Study design
This study was a prospective observational study that evaluated whether visual feedback during ultrasound-guided IFI influences the subsequent course of pain in patients with MPS, with expectations for treatment as a mediator. Visual feedback was defined as IFI administration while patients viewed the ultrasound image with their doctor.
Study Participants And Eligibility Criteria
This study was conducted from March 2019 to March 2020 at two facilities in Japan: Kansai Medical University Hospital and Omotesando Pain Clinic. During this period, we continuously enrolled outpatients who received ultrasound-guided IFI for MPS of the upper back. MPS was diagnosed based on the following criteria: 1) a tender spot located with palpation, with or without referral of pain; 2) recognition of symptoms by the patient during palpation of the tender spot; and 3) at least three of the following: a) muscle stiffness or spasm, b) limited range of motion (ROM) of an associated joint, c) pain worsening with stress, and d) palpation of a taut band and/or nodule associated with the tender spot [13]. Diagnoses of MPS required careful manual examination, which is considered highly reliable [14]. The upper back was defined as the region below the neck and above the costal margin [15].
The exclusion criteria were: (1) patients who were younger than 20 years and (2) patients who had any comorbid psychiatric disease or condition that made communication difficult, such as cognitive impairment or delirium.
Intervention
IFIs (5 mL saline per injection) were administered by a doctor at each institution to four bilateral upper and lower sites at the boundary between the levator scapula and trapezius muscle. Doctors used ultrasound to confirm the following points in real-time: 1) boundary between the levator scapula and trapezius muscle; 2) position of the needle tip; 3) saline injection; and 4) removal of the fascia. A single IFI session was performed on the day of intervention.
Before starting treatment, a doctor at each institution verbally explained the ultrasound-guided IFI procedure (i.e., points 1) to 4) above) to the patients. Patients were then asked whether they would like to request to view the real-time ultrasound images during the procedure. The merits and demerits of visual feedback were not shared with the patients. Patients who viewed the real-time ultrasound images were defined as the visual feedback group, and patients who did not view the real-time ultrasound images were defined as the non-visual feedback group. Both groups spent equal amounts of time in each session. After the ultrasound-guided IFI, patients were instructed to perform two 3-minute sessions of a self-stretching exercise (right-left cervical rotation) per day. Assessments were carried out by a doctor in the outpatient department or clinic of each institution on the following days: before intervention (T0), the day of the IFI (T1), and 14 days after the IFI (T2).
Measures
Clinical demographic characteristics
Clinical demographic information was obtained from all subjects and included age, sex, institution (hospital or clinic), primary illness (cancer or non-cancer), pain numerical rating scale (NRS) before the IFI, NRS to assess expectations for treatment before IFI, and analgesic drug use.
Measures of pain intensity and criterion for pain reduction
Average pain intensity was assessed using an 11-point NRS for pain, which ranged from 0 (no pain) to 10 (worst possible pain) [16]. The questionnaire was self-administered and contained the following question: ‘‘How intense was your average pain over the past 24 hours?’’ For patients with multiple MPS sites, we used the average pain NRS score. For pain at T1, pain intensity was evaluated as pain at the time. The reliability and validity of this scale have been established previously [17]. The criterion for pain reduction was determined as ≥ 50% improvement in pain NRS score after the intervention. The optimal cut-off point for NRS rate of change has been reported to be 50% when determining the proportion of patients with clinically significant pain improvement [18].
Measures of expectations for treatment and criteria for increased expectations
Expectation intensity was determined using an NRS to assess expectations for treatment, which ranged from 0 (no expectations) to 10 (highest expectations). The questionnaire was self-administered and contained the following question: “How well do you expect this treatment to reduce pain?” The validity of this questionnaire is not clear, but it has frequently been used in research [10, 12]. The criterion for increased expectations was determined as an NRS score ≥ 8 or ≥ 33% improvement in NRS score for expectations for treatment after IFI.
Measures of cervical ROMs
Cervical ROM for flexion, extension, lateral flexion, and rotation were measured using a goniometer (TAKUMED, Kyoto, Japan), which is an objective and reliable method [19]. All measurements were obtained by one doctor at each institution.
Outcomes
The primary endpoint was the proportion of patients who showed an improvement in pain NRS score by ≥ 50% at 14 days after ultrasound-guided IFI, between patients with and without the visual feedback. The secondary outcomes were NRS score change for expectations for treatment and pain, cervical ROM, correlation between increased expectations and pain NRS reduction using path analysis, and adverse events.
Sample Size Calculation
Because previous studies on this subject are limited, sample size calculation was performed based on reports using similar therapeutic methodologies [2, 5–8]. We assumed that the response rates of IFI with and without visual feedback are 65% and 40%, respectively. The sample size required to achieve 80% statistical power at a 5% two-sided significance level was 62 patients per group, which was calculated using a chi-square test. Considering 10% rate of withdrawal, we determined a total sample size of 136 patients. The effective rate described here is the proportion of patients whose NRS scores reduced by ≥ 50% after treatment.
Statistical analysis
Data are reported as means and standard deviations, medians with interquartile ranges, or frequencies (%), as appropriate. The proportion of study participants with visual feedback among all participants were estimated, including the exact 95% confidence intervals (95% CI). The rate of participants with pain reduction at 14 days after IFI for each group were estimated, including the exact 95% CI.
The study participants were classified into two groups: the visual feedback and non-visual feedback groups. Unpaired t-tests were used for comparisons of the dependent variables of age, pain NRS score before IFI, and NRS score for expectations for treatment before IFI. Pearson’s chi-square tests were used to analyze the dependent variables of sex, institution, primary illness (cancer), and analgesic drug use.
The proportion of patients who showed ≥ 50% improvement in pain NRS score at 14 days after intervention was analyzed using a chi-square test.
Changes in the course (T0, T1, and T2) of NRS scores for expectations for treatment, pain NRS scores, and cervical ROM scores were analyzed using one-way repeated measures analyses of variance (ANOVA) for each group. To conduct comparisons between groups, time course was used as the within-subjects factor and group was used as the between-subjects factor in a two-way repeated measures ANOVA. Multiple comparisons were corrected using the Bonferroni method. If participants withdrew from the study, NRS scores after withdrawal were substituted with scores immediately before withdrawal. Change in analgesic drug use during the period and loss to follow up were classified as withdrawals from the study.
Path analyses were conducted to estimate the direct and indirect paths with reference to correlation coefficients. A hypothetical model was created in which visual feedback, pain reduction at T1, expectations at T0, and increased expectation at T1 predicted pain reduction at T2. Visual feedback, age, analgesic drug use, NRS at T0, expectations at T0, pain reduction at T1, and increased expectations at T1 were mediators of increased expectations and pain reduction at T2. Figure 1 shows the hypothetical model (Akaike information criterion [AIC] = 116.832). Path analyses were performed by removing paths with p < 0.05, adjusting paths with reference to the modification index, repeating model correction while checking the goodness of fit index (GFI), and investigating correlations between factors specifying pain reduction at T2. To assess fit, we used model chi-square values, GFI, comparative fit index (CFI), root mean square error of approximation (RMSEA), and AIC. Smaller chi-square values, > 0.95 for GIF and CFI values, and ≤ 0.08 RMSEA values indicate good model fit [20]. The AIC was used to compare the hypothetical model with the modified model; a lower AIC value indicated a better model.
A value of p < .05 was considered statistically significant. Statistical analyses were performed using SPSS version 25.0 and Amos version 25.0 for Macintosh (SPSS, Inc., IBM, Chicago, IL, USA).