Does adding cervical mobilization to shoulder mobilization improve pain, sensation, and function in subacromial impingement syndrome? A three-arm randomized controlled trial

Abstract

Background

The impact of adding cervical mobilization in Subacromial Impingement Syndrome (SIS) remains underexplored. This study investigated the effects of adding cervical to shoulder mobilization on pain, sensation, and function in patients with Subacromial Impingement Syndrome.

Methods

The study was registered at ClinicalTrials.gov (NCT06602206) on 30/07/2024. Participants (n = 45, aged 27–65) were randomly allocated into three groups (n = 15 each): a conventional treatment group (6 males, 9 females), a shoulder mobilization group (6 males, 9 females), and a cervical mobilization group (5 males, 10 females). Pain was assessed using the visual analog scale (VAS), pressure pain threshold (PPT) with a handheld dynamometer, tactile sensation with a two-point discrimination test using an esthesiometer, range of motion (ROM) and pain-free ROM with a goniometer, and functional status with the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire. All participants underwent a 15-session treatment program. Assessments were conducted before and after the intervention.

Results

All groups improved in pain, ROM, sensation, and function (p < 0.05). The cervical mobilization group showed a greater reduction in VAS scores than the others (F = 5.251, p = 0.009). Pain-free ROM improved most in the cervical and least in the control group (F = 3.501, p = 0.039). PPT and tactile sensation increased significantly in the cervical group (F = 3.256, p = 0.048; F = 3.529, p = 0.038). Both mobilization groups improved more in DASH scores than controls (F = 3.933, p = 0.027). All outcomes had large effect sizes.

Conclusion

While all interventions were beneficial in SIS, cervical mobilization combined with conventional treatment and shoulder mobilization proved most effective in reducing pain, enhancing pain-free ROM, and improving sensory function. Minimal Clinically Important Differences values were established for all assessment tools.

Highlights

This study used a three-arm design to assess treatments for subacromial impingement.

First to report MCID for pain-free arc, ROM, and sensation providing novel references.

First to show cervical plus shoulder mobilization improves tactile sensation in SIS.

Shoulder plus cervical mobilization reduced pain, improved pain-free arc & sensation in SIS.

Introduction

Shoulder pain is prevalent across all ages and activity levels, with Subacromial Impingement Syndrome (SIS) being the most common cause¹. SIS is a common musculoskeletal disorder involving abnormalities of subacromial bursa and coracoacromial ligament, along with repeated impingement or abrasion of the rotator cuff or long head of the biceps beneath the anterior acromion [1–3].

Several studies in the literature support the association between SIS and cervical dysfunction and emphasize that the cervical and thoracic regions should not be overlooked during shoulder rehabilitation [4, 5]. The interaction between cervical dysfunction and subacromial impingement syndrome (SIS) is a bidirectional process occurring through both neurological and biomechanical pathways. Recent evidence further supports this relationship, demonstrating a significant association between shoulder dysfunction and concomitant neck disability, as well as a notable incidence of impingement syndrome in patients with cervical radiculopathy [6, 7]. However, findings remain inconsistent: while some report beneficial effects of cervicothoracic manual therapy others find no significant outcomes [8–11]. Therefore, the effectiveness of cervicothoracic manual therapy techniques in conjunction with exercise remains inconclusive. Previous studies investigating cervicothoracic manual therapy in patients with SIS have primarily reported short-term improvements in pain intensity, ROM, and functional outcomes; however, substantial heterogeneity exists in intervention protocols, targeted spinal regions, and outcome measures [12–14]. This heterogeneity limits the generalizability of existing findings. Moreover, only a limited number of studies have explored underlying sensory mechanisms, leaving a clear knowledge gap regarding the effects of cervical mobilization on sensory processing in SIS.

From a mechanistic perspective, cervical mobilization may influence shoulder-related sensory and clinical outcomes through modulation of central pain processing mechanisms [15]. In individuals with chronic shoulder pain due to SIS, evidence of secondary hyperalgesia suggests the presence of central sensitization, characterized by an amplification of nociceptive signalling within the central nervous system [16–19]. Central sensitization is associated with lowered pain thresholds, enhanced responsiveness to both noxious and non-noxious stimuli, and pain hypersensitivity that extends beyond the site of tissue injury [15]. Altered afferent input from cervical spinal segments, which share convergent neural pathways with the shoulder region, may contribute to increased central excitability and persistent pain experiences. Cervical mobilization has been proposed to modulate spinal and supraspinal pain processing, potentially reducing central hypersensitivity and leading to improvements in sensory disturbances and pain perception at remote regions such as the shoulder. This supports a predominantly neurophysiological mechanism underlying the effects of cervical mobilization rather than a purely biomechanical explanation [20–22].

Emerging evidence suggests that chronic shoulder pain associated with SIS involves both peripheral and central sensitization mechanisms, which may lead to sensory disturbances around the shoulder region [23]. Despite this, many studies have predominantly focused on pain intensity, range of motion, and functional outcomes, while investigations addressing sensory alterations remain limited and inconclusive [11, 24–26]. This highlights an important gap in the literature regarding sensory processing in patients with SIS.

Despite these theoretical considerations, evidence regarding the effects of shoulder and cervical mobilization on sensory outcomes in patients with SIS remains limited, and sensory assessments are often absent from clinical trials. Given that sensory disturbances represent a major gap in the existing literature, sensory outcomes were defined as the primary outcome measures of this study, with pain intensity, ROM, and functional outcomes considered secondary. Accordingly, this study aimed to investigate the effectiveness of adding cervical mobilization to shoulder mobilization in patients with SIS using a comprehensive assessment framework that includes sensory parameters. It was hypothesized that the addition of cervical mobilization would be superior to shoulder mobilization alone in improving sensory outcomes, pain, ROM, and functional performance.

Methods

Study design

This study was designed in accordance with the CONSORT 2010 guidelines and was registered at ClinicalTrials.gov (NCT06602206) on 30/07/2024. The protocol adhered to the recommendations of the EQUATOR network, specifically the “Consensus on Interventions Reporting Criteria List for Spinal Manipulative Therapy (CIRCLe SMT)” guideline for spinal manipulative interventions [27].

