Introduction

Despite high prevalence and the socio-economic burden due to the associated pain and disability, neck pain (NP) receives much less attention when compared to low back pain (LBP) [1, 2]. The most common cause of NP is mechanical NP which is associated with long-standing poor posture and reduction in function and strength of cervical muscles [3, 4]. Although several rehabilitation treatment modalities are commonly employed, there is no consensus regarding the most effective management strategies for mechanical NP. Several reports have supported the use of a combination of different physical therapy modalities in the treatment of mechanical NP [5]. A recent trial by Domingues et al. [6] reported that a combination of manual therapy and exercises was more effective than usual care in improving disability and pain in patients with non-specific chronic NP.

Although mechanical diagnosis and treatment (MDT)-based treatment has been reported to be effective and superior to other interventions such as manual therapy and exercises for chronic LBP [7], few reports have studied the effect of MDT on NP. A systematic review of 5 trials by Takasaki and May et al. [8] reported that although MDT may improve pain than ‘wait and see’ or other treatment approaches such as strengthening exercises and cognitive behaviour therapy, the difference in effect may not be clinically important and that MDT does not have a greater effect on disability than ‘wait and see’ or other treatment approaches. Although centralisation of pain on repeated movement testing and patient age have been reported as an important prognostic factors affecting outcomes after MDT in patients with LBP [9, 10], studies have not reported prognostic factors affecting outcomes in patients treated with MDT for NP, especially as part of a multimodal rehabilitation program.

Baseline variables can help to predict treatment outcomes in patients undergoing rehabilitation treatment for NP and several prognostic factors can affect outcomes after treatment. Improvement in pain and functional ability in the short term (at the end of treatment) is perceived by patients as a sign of improvement in their spine condition, and is a useful indicator of clinical outcomes as a direct effect of the treatment intervention [11, 12]. Furthermore, a minimal clinical important difference (MCID) between the pre- and post-treatment outcome variables is recommended to identify patients who achieved a clinically significant improvement with rehabilitation treatment for spine pain [13, 14].

Hence, the aim of the present study was to determine the predictors of failure to achieve an MCID for pain and disability at the end of an MDT-based multimodal rehabilitation treatment for mechanical NP.

Materials and methods

Study design

In this retrospective study, electronic records of routinely collected clinical data of patients treated with physical rehabilitation for NP from 2016 to Dec 2020 at a chain of outpatient spine rehabilitation clinics (QI Spine Clinic, India) were analysed. This study was approved by an Institutional Review Board and Ethics Committee, and informed consent was taken from all patients at the start of treatment regarding the use of their anonymized demographic and clinical data for research.

Study participants

The inclusion criteria were all patients > 20 years of age presenting with NP who underwent rehabilitation treatment at our clinics. The exclusion criteria were patients with red flags conditions (tumour, infection, fracture, and instability), inflammatory NP (rheumatoid arthritis, spondyloarthropathy), cervical myelopathy or cord compression, patients with incomplete clinical records, and patients who took < 6 sessions or > 36 sessions of treatment at the centre. Patients who did not complete a minimum of 6 treatment sessions were excluded. Similarly, patients who underwent treatment of > 36 sessions were excluded to minimize the possibility that improvement in NP and function was due to time rather than a direct effect of the rehabilitation treatment. Mechanical NP was defined as pain which arose from the cervical spine or its surrounding soft-tissue structures, often showed improvement with rest, and aggravated during specific neck movements or activities [15]. Non-mechanical NP secondary to inflammatory condition (e.g. rheumatoid arthritis or spondyloarthropathies) or any red flag condition (e.g. infection, tumour, and fractures) were ruled out during first consultation based on history and presentation, clinical examination, and investigations (radiological and blood) wherever available.

