We carried out a prospective cohort study in a private physiotherapy clinic located in Athens, Greece, between April 2023 and June 2025. Participants’ recruitment was conducted through invitations sent from the University of West Attica and referrals from private clinics located in Athens, Greece. Each participant signed an informed consent form in compliance with the World Medical Association Declaration of Helsinki (2013) [19]. A single physical therapist (ND), who had 23 years of clinical experience in musculoskeletal problems and had completed a 2-year postgraduate training in orthopedic manipulative therapy, conducted all treatments. The design of our study was in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [20].

To be included subjects should be between 18 and 60 years of age, with non-specific LBP, and without symptoms of lower-extremity radicular pain.

Subjects between 18 and 60 years of age with non-specific LBP without symptoms of lower-extremity radicular pain that exceeded 12 weeks were included in our study. Participants were excluded if they had LBP with neurological deficits, spinal fractures, trauma or surgery in the spine, cancer history, osteoporosis, prolonged steroid use, sudden weight loss, or cauda equina syndrome [2].

The initial assessment was conducted by a musculoskeletal physiotherapist (SK) with 18 years of experience. Demographic characteristics such as sex, age, height, weight, symptom duration, and previous experience of lumbar spine manipulation were collected. Patients reported the pain intensity of their last episode of low back pain within the past 7 days, using a Numeric Rating Pain Scale (NRPS) from 0 to 10. Additionally, patients completed the Expectations for Treatment Scale (ETS) questionnaire and the Oswestry Disability Index (ODI) questionnaire.

Physical examination included patient history and observation, along with special tests, palpation, and movement examination. All the assessment procedures were aligned with the International Framework endorsed by the International Federation of Manual and Musculoskeletal Physical Therapists (IFOMPT) for identifying red flags for serious spinal pathologies [21]. Subsequently, all patients were familiarized with the PPTs testing procedure at both local and remote sites. An independent assessor (NZ) conducted all PPTs measurements at each site prior to and following the intervention.

All patients were lying in a relaxed side-lying position. The treating physiotherapist (ND) performed a high-velocity, low-amplitude (HVLA) thrust at the most hypomobile segment of the lumbar spine as assessed by the same physiotherapist. A second attempt was performed if needed, when the first application was not accompanied by an audible pop. We theorized that the sound of an audible pop was indicative of a successful attempt. PPTs were measured immediately after the technique application by the same assessor (NZ).

Participants completed the ETS and ODI questionnaires before baseline PPT testing. Baseline PPT assessment required approximately 10 - 15 minutes. The lumbar manipulation was then performed, and post-manipulation PPT measurements commenced immediately after the intervention and were completed within approximately 10 - 15 minutes.

Treatment-related outcome expectations were assessed by the ETS questionnaire. The ETS comprises 5 items that are intended to evaluate patients’ expectations related to a specific treatment in terms of symptom reduction and general health status [22]. Each item is scored using a 4-point scale from 1 (partially disagree) to 4 (definitely agree). Consequently, the total score ranges from 5 to 20, with higher scores indicating higher treatment expectations. For this study, the Greek version of ETS was used. The Greek version of the ETS questionnaire has shown a high level of content validity, an excellent test-retest reliability (ICC: 0.96), and an acceptable internal consistency (Cronbach’s alpha: 0.84) in patients with musculoskeletal conditions [23].

The ODI questionnaire is a short and simple to complete instrument that aims to assess self-reported disability in patients with low back pain. The ODI questionnaire consists of 10 items, with each of them can be scored on a 6-grade scale from 0 to 5. Total scores range from 0 to 50, with higher scores indicating greater disability and limitations. The final score is expressed as a percentage. The Greek version of the ODI questionnaire has shown good i nternal consistency (Cronbach’s alpha: 0.833) in Greek-speaking patients with low back pain [24].

In this study, PPTs at both local and remote sites were measured. The algometer was placed at the spinous process of L5, at the paravertebral musculature bilaterally of L5, at the midpoint of the upper trapezius muscle bilaterally, and at the anterior tibialis muscle bilaterally.