Participant selection

The study was approved by the Non-Interventional Clinical Research Ethics Committee of Kutahya Health Sciences University (Decision No: 2024/06–29) and conducted per the Declaration of Helsinki. Participants diagnosed with SIS by a physical therapy specialist at Kutahya State Hospital were included in the study. The diagnosis was established based on clinical examination supported, when necessary, by imaging findings such as magnetic resonance imaging or X-ray, in accordance with routine clinical practice.

Sample size estimation was conducted using G*Power software (version 3.1) based on the primary sensory outcome (pressure pain threshold). A moderate effect size (d = 0.5), derived from previous studies on cervical or cervicothoracic mobilization in shoulder or neck disorders [22, 28], was assumed. Group differences among the three interventions were analysed using a one-way ANOVA, with an alpha level of 0.05 and 80% power. No adjustment for multiple comparisons was applied, resulting in a required sample size of 45 participants (15 per group). A total of 57 individuals were initially assessed for eligibility. Twelve participants who did not meet the inclusion criteria were excluded, resulting in 45 participants enrolled in the study. During the intervention period, five participants withdrew from the study (three from the Conventional Treatment group and one from each of the Shoulder and Cervical Mobilization groups). Reasons for withdrawal included inability to obtain leave from work, self-reported symptomatic improvement, and long travel distance to the study site. Participants who withdrew were replaced with new eligible individuals using age- and sex-stratified randomization to maintain balance between the groups (Fig. 1.). Prior to inclusion, all participants were re-evaluated by the researcher using a standardized cluster of clinical tests, including the Neer, Hawkins–Kennedy, Jobe, and painful arc tests. Although individual tests demonstrate limited diagnostic accuracy, their combined use has been shown to improve the identification of subacromial impingement syndrome. Reported sensitivity and specificity values are approximately 79% and 59% for the Hawkins–Kennedy test, 72% and 60% for the Neer test, 71% and 74% for the Jobe test, and 53% and 76% for the painful arc test, respectively [29, 30].

Fig. 1.

Participant flow diagram

Inclusion criteria

Diagnosis of SIS, Age 18 or older, Shoulder pain persisting for at least 3 months, VAS score ≥ 4, No shoulder treatment within the past 6 months [4, 8, 10, 11].

Exclusion criteria

History of surgery in the shoulder, cervical, or thoracic regions, Frozen shoulder, shoulder instability, or full-thickness rotator cuff tear, Systemic rheumatologic or neurological disease, History of upper extremity fracture, Diagnosed scoliosis [4, 8, 10, 11].

Randomization

Participants were stratified by age (below and above the median age of 50 years) and sex. Within each stratum, a computer-generated randomization sequence using permuted blocks was created by an independent researcher who was not involved in participant recruitment, assessment, or treatment allocation. Randomization was performed using an online randomization tool (https://www.random.org/lists; generated on 30 September 2024).

Eligible participants were enrolled by the principal investigator. Group allocation was concealed using sequentially numbered, opaque, sealed envelopes, which were opened only after baseline assessments were completed. Participants were stratified by age and sex and randomly assigned (via https://www.random.org/lists, 30/09/2024) to three groups (n = 15 each).

1. Conventional Treatment.

2. Conventional Treatment + Shoulder Mobilization.

3. Conventional Treatment + Shoulder and Cervical Mobilization.

Intervention

Assessment

Assessment criteria

All participants underwent pre-treatment and 15-session post-treatment assessments.

Sociodemographic form

Collected information on age, sex, height, weight, marital status, education level, occupation, work conditions, and chronic diseases. Chronic conditions reported included 7 cases of diabetes, 5 hypertension, and 1 hyperlipidaemia. All medications were prescribed by the same physician. Medication use was confirmed only before treatment; no participants used medication during the intervention phase.

Pressure pain threshold

Pressure pain threshold (PPT) was assessed using a handheld dynamometer (MicroFET 2, Hoggan Scientific, USA) equipped with a 1 cm² rubber probe, which has been used in previous studies as an alternative method for algometry-based measurements. PPT is commonly evaluated using algometers; however, handheld dynamometers have also been reported in the literature as a valid and reliable tool for assessing pressure pain sensitivity in musculoskeletal conditions. Measurements were taken over the upper trapezius, supraspinatus, and middle deltoid muscles. Pressure was applied perpendicular to the skin at a rate of approximately 30 kPa/s until the participant verbally reported pain onset. Each measurement was repeated three times with 30–60 s rest between trials, and the mean value was used for analysis [31–33].

Tactile sensation

Two-point discrimination (2PD) was assessed using a Baseline aesthesiometer over the C5, C6, and C7 dermatomes of the shoulder region. To standardize measurement sites, vertical reference lines were drawn from the anterior, middle, and posterior borders of the acromion toward the elbow. Testing was performed using an ascending method starting from 0 mm, with inter-point distance gradually increased in 2–3 mm increments. The distance at which the participant first perceived two distinct points was recorded in millimetres. Only ascending trials were performed; therefore, measurements may be influenced by perceptual response bias, which is acknowledged as a methodological limitation [31, 34]. While intra-rater reliability was not directly calculated in the current study, the assessment was conducted in accordance with standardized protocols previously described by Moberg (1958) and Dellon (1987), which have been widely used to ensure consistency in two-point discrimination measurements [35, 36].

Pain assessment

Pain intensity was assessed using the VAS, a validated and reliable self-report tool ranging from 0 (no pain) to 10 (worst imaginable pain) [37]. Participants were instructed to rate their worst shoulder pain experienced during active shoulder movements within the previous week.

Pain-free range of motion

Pain-free ROM was measured using a universal goniometer during active shoulder abduction in the frontal plane. Participants were instructed to actively elevate their arm at a comfortable speed until the first onset of pain was perceived, and this point was recorded as the beginning of the painful range. Participants were then asked to continue the movement to their maximum available range, and the point at which pain subsided was also verbally reported. The angular difference between the onset and cessation of pain was recorded in degrees. All measurements were performed in a standardized seated or standing position by the same assessor, with attention given to minimizing compensatory trunk movements. Shoulder ROM measurements using a universal goniometer have shown excellent intra-rater (ICC = 0.89–0.99) and good to excellent inter-rater reliability [38–40].

