeneity amongst them. Therefore, the objective of this study was to review all the economic publications about DN and BTX-A for post-stroke spasticity.

Methodology

To perform this narrative review, studies were included if they met the following eligibility criteria: 1) involved patients with post-stroke spasticity, with no restrictions regarding race, age, or sex; 2) included an intervention with BTX-A infiltration or DN; 3) reported economic outcomes; 4) consisted of a clinical trial (randomized clinical trial, matched-controls, cohorts) published in a peer-reviewed journal; and 5) written in English or Spanish. Studies were excluded if they met the following criteria: 1) publications that were not specific to the post-stroke population and 2) other types of publications such as book sections, conference abstracts, reviews, or meta-analyses.

Results

A total of 9 studies were found (see Table 1). Seven studies analyzed BTX-A infiltration and two studies analyzed DN. No studies compared both interventions.

The most frequent outcomes used in the economic studies and the results of the cost-effectiveness analysis carried out were reviewed. The economic analysis is usually performed through the ICER which is a summary measure representing the economic value of an intervention, compared with an alternative. It is calculated by dividing the difference in total costs (incremental cost) by the difference in the chosen measure of health outcome or effect (incremental effect), which in the case of the studies selected in this review were specific clinical scales or QALYs.

For the studies that used specific clinical scales, the most commonly used to verify improvements in spasticity and allow direct assessment of the response to treatment were the AS and subsequent modifications of this scale (MAS and MMAS). A prospective multi-centre study compared incoBoNT-A with conventional antispastic treatments for upper limb spasticity after stroke and found more than 56.4% of treatment responders with incoBoNT-A (vs. 26.9% with conventional treatment) after a year [28], with the resulting ICERs favourable to the incoBoNT-A treatment.

Two publications on DN treatment showed similar results with these scales [22, 23]. The first study showed that a single DN session in patients with chronic stroke resulted in statistically significant improvements in elbow extensor spasticity, with 73% of patients responding to the MMAS in the intervention group vs. 8% responding in the sham group, considering the values taken before and just after the session with DN [23]. Similarly, in another study on upper limb rehabilitation in patients with subacute stroke that performed DN with the DNHS^® technique, an average of 92% of patients responded favourably to the MMAS at 4 weeks vs. 18% in the control group, and 70% vs. 17% at 8 weeks, with significant differences in elbow flexion, forearm supination, and wrist extension spasticity [22].

In the case of studies that used economic outcomes, the preferred variable for cost-effectiveness analysis was QALYs. This outcome synthesizes the relationship between the number of years of life and the quality, or desirability of health state. In terms of costs, analyses were made from the point of view of society, the health care system, or at the clinic level. A study in Italy comparing treatment with aboBoNT-A and rehabilitation vs. rehabilitation alone found an improvement of 0.47 QALY more in the aboBoNT-A group over a 2-year period [29]. However, another study carried out by Doan et al. [27] only found an improvement of 0.107 over 5 years of treatment with usual care and onaBoNT-A. Another study evaluated the cost-effectiveness of treatment with incoBoNT-A vs. conventional anti-spasticity treatment using QALY in Romania in 2020 [30] and showed an improvement of more than 1 QALY for the incoBoNT group. In the case of botulinum toxin, some studies found no significant differences compared to rehabilitation alone [25], while for DN significant intra-group differences were found in patients with subacute stroke [22] and inter- and intra-group differences at 2 weeks after a single DN session in patients with chronic stroke [23].

One of the most comprehensive studies was BoTULS 2010 [25]. In this multi-centre study with more than 300 patients, outcomes of upper limb function such as the ARAT, MAS, Barthel ADL Index, troke-related QoL/Stroke Impact Scale, and EQ-5D among others, were observed for one year. Only the QoL data were used for the cost-effectiveness analysis. In the results, we observe the complexity of the cost-effectiveness analysis in this type of treatment as most of the variables show a certain advantage with the use of the toxin. Although there was an improvement in muscle tone in the first month, greater strength after 3 months, and an improvement in pain after 12 months, the authors conclude that “however, these differences were small and of uncertain clinical relevance.” BTX-A combined with the upper limb therapy programme indicated that there was only a 0.36 probability of it being cost-effective at a threshold ceiling ratio of 20,000 per QALY [25]. Although the results of the BoTULS study were not very encouraging for the use of the toxin, in 2013, Doan et al. [27] modeled the same data in different scenarios, giving results below the acceptable willingness to pay (established at €20,000/QALY) and even suggesting that savings could be achieved from a societal point of view. Comparing the likelihood of BTX-A being cost-effective with DN treatment, the study carried out by Fernández-Sanchis et al. [23] observed that DN had a 0.5 probability of being cost-effective in terms of QoL with a much lower threshold than BTX-A, and that these results were even more favourable to DN when considering the percentage of responders to the MMAS.