Demographic data including age, gender, body mass index (BMI), lifestyle, and history of diabetes and osteoporosis were collected at the time of the first consultation. Lifestyle was categorised as sedentary if it involved prolonged sitting at work, during leisure time and lack of physical activity/exercise or movement during most part of the day; as active if there was daily moderate-to-vigorous physical activities as per the American College of Sports Medicine (ACSM) recommendations [16, 17]; and as semi-active if the lifestyle was more active than sedentary but did not involve performing daily moderate-to vigorous physical activities. Based on the presentation, patients with mechanical NP were categorised into “central NP with or without above elbow pain” (along the extent of the cervical spine posteriorly with or without pain radiating along the trapezius and into the arm till the elbow) similar to the “above elbow derangement” subgroup used in the MDT classification, and “central NP with below elbow pain” (along the extent of the cervical spine posteriorly with pain radiating into the upper limb below the elbow) similar to the “below elbow derangement” subgroup used in the MDT classification.

All patients were clinically examined for head and neck posture, cervical spine range of motion, motor and sensory function (myotomal and dermatomal loss) by a physiotherapist in the clinic. Repeated movement testing was performed using the MDT method to determine directional preference and response [18]. Before and at the end of treatment, NP intensity was measured using the numerical pain rating scale (NPRS), with pain intensity ranging from “0” (no pain) to “10” (worst pain imaginable), and functional disability was measured using the Neck Disability Index (NDI) [19, 20]. The total duration of treatment (in days) was also derived from the records for all patients.

Rehabilitation protocol

An MDT-based multimodal rehabilitation protocol using a combination of patient education, pain management with mechanical diagnosis and treatment (MDT)-based directional movements [18] and adjuvant frequency-specific microcurrents (FSM) application [21], and strengthening and stabilisation of cervical and upper back muscles were used in all patients. All treatment was provided by senior spine physiotherapists at each clinic who had undergone internal training in the MDT-based multimodal rehabilitation protocol conducted by a central team. The central team consisted of a clinic director and a national clinical expert (both with > 10 years of experience in spine rehabilitation) who were trained in the MDT technique and had a certified diploma from the McKenzie Institute International® (Paraparaumu, New Zealand). The internal training program was designed and delivered by these MDT-certified professionals and involved training all physiotherapists within the system on the multimodal rehabilitation treatment protocol. After the initial training, all physiotherapists underwent continued education training and internal assessment by the central team to qualify for promotion to senior positions within the system. Furthermore, an internal referral or escalation system was part of our clinical care delivery system where senior physiotherapists at each clinic can refer/escalate patients (who were non-responders or slow responders) to their clinic heads or the central team to confirm the MDT diagnosis and treatment technique being followed for the patient.

Patient education involved information on avoiding pain-aggravating neck movements, postures and activities, and ways of self-care at home during the course of treatment. Pain management was done using MDT-based repeated directional movements and adjuvant FSM application. The FSM modality involved application of low-voltage pulsed microamperage (one millionth of ampere) current generated by an FSM unit and delivered using 4 leads applied around the cervical spine (one on the posterior aspect of the neck on midline, one on either shoulder, and one on the upper back midline) [21]. This was administered to patients with higher baseline pain intensity (NPRS score > 3). Previous studies have reported the efficacy of FSM as an adjuvant for pain management in spine and musculoskeletal pain [22,23,24,25]. Based on the directional preference and response, patients were instructed to perform directional movements under the supervision of the treating physiotherapist, which was to be repeated at home at regular intervals. At home, patients were asked to repeat their directional preference movements 4–5 times a day (10 repetitions per session). Patients were asked to avoid or discontinue the movement if it aggravated or peripheralized their pain (i.e. symptoms migrating away from the midline of the body and towards the upper limb or from a proximal to distal direction in the upper limb) and inform their therapist. Once the pain reduced to mild (NPRS score ≤ 3), paraspinal muscle strength and spine mobility were assessed using a pressure biofeedback cuff and cervical goniometer and the patient was put on a customised strengthening and stabilisation exercise regimen. For patients without a directional preference (non-responders), pain was managed using FSM application, and strengthening and stabilisation exercises were initiated early and gradually progressed from basic to intermediate to advanced intensity of exercises based on patient response and tolerance. All patients were advised to undergo a minimum of 6 supervised rehabilitation sessions at the clinic. Patients included in the study took anywhere between 6 and 36 treatment sessions. The number of sessions done by the patient was decided by the treating physiotherapist or by the clinic head or central team (during escalation) based on baseline pain and disability, response to repeated movement testing, and progress in improvement from baseline pain and disability levels during the treatment. Hence, patients with higher baseline disability, partial or non-response to repeated movement testing, and slower progression in improvement of pain and disability during the treatment were prescribed more number of treatment sessions and had a longer treatment duration.