The painful arc test has shown moderate diagnostic utility for subacromial impingement syndrome, with sensitivity of 0.53–0.62 and specificity of 0.60–0.76. Kelly et al. (2010) reported an overall accuracy of 62.1% and highlighted that its diagnostic value is limited when used alone but increases when combined with other clinical tests [41].

Range of motion

Active ROM of the shoulder (flexion, extension, abduction, adduction, internal and external rotation) was measured using a universal goniometer according to international guidelines [42]. Measurements were performed within the pain-free ROM and were terminated at the point where pain first limited further movement, rather than maximal stiffness.

Functionality

Upper extremity functions were assessed using the Turkish version of the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire. The main section of the questionnaire, the DASH disability/symptom score, consists of 30 items. To be considered valid, at least 27 of the 30 items must be completed. Each item is rated on a 5-point Likert scale, and the total score is calculated on a scale from 0 to 100 [43]. In addition, the DASH includes two optional modules designed to assess work and sport/music performance. Each module consists of four items, with each item scored out of 25 points, resulting in a total score ranging from 0 to 100 for each modüle.

Treatment

All participants received a conventional physiotherapy program consisting of hot pack, therapeutic ultrasound, and transcutaneous electrical nerve stimulation (TENS), administered by trained hospital physiotherapy staff. Conventional TENS was applied using a frequency of 80–100 Hz and a pulse duration of 100 µs at a sensory-level intensity, without muscle contraction. The application lasted 20 min, and electrodes were placed around the painful shoulder region. Therapeutic ultrasound was applied using a frequency of 1 MHz and an intensity of 1.0–1.5 W/cm² in continuous mode for 5 min. The ultrasound head was moved in circular motions over the subacromial region. Sessions were conducted 5 times per week for 3 weeks, with each session lasting approximately 30 min.

Exercise therapy was delivered individually by an experienced physiotherapist using a standardized protocol to ensure consistency across groups.

All shoulder and cervical mobilization techniques were delivered by the same physiotherapist at each session according to predefined protocols to ensure treatment fidelity; however, therapist-specific effects could not be fully isolated. No adverse events or treatment-related complications were reported during the intervention period.

Regarding co-interventions, medication use was assessed at baseline and monitored throughout the intervention via participant self-report at each session. Participants were instructed not to use analgesic or anti-inflammatory medication during the intervention period, and no medication related to shoulder pain was reported. Chronic conditions such as diabetes or hypertension were recorded, and participants with these conditions continued their routine medications without any changes during the study.

Conventional treatment

All participants received hot pack, TENS, and therapeutic ultrasound administered by hospital staff. The exercise program included stretching and strengthening components. Stretching involved pectoralis muscle stretches at 90° and 135° shoulder elevation against a wall and posterior capsule stretches. Stretching was performed as one set consisting of five repetitions, each held for 15 s. Strengthening consisted of resisted internal/external rotation, scapular retraction at 90° elbow flexion and full extension, and scapular plane elevation in the “full can” position. Strengthening was performed as three sets of ten repetitions, supervised by an experienced physiotherapist (7 years of clinical practice).

The exercise program was structured according to the FITT-VP principles.

Frequency

Exercises were performed five sessions per week for 3 weeks (total of 15 sessions).

Intensity

Resistance was individually adjusted based on patient tolerance using TheraBand (red or green) for scapular exercises and 0.5–2.5 kg dumbbells for shoulder elevation exercises, determined by 10-repetition maximum.

Time

Each session included stretching and strengthening exercises lasting approximately.

one session per day.

Type

The program included stretching (pectoralis muscle and posterior capsule stretching).

and strengthening exercises (resisted internal/external rotation, scapular retraction, and scapular plane elevation in the “full can” position).

Volume

Stretching was performed as one set of five repetitions with 15-second holds, while strengthening exercises were performed as three sets of ten repetitions.

Progression

Resistance was kept constant throughout the intervention period based on initial tolerance level, ensuring standardized loading across sessions. All exercises were supervised by an experienced physiotherapist with 7 years of clinical experience.

Shoulder mobilisation

Maitland mobilization, a passive, rhythmic oscillatory technique, was applied at Grade III intensity, characterized by large-amplitude movements within the available ROM [44, 45]. In the group that received shoulder mobilization in addition to conventional therapy, lateral distraction, anterior-posterior glide, and inferior glide techniques were applied to the glenohumeral joint. For lateral distraction, patients lay supine while the therapist grasped the humerus and mobilized the humeral head laterally, anteriorly, and inferiorly, perpendicular to the glenoid surface. In inferior gliding, with the shoulder at 45° abduction and under slight traction, the therapist applied a downward force to the humeral head. For anterior-posterior gliding, the same position was maintained, and the therapist mobilized the humeral head in anterior and posterior directions. All mobilizations were performed with 30 oscillations per set, each lasting 1–2 s, for three sets. All mobilizations were performed by a physiotherapist with 7 years of clinical experience.

Cervical mobilisation

In the cervical mobilization group, bridge technique, lateral flexion traction, anteroposterior sliding and lateral sliding were applied in the grade 3 range. In the bridge technique, the patient lay supine while the therapist, seated at the head of the table, positioned fingers under the occiput, gently extended the head, and applied brief traction. For traction with lateral flexion, the therapist supported the head and chin, applied slight traction, and induced lateral flexion bilaterally. In the anteroposterior glide, anterior and posterior movements were performed via chin control under slight traction. The final lateral glide technique, without traction, involved resting the patient’s head on the therapist’s abdomen, using body weight shifts to induce controlled elliptical movements without lateral flexion. All mobilizations were performed with 30 oscillations per set, each lasting 1–2 s, for three sets. All mobilizations were performed by a physiotherapist with 7 years of clinical experience.

Statistical analysis

Data were analyzed using IBM SPSS Statistics Standard Concurrent User Version 26 (IBM Corp., Armonk, NY, USA). Descriptive statistics were presented as number (n), percentage (%), mean (X), standard deviation (SD), median (M), minimum, and maximum values. Normality of continuous variables was assessed using the Shapiro–Wilk test. A skewness value within ± 2.0 and a kurtosis value below 7.0 were accepted as indicators of normal distribution. Accordingly, the data met the assumptions of normality and parametric tests were applied.