All measurements were carried out in a seated and prone-lying position. The sites located in the lumbar spine were assessed in the prone-lying position, whereas anterior tibialis and upper trapezius were assessed in a seated position. For each site, three repetitions with a 30-second interval were applied, and the average value was recorded [25]. During the measurement, patients were guided to respond with the word “now” when the pressure stimulus became painful. The same procedure was held prior to and following the intervention, and all measurements were taken by the same blinded assessor (NZ). A hand-held digital algometer with a 1-cm diameter probe (Baoshishan ZP-1000 N 20/22806, Baoshishan, Shenzhen, China) was used for this study. Pressure application rate was set at 0.5 kgf/cm² per second (50 kPa/s) [25]. Based on an intra-rater reliability pilot-test conducted in healthy subjects prior to the main study we found excellent intraclass correlation coefficient (ICC) values for L5 spinous process (ICC: 0.93), left (ICC: 0.89) and right (ICC: 0.97) paravertebral area, left (ICC: 0.89) and right (0.95) upper trapezius, left (ICC: 0.97) and right (0.89) tibialis anterior.

Participants’ history, physical examination, demographic characteristics, NRPS, ODI, and ETS scores were collected by an independent physiotherapist who did not participate in other measurements and treatment. Also, the PPTs assessor and treating physiotherapist were blinded to the participants’ NRPS, duration of symptoms, ETS, and ODI scores. During the study, the PPTs assessor and treating physiotherapist were in different rooms in order to minimize possible interaction with the patients and to maximize adherence to the study protocol. Therapists followed the protocol by avoiding discussions about treatment expectations. While this raised ethical concerns by limiting the therapeutic relationship essential to patient care, it preserved the study’s integrity and scientific rigor.

An a priori sample-size estimation was performed during study planning using G*Power (version 3.1.9.7). At the time of study design, no previous studies had evaluated the association between treatment-related outcome expectations and changes in pressure pain thresholds following spinal manipulation in patients with chronic NSLBP. Therefore, the calculation was based on the closest available evidence by Wang et al. (2012), reporting between-group differences in lumbar PPTs following spinal manipulation in healthy individuals [26]. Using an anticipated effect size (Cohen’s d) of 0.70, power of 80%, and α = 0.05, a minimum sample of 56 participants was estimated [27]. Because the present study primarily investigated associations between expectations and changes in PPTs rather than between-group differences, this calculation should be considered approximate. Accordingly, the study should be interpreted as exploratory and hypothesis-generating.

The normality of the data was evaluated using the Shapiro-Wilk test, along with visual examination of Q-Q plots and boxplots as well. Normally distributed data were presented with mean values and standard deviations (SD), while non-normal data were presented with median values and interquartile ranges (IQR).

The primary outcome was the change in PPTs (post-intervention PPT minus pre-intervention PPT) following lumbar spinal manipulation. The primary hypothesis was that higher treatment-related outcome expectations, measured using the ETS, would be associated with greater increases in PPTs. Therefore, the primary analysis consisted of Wilcoxon signed-rank test was used to assess changes between ETS scores and changes in PPTs. Analyses based on ETS categorization (high versus low expectations) were considered exploratory sensitivity analyses.

Associations between the dichotomous variables of sex and previous experience of manipulation with changes in PPTs and ETS scores were explored using the Mann-Whitney U test. Spearman’s rank correlation coefficients (rho) were calculated to investigate associations between changes in PPTs and the following continuous variables: ETS score, NRPS, ODI score, age, body mass index (BMI), and symptom duration. Spearman’s correlation coefficient was classified as weak (≤0.39), moderate (0.4 to 0.69), or strong (0.7 to 1.0) correlation (Dancey & Reidy 2007). Due to the absence of established cut-off points for the ETS score, sensitivity analyses (Mann-Whitney U test) were performed based on the mean value and the 33^rd and 66^th percentiles. It should be noted that the association between ETS scores and changes in PPTs constituted the primary analysis. All additional analyses examining individual body sites, demographic and clinical variables, previous manipulation experience, sex, and alternative ETS categorizations were considered exploratory.

All forementioned analyses were performed in SPSS (v25), with p < 0.05 considered significant.

The prespecified primary analysis examined the association between treatment-related outcome expectations (ETS scores) and changes in PPTs following lumbar manipulation. No statistically significant associations were observed between ETS scores and PPT changes at any local or remote measurement site (Table 3).