Apart from the aforementioned measures which were the most frequently used, other measures of effectiveness were identified. For example, the STM, which is defined as the percentage of patients who had met or had sufficient improvement in pre-treatment functional targets to warrant continuation of therapy [24], was only used in studies carried out with BTX-A. In the study by Ward et al. [24], the measurement of favourable or unfavourable outcomes to treatments was decided according to the criteria of clinicians participating in a Delphi panel (however, outcomes were based solely on expert opinion and as such may be subject to bias and inaccuracy) with no objective clinical or QoL measures. This cost-effectiveness study of BTX-A vs. oral treatment alone for spasticity showed that first-line use of BTX- A resulted in a higher success rate (STM) and lower costs associated with decreased nursing costs. Although nursing hours decreased, physical therapy hours increased, but not at the same level of cost. Moreover, in this review, we found other ways of assessing the efficacy of treatments of spasticity. Burbaud et al. [26] reported an improvement of 2.4 points with BTX-A on their own 4-point scale SRS in which the patient was simply asked to rate improvement based on a 4-point scale (no, modest, moderate, strong).

Discussion

Some of the cost-effectiveness analyses of BTX-A have been performed on the basis of models whose data came from reviews and network-meta-analyses that, combined with a good decision tree design and a probabilistic sensitivity analysis, give good reliability to the results. However, in other cases, the results of treatment effectiveness were based solely on expert opinion [29]. The use of a Delphi panel to determine the treatment outcome may result in unreliable and irreproducible results. Although the development of models can provide important information and confidence in the analysis, it is necessary to provide a reliable and verifiable source of data so that the results can be trusted.

Both BTX-A and DN studies were based on the same effectiveness outcomes for economic analyses in spasticity, permitting direct comparisons. Both interventions have a good cost-effectiveness ratio compared to placebo or conventional treatments. However, the large differences in cost assessment and treatment duration and lack of a direct comparison between BTX-A and DN make conclusions difficult. A budget impact study in the United Kingdom evaluated the impact of incorporating the new aboBoNT-A vs. the established onaBoNT-A and incoBoNT-A, indicating that a 5-year savings of £6,283,829 could be produced [31]. Since improvements in both QoL and responders to the MMAS scale have yielded similar results for the BTX-A and DN studies, it could be expected that the savings, using DN treatment, could be equal to or greater than with BTX-A.

In addition to different costs, there are also different cycle durations in the case of toxins and the number of sessions in DN that affect cost-effectiveness analyses. A study in Australia indicated that treatment with incoBoNT-A could be cost effective beyond 4 cycles per patient in those who were responders to treatment [32]. However, in a DN study, it was observed that a 4-week treatment could be more cost-effective than an 8-week treatment [22]. In another study, in addition to the direct or indirect costs themselves, the increased probability of fractures was related to the use of medication for spasticity such as diazepam or other sleeping pills, indicating that BTX-A can reduce their use and that there would also be a reduction in the costs of surgery and care related to fractures [33].

The diversity of the costs evaluated, the different countries and currencies, and the temporal analysis make it difficult to draw comprehensive conclusions. Moreover, the different protocol applications may also impose cost differences, for example, DN can be included as part of the standardized rehabilitation treatment, without involving any additional costs, or as a specific and complementary treatment as is the case for BTX-A. Apart from the differences derived from this treatment approach, in the case of DN the number of sessions that have the optimal cost-effectiveness has not yet been determined, whereas in the case of BTX-A only one session is needed. The difference in the number of sessions is related to the different mechanism of action whereby its effects last for about 3 months [34]. In the case of patients with subacute stroke, a recent study [22] determined that 4 DN sessions resulted in more cost-effectiveness than 8, but this has not been analyzed in the case of patients with chronic spasticity. Apart from the aforementioned limitations, derived from the heterogeneity of studies and the lack of direct comparisons between the two treatments, there were very few studies of DN. Thus, there is a need for more robust placebo-controlled studies about the effectiveness of DN on spasticity, including more sensitive and specific spasticity outcome measures. Nevertheless, cost-effectiveness studies comparing BTX-A and DN using the most frequently used outcomes such as MMAS and QALYs would be of great interest to allow for a more direct comparative analysis.

Conclusions

Both DN and botulinum toxin treatments present good cost-effectiveness results. However, the appearance of new therapeutic routes requires new studies that compare their efficiency. It is recommended that comparative studies have equivalent durations, effectiveness outcome measures, and a selection of costs from the same perspectives in different health systems.