One-way analysis of variance (ANOVA) was used for between-group comparisons of continuous variables, while categorical variables were analyzed using Pearson’s chi-square or Fisher’s exact tests. A mixed-design ANOVA was conducted to examine group differences across follow-up time points. Since no significant differences were observed in baseline measurements between groups, analysis of covariance (ANCOVA) was not performed. Bonferroni correction was applied for multiple comparisons of main effects. Effect size was interpreted using η² values, and a p-value < 0.05 was considered statistically significant.

Minimal Clinically Important Difference (MCID) values were determined using both distribution-based and anchor-based approaches. In the distribution-based method, MCID was calculated as 50% of the standard deviation of baseline measurements (0.5 × SD). The 95% confidence interval (CI) for the MCID was calculated using the standard error (SE), with lower and upper bounds determined using the formula: 95% CI = MCID ± 1.96 × SE [46].

The standard error was calculated based on the sample size and standard deviation. These confidence intervals were reported to reflect the precision of the MCID estimates. In the anchor-based analysis, participants perceived level of change was used as the reference, and the mean change score of the subgroup reporting minimal improvement was accepted as the MCID value [46].

Results

The study included 45 participants aged 27–65 years: 15 in each group. Group distributions were similar in sex (Conventional: 6 M/9F, Shoulder: 6 M/9F, Cervical: 5 M/10F). Demographic characteristics such as age, sex, BMI, chronic disease status, and dominant/affected side are summarized in Table 1. Shapiro–Wilk test results indicated that all continuous variables were normally distributed (p > 0.05 for all), supporting the use of parametric analyses.

Table 1.

Comparison of descriptive characteristics of the participants by groups (N = 45)

	Group
	Conventional Treatment	Shoulder Mobilization	Cervical Mobilization	Test (p)
	n = 15	n = 15	n = 15
Age
X ± SS	49.27 ± 10.11	46.93 ± 8.46	51.80 ± 8.50	F = 1.083 p = 0.348
M (min-max)	52 (27–65)	45 (33–62)	52 (37–65)
Gender, n (%)
Male	6 (%40)	6 (%40)	5 (%30)	χ2 = 0.189 p = 0.910
Female	9 (%60)	9 (%60)	10 (%70)
Body Max index, (kg/m2)
X ± SS	28.70 ± 7.22	27.18 ± 5.10	29.43 ± 5.61	F = 0.544 p = 0.585
M (min-max)	25.8 (18.2–43.1)	26.1 (19.1–38.1)	29 (19.5–38.9)
Chronic Disease, n (%)
Yes	4 (%30)	4 (%30)	5 (%30)	χ2 = 0.216 p = 0.897
No	11 (%70)	11 (%70)	10 (%70)
Dominant side, n (%)
Right	14 (%90)	15 (%100)	15 (%100)	χ2 = 2.045 p = 0.360
Left	1 (%10)	0 (%0)	0 (%0)
Affected Side, n (%)
Right	12 (%80)	8 (%50)	11 (%70)	χ2 = 2.696 p = 0.260
Left	3 (%20)	7 (%50)	4 (%30)

ANOVA (F); Chi-square test (χ²). Descriptive statistics are presented as mean (X̄), standard deviation (SD), median (M), minimum (min), maximum (max), number (n), and percentage (%)

A mixed-methods ANOVA revealed no significant baseline differences between groups for VAS (F = 0.006, p = 0.994) and pain-free ROM (F = 0.390, p = 0.680). A significant main effect of time was observed for both outcomes across all groups (VAS: p < 0.001; pain-free ROM: p < 0.001), indicating significant pre- to post-treatment improvements. A significant group × time interaction was found for VAS (F = 5.251, p = 0.009) and pain-free ROM (F = 3.501, p = 0.039), with effect sizes of η² = 0.20 and η² = 0.14, respectively. Post-hoc Bonferroni-adjusted comparisons showed that the cervical mobilization group demonstrated significantly greater improvements in both VAS and pain-free ROM compared to the other groups (p < 0.05) (Table 2).

Table 2.

Comparison of VAS and pain-free rom measurements across follow-up periods by groups

	Group
	Conventional Treatment a	Shoulder Mobilization b	Cervical Mobilization c	Test Statistics†	Post Hoc (Bonferroni)	MCID (95% CI)
	n = 15	n = 15	n = 15
VAS (cm)	X ± SS	X ± SS	X ± SS
Pre-treatment  Post-treatment	6.82 ± 1.27 cm 3.32 ± 1.13 cm	6.77 ± 1.38 cm 2.87 ± 1.96 cm	6.77 ± 1.41 cm 1.36 ± 1.68 cm	F = 0.006 p = 0.994 η²=0.00 F = 5.950 p = 0.005 η²=0.22		-3.08 cm (-4.45; -1.7)
					(a = b) > c
Test Statisticsϕ	F = 63.307 p < 0.001 η²=0.601	F = 78.873 p < 0.001 η²=0.653	F = 151.44 p < 0.001 η²=0.783
Difference& (post-pre)	-3.50 ± 1.31 cm	-3.91 ± 1.82 cm	-5.41 ± 1.92 cm	F = 5.251 p = 0.009 η²=0.20	(a = b) > c
Pain-Free Rom (°)	X ± SS	X ± SS	X̄± SS
Pre-treatment  Post-treatment	90.00° ± 26.73° 114.67° ± 23.94°	90.00° ± 28.03° 124.33° ± 28.28°	97.00° ± 19.62° 140.67° ± 16.02°	F = 0.390 p = 0.680 η²=0.02 F = 4.770 p = 0.014 η²=0.19	(a = b) < c	7.2° (2.5; 11.91)
Test Statisticsϕ	F = 23.603 p < 0.001 η²=0.36	F = 45.728 p < 0.001 η²=0.521	F = 73.97 p < 0.001 η²=0.638
Difference& (Post-Pre)	24.67° ± 18.17°	34.33° ± 21.87°	43.67° ± 18.75°	F = 3.501 p = 0.039 η²=0.14	a < b< c

MCID Minimal Clinically Important Difference, CI Confidence Interval, cm Centimetre

ANOVA (F), Effect Size (η²), ^ϕWithin-group comparisons, ^†Between-group comparisons, ^&Comparison of differences between pre- and post-test scores. Descriptive statistics are presented as mean (X̄), standard deviation (SD)

Statistically significant results are shown in bold (p < 0.05)

A mixed-design ANOVA revealed no significant baseline differences between groups for pressure pain threshold (PPT) and tactile sensation across all measurement sites (p > 0.05). A significant main effect of time was observed for all PPT measurements (upper trapezius, middle deltoid, and supraspinatus) and tactile sensation (anterior, middle, and posterior shoulder), indicating significant improvements following treatment in all groups (p < 0.05). A significant group × time interaction was found only for upper trapezius PPT (F = 3.256, p = 0.048, η² = 0.13) and anterior shoulder tactile sensation (F = 3.529, p = 0.038, η² = 0.14). Post-hoc Bonferroni-adjusted comparisons showed that the cervical mobilization group demonstrated significantly greater improvements in upper trapezius PPT and anterior shoulder tactile sensation compared to the other groups (p < 0.05). No significant group × time interaction was observed for middle deltoid or supraspinatus PPT, or for middle and posterior shoulder tactile sensation (p > 0.05) (Table 3).