Study outcome variables

Our chain of outpatient spine rehabilitation clinics used a dedicated EMR software where treating physiotherapists were required to enter pre- and post-treatment NPRS and NDI scores for all patients who underwent treatment at our clinics. Furthermore, a clinic head (who was a senior physiotherapist in charge of the clinic) and a central team performed frequent reviews of patient records and ensured that patient information such as demographic, clinical presentation, response to MDT testing, treatment, and pain (NPRS score) and disability (NDI score) outcome details were complete before the discharge of the patient.

Pre- and post-treatment (at the time of discharge) NRPS and NDI scores were collected in all patients. The NPRS and NDI scores were collected from all patients using self-administered scales in the clinic before the consultation (pre-treatment) and at the time of discharge or end of treatment (post-treatment). To determine significant change in NPRS and NDI scores at the end of treatment, an MCID between pre-and post-treatment NPRS and NDI scores were calculated. Based on the recommendations by Ostelo et al. [10], MCID threshold was set at ≥ 30% improvement from baseline or pre-treatment NPRS and NDI scores. Based on the NPRS score, patients were categorised into 3 subgroups (≤ 3, 4–7, > 7), and based on NDI score into 4 subgroups (minimal, moderate, severe, crippled/bed-bound). Treatment compliance rate was calculated as the percentage of treatment sessions completed out of their prescribed total treatment sessions. Patients who attended less than 80% of their treatment sessions were termed as non-compliant to the treatment. The primary treatment outcomes in this study were MCID of ≥ 30% improvement for NPRS and NDI scores from pre-treatment values at the end of treatment, whereas the secondary treatment outcomes were NPRS and NDI scores at the end of treatment.

Statistical analysis

Demographic data including age, gender, body mass index (BMI), lifestyle, prevalence of diabetes and osteoporosis, response to repeated movements, number of treatment sessions, and treatment duration were analysed. Percentage of patients who achieved MCID of ≥ 30% for NPRS and NDI after treatment were calculated using pre- and post-treatment NPRS and NDI scores. Categorical data were compared using the Fisher’s test or Chi-square test and continuous data using the two-way analysis of variance (ANOVA) or Kruskal–Wallis test (for not normal distributed or ordinal data). For the univariate logistic regression analysis, pre-treatment variables were sub grouped and the percentage of patients who achieved MCID for NPRS and NDI after treatment were compared between subgroups for each variable using Chi-square test. The pre-treatment variables which were sub grouped included gender (male and female), age (< 30, 30–49, 50–69, and ≥ 70 years), BMI (< 25, 25.0–29.9, and ≥ 30 kg/m2), lifestyle (sedentary and active/semi-active), diabetes (yes and no), osteoporosis (yes and no), clinical presentation (central NP with or without above elbow pain and central NP with below elbow pain), response to repeated movements (responder, partial responder, and non-responder), number of treatment sessions (6, 7–12, 13–18, and > 18), treatment compliance rate (< 80% and ≥ 80%), pre-treatment NPRS score (≤ 3, 4–7, and > 7), and pre-treatment NDI score (≤ 20, > 20–40, > 40–60, and > 60). Pre-treatment variables which were found to be significant on univariate logistic regression analysis were included in a multivariate logistic regression analysis to determine odd’s ratio for these variables as risk factors for not achieving the ≥ 30% MCID for post-treatment NPRS and NDI scores. An Odds ratio (OR) > 1 indicated a higher probability of not achieving MCID, and an OR < 1 indicated a lower probability of not achieving MCID or higher probability of achieving MCID for NPRS and NDI scores. A p-value of < 0.05 was considered significant. Statistical analysis was performed using the SPSS (ver. 20.0) statistical analysis software (SPSS Science Inc, Chicago, IL).