Secondary exploratory analyses examined associations between PPT changes and age, disability, pain intensity, BMI, symptom duration, sex, previous manipulation experience, and alternative ETS categorizations. A weak correlation between baseline ODI scores and changes in PPTs at the L5 spinous process was discovered (r = −0.273, p = 0.042), but this did not occur at any of the other sites. Similarly, no correlation was observed between age and PPTs changes, except at the left trapezius muscle, where a weak correlation was presented (r = 0.370, p = 0.005). Detailed results of the Spearman’s RHO test are presented in Table 3.The dichotomous variables of sex and previous experience of manipulation did not appear to correlate with the changes in PPTs across all measurement sites. The corresponding results of the Man-Whitney U test are presented in Table 4 and Table 5, respectively.

Table 3. Pressure pain thresholds (kg/cm²) between pre- and post-HVLA thrust at local and distal sites ofthe body in patients with high and low treatment expectations.

Abbreviations: r, correlation coefficient; p, p-value; kg/cm², kilograms/centimeters²; ETS, expectations for treatment score; ODI, Oswestry disability index; BMI, body mass index.

Table 4. Pressure pain thresholds (kg/cm²) between pre- and post-HVLA thrust at local and distal sites of the body in female and male patients.

^aValues are expressed as median (25 - 75% IQR). ^bMann-Whitney U test, between-group comparison of the change in PPTs at each site of measurement. Abbreviations: kg/cm², kilograms/centimeters².

Based on the sensitivity analyses, there was no difference in changes in PPTs between patients with low and high treatment expectations, irrespective of the cut-off point used. The results of the Man-Whitney U test using a cut-off of ETS = 12.7 are presented in Τable 6. For the second analysis, the 33rd (ETS = 9.5) and the 66^th (ETS = 14) percentiles were used to assign participants to low, medium, and high treatment expectations groups. The results of the second Mann-Whitney U test between high and low treatment expectation groups are presented in Table 7.

Table 5. Pressure pain thresholds (kg/cm²) between pre- and post-HVLA thrust at local and distal sites in patients with and without previous experience of manipulation.

^aValues are expressed as median (25 - 75% IQR). ^bMann-Whitney U test, between-group comparison of the change in PPTs at each site of measurement. Abbreviations: kg/cm², kilograms/centimeters².

Table 6. Pressure pain thresholds (kg/cm²) between pre- and post-HVLA thrust at local and distal sites in patients with high and lowtreatment expectations (cut-off point = 12.7).

Values are expressed as median (25% - 75% IQR). Abbreviations: kg/cm², kilograms/centimeters².

Table 7. Pressure pain thresholds (kg/cm²) between pre- and post HVLA thrust at local and distal sites in patients with high and low treatment expectations (High expectations > 14, Low expectations ≤ 9.5).

^aValues are expressed as median (25% - 75% IQR). ^bMann Whitney U test, between group comparison of the change in PPTs at each site of measurement. Abbreviations: kg/cm², kilograms/centimeters².

The findings of this study should be interpreted with caution regarding its limitations. Our results can be generalized only in mostly young females with low levels of LBP-related disability and mild to moderate levels of pain intensity. In general, chronic NSLBP population consists of patients typically older in age with greater levels of pain and disability [45]. Thus, our sample is not entirely representative of the population being studied.

Furthermore, multiple statistical tests were performed across several measurement sites and explanatory variables. As these analyses were exploratory and no formal adjustment for multiplicity was applied, there is an increased risk of Type I error. Therefore, isolated statistically significant findings, particularly those not directly related to the primary hypothesis, should be interpreted with caution and require confirmation in future studies.

We did a single SMT session and measured the PPTs immediately after, with no follow-up measurements. This could have influenced the ability of this study to detect any correlation between expectations and PPT changes, as the expectation-induced placebo analgesia can take up to a few minutes to begin manifesting, and it peaks several minutes after the intervention in some cases [46]. Also, it can be strengthened with repeated exposure to the intervention through conditioning and reinforcing expectations. Following the same trend, some data suggest that PPTs changes can be observed 5-30 minutes after lumbar SMT in healthy individuals [47].