Table 3.

Comparison of sensory evaluation measurements across follow-up periods by groups

Pressure Pain Threshold (kPa)
		Conventional Treatment a	Shoulder Mobilization b	Cervical Mobilization c	Test Statistics†	Post Hoc (Bonferroni)	MCID (95% CI)
Upper Trapezius
Pre-treatment  Post-treatment		7.67 ± 4.45 9.62 ± 3.11	5.47 ± 2.02 8.45 ± 2.27	6.67 ± 2.66 11.19 ± 3.67	F = 1.774 p = 0.182 η²=0.08 F = 3.009 p = 0.060 η²=0.13		1.73 (-0.07; 3.53)
Test Statisticsϕ		F = 7.324 p = 0.010 η²=0.148	F = 17.239 p < 0.001 η²=0.291	F = 39.601 p < 0.001 η²=0.485
Difference& (Post-Pre)		1.95 ± 3.34	2.99 ± 2.26	4.53 ± 2.65	F = 3.256 p = 0.048 η²=0.13	(a = b) < c
Middle Deltoid
Pre-treatment  Post-treatment		7.20 ± 2.38 10.25 ± 2.73	6.42 ± 2.35 10.2 ± 3.74	6.18 ± 2.43 8.83 ± 2.38	F = 0.750 p = 0.479 η²=0.03 F = 1.076 p = 0.350 η²=0.05		1.89 (0.23; 3.55)
Test Statisticsϕ		F = 18.142 p < 0.001 η²=0.302	F = 27.805 p < 0.001 η²=0.398	F = 13.7 p < 0.001 η²=0.246
Difference& (Post-Pre)		3.88 ± 3.58	3.69 ± 4.77	4.37 ± 2.99	F = 0.635 p = 0.535 η²=0.03
Supraspinatus
Pre-treatment  Post-treatment		8.76 ± 4.60 12.64 ± 3.81	7.50 ± 3.08 11.19 ± 3.74	8.11 ± 3.69 12.47 ± 4.38	F = 0.403 p = 0.671 η²=0.02 F = 0.596 p = 0.555 η²=0.03		1.93 (-0.1; 3.97)
Test Statisticsϕ		F = 15.234 p < 0.001 η²=0.266	F = 13.754 p < 0.001 η²=0.247	F = 19.296 p < 0.001 η²=0.315
Difference& (Post-Pre)		3.05 ± 2.93	3.78 ± 3.06	2.65 ± 2.28	F = 0.124 p = 0.883 η²=0.01
Tactile Sensation (mm)
Anterior shoulder
Pre-treatment  Post-treatment	52.33 ± 11.32 37.33 ± 10.83		58.33 ± 20.85 36.00 ± 12.28	63.67 ± 18.37 32.67 ± 7.76	F = 1.607 p = 0.212 η²=0.07 F = 0.792 p = 0.460 η²=0.04		-4.42 (-9.52; 0.69)
Test Statisticsϕ	F = 12.376 p = 0.001 η²=0.228		F = 27.436 p < 0.001 η²=0.395	F = 52.861 p < 0.001 η²=0.557
Difference& (Post-Pre)	-15.00 ± 14.02		-22.33 ± 20.43	-31.00 ± 14.29	F = 3.529 p = 0.038 η²=0.14	(a = b) > c
Middle Shoulder
Pre-treatment  Post-treatment	52.00 ± 19.62 38.00 ± 11.77		48.67 ± 20.13 31.67 ± 7.24	54.00 ± 17.95 31.33 ± 10.43	F = 0.294 p = 0.747 η²=0.01 F = 2.118 p = 0.133 η²=0.09		-3.42 (-8.6; 1.77)
Test Statisticsϕ	F = 9.185 p = 0.004 η²=0.179		F = 13.544 p < 0.001 η²=0.244	F = 24.077 p < 0.001 η²=0.364
Difference& (Post-Pre)	-14.00 ± 20.11		-17.00 ± 21.53	-22.67 ± 9.61	F = 0.908 p = 0.411 η²=0.04
Posterior Shoulder
Pre-treatment  Post-treatment	54.00 ± 19.20 40.00 ± 13.89		52.00 ± 21.53 35.67 ± 14.13	53.67 ± 16.20 31.67 ± 11.29	F = 0.047 p = 0.954 η²=0.00 F = 1.504 p = 0.234 η²=0.07		-3.72 (-8.36; 0.93)
Test Statisticsϕ	F = 10.809 p = 0.002 η²=0.205		F = 14.713 p < 0.001 η²=0.259	F = 26.693 p < 0.001 η²=0.389
Difference& (Post-Pre)	-14.00 ± 14.29		-16.33 ± 21.75	-22.00 ± 11.77	F = 0.933 p = 0.401 η²=0.04

MCID Minimal Clinically Important Difference, CI Confidence Interval, mm millimeter, kPa kilopascal

ANOVA (F), Effect Size (η²), ^ϕWithin-group comparisons, ^†Between-group comparisons, ^&Comparison of differences between pre- and post-test scores. Descriptive statistics are presented as mean (X̄), standard deviation (SD)

Statistically significant results are shown in bold (p < 0.05)

A mixed-methods ANOVA revealed no significant baseline differences between groups for any range of motion (ROM) parameters (p > 0.05). A significant main effect of time was observed for all ROM outcomes (flexion, extension, abduction, adduction, internal rotation, and external rotation), indicating significant improvements from pre- to post-treatment in all groups (p < 0.05). No significant group × time interaction was found for any ROM parameter (p > 0.05), suggesting that the magnitude of improvement was similar across groups. A significant group effect was observed only for post-treatment external rotation (F = 4.816, p = 0.013, η² = 0.19), where the conventional treatment group showed significantly lower values compared to the shoulder and cervical mobilization groups (p < 0.05). No other between-group differences were found (Table 4).