Results

Based on the inclusion criteria, a total of 6402 patients with NP underwent rehabilitation treatment at our clinics. Based on the exclusion criteria, 421 patients were excluded due to red flag conditions, cord compression, or inflammatory NP and 5981 patients were treated for mechanical NP. In patients with mechanical NP, 915 patients either completed < 6 or > 36 sessions, and 68 patients took telerehabilitation treatment. Therefore, data from 4998 patients with NP who underwent in-clinic multimodal rehabilitation treatment were analysed for this study. The patients analysed included 2698 males and 2300 females with a mean age of 46.1 ± 13.3 years who took a mean 15 ± 7 treatment sessions at a mean treatment (or follow-up) duration of 84.5 ± 107 days. The MCID for NPRS score was achieved in 93% of patients, and for NDI score was achieved in 85.5% of patients at the end of treatment. The mean treatment of 15 ± 7 sessions was done in patient group which achieved MCID for NPRS and NDI was significantly higher (p < 0.0001) when compared to the mean treatment of 11 ± 5 sessions done in the patient group which did not achieve MCID for NPRS and NDI. The baseline characteristics of patients in this study are summarised in Table 1

Table 1 Demographic parameters of the study group
Full size table

Univariate analysis of MCID achieved for NPRS and NDI scores

The MCID achieved for NPRS score at the end of treatment were significantly higher in younger patients (p < 0.0001), in patients without diabetes (p = 0.02) and osteoporosis (p = 0.002), among responders to repeated movement testing (p < 0.0001), among patients with pre-treatment NPRS score 4–7 (p = 0.01), among patients with pre-treatment NDI score < 40 (p < 0.0001), among patients who did higher number of treatment sessions (p < 0.0001), and among patients with compliance rate of ≥ 80% (p < 0.0001) (Table 2). However, there were no significant difference when subgroups were compared based on gender, BMI, lifestyle, and presentation.

Table 2 Univariate analysis of MCID achieved for NPRS and NDI scores in pre-treatment variable subgroups
Full size table

Similarly, the MCID achieved for NDI score at the end of treatment were significantly higher in younger patients (p < 0.0001), in patients without diabetes and osteoporosis (p < 0.0001), among responders to repeated movement testing (p < 0.0001), among patients with pre-treatment NPRS score 4–7 (p = 0.01), among patients with pre-treatment NDI score > 20 (p < 0.0001), among patients who did higher number of treatment sessions (p < 0.0001), and among patients with compliance rate of ≥ 80% (p < 0.0001) (Table 2). However, there were no significant difference when subgroups were compared based on gender, BMI, lifestyle, and presentation.

Predictors of failure to achieve MCID for post-treatment NPRS and NDI scores

Multivariate analysis showed that age > 70 years, diabetes, osteoporosis, partial or non-response to repeated movement testing, lesser number of treatment sessions, lower compliance rate, and higher pre-treatment NDI score were associated with increased risk for failure to achieve MCID for NPRS score after treatment (Table 3). Similarly, age > 70 years, diabetes, osteoporosis, partial or non-response to repeated movement testing, lesser number of treatment sessions, lower compliance rate, and lower pre-treatment NPRS and NDI scores were associated with increased risk for failure to achieve MCID for NDI score after treatment (Table 3).

Table 3 Multivariate analysis to determine risk factors for failure to achieve MCID for post-treatment NPRS and NDI scores
Full size table

Discussion

The results of this study showed that 7% of patients failed to achieve MCID of ≥ 30% for NPRS score, whereas 14.5% of patients failed to achieve MCID of ≥ 30% for NDI score at the end of treatment. Age > 70 years, diabetes, osteoporosis, partial or non-response to repeated movement testing, lesser number of treatment sessions, and lower compliance rate were associated with increased risk for failure to achieve MCID for NPRS and NDI scores after treatment. A higher pre-treatment NDI score was associated with increased risk for failure to achieve MCID for NPRS score after treatment, whereas lower pre-treatment NPRS and NDI scores were associated with increased risk for failure to achieve MCID for NDI score after treatment.