Table 4.

Comparison of ROM measurements across follow-up periods by groups

	Group
	Conventional Treatment a	Shoulder Mobilization b	Cervical Mobilization c	Test Statistics†	PostHoc(Bonferroni)	MCID %95 CI)
	n=15	n=15	n=15
Flexion (°)	X̄ ± SS	X̄± SS	X̄ ± SS
Pre-treatment	139.67° ± 18,27°	142.33° ± 27,44°	141,00° ± 24,73°	F = 0.047 p = 0.954 η2 = 0,00		3.96° (-1.76; 9.68)
Post-treatment	156.67°± 20.50°	165.00° ± 13.23°	170.00° ± 9.64°	F = 2.967 p = 0.062 η2 = 0.12
Test Statisticsϕ	F = 8.059 p = 0.007 η2 = 0.161	F = 14.326 p  < 0.001 η2 = 0.254	F = 23.451 p < 0.001 η2 = 0.358
Difference& (Post-Pre)	17.00° ± 17.40°	22.67° ± 25.27°	29.00° ± 25.93°	F = 1.005 p = 0.375 η2 = 0.05
Extansion (°)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	41.67° ± 7.48°	40.33° ± 10.43°	43.00° ± 8.19°	F = 0.345 p = 0.710 η2 = 0.02		4.44° (0.95; 7.93)
Post-treatment	54.33° ± 9.04°	57.33° ± 11.00°	60.33° ± 6.67°	F = 1.639 p = 0.206 η2 = 0.07
Test Statisticsϕ	F = 22.086 p < 0.001 η2 = 0.345	F = 39.782 p < 0.001 η2 = 0.486	F = 41.358 p < 0.001 η2 = 0.496
Difference& (Post-Pre)	12.67° ± 7.99°	17.00° ± 10.99°	17.33°± 11.93°	F = 0.933 p = 0.401 η2 = 0.04
Abduction (°)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	119.47° ± 26.91°	118.33° ± 28.39°	125.00° ± 23.22°	F = 0.277 p = 0.760 η2 = 0.01		5.35° (-0.22; 10.91)
Post-treatment	143.33° ± 23.43°	149.00° ± 21.23°	160.67°± 15.45°	F = 2.838 p = 0.070 η2 = 0.12
Test Statisticsϕ	F = 13.987 p < 0.001 η2 = 0.25	F = 23.093 p < 0.001 η2 = 0.355	F = 31.237 p < 0.001 η2 = 0.427
Difference& (Post-Pre)	23.87° ± 23.33°	30.67° ± 22.27°	35.67° ± 28.15°	F = 0.861 p = 0.430 η2 = 0.04
Adduction (°)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	31.67° ± 10.12°	31.67° ± 8.59°	34.67° ± 9.35°	F = 0.512 p = 0.603 η2 = 0.02		3.6° (0.46; 6.75)
Post-treatment	42.67° ± 7.76°	44.00° ± 9.30°	45.67° ± 5.30°	F = 0.582 p = 0.563 η2 = 0.03
Test Statisticsϕ	F = 26.623 p < 0.001 η2 = 0.388	F = 33.468 p < 0.001 η2 = 0.443	F = 26.623 p < 0.001 η2 = 0.388
Difference& (Post-Pre)	11.00° ± 8.70°	12.33° ± 8.42°	11.00° ± 7.61°	F = 0.130 p = 0.878 η2 = 0.01
Internal rotation (°)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	60.00° ± 15.00°	65.33 ± 17.06	68.67 ± 12.02	F = 1.302 p = 0.283 η2 = 0.06		3.21° (-1.12; 7.53)
Post-treatment	74.67°± 14.45°	79.00° ± 12.85°	82.33° ± 10.83°	F = 1.354 p = 0.269 η2 = 0.06
Test Statisticsϕ	F = 13.661 p < 0.001 η2 = 0.245	F = 11.862 p = 0.001 η2 = 0.22	F = 11.862 p=.,001 η2 = 0.22
Difference& (Post-Pre)	14.67° ± 16.42°	13.67°± 17.97°	13.67° ± 10.77°	F = 0.021 p = 0.979 η2 = 0.00
External rotation (°)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	52.33° ± 17.31°	61.33° ± 19.59°	67.33° ± 12.23°	F = 3.08 p = 0.056 η2 = 0.13		3.59° (-0.76; 7.94)
Post-treatment	67.67° ± 18.60°	77.67° ± 11.16°	83.00° ± 9.78°	F = 4.816 p = 0.013 η2 = 0.19	a< (b = c)
Test Statisticsϕ	F = 13.664 p < 0.001 η2 = 0.245	F = 15.505 p < 0.001 η2 = 0.27	F = 14.265 p < 0.001 η2 = 0.254
Difference& (Post-Pre)	15.33° ± 15.86°	16.33° ± 19.22°	15.67°± 12.37°	F = 0.015 p = 0.985 η2 = 0.00

MCID Minimal Clinically Important Difference, CI Confidence Interval

ANOVA (F), Effect Size (η²), ^ϕWithin-group comparisons, ^†Between-group comparisons, ^&Comparison of differences between pre- and post-test scores. Descriptive statistics are presented as mean (X̄), standard deviation (SD)

Statistically significant results are shown in bold (p < 0.05)

A mixed-methods ANOVA revealed no significant baseline differences between groups for DASH subdomains (disability/symptom, sport/music, and work) (p > 0.05). A significant main effect of time was observed for all subdomains, indicating significant improvements from pre- to post-treatment in all groups (p < 0.05). A significant group × time interaction was found only for the disability/symptom subdomain (F = 3.933, p = 0.027, η² = 0.16). Post-hoc Bonferroni-adjusted comparisons showed that the Shoulder Mobilization and Cervical Mobilization groups demonstrated significantly greater improvements compared to the Conventional Treatment group (p < 0.05). Additionally, a significant group effect was observed for post-treatment disability/symptom scores (F = 4.029, p = 0.025, η² = 0.16), with the Conventional Treatment group showing higher (worse) scores than the other groups. No significant group × time interaction or between-group differences were observed for the sport/music and work subdomains (p > 0.05) (Table 5).

Table 5.

Comparison of DASH measurements across follow-up periods by groups

	Group
	ConventionalTreatment a	Shoulder Mobilization b	Cervical Mobilization c	Test Statistics†	Post Hoc(Bonferroni)	MCID (%95 CI)
	n=15	n=15	n=15
Disability/symptom score (0–100)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	37.93 ± 18.27	39.54 ± 11.82	40.01 ± 15.18	F = 0.076 p = 0.927 η2 = 0.01		-6.08 (-10.24; -1.92)
Post-treatment	19.74 ± 13.27	10.99 ± 9.90	10.09 ± 6.59	F = 4.029 p = 0.025 η2 = 0.16	a> (b = c)
Test Statisticsϕ	F = 31.616 p < 0.001 η2 = 0.429	F = 77.892 p < 0.001 η2 = 0.65	F = 85.565 p < 0.001 η2 = 0.671
Difference& (Post-Pre)	-18.19 ± 10.34	-28.55 ± 13.04	-29.92 ± 13.93	F = 3.933 p = 0.027 η2 = 0.16	a> (b = c)
sport/music score (0–100)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	13.75 ± 24.57	10.00 ± 18.42	3.33 ± 8.80	F = 1.227 p = 0.303 η2 = 0.06		-1.43(-7.3; 4.45)
Post-treatment	1.67 ± 6.45	0.00 ± 0.00	0.00 ± 0.00	F = 1.000 p = 0.76 η2 = 0.05
Test Statisticsϕ	F = 6.429 p = 0.015 η2 = 0.133	F = 4.403 p = 0.042 η2 = 0.095	F = 0.489 p = 0.488 η2 = 0.012
Difference& (Post-Pre)	-12.08 ± 24.60	-10.00 ± 18.42	-3.33 ± 8.80	F = 0.920 p = 0.406 η2 = 0.04
Work score (0–100)	X̄ ± SS	X̄ ± SS	X̄ ± SS
Pre-treatment	44.17 ± 24.37	48.42 ± 25.22	35.42 ± 25.95	F = 1.039 p = 0.363 η2 = 0.05		-5,34 (-10,67; -0,01)
Post-treatment	22.08 ± 22.39	13.75 ± 20.07	5.83 ± 11.20	F = 2,885 p = 0.067 η2 = 0.12
Test Statisticsϕ	F = 14.723 p < 0.001 η2 = 0.26	F = 36.281 p < 0.001 η2 = 0.463	F = 26.421 p < 0.001 η2 = 0.386
Difference& (Post-Pre)	-22.08 ± 15.82	-34.67 ± 25.42	-29.58 ± 24.37	F = 1.21 p = 0.308 η2 = 0.05

MCID Minimal Clinically Important Difference, CI Confidence Interval

ANOVA (F), Effect Size (η²), ^ϕWithin-group comparisons, ^†Between-group comparisons, ^&Comparison of differences between pre- and post-test scores. Descriptive statistics are presented as mean (X̄), standard deviation (SD)

Statistically significant results are shown in bold (p < 0.05)

Effect sizes varied across outcomes, ranging from small to large, with most measures demonstrating small to medium between-group effects.

Discussion

This study investigated the effects of shoulder and cervical mobilization on pain, ROM, sensation, and function in individuals with Subacromial Impingement Syndrome (SIS; n = 45). The results of our study demonstrated statistically significant improvements in pain, pain-free ROM, ROM, sensation, and functionality in all three treatment groups. In terms of pain, the Cervical Mobilization group showed a greater reduction compared to both the Conventional Treatment and Shoulder Mobilization groups. Similarly, improvements in the pain-free ROM were greatest in the Cervical Mobilization group and least in the control group. External rotation ROM increased more significantly in the mobilization groups than in the control group. The most notable gains in pressure pain threshold of the upper trapezius and in tactile sensation of the anterior shoulder were also observed in the Cervical Mobilization group. Regarding the DASH disability/symptom subscale, higher post-treatment scores in the control group indicated less improvement, while the Shoulder and Cervical Mobilization groups exhibited significantly better outcomes.

Evidence from systematic reviews supports the effectiveness of conservative interventions for SIS, with exercise therapy recommended as the first-line treatment and manual therapy considered a valuable adjunct. Meta-analytic findings suggest that the addition of manual therapy to exercise may result in greater short-term pain relief and functional improvement compared with exercise alone, although the overall quality of evidence remains low and effect sizes range from small to moderate [47, 48]. Consistent with these findings, mobilization techniques are frequently incorporated into rehabilitation programs for SIS to reduce pain, improve ROM, and enhance functional outcomes [49]. Approaches such as the Maitland and Mulligan techniques, particularly when combined with exercise therapy, have been associated with improvements in pain, ROM, and functional scores [4]. Nevertheless, methodological and anatomical variability has contributed to inconsistent findings across studies, particularly regarding cervicothoracic mobilization. In the present study, application of the Maitland technique to both the shoulder and cervical regions resulted in greater improvements in the mobilization groups, especially in pain-related and functional outcomes, supporting the value of a multimodal physiotherapy approach in SIS.

Chronic SIS-related pain involves both peripheral and central sensitization mechanisms [50]. PPT assessment helps identify central sensitization, as reduced PPT levels are typically seen in SIS patients [22]. Although interventions like mobilization, TENS, and hot packs can improve PPT, findings are variable [16]. PPT was measured at the supraspinatus, middle deltoid, and upper trapezius. All groups showed significant improvement; however, the Cervical Mobilization group exhibited superior improvement in the upper trapezius relative to the other groups. SIS patients typically show higher upper trapezius and lower middle/lower trapezius activation than healthy individuals, which may explain this finding. Furthermore, the MCID values exceeded the predefined thresholds, thereby reinforcing the clinical relevance of mobilization in reducing sensitivity.