The effect of pre-treatment or baseline NPRS and NDI scores on treatment outcome in NP is unclear with contrary findings in the literature. De Pauw et al. [26] in a retrospective analysis of 437 patients with chronic NP treated with multimodal treatment (back school education and exercise therapy), reported that a high NDI score, high NRS score for radicular pain in the upper extremities, a low NRS score for NP, and history of trauma decreased the odds of patient responding to the given treatment. However, Bohman et al. [27], in a longitudinal cohort study of 89 women with chronic, non-specific NP, reported increasing age and lower baseline neck disability as important predictors associated with poor short- and long-term improvement after 11 weeks of physiotherapy interventions (coordination exercise, strength training and massage) similar to our study. A possible explanation for our finding of lower pre-treatment NPRS and NDI scores associated with risk for failure to achieve MCID for NDI could be due to difference in study population demography (larger patient numbers, lower mean age, gender ratio, type of treatment administered) and other confounding factors such as duration of pain, episodes of recurrence, cervical active ROM, and baseline mental health which may have significantly affected post-treatment disability [27, 28].

Our results were in accordance with the findings of previously published studies regarding the effect of risk factors such as number of treatment sessions [29], and response to repeated movement testing [30] on treatment outcome. Although diabetes mellitus and osteoporosis have been associated with increased incidence of NP [31, 32], their effect as risk factors for treatment outcomes has seldom been analysed. Weigl et al. [28], in a prospective cohort study of 112 patients with chronic NP treated with multidisciplinary rehabilitation, reported no correlation between co-morbidities and treatment outcome. However, our results indicate that the presence of baseline diabetes and osteoporosis were risk factors for failure to achieve MCID for pain and disability. Diabetes can negatively alter the structural composition and mechanical properties of the intervertebral disc increasing susceptibility to disc prolapse [31], whereas osteoporosis can aggravate degenerative changes in the cervical spine, especially with increasing age, in patients with NP [32].

To the best of our knowledge, the current study is the largest study which determined predictors of MCID for pain and disability after an MDT-based multimodal rehabilitation treatment and the first study to evaluate the association between diabetes, osteoporosis, number of treatment sessions, and treatment compliance on treatment outcome. However, there are some limitations to this study. First, the retrospective study design had its inherent limitations and biases such as selection bias. Second, predictors of failure to achieve minimal clinical important difference were analysed at the end of a mean treatment duration of 3 months in the current study and not for longer duration of 6 and 12 months. However, previous studies have reported that predictors of treatment outcome at the end of treatment were similar to factors which affected treatment outcomes after 6 and 12 months [26,27,28]. Third, at-home self-treatment compliance for the patient of the prescribed MDT-based repeated movements, an integrated part of the MDT technique, maybe an important factor which may influence MCID and has been not been factored in in this study. Fourth, the duration of the pain or the disability was not used for analysis in the current study. Patients presented at our clinic with a wide variety of self-reported patterns of symptom duration which included acute mild, acute severe, chronic mild, chronic severe, acute aggravation of chronic pain, subacute pain, episodic pain, and recurrent pain. Furthermore, pain duration may change with the use of pain medications or activity modification done before presentation at our clinic. Therefore we avoided subgrouping our patients into the broad categories of acute, subacute, chronic NP based on symptom duration as it may have not represented the actual clinical course of their NP. Hence we focused on the characteristics of the current pain episode in terms of pain intensity and disability and classified them into mechanical or non-mechanical NP from a pathological point of view and as responders or non-responders to repeated movement testing for mechanical NP. Therefore, the results of our study are applicable to mechanical NP with or without a response to repeated movement testing and the effect of treatment based on baseline pain intensity and disability rather than to NP categorised as acute, subacute, or chronic. Lastly, while the current study has included several variables, including diabetes, osteoporosis, number of treatment sessions, and compliance rate in the model, there may still be other factors such as tobacco use, and psychosocial factors that could help predict the outcome more precisely and needs to be included in the future studies.

Conclusion

In conclusion, age > 70 years, diabetes, osteoporosis, partial or non-response to repeated movement testing, lesser number of treatment sessions, and lower compliance rate, were associated with increased risk for failure to achieve MCID for NPRS and NDI scores after treatment. A higher pre-treatment NDI score was associated with failure to achieve MCID for NPRS score, and lower pre-treatment NPRS and NDI scores were associated with failure to achieve MCID for NDI score. Identifying and addressing some of the modifiable risk factors as part of rehabilitation treatment may help achieve better outcomes in mechanical NP.