While traditionally linked to nociceptive input, shoulder pain is also influenced by central mechanisms, as structural damage often poorly correlates with pain severity and may be present in asymptomatic individuals [51]. Two-point discrimination (2PD) thresholds are considered to reflect cortical reorganization in chronic pain conditions. In the present study, 2PD was assessed across three shoulder regions, and significant post-treatment improvements were observed in all groups. The greatest reduction in the anterior region was noted in the Cervical Mobilization group. This observation may be explained by the anatomical and neurophysiological relationship between the cervical spine and shoulder region. The suprascapular nerve, originating primarily from the C5–C6 nerve roots, provides both motor innervation to the supraspinatus muscle and sensory input from structures in the anterior shoulder region. Cervical mobilization may influence afferent input at the segmental level, potentially modulating sensory processing and improving tactile discrimination in related dermatomes. Similar segmental interactions between cervical interventions and shoulder sensory outcomes have been suggested in previous neuroanatomical and manual therapy literature, although direct causal mechanisms remain to be fully elucidated [18, 52]. The use of a non-bidirectional (ascending and descending) trial for TPD measurement may be considered a limitation, as it may not fully account for the error of anticipation. However, consistency was maintained by using the same assessor and standardized instructions for all participants. MCID values for anterior, middle, and posterior regions (4.42 mm, 3.42 mm, and 3.72 mm, respectively) offer the first reference data for SIS.

Pain in SIS is frequently associated with narrowing of the subacromial space and inflammation of surrounding tissues [50]. Pain experienced at specific angles during shoulder elevation is commonly assessed through the pain-free ROM, and both exercise and mobilization have been reported to improve this measure in patients with subacromial pain. For example, guided exercise combined with joint mobilization resulted in decreased pain during active movements compared with exercise alone or no treatment in a three-armed controlled trial of patients with subacromial pain syndrome [53]. Similarly, a randomized clinical trial comparing manual therapy to therapeutic exercise found significant improvements in shoulder pain and ROM after interventions, with manual therapy showing at least comparable benefits [54]. In our study, all groups improved significantly in the pain-free ROM, with the Cervical Mobilization group showing the greatest gains, suggesting that multimodal conservative approaches may enhance specific aspects of pain-related functional movement. The MCID value identified in this study constitutes the first reference for the pain-free ROM in SIS.

Shoulder ROM is critical for daily activities and is a key indicator of therapeutic progress. Reduced ROM in SIS is associated with decreased function and quality of life. Numerous studies have reported that exercise and manual therapy improve active ROM [10, 12, 21, 55]. Recent evidence supports the potential benefits of combining manual therapy with exercise to improve scapular ROM and pain in patients with shoulder impingement. A randomized controlled trial demonstrated that adding manual therapy to stretching and strengthening exercises led to statistically significant improvements in scapular protraction and upward rotation, as well as pain and functional capacity, compared with exercise alone [8]. Another recent RCT found that both manual therapy and therapeutic exercise improved shoulder pain, disability, and active ROM, with manual therapy producing greater pain reduction [42]. These findings align with our own reported ROM and pain improvements, reinforcing the role of multimodal conservative interventions in SIS management. This study also calculated MCID values for ROM in SIS for the first time, offering important baseline data.

Previous studies have shown that manual therapy, strengthening, and stretching interventions—particularly when combined with shoulder and cervical mobilization—lead to significant improvements in daily living activity performance [8, 56]. Eliason et al., the addition of joint mobilization to an exercise program was reported to have positive effects on shoulder function, pain, and active ROM, with functional outcome scores showing progressive improvement over time [57]. Similarly, Tauqeer et al., in their randomized controlled study, suggested that manual therapy combined with exercise may further enhance functional outcomes in individuals with subacromial impingement syndrome (SIS) [8]. In another study, manual therapy interventions were found to improve functional in patients with SIS [48]. In line with previous studies, our findings indicate that the applied intervention effectively improved functional performance as measured by the Disabilities of the Arm, Shoulder and Hand (DASH) questionnaire in the study population.

All Effect sizes varied across outcomes, ranging from small to large, with most measures demonstrating small to medium between-group effects. Furtermore all MCID values with 95% confidence intervals crossing zero should not be interpreted as statistically established thresholds. Nevertheless, the reported MCID values offer clinically informative estimates and should be interpreted with caution and in conjunction with other outcome measures.

Study limitations

This study has several limitations. Participant and therapist blinding was not feasible, and all manual therapy interventions were delivered by a single physiotherapist, which may have introduced therapist effects. The single-centre design further limits causal interpretation and generalizability.

The small sample size may have reduced statistical power and increased the risk of Type I error given the multiple outcomes assessed. Participants who withdrew were replaced to maintain group balance, which deviates from a strict intention-to-treat approach. Additionally, no long-term follow-up was conducted, sensory measures may be subject to measurement variability, and the inclusion criteria (VAS ≥ 4, ≥3 months of symptoms) limit generalization to milder or acute cases.

Conclusions

This study examined the effects of adding cervical mobilization to shoulder mobilization in patients with SIS using a comprehensive assessment that included sensory outcomes. Improvements over time were observed in pain, range of motion, function, and selected sensory measures; however, between-group differences were limited.

The findings suggest that sensory outcomes may show measurable changes following rehabilitation interventions and can provide complementary information alongside traditional clinical measures. Within the limitations of this study, the additional effects of cervical mobilization should be interpreted cautiously, and further research is required to clarify its clinical relevance in the management of SIS.

Abbreviations

SIS

Subacromial Impingement Syndrome

VAS

Visual analog scale

PPT

Pressure pain threshold

ROM

Range of motion

DASH

Disabilities of the arm, shoulder and hand

TENS

Transcutaneous electrical nerve stimulation

2PD

Two-point discrimination

Authors’ contributions

O.U. and M.I. contributed to the conceptualization and methodology of the study. O.U. conducted the formal analysis, investigation, and data curation. O.U. prepared the original draft of the manuscript. M.I. reviewed and critically edited the manuscript and supervised the study. All authors read and approved the final manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to patient privacy considerations but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Non-Interventional Clinical Research Ethics Committee of Kutahya Health Sciences University (Decision No: 2024/06–29), with additional permission from the Kutahya Provincial Health Directorate. Written informed consent was obtained from all participants prior to participation. The study was conducted in accordance with the Declaration of Helsinki. Data were stored on password-protected computers accessible only to the research team, and participants retained the right to access their data and request removal of identifying information.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.